Biophysical Techniques in Photosynthesis

Biophysical Techniques in Photosynthesis

Advances in Photosynthesis VOLUME 3

Series Editor: GOVINDJEE Department of Plant Biology University of Illinois, Urbana, Illinois, U.S.A. Consulting Editors: Jan AMESZ, Leiden, The Netherlands Eva-Mari ARO, Turku, Finland James BARBER, London, United Kingdom Robert E. BLANKENSHIP, Tempe, Arizona, U.S.A. Norio MURATA, Okazaki, Japan Donald R. ORT, Urbana, Illinois, U.S.A. Advances in Photosynthesis provides an up-to-date account of research on all aspects of photosynthesis, the most fundamental life process on earth. Photosynthesis is an area that requires, for its understanding, a multidisciplinary (biochemical, biophysical, molecular biological, and physiological) approach. Its content spans from physics to agronomy, from femtosecond reactions to those that require an entire season, from photophysics of reaction centers to the physiology of the whole plant, and from X-ray crystallography to field measurements. The aim of this series of publications is to present to beginning researchers, advanced graduate students and even specialists a comprehensive current picture of the advances in the various aspects of photosynthesis research. Each volume focusses on a specific area in depth.

The titles to be published in this series are listed on the backcover of this volume.

Biophysical Techniques in Photosynthesis Edited by

Jan Amesz and Arnold J. Hoff Department of Biophysics Huygens Laboratory, University of Leiden, 2300 RA Leiden The Netherlands

KLUWER ACADEMIC PUBLISHERS NEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW

eBook ISBN: Print ISBN:

0-306-47960-5 0-7923-3642-9

©2004 Kluwer Academic Publishers New York, Boston, Dordrecht, London, Moscow Print ©1996 Kluwer Academic Publishers Dordrecht All rights reserved No part of this eBook may be reproduced or transmitted in any form or by any means, electronic, mechanical, recording, or otherwise, without written consent from the Publisher Created in the United States of America Visit Kluwer Online at: and Kluwer's eBookstore at:

http://kluweronline.com http://ebooks.kluweronline.com

Contents

Preface

xi

Part One: Optical Methods 1

Developments in Classical Optical Spectroscopy Jan Amesz Summary I. Introduction II. Absorption and Absorption Difference Spectroscopy III. Fluorescence References

2

3

3 3 3 4 6 8

Linear and Circular Dichroism Garab

11

Summary I. Introduction II. Linear Dichroism III. Circular Dichroism IV. Concluding Remarks Acknowledgements References

11 12 16 24 35 35 35

Fluorescence Kenneth Sauer and Martin Debreczeny Summary I. Introduction II. Steady-State Fluorescence III. Time-Resolved Fluorescence IV. Conclusion Acknowledgements References

41 41 42 45 52 59 59 60

v

vi

4

5

6

Contents Ultrafast Spectroscopy of Photosynthetic Systems Ralph Jimenez and Graham R. Fleming Summary I. Introduction II. Laser Sources III. Fluorescence Upconversion IV. Transient Absorption V. Concluding Remarks References

63 63 64 66 70 72 72

Data Analysis of Time-Resolved Measurements Alfred R. Holzwarth

75

Summary I. Introduction II. Methods for Time-Resolved Data Analysis III. Some Applications to Photosynthesis IV. Conclusions Acknowledgements References

75 76 76 87 89 91 91

Photosynthetic Thermoluminescence as a Simple Probe of Photosystem II Electron Transport Yorinao Inoue Summary I. Introduction II. Origins of TL from Photosynthetic Apparatus III. Application of TL as a Probe of PSII Photochemistry IV. Perspective: Merits and Demerits of TL Technique Acknowledgements References

7

63

Accumulated Photon Echo Measurements of Excited State Dynamics in Pigment– Protein Complexes Thijs J. Aartsma, Robert J.W. Louwe and Peter Schellenberg Summary I. Introduction II. Homogeneous and Inhomogeneous Linewidths III. Photon Echo Phenomena IV. Accumulated Photon Echo: Experimental V. Energy Transfer VI. Photon Echo Experiments on Reaction Centers VII. Conclusion and Perspectives Acknowledgements References

93 93 94 94 103 104 105 105

109 109 109 110 111 114 116 118 119 120 120

Contents

8

vii

Spectral Hole Burning: Methods and Applications to Photosynthesis N. Raja S. Reddy and Gerald J. Small Summary I. Introduction II. Experimental Methods III. Applications Acknowledgements References

9

Infrared and Fourier-Transform Infrared Spectroscopy Werner Mäntele Summary I. Introduction: Looking Back 100 Years II. What can Infrared Spectroscopy Tell us about the Processes in Photosynthetic Membranes and Reaction Centers? III. From Bands to Bonds: Strategies for Band Assignments IV. Fourier-Transform Infrared (FTIR) Spectroscopy V. Single Wavelength IR Techniques VI. Sample Preparation for Infrared Spectroscopy VII. Conclusions and Outlook Acknowledgements References

10

Resonance Raman Studies in Photosynthesis – Chlorophyll and Carotenoid Molecules Bruno Robert Summary I. Introduction II. Introduction to Raman and Resonance Raman Spectroscopy III. Resonance Raman Spectroscopy of Photosynthetic Pigments IV. Resonance Raman Spectroscopy as Method of Chemical Analysis in Photosynthesis V. Resonance Raman Spectroscopy as a Probe for Molecular Conformation VI. Resonance Raman Spectroscopy as a Probe for Intermolecular Interactions VII. Time-Resolved Resonance Raman Studies VIII. Resonance Raman Spectroscopy as a Probe for Studying the Nature of Electronic Transitions IX. Perspectives Acknowledgements References

11

Stark Spectroscopy of Photosynthetic Systems Steven G. Boxer Summary I. Introduction II. Methods III. Limitations and Conceptual Issues IV. Examples of Recent Results for Photosynthetic Systems Acknowledgements References

123 123 124 126 129 134 135

137 137 138 139 139 141 152 155 157 157 157

161 161 162 162 163 167 168 169 171 172 173 174 174

177 177 177 178 181 184 188 188

viii

12

Contents The Photoacoustic Method in Photosynthesis – Monitoring and Analysis of Phenomena Which Lead to Pressure Changes Following Light Excitation Shmuel Malkin Summary Introduction – Historical Notes and Main Aspects I. II. Experiments and Results with the Gas-phase Coupled Microphone Time Domain with a Sample Coupled III. Experiments and Results in the Piezoelectric Sensor IV. Applications to Physiological Studies References

191 191 192 194 202 204 204

Part Two: Magnetic Resonance 13

Magnetic Resonance: An Introduction Arnold J. Hoff

14

Time-Resolved Electron Paramagnetic Resonance Spectroscopy – Principles and Applications Haim Levanon Summary Introduction I. II. Experimental III. Results IV. Concluding Remarks Acknowledgements References

15

Electron Spin Echo Methods in Photosynthesis Research R. David Britt Summary I. Introduction II. ESEEM III. ESE-ENDOR IV. Additional Examples of ESE Applications in Photosynthesis V. Instrumentation Acknowledgements References

16

ENDOR Spectroscopy Wolfgang Lubitz and Friedhelm Lendzian Summary I. Introduction II. Principles of Electron–Nuclear Multiple Resonance Spectroscopy III. Selected Applications of ENDOR to Photosynthesis IV. Concluding Remarks Acknowledgements References

209

211 211 212 213 218 229 229 230

235 235 235 238 243 246 249 252 252

255 255 256 258 268 272 272 272

Contents

17

ix

Optically Detected Magnetic Resonance (ODMR) of Triplet States in Photosynthesis Arnold J. Hoff Summary I. Introduction II. The Triplet Spin Hamiltonian in Zero Magnetic Field III. Optical Detection of Magnetic Resonance, ODMR IV. Double Resonance V. ODMR in Photosynthesis VI. Concluding Remarks References

18

Magic Angle Spinning Nuclear Magnetic Resonance of Photosynthetic Components Huub J.M. de Groot Summary I. Introduction II. Magic Angle Spinning NMR Spectroscopy III. Probing the Local Environment of M(Y)210 in Rb. sphaeroides R26 RC with MAS IV. The Configuration of the Spheroidene in the Rb. sphaeroides RC V. The Asymmetric Binding in Rb. sphaeroides R26 VI. New Developments. CIDNP and Correlation Spectroscopy VII. Concluding Remarks Acknowledgements References

277 277 278 278 279 284 288 295 295

299 299 300 300 302 305 306 309 312 312 312

Part Three: Structure and Oxygen 19

Structure Determination of Proteins by X-Ray Diffraction Marianne Schiffer Summary I. Introduction II. Theory, Equations, and Some of the Terms Used in X-Ray Structure Determination III. Determination of Protein Structure IV. Quality of the Structure V. Comparison with Structural Information Obtained with Other Techniques Acknowledgements References

20

317 317 317 318 318 323 323 323 324

Electron Microscopy Egbert J. Boekema and Matthias Rögner

325

Summary I. Principles II. Periodic Averaging III. Single Particle Averaging IV. Concluding Remarks Acknowledgements References

325 326 330 332 335 335 335

x

21

Contents X-Ray Absorption Spectroscopy: Determination of Transition Metal Site Structures in Photosynthesis Vittal K. Yachandra and Melvin P. Klein Summary I. Introduction II. X-Ray Absorption Spectroscopy (XAS) III. Applications of XANES and EXAFS in Photosynthesis IV. Future Directions Acknowledgements References

22

Mössbauer Spectroscopy Peter G. Debrunner Summary Introduction I. II. Mössbauer Spectroscopy: Physics and Formalism III. Applications References

23

Characterization of Photosynthetic Supramolecular Assemblies Using Small Angle Neutron Scattering David M. Tiede and P. Thiyagarajan Summary I. Introduction II. Small Angle Neutron Scattering III. SANS Studies of Photosynthetic Complexes IV. Concluding Remarks Acknowledgements References

24

Measurement of Photosynthetic Oxygen Evolution Hans J. van Gorkom and Peter Gast Summary I. Introduction II. Polarography III. EPR Oximetry IV. Mass Spectrometry V. Photoacoustic Spectroscopy VI. Galvanic Sensors VII. Prospects Acknowledgements References

Index

337 337 338 338 345 350 351 352

355 355 356 357 365 371

375 375 376 377 379 388 388 389

391 391 392 392 398 401 402 402 402 403 403

407

Preface Progress in photosynthesis research is strongly dependent on instrumentation. It is therefore not surprising that the impressive advances that have been made in recent decades are paralleled by equally impressive advances in sensitivity and sophistication of physical equipment and methods. This trend started already shortly after the war, in work by pioneers like Lou Duysens, the late Stacy French, Britton Chance, Horst Witt, George Feher and others, but it really gained momentum in the seventies and especially the eighties when pulsed lasers, pulsed EPR spectrometers and solid-state electronics acquired a more and more prominent role on the scene of scientific research. This book is different from most others because it focuses on the techniques rather than on the scientific questions involved. Its purpose is three-fold, and this purpose is reflected in each chapter: (i) to give the reader sufficient insight in the basic principles of a method to understand its applications (ii) to give information on the practical aspects of the method and (iii) to discuss some of the results obtained in photosynthesis research in order to provide insight in its potentalities. We hope that in this way the reader will obtain sufficient information for a critical assessment of the relevant literature, and, perhaps more important, will gain inspiration to tackle problems in his own field of research. The book is not intended to give a comprehensive review of photosynthesis, but nevertheless offers various views on the exciting developments that are going on. The methods discussed in the book can be roughly divided in three categories: methods of optical electronic and vibrational spectroscopy, magnetic resonance methods and methods mainly aimed at obtaining structural information. Optical methods and phenomena that are discussed include linear and circular dichroism, time resolved fluorescence and absorbance measurements, including their data analysis, vibrational spectroscopy (infrared and resonance Raman) and specialized techniques, such as photon echo, hole burning, Stark and photoacoustic spectroscopy. The section on magnetic resonance is mainly devoted to electron spin resonance and the various techniques that apply: EPR, ESE, ENDOR and ODMR. One chapter is devoted to magic angle spinning NMR. Knowledge of the structure of the photosynthetic apparatus is a prerequisite for obtaining insight in its function. With one exception (oxygen measurements) the third section of the book concerns methods that are primarily aimed at obtaining such structural information. A variety of techniques is discussed: Xray diffraction, X-ray absorption, electron microscopy, Mössbauer spectroscopy and neutron scattering. Although the book does not pretend to give an exhaustive overview of all types of physical measurements in photosynthesis, we feel that it gives a fairly comprehensive picture of the most important techniques and of their applications. This book has been made possible by the help and effort of many. First of all we are indebted to the authors of the various chapters. All of them, we think, furnished us with first-rate contributions highlighting their field of specialization. Second, we would like to thank the editor-in-chief of the series, Govindjee, who engendered the idea for this book, and with incessant and unfailing enthusiasm guided us with his electronic messages. Third, we thank the secretarial staff of our department, Mrs. B.C. van Dijk and Mrs. M.J. Gouw who helped us in various ways. Finally we want to acknowledge the skill of Gilles Jonker and his staff at Kluwer in producing the book. Jan Amesz Arnold J. Hoff xi

This page intentionally left blank

PART ONE

Optical methods

This page intentionally left blank

Chapter 1 Developments in Classical Optical Spectroscopy Jan Amesz Department of Biophysics, Huygens Laboratory, University of Leiden, P.O. Box 9504, 2300 RA Leiden, The Netherlands

3 3 4 6 8

Summary I. Introduction II. Absorption and Absorption Difference Spectroscopy III. Fluorescence References

Summary An overview is given of the development of optical techniques as applied to photosynthesis research during the last 50 years and their importance for present day research is discussed. The review concerns the “classical” techniques, i.e. measurements of absorbance and of light-induced changes of absorbance and fluorescence emission and excitation spectroscopy. Abbreviations: BChl – bacteriochlorophyll; Chl – chlorophyll; FMO protein – Fenna–Matthews–Olson protein of green sulfur bacteria

I. Introduction

devices and new light sources became available, including pulsed lasers for flash spectroscopy as well as computers for data processing and registration. Hand in hand with these developments optical studies of photosynthesis have acquired growing importance for gaining insight in the molecular mechanisms of phosynthesis. In the chapters that follow, accounts will be given of the present state of the art, and examples will be given of the information obtained by modern optical methods. This chapter will survey some of the developments during the last 50 years and will discuss some of the “traditional” methods that still play an important role in photosynthesis research. Pioneers of the early days were H. Kautsky and E.C. Wassink and, at a somewhat later stage, L.N.M. Duysens, C.S. French, B. Chance, H.T. Witt and B. Kok. The first two studied the fluorescence properties of photosynthetic material (Kautsky and Hirsch, 1931; Kautsky and Franck,

In view of the key role of pigments in photosynthesis, it is not surprising that optical methods have played, and continue to play, an important role in photosynthesis research. Engelmann (1882) was the first to show, by means of action spectra of oxygen production, that chlorophyll and the so-called accessory pigments are involved in photosynthesis. Progress in optical research on photosynthesis, however, was for a long time arrested, mainly because simple and sensitive techniques for measuring and recording light intensities were lacking. About 50 years ago, however, a rapid development of optical techniques set on. World War II saw the development of the photomultiplier, and in the years that followed faster and more reliable and sensitive electronic Correspondence: Fax: 31-71-5275819; E-mail: [email protected]

3 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 3–10. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

4

1943; Vermeulen et al., 1937), while spectroscopy of light-induced absorbance changes was pioneered by Duysens (1952). In particular, the latter technique has proved invaluable to study the components of photosynthetic electron transport. One of the early results of such studies was, around 1960, the discovery of the two photosystems in plant photosynthesis (Duysens et al., 1961). The use of pulsed lasers was initiated in the sixties (DeVault and Chance, 1966; Wolff et al., 1969; Netzel et al., 1973) and has now progressed into the femtosecond region, enabling the study of early processes of energy transformations in excited pigments (see Chapter 4 by Jimenez and Fleming). French (French and Young, 1952) and in particular Duysens (1952) were the first to apply fluorescence spectroscopy to study energy transfer in photosynthetic organisms. French also devised various ingenious apparatus for the deconvolution of absorption spectra, the automatic recording of action spectra and for the measurement of so-called derivative spectra (French et al., 1954; French, 1955; French and Harper, 1957; Allen et al., 1960). Today, such measurements are routinely, and much more conveniently, performed with the aid of computer analysis. In the next two sections, we shall survey these developments in some more detail, and briefly discuss the importance of these “classical” optical techniques in modern photosynthesis research. II. Absorption and Absorption Difference Spectroscopy Measurement of the absorption spectrum is one of the basic methods to obtain information about the characteristics of photosynthetic material. As this is normally done with commercial apparatus an extensive discussion of the method should not be necessary here. Nevertheless, absorption spectra of rather poor quality are being published occasionally even today, and this is mainly due to the fact that these commercial apparatus are not designed for scattering material. Light scattering, if not properly corrected for, not only causes an upward shift and distortion of the absorption spectrum, but it may also decrease the amplitude of absorption bands (Amesz et al., 1961; Latimer and Eubanks, 1962). Moreover,

Jan Amesz additional distortion may occur due to selective scattering near the absorption bands (Latimer, 1959). The effects can be minimized by collecting the transmitted light over a relatively large angle. Other methods that may be applied are adjusting the refractive index of the medium and the socalled opal glass method (Shibata, 1958), the latter, however, at the expense of sensitivity. In some cases reliable data can be obtained by fluorescence detected absorbance (Kramer et al., 1985). Of course, the same principles apply to more specialized absorption measurements, such as linear and circular dichroism and absorption difference spectroscopy. It should be noted, however, that even a properly measured absorption spectrum is not identical to that of the same pigments in solution, if the pigments are contained in particles that have a non-negligible absorption. This is the so-called “flattening effect” (Duysens, 1956). Due to the presence of different “pools” of chemically identical pigments or to excitonic interactions the in vivo absorption spectra of photosynthetic pigments nearly always consist of strongly overlapping absorption bands which are, moreover, inhomogeneously broadened. French and coworkers (Allen et al., 1960) determined the first derivatives of the absorption spectra to distinguish the various in vivo absorption bands of chlorophyll. A more convenient and nowadays extensively used method to enhance the resolution of absorption (or other) spectra is by measuring the second or even fourth derivatives (Martin, 1959; Butler and Hopkins, 1970a, 1970b; see Fig. 1). Caution, however, is needed in the interpretation because the resulting bands are not only sharpened, but side bands are also generated by the differentiation. Duysens (1952,1957) was the first to apply the measurement of changes of absorbance, induced by illumination, to the study of photosynthesis. Since then, this method has continuously gained importance and it is still one of the most effective methods to study molecular processes in photosynthesis. Although pump-probe measurements with high time resolution are now in the forefront of research (see Chapter 4 by Jimenez and Fleming), the classical methods, using a continuous or semi-continuous measuring beam, are still being extensively used in photosynthesis research, and

Classical optical spectroscopy

these are still of sufficient importance to warrant a brief discussion. In the “older” apparatus, such as described by Duysens (1957) and Chance (1951), the measuring beam was modulated mechanically and was either split into a measuring and a reference beam or alternatively passed two monochromators set at different wavelengths. The resulting a.c. photomultiplier signal was then fed into a lock-in amplifier to reduce effects of the non-modulated “actinic” illumination and to provide enhanced stability and sensitivity. Alternatively, modulated actinic light has been used for generating modulated signals of intermediates with a sufficiently

5

short lifetime (Kok, 1959; Spruit, 1971; Nishimura et al., 1969). The relatively slow modulation employed in the above mentioned apparatus precludes measurements of absorbance changes faster than a few ms. In fact, the development of rapid and stable d.c. amplifiers has largely obviated the need for light modulation, and apparatus with a continuous measuring beam, employing xenon or laser flash actinic illumination have been extensively used in photosynthesis research. Fluorescence artifacts can be corrected for, if necessary, by subtracting the signal obtained without a measuring beam. Flash spectroscopy was pio-

6 neered by Porter and Norrish (see Porter, 1968) to study reactions in gases and liquids. Early apparatus for use in photosynthesis research have been described by Witt et al. (1959), Wolff et al. (1969) and Ke et al. (1964). A time resolution of is easily obtained, but can be extended to about 20 ns (Wolff et al., 1969). A 2 ns resolution has been obtained by using a xenon flash as a quasi-continuous light source (van Bochove et al., 1984; Kleinherenbrink, 1992). An apparatus for measuring time resolved difference spectra based on an array of pulsed light emitting diodes has been described by Klughammer et al. (1990).

III. Fluorescence Fluorescence from leaves and photosynthetic pigments was first observed by D. Brewster and G.G. Stokes, in the mid-nineteenth century (see Rabinowitch, 1951), but it was only about a century later when fluorescence measurements became an increasingly important tool in photosynthesis research. An interesting quantitative survey of the various topics studied by fluorescence in 200 publications of the period 1978– 1983 can be found in the review of Lavorel et al. (1986). Measurements with high time resolution will not be discussed in this chapter; and neither will be those of fluorescence polarization; for a discussion of those measurements the reader is referred to Chapter 3 by Sauer and Debreczeny. But also the “classical” fluorescence techniques still provide an important tool in photosynthesis research. They may be roughly divided in three types: measurements of emission spectra, of excitation spectra and measurements of (relative) fluorescence yields. All three methods require relatively simple equipment, which nevertheless have been perfected steadily during the last five or six decades, (Lavorel et al., 1986; Schreiber, 1986; Schreiber et al., 1993) with digital processing being standard nowadays. A discussion of some of the pitfalls and possible errors, like those caused by false light and self-absorption of fluorescence can be found in Amesz (1973). An apparatus specially designed for photosynthesis research is now commercially available from Walz, Effeltrich, Germany. Apparatus have also been

Jan Amesz devised for field studies and productivity measurements (Renger and Schreiber, 1986; Öquist and Wass, 1988). The prominent emission bands in photosynthetic material are normally those of the longestwavelength Chls or BChls of the system, due to energy transfer from short-wavelength to longwavelength absorbing pigments. Emissions from shorter-wavelengths absorbing pigments are usually weaker because thermal equilibrium favors emission from the pigment with the lowest energy. Conspicuous exceptions are found in the fluorescence spectra of green bacteria, red algae and cyanobacteria, where relatively strong bands are observed from the chlorosomes and the phycobilisomes (Amesz and Vasmel, 1986; Fork and Mohanty, 1986), showing that the efficiency of energy transfer from these extramembranous antenna systems to the pigments in the photosynthetic membranes is less than 100%. The same applies to the so-called FMO protein, so that the fluorescence spectra of green sulfur bacteria are dominated by the fluorescence bands of chlorosomes and FMO protein, whereas emission from the BChls in the core complex is hardly observable (Amesz and Vasmel, 1986; Otte et al., 1991). A quantitative determination, however, of the transfer efficiencies from such measurements requires knowledge of the “intrinsic” fluorescence yield of these antenna components, i.e. the yield in the absence of energy transfer to other pigments. Such knowledge is normally not available, and therefore one has to rely on fluorescence excitation spectra to obtain such information. Any phenomenon brought about by light can in principle be characterized by its excitation (action) spectrum. Such a spectrum defines the relative efficiencies of absorbed or incident photons of various wavelengths to bring about the phenomenon under study. Unfortunately, published action spectra are often poorly defined. A properly measured action spectrum gives quantitative information about the pigment or pigments that sensitize the reaction, but care must be taken to avoid errors, such as those caused by a nonlinearity of the response with light intensity and self-screening within the sample (Amesz, 1973). Quantitative information on efficiencies of energy transfer is obtained by comparison of the excitation spectrum with the absorption spec-

Classical optical spectroscopy

trum. Unavoidable imperfections and differences in the optical arrangement used for measuring these two spectra, especially with scattering samples, set a limit to the accuracy of such a comparison, and this means that an excitation spectrum will not normally give reliable information in the range between, say, 90 and 100%

7

transfer efficiency. Fig. 2 shows fluorescene excitation spectra of chloroplasts, measured at low temperature. By choosing the proper emission wavelength, the excitation spectra of Photosystems I and II can be measured independently in such a preparation, and this allows a distinction between the chlorophylls associated with the two

8

photosystems (note e.g. the predominance of the Chl b bands in the Photosystem II spectrum). Corresponding absorption spectra of the two photosystems are not available, unless one resorts to fractions solubilized by detergents, but the general impression from these and similar experiments is that the transfer efficiency from shortwavelength to long wavelength Chls is close to 100% in both photosystems. This means that the excitation spectra provide us with a means to determine the in situ absorption spectra of the two photosystems, information that cannot be obtained in any other way. The anoxygenic photosynthetic bacteria have only one photosystem and do not pose such problems. Fig. 3 shows a fluorescence excitation spectrum of a purple bacterium illustrating the lower limit of accuracy that can be obtained in transfer efficiency measurements (Kleinherenbrink et al., 1992). The absorption spectrum shows the antenna and reaction center bands in chromatophores of the BChl b containing purple bacterium Rhodopseudomonas viridis. The reaction center bands are completely lacking in the excitation spectrum of antenna fluorescence, showing that the efficiency of energy transfer from the reaction center to the antenna does not exceed 2% (Otte et al., 1993). The widely held “trap–

Jan Amesz limited” model for energy conversion clearly does not apply in this case. Measurement of absolute yields of fluorescence in photosynthetic material is notoriously difficult, since it requires absolute measurements of light intensities and a representative sampling of all fluorescence emitted (Weber and Teale, 1957). In fact, one may doubt if accurate numbers have ever been published for photosynthetic material. Relative yields can be measured much more easily, and such measurements, as a function of time, of added ions and inhibitors and of light intensity, have been done extensively during the last decades. Starting with the work of Duysens and Sweers (1963), studies of the so-called variable fluorescence and induction effects have yielded a wealth of information on various aspects of the mechanism of photosynthesis. A discussion is beyond the scope of this chapter; the reader may be referred to reviews of van Gorkom (1986), Renger and Schreiber (1986), Krause and Weis (1991) and Dau (1994). Historical aspects have been reviewed by Duysens (1986) and Govindjee (1995). References Allen MB, French CS and Brown JS (1960) Native and extractable forms of chlorophyll in various algal groups. In: Allen MB (ed) Comparative Biochemistry ofPhotoreactive Systems, pp 33–51. Academic Press, New York. Amesz J (1973) Spectrophotometric methods in photobiology. In: Checcucci A and Weale RA (eds) Primary Molecular Events in Photobiology, pp 21–43. Elsevier, Amsterdam. Amesz J and Vasmel H (1986) Fluorescence properties of photosynthetic bacteria. In: Govindjee, Amesz J and Fork DC (eds) Light Emission by Plants and Bacteria, pp. 423– 450. Academic Press, Orlando, FL. Amesz J, Duysens LNM and Brandt DC (1961) Methods for measuring and correcting the absorption spectrum of scattering suspensions. J Theor Biol 1: 59–74. Butler WL and Hopkins DW (1970a) Higher derivative analysis of complex absorption spectra. Photochem Photobiol 12: 439–450. Butler WL and Hopkins DW (1970b) An analysis of fourth derivative spectra. Photochem Photobiol 12: 451–456. Chance B (1951) Rapid and sensitive spectrophotometry. III. A double beam apparatus. Rev Sci Instr 22: 634–638. Dau H (1994) Short-term adaptation of plants to changing light intensities and its relation to Photosystem II photochemistry and fluorescence emission. J. Photochem Photobiol B: Biol 26: 3–27. DeVault D and Chance B (1966) Studies of photosynthesis using a pulsed laser. I. Temperature dependence of cyto-

Classical optical spectroscopy chrome oxidation rate in Chromatium. Evidence for tunneling. Biophys J 6: 825–847. Duysens LNM (1952) Transfer of Excitation Energy in Photosynthesis. Doctoral Thesis, University of Utrecht. Duysens LNM (1956) The flattening of the absorption spectrum of suspensions, as compared to that of solutions. Biochim. Biophys. Acta 19: 1–12. Duysens LNM (1957) Methods for measurement and analysis of changes in light absorption occurring upon illumination of photosynthesizing organisms. In: Gaffron H (ed) Research in Photosynthesis, pp 59–61. Interscience Publishers, New York. Duysens LNM (1986) Introduction to (bacterio)chlorophyll emission: a historical perspective. In: Govindjee, Amesz J and Fork DC (eds) Light emission by Plants and Bacteria, pp 3–28. Academic Press, Orlando. Duysens LNM and Sweers HE (1963) Mechanism of the two photochemical reactions in algae as studied by means of fluorescence. In: Studies on Microalgae and Photosynthetic Bacteria, pp 353–372. Univ of Tokyo Press, Tokyo. Duysens LNM, Amesz J and Kamp BM (1961) Two photochemical systems in photosynthesis. Nature 190: 510–511. Engelmann TW (1882) Über Sauerstoffausscheidung von Pflanzencellen im Mikrospectrum. Botan Z 40: 419–426. Fork DC and Mohanty P (1986) Fluorescence and other characteristics of blue-green algae (cyanobacteria), red algae, and cryptomonads. In: Govindjee, Amesz J and Fork DC (eds) Light Emission by Plants and Bacteria, pp 451–496. Academic Press, Orlando. French CS (1955) Fluorescence spectrometry of photosynthetic pigments. In: Johnson FH (ed) The Luminescence of Biological Systems, pp 51–74. Am Ass for the Advancement of Science, Washington DC. French CS and Harper GE (1957) Derivative spectrophotometry. Carnegie Institution of Washington Year Book 56: 281–283. French CS and Young VK (1952) The fluorescence spectra of red algae and the transfer of energy from phycoerythrin to phycocyanin and chlorophyll. J Gen Physiol 35: 873–890. French CS, Towner H, Bellis DR, Cook RM, Fair WR and Holt WW (1954) A curve analyser and general purpose graphical computer. Rev Sci Instr 25: 765–775. Govindjee (1995) Sixty-three years since Kautsky: chlorophyll a fluorescence. Aust J Plant Physiol 22: 131–160. Kautsky H and Franck U (1943) Chlorophyll Fluoreszenz und Kohlensäure Assimilation XII. Zusammenfassung der bisherigen Ergebnisse und ihre Bedeutung für die Kohlensäureassimilation. Biochem Z 315: 207–232. Kautsky H and Hirsch A (1931) Chlorophyll Fluoreszenz und Kohlensäure Assimilation. Biochem Z 274: 423–434. Ke B, Treharne RW and McKibben C (1964) Flashing light spectrophotometer for studying the fast reactions during photosynthesis. Rev Sci Instr 35: 296–300. Kleinherenbrink FAM (1992) Trapping Efficiencies and Electron Transfer in Photosynthetic Bacteria. Doctoral Thesis, University of Leiden. Kleinherenbrink FAM, Deinum G, Otte SCM, Hoff AJ and Amesz J (1991) Energy transfer from long-wavelength absorbing antenna bacteriochlorophylls to the reaction center. Biochim Biophys Acta 1099: 175–181.

9 Klughammer C, Kolbowski J and Schreiber U (1990) LED array spectrophotometry for time resolved difference spectra in the 530–600 nm wavelength region. Photosynth Res 25: 317–327. Kok B (1959) Light-induced absorption changes in photosynthetic organisms. II: A split-beam difference spectrophotometer. Plant Physiol 34: 184–192. Kramer HJM, Amesz J and Rijgersberg CP (1981) Excitation spectra of chlorophyll fluorescence in spinach and barley chloroplasts at 4 K. Biochim Biophys Acta 637: 272–277. Kramer HJM, Westerhuis WHJ and Amesz J (1985) Low temperature spectroscopy of intact algae. Physiol Végét 23: 535–543. Krause GH and Weis E (1991) Chlorophyll fluorescence and photosynthesis – the basics. Ann Rev Plant Phys Plant Mol Biol 42: 313–349. Latimer P (1959) Influence of selective light scattering on measurement of absorption spectra of Chlorella. Plant Physiol 34: 193–199. Latimer P and Eubanks CAH (1962) Absorption spectrophotometry of turbid suspensions: a method for correcting for large systematic distortions. Arch Biochem Biophys 98: 274–285. Lavorel J, Breton J and Lutz M (1986) Methodological principles of measurement of light emitted by photosynthetic systems. In: Govindjee, Amesz J and Fork DC (eds) Light Emission by Plants and Bacteria, pp 57–98. Academic Press, Orlando. Martin AE (1959) Multiple differentiation as a means of band sharpening. Spectrochim Acta 14: 97–103. Netzel TL, Rentzepis P and Leigh J (1973) Picosecond kinetics of reaction centers containing bacteriochlorophyll. Science 182: 238–241. Nishimura M, Legallais V and Mayer D (1969) Multipurpose phosphoroscopic instrument for the study of phosphorescence, induction of fluorescence and absorbance change of turbid biological materials. Rev Sci Instr 40: 271–273. Otte SCM, van der Heiden JC, Pfennig N and Amesz J (1991) A comparative study of the optical characteristics of intact cells of photosynthetic green sulfur bacteria containing bacteriochlorophyll c, d or e. Photosynth Res. 28: 77–87. Otte SCM, Kleinherenbrink FAM and Amesz J (1993) Energy transfer between the reaction center and the antenna in purple bacteria. Biochim Biophys Acta 1143: 84–90. Öquist G and Wass R (1988) A portable, microprocessor operated instrument for measuring fluorescence kinetics in stress physiology. Physiol Plantarum 73: 211–217. Porter G (1968) Flash photolysis and some of its applications. Science 160: 1299–1307. Rabinowitch EI (1951) Photosynthesis and Related Processes, Vol II, part 1, Spectroscopy and Fluorescence of Photosynthetic Pigments; Kinetics of Photosynthesis. Interscience Publ, New York. Renger G and Schreiber U (1986) Practical applications of fluorometric methods to algae and higher plant research. In: Govindjee, Amesz J and Fork DC (eds) Light Emission by Plants and Bacteria, pp 587–619. Academic Press, Orlando. Schreiber U (1986) Detection of rapid induction kinetics with

10 a new type of high-frequency modulated chlorophyll fluorometer. Photosynth Res. 9: 261–272. Schreiber U, Neubauer C and Schlina U (1993) PAM fluorometer based on medium- frequency pulsed Xe- flash measuring light: A highly sensitive new tool in basic and applied photosynthesis research. Photosynth Res. 36: 65–72. Shibata K (1958) Spectrophotometry of intact biological materials. Absolute and relative measurements of their transmission, reflection and absorption spectra. J Biochem (Tokyo) 45: 599–623. Spruit CJP (1971) Sensitive quasi-continuous measurement of photoinduced transmission changes. Meded Landbouwhogeschool Wageningen 71: 1–6. Van Bochove AC, Swarthoff T, Kingma H, van Grondelle R, Duysens LNM and Amesz J (1984) A study of the primary charge separation in green bacteria by means of flash spectroscopy. Biochim Biophys Acta 764: 343–346.

Jan Amesz Van Gorkom H (1986) Fluorescence measurements in the study of Photosystem II electron transport. In: Govindjee, Amesz J and Fork DC (eds) Light Emission by Plants and Bacteria, pp 267–289. Academic Press, Orlando. Vermeulen D, Wassink EC and Reman GH (1937) On the fluorescence of photosynthesizing cells. Enzymologia 4: 254–268. Weber G and Teale FWJ (1957) Determination of the absolute quantum yield of fluorescent solutions. Trans Faraday Soc 53: 646–655. Witt HT, Moraw R and Müller A (1959) Blitzlichtphotometrie. Z Physik Chem NF 20: 193–205. Wolff C, Buchwald H-E, Rüppel H, Witt K and Witt HT (1969) Rise time of the light induced electrical field across the function membrane of photosynthesis. Z Naturforsch 24b: 1038–1041.

Chapter 2 Linear and Circular Dichroism Garab Institute of Plant Biology, Biological Research Center, Hungarian Academy of Sciences, Szeged, P.O. Box 521, H-6701, Hungary

Summary I. Introduction A. Polarized Light B. Absorbance of Light by Molecules II. Linear Dichroism A. Polarization of the Electronic Transitions of Chls B. Anisotropy in Protein Complexes and Membranes 1. LD of Absorbance 2. Photoacoustic Linear Dichroism 3. Polarized Fluorescence Emission C. Methods of Orientation of Membranes and Particles 1. Mechanical Orientation Techniques 2. Orientation in a Magnetic Field 3. Orientation in an Electric Field 4. Photoselection D. Determination of the Orientation Angle of Dipoles in Realistic Systems 1. Degree of Orientation, Distribution Functions 2. Membrane Curvature, Microscopic LD 3. Fluctuations E. Miscellaneous Applications of LD 1. Estimation of Shape and Size of Particles 2. Spatial Position of Organelles 3. Information on the Band Structure III. Circular Dichroism A. Physical Origins of CD Signals B. CD of Photosynthetic Pigments 1. Intrinsic CD of Isolated Molecules 2. Excitonic Interactions 3. Differential Scattering, Psi-type CD, and Differential Polarization Imaging C. Secondary Structure of Chl-Containing Proteins D. Artifacts IV. Concluding Remarks Acknowledgements References

11 12 13 15 16 16 17 17 18 18 19 19 20 20 20 21 21 22 23 23 24 24 24 24 25 26 27 27 30 33 33 35 35 35

Summary The efficiency of photosynthetic light energy conversion depends largely on the molecular architecture of the photosynthetic membranes. Linear and cicular dichroism (LD and CD) techniques have contributed Correspondence: Fax: 36-62-433434; E-mail: [email protected]

11 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 11–40. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

12

Garab

significantly to our knowledge of the molecular organization of the pigment system in various complexes and membranes. Systematic LD studies have led to the recognition of an apparently universal property of pigment systems in vivo: all pigments in all photosynthetic organisms display non-random orientation with respect to each other, to the protein axes and to the membrane plane. This molecular organization plays an important role in the energy transfer between pigment molecules. CD spectroscopy is widely used for the detection of excitonic interactions, which have been found to occur in virtually all reaction center and antenna complexes. Excitonic CD carries information on the distances and orientation of the interacting pigment molecules. CD is also capable of revealing information about certain macroorganizational parameters in molecular aggregates with sizes commensurate with the wavelength of visible light. These non-invasive techniques can be used for systems in a wide range of structural complexity, from isolated pigment molecules to whole organelles. CD and LD techniques have been extended to the (sub)picosecond time range. Combined with the methods of quantitative evaluation of data, these techniques will certainly remain indispensable in elucidation of the structure and function of the photophysical and photochemical apparatus. The purpose of this chapter is to provide an introduction to the theory and practice of LD and CD methods in photosynthesis. The main emphasis will be placed on the underlying principles and the basics of the experimental procedures, complemented with a few illustrations of results. I would also like to draw attention to a few recently introduced polarization techniques which are ripe for application in photosynthesis. Abbreviations: BChl – bacteriochlorophyll; Chl – chlorophyll; CB – circular birefringence; CD – circular dichroism; CDS – circular differential scattering; CIDS – circular intensity differential scattering; CPL – circularly polarized luminescence; DPI – differential polarization imaging; DR – dichroic ratio; FDCD – fluorescence-detected circular dichroism; FMO – Fenna–Matthews–Olson [complex]; FP – fluorescence polarization ratio; LB – linear birefringence; LD – linear dichroism; LHCII – light-harvesting chlorophyll-a/b-protein complex of photosystem 2; MCD – magnetic circular dichroism; ORD – optical rotatory dispersion; PALD – photoacoustic LD; Pheo – pheophytin; PSI or II – photosystem I or II; PChl – protochlorophyll; psi – polymer and salt-induced

I. Introduction In investigations of the primary processes of photosynthesis, the ultimate goal is to understand the structure and function of the photophysical and photochemical machinery. The efficiency of the primary step in the conversion of light energy to chemical energy depends largely on the molecular architecture of the reaction centers and the antenna system. The high efficiency of the primary charge separation and stabilization is in large part due to the special organization of the reaction centers. The details of the operation of the reaction centers, however, are still not fully understood. Energy migration in the antenna is largely determined by the molecular architecture of the pigment system. An optimized antenna organization should minimize quantum losses and ensure an efficient energy supply to the reaction

centers under a wide range of environmental and physiological conditions. On the other hand, a controlled dissipation of the absorbed energy in the antenna can prevent photoinhibitory damage of the photosynthetic machinery and thereby play a protective role. This requires a highly organized molecular architecture, which should nevertheless be capable of structural reorganizations. For an understanding of the structure and function of the building blocks and also the mechanism of operation of the entire photosynthetic apparatus, non-invasive techniques, such as LD and CD, are of special value. For a complete understanding of the function, the structural and optical information must be combined. Full structural information can be provided only by atomic resolution crystallography. The information content of polarization spectroscopy is substantially less than that of crystallography. However, many

LD and CD

Chl-containing complexes or other constituents of the photosynthetic apparatus appear to resist crystallization. Furthermore, in complex systems, e.g. membranes or organelles, a number of structural parameters can be determined only by means of LD and CD investigations. These methods are also indispensable when structural reorganizations and ultrafast processes are to be monitored. Full structural information is available on the bacterial reaction center of Rhodopseudomonas viridis (Deisenhofer et al., 1984). As pointed out by Breton and Nabedryk (1987), conclusions from polarization spectroscopy are in good qualitative accord with the results of X-ray crystallography. A recent systematic study on bovine gamma-crystallines showed the excellent agreement of LD and X-ray data (Bloemendal et al., 1990). The first part of this chapter will survey the basic principles related to polarized light and the interaction of light with absorbing matter. Then, the methods of LD and CD will be overviewed, together with a few examples of their application; these merely serve for illustration and cannot substitute a systematic review. Throughout the chapter, the emphasis will be placed on the concepts, and the mathematical formalism will be used sparingly. LD and CD spectroscopy have developed significantly in the past decade. LD spectroscopy is now used routinely to monitor ultrafast processes in the reaction center complex (Kirmaier et al., 1985) and energy migration pathways in the antenna (reviewed by van Amerongen and Struve, 1995). The method of recording transient CD spectra at picosecond time resolution has also been elaborated (Xie and Simon, 1991). The “standard” techniques have recently been combined with novel methods, such as photoacoustic LD et al., 1985), differential polarization microscopy (Finzi et al., 1989) and circular differential polarization scattering (Garab et al., 1988c). Some other techniques, such as vibrational CD, CPL, FDCD have as yet been little used in photosynthesis, but hold a promise for future applications. The pioneering work and basic conclusions on the orientation of photosynthetic pigments in vivo were reviewed by Breton and Verméglio (1982),

13

and later developments were summarized by Breton (1986) and Garab et al. (1987). LD methods have been in the focus of many textbooks and reviews (Hofrichter and Eaton, 1976; Johansson and Lindblom, 1980; Clayton, 1980; Nordén et al., 1992; Bloemendal and van Grondelle, 1993; van Amerongen and Struve, 1995). CD and MCD in different Chl-containing systems were first reviewed by Sauer (1972). The theory of excitonic CD and results on Chl-containing complexes were dealt with in depth by Pearlstein (1982, 1987, 1991).

A. Polarized Light Light is an electromagnetic wave which oscillates periodically in both time and space. In the wave, the electric and magnetic vectors, which are proportional to each other in magnitude, are mutually perpendicular, and also perpendicular to the direction of propagation. Non-polarized light consists of vibrations in many different polarization directions. In linearly polarized light, (often called plane polarized light), the electric vector, E (“the light vector”), oscillates sinusoidally in a direction (plane) which in spectroscopy is conventionally called the polarization direction (plane). In circularly polarized light, the magnitude of E remains constant, but it traces out a helix as a function of time. In accordance with the convention used in CD spectroscopy, in right and left circularly polarized light beams, when viewed by an observer looking toward the light source, the end-point of E would appear to rotate clockwise and counterclockwise, respectively. It is useful to apply the principle of superpositions to conceptualize the state of polarization of a light beam. As shown in Fig. 1A and B, circularly polarized light can be represented as the sum of two orthogonal linearly polarized beams in which the amplitudes are equal and the phases are shifted by exactly + or With a phaseretardation of another linearly polarized beam is obtained, the polarization of which is orthogonal to the original polarization direction (Fig. 1C). With retardation angles different from or or if the amplitudes of the two orthognal linearly polarized components are not equal, the polarization will become elliptical, i.e. the gen-

14

eral form. (For a more elaborate treatment of polarized light and the basic principles of the classical theory of optics, the reader is referred to textbooks, e.g. by Born and Wolf, 1980.) Through the use of the principle of superpositions, circularly or linearly polarized light can easily be constructed. In the following the modulation method used in most dichrographs will be outlined. Let us consider a linearly polarized light produced, for example, by a birefringent crystal. Let us transmit this beam through a slab of an isotropic transparent material (e.g. fused silica) with edges oriented at 45° with respect to E. If the slab is pressed and drawn in one direction by applying a.c. voltage (V) on a piezoelectric transducer (M, modulator), LB is induced, which results in phase shifts of one of the linearly polarized component beams; the phase shift is linearly

Garab

proportional to the applied voltage (Fig. 2). For samples which show anisotropy of absorbance for orthogonal circularly or linearly polarized beams, the light intensity (I) transmitted by the sample (S) varies periodically with f or 2f, respectively. I and can be measured with a photomultiplier (PMT) and an appropriate demodulation technique (DEM), and thus CD and LD signals can be recorded. (For typical block diagrams and some technical aspects, see Bloemendal and van Grondelle, 1993; Johnson, 1985.) As will be evident in section III, it is useful to envisage the linearly polarized light as the sum of right and left circularly polarized light beams of equal intensity. In Figs. 1A and B the sum of the horizontal components is zero and thus the sum of the two orthogonally circularly polarized beams indeed reduces to a linearly (vertically) polarized beam. A general light beam can also be conveniently represented by the Stokes’ parameters, a 4 × 1 column matrix, and the light-matter interaction can be described by the 4 × 4 Mueller matrix:

LD and CD

The Stokes parameters (I,Q,U,V) characterize the monochromatic light beam propagating along z axis: its intensity, degree of linear polarization in xz/yz, at ±45° and circular polarization, respectively. In most samples, there are correlations between different elements of the matrix. Generally, however, all 16 elements can be independent and yield useful structural information on the sample (Kim et al., 1987a). Although there is an impetus to measure more elements of the matrix (Tinoco et al., 1987), in most cases only the absorption LD and CD are determined, mainly because of technical difficulties and the poor understanding of the physical meaning of other elements.

B. Absorbance of Light by Molecules During an optically induced transition, the electron distribution of the molecule oscillates periodically with the frequency of the absorbed light. This means a transient oscillation of the electric and magnetic moments, which can generally be regarded as transition dipole moments, and m, respectively. However, in the UV to IR spectral range, only the electric dipole transitions have significant intensities and thus the absorption can be satisfactorily described by the electric transition dipole moment, which for an optically induced electronic transition between the ground and excited states, a and b, is defined by the vector integral: Here and are the wave functions of the corresponding states of the molecule, and the electric dipole operator contains the sum of the products of each charged particle (electron or nucleus), and their position vector, In classical terms, the interaction is described by the induction of an oscillating dipole by the oscillating

15

electric field vector of the light. (In order to simplify the physical interpretation, the classical and quantum mechanical pictures will be used interchangeably.) In accordance with the Born–Oppenheimer approximation, the wave functions are written as the product of the electronic (e) and nuclear (n) wave functions:

In the first approximation, the electronic transitions are considered for fixed nuclear positions, i.e. for no vibronic coupling of the transition. The admixture of vibronic components may complicate the interpretation of the polarization measurements. The probability of absorbance is proportional to the square of the scalar product of the electric vector of the light and the transition dipole vector of the molecule:

This means that a light beam polarized parallel to the transition dipole vector has the maximum probability of absorption, whereas if it is polarized perpendicular to no absorption can take place. This serves the basis for LD spectroscopy. Let us consider an ‘oriented gas’, a set of noninteracting molecules in which all molecules are aligned parallel to each other. For a concentration (c) of 1 M and a pathlength (l) of 1 cm, let the absorbance with polarization parallel to the direction of the molecular dipoles be For the other two orthogonal linear polarizations the absorbance is evidently zero. On the other hand, in a random gas, after averaging, we obtain with any direction of the polarization of the light. is the isotropic molar extinction coefficient; The length of the transition moment vector (in the dipole strength, is correlated with the molar extinction coefficient:

where is the frequency of the light, i.e. the dipole strength is determined by the integrated area of the absorbance band.

16

Garab

II. Linear Dichroism Linear dichroism is the differential absorption of two orthogonal linearly polarized light beams and in a macroscopically oriented sample: There are two basic cases for LD investigations: (i) If the spatial position of the molecules is known in the laboratory coordinate system (i.e. the molecules are macroscopically aligned), the orientation of the molecular dipoles can be determined with respect to the molecular coordinates, (ii) When the orientation of a transition dipole is known with respect to the molecular coordinate system, LD carries information on the orientation of the molecule, or at least on that of the transition dipole with respect to the symmetry axis of the sample. In most applications, the orientation of dipoles with respect to the coordination system of the object is at the focus of interest. However, the depth of information provided by these studies depends on the knowledge of the nature of the electronic transitions of the molecules and on the reliability of the data concerning the orientation of the transition dipoles with respect to the molecular coordinate system. It should be noted that, for symmetry reasons, not all molecules can exhibit LD (Hofrichter and Eaton, 1976) but Chls, carotenoids, cytochromes and phycobilins possess linearly polarized electronic transitions, and thus are readily accessible for LD studies (Breton and Vermeglio, 1982; Juszcak et al., 1987). On the other hand, the method of LD is not confined to optical transitions. The anisotropy properties of magnetic transitions, which have proved very useful in the characterization of triplet transitions, can be studied with polarized microwaves (Hoff, 1990). X-ray LD acquires information about the orientation of specific chemical bonds (Ade and Hsiao, 1993).

A. Polarization of the Electronic Transitions of Chls The interpretation of the electronic spectra of Chls and the determination of the polarizations of the major electronic transitions with respect to the molecule-fixed coordinate system are far from

definitive. Our knowledge is based mainly on the results of LD studies of molecular Chl solutions oriented in different systems, e.g. stretched film (Breton et al., 1972) multilayers (Hoff, 1974), host crystal (Moog et al., 1984) and liquid crystals (Bauman and Wrobel, 1980; et al., 1987). Recently Langmuir Blodgett film was used to orient plastoquinone molecules (Kruk et al., 1993). The molecules in these systems must not interact with each other or with the host matrix, because interactions may lead to changes in the polarization directions. The orientation of different transition dipoles of Chls and Pheos are given with respect to the X-Y molecular framework of the tetrapyrrole plane, and is measured in degrees, clockwise from the X axis of the molecular frame. Conventionally, the X passes through C7, the position of phytyl substitution, and the Y axis through the N atoms of pyrroles I and III in Fig. 3). (The axes, are selected on the basis of the electronic symmetry of the system.) The main electronic transitions of Chls are labeled with indices X and Y, and and and for the red and the Soret bands, respectively, to indicate that they are polarized along these axes. However, Bauman and Wrobel (1980)

LD and CD

17

showed that the red band of Chl a and BChl a have mixed X and Y character. Fragata et al. (1988) calculated the polarization of different UV–VIS transitions of Chl a and Pheo a and showed that the main transition, (0 – 0) of Chl a which absorbs at 670 nm is found at 70° (Fig. 3). Van Gurp et al. (1989) also determined about 20° for the deviation of of Chl a from the Yaxis but preferred to take the transition moment on the other side of the symmetry axis, i.e. at 109°, thereby closer to Furthermore, it was concluded that other transitions of Chl a cannot be characterized in simple terms of a transition moment in the molecular frame but must be described in terms of averages of goniometric functions.

B. Anisotropy in Protein Complexes and Membranes With proper selection of the orientation method (see below) it can be ensured that the orthogonal polarizations of the measuring beam, hereafter referred to as and coincide with the preferential alignment of the sample, e.g. the plane of membranes or the long axis of complexes or aggregates. This simplifies interpretation of the data appreciably. For the general case, the Euler transformation must be applied. 1. LD of Absorbance (i) Let us consider oriented planar membranes which contain absorbing dipoles with well-defined orientation angle with respect to the membrane plane (Fig. 4A). The orientation of and its unit vector, can be characterized with the azimuthal and orientational angles, and and

where u, v and n are also unit vectors. Since inside the membrane cannot be fixed with respect to the coordinate system, averaging must be performed:

(ii) For aligned protein complexes (Fig. 4B):

Hence,

and

18

The quantities which are directly related to the orientation angle are the parameter S, the reduced dichroism and the dichroic ratio (DR):

If in membranes, S > 0 (DR > 1) it can be concluded that the dipoles tend to lie in the membrane plane, with with respect to the normal to the plane is called the magic angle, an orientation which cannot be distinguished from random orientation in an LD experiment. The dichroism of the sample is often characterized by the quantity which in some papers is referred to as dichroic ratio and in others called reduced LD. However, it must be noted that, although this expression is well suited for the general case when the symmetry of the system is not known, its use may be misleading for systems with uniaxial symmetry where 2. Photoacoustic Linear Dichroism PALD is the differential dissipation of irradiation with orthogonal linear polarizations of light. This method provides a special tool to measure the orientation dependence of the radiationless deexcitation processes. (Contributions from the anisotropic character of heat conductivity have to be determined separately.) PALD has been used for both model systems and native particles et al.,1985; et al.,1984; et al., 1990). For a recent review, see et al. (1991). 3. Polarized Fluorescence Emission In a dipolar approximation, the fluorescence emission is polarized because, for the same electronic transition, the orientation of the emitting dipole is parallel to the absorbing dipole. The intensity of the emission in a given polarization direction is proportional to the squares of the scalar products. In a coordinate system as in Fig.

Garab In and a macroscopically oriented sample, the dichroic ratio of fluorescence, which is often called the fluorescence polarization ratio (FP), can be used to calculate the orientation angle:

For this correlation to be valid, FP must be excited with non-polarized light, and the system must involve perfect energy transfer. For the general case, FP depends not only on the fluorescence, but also on the absorbance of the sample, i.e. photoselection may play an important role (Szitó et al., 1985). A general theoretical analysis has been presented by van Gurp et al.(1988). For steady-state fluorescence, with excitation in the blue, this effect is usually small, as can be demonstrated by comparing FP spectra recorded with nonpolarized and linearly polarized excitations, respectively. The magnitude of polarization of the fluorescence emission of intrinsically anisotropic and oriented sample upon excitation with non-polarized light is sometimes evaluated in the form of This representation, however, may cause confusion. Conventionally, stands for the degree of polarization of fluorescence after polarized excitation of a randomly oriented sample, the indices referring to the polarization directions of the observation with respect to the direction of polarization of the exciting beam (see Chapter 3). In complex systems, e.g. photosynthetic membranes, it is technically difficult to separate FP from P. As recognized by Breton et al. (1973), the intrinsic anisotropy of dipoles, which in oriented samples gives rise to can contribute significantly to the measured value of P even if macroscopically FP = 1. This “residual polarization” is due to the fact that differently oriented membranes contained in a randomly oriented suspension exhibit different P values. In order to minimize this effect, the membranes should be oriented with their planes facing the observation. (For further comment, see II.D.2). Polarized fluorescence emission measurements have recently been used to determine the degree

LD and CD

19

of orientation of LHCII in compressed gel, and aided the precise determination of the orientation angles of different absorbance transition dipoles (van Amerongen et al., 1994).

C. Methods of Orientation of Membranes and Particles Orientation of a sample can be achieved with various techniques. Which method is applied depends on the experimental conditions and on the goal of the investigation, and a general recipe cannot be given. 1. Mechanical Orientation Techniques In one of the simplest cases, the gravitational field is used to align the particles during drying. With flat membranes, a high degree of orientation can be achieved. However, since membranes tend to lie face-down on a supporting quartz or glass plate, LD can be measured only at a tilt angle with respect to the plate (Breton et al., 1973). This may easily introduce artifacts due to reflections. This method is useful when the water content of the sample must be low, e.g. in polarized IR spectroscopic studies (Nabedryk et al., 1984). The spreading of oblate particles with a fine brush over a quartz plate also yields a dry sample. The particles are aligned reasonably well, with their long axis parallel to the direction of spreading (Breton et al., 1973). Some artifacts may be present (Haworth et al., 1982), which are probably due to the uneven surface of the film. The method of stretching films (e.g. polyvinyl alcohol) containing the particles can be used in a wide range of sizes, from pigments (Breton et al., 1972) to whole chloroplasts et al., 1985). The main advantage of this method is that the degree of orientation can readily be evaluated as a function of the extent of stretching (Nordén, 1980). Orientation by flow, i.e. orientation in a hydrodynamic gradient, is used mainly for long, cylindrical molecules or particles. The shear gradient is usually formed in the annular gap between a rotating and a fixed cylinder. In flow orientation, the particles can be suspended at low concentration, and there is essentially no restriction as concerns the reaction medium. (For a review, includ-

ing technical and theoretical questions, see Nordén et al., 1992.) The currently most versatile and probably most widely used method is the gel-squeezing technique. This method was invented by Abdourakhmanov et al. (1979) and was described for polyacrylamide gel, which is a continuum gel, i.e. it permits the alignment of particles of different sizes and shapes. Orientation by gel squeezing combines the advantages of film stretching and flow orientation: the degree of orientation can be determined precisely, while the aqueous environment of the sample is preserved. As illustrated in Fig. 5A, flat membranes (or disc-like particles) tend to align themselves with their plane perpendicular to the unidirectional compression. Rod-like particles also tend to align in the plane perpendicular to the compression, but otherwise they remain randomly oriented. Thus, for rod-like particles it is useful to apply a second direction of squeezing (Fig. 5B). With gel-compression, deformable membrane spheres (e.g. vesicles, chromatophores and chlo-

Garab

20

roplast ‘blebs’) can be deformed into ellipsoids and the intrinsic dichroism of the sample can thereby be made apparent and quantitatively evaluated (Kiss et al., 1985; Abdourakhmanov and Erokhin, 1980). Polyacrylamide gel is transparent in the entire visible and near-IR spectral regions and with the admixture of glycerol can be used for low temperature measurements. Increasing the concentration of acrylamide and increasing the ratio of bisacrylamide to acrylamide renders the gel more rigid, with a smaller mesh size. Rigid and soft gels are suitable for pigment-protein complexes and membranes, respectively (A.O. Ganago, personal communication). For many applications, the components of the gel do not perturb the functions of the embedded particles (Ganago et al., 1982; Vermeglio et al., 1990; Breton and Ikegami, 1989). However, this was not the case for the oxidation state of P700 (Breton and Ikegami, 1989), and for the electric properties of chloroplast thylakoid membranes (Osváth et al., 1994). Acrylamide also diminished the big CD of chloroplasts. The fact that polyacrylamide significantly reduces the intensity of light scattering (Haworth et al., 1982) may be indicative of a partial disintegration of the sample due to incorporation of the gel into the membrane. This can probably account for observations that in heliobacterial membranes BChl g was bleached (van Dorssen et al., 1985) and artifacts appeared in complexes of fucoxanthin-containing algae (Hiller and Breton, 1992). 2. Orientation in a Magnetic Field The magnetically induced orientation of a photosynthetic system was first reported in Chlorella by Geacintov et al. (1972). A major advantage of this method is that practically no restriction applies as concerns the composition of the medium, and the degree of alignment can be nearly 100%. However, as the method is based on the diamagnetic susceptibility anisotropy of the sample, which is a collective property of the particle, both the shape and size of the particle may limit orientability in normally available fields of 1–2 T. (For a detailed analysis of the mechanism of orientation see Knox and Davidovich, 1978; Papp and Meszéna, 1982.) In practice, this method is lim-

ited to granal chloroplasts and large aggregates of LHCII (Kiss et al., 1986), which are aligned with their membrane planes and sheets perpendicular to the field vector (Geacintov et al., 1972; Breton et al., 1973; Garab et al., 1981; Kiss et al., 1986). Orientation in a magnetic field can be trapped at low temperature (Vermeglio et al., 1976) or in a gel (Finzi et al., 1989). 3. Orientation in an Electric Field An electric dipole placed in a unidirectional electric field orientates so as to minimize the total energy of the system (Charney, 1988). The measurements are usually restricted to low ionic strength. The electrophoretic movement of particles can be prevented by using alternating voltage or pulses. It has been shown that the frequency and the magnitude of the voltage may influence not only the degree, but also the mechanism of orientation through permanent and induced dipole moments (Gagliano et al., 1977; Charney, 1988). Chloroplasts can be oriented in an a.c. (50 Hz) field of about while small particles, e.g. chromatophores or isolated complexes, can be easily oriented by millisecond electric pulses of A quasisteady-state alignment of particles can be trapped in a gel, and thus the orientation can be studied in the absence of an external electric field (Dér et al., 1986). 4. Photoselection The technique of photoselection is based on the selective excitation of molecules by linearly polarized light, which induces anisotropy in the sample. This occurs alike in intrinsically isotropic samples (e.g. solutions of molecules) and in samples containing randomly oriented intrinsically anisotropic particles (e.g. a membrane suspension). The anisotropy function, r, for absorbance difference is defined as

where the indices refer to the orientation of the polarization direction of the probe light with respect to that of the excitation and the iso-

LD and CD

tropic absorbance change. The angle between the sensitizing molecular absorbance dipole and the detected absorbance or emission dipole can be calculated by measuring the absorbance or emission of the detected chromophore with polarization directions parallel and perpendicular to the actinic polarization (Breton and Vermeglio, 1982). This technique has become increasingly important in determination of the mutual orientation of the dipoles in the reaction center preparations containing small number of Chl molecules (Kwa et al., 1994). The method of photoselection is also ideally suited to the monitoring of energy transfer processes (van Amerongen and Struve, 1995).

D. Determination of the Orientation Angle of Dipoles in Realistic Systems In idealized systems, which were considered in the examples above, it is assumed that the degree of orientation is 100%, that the membranes are flat, that the rod- or disc-like particles or membranes cannot be deformed, that there is no fluctuation in the orientation angles and that both LD and absorbance can be measured with high precision. With the exception of the last conditions, in realistic systems these assumptions do not hold and the conclusions from idealized systems remain qualitative. (Although LD and absorbance can be measured with high precision, when they are measured in separate instruments, minor wavelength shifts may introduce quite large distortions in S. Smaller but significant distortions may be caused if LD exhibits too large amplitudes. Most dichrographs measure and use the approximation of which is not true if 1. Degree of Orientation, Distribution Functions In some systems, a large proportion of the particles are found to be perfectly aligned whereas the complementary population remains at random, thus the degree of orientation can easily be defined. For instance in an external magnetic field of about 1 T, the orientation is saturated for whole chloroplasts, whereas chloroplasts fragments can not be oriented even at much higher

21

field strengths (Garab et al., 1981). In other systems, e.g. with gel-squeezing or film-stretching techniques, the random suspension is gradually shifted toward the ‘perfect’ alignment, but saturation cannot be attained for finite values of the deformation parameters. These systems are characterized by distribution functions which characterize the alignment of the particles. The distribution functions for rod-shaped and disc-shaped particles can be calculated on the basis of the behavior of rigid particles in the squeezed gel. It is envisioned that rigid rods in an amorphous, uniform, continuum matrix, rotate in such a way that the ratio of the projections of their long axis changes identically to the ratio of the corresponding sample dimensions. A similar correlation is applied for the plane of discs (Ganago and Fock, 1981). For disc-shaped particles after unidirectional compression:

and for rod-shaped particles after two-directional compression:

In both cases:

where M (>1) is the compression parameter (or often called squeezing parameter) (see Fig. 5). It is easy to show that if and if and thus and vanish, and reduce to the idealized cases, respectively. (For the distribution function and calculations for the unfavorable cases, i.e. rod-shaped particles with unidirectional compression and disc-shaped particles with two-directional compression, see Ganago and Fock, 1981.) Fig. 6 shows and as a function of the orientation angle for the idealized and realistic cases. In our laboratory S values in chloroplasts with M = 2, typically increase from – 0.01 to about + 0.23 between 650 and 690 nm; FP

22

values are found between 1.1. and 1.7 (Szitó et al., 1984, 1985). It is assumed that no friction occurs between the particle and the gel, and there is no deformation of the shape. Since, however, distortions may occur it is advisable to carry out the experiments with different known values of the squeezing parameter, with cells of defined dimensions and extrapolate the value of the orientation angle to the non-squeezed case, and/or carry out the calculations with different presumptions (Kiss et al., 1985). In practice, distortions are small if (Ganago et al., 1983; Kiss et al., 1985). A properly prepared sample must be homogenous, and cracks, which may occur during too fast cooling or too large squeezing, for instance, should be avoided. Due to the compression, strains may be induced in the cell wall, which can introduce artifacts. For a homogenous sample without cracks and strains, the color pattern between two crossed polarizers is bright and uniform. Although the technique of gel-squeezing has been shown to yield reliable data on the orientation angles, it is difficult to prove that the basic assumptions are correct for the general case. Further complications may arise if the shape of the particle has mixed character. Thus, for a

Garab

quantitative analysis the best strategem is to apply independent orientation techniques (see e.g. van Amerongen et al., 1988). 2. Membrane Curvature, Microscopic LD Membrane curvature can be taken into account by means of simple geometrical models. For chloroplasts, such model calculations led to the conclusion that the long-wavelength emitting dipoles of Chl a lie essentially in the plane of the membranes. It was also shown that dipoles of Chl a span a much larger angular interval than previously thought (Garab et al., 1981). (For the estimated range of the orientation angle, see Fig. 6 and data above.) The importance of structural factors is evident in differential polarization images of chloroplasts (Finzi et al., 1989; Garab et al., 1991a). As can be seen in Fig. 7, the magnitude of the local LD depends strongly on the curvature of membranes, and the macroscopic LD evidently represents only an averaged value. LD microscopy likewise revealed that, due to the curvature of membranes, the local LD does not vanish even in face-aligned chloroplasts, despite the fact that LD = 0 in a suspension and in the plane of the image (Finzi et al., 1989). The

LD and CD

fact that residual polarization due to intrinsic anisotropy (Breton et al., 1973) may contribute significantly to the degree of polarization of fluorescence (see II.B.3) points to the need for microscopic fluorescence polarization investigations, possibly in combination with fluorescence lifetime imaging (Gadella et al., 1993). Such techniques would probably permit estimation of the magnitude of local order in native membranes and/or in lamellar aggregates of purified complexes, provided the local order is longranged enough compared to the resolution. 3. Fluctuations The orientation of a transition dipole may fluctuate in a certain angular interval. The fluctuation of the pigment dipoles can originate from (i) the pigment-protein complexes and/or (ii) the fluctuation of the protein axes in the membrane. In both cases the fluctuation can be either dynamic or statistical, i.e. it can originate from a rocking type of motion or is due to statistical disorder. If the long axis of the protein is embedded in the membrane at an angle with respect to the membrane normal, and there is a fluctuation of the orientation of the protein axis in the interval between and but fluctuation of the orientation angle of the dipole with respect to the protein axis is not permitted, we obtain:

23

where,

denotes averaging. This shows that spectral variations of S for the same set of dipoles can be equally ascribed to the variation in the orientation of the protein axis with respect to the membrane plane and the orientation of the dipole with respect to the protein axis. It may be speculated that major structural rearrangements in the antenna are accompanied by reorientation of some complexes. Such changes may be responsible for the observed LD changes due to state transitions in cyanobacteria (Homer-Dixon et al., 1994). Data obtained on algal mutants and on chloroplasts treated with linolenic acid suggested the importance of fluctuations due to the increased fluidity of membranes (Szitó et al., 1984, 1985). On the other hand, large fluctuations were not encountered in untreated wild-type chloroplasts.

E. Miscellaneous Applications of LD Besides the two basic uses of LD spectroscopy outlined above (II.A and B), LD measurements can yield further structural information on the system.

Garab

24

1. Estimation of Shape and Size of Particles

3. Information on the Band Structure

Uni- and bidirectional compressions of a gel result in a better alignment for disc- and rod-shaped particles, respectively. Thus, if the shape of the (sufficiently rigid) particle is unknown, measurement of the LD as a function of the squeezing parameters can lead to discrimination between disc-shaped and rod-shaped particles. This is called the reverse problem of LD (Ganago and Fock, 1981). From an analysis of the relaxation kinetics after electric or magnetic orientation, the size of the particle can be estimated (Geacintov et al., 1972; Kiss et al., 1986). In an electric field, the kinetics of the rise of the LD signal also carries information on the mechanism of the alignment (Charney, 1988), and therefore on the electric properties of the particle (e.g. contribution and orientation of the permanent dipole vector). Van Haeringen et al. (1994) determined the electric dipole moment of PSI trimers. (They also achieved the experimental separation of LB from LD.) Via the magnetic orientability, the size of the particle and the relative order inside the macrostructure can be estimated (Barzda et al., 1994).

For fully allowed, intense electronic transitions, the polarization is generally constant across an isolated absorbance or fluorescence band. (For cases involving the admixture of vibronic transitions, however, see Nordén et al., 1992.) Thus, the measurements can be used to resolve overlapping bands (Garab and Breton, 1976; Kramer and Amesz, 1982; Mimuro et al., 1990). Essentially the same concept was applied recently in the deconvolution of LD spectra (Zucchelli et al., 1994). This analysis showed that all of the major absorbance forms of Chl a display considerable orientational homogeneity across the band. Hemelrijk et al. (1992) used absorbance, LD and CD spectra to identify different spectral forms of Chl a and b in LHCII. A similar analysis was performed by Matsuura et al. (1993) in chlorosomes. The knowledge of band-structure is also necessary for the determination of the orientation angles of different absorbance and fluorescence transition dipoles (e.g. Garab et al., 1981; van Amerongen et al., 1994).

III. Circular Dichroism

2. Spatial Position of Organelles

CD is the differential absorbance of left and right circularly polarized light:

If the anisotropy of the transition dipoles of a complex system is well characterized, e.g. that of a membrane or a whole chloroplast, information can be obtained on the spatial position of the system, e.g. on the position of a chloroplast in a cell, and the light-induced orientation of chloroplasts inside the cell can be monitored (Tlalka and Gabrys, 1993). Polarization microscopy played a special role in the early studies of the optical anisotropy of chloroplasts (reviewed by Breton and Vermeglio (1982)). With the advance of differential polarization imaging techniques (Kim et al., 1987a) and laser scanning microscopy (Shotton and White, 1989), a more refined use of polarization microscopy seems possible. The fact that chloroplasts can be sliced optically during the recording of LD images (Garab et al., 1991a) opens up the possibility for 3–dimensional reconstruction of the thylakoid membrane system through the use of LD.

CD arises from the intra- or intermolecular asymmetry (helicity) of the molecular structure. The helicity (chirality) of the structure means that it cannot be superimposed on its mirror image; this property is also often called handedness. This lack of symmetry, which arises, for example, if a carbon atom in the molecule is bonded to four different residues, is the property of nearly all organic molecules synthesized in biology. As the handedness of a molecule is the same from any direction, the selective absorbance of left and right circularly polarized light can be observed in a randomly oriented sample. (For oriented systems, see below.) Although CD is the most commonly determined chiroptical quantity, it is necessary to recall the correlations between CD and the other manifestations of optical activity: ellipticity, ORD and CB.

LD and CD

25

Absorbance by an optically active sample induces ellipticity in the linearly polarized measuring beam. This can be understood if it is taken into account that linearly polarized light can be decomposed into two oppositely rotating circularly polarized light components of equal amplitudes (see I.A). After the selective absorbance of one of the components, the two intensities do not remain equal and thus the beam will become elliptically polarized. It is obvious that CD and ellipticity are equivalent quantities. Although modern home-built or commercially available dichrographs measure absorbance differences, it is common practice to express CD in units of ellipticity, millidegrees (m°, mdeg; absorbance unit). (It is interesting to note that the method elaborated for the measurement of time-resolved CD with nanosecond resolution is based on ellipsometry and not (Lewis et al., 1992).) In ORD measurements, the rotation of the orientation of linearly polarized light is measured as a function of wavelength. In an optically active material, and the electric vectors, rotate at different speeds This results in a net rotation in the direction of of the linearly polarized measuring beam. (For a weakly absorbing sample, the elliptically polarized light is considered to be linearly polarized along the major axis of the ellipse.) The optical rotation of a sample can be measured at any wavelength, i.e. also outside the absorbance bands. This is the main advantage of ORD over CD. The relation between CD and ORD is given by the Krönig–Kramers transforms (see Born and Wolf, 1980). Thus, the information from CD and ORD is redundant. The optical activity of a chiral molecule for each electronic transition is characterized by the rotational (or rotatory) strength of the transition. This is analogous to the dipole strength (see I.B) and is measured via the area under the CD band. As concerns the physical meaning of rotational strength it must be stressed that this does not depend solely on the electric dipole of the transition, but also on the magnetic dipole (m): (19)

The magnetic transition dipole is a purely imaginary vector. In the Rosenfeld equation (Eq. 19),

Im indicates that the rotational strength corresponds to the imaginary part of the scalar product, and thus it is a real number. It is evident that, for a molecule to be optically active, both and m must be non-zero, and m must have a component parallel to For these to occur, the molecule must have a non-zero absorbance and be asymmetric. To explain this latter condition, it may be recalled that the electric dipole transition moment corresponds to a linear oscillatory motion of charge induced by the electric field of light (see section I.B), whereas the magnetic transition dipole can be regarded as a light-induced current loop. In asymmetric molecules, light induces a circular motion about the direction of which corresponds to a helical displacement of charge. In molecules that contain a plane or center of symmetry, rotational strength vanishes. (E.g. for a ring, m is perpendicular to the plane of the ring, while is in the plane.) This explains the correlation between the magnitude of the CD and the helicity of molecules, which facilitates the helical flow of charges (Woody, 1985; Charney, 1979).

A. Physical Origins of CD Signals (i) In the basic case, CD arises from intrinsic asymmetry or the asymmetric perturbation of a molecule (Woody, 1985). For a single electronic transition, CD has the same band shape as the absorption, and its sign is determined by the handedness of the molecule (positive or negative Cotton effect). (ii) In molecular complexes or small aggregates, CD is generally induced by short-range, excitonic coupling between chromophores (Tinoco, 1962; DeVoe, 1965). Excitonic interactions give rise to a conservative band structure (i.e. the positive and negative bands of the split spectrum, plotted on an energy scale, are represented with equal areas). (iii) In complex systems, such as DNA aggregates, condensed chromatins, viruses, etc., very intense CD signals have been observed, with non-conservative, anomalously shaped bands accompanied by long tails outside the absorbance. The CD signals of these samples have been shown to originate from the differential absorbance and differential scattering of left and right circularly polarized light:

Garab

26

Systematic studies have revealed that these signals are associated with the macro-organization of the system; they provide valid and unique structural information about large chiral objects and carry information on the long-range chiral organization of chromophores (Keller and Bustamante, 1986a,b; Tinoco et al., 1987). CD signals (i)–(iii) originate from different levels of structural complexity. These different types of CD will be treated in somewhat more detail in the following section. Two other types of CD signals must be added to complete the list of CD due to different physical origins. These latter CD signals can be combined with any level of structural complexity of the sample. (iv) If a chiral molecule (or complex) is luminescent, the emitted light will be partly circularly polarized (Steinberg, 1978). CPL provides a tool for studies of the chirality of the excited state. CPL data on photosynthesis are scarce. Gafni et al. (1975) conducted comparative studies of CD and CPL on Chl dimers in solution, subchloroplast particles and chloroplasts, and demonstrated large differences in both magnitude and sign. Additionally, it was concluded that in chloroplast the emission anisotropy did not depend on the fluorescence yield, which can probably be interpreted as an indication that CPL reflects mainly the organization of the antenna rather than that of the reaction center. Since the sensitivity of CPL to scattering and sieve effects (Duysens, 1956) is different from that of CD, CPL may be a complementary tool in the elucidation of the macro-organization of chromophores in complex systems. CPL should not be confused with FDCD (Tinoco et al., 1987). FDCD detects the difference between the fluorescence intensities due to left and right circularly polarized light. FDCD can also be used to separate CDS from CD. In photosynthetic systems, complications may arise from the large number of fluorescing Chl molecules, and from the intense energy transfer among Chls. (This may explain the lack of data.) FDCD technique has recently been combined with lifetime measurement (Wu et al., 1993). (v) An external magnetic field parallel to a direction of propagation of light represents a perturbation that induces CD in chiral or achiral

samples. Since MCD arises via a different mechanism, it is superimposed on the ‘natural’ CD (Sutherland and Holmquist, 1980). Its magnitude is linearly proportional to the field strength. MCD has contributed significantly to the knowledge of the fine structure of porphyrins (Sutherland, 1978)). MCD is able to detect very weak transitions; hence, certain choromophores can be used as markers. Some weak absorbance and CD bands, e.g. Chl a at 580 nm, exhibit very intense MCD. Fluorescence detection of MCD is also a commonly used tool, e.g. for the separation of MCD signals originating from fluorescing chromophores from the total MCD signal. In analogy to CPL, the magnetically induced chirality of the excited state can be detected as magnetic circular emission. In magnetically orientable systems, MCD and orientation-dependent CD can be combined et al., 1982; Garab et al., 1988a). In general, the CD of oriented systems requires special attention as concerns both the possible artifacts (Davidsson et al., 1980) and the interpretation of data. CD in oriented systems contains additional information due to the fact that chiroptical effects develop along different molecular and crystal axes in specific and different ways (Charney, 1979). For instance, the orientation dependence of excitonic bands can reveal information on the symmetry of the excitonic aggregate. In photosynthesis, the conclusion reached by Charney in 1979 still applies: “This field remains ripe for application”. Vibrational CD (Keiderling et al., 1993) and Raman optical activity techniques (Barron, 1982) combine CD with vibrational methods.

B. CD of Photosynthetic Pigments Different levels of molecular organizations which give rise to different CD signals by different physical mechanisms can also be recognized in Chl-containing systems (Fig. 8). (i) In monomeric solutions, the intrinsic CD of Chls, with band shapes identical to those of the absorbance bands, is very weak (Dratz et al., 1966). (ii) In pigment-protein complexes, Chls typically exhibit CD with a conservative band structure, which arises from excitonic interactions (Pearlstein, 1982). (iii) Granal thylakoid mem-

LD and CD

27

ecular structure by Houssier and Sauer (1970). The intrinsic CD of most open-chain tetrapyrroles (Scheer, 1982) is much greater than that of Chls. Carotenoids are achiral in solution, but when bound to protein they display significant optical activity (Frank et al., 1989). In BChl-containing organisms, CD can be used for the investigation of carotenoid binding (Cogdell and Scheer, 1985). CD suggests asymmetry in the binding environment, but the participation of excitonic interactions cannot be ruled out (Frank and Cogdell, 1993). 2. Excitonic Interactions

branes and macroaggregates of LHCII (Gregory et al., 1980; Garab et al., 1988a) are characterized by non-conservative CD signals with extremely large amplitudes and long scattering tails, attributed to the long-range coupling of chromophores in chirally organized macrodomains (Garab et al., 1988c; Finzi et al., 1989). In a hierarchic system, CD signals of different physical origins are superimposed on each other. Thus, the observed CD spectra of Chl dimers, for example, always contain the spectra of the monomers; this causes some deviation from the conservative band structure, which can be corrected by subtracting the intrinsic CD from the observed signal (for details, see Houssier and Sauer, 1970). Psi-type CD bands of chloroplasts and LHCII were shown not to interfere significantly with the excitonic bands (Garab et al., 1988a; Garab et al., 1991b; Barzda et al., 1994). Thus, the CD of complex systems, at least in principle, can be deconvoluted to the component spectra of different physical origins. 1. Intrinsic CD of Isolated Molecules The CD spectra of several Chl and PChl pigments and their Pheos in diethyl ether were recorded and their origins explained in terms of the mol-

The coupling of two or more molecular dipoles with one another leads to shifts in absorbance bands and can generate CD. Let us consider two identical pigment molecules separated by a distance vector R. If the two molecules are brought close enough to each other to interact electronically, but are still sufficiently apart for the electrons to remain localized on each of the molecules, the absorbance band splits into two bands. The degree of separation of the bands depends on the interaction energy between the two dipoles and

where vector

is in debye, R is in nm; with the unit

This means that the interaction energy depends not only on the dipole strength of the molecules, but also on their distance and mutual orientation. It is to be noted here that the rate of the Förster-type of resonance energy transfer (Förster, 1965) is proportional to The preferentially in-plane orientation of dipoles with respect to the membrane plane (as shown by LD) largely facilitates energy migration in directions parallel to the membrane plane (Garab et al., 1981). Energy transfer interactions are dealt with in a recent review by van Grondelle et al. (1994). The rotational strength (in units of DebyeBohr magnetons) for the excitonic CD of a dimer is given as:

28

where + and – designate the lower and higher energy transitions of the two exciton states, respectively, is the band center energy, and the vectors in scalar triple products are unit vectors, It is clear that for the dimer is independent of for the monomers, and i.e. the CD of the couplet is conservative. In general, an additional term can originate from electric-magnetic coupling (Cantor and Schimmel, 1980). However, in most cases the contribution from this source, also with conservative band structure, is weak. The peak positions of the excitonic CD bands coincide with those of the split absorbance bands, as illustrated by the “stick” spectra (Fig. 9), which can be “dressed” by Gaussian bands (Scherer and Fischer, 1991). In complex systems, the absorbance may exhibit numerous sub-bands

Garab

due to environmental effects, which makes identification of the different bands very difficult. In contrast, CD selectively detects the excitonic interactions on a weak background of the intrinsic CD of the pigment molecules, thereby offering a much better sensitivity than absorbance for identification of the the excitonic bands . On the other hand, in some geometries no excitonic CD signal appears, e.g. if the two dipoles are coplanar or if they are oriented parallel to one another. In many conformations of the dimer and favorable orientation angles of the dipoles with respect to the symmetry axis of the complex, the excitonic nature of a dimer can also be recognized in LD (Pearlstein, 1982). Thus, at least for excitonic interactions, the absorbance, CD and LD spectra are correlated with each other. For excitonic interactions with known geometry it is possible to calculate these spectra, as in purple bacterial reaction centers (Scherer and Fischer, 1991). In most applications, however, LD and CD spectra

LD and CD

are recorded with the aim of determining the spatial relationship and interactions among pigment dipoles, respectively. It must also be realized that in complex systems intense excitonic interactions may be confined to a relatively small cluster of pigment molecules. On the other hand, the majority of pigment dipoles are non-randomly oriented with respect to the symmetry axis (Breton and Vermeglio, 1982; Garab et al., 1987). The characteristic CD spectra of different particles and isolated complexes have been studied by many authors. The experimental results and their interpretation have been reviewed (Pearlstein, 1987, 1991). The brief summary below serves to illustrate most typical applications. Although the six pigments in the purple bacterial reaction center are in close contact, the CD of the complex is dominated by the signal of the special pair, and the remaining BChl and BPheo molecules exhibit less intense bands. As discussed in depth by Pearlstein (1991), since the diameter of the macrocycles is larger than the center to center distance of the two molecules, the point dipole approximation is not satisfactory. Further corrections are necessary if the Soret dipoles interact with the dipole of a nearby molecule (Scherz and Parson, 1986) or if a charge transfer complex is formed (Parson and Warshel, 1987). Picosecond CD transients also indicated the role of non-excitonic CD contributions (Xie and Simon, 1991). Model systems, such as BChl aggregates in micelles, permit characterization of both the excitonic and non-excitonic CD bands, which are similar to those in the reaction center (Scherz, 1992). The CD spectra of in vitro PChl aggregates also display many similarities to those in native systems (Böddi and Shioi, 1990). The pigment organization in the reaction center of a green bacterium, Chloroflexus aurantiacus, was analyzed by using exciton theory and the structure of Rps. viridis (Deisenhofer et al., 1984). The data suggested that the arrangement of the chromophores is very similar to that in purple bacteria, which explains the functional similarity of the two reaction centers (Vasmel et al., 1986). The contribution of P680 to the CD of the PSII reaction center D1-D2-cyt b559 was investigated by Braun et al. (1990): the split (+)679 nm and (–)669 nm bands were interpre-

29

ted as indicating a loosely coupled dimer of Chl a, in which the interaction is much weaker than in the purple bacterial reaction centers (cf. also van der Vos et al., 1992). In an exciton-coupled aggregate, many interactions occur, which affect the absorbance, CD and LD of the aggregate (Pearlstein, 1991). The water-soluble BChl a-containing complex of the green photosynthetic bacterium, Prosthecocloris aestuarii, the Fenna–Matthews–Olson (FMO) complex was the first photosynthetic pigmentprotein complex to be structurally characterized by atomic resolution crystallography (Fenna and Matthews, 1975) and is one of the best-studied complex in photosynthesis. However, interpretation of the absorbance and CD of the FMO complex turned out to be more difficult than anticipated, and in fact CD has been explained satisfactorily (Lu and Pearlstein, 1993). In the simulation, the distinct interaction of each of the seven BChl molecules with protein moieties, and the trimeric nature of the complex had to be taken into account (Pearlstein, 1992). Lu and Pearlstein (1993) found that some cryosolvents perturb the structure. Zhou et al. (1994) have reported the occurrence of redox sensitive structural changes in the FMO complex which could be detected by CD. For LHCII, in which the structure is known at 3.4 Å resolution (Kühlbrandt et al., 1994) an exact interpretation of the CD data has not been presented. The characteristic spectra of CP2 and LHCII were interpreted in terms of a Chl b trimer (van Metter, 1977; Gülen and Knox, 1984). However, it was later concluded (Hemelrijk et al., 1992) that the CD signal probably originates from an array of pigment molecules in which not only Chl b-Chl b, but also Chl b-Chl a interactions play an important role. For intramembrane purple bacterial complexes, the lack of high-resolution structure hampered analysis of the CD spectra in terms of exact structural parameters. Analyses of the CD spectra and the amino acid sequences led to the construction of models which satisfactorily described the structure and function of these antenna complexes (Scherz and Parson, 1986; Zuber and Brunischolz, 1991; Visshers et al., 1991; see also Pearlstein, 1992). The crystal structure of the B800-850 antenna complex from Rps. acidophila

30

has been determined to a resolution of 2.5 Å by McDermott et al. (1995). This opened up the possibility to use CD toward the understanding of the excitation energy transfer processes. In heliobacteria, several BChl g forms have been identified which transfer energy efficiently to the longest wavelength (808 nm) form, but CD indicates that only a relatively small number of pigments participate in excitonic interactions (van Dorssen et al., 1985), which emphasizes the importance of pigment clusters in bacterial antenna complexes. CD spectra have been recorded for most subchloroplast particles and complexes (Bassi et al., 1985), but most of them have not been analyzed in terms of exciton theory. These CD spectra are clearly dominated by different excitonic bands characteristic of the different pigment-protein complexes, suggesting that CD can be used for the “fingerprinting” of complexes (for an overview, see Garab et al., 1987). The Chl b-containing antenna of Prochlorotix hollandica has been shown not to contain the characteristic CD bands of LHCII, which indicates that this complex is not closely related to LHCII (Matthijs et al., 1989). The similarity of the CD spectra of the antenna Chls in the native membrane and the trimeric form of the isolated PSI reaction center complex suggests that trimeric PSI pre-exists in the membrane (Shubin et al., 1993). In all the above systems, proteins provide the binding sites, which in turn ensure the appropriate distances and mutual orientations of the Chl molecules. In chlorosomes, BChl c oligomers appear to be the main building blocks, thus investigations of model systems of large Chl aggregates are of great interest (Scherz et al., 1991; Gottstein et al., 1993). Studies of macroaggregates of PChl also revealed many similarities to some in vivo systems (Böddi and Láng, 1984; Sundqvist et al., 1980). In chlorosomes and in macroaggregates of Chls and PChl (i) the CD is sensitive to the conditions of preparation, (ii) the size of the aggregates is commensurate with the wavelength of visible light, and (iii) the pigment molecules appear to be assembled with long-range order. Thus, the CD signals may originate in part from the long-range chiral order of the pigment molecules, as suggested by Lehmann et al. (1994) who observed a giant CD signal in chlorosomes

Garab treated with protease. Identification of the origin of the CD signals in these highly-organized macrosystems and determination of the possible contribution of psi-type effects requires systematic investigations. 3. Differential Scattering, Psi-type CD, and Differential Polarization Imaging The theoretical framework for understanding the CD in large chiral assemblies has been extensively developed over the past two decades. Preferential scattering of one of the circular polarizations of the light by chiral samples has been interpreted within the framework of the CIDS theory (Bustamante et al., 1985). The theory for psi-type aggregates describes the interaction of light with large inhomogenous molecular aggregates containing a high density of intensely interacting chromophores (Keller and Bustamante, 1986a,b). The theory of imaging macrodomains that have different optical properties and different molecular structures has been elaborated for CD (Keller et al., 1985) and extended to all transmisson or scattering Mueller images (Kim et al., 1987a,b). CIDS results from interference effects of wavelets generated at different points in the object in which the point polarizable groups are helically arranged, and the pitch and the diameter of the helix are commensurate with the wavelength of the measuring light. Since the wavelets maintain a well-defined polarization and phase relationships to each other, the interference phenomenon is greatest when the wavelength of the circularly polarized light closely matches the dimensions of the macrohelix. In the first Born approximation, each dipole is considered independent of the others. The interaction of subunits can be taken into account in higher Born approximations (Tinoco et al., 1987). Theory predicts that CIDS as a function of scattering angle:

exhibit lobes of alternating sign (Bustamante et al., 1985), the profile of which is determined by the helical parameters of the chiral macrostructure. It must be stressed that the angle-depen-

LD and CD

dence of the non-polarized scattering is not related to the profile of CIDS. Non-polarized scattering carries information on the shape and size of the particle rather than on the organization of chromophores inside the scattering particle. Nevertheless, both non-polarized and differential polarization scattering signals, especially when measured as a function of the angle of observation, carry useful information on the macrostructural parameters of large objects. The psi-type CD theory (Keller and Bustamante, 1986a,b) is based on the classical theory of coupled oscillators (DeVoe, 1965). The theory of DeVoe considers that light induces oscillating (transition) dipoles in the polarizable groups of the object, and the induced dipoles interact as static dipoles, with a distance-dependence of However, in large objects, it is necessary to consider not only these short-range interactions but also long-range effects. In psi-type aggregates the full electrodynamic interaction between the dipoles must be taken into account (Keller and Bustamante, 1986a). In small aggregates, the entire aggregate at any instant is in the same phase of the wave upon the interaction with the light. In contrast, in large aggregates which are commensurate with the wavelength this is not true and retardation effects can play an important role. At distant points of observation the oscillating dipole can be regarded as a radiating spherical wave. Thus, the chromophores at large distances can be coupled via a radiation coupling mechanism between the dipoles. In large chirally organized aggregates the significance of the radiation coupling, which is essentially due to multiple scattering inside the particle, can be comparable to that of the static dipole coupling. The electric field at any point in space x, due to an oscillating electric dipole, located at can be written as: Furthermore,

which means that the electric field at any point is the superposition of the incident electric field and the sum of the fields produced by all oscillating dipoles. This shows that for the general case the quantity of interest in understanding the

31

CD (and other optical properties) of large aggregates is not the coupling between individual pairs of chromophores, but the coupling between any given chromophore and the rest of the chromophores in the macrodomain. The interaction tensor has been given in an explicit form (Keller and Bustamante, 1986a):

where and The first and third terms of the tensor, with and dependences, describe the static dipole coupling and the radiation term, respectively. The second term, with is called the intermediate coupling. The final term ensures that the self-interaction is zero. Due to static, intermediate and radiation coupling between dipoles, intense “anomalous” CD signals are generated (Keller and Bustamante, 1986b). The magnitude of the signal is controlled by the volume, chromophore density and pitch of the helically organized macrodomain (Kim et al., 1986). Further, the shape of the spectra depends mostly on the pitch and the handedness, with sign-inverted mirror spectra for opposite handedness. In psi-type aggregates, the theoretical prediction is that if the long-range coupling between the dipoles is strong, the excitation generated at one chromophore can delocalize for the entire aggregate. This is called the collective absorbance, which increases or decreases at a given wavelength depending on how well the light is able to produce a collective excitation in the system. When the group polarizabilities are made weaker, the groups are moved farther apart or the density of the chromophores inside the aggregate is diminished, the ability of an excitation created at a given position in the aggregate to transfer to a different part becomes less and less efficient. Finally, for the case of a small system the theory of psi-type CD (Keller and Bustamante, 1986a,b) reduces to the classical theory of DeVoe (1965), which describes excitonic interactions. Since psi-type aggregates also satisfy the criteria for CIDS, psi-type CD is always accompanied

32

by CIDS. In CIDS alone, the static and intermediate coupling fields become insignificant and can be omitted from Differential polarization imaging (DPI) is a method suitable for structural investigations of large anisotropic objects. It can provide information on the macro-organization of large molecular structures, and resolve microscopic domains of distinct optical anisotropy. In DPI, the image is composed of points carrying information on a differential polarization quantity such as CD or LD or other elements of the Mueller matrix (Kim et al., 1987a,b; Kim and Bustamante, 1991). Philipson and Sauer (1973) recognized that CDS contributes to the CD of chloroplasts. Later, it was shown that scattering does not significantly distort the “true” CD bands in chloroplasts and LHCII macroaggregates, but is superimposed on the excitonic CD signals and carries distinct physical information on the macro-organization of pigments (Garab et al., 1988a,c). A comparison of non-polarized scattering with the anomalous CD in thylakoid membranes subjected to different ionic strengths and osmotic pressure revealed that while CD selectively responds to structural rearrangements of the pigment-protein complexes, non-polarized scattering yields far less specific information (Garab et al., 1991b). Further, intense light scattering per se, e.g. in a suspension of thylakoid membranes does not noticeably affect the excitonic CD bands. of chloroplast thylakoid membranes was measured in a set-up involving an Ar-ion laser, a Pockel’s cell and goniometric detection of the scattered light. It was found that chloroplasts exhibited four CD lobes with alternating signs, which was attributed to a left-handed helix with an estimated pitch and radius of 200–400 nm (Garab et al., 1988c). The helically organized macrodomains were imaged at different wavelengths in a confocal CD microscope (Finzi et al., 1989, 1991). The images displayed huge signals emerging from discrete “islands” of the chloroplasts; the diameter of which could be estimated to be between 0.3 and 0.6 Local CD spectra recorded in different pixels of individual chloroplasts exhibited broad positive and negative bands from different domains (Finzi

Garab et al., 1989). Microscopic data ruled out the interpretation of the anomalous CD of chloroplast suspensions in terms of short-range interactions and lent support to the notion that the main bands originate from helically organized macrodomains of the pigment system. The chirally organized macrodomains in chloroplasts have also been shown to undergo gross (up to 80–90%) lightinduced reversible structural changes which can be detected in the major “anomalous” CD bands (Garab et al., 1988b). The significance of these structural changes is not understood exactly, but these changes are likely to be correlated with the energy dependent non-photochemical quenching which is capable to dissipate excess excitation energy in the antenna (see e.g., Horton et al., 1991; Istokovics et al., 1992). In granal chloroplasts and LHCII macroaggregates, the density of chromophores is high, the dipoles interact intensely with each other, and the complexes are assembled in large 3–dimensional aggregates (Barzda et al., 1994). Hence, if asymmetry is introduced, e.g. during macroaggregation, these systems satisfy the conditions for psitype aggregates. Barzda et al. (1994) showed that, in accordance with the prediction of psi-type theory, the magnitude of the major CD bands of LHCII and chloroplasts increased with the size of the macroaggregates. A similar correlation was found for the B880 antenna complex of Rps. marina (R. Meckenstock, G. Garab, R. A. Brunisholz and H. Zuber, unpublished.) With LHCII and B880 the size of the aggregate can be varied in a broad range. These systems offer the convenience that the measurements can be carried out in the visible and near-IR spectral regions. Furthermore, our knowledge concerning the structure of the pigment system is usually more advanced than that on the chromophores of most non-photosynthetic psi-type aggregates. Thus, photosynthetic systems appear to be ideally suited for systematic studies of psi-type CD. Without systematic studies under well defined experimental conditions and on systems permitting realistic model calculations quantitative interpretation of psi-type spectra does not seem possible. It would be equally important to understand the functional significance of the psi-type organization of the pigment system in granal chloroplasts.

LD and CD

C. Secondary Structure of Chl-containing Proteins It has been established that the CD of proteins is sensitive to the secondary structure. The signal observed in the far-UV spectral range is primarily determined by the spatial arrangement of the amide chromophores, and therefore the CD reflects the backbone conformation of the polypeptides. The amide chromophore has plane symmetry and is itself not optically active. In a peptide, a chiral electrostatic field can be provided by the surrounding amides and other polar groups; exciton interactions between chromophores also contribute significantly to the CD signal (Woody, 1985; Johnson, 1990). Some of the interactions can be specifically assigned to certain conformations (Woody, 1985). The conformations are characterized by the 222 nm negative band and an excitonic couplet at (–)208 nm/(+)192 nm. The spectrum of the also depends on the length of the chain. are usually characterized by a (–)216 nm band and a much stronger positive band between 195 and 200 nm. Different types of which play an important role in the folding of proteins, have been shown to occur in equilibrium (Perczel et al., 1991). These are usually characterized by negative bands at 220–230 nm and 180–190 nm, and a positive band between 200 and 210 nm. Aperiodic (random coil) proteins have an unordered structure and usually exhibit weak and highly variable CD spectra, often with a negative band at around 200 nm. In accordance with the principle of additivity of CD signals, the CD spectrum of a protein can be considered to be the weighted sum of the spectra of the secondary conformations. This is the basis for analysis of the spectra for prediction of the secondary structure of proteins. Semiempirical methods for this analysis are based on different sets of CD spectra of proteins with known secondary structure (Woody, 1985; Johnson, 1990). The prediction methods, and especially those which use linear combinations of known structures and CD spectra and also apply statistical methods in order to account for variabilities (Provencher and Glöckner, 1981; Johnson, 1990), yield a highly reliable prediction of the

33

content, while those relating to and turns are less reliable (Johnson, 1990). This conclusion of Johnson (1990) has been confirmed by a systematic analysis of UV-CD and Fourier transform infrared spectra (Hollósi et al., 1993; Pribic et al., 1993). Conformational analyses of the proteins for the content of purple bacterial reaction centers have yielded values comparable to those deduced from X-ray data (47–55% and 42%, respectively) . Similar contents were found in other isolated complexes (Breton and Nabedryk, 1987).The orientation of the with respect to the membrane normal was found to be less than 30°, as determined from IR LD measurements (Nabedryk et al., 1984). Based on UV-CD measurements Paulsen et al. (1993) concluded that pigment binding plays an important role in the determination of the secondary structure of LHCII.

D. Artifacts In molecular solutions or small aggregates, most of the artifacts originate from the optical trail of the set-up and the cell, which contain residual strains that cannot be fully eliminated. These can be taken into account by subtracting the baseline measured with the same cell in the same orientation and with a mimicked absorbance. In the presence of large particles, which “concentrate” the chromophores and are commensurate with the wavelength of the measuring light, both the absorbance and the CD are affected by flattening and light scattering (Duysens, 1956; Bustamante and Maestre, 1988). The degree of flattening of the CD spectrum is twice as large as the flattening of the absorbance, and both are increasingly more significant toward shorter wavelengths (Bustamante and Maestre, 1988). However, in the presence of CDS contributions, the flattening effects can be more complex; corrections can then be difficult and efforts must be concentrated first on correcting for CDS. In a conventional dichrograph, CD of absorbance and CDS are combined into the apparent CD signal. CDS can most easily be recognized by varying the acceptance angle of the photomultiplier, e.g. by changing the distance between the

34

sample and the photomultiplier (Philipson and Sauer, 1973). It is relatively simple to separate CDS from CD of absorbance. Outside the absorbance band, CD is produced by CDS and exhibits a monotonously decreasing signal (“long tail”) on the long wavelength side of the absorbance band. CDS can also contribute to the CD signal inside the absorption band. (Inside the absorbance band, the anomaly can also originate from psi type CD, which is always accompanied by CDS,) As CDS can be intense in the forward direction, and as there is a ‘cross term’ between the absorbance and scattering (Bustamante and Maestre, 1988) simple corrections for scattering contributions do not provide reliable results. FDCD and photoacoustic techniques have been proposed for such corrections. FDCD, however, may not easily be adapted for the highly fluorescing photosynthetic systems, and to my knowledge, photoacoustic CD measurements have not been attempted. In chloroplasts and LHCII macroaggregates, MCD of Chls, which originates from inside the complexes and is therefore subjected to the same alterations as natural CD, has been shown not to be distorted significantly by CDS (Garab et al., 1988a). In chloroplast suspensions, “regular” scattering (which does not discriminate between left and right circularly polarized light) does not distort the CD signal to any noticeable extent. This may be observed, for example, in turbid chloroplast suspensions in which the electrostatic conditions do not permit the formation of large aggregates (Garab et al., 1991b). CD is correlated with CB, and thus CB is often suspected to contribute to the CD signal. It has been demonstrated, however, that both in conventional CD measurements and in ellipsometry CB in first order does not contribute to the signal (Björling et al., 1991; Lewis et al., 1992). LD and LB can easily cause artifacts. Since all photosynthetic membranes are intrinsically anisotropic, the extent of interference to CD by LD and LB must always be investigated in oriented systems. The facts that LD is usually more than an order of magnitude stronger than CD, and that imperfections may occur in the optical trail, make CD sensitive to LD and LB. This problem of CD in oriented systems has been addressed by

Garab many authors (Davidsson et al., 1980; Shindo et al., 1985; Shindo, 1985). (In randomly oriented samples LD and LB can be induced by a linearly polarized excitation. Such excitation can cause artifacts in pico- and nanosecond CD transients.) Artifacts in CD due to LD and LB and their ‘cure’ were analyzed in detail by Björling et al. (1991) and Lewis et al. (1992). CD was found to be most sensitive to the coupling of LD with stray LB in the optical components before the sample. An artifact due to coupling between the LB of the optical system and the LD of the sample was reported by Francke et al. (1994), who observed spurious CD signals in intact cells of heliobacteria. The LD was caused by gravitational orientation of the cells, whereas the LB originated mainly from the strain in the window of the cryostat. The magnitude of the spurious signal increased dramatically in response to lowering of the temperature, which suggested an additional (probably internal pressure-dependent) LB inside the sample. The effects of different contributions to the CD images from sources, such as LD, linear differential scattering and LB were analyzed by Kim et al. (1987b). It was shown that most of these contributions can be separated on the basis of the symmetry behaviour of different Mueller matrix elements upon rotation of the optical components. In particular, CD image artifacts due to imperfections in circular polarization can be eliminated by using the fact that the CD-related images or are invariant on the rotation whereas contributions from a linear anisotropic signal are sensitive to rotation. CD images of chloroplasts were tested for linear dichroic contributions in both face-aligned (LD = 0) and edge-aligned positions, and it was found that the main features of the CD images were unaffected. (The LD images changed sign upon 90° rotation of the sample or the photoelastic modulator (Finzi et al., 1989).) Similar experiments were performed on macroscopic samples with magnetically aligned chloroplasts trapped in gel. CD spectra were recorded in vertical position and at ± 45° and ± 90° with respect to the vertical direction. The contribution of LD to CD (in position) was found to be smaller than the CD signal (Garab et al., 1991a).

LD and CD

IV. Concluding Remarks The application of LD and CD spectroscopic techniques have provided a rich yield of structural information on the pigment systems in various photosynthetic complexes, membranes and organelles. For most applications, LD and CD have solid theoretical background and elaborated experimental procedures. Thus these techniques are suitable for routine applications in a wide range of studies. For special applications, however, further work is needed. In LD, earlier investigations concentrated mainly on qualitative conclusions and establishing the general rules of dipole orientations in vivo. Now, there appears a demand for a quantitative approach, and its use in monitoring the primary photophysical and photochemical processes. Conclusions on the orientation of the molecules inside Chl-containing complexes may, however, remain tentative because of some uncertainities in the polarization directions of the electronic transitions of Chls. The importance of excitonic CD has been demonstrated thoroughly in almost all photosynthetic complexes studied so far. As shown for the case of the FMO complex, CD contains information on virtually all possible pigment–pigment interactions as well as the interactions of the pigments with the protein moieties. This information, with so much details, is clearly impossible to extract from the CD spectrum alone. For less characterized structural entities, the conclusion from CD may be confined to the identification of the most important excitonic interactions. This can be improved by analyzing LD, CD and absorbance spectra on the basis of the physical correlations between the bands of excitonic origin, and also by using independently determined structural parameters. CD and LD studies can be extended to intact systems where unique structural information appears to be available on the macro-organization of the pigments. Signals originating from highly organized systems, however, may be combined with artifacts, which suggests a cautious approach. Our understanding of the CD features of highly organized objects with sizes commensurate with the wavelength is far from being complete

35

but progress is very likely, for these questions are in the focus of interest in many spectroscopic laboratories. It is somewhat surprising that only three out of sixteen elements of the Mueller matrix are determined routinely. In highly organized large objects which contain high density of interacting transition dipoles, and where these dipoles are distributed in a well defined anisotropic pattern, and the sample exhibits multilevel optical activities, these elements may provide insight into the molecular organization of macrodomains. Acknowledgements I thank Dr. H. van Amerongen and Prof. R. van Grondelle for critical reading of the manuscript and helpful suggestions. I am indebted to my coworkers, Virginijus Barzda and Tamás Jávorfi, for their help in the preparation of the figures. I am grateful to Profs. M. Hollósi and H. Scheer, and Dr. L. Zimányi for stimulating discussions. Thanks are due to Prof. R. Jennings and Dr. H. van Amerongen for providing their preprints. This work was supported by a grant from the Hungarian Research Fund, OTKA (IV/2999). References Abdourakhmanov I and Erokhin YE (1980) Linear dichroism of pigments associated with spherical chromatophores. Model of orientation in polyacrylamide gels. Mol Biol (USSR) 14: 539–548. (Russian edition). Abdourakhmanov I, Ganago AO, Erokhin YuE, Solov’ev A and Chugunov V (1979) Orientation and linear dichroism of the reaction centers from Rhodopseudomonas sphaeroides R-26. Biochim Biophys Acta 546: 183–186. Ade H and Hsiao B (1993) X-ray linear dichroism microscopy. Science 262: 1427–1429. Barron LD (1982) Molecular Light Scattering and Optical Activity. Cambridge Univ. Press, Cambridge. Barzda V, Mustárdy L and Garab G (1994) Size dependency of circular dichroism in macroaggregates of photosynthetic pigment-protein complexes. Biochemistry 33: 10837– 10841. Bassi R, Machold O and Simpson D (1985) Chlorophyllproteins of two photosystem I preparations from maize. Carlsberg Res Commun 50: 145–162. Bauman D and Wrobel D (1980) Dichroism and polarized fluorescence of chlorophyll a, chlorophyll c, and bacteriochlorophyll a dissolved in liquid crystals. Biophys Chem 12: 83–91. Björling SC, Goldbeck RA, Milder SJ, Randall CE, Lewis

36 JW and Kliger DS (1991) Analysis of optical artifacts in ellipsometric measurements of time-resolved circular dichroism. J Chem Phys 95: 4685–4694. Bloemendal M and Van Grondelle R (1993) Linear-dichroism spectroscopy for the study of structural properties of proteins. Mol Biol Reports 18: 49–69. Bloemendal M, Leunissen JAM, Van Amerongen H and Van Grondelle R (1990) Average orientation of aromatic residues in proteins determined from linear dichroism spectroscopy. A comparison of results on bovine with X-ray data. J Mol Biol 216: 181–186. Böddi B and Láng F (1984) Molecular structure of protochlorophyll forms. In: Sironval C and Brouers M (eds) Protochlorophyllide Reduction and Greening, pp 341–349. Martinus Nijhoff/Dr W Junk Publ, The Hague. Böddi B and Shioi Y (1990) Spectroscopic forms of oleinylprotochlorophyllide in solid films. Biochim Biophys Acta 1015: 116–120. Born M, Wolf E (1980) Principles of Optics. 6th edn. Pergamon Press, Oxford. Braun P, Greenberg BM and Scherz A (1990) D1–D2–cytochrome b559 complex from the aquatic plant Spirodela oligorrhiza: Correlation between complex integrity, spectroscopic properties, photochemical activity, and pigment composition. Biochemistry 29: 10376–10387. Breton J (1986) Molecular orientation of the pigments and the problem of energy trapping in photosynthesis In: Staehelin A and Arntzen (eds) Encyclopedia of Plant Physiology, pp 319–326. Springer-Verlag, Berlin. Breton J and Ikegami I (1989) Orientation of photosystem I pigments: low temperature linear dichroism spectroscopy of a highly-enriched P700 particle isolated from spinach. Photosynth Res 21: 27–36. Breton J and Nabedryk E (1987) Pigment and protein organization in reaction center and antenna complexes. In: Barber J (ed) The Light Reactions, pp 159–195, Elsevier, Amsterdam. Breton J and Verméglio A (1982) Orientation of photosynthetic pigments in vivo. In: Govindjee (ed) Photosynthesis, pp 153–193. Academic Press, New York. Breton J, Michel-Villaz M, Paillotin G and Vandevyver M (1972) Application of linear dichroism to the study of the distribution of pigments in monomolecular layers. Thin Solid Films 13: 351–357. Breton J, Becker JF and Geacintov N (1973a) Fluorescence depolarization study of randomly oriented and magnetooriented spinach chloroplasts in suspension. Biochem Biophys Res Comm 54: 1403–1409. Breton J, Michel-Villaz M and Paillotin G (1973b) Orientation of pigments and structural proteins in the photosynthetic membrane of spinach chloroplasts: A linear dichroism study. Biochim Biophys Acta 314: 42–56. Bustamante C and Maestre MF (1988) Statistical effects in the absorption and optical activity of particulate suspension. Proc Natl Acad Sci USA 85: 8482–8486. Bustamante C, Maestre MF and Keller D (1985) Expressions for the interpretation of circular intensity differential scattering of chiral aggregates. Biopolymers 24 (8): 1595–1612. Cantor CR and Schimmel PR (1980) Biophysical Chemistry Part II: Techniques for the Study of Biological Structure

Garab and Function. W. H. Freeman and Company, San Francisco. Charney E (1979) The Molecular Basis of Optical Activity. John Wiley and Sons, New York. Charney E (1988) Electric linear dichroism and birefringence of biological polyelectrolytes. Q Rev Biophys 21: 1–60. Clayton RK (1980) Photosynthesis. Physical Mechanisms and Chemical Patterns. Cambridge University Press, Cambridge. Cogdell RJ and Scheer H (1985) Circular dichroism of lightharvesting complexes from purple photosynthetic bacteria. Photochem Photobiol 42: 669–678. Davidsson A, Nordén B and Seth S (1980) Measurement of oriented circular dichroism. Chem Phys Lett 70: 313–316. Deisenhofer J, Epp O, Miki K, Huber R and Michel H (1984) X-ray structure analysis of a membrane protein complex. Electron density map at 3 Å resolution and a model of the chromophores of the photosynthetic reaction center from Rhodopseudomonas viridis. J Mol Biol 180: 385–398. DeVoe H (1965) Optical properties of molecular aggregates. II. Classical theory of the refraction, absorption, and optical activity of solutions and crystals. J Chem Phys 43: 3199–3208. Dratz EA, Schultz AJ and Sauer K (1966) Chlorophyll-chlorophyll interactions. Brookhaven Symp Biol 19: 303–318. Duysens LNM (1956) The flattening of the absorption spectrum of suspensions, as compared to that of solutions. Biochim Biophys Acta 19: 1–12. Fenna RE and Matthews BW (1975) Chlorophyll arrangement in a bacteriochlorophyll protein from Chlorobium limicola. Nature 258: 573–574. Finzi L, Bustamante C, Garab G and Juang C-B (1989) Direct observation of large chiral domains in chloroplast thylakoid membranes by differential polarization microscopy. Proc Natl Acad Sci USA 86: 8748–8752. Finzi L, Ulibarri L and Bustamante C (1991) Differential polarizarion imaging. V. Numerical aperture effects and the contribution of preferential scattering and absorption to the circular dichroism images. Biophys J 59: 1183–1193. Förster T (1965) Delocalized excitation and excitation transfer. In: Sinanoglu O (ed) Modern Quantum Chemistry. Part III: Action of Light and Organic Crystals, pp 93–137. Academic Press, New York. Bauman D and Stillman MJ (1982) Circular dichroism and magnetic circular dichroism spectra of chlorophyll in nematic liquid crystals. Biochim Biophys Acta 681: 273–285. Hotchandani S, Leblanc RM, Szurkowski J (1984) Photoacoustic spectra of chlorophyll a and chlorophyll b in liquid crystal matrix. In: Sybesma C (ed) Advances in Photosynthesis Research, Vol I, pp 713–716. Martinus Nijhoff/Dr. W. Junk Publ., The Hague. Lorrain L, Wrobel D and Leblanc RM (1985) Polarized photoacoustic, absorption and fluorescence spectra of chloroplasts and thylakoids oriented in polyvinyl alcohol films. Biochem Biophys Res Comm 126: 254–261. Bauman D, Manikowski H, Browett WR and Stillman MJ (1987) Circular dichroism and magnetic circular dichroism spectra of chlorophylls a and b in nematic

LD and CD liquid crystals. II. Magnetic circular dichroism spectra. Biophys Chem 28: 101–114. Cegielski R and Leblanc RM (1990) Thermal deactivation of excitation of Rhodospirillum rubrum cells and their fragments immobilized on polymer films. Photosynthetica 24: 85–95. Wrobel D and Dudkowiak A (1991) Application of the polarized light photoacoustic in photosynthesis research. Trends Photochem Photobiol 2: 411–420. Fragata M (1991) Resolved and electronic transitions in chlorophyll a. In: Douglas RH, Moan J and Rontó G (eds) Light in Biology and Medicine, Vol 2, pp 393–399. Plenum Press, New York. Fragata M, Nordén B and Kurucsev (1988) Linear dichroism (250–700 nm) of chlorophyll a and pheophytin a oriented in a lamellar phase of Characterization of electronic transitions. Photochem Photobiol 47: 133. Francke C, Otte SCM, Van der Heiden JC and Amesz J (1994) Spurious circular dichroism signals with intact cells of heliobacteria. Biochim Biophys Acta 1186: 75–80. Frank HA, Cogdell RJ (1993) Photochemistry and function of carotenoids in photosynthesis. In: Young A and Britton G (eds) Carotenoids in Photosynthesis, pp 253–326. Chapman and Hall, London. Frank HA, Violette CA, Taremi SS and Budil DE (1989) Linear dichroism of single crystal of the reaction centre from Rhodobacter sphaeroides wild type strain 2.4.1. Photosynth Res 21: 107–116. Gadella T, Jovin T and Clegg R (1993) Fluorescence lifetime imaging microscopy (FLIM): spatial resolution of microstructures on the nanosecond time scale. Biophys Chem 48: 221–239. Gafni A, Hardt H, Schlessinger J and Steinberg IZ (1975) Circular polarization of fluorescence of chlorophyll in solution and in native structures. Biochim Biophys Acta 387: 256–264. Gagliano A, Geacintov N and Breton J (1977) Orientation and linear dichroism of chloroplasts and sub-chloroplast fragments oriented in an electric field. Biochim Biophys Acta 461: 460–474. Ganago AO and Fock M (1981) Direct and reverse problems in linear dichroism studies. Spectr Lett 14: 405–414. Ganago IB, Klimov W, Ganago AO, Shuvalov VA and Erokhin YE (1982) Linear dichroism and orientation of pheophytin, the intermediary electron acceptor in photosystem II reaction centers. FEES Lett 140: 127–130. Ganago AO, Garab G and Faludi-Dániel A (1983) Analysis of linearly polarized fluorescence of chloroplasts oriented in polyacrylamide gel. Biochim Biophys Acta 723: 287– 293. Garab G and Breton J (1976) Polarized light spectroscopy on oriented spinach chloroplasts. Fluorescence emission at low temperature. Biochem Biophys Res Comm 71: 1095–1102. Garab G, Kiss JG, Mustárdy L and Michel-Villaz M (1981) Orientation of emitting dipoles of chlorophyll a in thylakoids. Consideration on the orientation factor in vivo. Biophys J 34: 423–437. Garab G, Szitó T, Faludi-Dániel A (1987) Organization of pigments and pigment-protein complexes of thylakoids re-

37 vealed by polarized light spectroscopy. In: Barber J (ed) The Light Reactions, pp 305–339. Elsevier, Amsterdam, New York, Oxford. Garab G, Faludi-Dániel A, Sutherland JC and Hind G (1988a) Macroorganization of chlorophyll a/b light-harvesting complex in thylakoids and aggregates: Information from circular differential scattering. Biochemistry 27: 2425– 2430. Garab G, Leegood RC, Walker DA, Sutherland JC and Hind G (1988b) Reversible changes in macroorganization of the light-harvesting chlorophyll a/b pigment-protein complex detected by circular dichroism. Biochemistry 27: 2430– 2434. Garab G, Wells KS, Finzi L and Bustamante C (1988c) Helically organized macroaggregates of pigment-protein complexes in chloroplasts: Evidence from circular intensity differential scattering. Biochemistry 27: 5839–5843. Garab G, Finzi L, Bustamante C (1991a) Differential polarization imaging of chloroplasts: Microscopic and macroscopic linear and circular dichroism. In: Douglas RH, Moan J and Rontó G (eds) Light in Biology and Medicine, pp 77–88. Plenum Press, New York, London. Garab G, Kieleczawa J, Sutherland JC, Bustamante C and Hind G (1991b) Organization of pigment-protein complexes into macrodomains in the thylakoid membranes of wild-type and chlorophyll b-less mutant of barley as revealed by circular dichroism. Photochem Photobiol 54: 273–281. Geacintov N, Van Nostrand F, Becker JF and Tinkel J (1972) Magnetic field induced orientation of photosynthetic systems. Biochim Biophys Acta 267: 65–79. Gottstein J, Scherz A and Scheer H (1993) Bacteriochlorophyll aggregates in positively charged micelles. Biochim Biophys Acta 1183: 413–416. Gregory RPF, Demeter S and Faludi-Dániel A (1980) Macromolecular organization of chlorophyll a in aggregated chlorophyll a/b protein complex as shown by circular dichroism at room and cryogenic temperatures. Biochim Biophys Acta 591: 356–360. Gülen D and Knox RS (1984) Absorption and circular dichroism of the chlorophyll–protein CPII: extensions of a trimeric exciton model. Photobiochem Photobiophys 7: 277– 286. Haworth P, Arntzen CJ, Tapie P and Breton J (1982) Orientation of pigments in the thylakoid membrane and in the isolated chlorophyll-protein complexes of higher plants. I determination of optimal conditions for linear dichroism measurements. Biochim Biophys Acta 679: 428–435. Hemelrijk PW, Kwa SLS, Van Grondelle R and Dekker JP (1992) Spectroscopic properties of LHCII, the main lightharvesting chlorophyll a/b protein complex from chloroplast membranes. Biochim Biophys Acta 1098: 159–166. Hiller RG and Breton J (1992) A linear dichroism study of photosynthetic pigment organisation in two fucoxanthincontaining algae. Biochim Biophys Acta 1102: 365–370. Hoff A (1974) The orientation of chlorophyll and bacteriochlorophyll molecules in an oriented lecitin multilayer. Photochem Photobiol 19: 51–57. Hoff A (1990) Triplet states in photosynthesis: linear dichroic optical difference spectra via magnetic resonance. In: Lin-

38 skens HF and Jackson JF (eds) Physical Methods in Plant Sciences, pp 23–57. Springer-Verlag, Berlin. Hofrichter J and Eaton WA (1976) Linear dichroism of biological chromophores. Ann Rev Biophys Bioeng 5: 511– 560. Hollósi M, Holly S, Majer Z, Kajtár J, Perzcel A and Fasman GD (1993) Analysis of the circular dichroism spectra of peptides guided by Fourier-transform infrared spectroscopy. 5th International Conference on Circular Dichroism, Book of Abstracts, pp 59–65. Homer-Dixon JA, Gantt E and Bruce D (1994) Pigment orientation changes accompanying the light state transition in Synechococcus sp. PCC 6301. Photosynth Res 40: 35– 44. Horton P, Ruban AV, Rees D, Pascal AA, Noctor G and Young AJ (1991) Control of the light-harvesting function of chloroplasts membranes by aggregation of the LHCII chlorophyll-protein complex. FEBS Lett 292: 1–4. Houssier C and Sauer K (1970) Circular dichroism and magnetic circular dichroism of the chlorophyll and protochlorophyll pigments. J Am Chem Soc 92: 779–791. Istokovics A, Lajkó F, Liker E, Barzda V, Simidjiev I, Garab G (1992) Inhibition of the light induced reversible structural rearrangements of the macrodomains and the phosphorylation of membranes by quinone antagonists. In: Murata N (ed) Research in Photosynthesis, Vol II, pp 631– 634. Kluwer Academic Publishers, Dordrecht, Boston, London. Johansson L and Lindblom G (1980) Orientation and mobility of molecules in membranes studied by polarized light spectroscopy. Q Rev Biophys 13: 63–118. Johnson WCJ (1985) Circular dichroism and its empirical application to biopolymers. Meth Biochem Anal 31: 61–163, Johnson WCJ (1990) Protein secondary structure and circular dichroism: A practical guide. Proteins: Struct Funct Gen 7: 205–214. Juszcak L, Geacintov NE, Zilinkas BA and Breton J (1987) Linear dichroism and orientation of pigments in phycobilisomes and their subunits In: Scheer H and Schneider S (eds) Photosynthetic Light-Harvesting Systems. Organization and Function, pp 281–292. Walter de Gruyter, Berlin. Keiderling TA, Pancoska P, Yasui SC, Urbanova M, Baumruk V, Gupta VP, Duko RK, Bour P and Huo D (1993) Application of vibrational circular dichroism to protein secondary strusture – empirical and theoretical comparisons. 5th International Conference on Circular Dichroism, Book of Abstracts, pp 73–77. Keller D and Bustamante C (1986a) Theory of the interaction of light with large inhomogeneous molecular aggregates. I. Absorption. J Chem Phys 84: 2961–2971. Keller D and Bustamante C (1986b) Theory of the interaction of light with large inhomogeneous molecular aggregates. II. Psi-type circular dichroism. J Chem Phys 84: 2972–2979. Keller D, Bustamante C, Maestre MF and Tinoco IJ (1985) Imaging of optically active biological structures by use of circularly polarized light. Proc Natl Acad Sci USA 82: 401– 405. Kim M and Bustamante C (1991) Differential polarization imaging. IV. Images in higher Born approximations. Biophys J 59: 1171–1182.

Garab Kim M, Ulibarri L, Keller D, Maestre MF and Bustamante C (1986) The psi-type circular dichroism of large molecular aggregates. III. Calculations. J Chem Phys 84: 2981–2989. Kim M, Keller D and Bustamante C (1987a) Differential polarization imaging. I. Theory. Biophys J 52: 911–927. Kim M, Ulibarri L and Bustamante C (1987b) Differential polarization imaging. II. Symmetry properties and calculations. Biophys J 52: 929–946. Kirmaier C, Holten D and Parson WW (1985) Picosecondphotodichroism studies of the transient states in Rhodopseudomonas sphaeroides reaction centers at 5 K. Effects of electron transfer on the six bacteriochlorin pigments. Biochim Biophys Acta 810: 49–61. Kiss L, Ganago AO and Garab G (1985) Quantitative method for studying orientation of transition dipoles in membrane vesicles of spherical symmetry. J Biochem Biophys Meth 11: 213–225. Kiss JG, Garab G, Tóth Z and Faludi-Dániel A (1986) The light-harvesting chlorophyll a/b protein acts as a torque aligning chloroplasts in a magnetic field, Photosynth Res 10: 217–222. Knox RS and Davidovich MA (1978) Theory of fluorescence polarization in magnetically oriented photosynthetic systems. Biophys J 24: 689–712. Kramer H and Amesz J (1982) Anisotropy of the emission and absorption bands of spinach chloroplasts measured by fluorescence polarization and polarized excitation spectra at low temperature. Biochim Biophys Acta 682: 201-207. Kruk J, Strzalka K and Leblanc M (1993) Linear dichroism and molecular orientation in Langmuir-Blodgett films of plastoquinones and quinone. Biochim Biophys Acta 1142: 6–10. Kühlbrandt W, Wang DN and Fujiyoshi Y (1994) Atomic model of plant light-harvesting complex by electron crystallography. Nature 367: 614–621. Kwa SLS, Eijckelhoff C, Grondelle R and Dekker JP (1994) Site-selection spectroscopy of the reaction center complex of photosystem II. I. Triplet-minus-singlet absorption difference: a search for a second exciton band of P-680. J Phys Chem 98: 7702–7711. Lehmann RP, Brunisholz RE and Zuber H (1994) Giant circular dichroism of chlorosomes from Chloroflexus aurantiacus treated with 1–hexanol and proteolytic enzymes. Photosynth Res 41: 165–173. Lewis JW, Goldbeck RA and Kliger DS (1992) Time resolved circular dichroism spectroscopy: Experiment, theory, and applications of biological systems. J Phys Chem 96: 5243– 5254. Lu X and Pearlstein RM (1993) Simulations of Prosthecochloris bacteriochlorophyll a protein optical spectra improved by parametric computer search. Photochem Photobiol 57: 86–91. Matsuura K, Hirota M, Shimada K and Mimuro M (1993) Spectral forms and orientation of bacteriochlorophylls c and a in chlorosomes of the green photosynthetic bacterium Chloroflexus aurantiacus. Photochem Photobiol 57: 92–97. Matthews BW, Fenna RE, Bolognesi MC, Schmid MF and Olson JM (1979) Structure of a bacteriochlorophyll a protein from the green photosynthetic bacterium Prosthecochloris aestuarii. J Mol Biol 131: 259–285.

LD and CD Matthijs HCP, Van der Staay GWM, Van Amerongen H, Van Grondelle R and Garab G (1989) Structural organisation of chlorophyll b in the prochlorophyte Prochlorothrix hollandica. Biochim Biophys Acta 975: 185–187. McDermott G, Prince SM, Freer AA, Hawthornthwaite-Lawless AM, Papiz MZ, Cogdell RJ and Isaacs NW (1995) Crystal structure of an integral membrane light-harvesting complex from photosynthetic bacteria. Nature 374: 517– 521. Mimuro M, Katoh T and Kawai H (1990) Spatial arrangement of pigments and their interaction in the fucoxanthin-chlorophyll a/c protein assembly (FCPA) isolated from the brown alga Dictyota dichotoma. Analysis by means of polarized spectroscopy. Biochim Biophys Acta 1015: 450–456. Moog RS, Kuki A, Fayer MD and Boxer SG (1984) Excitation transport and trapping in a synthetic chlorophyllide substituted hemoglobin: Orientation of the chlorophyll S1 transition dipole. Biochemistry 23: 1564–1571. Nabedryk E, Andrianambinintsoa S and Breton J (1984) Transmembrane orientation of in the thylakoid membrane and in light-harvesting complex. A polarized infrared spectroscopy study. Biochim Biophys Acta 765: 380–387. Nordén B (1980) Simple formulas for dichroism analysis. Orientation of solutes in stretched polymer matrices. J Chem Phys 72: 5032–5036. Nordén B, Kubista M and Kurucsev T (1992) Linear dichroism spectroscopy of nucleic acids. Q Rev Biophys 25: 51– 170. Osváth S, Meszéna G, Barzda V and Garab G (1994) Trapping magnetically oriented chloroplasts thylakoid membranes in gels for electric measurements. J Photochem Photobiol 26: 287–292. Papp E and Meszéna G (1982) Field concentration and temperature dependence of fluorescence polarization of magnetically oriented chloroplasts. Biophys J 39: 1–5. Parson WW and Warshel A (1987) Spectroscopic properties of photosynthetic reaction centers II. Application of the theory to Rhodopseudomonas viridis. J Am Chem Soc 109: 6152–6163. Paulsen H, Finkenzeller B and Kuhlein N (1993) Pigments induced folding of light-harvesting chlorophyll a/b-binding protein. Eur J Biochem 215: 809–816. Pearlstein RM (1982) Chlorophyll singlet excitons. In: Govindjee (ed) Photosynthesis, Vol. 1, pp 293–330. Academic Press, New York. Pearlstein RM (1987) Structure and exciton effects in photosynthesis. In: Amesz J (ed) Photosynthesis, pp 299–317. Elsevier, Amsterdam. Pearlstein RM (1991) Theoretical interpretation of antenna spectra. In: Scheer H (ed) Chlorophylls, pp 1047–1078. CRC Press, Boca Raton. Pearlstein RM (1992) Theory of the optical spectra of the bacteriochlorophyll a antenna protein trimer from Prosthecochloris aestuarii. Photosynth Res 31: 213–226. Perczel A, Hollósi M, Tusnády G and Fasman GD (1991) Convex constraint analysis: A natural deconvolution of circular dichroism curves of proteins. Protein Eng 4: 669– 679. Philipson KD and Sauer K (1973) Light-scattering effects on

39 the circular dichroism of chloroplasts. Biochemistry 12: 3454–3458. Pribic R, Van Stokkum IHM, Chapman D, Haris PI and Bloemendal M (1993) Protein secondary structure from Fourier transform infrared and/or circular dichroism spectra. Anal Biochem 214: 366–378. Provencher SW and Glöckner J (1981) Estimation of globular protein secondary structure from circular dichroism. Biochemistry 20: 33–37. Sauer K (1972) Circular dichroism and optical rotatory dispersion of photosynthetic organelles and their component pigments. Meth Enzymol 24: 206–218. Scheer H (1982) Phycobiliproteins: molecular aspects of photosynthetic antenna system. In: Fong FK (ed) Light Reaction Path of Photosynthesis, pp 7–45. Springer Verlag, Berlin. Scherer POJ, Fischer SF (1991) Interpretation of optical reaction center spectra. In: Scheer H (ed) Chlorophylls, pp 1079–1093. CRC Press, Boca Raton. Scherz A (1992) Exciton coupling calculations for optical spectra. Isr J Chem 32: 491–495. Scherz A and Parson WW (1986) Interactions of the bacteriochlorophylls in antenna bacteriochlorophyll-protein complexes of photosynthetic bacteria. Photosynth Res 9: 21– 30. Scherz A, Rosenbach-Belkin V, Fisher JRE (1991) Chlorophyll aggregates in aqueous solutions. In: Scheer H (ed) Chlorophylls, pp 238–268. CRC Press, Boca Raton. Shindo Y (1985) On the problems of CD spectropolarimeter (IV) Artifacts due to the light scattering by small particles. Appl Spectr 39: 713–715. Shindo Y, Nakagawa M and Ohmi Y (1985) On the problems of CD spectropolarimeters. II. Artifacts in CD spectrometers. Appl Spectr 39: 860–868. Shotton D and White N (1989) Confocal scanning microscopy: three-dimensional biological imaging. Trends Biochem Sci 14: 435–439. Shubin VV, Tsuprun VL, Bezsmertnaya IN and Karapetyan NV (1993) Trimeric forms of the Photosystem I reaction center complex pre-exist in the membranes of the cyanobacterium Spirulina platensis. FEES Lett 334: 79–82. Steinberg IZ (1978) Circular polarized luminescence. Meth Enzymol 49: 179–199. Sundqvist C, Ryberg H, Böddi B and Láng F (1980) Spectral properties of a long-wavelength absorbing form of protochlorophyll in seeds of Cyclanthera explodens. Physiol Plant 48: 297–301. Sutherland JC (1978) The magnetic optical activity of porphyrins. In: The Porphyrins. Vol III, pp 225–247. Academic Press, New York. Sutherland JC and Holmquist B (1980) Magnetic circular dichroism of biological molecules. Ann Rev Biophys Bioeng 9: 293–326. Szitó T, Garab G, Mustárdy L, Kiss JG and Faludi-Dániel Á (1984) Increasing fluctuation in orientation of pigmentprotein complexes within photosynthetic membranes treated with linolenic acid. Photobiochem Photobiophys 8: 239–249. Szitó T, Zimányi L and Faludi-Dániel A (1985) Fluorescence polarization spectra of granal and stromal membranes tre-

40 ated with linolenic acid. Orientation of the Photosystem I core complex within the membrane. Biochim Biophys Acta 808: 428–436. Tinoco IJ (1962) Theoretical aspects of optical activity. Part Two: Polymers. Adv Chem Phys 4: 113–160. Tinoco IJ and Williams ALJ (1984) Differential absorption and differential scattering of circularly polarized light: Applications to biological macromolecules. Ann Rev Phys Chem 35: 329–355. Tinoco IJ, Mickols W, Maestre MF and Bustamante C (1987) Absorption, scatering, and imaging of biomolecular structures with polarized light. Ann Rev Biophys Biophys Chem 16: 319–349. Tlalka M and Gabrys H (1993) Influence of calcium on bluelight-induced chloroplast movement in Lemna trisulca L. Planta 189: 491–498. Van Amerongen H and Struve WS (1995) Polarized optical spectroscopy of chromoproteins. Meth Enzymol 246: 259– 283. Van Amerongen H, Vasmel H and Van Grondelle R (1988) Linear dichroism of chlorosomes from Chloroflexus aurantiacus in compressed gels and electric fields. Biophys J 54: 65–76. Van Amerongen H, Kwa SLS, Van Bolhuis BM and Van Grondelle R (1994) Polarized fluorescence and absorption of macroscopically aligned light harvesting complex II. Biophys J 67: 837–847. Van der Vos R, Van Leeuwen PJ, Braun P and Hoff A (1992) Analysis of the optical absorbance spectra of D1–D2–cytochrome b-559 complexes by absorbance-detected magnetic resonance. Structural properties of P680. Biochim Biophys Acta 1140: 184–198. Van Dorssen RJ, Vasmel H and Amesz J (1985) Antenna organization and energy transfer in membranes of Heliobacterium chlorum. Biochim Biophys Acta 809: 199–203. Van Grondelle R, Dekker JP, Gillbro T and Sundström V (1994) Energy transfer and trapping in photosynthesis. Biochim Biophys Acta 1187: 1–65. Van Gurp M, Van Ginkel G and Levine YK (1988) Orientational properties of biological pigments in ordered systems studied with polarized light: photosynthetic pigmentprotein complexes in membranes. J Theor Biol 131: 333– 349. Van Gurp M, Van der Heide U, Verhagen J, Piters T, Van Ginkel G and Levine YK (1989) Spectroscopic and orientational properties of chlorophyll a and chlorophyll b in lipid membranes. Photochem Photobiol 49: 663–672. Van Haeringen B, Dekker JP, Bloemendal M, Rögner M,

Garab Van Grondelle R and Van Amerongen H (1994) The simultaneous measurement of electric birefringence and dichroism. A study on Photosystem 1 particles. Biophys J 67: 411–417. Van Metter RL (1977) Excitation energy transfer in the lightharvesting chlorophyll a/b protein. Biochim Biophys Acta 462: 642–658. Vasmel H, Van Dorssen RJ, de Vos GJ and Amesz J (1986) Pigment organization and energy transfer in the green photosynthetic bacterium Chloroflexus aurantiacus. I. The cytoplasmic membrane. Photosynth Res 7: 281–294. Verméglio A, Breton J and Mathis P (1976) Trapping at low temperature of oriented chloroplasts: Application to the study of antenna pigments and of the trap of Photosystem1. J Supramol Struct 5: 109–117. Verméglio A, Garcia D, Breton J (1990) Cytochrome arrangement in reaction centers of different species of photosynthetic bacteria. In: Michel-Beyerle M-E (ed) Reaction Centers of Photosynthetic Bacteria, pp 19–29. Springer-Verlag, Berlin. Visschers RW, Chang MC, Van Mourik F, Parkes-Loach PS, Heller BA, Loach PA, Van Grondelle R (1991) Fluorescence polarization and low-temperature absorption spectroscopy of a subunit form of light-harvesting complex I from purple photosynthetic bacteria. Biochemistry 30: 5734–5742. Woody RW (1985) Circular dichroism of peptides. In: Hraby VJ (ed) The Peptides, pp 15–114. Academic Press, New York. Wu K, Geng L, Joseph MJ and McGown LB (1993) Steadystate and lifetime-resolved fluorescence-detected circular dichroism using a modified phase-modulation spectrofluorometer. Anal Chem 65: 2339–2345. Xie X and Simon JD (1991) A picosecond circular dichroism study of photosynthetic reaction centers from Rhodobacter sphaeroides. Biochim Biophys Acta 1057: 131–139. Zhou W, LoBrutto R, Lin S and Blankenship RE (1994) Redox effects on the bacteriochlorophyll a-containing Fenna–Matthews–Olson protein from Chlorobium tepidum. Photosynth Res 41: 89–96. Zuber H, Brunisholz RE (1991) Structure and function of antenna polypeptides and chlorophyll-protein complexes: principles and variability. In: Scheer H (ed) Chlorophylls, pp 627–703. CRC Press, Boca Raton. Zucchelli G, Dainese P, Jennings RC, Breton J, Garlaschi FM and Bassi R (1994) Gaussian decomposition of absorption and linear dichroism spectra of outer antenna complexes of photosystem II. Biochemistry 33: 8982–8990.

Chapter 3 Fluorescence Kenneth Sauer* and Martin Debreczeny1 Department of Chemistry and Structural Biology Division, Lawrence Berkele.y Laboratory, University of California, Berkeley, Ca 94720–1460, USA; 1 Chemistry Division, Argonne National Laboratory, Argonne, IL 60439, USA

Summary I. Introduction A. Electronic Excitation and Emission B. Excited State Formation, Relaxation and Decay II. Steady-State Fluorescence A. Fluorescence Spectrophotometer B. Excitation Spectra C. Emission Spectra D. Light Scattering Interference E. Polarization/Depolarization F. VariableFluorescence 1. Photochemical Trapping Competes with Fluorescence 2. Electric Fields Influence Fluorescence Associated with Photosynthetic Materials 3. Temperature Dependence of Fluorescence III. Time-Resolved Fluorescence A. Experimental Considerations 1. Time-Correlated Single Photon Counting (TCSPC) 2. Fluorescence Upconversion 3. Deconvolution 4. Exciton Annihilation B. Isotropic Time-Resolved Fluorescence C. Anisotropic Time-Resolved Fluorescence IV. Conclusion Acknowledgements References

41 42 42 42 45 45 46 47 48 49 50 50 51 52 52 52 52 53 54 54 55 57 59 59 60

Summary

Fluorescence emission is a direct reflection of the properties of excited electronic states of molecules as they return radiatively to the ground electronic state. Fluorescence provides information about the (1) energy of the emitting state relative to the ground state, (2) lifetime of the excited state, (3) orientation of transition dipole moments and (4) symmetry properties of the ground and excited states. Fluorescence is especially valuable as a probe of photosynthetic systems, because it constitutes a sensitive competitive path to photochemical energy conversion, resulting in fluorescence quenching. Fluorescence spectra provide knowledge of the energy levels of different pigment pools in the light-harvesting antenna and reaction center complexes. Steady-state and time-resolved depolarization studies reflect the rapid excitation transfer processes that occur within these multi-pigment arrays in photosynthetic membranes. Using theoretical formulations, such as the Förster inductive resonance transfer mechanism, these *Correspondence: Fax: 1-510-4866059; E-mail: [email protected]

41 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 41–61. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

42

Kenneth Sauer and Martin Debreczeny

transfer rates can be related directly to molecular geometries derived from X-ray crystallography and to fundamental spectroscopic properties of the molecules involved. The consequences of electric fields formed by primary charge separation across photosynthetic membranes can be seen in the influence of an applied electric field on the fluorescence intensity and relaxation kinetics. Much of our current knowledge of the primary processes of photosynthetic energy conversion has derived from fluorescence measurements. Abbreviations: Chl – chlorophyll; FWHM – full-width half-maximum; ic – internal conversion; IRF – instrument response function; is – intersystem crossing; LHC – light-harvesting complex; PC – phycocyanin; TCSPC – time-correlated single-photon counting

I. Introduction Fluorescence emission derives from excited electronic states of molecules. In photosynthesis the molecules of interest are associated with the antenna and the reaction centers. Fluorescence from chlorophylls, bacteriochlorophylls and phycobiliproteins in whole organisms or in preparations of active membrane fragments or sub-complexes provides information about the roles of these molecules in primary photosynthetic energy conversion. Because the excited electronic states exist between the initial absorption of photons and the ultimate charge separation that completes the conversion of light energy into chemical energy, monitoring the fluorescence provides direct evidence of the mechanism and dynamics of the primary events in photosynthesis.

A. Electronic Excitation and Emission First we will look at the sequence of events involved in the evolution of electronic excitation and relaxation (decay) that are monitored using fluorescence. We will begin with a summary of the “intrinsic” properties of the excited electronic states of an isolated pigment molecule, like chlorophyll or bacteriochlorophyll. Then we will examine the consequences of collecting chromophores in the pigment-protein complexes that are ubiquitous in photosynthetic membranes and their associated components. The resulting delocalization of the excitation is essential to the function of the antenna in collecting light. Finally, we consider the influence of the reaction centers or

photochemical traps that extract the excitation energy for conversion into chemical potential. The time scale of this sequence of events ranges from a few femtoseconds required for transforming a ground electronic state into an excited state, picoseconds for collecting the excitation at the reaction centers, and picoseconds to nanoseconds in the reaction centers to accomplish the charge separation steps and occasionally the reversal of these steps to produce delayed fluorescence on a still longer time scale. We see from this scenario that the molecule whose fluorescence is detected may have been excited by photon absorption directly, may have received its excitation energy by transfer from other pigment molecules or it may be excited by the return of excitation from the traps. Each of these paths has a distinctive signature in the time dependence, wavelength dependence, depolarization, etc. of the fluorescence. In many cases the steps in the path can be elucidated using time-resolved measurements; however, steady-state measurements, which average the time-dependent behavior, also provide useful guides to investigating and interpreting the excited state relaxation. Both of these approaches will be explored in this chapter.

B. Excited State Formation, Relaxation and Decay The absorption of electromagnetic radiation by a molecule is properly described using quantum mechanics. Useful descriptions of how to characterize this process are given in several monographs (Cantor and Schimmel, 1980; Lakowicz,

Fluorescence

1983; Struve, 1989). For large chromophoric molecules like the photosynthetic pigments in protein environments, the course of events is extremely complex, and it cannot be described precisely. It is useful, therefore, to separate the overall process into a series of stages or influences that can be considered separately. One such sequence is illustrated, in part, in Fig. 1. 1. The ground state G of the molecule that is sampled by the incident radiation is a thermally equilibrated ensemble of configurations. 2. The absorption spectrum of a molecule reflects the energy (frequency) dependence of the probability of achieving a particular excited electronic state configuration, etc. The photon energy must correspond to the difference in energy between the initial and final states of the molecule, and the transition dipole moment describes the quantum mechanical coupling between the ground and excited states.

43 3. Excited electronic states of photosynthetic pigments invariably involve delocalized trons associated with the conjugated or aromatic bonding systems in chlorophylls, openchain tetrapyrroles or carotenoid polyenes. Because the mass of the electron is small compared with the nuclear mass of the atoms in the molecule, the redistribution of the electron in the excited state orbital occurs rapidly in comparison with nuclear motion. Thus, the excited electronic state of the molecule is produced with a nuclear configuration that is initially the same as that of the ground electronic state at the time of arrival of the photon (Franck–Condon Principle). 4. Inhomogeneous broadening of the absorption bands results from the large variety of microstates that is present in the initial thermal distribution of ground state configurations, together with the equally large variety of excited state configurations that can result from the absorption of photons of a particular energy by a large population of molecules. 5. Relaxation of the nuclear configuration in the excited electronic state is a consequence of the change in charge distribution produced by the transfer of the electron from the ground state orbital to the excited state orbital. This relaxation results from an exchange of energy among internal modes of motion of the chromophore, as well as interchange with other molecules in the surroundings – especially the protein matrix and other nearby chromophores. A component of this relaxation process, which is typically complete within a few picoseconds in photosynthetic pigments in condensed media, is the decay or transfer by internal conversion, ic, from higher energy excited electronic states etc. to the lowest energy excited state having the same spin multiplicity (typically singlet) as the ground state. 6. Thermal equilibration occurs as the excited state configurations attain the distribution corresponding to the temperature of the environment. This occurs typically within a few picoseconds of photon absorption and, except for molecules such as carotenoids with very short excited state lifetimes, is complete prior to most of the fluorescence emission. Where sig-

44 nificant emission occurs prior to excited state thermalization, this is detectable using a comparison of absorption or excitation spectra with fluorescence emission spectra using relations derived by Stepanov (1957). 7. Several distinct fates are possible for the thermally relaxed excited electronic state. a) Fluorescence – radiative decay from the excited state back to the ground state G. The probability for this to occur is governed by the same quantum mechanical principles that are involved in the absorption of radiation by the ground state. The spectrum of fluorescence is typically red-shifted (Stokes shift) relative to the longest wavelength absorption band, because the excited electronic state of the molecule has an altered (relaxed) nuclear configuration relative to that of the ground state. This results in a lower excited state energy and a slightly higher ground state energy relative to those involved in absorption and, because the Franck–Condon Principle applies to fluorescence also, this provides the basis for the red-shift in the fluorescence spectrum. b) Internal conversion (ic) – radiationless transition to the ground electronic state manifold. In this case the excited state energy is dissipated thermally by relaxation to the surrounding medium. c) Intersystem crossing (is) – singlet-to-triplet conversion. The resulting change in electron spin multiplicity is formally forbidden quantum mechanically, but for complex molecules this can nevertheless be a major route for decay of excited singlet states. d) Excitation transfer – migration of excitation from donor (D) to acceptor (A) molecules. In situations commonly encountered in photosynthetic materials this occurs by an inductive resonance process described initially by Förster (1948, 1967). The inductive resonance mechanism results in transfer among molecules which may be either the same or different chemically. The probability or rate of transfer depends on (1) the spectral energy overlap between D and A, (2) the inverse sixth power of the distance between D and A, (3) their relative orientation, and (4) the intrinsic fluorescence lifetime of D.

Kenneth Sauer and Martin Debreczeny The important range of spatial distances involved is 1.5 to 10 nm, corresponding to transfer rates of 1 to 1000 per nanosecond. e) Quenching – any process that decreases the excited state lifetime from that of the “isolated” molecule. Trapping of excitation in reaction centers resulting in productive charge separation is an example of photochemical quenching. Other types of quenching can occur from the proximity or addition of quencher molecules Q, which may be paramagnetic species, heavy atoms or ions, excitation transfer acceptors [section (d), above] or pigment aggregates. To the extent that these compete with natural relaxation processes such as fluorescence, they serve to decrease the fluorescence yield, thereby providing an indirect monitor of the quenching. 8. Polarization is a consequence of the vectorial nature of the absorption and emission of radiation. The transition dipole moments associated with these processes have both a magnitude (related to the allowedness or the “oscillator strength” of the transition) and a direction relative to molecular axes. For a molecule with a fixed orientation in space, the probability that it will absorb light depends on the direction of propagation of the incident radiation as well as on the relative orientation of the oscillating electric field vector. Subsequent emission from the excited state is polarized in a manner that depends on the transition moment vector for fluorescence and on the direction of observation relative to that of excitation. Depolarization of the fluorescence can result from: a) Intramolecular relaxation ic from higher excited electronic states having absorption transition moments oriented differently from that of fluorescence. b) Rotation of the molecule during the excited state lifetime. c) Excitation transfer to a differently oriented molecule, which then emits the fluorescence. 9. Fluorescence lifetimes are determined by the intrinsic fluorescence lifetime of the molecule as modified (decreased) by competing processes. The intrinsic lifetime (15–20 ns for

Fluorescence most chlorophylls) is governed by the transition dipole moment for spontaneous emission (Einstein A coefficient). For molecules having similar ground- and excited-state nuclear configurations, the intrinsic lifetime can often be calculated from the absorption properties of the molecule. 10. The fluorescence yield describes the fraction of the excited state population that results in fluorescence. Because fluorescence is decreased by competing excited-state relaxation processes, the reduction of fluorescence yield is directly proportional to the decrease in the fluorescence lifetime relative to the intrinsic lifetime (the lifetime in the absence of competing processes). 11. Fluorescence relaxation for simple molecules in dilute solutions and at low incident light intensities is kinetically first-order in the excited state concentration or population. This results in a simple single-exponential decay. Modifications of this simple kinetic behavior can result from a) inhomogeneous excited-state populations, typically reflecting molecules in different local environments giving rise to components with different relaxation rates, b) high local concentrations of excited states that result in excitation annihilation (bimolecular), often a consequence of high intensities used in pulsed-laser experiments, c) significant depletion of the ground-state population accompanying repeated high-intensity pulsed excitation, d) long-lived quenching species (triplets, oxidized or reduced reaction centers, etc.). In photosynthetic membranes or pigment complexes the fluorescence decay is never found to be simple single-exponential. It is, in fact, this complexity that often provides insight into the details of the excitation transfer and trapping processes associated with photosynthesis, as we shall explore in the remainder of this chapter. 12. Phosphorescence is emission associated with relaxation from the lowest energy state of (typically) a triplet manifold of states to the ground state. The spectrum of phosphorescence emission from a particular molecule is different from that of the fluorescence. Phosphorescence often appears at longer wave-

45

lengths than fluorescence and occurs with a much longer relaxation time (milliseconds to seconds), because it involves a process that is formally quantum mechanically forbidden. Although phosphorescence from Chl a has been reported, it has not played a significant role in photosynthesis research.

II. Steady-State Fluorescence

A. Fluorescence Spectrophotometer Fluorescence is commonly measured using a fluorescence spectrophotometer. Many commercial instruments are available, and a representative configuration is shown schematically in Fig. 2. The basic components are (1) a source of exciting light, (2) a sample containing a fluorescent ma-

46 terial, (3) a device for detecting the fluorescence intensity in a particular direction, and (4) electronic components for displaying, recording or storing output information in a form that is accessible using computer software. The configuration shown in Fig. 2 includes monochromators in both the excitation and emission beams. For a general purpose instrument it is desirable to be able to scan either monochromator to record excitation or emission spectra. Fluorescence is emitted in all directions, but with an intensity distribution that depends strongly on the angle with respect to the excitation beam, the polarization of the light and intrinsic properties of the fluorescing species. The detection system is typically arranged to observe fluorescence emitted at an angle such as 90° to the incident intensity. This minimizes interference from the transmitted light propagated in the forward direction, which is much more intense than the fluorescence. Polarizers can be inserted into the excitation or emission beam. A polarizer defines the orientation of the electric vector of the transmitted light in the plane perpendicular to its direction of propagation. The detector, which may be a vacuum-tube photomultiplier or a solid state photodiode, must be sensitive at the wavelengths of the fluorescence. Two important modes of treating the output signal from the detector are (1) analog detection of the output voltage or current, and (2) single-photon counting. The latter is particularly effective in suppressing the contributions of low-level noise (dark current) contributed by the detector electronics, thereby enabling the detection of weak fluorescence signals. For special purposes one or more of the design components or features shown schematically in Fig. 2 can be varied. Some of these modifications are of considerable importance for photosynthesis studies, and they will be mentioned at appropriate places in this chapter.

B. Excitation Spectra A fluorescence excitation spectrum displays the relative efficiency of different wavelengths of exciting light in generating fluorescence. Light that is incident on a homogeneous, clear sample with an intensity (see Fig. 2) is partially transmitted, and partially absorbed, (We will consider presently the complications introduced

Kenneth Sauer and Martin Debreczeny by samples that scatter a significant portion of the incident light.) Light which is absorbed by the sample may result in fluorescence, phosphorescence, non-radiative return to the ground state or photochemistry. The fluorescence quantum yield, is the fraction of light absorbed that is emitted as fluorescence. Thus,

for a homogeneous (non-scattering) sample. For samples that are sufficiently dilute and illuminated at low intensities, both and are linearly dependent on the incident light intensity; under these conditions the ratio is independent of light intensity. However, the quantum yield will, in general, depend on the wavelength of the incident light. Comparison of the fluorescence excitation spectrum with the absorption spectrum provides direct evidence of the excitation-wavelength dependence of the quantum yield. An example of the usefulness of this property is seen in Fig. 3. Photosynthetic materials such as chloroplast thylakoids that contain several different light-harvesting pigments often exhibit efficient excitation transfer from short wavelength-absorbing pigments to Chl a, which has a lower-energy excited state and hence a longer wavelength absorption band. A portion of the excitation arriving at Chl a is then emitted as fluorescence. If excitation transfer from the accessory pigments in the native photosynthetic membranes is highly efficient, then the fluorescence excitation spectrum will be superimposable on the absorption spectrum, indicating that the quantum yield of fluorescence is wavelength independent. By contrast, when the pigments are extracted into organic solvents or when the attachment of phycobilisomes or chlorosomes to the membranes is disrupted by cell breakage excitation transfer from the accessory pigments is no longer effective. Thus, wavelengths absorbed by the accessory pigments, mainly Chl b and carotenoids in the case of spinach chloroplast thylakoids (Fig. 3), do not lead to Chl a fluorescence, and the fluorescence excitation spectrum resembles only that portion of the absorption owing to the Chl a itself, as is seen in the excitation and absorption spectra of the

47

Fluorescence

C. Emission Spectra

extracted pigments (solid curves, especially in the region between 450 and 500 nm in Fig. 3). This approach has been used to explore the relative efficiencies of different accessory pigments in transferring excitation to chlorophyll in a variety of in vivo situations.

The fluorescence emission spectrum of a single substance in solution reflects radiative transitions typically from the lowest excited electronic state to the ground state. Because of relaxation of the nuclear configuration in the excited state prior to emission, the fluorescence typically undergoes a Stokes shift to longer wavelength, amounting to 4 to 7nm in the case of chlorophylls at room temperature, as seen in Fig. 3. To the extent that the transitions involving the lowest excited state show vibrational sub-structure (always incompletely resolved for large chromophores in condensed media), the fluorescence emission spectrum and the absorption spectrum in the long wavelength region exhibit a mirror-image relation to one another on a scale where the spectra are plotted as a function of the frequency (energy) of the radiation. Fig. 3 shows examples of such behavior for thylakoids and for the pigment extract, comparing the emission spectra at the righthand side of Fig. 3(a) with the absorption spectra in the long-wavelength region of Fig. 3(b). Failure of this relation may indicate situations where there is (1) significant overlap of more than one electronic transition in the absorption spectrum, as is the case for the and transitions of Chl a or Chl b, (2) incomplete relaxation or thermalization of the excited state prior to emission, or (3) excitation transfer among identical molecules in different local environments that produce different spectral shifts, or among chromophores that are chemically distinct but have overlapping absorption spectra. A particularly sensitive test for the presence of any of these contributions is achieved using the method devised by Stepanov (1957), and which has been applied to a study of excitation equilibration in Photosystem II (H. Dau and K. Sauer, Biochim Biophys Acta, submitted). Photosynthetic membranes also contain nonfluorescing pigments. Carotenoids, for example, absorb strongly in the visible and near-UV region of the spectrum but have almost undetectable fluorescence. The cause of this behavior is a lowlying electronic state that cannot be populated by direct light absorption from the ground state but that can be reached efficiently by intersystem crossing from a higher energy excited state. Such a state which quenches the fluorescence of the

48 molecule may, in general, be a paramagnetic triplet state or may have symmetry elements that cause it not to couple radiatively with the ground state. Nevertheless, molecules such as carotenoids can transfer excitation to “sensitize” the fluorescence of nearby chlorophyll molecules in photosynthetic membranes. The fluorescence of mixtures of chromophores of different chemical types is, in general, an additive superposition of the contributions of each of the molecules involved, if the chromophores do not interact with one another and if the sample is sufficiently dilute to avoid optical distortions. However, because the fluorescence of each molecular species is also determined by its distinctive absorption properties, the measured emission spectrum from a mixture of fluorophores depends critically on the excitation wavelength that is selected. Useful assays of mixtures of photosynthetic pigments have been devised in this way. For example, chlorophyll b can be detected at very low levels in the presence of a much larger concentration of chlorophyll a in a pigment extract using the facts that (1) absorption (excitation) by Chl b at wavelengths between 450 and 460 nm occurs in a region of the spectrum where Chl a absorbs hardly at all, and (2) emission from Chl b between 654 and 650 nm occurs in a region where Chl a emission is small. (Boardman and Thorne, 1971) For this assay to work, it is obviously necessary to avoid higher pigment concentrations where excitation transfer from chlorophyll b to chlorophyll a would quench the fluorescence of the former. Self-absorption of the fluorescence may occur in the case of strongly absorbing samples. The fluorescence spectrum undergoes distortion owing to fluorescence re-absorption by the sample itself, primarily in the wavelength region where there is the greatest overlap between the absorption and the fluorescence emission of the sample. As a consequence, the fluorescence signal is suppressed at these wavelengths, and the apparent fluorescence emission maximum is shifted to longer wavelengths where the spectral overlap is less. Although conditions differ for different experimental set-ups, a good rule of thumb is to use samples with absorbance (optical density) less than 0.05 at both the exciting wavelength and in the region of maximum spectral overlap.

Kenneth Sauer and Martin Debreczeny It is usually impossible to avoid self-absorption distortions of the fluorescence spectra for intact leaves or even for chloroplast suspensions, because the local concentrations of pigments are quite high even for diluted suspensions. For samples such as heavily pigmented leaves, where very little light passes all the way through the sample, fluorescence can nevertheless be detected by using a front-face illumination geometry, either by moving the detector or by turning the sample so that fluorescence emitted from the directly illuminated surface is detected. If the light at the excitation wavelength is totally absorbed by the sample, then the fluorescence observed is essentially independent of the sample concentration (or leaf thickness); however, self-absorption effects are still present in the emission spectrum.

D. Light Scattering Interference Samples that are inhomogeneous (in terms of their refractive index) on a scale of the order of the wavelength of light and longer are subject to light scattering. This is a prominent property of plant leaves, cell suspensions and, to a lesser extent, of suspensions of sub-membrane complexes such as reaction centers and antenna pigmentproteins. As a consequence, some or even most of the incident light is redirected into all directions, much the same as fluorescence, although the detailed dependence on angle is different. Scattering can be elastic (Rayleigh scattering), with no change in wavelength of the incident light, or inelastic, where the shift is both to longer (Stokes scattering) and to shorter (anti-Stokes scattering) wavelengths. The Raman effect is an important form of inelastic scattering that results from coupling to vibrational transitions in the chromophore or in the surrounding matrix. It is not necessary for the material to absorb at the wavelength of incident light for either elastic or inelastic scattering to occur. The scattered light is readily detected by sensitive fluorescence spectrophotometers, is usually strongly polarized, and can serve as a strong source of interference. Rayleigh scattering is sufficiently strong in any sample that it is impossible using steady-state methods to measure the fluorescence at the same wavelength as the exciting light, even for homogeneous solutions that are “dust-free”. However,

Fluorescence when the scattering is not too intense, the fluorescence excitation and emission wavelengths can be within a few nanometers of one another, depending on the quality of the monochromators used. Double- and even triple-monochromators, often supplemented with blocking optical filters, are required to suppress the transmission of stray light at wavelengths away from the one selected for excitation and to prevent the exciting light wavelength from reaching the detector. Another way to separate light-scattering from fluorescence signals is to use time-resolved measurements; light-scattering occurs essentially instantaneously, whereas fluorescence relaxation is measurably slower.

E. Polarization/Depolarization Because of the vectorial nature of the interaction between electromagnetic radiation and the molecular transition dipole moments for absorption and emission, fluorescence is typically polarized. The polarization occurs at two stages. (1) During the excitation process of an isotropic sample of randomly oriented chromophores, a sub-population of molecules is “photo-selected” for excitation by the projection of the electric vector of the incident radiation on the transition moment of each absorbing molecule. The excited state population is therefore anisotropic, having a vectorial character. (2) Each excited molecule, in turn, emits radiation that is polarized parallel to its fluorescence transition dipole moment, but propagation of the fluorescence occurs predominantly in directions that are not along the transition moment vector. Between the excitation and emission processes, events such as internal conversion, chromophore rotation or excitation transfer [see Section I.B.8] may occur in the sample. Each of these processes leads either to an alteration of the polarization direction or to depolarization (randomization of orientation). Detailed analyses of the consequences of each of the effects noted above have been published. As indicated in Fig. 2, optical polarizers can be inserted into the excitation and/or the emission beam to test the extent of polarization and its dependence on wavelength for a particular sample. For purposes of illustration, let us suppose that the excitation and emission propagation di-

49 rections and in Fig. 2) are at 90° to one another and lie in the horizontal plane. Unpolarized light incident on the sample has its electric vector oscillating in the plane perpendicular to the propagation direction. Even in the absence of inserted polarizers, photoselection will occur in the sample owing to the fact that the oscillating electric field does not have a component in the direction of propagation of the light and, hence, will be biased against exciting those molecules whose absorption transition moments happen to be oriented in that direction. For this reason the population of excited molecules in a sample illuminated from one direction is always anisotropic. Addition of a polarizer to the excitation beam selects a polarization direction for the electric vector in a plane that includes the direction of propagation and the polarizer transmission axis. Absorption of this plane-polarized radiation by an isotropic sample in turn modifies the anisotropy of the excited-state population initially produced. It is useful and sufficient for our purposes to consider only two orientations of the polarizer axis, one producing light polarized in the vertical plane and the other in the horizontal plane (obtained by rotating the polarizer 90° about an axis parallel to the propagation direction). Depending on which orientation of the polarizer is chosen, a different subset of the chromophores will be excited, determined by the projection of the oscillating electric field of the exciting light on the absorption transition dipoles of the individual chromophores. Depending on which subset is excited, the projection of the ensemble of molecular vectors is different in the direction of propagation of the fluorescence emission. This can be readily detected by inserting an analyzing polarizer into the fluorescence beam and orienting it so as to transmit light polarized in either the vertical or in the horizontal plane. If the polarizer in the excitation beam is oriented vertically, the degree of polarization of the sample fluorescence can be readily determined from the relative intensities of emitted light detected when the analyzer is oriented eithe parallel vertical) or perpendicular horizontal) to the direction of polarization of the exciting beam. (Because each of the instrument components in a fluorescence spectrophotometer, from the light

50 source through to the detector, introduces polarization effects quite apart from those of the sample, it is necessary to correct for this instrument polarization function, including its dependence on the wavelengths of excitation and emission. A straightforward way of accomplishing this correction is described by Houssier and Sauer (1969) and by Lakowicz (1983).) For purposes of interpretation of polarization measurements for a particular sample, the measured (and corrected) polarized fluorescence intensities are combined to give a value for the polarization anisotropy, A, where

Each of the quantities involved in Eq. (2) is dependent on the wavelengths of excitation and emission. (The reader should be aware that an alternative description of the polarization anisotropy is sometimes seen, especially in the earlier literature, where the factor of 2 in the denominator is not included.) Analyses of the relation between polarization/depolarization properties of particular samples have been described in detail in the literature. (Van Amerongen and Struve, 1995) Some relevant properties that influence the interpretation are (1) whether the sample is truly isotropic or whether there is partial or complete ordering of the directions of the transition moments, as in a single crystal or in a sample of oriented membrane fragments, (2) rotation of the chromophore during the excited state lifetime, and (3) excitation transfer among the chromophores within the sample. If either (2) or (3) is extensive for an isotropic sample, this can lead to complete depolarization of the fluorescence and an anisotropy value of zero. For an isotropic sample where the individual molecules are (1) effectively fixed in their orientation and (2) unable to transfer excitation to their neighbors during the excited state lifetime, the anisotropy value can range between + 2/5, when the absorption and emission transition dipoles are parallel to one another, to – 1/3, when they are mutually perpendicular. Intermediate values of the polarization anisotropy can be interpreted in terms of the angle between the ab-

Kenneth Sauer and Martin Debreczeny sorption and emission transition dipoles. If excitation transfer occurs among the molecules in the sample, the polarization anisotropy can be interpreted in terms of the relative orientation of the absorption transition moment of the chromophore initially excited, D, and the fluorescence transition moment of the acceptor chromophore, A, that emits the fluorescence. It is important for purposes of this analysis that the system of chromophores does not physically rotate to a significant extent during the excited state lifetime. For chromophores with fluorescence lifetimes of a few nanoseconds in aqueous solution at room temperature, the effective molecular weight should be in excess of 100 kDa to avoid serious contributions from rotational diffusion (Rigler and Ehrenberg, 1976). Polarization anisotropy measurements have been used for a variety of purposes in connection with photosynthesis studies. The relative directions of the absorption transition moments for transitions to different electronic excited states of pigment molecules like chlorophyll, bacteriochlorophyll or protochlorophyll have been derived from experimental measurements and compared with theoretical deductions (Gouterman and Stryer, 1962). Studies of fluorescence depolarization of pigment molecules in solution as a function of concentration have provided evidence that the range of excitation transfer for chlorophylls, for example, extends to 60 to 100 Å (Knox, 1975). For multiple chromophores present in pigment proteins or photosynthetic membrane preparations, the extent of depolarization reflects both the relative orientations of the chromophores and their ability to transfer excitation. Such analyses are aided by knowledge of the structural arrangement of the chromophores where such information is available from X-ray crystallography, and from time resolved measurements of the relaxation of the fluorescence anisotropy. We shall describe an example of such a study that has been applied to C-phycocyanin (PC) in Section III.

F. Variable Fluorescence 1. Photochemical Trapping Competes with Fluorescence An important application to photosynthetic sys-

Fluorescence tems arises from the competition between fluorescence and photochemical trapping in reaction centers. As a consequence of this competition, fluorescence yields are complementary to the yields of productive electron transport initiated by the reaction centers (Latimer et al., 1956; Govindjee et al., 1986; Krause and Weis, 1991; Dau, 1994). In Photosystem II and in purple bacteria, closing the reaction centers by strong illumination, using inhibitors of primary electron transport or adding reducing agents that keep the endogenous secondary electron acceptors in the reduced state, increases the yield of fluorescence from 3 to 5 fold. The maximum fluorescence yield is still less than 10%, indicating that there are other important alternative paths of excited state relaxation in these closed or blocked preparations. Special instrumentation has been developed to measure variable fluorescence yields and the associated kinetics of fluorescence induction. Depending on the light intensity used to close the traps and on the conditions of inhibition, the process occurs primarily on the nanoseconds to seconds time scale (Mauzerall, 1972). This is the time required for electron acceptors close to the reaction centers to become reduced or electron donor pools to become exhausted. In plants, whole cells, or intact chloroplasts, longer term adjustments in the redox levels of the donor and acceptor pools occur, and slower changes can be readily seen over intervals of minutes to hours (Kautsky and Hirsch, 1931; Büchel and Wilhelm, 1993). For reasons that are not yet well understood, Photosystem 1 makes little or no contribution to the variable fluorescence signals (Briantais et al., 1986). Instrument requirements for these studies differ significantly depending on the time scale involved. Because light plays the dual role of stimulating fluorescence and providing the mechanism for closing the photochemical traps, it is necessary for the fluorescence measurement to be able to distinguish between the two effects. One common approach is to use two different sources of illumination, one of which provides “exciting” light that is chopped or modulated at a frequency of several hundred and the other provides steady (unmodulated) “actinic” light when it is turned on. The exciting light intensity is set

51

sufficiently low that it does not significantly alter the photochemical state of the sample. Thus, the component of the fluorescence that is detectable using a lock-in amplifier tuned to the exciting light modulation frequency serves as a probe of the fluorescence efficiency or yield, regardless of the intensity of the fluorescence produced simultaneously by the unmodulated actinic light, even though the latter is by far the larger contribution to the overall fluorescence signal. Clearly it is necessary to pay careful attention to the design of the electronic circuitry to insure the isolation of these two signals. Commercial instrumentation has been developed for this purpose, including devices that can be used to monitor fluorescence induction in the leaves of plants growing in the field (Büchel and Wilhelm, 1993). In addition to studies of the kinetics of electron transfer reactions, the effectiveness and mode of operation of electron transport inhibitors and the role of different electron donors or acceptors, fluorescence induction has been used to analyze the heterogeneity of Photosystem II in higher plants and how this heterogeneity reflects the distribution of Photosystem II in thylakoid membranes (Melis, 1991). 2. Electric Fields Influence Fluorescence Associated with Photosynthetic Materials Because the transport of electrons across photosynthetic membranes is vectorial, trans-membrane electric fields are generated or modulated during the photosynthetic light reactions. These electric fields provide not only a source of chemical potential for driving some of the dark energyconserving biochemical processes, but also an interaction with pigment molecules in the membranes that absorb or fluoresce strongly. (See chapter by S. Boxer in this volume.) The direct effect of the field produces carotenoid and chlorophyll absorption band shifts that are sensitive to both the magnitude and direction of the field, which in turn provides a calibration of the strength of the field at those molecules (Junge, 1982). Fluorescence yield changes´ also occur, both as a direct effect of the field and as an indirect consequence of alterations in the rates of primary charge separation and recombination in the reaction centers which are competing with the

52 fluorescence. Several studies have focused on the consequences of externally applied electric fields. Macroscopic samples containing whole cells, membranes, sub-membrane fragments or isolated complexes placed between external electrodes can be subjected, at least briefly, to applied fields of (Lockhart et al., 1988). In such experiments the samples are generally spatially isotropic, so that the effect of the directionality of the field is impossible to determine. A second approach makes use of trans-membrane fields produced by transient ion gradients generated between the inside and outside spaces of enclosed membrane vesicles like chloroplast thylakoids or bacterial chromatophores. The direction of the electric field can be changed by altering the relative salt concentrations of the solutions used in rapid mixing experiments (Dau and Sauer, 1991). Based on both steady-state changes in fluorescence intensity and time-resolved fluorescence relaxation measurements, the effect of the applied electric field on the kinetics of primary charge separation and recombination has been investigated.

Kenneth Sauer and Martin Debreczeny not most, light-harvesting complexes isolated from a wide variety of photosynthetic organisms. The absorption bands associated with these long wavelength-emitting pigments are difficult to detect, indicating that they reflect only a small portion of the pigment molecules present. Because the excited states responsible for the long wavelength fluorescence are at energies close to or even lower than the excited states of the associated reaction centers, the role of these pigments in the collection of light and the funneling of excitation to the reaction centers is of considerable interest. Nevertheless, little is known at present about the mechanism of the very effective quenching of this fluorescence at room temperature. It has not been possible, for example, to provide a clear link between the photochemical (open/closed) state of the reaction centers and the intensity of long wavelength fluorescence. III. Time-Resolved Fluorescence

A. Experimental Considerations

3. Temperature Dependence of Fluorescence

1. Time-Correlated Single Photon Counting (TCSPC)

For most fluorescent molecules present in homogeneous dilute solution, the effect of lowering the temperature of the sample is primarily to narrow the spectral widths of the components of the emission spectrum (also of the absorption and excitation spectrum), thus improving the spectral resolution. Apart from this sharpening of the incompletely resolved spectra, there are no dramatic changes in fluorescence yield. Many photosynthetic organisms and isolated pigment-protein complexes, however, show dramatic increases in fluorescence yield upon lowering the temperature (see e.g., Butler et al., 1979; Mukerji and Sauer, 1989). The increase can be 20 fold or more between room temperature and 77 K. Furthermore, the spectrum of the fluorescence increase is typically shifted significantly to long wavelength, in comparison with the bulk of the fluorescence at room temperature. In fact, this is a reflection of the heterogeneity of the molecules giving rise to fluorescence emission in such samples; low temperature-enhanced, long wavelength-emitting pigments appear to be associated with many, if

The traditional instrument for time-resolving fluorescence is schematically the same as the steady-state apparatus shown in Fig. 2. However, the light source in a time-resolving instrument must be pulsed or modulated rather than continuous in intensity. We will limit our discussion here to techniques employing pulsed light sources; for a discussion of time resolution of fluorescence by modulation of a continuous light source, see Lakowicz (1983) or Lakowicz et al. (1990). If a pulsed laser is used as the excitation source, the natural line width of the laser is usually narrow enough that it is unnecessary to use an excitation monochromator. The fluorescence induced in the sample by the excitation pulse is time-resolved by the detector. The time of arrival of a fluorescence photon at the detector is compared with the time of the excitation pulse. The intensity of photons striking the detector must be limited so that single photons can be detected without distortion from multiple photon events. Fluorescence photons arriving at particular time delays relative to the excitation pulse are gated into channels of a

Fluorescence chosen temporal width and a fluorescence decay profile is collected on a multi-channel analyzer. What we have described above is known as the time-correlated single photon counting (TCSPC) technique. This method of time-resolving fluorescence is widely used and its practical implementation has been described in detail (O’Connor and Phillips, 1984). By the simultaneous acquisition of fluorescence photons at multiple time delays, TCSPC has the advantage of allowing for the collection of high signal-to-noise data in a relatively short time. The time resolution of this technique is limited by a combination of the temporal width of the excitation pulse and the time-response of the fluorescence detector. In recent years the temporal width of lasers has rapidly decreased to a point where picosecond and subpicosecond laser pulses are readily achievable with commercially available systems. If maximal time resolution of fluorescence is desired, the limiting factor for the TCSPC technique is usually the response time of the detector. Micro-channel plate detectors, a type of photomultiplier designed to achieve optimal time resolution, currently have time resolution (transit-time spread) as fast as 25 ps. (Hamamatsu Photonics, R3809U series)

2. Fluorescence Upconversion A more recently developed technique of timeresolving fluorescence, known as fluorescence upconversion, has the advantage over the TCSPC technique of being theoretically limited in time resolution by the temporal pulse width of the excitation source rather than the detector response time. In brief, a train of laser pulses is split into two beams, one of which is used to excite the sample. The subsequent fluorescence from the sample is collected and focused onto a non-linear crystal. The other laser beam is used as a variable delay gating pulse and is focused onto the same area of the crystal (see Fig. 4a). If the angle of the incoming light relative to the optical axis of the crystal is such that the phase matching requirement is satisfied, some of the light exiting from the crystal will be at a frequency equal to the sum of the laser and fluorescence frequencies. Since fluorescence upconversion will occur only when both sample fluorescence and

53

54 the laser gating pulse are present in the crystal, the time resolution of the experiment is laserpulse width limited (see Fig. 4b). A fluorescence decay can be recorded by incrementally optically delaying the gating pulse relative to the excitation pulse. Because only a narrow band of fluorescence is upconverted at a particular crystal orientation, a time-resolved fluorescence spectrum can be recorded by tuning the angle of the crystal. The practical implementation of such an instrument has been described elsewhere (Shah, 1988; Doust, 1982; Kahlow et al., 1988; Debreczeny, 1994). The chief disadvantage of the upconversion technique is the low efficiency with which currently employed crystals upconvert fluorescence. This means that unless time resolution of a few picoseconds or better is needed, the TCSPC technique is still the method of choice to obtain high signal-to-noise time-resolved fluorescence data. The fluorescence upconversion technique of time-resolving fluorescence is similar to the pump-probe technique of measuring transient absorption in that it is a two-photon technique and theoretically allows for pulse-width limited time resolution. However, fluorescence upconversion has the advantage over pump-probe techniques that the two photon process occurs in a non-linear crystal, not in the sample. This means that the phenomena of stimulated emission and excitedstate absorption, which often complicate the interpretation of pump-probe experiments, are avoided in the fluorescence upconversion experiment. (See chapter by G Fleming in this volume.) 3. Deconvolution An experimentally observed time-resolved fluorescence signal consists of the molecular fluorescence signal of interest convoluted with the instrument response function (IRF) (O’Connor and Phillips, 1984). If the kinetics of interest occur on a time scale much longer than the temporal width of the IRF, the IRF can be treated as instantaneous and the convolution integral ignored. However, because interesting events in photosynthesis are known to occur on time scale of picoseconds and shorter, practitioners of the TCSPC technique in this field have frequently relied on deconvolution to extend their time res-

Kenneth Sauer and Martin Debreczeny olution. Typically, when using the TCSPC technique, the IRF is collected at the laser excitation frequency by using a scattering solution in place of the sample. It has been found experimentally that in order to achieve the best fits to data, a wavelength dependent time shift between the IRF and the decay must be introduced (O’Connor and Phillips, 1984). This effect has been attributed to a wavelength dependence of the time response of the detector, because the fluorescence decay and IRF are necessarily collected at different wavelengths. As a consequence there is some uncertainty in the designation of time zero. Data from individual decays at different fluorescence wavelengths can be combined to produce time-resolved emission spectra, but at short times after the excitation pulse the time uncertainty leads to large spectral uncertainties, especially if the sample fluorescence changes rapidly at early times. In addition to providing much shorter IRFs, the upconversion technique has the advantage over the TCSPC technique of a more precise measurement of time zero. The IRF can be measured with the same geometric arrangement as the fluorescence upconversion signal by tuning the nonlinear crystal to the optimal angle for generation of sum frequency light from the residual exciting pulse and the gating pulse. The peak of this IRF is the delay setting at which the exciting and gating pulses are exactly temporally overlapping (time zero). As with the TCSPC technique, it can be shown that the observed fluorescence upconversion signal is the molecular fluorescence signal of interest convoluted with the IRF measured in the above manner (Doust, 1982). 4. Exciton Annihilation Because the upconversion signal is dependent on the square of the energy of the laser pulses, (Doust, 1982) the upconverted fluorescence power will be larger for high energy laser pulses at a low repetition rate than for low energy pulses at a high repetition rate, for a fixed level of average laser power. High pulse energies at low repetition rates have been used to study photostable dye molecules. However, in multi-chromophore systems like photosynthetic proteins, high energy pulses can lead to exciton annihilation

Fluorescence (Geacintov and Breton, 1982). Exciton annihilation can occur when two or more chromophores that are coupled by energy transfer each absorbs a photon. The experimental result is that the fluorescence decay profile shows an excitation intensity dependence and a decay rate governed by bi-excitonic annihilation (Geacintov and Breton, 1982). A simple estimate of the extent of exciton annihilation can be obtained by dividing the number of photons absorbed in the beam spot per laser pulse by the number of light-harvesting complexes (LHC) in the beam spot. In this context a LHC is a group of chromophores that are coupled by excitation transfer.

where is the molar decadic extinction coefficient of the entire LHC at the laser frequency, h is Planck’s constant, is the frequency of the laser, r is the radius of the laser beam waist, and is Avogadro’s number. The solution is assumed to be dilute. The average number of photons absorbed per LHC should be less than unity if exciton annihilation effects are to be avoided. Using the above equation, the extent of exciton annihilation in the TCSPC technique and the upconversion technique are compared, choosing the trimeric aggregate of C-phycocyanin (PC) excited at 624 nm with a repetition rate of 4 MHz as an example. In a typical TCSPC experiment the laser power at the sample is 0.5 mW, and the beam waist is roughly (unfocused). The average number of photons absorbed per pulse per PC timer in such an experiment is In a fluorescence upconversion experiment a typical power seen by the sample is 1 mW and the laser is focused onto the sample to have a beam waist. The number of photons absorbed per pulse per PC trimer under these conditions is If much larger complexes (for example, whole phycobilisomes which contain hundreds of coupled chromophores) are studied, exciton annihilation must be considered more carefully if the fluorescence upconversion

55 technique is to be employed. The amount of light pumping the sample can be reduced by diverting more power into the gating pulse. But since the upconverted power is a product of the gating and fluorescence powers, the reduction in pump power relative to gate power will ultimately lead to reduction in the signal level.

B. Isotropic Time-Resolved Fluorescence As mentioned in the section on steady-state fluorescence and shown below in Eq. (4), the fluorescence from a mixture of non-interacting and optically dilute chromophores can be described additively. The initial excited-state populations of the x different chromophore types are determined by the relative extinction coefficients, of the chromophores at the excitation wavelength, The evolution of the chromophore excited-state populations as a function of time are described by P(t), which has a value of 1 at the time of excitation and eventually decreases to 0, at which point the excited state has been entirely depleted. In the case of a non-interacting mixture of homogeneous chromophores, P(t) will decay as a single exponential. The contribution of each chromophore to the observed fluorescence signal is weighted by its net fluorescence quantum yield, and the shape of its fluorescence spectrum, f, as a function of the emission wavelength,

From Eq. (4) we can see that, in addition to resolving mixtures of chromophores by differences in their excitation and emission spectra, the lifetime of each chromophore species can be used as a further discriminating factor. If the requirement that chromophores be noninteracting is relaxed somewhat to include weak interactions in which the exchange of excitation energy occurs but energetic coupling is not so great as to affect the individual chromophore spectra (these are the conditions under which Förster’s mechanism of inductive resonance is applicable), Eq. (4) still holds. However, the population term, P(t), is no longer necessarily described by a single exponential, nor will it necessarily decay monotonically. If excitation en-

56 ergy is transferred preferentially from a donor chromophore type to an acceptor, the population term of the acceptor chromophore initially increases as a function of time if this transfer is rapid compared to decay of its excited state by means other than energy transfer. Electron transfer reactions can also be included in Eq. (4) by treating the excited state and ionized product of each molecule as separate species, although the photosynthetic molecular ions are typically non-fluorescent. For this reason time-resolved fluorescence and transient absorption measurements often provide complementary information. In such situations the ability to monitor the disappearance of the fluorescence as the molecule becomes ionized can provide an unambiguous measure of excited-state decay without interference from product formation. Unlike dye molecules free in solution, the chromophores in photosynthetic systems are organized in association with proteins to have fixed relative orientations and distances of separation. The geometry of the chromophore interactions provides high efficiency energy and electron transfer. The fixed geometry means that the energy transfer from a donor to acceptor chromophore in a light-harvesting complex can be described by a narrow range of rate constants and is often well approximated by a single rate constant. Similarly, electron transfer reactions in photosynthetic systems are often well represented by a single rate constant. This being the case, the time dependence of the chromophore excited-state or ionized-state populations can be described by Eq. (5). P is a vector describing excited-state and ionized-state populations of each chromophore type, while M is a square matrix containing the rate constants for energy transfer or electron transfer between different chromophore types. The diagonal elements of M (for which the donor and acceptor are the same chromophore type) contain the negative of the sum of the rate constants for all means of decay including energy transfer and electron transfer from the excited state of a particular chromophore type.

If Eq. (5) is solved for the excited-state or ionstate population of a particular chromophore spe-

Kenneth Sauer and Martin Debreczeny cies, the result is a sum of exponential terms described by Eq. (6):

Within a given system, all of the chromophore excited-state and ion-state populations contain the same number of exponential terms with the same rate constants, However, the amplitudes, associated with each exponential term will be different for the different chromophore species. By incorporating Eq. (6) into Eq. (4) we can see that the isotropic time-resolved fluorescence signal is described by a sum of exponentials. It is for this reason that the most common method of analysis of isotropic time-resolved fluorescence measurements by workers in the field of photosynthesis is to fit the data to a sum of exponentials. The fitted exponential amplitudes and rate constants are functions of the rate constants for energy transfer and electron transfer between chromophores. A common analysis technique that relies on the assumption that the rate constants are independent of the probed emission wavelength is to fit data simultaneously at multiple emission wavelengths with the exponential rate constants being linked at all wavelengths while the amplitudes are varied freely (global analysis) (Knorr and Harris, 1981; Knutson et al., 1983; Holzwarth et al., 1987; also, see chapter 5 by A.R. Holzwarth in this volume). Such techniques can increase the number of resolved kinetic components. Unfortunately, without simplifying assumptions, it is often difficult to relate these experimentally observed exponential rate constants to the molecular rate constants that are of interest. Recently algorithms have been developed which allow one to directly extract rate constants of interest for a given model (Roelofs et al., 1992; also, see chapter 5 by A.R. Holzwarth in this volume). In this case the difficulty often lies in establishing the uniqueness of a particular solution or model. When using the summed exponential model to describe time-resolved fluorescence one should keep in mind that the model is based on the assumption that kinetic processes can be described by discrete rate constants, and this will only be true to the extent that the chrom-

Fluorescence

ophores are rigidly held by the protein lattice during the relevant photo-physical events. In addition to the extraction of rate constants, the ability to observe the evolution of the fluorescence spectrum as the excited-state populations of a mixture of weakly coupled chromophores equilibrate allows one to extract information about the fluorescence spectra of the individual chromophores. Fig. 5a shows the time-resolved emission spectrum of the subunit of the lightharvesting protein C-phycocyanin at 77 K, excited at 580 nm (Debreczeny et al., 1993). The subunit contains only two chromophore types, referred to as and according to the amino acid residue to which the linear tetrapyrrole chromophore is covalently attached. Two peaks are evident in the time-resolved spectra; one peak at about 620 nm showing greatest intensity at early times, and a second peak at about 650 nm increasing in intensity concurrently with the decrease of the 620 nm peak. The spectra at the earliest times are representative of the chromophore excited-

57

state population induced by the 580 nm excitation pulse. It is evident that although both chromophores are excited by the laser pulse, the higher energy chromophore is preferentially excited. Energy transfer is most favorable in an energetically downhill direction (as predicted by Förster’s theory) so that with time we see an increase in the excited-state population of the lower-energy chromophore with concurrent loss of the excited-state population of the higherenergy chromophore With prior knowledge of the relative extinction coefficients of the two chromophore types, the time-resolved spectra at all times were modeled, and information about the rate constants for energy transfer and the fluorescence spectra of the individual chromophores were extracted. The inverse of the sum of the forward and back rate constants for energy transfer was found to be 64 ps with the ratio of the back-to-forward rate constants being <0.1 (Debreczeny et al., 1993). The extracted fluorescence spectra of the and chromophores are shown in Fig. 5b. The discussion above is restricted to the case of weak coupling in which individual chromophore spectra are not affected by inter-chromophore interactions. At inter-chromophore separation distances of <20 Å, however, strong energetic interactions between chromophores can lead to spectra which are not additive in terms of the spectra of the individual chromophores. In such cases the excitation energy becomes delocalized over the strongly coupled aggregates of chromophores and Eq. (4), which treats the net fluorescence signal as a sum of the fluorescence from the individual chromophores in a mixture, no longer applies. However, Eq. (4) may still be useful in systems in which chromophores are arranged in clusters within which chromophores are excitonically coupled, but between which energy transfer occurs by the Förster mechanism (Sauer, 1975).

C. Anisotropic Time-Resolved Fluorescence For a photo-physical event to be kinetically observable by isotropic time-resolved fluorescence spectroscopy, it must involve a change in the shape of the observable fluorescence spectrum or the fluorescence quantum yield. This means, for

58 example, that energy transfer between spectrally identical chromophores will be invisible by isotropic fluorescence spectroscopy. If, however, the polarization of the emitted fluorescence is measured in addition to its intensity and wavelength, much more can be learned about the nature of the photo-physical event, including resolution of energy transfer rate constants among identical chromophores. Tao (1969) has shown that the time-resolved fluorescence anisotropy of a single molecule (calculated according to Eq. (2)) can be interpreted as being a function of the dot product of the absorption transition dipole, induced by the excitation pulse, with the emission transition dipole, observed at some later time, t.

is the second Legendre polynomial in x, and the brackets around represent a correlation function that includes the rotational averaging necessary to describe an isotropic solution and an orientation evolution function for the emissive transition dipole. Eq. (7) shows that time-resolved fluorescence anisotropy is sensitive to rotation of the net emission transition dipole as a function of time and that the net angle by which the emission dipole is rotated relative to the absorption transition dipole can be extracted from the residual anisotropy level. If energy transfer between chromophores is being probed, fluorescence anisotropy will be sensitive to the relative orientation of the emission dipoles of the donor and acceptor chromophores. If the initially absorbing chromophore is also the emitting chromophore, the anisotropy value will remain high (at 0.4 if the absorption and emission transition dipoles are parallel, neglecting rotational diffusion). If the emitting chromophore is different from the initially absorbing chromophore, the fluorescence anisotropy value will decrease according to the extent by which the emission transition dipole is rotated away from the orientation of the absorbing transition dipole. In a typical case, the observed fluorescence spectrum contains contributions both from initially absorbing molecules and from molecules which received excitation through energy transfer. In

Kenneth Sauer and Martin Debreczeny this case, the anisotropic emission spectrum will dominate at early times when energy transfer has not had time to occur; then as energy transfer occurs, the anisotropy will decrease. If energy transfer occurs between chromophores of different energies, the anisotropy in the long wavelength region of the emission spectrum will normally decrease more rapidly than that in the short wavelength region, because energy transfer occurs preferentially downhill energetically. The reader is referred elsewhere for derivations of the model function used to fit timeresolved anisotropic fluorescence of photosynthetic systems (Lyle and Struve, 1991; Cross et al., 1983; Van Amerongen and Struve, 1995). Instead we give an example of the utility of this technique in extracting rate constants for energy transfer in C-phycocyanin. Fig. 6a shows the time-resolved fluorescence anisotropy of PC monomers and trimers isolated from a genetically engineered strain of Synechococcus sp. PCC 7002 (Debreczeny et al., 1995). The mutant strain was engineered to be missing one of the three chromophore types found in PC, to simplify the kinetic analysis (Debreczeny et al., 1993). A model of the PC trimer isolated from the mutant strain, based on the known crystal structure of wild-type PC (Schirmer et al., 1987), is shown in Fig. 6b. Note that the monomer aggregates into the trimer with symmetry (axis of symmetry perpendicular to the plane of the diagram). Trimers of PC stack to form the “rods” of the phycobilisome, a well-characterized light-harvesting complex found in cyanobacteria and some algae. The fluorescence anisotropy of PC monomers in Fig. 6a decays from an anisotropy value of 0.4 to a value of 0.33 with a characteristic time of 200 ± 70 ps. Because only two chromophores are present in this system (the and chromophores), the decay time can be directly related to the rate of energy transfer between them, and the small amount of anisotropy decay indicates that the chromophore transition dipoles are oriented nearly parallel to each other (27° ± 1.5° from parallel). Using Förster theory, a decay time of 158 ± 12 ps can be predicted, in reasonable agreement with the measured value (Debreczeny et al., 1995). When the monomer PC is aggregated into trimers, a dramatic effect is seen in the fluorescence

Fluorescence

59 sured values are again in excellent agreement with predictions based on the Förster theory (1.4 ± 0.1 ps and 46 ± 5 ps, respectively). Also from the experimental anisotropy decay it is possible to estimate the angle between the transition dipoles of the and chromophores on adjacent monomers (52°) and to restrict the allowed values of the angles formed between the axis of symmetry and the transition dipoles of the and chromophores. Further structural information could be obtained if oriented rather than isotropic samples of PC were used in these experiments. (See chapter by G. Garab in this volume.) IV. Conclusion

anisotropy decay. The PC trimer anisotropy decays to a much lower value and reaches its residual anisotropy level much more quickly than does the monomer. Two distinct regions of the PC trimer anisotropy are evident, showing characteristic decay times of 1.0 ± 0.2 ps (at the limit of the instrument time response) and 40 ± 2 ps. These two decay times, not seen in the monomer relaxation, can be assigned to energy transfer between the and chromophores on adjacent monomers and to energy transfer around the trimer ring, respectively (Fig. 6b). These mea-

Fluorescence has proved to be a valuable tool in investigating many questions related to photosynthetic energy conversion. Knowledge of the energies and lifetimes of the excited states of pigment molecules provides insights into events associated with excitation transfer among antenna chromophores, trapping and charge separation in reaction centers, charge recombination, quenching processes, etc. Existing quantum mechanical theory is adequate, in most cases, for understanding these processes, at least on a time scale longer than a few picoseconds. Where detailed structural information about the pigment-protein complexes is available, for example, calculated and experimental values for excitation transfer rates in chromophore arrays are in good agreement. Fast relaxation events on the sub-picosecond time scale are less well understood, and this regime is now being actively explored. It is a challenge to both experimentalists and theorists to design studies that address the major outstanding questions relating to our fundamental knowledge of excited electronic states and how their properties can be used to inform our understanding of the photosynthetic process. Acknowledgements The authors thank their colleagues Mary Talbot, Shelly Pizarro and Dr. Theo Roelofs at UC Berkeley for very helpful comments on this contribution. We are especially grateful to Mary Talbot for providing the spectra shown in Fig. 3. The original research carried out in the authors'

60 laboratory was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Energy Biosciences Division, of the U.S. Department of Energy under Contract No. DEAC03–76SF00098.

References Boardman NK and Thorne SW (1971) Sensitive fluorescence method for the determination of chlorophyll a / chlorophyll b ratios. Biochim Biophys Acta 253: 222–231. Briantais J-M, Vernotte C, Krause GH and Weis E (1986) Chlorophyll a fluorescence of higher plants: Chloroplasts and leaves. In: Govindjee, Amesz J and Fork DC (eds) Light Emission by Plants and Bacteria, pp 539–583 Academic, Orlando, FL. Büchel C and Wilhelm C (1993) In vivo analysis of slow chlorophyll induction kinetics in algae: Progress, problems and perspectives. Photochem Photobiol 58: 137–148. Butler WL, Tredwell CJ, Malkin R and Barber J (1979) The relationship between the lifetime and yield of the 735 nm fluorescence of chloroplasts at low temperatures. Biochim Biophys Acta 545: 309–315. Cantor CR and Schimmel PR (1980) Biophysical Chemistry. Part II. Techniques for the Study of Biological Structure and Function. WH Freeman, San Francisco, CA. Cross AJ, Waldeck DH and Fleming GR (1983) Time resolved polarization spectroscopy: Level kinetics and rotational diffusion. J Chem Phys 78: 6455–6467. Dau H (1994) Molecular mechanisms and quantitative models of variable Photosystem II fluorescence. Photochem Photobiol 60: 1–23. Dau H and Sauer K (1991) Electric field effect on chlorophyll fluorescence and its relation to Photosystem II charge separation reactions studied by a salt-jump technique. Biochim Biophys Acta 1098: 49–60. Debreczeny MP (1994) Comparison of the rate constants for energy transfer in the light-harvesting protein, C-phycocyanin, calculated from Förster’s theory and experimentally measured by time-resolved fluorescence spectroscopy. Ph.D. Thesis, University of California, Berkeley, Lawrence Berkeley Laboratory Report LBL-35672. Debreczeny MP, Sauer K, Zhou J and Bryant DA (1993) Monomeric C-phycocyanin at room temperature and 77 K: Resolution of the absorption and fluorescence spectra of the individual chromophores and the energy-transfer rate constants. J Phys Chem 97: 9852–9862. Debreczeny MP, Sauer K, Zhou J and Bryant DA (1995) Comparison of calculated and experimentally resolved rate constants for excitation energy transfer in C-Phycocyanin. 1. Monomers. 2. Trimers. J Phys Chem 99: 8412–8419; 8420–8431. Doust TAM (1982) Fluorescence lifetime and depolarisation measurements using frequency up-conversion. In: Doust TAM and West MA (eds) Picosecond Chemistry and Biology, pp 1–34. Science Reviews, Northwood, Middlesex, England.

Kenneth Sauer and Martin Debreczeny Förster T (1948) Zwischenmoleculare Energiewanderung und Fluoreszenz. Ann Physik 2: 55–75. Förster T (1967) Mechanisms of energy transfer. In: Florkin M and Stotz (eds) Comprehensive Biochemistry, Vol 22, pp 61–80, Elsevier, Amsterdam. Geacintov NE and Breton J (1982) Exciton annihilation and other non-linear high-intensity excitation effects In: Alfano RR (ed) Biological Events Probed by Ultrafast Laser Spectroscopy, pp 157–191. Academic Press, New York. Gouterman M and Stryer L (1962) Fluorescence polarization of some porphyrins. J Chem Phys 37: 2260–2266. Govindjee, Amesz J and Fork DC (eds) (1986) Light Emission by Plants and Bacteria. Academic Press, Orlando, FL. Holzwarth AR, Wendler J and Suter GW (1987) Studies on chromophore coupling in isolated phycobiliproteins. II. Picosecond energy transfer kinetics and time-resolved fluorescence spectra of C-phycocyanin from Synechococcus 6301 as a function of the aggregation state. Biophys J 51: 1–12. Houssier C and Sauer K (1969) Optical properties of the protochlorophyll pigments. II. Electronic absorption, fluorescence and circular dichroism spectra. Biochim Biophys Acta 172: 492–502. Junge W (1982) Electrogenic reactions and proton pumping in green plant photosynthesis. In: Slayman CL (ed) Current Topics in Membranes and Transport, Vol 16, Electrogenic Ion Pumps, pp 431–465. Academic Press, New York. Kahlow MA, Jarzeba W, DuBruil TP and Barbara PF (1988) Ultrafast emission spectroscopy in the ultraviolet by timegated upconversion. Rev Sci Instrum 59: 1098–1109. Kautsky H and Hirsch (1931) Neue Versuche zur Kohlensäureassimilation. Naturwissenschaften 19: 964. Knorr FJ and Harris JM (1981) Resolution of multicomponent fluorescence spectra by an emission wavelength-decay time data matrix. Anal Chem 53: 272–276. Knox RS (1975) Excitation energy transfer and migration: theoretical considerations. In: Govindjee (ed) Bioenergetics of Photosynthesis, pp 183–221. Academic Press, New York. Knutson JR, Beechem JM and Brand L (1983) Simultaneous analysis of multiple fluorescence decay curves: A global approach. Chem Phys Lett 102: 501–507. Krause GH and Weis E (1991) Chlorophyll fluorescence and photosynthesis: the basics. Annu Rev Plant Physiol Plant Mol Biol 42: 313–349. Lakowicz JR (1983) Principles of Fluorescence Spectroscopy. Plenum, New York. Lakowicz JR, Jayaweera R, Szmacinski H and Wiczk W (1990) Resolution of multicomponent fluorescence emission using frequency-dependent phase angle and modulation spectra. Anal Chem 62: 2005–2012. Latimer P, Bannister TT and Rabinowitch E (1956) Quantum yields of fluorescence of plant pigments. Science 124: 585– 586. Lockhart DJ, Goldstein RF and Boxer SG (1988) Structurebased analysis of the initial electron transfer step in bacterial photosynthesis: Electric field induced fluorescence anisotropy. J Chem Phys 89: 1408–1415. Lyle PA and Struve WS (1991) Dynamic linear dichroism in chromoproteins. Photochem Photobiol 53: 359–365.

Fluorescence Mauzerall D (1972) Light-induced fluorescence changes in Chlorella, and the primary photoreactions for the production of oxygen. Proc Natl Acad Sci USA 69: 1358–1362. Melis A (1991) Dynamics of photosynthetic membrane composition and function. Biochim Biophys Acta 1058: 87– 106. Mukerji I and Sauer K (1989) Temperature-dependent steadystate and picosecond kinetic fluorescence measurements of a Photosystem I preparation from spinach. In: Briggs WR (ed) Photosynthesis, pp 105–122. Alan R. Liss, New York. O’Connor DV and Phillips D (1984) Time-correlated Single Photon Counting. Academic Press, London. Rigler R and Ehrenberg M (1976) Fluorescence relaxation spectroscopy in the analysis of macromolecular structure and motion. Quart Rev Biophys 9: 1–19. Roelofs TA, Lee C-H and Holzwarth AR (1992) Global target analysis of picosecond chlorophyll fluorescence kinetics from pea chloroplasts. A new approach to the characterization of the primary processes in photosystem II a- and bunits. Biophys J 61: 1147–1163.

61 Sauer K (1975) Primary events and the trapping of energy. In: Govindjee (ed) Bioenergetics of Photosynthesis, pp 115–181. Academic Press, New York. Schirmer T, Bode W and Huber R (1987) Refined threedimensional structures of two cyanobacterial C-phycocyanins at 2.1 and 2.5 Å resolution. J Mol Biol 196: 677–695. Shah J (1988) Ultrafast luminescence spectroscopy using sum frequency generation. IEEE J Quantum Electron 24: 276– 288. Stepanov BI (1957) A universal relation between the absorption and luminescence spectra of complex molecules. Sov Phys Dokl 2: 81–84. Struve WS (1989) Fundamentals of Molecular Spectroscopy. Wiley-Interscience, New York. Tao T (1969) Time-dependent fluorescence depolarization and Brownian rotational diffusion coefficients of macromolecules. Biopolymers 8: 609–632. Van Amerongen H and Struve WS (1995) Polarized optical spectroscopy of chromoproteins. In: Sauer K (ed) Methods in Enzymology, Vol 246. Biochemical Spectroscopy, pp. 259–283. Academic Press, Orlando, FL.

This page intentionally left blank

Chapter 4 Ultrafast Spectroscopy of Photosynthetic Systems Ralph Jimenez and Graham R. Fleming* Department of Chemistry and the James Franck Institute, The University of Chicago, 5735 S. Ellis Avenue, Chicago, IL 60637, USA

Summary I. Introduction II. Laser Sources A. Modelocked Ti:sapphire Laser Systems 1. Ti:sapphire Oscillators 2. Ti:sapphire Amplifiers B. Methods of Extending the Tuning Range III. Fluorescence Upconversion A. Experimental Technique B. SomeExamples IV. Transient Absorption A. Experimental Technique B. Some Examples V. ConcludingRemarks References

63 63 64 64 64 65 65 66 66 68 70 70 71 72 72

Summary This Chapter discusses the use of fluorescence upconversion and transient absorption techniques for the study of photosynthetic systems. A description is given of the state-of-the-art laser systems available for ultrafast studies, along with examples of some common techniques for extending the wavelength range of these lasers. The experimental techniques of fluorescence upconversion and transient absorption are introduced, with the goal of describing the implementation, versatility, and limitations of these experiments. Recent experimental results are presented which illustrate applications of ultrafast spectroscopy to studies of excitation energy transfer in light harvesting complexes and electron transfer in bacterial reaction centers. Abbreviations: LH-I, LH-II – light-harvesting complexes I and II of purple bacteria; LHC-II – lightharvesting complex II of plants; P – primary electron donor; PS – photosystem; RC – reaction center; LBO – lithium borate;

I. Introduction

subpicosecond timescales. Thus ultrafast spectroscopy has played, and continues to play, a key role in characterizing these highly efficient processes. In this chapter we focus on the two most common spectroscopic techniques: time resolved absorption and fluorescence spectroscopies. Over the past ten years or so, a formidable array of

The elementary electron and energy transfer steps in photosynthesis occur on picosecond or *Correspondence: Fax: 1-312-7020805; E-mail: [email protected]

63 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 63–73. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

64 nonlinear spectroscopic techniques have been developed as have extensions into the infrared spectral region. The reader is referred to recent literature for details of these techniques (Diller et al., 1992; Maiti et al., 1993, Durrant et al. 1994). Solid state lasers, in particular Ti:sapphire lasers have greatly simplified the practice of, and extended the capabilities of ultrafast spectroscopy. We describe apparatuses based only on such lasers in the belief that dye laser-based systems will become obsolete within the next five to seven years. One aspect of ultrafast spectroscopy that contrasts it with longer timescale spectroscopy should be mentioned. Femtosecond pulses are shorter than the periods of some molecular vibrations; in other words, “vibrationally abrupt” (Scherer et al., 1993; Jonas and Fleming, 1994; Jonas et al., 1994). For example, a 50 fs pulse is abrupt with respect to a vibration. In these cases, vibrational quantum beats may be observed in the pump probe or spontaneous emission signals. The beat frequencies allow measurement of vibrational frequencies in absorption spectra that are quite devoid of structure. It may, however, be quite tricky to assign the frequencies to the ground or excited state and to be sure that the fundamental frequency is being observed. Jonas and Fleming (1994) and Scherer et al. (1993) discuss these points in detail. The rate of disappearance of these beats reflects the timescale on which the environment perturbs the energy levels of the chromophore. A fascinating development in ultrafast photosynthesis research is the observation of coherent nuclear motion in bacterial reaction centers by Martin and coworkers (Vos et al., 1993). The beats are assigned to the excited state and suggest that vibrational dephasing may be incomplete on the timescale of the primary charge separation. The outline of this chapter is as follows. First, the newest laser sources are described. Various methods of extending the tuning range of these sources over the ultraviolet, visible, and nearinfrared regions are then outlined. Next, the principles of fluorescence upconversion are described, along with a discussion of the optical arrangements used for these experiments. Applications of fluorescence upconversion for monitoring the rates of energy transfer in bacterial and plant light

Ralph Jimenez, and Graham R. Fleming

harvesting systems, and measuring the rate of the primary charge separation in bacterial reaction centers are discussed. Finally, the experimental methodology of transient absorption is described, along with two applications of this technique to studies of bacterial reaction centers. II. Laser Sources

A. Modelocked Ti:sapphire Laser Systems 1. Ti:sapphire Oscillators Modelocked Ti:sapphire oscillators and amplifiers represent an enormous simplification over the traditional femtosecond dye laser/amplifier arrangements. Additionally, modelocked Ti:sapphire oscillators can directly generate sub-20 fs pulses (Huang et al., 1992a,b; Asaki et al. 1993). This performance could only be achieved with a dye laser after amplification, continuum generation and pulse compression. The cavity arrangement of a typical self-modelocked Ti:sapphire laser, given schematically in Fig. 1, shows the simplicity of these lasers. The oscillator is usually pumped by 5–8 Watts from a small frame argon ion laser, and modelocking is initiated by some mechanical disturbance of the cavity. Short pulse generation has been demonstrated from near 700 nm to above 1000 nm with essentially the same cavity design. These lasers produce pulses with 1–6 nJ energy at a repetition rate around 80 MHz, with pulses as short as 11 fs. Cavity

Ultrafast spectroscopy in photosynthesis

65

dumping of these oscillators has also been demonstrated, giving 40–60 nJ pulse energies at tens to hundreds of kHz repetition rate (Pschenichnikov et al., 1994). No increase in pulse duration results from cavity dumping. Extremely low noise (< 0.1% RMS) and high long term stability combined with these desirable pulse durations make these lasers ideal sources for the study of nearIR photosynthetic systems, e.g. purple bacteria, green bacteria, and heliobacteria. 2. Ti:sapphire Amplifiers For many applications, the tuning range of these lasers must be extended into the visible. There are several techniques for doing this, but higher pulse energies are usually necessary. Furthermore, lower repetition rate, higher pulse energy sources are often required for experiments in which the sample recovery time is longer than the modelocked frequency. Ti:sapphire amplifiers are most easily categorized by their repetition rates. It should be noted that all three types of amplifiers are available commercially. The first category is that of Nd:YAG pumped amplifiers. The pump source is the second harmonic of a Q-switched Nd:YAG laser. These amplifiers typically give hundreds of millijoules to several joules of pulse energy, at 10–20 Hz repetition rate. Amplified pulses as short as 21 fs have been obtained (Zhou et al., 1994). A pulse is selected from the oscillator, temporally stretched (in order to prevent damage to the amplifier due to high peak power) by a grating, routed into the amplifier cavity via electro-optic switching, and dumped out of the cavity in the same way after several round trips. A grating compressor is used after amplification to restore the pulse duration. The second type of Ti:sapphire amplifier is pumped by the second harmonic of a Q-switched Nd:YLF laser. These regenerative amplifiers operate at 1–10 kHz repetition rates and typically utilize pre-amplification pulse stretching, electrooptic switching, and post-amplification compression (Salin et al., 1991; Squier et al., 1993). These systems yield from tens of microjoules to one millijoule pulse energies, with pulse durations down to 30 fs (Wynne et al., 1994). The layout of a system designed and built in our laboratory

which does not employ a pulse stretcher is shown in Fig. 2 (Joo et al., 1995). The third type of amplifier system is the high repetition rate regenerative amplifier which is pumped with 14–15 Watts from a CW ion laser (Norris, 1992). In this amplifier, the pulse is injected and ejected acousto-optically (electro-optic devices are limited to 10 kHz or lower). Pulse stretching prior to amplification may be employed, but more commonly the pulse is stretched by dispersive elements within the amplifier cavity. As usual, the pulse is recompressed after amplification. These amplifiers typically give 1–5 microjoules pulse energies, at repetition rates from 50 kHz to 300 kHz. Pulses as short as 85 fs have been reported (Sosnowski et al., 1994).

B. Methods of Extending the Tuning Range A variety of methods may be employed for shifting the frequency of ultrafast laser sources. One

66

Ralph Jimenez and Graham R. Fleming

of the simplest methods is harmonic generation. Second and third harmonic generation of Ti:sapphire wavelengths is commonly done with LBO and BBO crystals. The wavelength range from 250 nm to 320 nm and 360 nm to 490 nm may be accessed in this way. Another important technique is continuum generation. When an amplified femtosecond pulse (pulsewidth <200 fs and energy >200 nJ) is focused into a transparent material (e.g. sapphire), the high intensity causes self focusing and subsequent self-phase modulation which creates a stable white light continuum extending across the entire visible spectrum. For pulse energies less than 2 microjoules, this produces a high-quality Gaussian beam at all wavelengths throughout the visible and near infrared. Typically, 1–5 nJ energy is produced for a given 10 nm bandwidth. The color may be selected with an interference filter, providing enough light to serve as the excitation pulse for a fluorescence upconversion experiment, or the probe pulse for a transient absorption experiment. Selected portions of the continuum may be amplified by optical parametric amplification (Reed et al., 1994). One way of doing this involves temporally and spatially overlapping a portion of the continuum into a BBO crystal, along with an intense (1 microjoule or more) pulse of second harmonic from an amplified Ti:sapphire beam. With a 250 kHz amplifier, it has been demonstrated that tens of nanojoules per pulse can be produced across the visible spectrum. III. Fluorescence Upconversion

A. Experimental Technique The principle of the fluorescence upconversion method is that a short pulse of light excites a sample whose fluorescence is focused into a nonlinear crystal along with a variably delayed ‘gate’ pulse, and the sum frequency is detected as a function of the delay between the two pulses (Fig. 3). Rotation of the crystal determines the wavelength of fluorescence upconverted. The advantage of using this gating technique is that the time resolution of the experiment is determined by the width of the pulses (see below), not by the time resolution of the detection system. Detailed de-

scriptions of the upconversion method were given by Shah (1988) and Kahlow et al. (1988). The upconverted signal is usually detected by a photomultiplier tube used with a photon counter. A double monochromator is very helpful because despite the non-collinear geometry, the signal is contaminated with doubled gate beam and doubled excitation beam, both of which are orders of magnitude more intense than the upconverted signal. A prism may also be used for dispersing the upconverted light prior to the monochromator. The use of nonlinear crystals which require orthogonally polarized interacting beams (Type II phase-matching) is also helpful in this sense, in reducing the background due to one beam signals. Two optical layouts for the upconversion experiment are shown in Fig. 4.

Ultrafast spectroscopy in photosynthesis

The setup with the elliptical mirror is the easiest to align. However, the scheme with the parabolic mirrors is more flexible since there is enough space to insert a small cryostat. In either arrangement, the excitation beam may be aligned so that it does not hit the reflector, and fluorescence is collected at a small angle off-axis from the excitation beam. This technique makes it easier to measure fluorescence wavelengths near the excitation wavelength without upconverting the excitation beam. The time resolution of the upconversion experiment is determined by the instrument response function (IRF), which is approximately given by the cross-correlation of the excitation pulse with the gate pulse:

67

where and are the intensity profiles of the excitation and gate pulses, respectively. Operationally, the IRF is measured by angle-tuning the crystal to upconvert transmitted or scattered excitation light. Utilizing the setup of Fig. 4 (left), M. Du, X. Xie and coworkers (Du et al., 1992) have achieved a 70 fs IRF. For pulses appreciably shorter than 100 fs, crystals much thinner than 1 mm are required. The sum-frequency is generated at all points through the thickness of the nonlinear crystal where the gate pulse and fluorescence (or excitation pulse) are temporally and spatially overlapped. But the group velocity mismatch between the fluorescence and gate beam wavelengths causes broadening of the IRF. For example, Du et al. (1992) calculate the group velocity mismatch for their 0.4mm crystal to be 115– 130 fs when upconverting 940 nm fluorescence

68 with a 608 nm gate pulse. This broadening needs to be added to the measured cross-correlation in considering the time resolution of the measurement (Shah, 1988). Thus, limitations of the upconversion technique make experiments with IRFs of significantly less than 100 fs difficult, even with the availability of shorter pulses. Furthermore, the necessity of using thin crystals with large upconverted bandwidths causes a difficulty in discriminating fluorescence from scattered or transmitted excitation beam. This contamination of the signal can be very difficult to overcome when measuring fluorescence much closer than 30–40 nm from the excitation wavelength. Fluorescence upconversion measurements with high time resolution (<100 fs) are most readily performed on samples with large Stokes shifts. Unfortunately, this is not usually the case for chlorophylls in, for example, light harvesting systems. Many fluorescence upconversion measurements resolve the isotropic emission from a sample in order to monitor the appearance or lifetime of an emitting species. In contrast, depolarization measurements can be used to resolve dynamics within a fluorescence band, such as energy transfer. Performing these measurements requires the addition of a polarizer to each beam and a waveplate for rotating the polarization of the excitation beam. In addition, it must be verified that the optical arrangement preserves the anisotropy of the fluorescence by ensuring that the anisotropy of a standard system (such as Rb. sphaeroides R26 reaction centers) is 0.4. A further complication in these studies is the presence of singlet–singlet and singlet–triplet annihilation processes whenever high excitation densities and/or high repetition rates are used (van Grondelle, 1985). These processes shorten the fluorescence lifetime of pigment-protein complexes, and may affect depolarization measurements when the systems are highly ordered. Low temperature fluorescence upconversion studies have not yet become common. Signal averaging considerations make fluorescence upconversion most convenient with medium and high repetition rate laser sources. But biological samples are easily damaged, and the sample must be moved throughout the experiment. The challenge of refreshing the sample volume excited by

Ralph Jimenez and Graham R. Fleming the beam has not been easily met. Stanley and Boxer (1995) have measured fluorescence decays of bacterial reaction centers at 80 K by mounting a light (4 oz.) Joule–Thompson refrigerator (MMR Technologies) on an audio speaker, in an optical arrangement similar to that shown in Fig. 4 (right).

B. Some Examples The first example of a fluorescence upconversion experiment concerns the dynamics of primary charge separation in the reaction centers of purple bacteria such as Rhodobacter capsulatus and Rhodobacter sphaeroides. The timescale for electron transfer from the primary donor, an excitonically coupled dimer of bacteriochlorophyll a (denoted P*) to the acceptor bacteriopheophytin is around 3 picoseconds at room temperature. This rate was measured by a number of groups by monitoring the decay of stimulated emission from P* (for example, Martin et al., 1986). In measurements of the decay of P* by stimulated emission, most workers have made measurements at or near the isosbestic point in the spectrum consisting of ground state bleaching and absorption of the radical cation of P and P*. However, such a procedure makes it difficult to observe longer decay components in the stimulated emission. Fluorescence upconversion measurements performed by Du and coworkers clearly showed behavior which had been difficult to observe in pump-probe experiments. After exciting the special pair’s or bands and time resolving the P* emission at 940 nm, M. Du, S.J. Rosenthal and coworkers (Du et al., 1992) observed nonexponential decay kinetics. An example of a P* decay is shown in Fig. 5. Attempts to explain this observation have ranged from considerations of heterogeneity in the protein environment of the reaction center to models based on protein fluctuations on the time scale of the electron transfer (Gehlen et al., 1994; Jia et al., 1994). A second application of fluorescence upconversion to photosynthesis is to the study of energy transfer in light harvesting systems. Energy transfer amongst differently oriented but otherwise similar or identical chromophores is conveniently monitored via fluorescence depolarization. In a series of polarized light experiments, Du and co-

Ultrafast spectroscopy in photosynthesis

workers (1993, 1994) have measured the depolarization of fluorescence from membranes of PSIonly and LHC-II-only strains of Chlamydomonas reinhardtii. The average single-step energy transfer in PSI was observed to occur in 200 fs. The depolarization process in LHC-II, on the other hand, was observed to be highly nonexponential, spanning 2 orders of magnitude from hundreds of femtoseconds to tens of picoseconds (see Fig. 6). As a final example of time resolved fluorescence studies of photosynthetic systems, we consider the energy transfer processes in bacterial light harvesting complexes. The purple photosynthetic bacteria contain two types of light harvesting complexes which channel excitations to the photochemical reaction center. In Rb. sphaeroides, these complexes are thought to be built from dimers of bacteriochlorophyll a, along with the carotenoid spheroidene. The LH-I complex is closely associated with the reaction center in a ringlike structure, and the LH-II complexes form pools which connect LH-I/RC structures.

69

Recently, depolarization measurements have resolved the single-step hopping of electronic excitation amongst the bacteriochlorophylls in LH-I (Branforth et al., 1994). The anisotropy of 940 nm fluorescence decays from a value around 0.4 to 0.1 with a 110 fs exponential time constant. The observed non-exponentiality may be a result of a distribution of energy transfer rates which result in a non-exponential depolarization process. As an example of an unusual anisotropy decay, Fig. 7 shows the decay of anisotropy from the bacteriochlorophyll fluorescence upon excitation of spheroidene in LH-I (Bradforth et al., 1994). This type of decay (starting near zero and going to a negative value) can result from energy transfer when the long axis of the spheroidene is tilted 30 degrees with respect to the normal of the plane containing the transition moments of two bacteriochlorophyll dimers. The projection of the spheroidene transition moment on the plane lies between the two dimer transition dipoles

70

Ralph Jimenez and Graham R. Fleming

IV. Transient Absorption

A. Experimental Technique Transient absorption methods are applicable to a far wider range of systems than fluorescence upconversion. The fact that several electronic states may contribute to the signal and that the temperature dependence of the signals can be rather subtle has recently been discussed by Jonas and coworkers (Jonas, 1994; Jonas et al., 1994). Conceptually, the experiment involves monitoring the intensity of a weak probe beam transmitted through a sample subsequent to the passage of a much stronger pump beam. The differential absorbance of the sample is monitored by a detector, and this signal is plotted as a function of delay between the pump and probe pulses. An experimental arrangement for pump-probe spectroscopy is shown in Fig. 8. The transmitted probe beam may be detected by a diode, or multichannel detection may be used in order to frequency resolve the signal. Lock-in detection may be used with high repetition rate sources, detection being referenced to a chopper placed in the pump beam. Experiments with low repetition rate sources cannot efficiently utilize lock-in detection, so this type of measurement can suffer from

high background levels. Shot-by-shot normalization of the signal to compensate for fluctuations in the laser power can also be employed, and is useful both for low repetition rate experiments, and for experiments in which a portion of white light continuum serves as the probe beam. Multicolor pump probe experiments require care to ensure that the pump and probe beams are focussed at the same spot in the sample. Pumpprobe experiments can easily be performed at low temperatures (with low repetition rate lasers), and in contrast to fluorescence upconversion, many studies have been reported at cryogenic temperatures.

Ultrafast spectroscopy in photosynthesis

71

B. Some Examples Extensive measurements of the rates of primary charge separation in reaction centers from Rhodobacter capsulatus have been performed at room temperature, and low temperatures, by Jia et al. (1994). A series of mutants was constructed in which the amino acid residues at sites L181 and M208 were modified in order to change the free energy of the electron transfer reaction. The redox potential of each mutant was measured electrochemically, and the rates of electron transfer were measured via monitoring the stimulated emission from P*. The fastest rate measured was for a mutant containing a tyrosine in place of the wild type phenylalanine at position L181 (< 3 ps) and the slowest rate was for the Thr(L181)– Thr(M208) mutant (around 30 ps). An example of their data is displayed in Fig. 9. The temperature dependence of the rates was also studied. A model based on a distribution of electron transfer rates arising from a distribution in free energy gaps was used to fit the data in order to estimate the reorganization energy arising from low-frequency protein and intermolecular modes coupled to the electron transfer. A remarkable series of pump-probe measurements by J.L. Martin and coworkers (Vos et al., 1993) suggests the need to re-examine standard

models of electron transfer for the primary charge separation step. Using the mutant of Rhodobacter capsulatus Martin and coworkers observed oscillations in the 77 K pump-probe signals (Fig. 10) which they interpret as arising from vibrational wavepackets in the excited state. These oscillations persist for many picoseconds, and indicate that vibrational coherence persists on the timescale of the primary charge separation. The mutant does not undergo electron transfer due to the absence of the acceptor pheophyin, but in more recent studies, the same authors have shown that these oscillations are present in pumpprobe signals from Rhodobacter sphaeroides reaction centers in their native membrane environment (Vos et al., 1994). Observation of an oscillatory contribution to the signal calls into question the conventional assumption that vibrational dephasing and relaxation occur on much shorter timescales than does the electron-transfer step. A full description of electron transfer incorporating a realistic model of the protein dynamics is still an active area of theoretical development (Jean et al., 1992; Skourtis et al., 1993).

72 V. Concluding Remarks Improved precision in ultrafast spectroscopy will allow more sophisticated data analysis such as singular value decomposition (Woodbury et al., 1994) to determine the number of components in a transient spectrum. With improved precision, decays will in many cases no longer be satisfactorily fit as single exponential processes and more complex models will be required to fit data and extract underlying system parameters. It seems likely that four and six wave mixing techniques such as the various echo methods (Nibbering et al., 1991; Cho and Fleming, 1993, 1994) will be increasingly applied to photosynthetic systems, to reveal dephasing timescales and the dynamics of spectral diffusion over wide dynamic ranges.

References Asaki MT, Huang CP, Garvey D, Zhou J, Kapteyn HC and Murnane MM (1993) Generation of 11-fs pulses from a self-mode-locked Ti:sapphire laser. Opt Lett 18: 977–980. Bradforth S, Jimenez R, Fidler V, Fleming GR, Nagarajan S, Norris J, van Mourik F and van Grondelle R (1994) Ultrafast energy transfer in the core light harvesting complex of photosynthetic bacterium Rhodobacter sphaeroides observed by fluorescence upconversion. In: Barbara PF, Knox WH, Mourou GA and Zewall AH (eds.) Ultrafast Phenomena Vol IX, pp 441–442. Springer-Verlag, Berlin. Bradforth S, Jimenez R, van Mourik F, van Grondelle R and Fleming GR (1995) Excitation transfer in the core light harvesting complex (LH1) of Rhodobacter sphaeroides: An ultrafast fluorescence depolarization and annihilation study. J Phys Chem 99: 16179–16191. Cho M and Fleming GR (1993) Photon echo measurements in liquids: Numerical calculations with model systems. J Chem Phys 98: 2848–2859. Cho M and Fleming GR (1994) Fifth-order three pulse scattering spectroscopy: Can we separate homogenous and inhomogenous contributions to optical spectra? J Phys Chem 98: 3478–3485. Diller R, Iannone M, Cowen BR, Maiti S, Bogomolni R and Hochstrasser RM (1992) Picosecond dynamics of bacteriorhodopsin, probed by time resolved infrared spectroscopy. Biochemistry 31: 5567–5572. Du M, Rosenthal SJ, Xie X, DiMagno TJ, Schmidt M, Hanson DK, Schiffer M, Norris JR and Fleming GR (1992) Femtosecond spontaneous emission studies of reaction centers from photosynthetic bacteria. Proc Natl Acad Sci USA 89: 8517–8521. Du M, Xie X, Jia Y, Mets L and Fleming GR (1993) Direct observation of ultrafast energy transfer processes in PS-I core antenna. Chem Phys Lett 210: 535–542.

Ralph Jimenez and Graham R. Fleming Du M, Xie X, Mets L and Fleming GR (1994) Direct observation of ultrafast energy transfer processes in light harvesting complex II. J Phys Chem 98: 4736–4741. Durrant JR, Knoester J and Wiersma DA (1994) Local energetic disorder in molecular aggregates probed by the one exciton to 2 exciton transition. Chem Phys Lett 222: 450– 456. Gehlen JN, Marchi M and Chandler D (1994) Dynamics affecting the primary charge transfer in photosynthesis. Science 263: 499–502. Huang CP, Asaki MT, Backus S, Murnane MM, Kapteyn HC and Nathel H (1992a) 17–fs pulses from a self-mode-locked Ti:sapphire laser. Opt Lett 17: 1289–1291. Huang CP, Kapteyn HC, McIntosh JW and Murnane MM (1992b) Generation of transform limited 32–fs pulses from a self-mode-locked Ti:sapphire laser. Opt Lett 17: 139– 141. Jean JM, Friesner RA and Fleming GR (1992) Application of a multilevel Redfield theory to electron transfer in condensed phases. J Chem Phys 96: 5827–5842. Jia Y, DiMagno TJ, Chan CK, Wang Z, Du M, Hanson DK, Schiffer M, Norris JR, Fleming GR and Popov MS (1994) Primary charge separation in mutant reaction centers of Rhodobacter capsulatus. J Phys Chem 97: 13180–13191. Jonas DM and Fleming GR (1994) Vibrationally abrupt pulses in pump-probe spectrsocopy. In: El Sayed MA (ed) Ultrafast Processes in Chemistry and Photobiology Blackwell Scientific Publications, Oxford. Jonas DM, Bradforth SE, Passino SA and Fleming GR (1994) Femtosecond wavepacket spectroscopy: Influence of temperature, wavelength, and pulse duration. J Phys Chem, in press. Joo T, Jia Y and Fleming GR (1995) Ti:sapphire regenerative amplifier for ultrashort high power multikilohertz pulses without external stretcher. Opt Lett 20: 389–391. Kahlow MA, Jarzeba W, DuBruil TP and Barbara PF (1988) Ultrafast emission spectroscopy in the ultraviolet by timegated upconversion. Rev Sci Instrum 59: 1098–1109. Maiti S, Cowen BR, Diller R, Iannone M, Moser CC, Dutton PL and Hochstrasser RM (1993) Picosecond infrared studies of the dynamics of the photosynthetic reaction center. Proc Natl Acad Sci USA 90: 5247–5251 . Martin JL, Breton J, Hoff AJ, Migus A and Antonetti A (1986) Femtosecond spectroscopy of electron transfer in the reaction center of the photosynthetic bacterium Rhodopseudomonas sphaeroides R26: direct electron transfer from the dimeric bacteriochlorophyll primary donor to the bacteriopheophytin acceptor with a time constant of 2.8 ± 0.2 psec. Proc Natl Acad Sci USA 83: 957–961. Nibbering ETJ, Wiersma DA and Duppen K (1991) Femtosecond non-Markovian optical dynamics in solution. Phys Rev Lett 66: 2464–2467. Norris TB (1992) Femtosecond pulse amplification at 250 kHz with a Ti:sapphire regenerative amplifier and application to continuum generation. Opt Lett 17: 1009–1011. Proctor B and Wise F (1993) Generation of 13–fs pulses from a mode-locked laser with reduced third-order dispersion. Appl Phys Lett 62: 470–472. Pschenichnikov MS, De Boey WP and Wiersma DA (1994)

Ultrafast spectroscopy in photosynthesis Generation of 13–fs, 5–MW pulses from a cavity dumped Ti:sapphire laser. Opt Lett 19: 572–575. Reed MK, Steiner-Shepard M and Negus DK (1994) Optical parametric amplification of white light continuum components at 250 kHz with a Ti:sapphire regenerative amplifier. In: Barbara PF, Knox WH, Mourou GA and Zewall AH (eds) Ultrafast Phenomena Vol IX, pp 215–216. SpringerVerlag, Berlin. Salin F, Squier J, Mourou G and Vaillancourt G (1991) Multikilohertz amplifier for high-power femtosecond pulses. Opt Lett 16: 1964–1966. Scherer NF, Jonas DM and Fleming GR (1993) Femtosecond wave packet and chemical reaction dynamics of iodine in solution: Tunable probe study of motion along the reaction coordinate. J Chem Phys 99: 153–168. Shah, J (1988) Ultrafast luminescence spectroscopy using sum frequency generation. IEEE J Quant Elect 24: 276–288. Skourtis SS, da Silva AJR, Bialek W, and Onuchic JN (1992) A new look at the primary charge separation in bacterial photosynthesis. J Phys Chem 96: 8034–8041. Sosnowski T, Klein PB, Norris TB, Bhargava RN and Gallagher D (1994) Femtosecond blue continuum generation and its application to the time-resolved study of emission in Mn-doped ZnS nanocrystals. In: Barbara PF, Knox WH, Mourou GA and Zewall AH (eds) Ultrafast Phenomena Vol IX, pp 389–390. Springer-Verlag, Berlin. Squier J, Korn G, Mourou G, Vaillancourt G and Bouvier

73 M (1993) Amplification of femtosecond pulses at 10 kHz repetition rates in Opt Lett 18: 625–627. Stanley R and Boxer SG (1995) Oscillations in the spontaneous fluorescence from photosynthetic reaction centers. J Phys Chem, 99: 859–869. van Grondelle R (1985) Excitation energy transfer, trapping and annihilation in photosynthetic systems. Biochim Biophys Acta 811: 147–195. Vos MH, Rappaport F, Lambry JC, Breton J and Martin JL (1993) Visualization of coherent nuclear motion in a membrane protein by femtosecond spectroscopy. Nature 363: 320–325. Vos MH, Jones MR, Hunter CN, Breton J, Lambry JC and Martin JL (1994) Coherent dynamics during the primary electron transfer in membrane bound reaction centers of Rhodobacter sphaeroides. Biochemistry 33: 6750–6757. Woodbury NW, Peloquin JM, Alden RG, Lin X, Lin S, Taguchi AKW, Williams JC and Allen JP (1994) The relationship between thermodynamics and mechanism during photoinduced charge separation in reaction centers from Rhodobacter sphaeroides. Biochemistry 33: 8101–8112. Wynne K, Reid GD and Hochstrasser RM (1994) Regenerative amplification of 30 fs pulses in Ti:sapphire at 5 kHz. Opt Lett 19: 895–898. Zhou J, Huang CP, Shi C, Murnane MM and Kapteyn HC (1994) Amplification of 21 fs pulses to the millijoule level. Opt Lett 19: 126–129.

This page intentionally left blank

Chapter 5 Data Analysis of Time-Resolved Measurements Alfred R. Holzwarth Max-Planck-lnstitut für Strahlenchemie; D-45470 Mülheim a.d. Ruhr, Germany

75 76 76 76 78 83 85 85 86 87 87 87 89 91 91

Summary I. Introduction II. Methods for Time-Resolved Data Analysis A. Single-Decay vs. Global Analysis Methods B. Data Fitting vs. Model Testing Methods C. Procedures for Optimization D. Importance of Proper Weighting Factors E. Importance of Error Analysis F. The Identifiability Problem III. Some Applications to Photosynthesis A. Purple Bacterial Energy Transfer B. Photosystem I Kinetics IV. Conclusions Acknowledgements References

Summary Advanced data analysis methods suitable for the analysis of kinetic spectroscopic data from complex chemical and biological systems are defined and reviewed. The aim of this chapter is on the one hand to introduce beginners dealing with complex data analysis problems into the matter. On the other hand it is intended as a relatively simple description, using only a minimum of mathematics, that should enable non-specialist readers from other fields to better understand the methods currently used to deal with complex kinetic problems. The first part introduces the basic concepts and terms by using a simple kinetic scheme as an example. The fundamentally different concepts of “mathematical data fitting” and “physical model testing” are defined and illustrated with some examples. Their capabilities and limitations are then discussed in detail. Further subsections deal with the mathematical optimization procedures, proper weighting of data points, and the importance of performing an error analysis. As a firsthand illustration to the use of these methods in current research problems some recent applications in the field of photosynthesis research are given. Abbreviations: DAS – decay–associated spectrum; LHI – B875 light-harvesting complex of photosynthetic bacteria; LHII – peripheral B800–850 light-harvesting complex of photosynthetic bacteria; PS – photosystem; RC – reaction center complex; SAAS – species-associated absorption spectrum; SAES – species-associated emission spectrum; SADS – species-associated absorption difference spectrum

Correspondence: Fax: 49-208-3063951; E-mail: [email protected]

75 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 75–92. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

76 I. Introduction The still increasing tendency to study more and more complex systems and/or phenomena with the help of time-resolved spectroscopic techniques can be observed in all areas of science, and photosynthesis research represents no exception. Using advanced time-resolved techniques researchers can now, often in combination with other methods, ask – and luckily answer in many cases – more sophisticated questions which are directed towards a better understanding of complex multicomponent systems and processes on a molecular level. The spectroscopic or other data sets acquired during such studies are generally large and have to be analyzed in terms of complex kinetic and/or other mathematical and physical models. Traditional data analysis methods prove increasingly insufficient to extract the maximal amount of relevant information contained in such data. The aim of this chapter is to give a brief overview on the powerful and sophisticated methods that have been developed over the recent years to deal with highly complex data analysis problems. The aim is actually twofold: Firstly, this chapter should clarify the terms and methods used in this area of research and thus provide a basis to enable non-specialist colleagues to read and better understand and judge papers on sophisticated kinetic measurements. Secondly, this contribution is aimed at scientists who want to learn about these methods and eventually apply them in their own research. Historically most of the methods described here have been developed in time-resolved fluorescence spectroscopy. All of these methods are quite general, however, and apply basically to all time-resolved and – with some appropriate minor modifications – also to all non-time-resolved spectroscopies and general data analysis problems. After the introduction and definition of terms like single decay analysis, global analysis, decay-associated and species-associated spectra, mathematical data fitting vs. physical model testing etc., some applications of these methods will be presented briefly. The examples presented have been chosen merely for their suitability to exemplify the principles and the capability of the discussed methods. They stem for the most part

Alfred R. Holzwarth from the author’s own research subjects, for the simple reason that an in-depth analysis of these problems has been carried out in the author’s laboratory. Within the limited space available for this more technically oriented chapter no attempt will be made to give rigorous proofs for the statements and relationships discussed. Rather the emphasis is on the principles. Numerous examples from other research groups who have applied similar methods could have been presented here. It is not possible nor intended, however, to give a complete overview of all papers where such techniques have been applied. The selected references cited should thus be taken merely as a guide for the more interested reader to go into the details of the methods.

II. Methods for Time-Resolved Data Analysis

A. Single-decay vs. Global Analysis Methods Suppose that one has measured some time dependent quantity (fluorescence intensity, absorption difference etc.) vs. time at a certain detection wavelength The signal is usually the response to an optical excitation by a brief exciting light pulse of wavelength If the system under study is a multicomponent one, as almost all biological systems and photosynthetic systems in particular one, the kinetics observed will in general depend both on the detection wavelength and on the excitation wavelength The time-dependent signal measured at one excitation/detection wavelength pair in a multicomponent system will not be sufficient to characterize the system fully. Suppose that the signal I(t) follows – which is true in many cases – a multiexponential law such that:

For a coupled N-component system or N-state system (a state could be an excited state of a chromophore or a group of chromophores, a radical pair state etc.) we have in general (without giving any proof) n = N, i.e. the number of kin-

Data analysis of time-resolved measurements

etic components (lifetimes) equals the number of (distinguishable) chemical components or states in the system. This is true if the kinetics of the system can be described in terms of a set of homogeneous first-order differential equations. How can we analyze the data from such a kinetics? In order to illustrate the procedures we will use in this and the following paragraphs the simple model example of a two-component system with reversible energy transfer (Fig. 1). Despite its simplicity this system is well-suited to allow us to understand the principles of the methods, define the terms used, and to understand the solutions. Extension to more complex systems is then straightforward. The solution of the kinetic equations for this system (Fig. 1) results in a biexponential decay

77

of the form given in Eq. (1) with n = 2. If we use the traditional way of analyzing each decay separately (single-decay analysis) we get a set of four parameters, two amplitudes and two lifetimes, for each experiment at a particular excitation/detection wavelength pair. One such experiment will not be sufficient to characterize the system. Carrying out the analysis for a set of M experiments will result in parameters for the whole data set, i.e. 2*M amplitudes and 2*M lifetimes. Inspection of the kinetic problem (Fig. 1) indicates that we expect only two lifetimes, however, which are expected to be the same at all excitation/detection wavelength pairs. This means that using conventional single-decay analysis we have determined more fitting parameters from the data set than required, since

78 we have ignored the fact that certain parameters (in this case the lifetimes) are connected or even constant across all individual (decay) experiments. This is the instant where “global lifetime analysis” becomes relevant. In “global analysis” one attempts to extract a single parameter set for all experiments from a combined analysis carried out in a single analysis run by taking explicit account of those parameters of the system that are connected across experiments. In our simple case the relationship of the lifetime parameters across experiments is a simple identity, i.e. and the same relationship hold for This is in fact the simplest form of a parameter connection across experiments but is also the most common one in time-resolved spectroscopy. It is important to realize that any parameter connection can in principle be achieved, provided it can be formulated in some mathematical terms, either analytically or numerically. The nature of this connection does not follow from the time-resolved experiment itself but must always be derived from the physics or chemistry underlying the problem at study. In our simple system (Fig. 1) the total number of parameters P determined from the full data set using global analysis is then while P for single-decay analysis is The latter is always larger than for M > 1. Thus global analysis allows one to extract, from the same data set, the same information with a smaller number of parameters. For an N-component system the relation will be M + N vs. Without giving any proof, we may expect as a consequence that global analysis will generally lead to i) a better accuracy in the values of the extracted parameters and thus ii) allow the analysis of more complex systems and/or more closely spaced lifetime components. How large is the improvement resulting from global over single-decay analysis? This is a difficult question that can not be answered in general. Rather the answer depends on the very details of the measurement. The degree of improvement can range from very large to actually non-existent in a few special cases. In many cases however the improvement is such that one can clearly state that the problem could be analyzed in a meaning-

Alfred R. Holzwarth ful way only by global analysis while single decay analysis fails completely. The important point here is that once the details of the experiment are known we can get a good estimate about the actual degree of improvement of global vs. single experiment analysis. For this purpose also numerical simulations of experimental data sets should be carried out taking into account the experimental details like S/N ratio, noise type on the data etc. (see below). Returning to our twocomponent system, Fig. 2 shows the improvement in the relative errors of the lifetimes. The figure shows that the degree of improvement will depend on the type of noise (i.e. constant, poissonian, signal-dependent etc. noise) which in turn depends on the type of experiment (transient absorption, fluorescence etc.) and on the difference in the spectra of the components. A similar relationship could be given for the error reduction in the amplitudes. The first application of global lifetime analysis to my knowledge has been reported by Knorr and Harris (1981). Later on the method has been further developed and applied to more complex problems by Brand and coworkers (Knutson et al. 1983; Beechem et al. 1984; 1985a). First applications in photosynthesis research were made soon thereafter (Wendler et al. 1986; Holzwarth et al. 1987; Wendler and Holzwarth, 1987). A general presentation of global analysis may be found in the publications of Beechem (1989) and Weidner et al. (1990), while the application of global analysis to photosynthesis problems has been discussed in part by Holzwarth (1987; 1988). Global lifetime analysis has been applied by a number of groups to photosynthesis problems, see e.g. Holzapfel et al. (1989); Finkele et al. (1992); Holzwarth et al. (1987); Roelofs et al. (1991); Roelofs et al. (1992); Mukerji and Sauer (1993). The method can be applied equally well without any essential modification to all kinds of kinetic analysis and applications to fluorescence kinetics, transient absorption kinetics, anisotropy decay analysis, to name just a few, and they have become quite popular.

B. Data Fitting vs. Model Testing Methods Global as well as single-decay analysis, as described in the previous paragraph, aims at an

Data analysis of time-resolved measurements

79

adequate mathematical description of the measured (kinetic) data. A mathematical model containing parameters is fitted to the data. Consequently the parameters resulting from such kind of analysis are primarily mathematical parameters which do not as such give direct information in terms of a physical/chemical model. Rather these fitting parameters – in most cases of time-resolved spectroscopy amplitudes and lifetimes, along with their temperature, wavelength, etc. dependence – have to be used to derive the parameters of real interest to the researcher, which I will call here the “physical parameters”. This situation is illustrated in Fig. 3 on the righthand branch. The purely mathematical “data fitting procedure” leads to the “mathematical parameters” that describe the kinetics. This is the point where data analysis often ends. Unfortunately, however, these “mathematical parameters” as such are hardly ever of interest to the researcher. This fact is generally not realized sufficiently well. We could, however, chose an entirely different approach. Rather than to first fit a purely mathematical function to the data and then try to indirectly calculate from those parameters the “physical parameters” of real interest, we could try to fit a “physical model” directly to the data. This “physical model” would have to be a mathematical description of the real physical/chemical behavior of the system and would contain the real “physical parameters” of interest, e.g. rate constants, activation energies, spectra, pK-values

80 and the like as fitting parameters. This procedure has several names in the literature like e.g. target analysis, compartmental analysis, kinetic modelling (Beechem et al., 1985b; Löfroth, 1986) depending on the application and also on the preferences of the authors. This variety in nomenclature is partly given by historic reasons and partly by some (minor) differences in the various methods. The underlying principle is always the same, however: Direct derivation of the physical/chemical parameters of interest from the original (lifetime) data rather than from any intermediate mathematical fitting parameters. This principle is illustrated in the left-hand branch of Fig. 3. What are the advantages of such methods over mere mathematical data fitting methods? These advantages are quite substantial and manifold. First, the direct method ensures that we try to extract from the original data set only the minimally required number of unknown parameters that are of real interest to understand the problem. The second advantage is even more important, however, and must be stressed strongly: It involves a drastic change in the “philosophy” of the data analysis such that mathematical data fitting is replaced by real “physical/chemical model testing” . Using these methods several alternative physical/chemical (kinetic) models can be tested directly on the original data. As a result a decision can be made which of these models fits the data best and which ones can be excluded on the basis of a particular data set. In practice then more than one model may remain that fits the data well. In such a case additional experiments or tests must be carried out in order to distinguish these models. Furthermore the physical parameters resulting from such model testing can then be judged using physical/chemical reasoning. For example any model, even if it fits the data formally well (see paragraph III for two examples) that would result in unrealistic values of physical parameters or violation of physical/chemical principles or laws, like e.g. negative rate constants, negative fluorescence spectra, or other unrealistic values could be excluded immediately as a reasonable description of the system. This possibility to definitely exclude certain models is a tremendous step forward over mere mathematical data fitting. Furthermore, the fact that the parameters of real physical interest are

Alfred R. Holzwarth fitted directly to the original data has the effect of structuring the multidimensional parameter space within which the best solution to the data fitting has to be found. This structuring has the effect that the physical parameters derived are obtained with both the highest possible accuracy (given the quality of the input data set) and with a realistic error margin (see below for error analysis). In practice several possible physical/chemical models, preferentially as many as one can think of, should be tested against the data. All this may sound rather theoretical so far. For this reason we will return to our simple example of a twocomponent mixture. Referring to Fig. 1 we can now define the terms decay-associated spectrum (DAS) and species-associated absorption/emission spectrum (SAAS/SAES). The following equations describe the relationships between these terms and between mathematical and physical parameters of the model: For the two-component mixture we would measure two DAS, and with corresponding lifetimes and (Fig. 1B). The DAS (amplitudes of Eq. (1) as a function of emission or excitation wavelength) and the lifetimes would result from a global data fitting within a sums of exponentials model. They do not represent the physical parameters of interest, however. Rather the physical parameters of interest are the SAES or SAAS (Fig. 1B), which actually are the real emission or absorption spectra, respectively, of the two species, and the rate constants The following equations show the relationships between these parameters (in case of a two-component system):

It follows from (Eq. (2)) that the DAS are generally linear combinations of the SAES (or SAAS if the measurement is carried out as a function of the excitation wavelength) with coefficients Furthermore, the lifetimes are functions only of the rate constants and the kinetic connecting scheme (See Fig. 1a) of the system under study. The coefficients for the SAES are complicated

Data analysis of time-resolved measurements functions of both the rate constants, the excitation wavelength, and the extinction coefficients of the species involved (Eq. (3)). The fluorescence decay is a linear combination of the exponential terms with the DAS as weighting factors (amplitudes). These equations correspondingly hold for transient absorption measurements if the SAES are replaced by the corresponding species-associated absorption difference spectra (SADS). The relationships defined in Eq. (3) for the example of the two-component mixture with energy transfer can be generalized in closed form in the following way for all kinetic systems that can be described by a set of homogeneous first order differential equations. Using the following definitions: X(t) = (n × 1) vector of concentrations of (excited state) species. time derivatve of X(t) T = (n × n) transfer matrix describes connectivity of system.

vector of time zero absorbances of species. l(t) = excitation function. We can then write the following equation for the time dependence of fluorescence (or likewise transient absorption ) spectra:

with eigenvalue of matrix T. U = matrix of eigenvectors of T. n = number of (excited state) species k. species-associated emission (SAES) or difference spectrum (SADS) of species. These equations fully describe the relationship between the measurable decay data as a function of excitation wavelength emission or (in transient absorption) detection wavelength the kinetic scheme (transfer matrix T) and the rate constants of the system on the one hand and the species-associated spectra on the other hand.

81 Also the relationship between the DAS and lifetimes (the mathematical parameters) and the physical parameters become clear from these equations. The relationship between the eigenvalues and the measurable lifetimes is The relation given in Eq. (3) is simply a special case of the closed and generalized form given in Eq. (5). This equation can be fitted directly to a time-resolved data sets after a specific kinetic model has been chosen. In this way the rate constants, the SAES or SADS and, under certain conditions, also the species-associated absorption spectra (SAAS) can be obtained from the data. There is one important restriction, however: The so-called identifiability conditions (see below) must be analyzed and obeyed since they might limit the number of independent physical parameters that can be fitted simultaneously. The methods outlined above are discussed in detail in several papers (Beechem et al. 1985b; Ameloot et al. 1986). Some applications to photosynthesis may be found elsewhere in (Holzwarth, 1990; Roelofs et al. 1992; Roelofs and Holzwarth, 1990; Müller et al. 1991; 1992; 1993). As an important extension to the basic kinetic or target analysis outlined above, further restrictions with respect to the values and range or the mutual relationships of the physical fitting parameters can be implemented easily if such information is available. It could come either from other measurements or could be introduced deliberately in order to test certain restricted models. The advantage is that, without any loss of the benefits of target analysis mentioned above, the general procedure is rather flexible and can be adapted easily even to highly complex situations, provided a mathematical description can be formulated. This mathematical description does not have to be given in a closed analytical form but can even be applied in a numerical form. It also follows easily that the formulation of the physical/chemical model to be tested against the data in addition allows one to put in all available other information on the system. This additional information may be spectroscopic, structural, thermodynamic or other in nature. In this sense the physical model testing represents a truly holistic approach. Fig. 4 shows an approach often used in the socalled compartmental analysis of data repre-

82

senting antenna energy transfer/charge separation processes. Due to the complexity of an antenna/reaction center complex certain simplifying assumption must be made. The level of approximation can be chosen, depending on the level of detail (time-resolution and completeness) present in the data. At one extreme level all chromoph-

Alfred R. Holzwarth ores and reaction center states could be taken into account explicitly. This would involve the full kinetic description of the system. However the measurable data hardly ever allow, except in the most simple systems like e.g. phycobiliproteins (Suter and Holzwarth, 1987; Holzwarth et al. 1987) such detail of analysis. One is thus forced to make simplifying assumptions in the definition of how many distinguishable compartments (a compartment is a “quasi-state”) should be or need to be taken into account in the model. Within certain limits it is up to the researcher how many compartments he wants to define and consider in the analysis. How complex or simple a “compartment”, which in the optimal case should encompass a group of chromophores which show similar kinetics and spectra, can be defined, will depend mostly on the amount and precision of data that is available (see e.g. Holzwarth et al. 1987). The proper “compartmentalization” of the problem at hand has to be found out by testing various kinetic models on the data and involves subsequent checking of the physical parameters for their internal consistency and physically reasonable behavior. The well-known exciton/radical pair equilibrium model for photosystem II and purple bacterial antenna/RC complexes (Schatz et al., 1988) is a typical case of a compartment model and has been used extensively in the past (see van Grondelle et al., 1994 for a review). It is important to note that in all cases where a full kinetic treatment (taking into account all chromophores or states separately) is not possible, the proper or “reasonable” compartmentalization of the system tells us a great deal about the physical nature and behavior of the system. Since any compartmentalization of a system represents an approximation, a compartment model can only answer a limited number of questions about a system. Thus as the most stringent restriction, no information can be obtained directly on the processes going on within a compartment, which in essence is considered a “black box” with input and output only. For example in the case of the exciton/radical pair equilibrium model, no information is obtained on the energy transfer rates between the antenna chromophores. Rather it is assumed that these rates, and thus the antenna excited state equilibration within a compartment, must be fast in relation to the other

Data analysis of time-resolved measurements rates in the system. If this assumption is not valid the model is not applicable and must be extended by increasing the number of compartments. In this discussion in has already been assumed implicitely that usually a physical model testing approach also involves and implies a global analysis, i.e. the physical model is fitted to a large set of decay data rather than a single decay, although the latter is not entirely excluded. However the amount of information contained in a single decay curve is usually not sufficient to solve more than the most simple kinetic models. It has become clear that any “target” or “compartment” analysis requires a detailed physical/chemical (kinetic) model before the analysis can start, in contrast to global or single-decay “lifetime fitting”. It is thus often erroneously assumed that the outcome of the analysis is already determined by the assumptions that have been put into the physical model. This is fortunately not true, however. As pointed out above, the model is tested against the data, rather than assumed as being correct. The testing of course has to be done with several different and alternative models if meaningful results are to be expected. This procedure then answers the question whether a certain model fits the data or not. It can not positively prove whether a model is the correct one. In practice several different models may turn out to provide an equally good fit to the data. It is important to realize that the “model testing approach” delivers qualitatively different answers than the mathematical fitting approach. In mathematical fitting one can usually always find a good fit, provided one allows for a large enough number of parameters. In physical model testing this is not the case.

C. Procedures for Optimization Nothing has been said so far about the mathematical procedures suitable to find the “optimal” values for the parameters of a given mathematical or physical model. For all kinetic problems the model equations are non-linear. Thus we require non-linear optimization methods for the optimal parameter estimation. Such optimization procedures are the subject of an extended area of numerical mathematics and no attempt will be made here for an extensive in-depth presentation. The

83 principal procedures may be found in monographs on numerical data analysis (see e.g. Bevington, 1969). Rather some general aspects will be discussed here. First of all it is not necessary to make any distinction between mathematical data fitting and physical/chemical model testing, since the mathematical procedures for parameter optimization are the same for both approaches. The first question to be answered is what we define as an “optimal” value for a fitting parameter. The objective is to find the parameter value that has the maximum likelihood of being correct in a statistical sense for the data set given. For practical purposes of interest we can say that the maximum likelihood of a parameter value can be obtained by minimizing the quadratic deviations of the fitting function from the data. Taking into account the non-linear nature of the problems this is called non-linear least squares minimization. Although in a strict mathematical sense the assumption that the maximum likelihood value of a parameter is found by a least squares minimization is not generally valid, this nevertheless is a useful and practical definition. We thus define a function (called the reduced which has to be minimized using appropriate methods in order to determine the optimal values for the elements of the parameter set

In this equation the double summation runs over all experiments and all data points of an individual experiment. The function denotes the value of the fitting function for a particular parameter set at the position qi while denotes the standard deviation from the mean of the particular data point N is the total number of data points in the fit, while m denotes the total number of independent fit parameters. Numerous algorithms exist for the minimization of the least squares function and the reader is referred again to numerical mathematics and data analysis text books (Bevington, 1969). These methods are usually iterative methods, i.e. one starts with some initial estimate for a parameter

84

value and computes in a next step corrections this parameter set such that the decreases. This operation in repeated until the minimal value has been found, i.e. until the method converges. The most common procedure is the Levenberg–Marquardt procedure (Marquardt, 1963) which involves a combination of the well-known Gauss–Newton method and the steepest-descent method. The steepest descent method only determines the slope of the with respect to the uncorrelated parameters Thus it has the advantage of converging quickly if the parameter estimate is far from the optimum where the slope of the surface is large. Coming closer to the minimum, this procedure either fails or at least shows very poor convergence. The Gauss–Newton method on the other hand takes into account the correlation between the different parameters. It converges well if close to the minimum, but has difficulties if the solution is still far from the minimum. Combining the two methods and smoothly switching from steepest descent to Gauss–Newton as the minimum is approached is achieved by the Levenberg–Marquardt method. Thus it combines the advantages of both simpler methods while avoiding their difficulties. An intricate problem involved in most minimization methods is illustrated in Fig. 5 for a singleparameter case. Starting at some point on the surface the minimization procedure should allow one to find the optimal parameter value at the

Alfred R. Holzwarth absolute (global) minimum of the (remember that the in general is a hypersurface of dimension p + 1 when p is the number of parameters). If the is smooth, the minimal value can usually be found without difficulties. If there are local minima on the way however (see Fig. 5), the minimization algorithm might well converge into such a local minimum from where it can not recover. This is true for essentially all non-linear minimization algorithms that are based on derivatives of the or the fitting function. Worst of all, for a complex multidimensional the chances of falling into a local minimum may be quite high. Fig. 5 also illustrates that falling into a local minimum might be prevented by the choice of different starting points for the iteration. This usually helps for fitting problems with a relatively small number of parameters but becomes increasingly difficult and computer time-consuming for a large parameter set. One has to realize that in practice one can not have the guarantee to find the local minimum of a complex by using any of the iterative methods and one has to be well aware of that problem. A vast literature exists on the use of methods other than the Levenberg–Marquardt method or even non-least-squares methods in the analysis of kinetic data (Jennrich and Ralston, 1979; Livesey et al., 1988). Despite many claims to the contrary it is however fair to say that, with the exception of the local minimum problem, which is however inherent to the other methods as well, in the overwhelming number of cases the Levenberg– Marquardt represents a well-suitable and reliable method for non-linear parameter optimization. Thus no really good justification exists to apply other methods except in cases where the local minimum problem becomes severe. If this is the case iterative minimization schemes based on derivatives of the or other fitting functions have to be abandoned entirely. A new and powerful method has been developed that does not suffer from the local minimum problem. This is the so-called group of “genetic code algorithms” (Goldberg and Richardson, 1987; Goldberg et al., 1989). To my knowledge so far hardly any applications of these algorithms to data analysis in time-resolved spectroscopy exist (see e.g. Trinkunas and Holzwarth, 1994b) but an increas-

Data analysis of time-resolved measurements ing number of applications are anticipated in the near future.

D. Importance of Proper Weighting Factors Eq. (6) contains the standard deviation of the data point qi. The inverse standard deviation in this equation provides the weighting factor with which the particular data point qi contributes to the The standard deviation is a measure of the noise in the data. Often in nonlinear fitting problems this weighting factor is set to one, resulting in an equal weighting of all data points and thus ignoring the fact that experimental points in general have different accuracy. It is highly important to note that by doing so one deliberately introduces a systematic error into the non-linear parameter optimization resulting in poor convergence, poor accuracy and in many cases even wrong results. In any case the capability to resolve complex fitting problems is impeded. Therefore the use of proper weighting factors in the definition of the (Eq. (6)) is absolutely required for a correct data analysis. The problem is serious already for a single decay but becomes even more severe if multiple experiments are being combined in a global analysis scheme. The interested reader is referred to the article by di Cera (1992). For practical purposes of fluorescence and transient absorption spectroscopy the standard deviations are often easily defined. In a singlephoton counting experiment the noise is of Poissonian nature and thus the standard deviation is given by the square root of the number of counts in a channel. In a transient absorption experiment measuring small absorption differences, the standard deviation is usually the same for all data points in a single experiment (constant noise independent of the signal). In the latter case the weighting could be ignored if analyzing a single experiment only. However if several experiments are being combined in a global analysis each of them might have a different standard deviation and the proper weighting factors must again be applied. For other kinds of experiment often the noise on the data is not so easily known. In order to get the proper weighting factors one must try to get as close an estimate of the standard deviation of the measured signals however.

85 Finally it is noted without proof here that the type of noise present in a signal (poissonian, constant noise, signal-dependent etc.) determines and limits the degree of improvement that can theoretically be achieved by the use of global analysis (including global physical model testing approaches) (see Fig. 2). This is quite generally the case and is independent of the amount of noise that is present. In any case, however, the maximal improvement can only be expected if a proper weighting scheme is applied.

E. Importance of Error Analysis Surprisingly, even in the most sophisticated data analysis the error analysis is often ignored entirely or is carried out on a minimal and thus often useless level. This happens despite the fact that it is well-known that a value of a measured or derived quantity is only as good as the knowledge about its confidence interval. The reason for this negligence is that in most cases the estimation of a reliable and realistic confidence interval of a parameter in a complex multiparameter model can easily consume more manpower and computer resources than the actual parameter optimization itself. Given this situation any systematic attempt of getting even a sub-optimal estimate of a confidence interval is better than none. Very reliable methods exist, among which the Monte Carlo method is the most prominent (Straume et al., 1992). However the tremendous amount of computer time that it requires usually makes it a non-method for the error analysis of complex models. Without spending very much additional computing time the matrix procedures involved in the Levenberg–Marquardt method allow one to calculate the so-called variance–covariance matrix (Johnson et al., 1992) and get the confidence intervals from its diagonal elements. The problem is, however, that the confidence intervals estimated in this way are usually by far too small since the correlation between different parameters is not taken into account properly. Furthermore the confidence interval is assumed to be symmetric to both sides of a parameter value which is usually not the case in a non-linear model. Quite on the contrary, the confidence interval will in general be highly non-symmetric about the most likely value (Johnson, 1983). A

86 much better, and from the point of computer time still acceptable, method is the exhaustive search method. This method searches the form of the surface in the immediate vicinity of the optimal values of all parameters. This is done by stepwise carrying away a parameter from its optimal value and, using the same algorithm as has been used for the optimization, to minimize the by letting the other parameters compensate the disturbance introduced by carrying away one parameter from its optimal value. This procedure is repeated several times for different values of a particular parameter, making deviation to both sides of the optimal value. As a criterion for the maximal deviation one defines a certain absolute increase in the value of the This procedure is carried out successively for each parameter. The estimates to both sides of the optimal value where the has increased by the same pre-defined amount then define the confidence interval for that parameter. The advantage of this procedure is that is can be implemented easily without writing much new code and that it takes full account of the correlations between parameters. Such an analysis has been carried out by Roelofs et al. (1992) for the case of a target analysis of a fluorescence lifetime experiment on higher plant thylakoids. The outcome showed that the confidence interval for certain parameters was by orders of magnitude larger than the error estimate from the variance–covariance matrix and that it was in fact highly non-symmetric. The analysis also showed large differences in percentage error between different parameters. This may serve as an example that error analysis is indeed important and essential for distinguishing different models on the basis of a given set of experiments. Additional methods are described by Johnson et al. (1992).

F. The Identifiability Problem The advantages of physical/chemical model testing vs. mathematical data fitting have been pointed out in paragraph II.B of this chapter. There is one important limitation however, that has not been discussed sufficiently so far, i.e. the socalled identifiability problem. This deals wilth the possibility that a given set of experimental data might not contain enough information to extract

Alfred R. Holzwarth all the physical parameters that are contained in a physical model to be tested. It is important to realize that this has nothing to do with the accuracy of the data set (a low accuracy would simply lead to a larger confidence interval). Rather this is an inherent mathematical or information theoretical problem. Let us take our simple example of a two-component mixture with energy transfer. Suppose we have measured the fluorescence kinetics as a function of emission wavelength. We now want to extract the rate constants and the SAES by kinetic modelling. The model contains four rate constants and two SAES. Even in this simple case not all four rate constants can be determined but only three of them. The other one must be fixed on some assumed or otherwise determined value. It is easy to see even for the mathematically less experienced reader why this limitation exists. The experiment determines only two lifetimes and, at each detection wavelength, two amplitudes. Since the fluorescence amplitudes are usually not measured as absolute values but only as relative numbers (due to the presence of unknown instrumental proportionality factors), for each detection wavelength there exist only three independent parameters, two lifetimes and one (relative) amplitude. The other amplitude is a dependent one, i.e. it is the difference from 100%. Thus not enough linearly independent equations exist in order to determine four rate constants. If one were to allow for freely adjustable rate constants in the fit, a mathematical solution (good fit) would definitely result. However, the fit would not be unique and three of the rate constants would be parametrically dependent on the fourth one. This means that the system is underdetermined and the fitting result in the rate constants would be to a large part arbitrary (though entirely within a mathematically good description of the data set). This may serve as an example that it is absolutely necessary in all cases of model testing to carefully investigate the identifiability conditions and to avoid carrying out a target analysis with an underdetermined model. For more complex models the identifiability analysis may be a tedious and not easily accomplished task. The space available here is not sufficient however to discuss this problem in more detail and the reader is

Data analysis of time-resolved measurements referred to the papers by Ameloot et al. (1986; Ameloot and Hendrickx, 1982) instead. III. Some Applications to Photosynthesis We will now give some practical examples from photosynthetic systems that are meant to provide more insight into the capability of the procedures and the kind of answers that may be expected. The data sets underlying the two examples given here have been analyzed thoroughly using both global lifetime analysis as well as global target analysis methods. Both examples have in common furthermore that only target analysis was capable of determining the underlying mechanisms and physical models properly. In the first example on the basis of global lifetime analysis alone one would have arrived at a completely wrong mechanism.

A. Purple Bacterial Energy Transfer The antenna systems of many purple bacteria consists of two pools, i.e. the inner core antenna pool B875 (LHI) and the peripheral antenna pool B800–850 (LHII) (van Grondelle et al., 1994) . Based mainly on transient absorption studies a sequential scheme of energy transfer (see Fig. 6 scheme I) had been proposed in the literature (Sundström et al., 1986). A key feature of this model is a quite slow energy transfer of about 35–40 ps between the LHII and the LHI pools. The global lifetime analysis of fluorescence lifetime data was in good agreement with such a model (Müller et al., 1993). Interestingly the global target analysis, based on this model, gave a perfect fit to the data as well (Müller et al., 1993). However, inspection of the resulting rate constants for the sequential model (Scheme I, Fig. 6) revealed that this kinetic model was entirely meaningless in physical terms since the ratio This ratio would have violated strongly the Boltzmann principle, given the fact that there are about 24 B875 BChls per P860. A ratio of about 1/30 was expected instead on the basis of the Boltzmann factor, i.e. about three orders of magnitude less than found in the target analysis of scheme I (Müller et al., 1993). We thus have the interesting situation that the target analysis of one particular model indicates perfect

87 fit with the data. Nevertheless the model is wrong. This conclusion could be arrived at only upon careful consideration of physical principles not explicitly imposed in the kinetic model (other models were also tested but were formally not in good agreement within the limitations of the target analysis). Thus a different kinetic scheme (shown in Fig. 6, scheme II) had to be searched for which would describe the data set well. Scheme II represents a heterogeneous model, assuming that there are two types of differently behaving reaction centers but the same antenna system. One such possibility was that the two types of reaction center would undergo charge separation with different rates, i.e. open and reduced closed reaction centers. This model (Müller et al., 1993) explained the data perfectly well without leading to any conflicts with other physical principles. Furthermore it is also in good agreement with other knowledge about this system. Note that in addition the two parts of the parallel scheme II are linked by assuming equation rate constants for transfer (rate constants This is also reasonable since any change in the redox state of the RC is not expected to influence the antenna energy transfer rates. This application of target analysis and critical checking of its predictions were essential to exclude the wrong model.

B. Photosystem I Kinetics Fig. 7 shows different compartment models (compartments are antenna pools 1–3) for the core antenna (RC complex of photosystem I from cyanobacteria (the same schemes should hold in principle for the corresponding higher plants core antenna complex) (Holzwarth et al. 1993). The kinetic models 7A and 7C contain two antenna pools each. However the RC and thus charge separation is coupled to these pools in two different ways: in 7A the RC is strongly coupled with (and is in fact part of) the small pool of longwave (red) Chl a, while in model 7C it is coupled to the main pool. Model 7B contains 3 antenna pools. This model, although theoretically reasonable, was already too complex given the details and resolution of the experimental data sets (two lifetime components were found by global lifetime analysis in fluorescence and three in transi-

88

ent absorption). For this reason model 7B had to be discarded from the analysis since it predicts at least 3 lifetime components for fluorescence and 4 components for transient absorption. However at a somewhat higher time resolution of the experimental data one could try to test this model again. Both models 7A and 7C were tested extensively on the fluorescence and the transient absorption data set. Again both models gave a (mathematically) perfect fit to the data. The resulting SAES (from the fluorescence data) and the SADS (from the transient absorption data) are shown in Fig. 8 for each model. The fluorescence data result in two SAES, one each for each antenna pool, and the transient absorption data in three SADS, one for each antenna pool and an additional one for the primary radical pair. The latter cannot be observed in fluorescence. Clearly only one of these models could be correct. Further analysis of the SAES and SADS revealed that, despite the good fit for both models, in fact only model 7A was physically reasonable while model 7C had to be discarded. The

Alfred R. Holzwarth

reason for this is that in the definition given in Eq. (5) the area under an SAES is directly proportional to the radiative rate constant of the corresponding excited state. Since in PS I only Chls a emit, the areas should be quite similar. Likewise the SADS give directly the excitedground state difference extinction coefficients. Again these should be roughly similar in strength for the various excited Chl pools. These conditions are fulfilled only for model 7A but not for 7C, however, as can be seen from Fig. 8. Further analysis also revealed that model 7C would violate the Boltzmann principle. In further analysis data about the pool sizes could be estimated from the target data which were in very good agreement with a decomposition of the absorption spectrum (Holzwarth et al., 1993; Trinkunas and Holzwarth, 1994a). We are thus again faced with a situation where more than one model gives a perfect fit even in target analysis. However only one of the models is physically reasonable. Such a situation is quite common in such kind of analysis. In contrast glo-

Data analysis of time-resolved measurements

bal lifetime analysis could not distinguish between these cases. This demonstrates clearly the strength of global target analysis. IV. Conclusions The principle and some applications of various kinds and levels of data analysis of time-resolved data have been presented and some of the pitfalls have been pointed out. It has been demonstrated that the physical interpretation of a given data set may very well depend on the proper application of sophisticated data analysis procedures. The importance of such methods will definitely grow with the complexity of the systems under study and the increasing detail at the molecular level at which answers are being sought.

89 Questions are often asked by novices in the field of data analysis: How many lifetime components can be separated, where is the limit, are you not trying to resolve too many components? These questions often imply that any analysis in terms of more than two or, in rare cases, perhaps three lifetime components is either impossible or would involve even some kind of unreliable “guessing”. The number of lifetime components that can be separated depends on many parameters and details of both the system under study as well as on the details of the experiment. Thus a general answer to that question can not be given! Rather any specific answer would be limited to a particular case and can only be given after a careful consideration of the details of the experiment and the knowledge of which different kind of models one would like to or ought to be able to distinguish. One important consequence of this situation is that it is advantageous to use the knowledge about the details of various data analysis methods already in the planning of an experiment in order to ensure optimal results. Furthermore it is always wise in case of a complex system to perform numerical simulations for various situations, thus trying to match as closely as possible the simulated data to the experimental ones. Then the same analysis methodology should be used to estimate the parameters and (equally important) their errors for both the experimental and the simulated data sets. Doing so one gains a detailed knowledge about the capabilities and limitation(s) of the applied procedures for that particular case. There is of course absolutely no guessing involved in these procedures, as is occasionally implied by novices. If the principles and rules are followed properly, knowing the limitations, the results are absolutely reliable. It is furthermore fair to note that the principal procedures described here are applicable as well to data analysis problems outside kinetic spectroscopy. Finally I should like to mention a couple of important additional papers from research areas outside photosynthesis where similar analysis procedures as treated here have been described and/or applied in a highly fruitful fashion (Nagle, 1991; Beechem, 1988; Lakowicz et al. 1984; Beechem et al. 1991; Christian et al. 1981; Halvorson 1981; Hessling 1992). The interested reader is referred

90

Alfred R. Holzwarth

Data analysis of time-resolved measurements to these papers if he/she wants to get more deeply involved in global data analysis procedures.

Acknowledgements I would like to thank all my colleagues who have over the past years worked with me and who have been involved in the development and the application of the data analysis procedures described here. Of those I would particularly like to mention Mrs. Iris Martin, a mathematician and computer programmer who has worked closely with me in the development of particular data analysis code, and Dr. Marc Müller who has spent much time on the computer developing and testing many complex data analysis cases and finding out about their capabilities and limitations.

References Ameloot M and Hendrickx H (1982) Model building in fluorescence relaxation experiments. Arch Int Physiol Biochim 90:46–47. Ameloot M, Beechem JM, and Brand L (1986) Compartmental modeling of excited-state reactions: Identifiability of the rate constants from fluorescence decay surfaces. Chem Phys Lett 129:211–219. Beechem JM (1989) A second generation global analysis program for the recovery of complex inhomogeneous fluorescence decay kinetics. Chem Phys Lipids 50:237–251 . Beechem JM, Knutson JR, and Brand L (1984) Simultaneous analysis of multiple fluorescence emission anisotropy decays. Biophys J 45:A127. Beechem JM, Ameloot M, and Brand L (1985a) Global analysis of fluorescence decay surfaces: Excited-state reactions. Chem Phys Lett 120:466–472. Beechem JM, Ameloot M, and Brand L (1985b) Global and target analysis of complex decay phenomena. Anal Instrum 14:379–402. Beechem JM, Gratton E, Ameloot M, Knutson JR, and Brand L (1991) The Global Analysis of Fluorescence Intensity and Anisotropy Decay Data – 2nd-Generation Theory and Programs. In: Lakowicz JR (ed) Topics in Fluorescence Spectroscopy. 2, pp 241–305, Plenum Press, New York. Bevington PR (1969) Data reduction and error analysis for the physical sciences. McGraw Hill, New York, pp 1–336. Christian GD, Callis JB, and Davidson ER (1981) Array detectors and excitation-emission matrices in multicomponent analysis. In: Wehry EL (ed) Modern Fluorescence Spectroscopy 4, Plenum Press, New York, pp 111–166. Di Cera E (1992) Use of weighting function in data fitting. In: Brand L and Johnson ML (eds) Methods in Enzymology, Vol 210 Numerical Computer Methods, Academic Press, San Diego, pp 68–87. Finkele U, Lauterwasser C, Struck A, Scheer H, and Zinth

91 W (1992) Primary electron transfer kinetics in bacterial reaction centers with modified bacteriochlorophylls at the monomeric sites BA,B. Proc Natl Acad Sci USA, 89:9514– 9518. Goldberg DE and Richardson J (1987) Genetic algorithms with sharing for multimodal function optimization. Genetic algorithms and their applications, Proc 2 Int Conf on Genetic Algorithms: 41–49. Goldberg DE, Korb B, and Deb K (1989) Messy genetic algorithms: Motivation, analysis, and first results. Complex Systems 3:493–530. Halvorson, HR (1981) Determining the number of interacting species: Significant factor analysis. Biophys Chem 14:177– 184. Hessling, B, Souvignier, G, and Gerwert, K (1992) A New Approach to Analyse Kinetic Data of Bacteriorhodopsin, Factor Analysis and Decomposition. In: Structures and Functions of Retinal Proteins 221:155–158 . Holzapfel W, Finkele U, Kaiser W, Oesterhelt D, Scheer H, Stilz HU, and Zinth W (1989) Observation of a bacteriochlorophyll anion radical during the primary charge separation in a reaction center. Chem Phys Lett 160:1–7. Holzwarth AR (1987) A model for the functional antenna organization and energy distribution in the photosynthetic apparatus of higher plants and green algae. In: Biggins J (ed) Progress in Photosynthesis Research. 1, Nijhoff Publishers, Dordrecht, pp 53–60. Holzwarth AR (1988) Time resolved chlorophyll fluorescence. What kind of information on photosynthetic systems does it provide?. In: Lichtenthaler HK (ed) Applications of Chlorophyll Fluorescence, Kluwer Academic Publishers, Dordrecht, pp 21–31. Holzwarth AR (1990) The functional organization of the antenna systems in higher plants and green algae as studied by time-resolved fluorescence techniques. In: Baltscheffsky M (ed) Current Research in Photosynthesis. II, Kluwer Academic Publishers, Dordrecht, pp 223–230. Holzwarth AR, Wendler J, and Suter GW (1987) Studies on chromophore coupling in isolated phycobiliproteins. II Picosecond energy transfer kinetics and time-resolved fluorescence spectra of C-phycocyanin from Synechococcus 6301 as a function of the aggregation state. Biophys J 51:1– 12. Holzwarth AR, Schatz G, Brock H, and Bittersmann E (1993) Energy transfer and charge separation kinetics in photosystem I: 1 Picosecond transient absorption and fluorescence study of cyanobacterial photosystem I particles. Biophys J 64:1813–1826. Jennrich RI and Ralston ML (1979) Fitting nonlinear models to data. Ann Rev Biophys Bioeng 8:195–238. Johnson ML (1983) Evaluation and propagation of confidence intervals in nonlinear, asymmetrical variance spaces. Analysis of ligand-binding data. Biophys J 44:101–106. Johnson ML and Faunt LM (1992) Parameter estimation by least-squares methods. In: Brand L and Johnson ML (eds) Methods in Enzymology 210 Numerical Computer Methods. Academic Press, San Diego, pp 1–37. Knorr FJ and Harris JM (1981) Resolution of multicomponent fluorescence spectra by an emission wavelength-decay time data matrix. Anal Chem 53:272–276.

92 Knutson JR, Beechem JM, and Brand L (1983) Simultaneous analysis of multiple fluorescence decay curves: A global approach. Chem Phys Lett 102:501–507. Lakowicz JR, Laczko G, Cherek H, Gratton E, and Limkeman M (1984) Analysis of fluorescence decay kinetics from variable frequency phase shift and modulation data. Biophys J 46:463–477. Livesey AK and Brochon J-C (1988) Maximum entropy data analysis of dynamic parameters from pulsed-fluorescent decays. In: Harding SE and Rowe AJ (eds) Dynamic Properties of Biomolecular Assemblies, Royal Society of Chemistry, pp 135–147. Löfroth J-E (1986) Time-resolved emission spectra, decayassociated spectra, and species-associated spectra. J Phys Chem 90:1160–1168. Marquardt DW (1963) An algorithm for least-squares estimation of nonlinear parameters. J Soc Ind Appl Math 11:431–441. Mukerji I and Sauer K (1993) Energy transfer dynamics of an isolated light harvesting complex of photosystem I from spinach: Time-resolved fluorescence measurement at 295K and 77K. Biochim Biophys Acta 1142:311–320. Müller MG, Griebenow K, and Holzwarth AR (1991) Primary processes in isolated photosynthetic bacterial reaction centers from Chloroflexus aurantiacus studied by picosecond fluorescence spectroscopy. Biochim Biophys Acta 1098:1– 12. Müller MG, Griebenow K, and Holzwarth AR (1992) Primary processes in isolated bacterial reaction centers from Rhodobacter sphaeroides studied by picosecond fluorescence kinetics. Chem Phys Lett 199:465–469. Müller MG, Drews G, and Holzwarth AR (1993) Excitation transfer and charge separation kinetics in purple bacteria: 1. Picosecond fluorescence of chromatophores from Rhodobacter capsulatus wild type. Biochim Biophys Acta 1142:49–58. Nagle JF (1991) Solving complex photocycle kinetics. Theory and direct method. Biophys J 59:476– 487. Roelofs TA and Holzwarth AR (1990) In search of a putative long-lived relaxed radical pair state in closed photosystem II. Kinetic modeling of picosecond fluorescence data. Biophys J 57:1141–1153. Roelofs TA, Gilbert M, Shuvalov VA, and Holzwarth AR (1991) Picosecond fluorescence kinetics of the D1–D2– cyt-b559 photosystem II reaction center complex. Energy

Alfred R. Holzwarth transfer and primary charge separation processes. Biochim Biophys Acta 1060:237–244. Roelofs TA, Lee C-H, and Holzwarth AR (1992) Global target analysis of picosecond chlorophyll fluorescence kinetics from pea chloroplasts. A new approach to the characterization of the primary processes in photosystem II alfaand beta-units. Biophys J 61:1147–1163. Schatz GH, Brock H, and Holzwarth AR (1988) A kinetic and energetic model for the primary processes in photosystem II. Biophys J 54:397–405. Straume M and Johnson ML (1992) Monte Carlo method for determining complete confidence probability distributions of estimated model parameters. In: Brand L and Johnson ML (eds) Methods in Enzymology. 210 Numerical Computer Methods, Academic Press, San Diego, pp 117–129. Sundström V, van Grondelle R, Bergström H, Akesson E, and Gillbro T (1986) Excitation-energy transport in the bacteriochlorophyll antenna systems of Rhodospirillum rubrum and Rhodobacter sphaeroides, studied by low-intensity picosecond absorption spectroscopy. Biochim Biophys Acta 851:431–446. Suter GW and Holzwarth AR (1987) A kinetic model for the energy transfer in phycobilisomes. Biophys J 52:673 – 683. Trinkunas G and Holzwarth AR (1994a) Kinetic modeling of exciton migration in photosynthetic systems. 2 Simulations of excitation dynamics in two-dimensional photosystem I core antenna/reaction center complexes. Biophys J 66:415– 429. Trinkunas G and Holzwarth AR (1994b) Modelling of energy transfer in photosystem I using genetic algorithm. Liet Fiz Zurn 34:287–292. van Grondelle R, Dekker JP, Gillbro T, and Sundström V (1994) Energy transfer and trapping in photosynthesis. Biochim Biophys Acta 1187:1–65. Weidner R and Georghiou S (1990) Global methods for timeresolved and steady-state fluorescence and steady-state absorption spectroscopy. In: Lakowicz JR, (ed) Time-Resolved Laser Spectroscopy in Biochemistry II, Proc of SPIE, SPIE, Bellingham, pp 717–726. Wendler J and Holzwarth AR (1987) State transitions in the green alga Scenedesmus obliquus probed by time-resolved chlorophyll fluorescence spectroscopy and global data analysis. Biophys J, 52:717–728. Wendler J, John W, Scheer H, and Holzwarth AR (1986) Energy transfer kinetics in trimeric C-phycocyanin studied by picosecond fluorescence kinetics. Photochem Photobiol 44:79–85.

Chapter 6 Photosynthetic Thermoluminescence as a Simple Probe of Photosystem II Electron Transport Yorinao Inoue Solar Energy Research Group, The Institute of Physical and Chemical Research (RIKEN), Wako, Saitama 351–01, Japan

Summary I. Introduction II. Origins of TL from Photosynthetic Apparatus A. Processes Involved in Photosynthetic TL 1. Historical Aspects and Basic Phenomenology 2. Formation and Stabilization of Charge Pairs in PSII 3. Emission of TL by Thermally Activated Recombination (Relationships with Delayed Fluorescence) 4. Measurements andAnalyses a. TL Setup and Measurements b. Simulation of Glow Curves c. Analysis of Oscillations in TL Intensity B. Assignments of TL Bands to Charge Pairs 1. The B-band 2. The Q-band 3. The A-band 4. The 5. The 6. The C-band 7. The Z-band 8. The TL-bands at Temperatures Below 77 K III. Application of TL as a Probe of PSII Photochemistry A. Acceptor Side of PSII B. Donor Side of PSII IV. Perspective: Merits and Demerits of TL Technique Acknowledgements References

93 94 94 94 94 95 96 98 98 98 99 100 100 101 101 102 102 102 102 103 103 103 103 104 105 105

Summary The measument of photosynthetic thermoluminesence (TL) is a low-cost high-performance technique. A minimum TL setup consisting of a cooled, red-sensitive head on-type photomultiplier, high-voltage power supply, X-Y recorder, heater-installed sample holder, Dewar flask, constant voltage power supply, thermocouple and a small flash device provides one with unique and suggestive data concerning the electron transport on both the donor and acceptor sides of photosystem II. If necessary, the setup can be graded up by use of a photon counter, digital thermometer and a computer-assisted data acquisition/analysis system. The technique is not always good at determining the absolute values of physical parameters, but is Correspondence: Fax: 81-48-4624685; E-mail: r19600%rkna50.riken.go.jp

93 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 93–107. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

94

Yorinao Inoue

good at detecting weak alterations in PSII functioning induced by some treatment or mutation, especially those manifest only in higher S-states. Measurements can usually be done with small amounts of algal whole cells, implying its useful application in screening or preliminary characterization of genetically engineered mutants. The chapter deals with its recent applications together with its basic phenomenology including assignments of respective emission bands. Abbreviations: D1 – the 32 kDa protein of PSII reaction center; D2 – the 34 kDa protein of PSII reaction center; DCMU – 3-(3,4-dichlorophenyl)-1,1-dimethylurea; LHC – light-harvesting chlorophyll a/b complex; P680 – the reaction center chlorophyll of PSII; Phe – Pheophytin, the primary electron acceptor of PSII; PQ – plastoquinone; PSI –.photosystem I; PSII – photosystem II; – the primary acceptor quinone of PSII; – the secondary acceptor quinone of PSII; – – redox states of the tetranuclear-Mn in water-oxidizing enzyme; TL – thermoluminescence; – auxiliary electron donor to P680, Tyr 160 residue of D2 protein; – regular electron donor to P680, Tyr 161 residue of D1 protein.

I. Introduction Thermoluminescence (TL) is an outburst of light emission occurring at characteristic temperatures when organic or inorganic materials preilluminated at low temperatures are warmed gradually in darkness. This phenomenon is common for light-responsive semiconductors, and is known to arise from thermally activated recombination of the electrons and positive holes that are generated by photoreactions and trapped or stabilized in frozen states by low temperature. In plant materials, the outburst of light occurs at several temperatures, showing variously shaped, complicated glow curves consisting of several emission bands depending on the condition of preillumination. Through extensive studies during 20 years, many of these photosynthetic TL components were separated and characterized in detail. It has now been firmly established that most of the photosynthetic TL components arise from the reversal of light-driven charge separation in photosystem II (PSII) through a thermally activated recombination between the positive charges accumulated in the intermediates of the water oxidation system on its donor side and the negative charges stabilized on primary or secondary quinone acceptors on its acceptor side (for reviews see Inoue and Shibata, 1982; Sane and Rutherford, 1986; Vass and Inoue, 1992). The assignments of charge pairs responsible

for respective TL components have now enabled us to utilize TL as a simple but effective probe of PSII electron transport that provides unique and valuable information on both sides of PSII. In this chapter, we will review the current knowledge about photosynthetic TL and discuss some possible new research areas to which TL technique can be applied.

II. Origins of TL from Photosynthetic Apparatus

A. Processes Involved in Photosynthetic TL 1. Historical Aspects and Basic Phenomenology TL from plant material was first observed by Arnold and Sherwood (1957) subsequent to the discovery of delayed luminescence from chloroplasts by Strehler and Arnold (1951) (see also Arnold, 1991). In their early studies, they have already indicated that the glow curve from plant materials consists of several TL components having different emission temperatures, indicative of participation of more than two species of charge pairs in photosynthetic TL. They have also indicated that these TL components originate mostly from PSII, but not from PSI, although detailed characterization of the phenomena had to await assignment

Thermoluminescence, a specific probe of PSII of respective charge pair species responsible for each TL component. The fact that the positive charges accumulating in the oxygen-evolving system of PSII are involved in the phenomenon of photosynthetic TL was first suggested by observations that darkgrown gymnosperm leaves, angiosperm leaves greened under widely-spaced intermittent flashes or algal cells grown in Mn-deficient medium do not exhibit major TL bands, unless their latent water-oxidation systems are photoactivated by exposure to continuous or shortly-spaced sequential flashes (Ichikawa et al., 1975; Inoue, 1976; Inoue et al., 1976). This suggestion was further confirmed by the finding that the intensity of major TL band exhibits a period-four oscillation (Inoue and Shibata, 1977a). This finding subsequently led to assignment of specific S-state intermediates as the positive charge carriers responsible for major TL bands (Rutherford et al., 1982), and consequently opened up recent applications of TL for studying the roles of extrinsic proteins and other cofactors in normal or modified turnovers of the S-state system. As to the negative charge carriers, the first indication of the involvement of two acceptor quinones of PSII was obtained in an early work by Rubin and Venediktov (1969), who observed a clear interconversion between the two TL bands by treatment with DCMU. This observation was subsequently extended by Demeter et al. (1985a) by studying herbicide-resistant mutant plants. They observed a clear downshift of TL peak temperature in the mutant plants, indicating that a significant alteration in properties occurs when some amino acids are replaced in the D1 protein. Supported by these as basic findings, photosynthetic TL has now become widely used as a simple but efficient probe of PSII functioning not only for biophysical purposes but also for assisting genetical and environmental studies. 2. Formation and Stabilization of Charge Pairs in PSII PSII reaction center consists of four membrane proteins, D1, D2, cytochrome b-559 and the product of psbI gene, in which the functional

95 prosthetic chromophores and acceptor quinone molecules are believed to be housed in a symmetric way like in the reaction center of nonsulfur purple bacteria. However, in order to retain the functional Mn-cluster and thereby the ability of water oxidation, several more membrane proteins and one extrinsic protein are required to be associated with the reaction center (Fig. 1). Absorption of a photon by chlorophyll in PSII results in excitation of the primary electron donor, P680, which elicits fast charge separation between P680 and Phe, the primary acceptor of PSII, within a few ps. The charge separated state is then stabilized in about 300 ps by rapid electron transfer from to the primary one-electron acceptor quinone of PSII, which is then followed by further transfer of the electron to the secondary two-electron acceptor quinone of PSII in Reduced is stable for tens of seconds or even to hours at room temperature and constitutes the major trap of negative charges in photosynthetic TL phenomena, while does not, because of its short life, unless the electron transport between and is blocked by some means, e.g. by herbicides. Upon the second photochemistry in the same reaction center, is further reduced to but this species is readily replaced by a plastoquinone (PQ) of the pool and does not participate in photosynthetic TL under normal conditions. Stabilization of the charge separated state occurs also on donor side of PSII. The cation radical of P680, is reduced by Yz, the Tyr 161 residue of D1 protein, and is then reduced by one electron transport from the Mn-cluster of the water-oxidizing apparatus, which is believed to consist of four Mn atoms. The Mn-cluster is known to assume five redox states denoted Si (i = 0–4) with and as the lowest and highest oxidation states, respectively. Repeated photochemistry in PSII reaction center advances the Sstates by one step each, and abstraction of four electrons from two molecules of water finally results in release of one molecule of Among the five intermediate S-states, and are stable in dark-adapted condition, but they do not participate in TL because they do not carry any positive charge, whereas and carry a positive equivalent (see later) and are stable for tens of

96

seconds at room temperature, so that they constitute the major source of positive charges for photosynthetic TL. The proton release pattern coupled with the cycling of the S-state system has been postulated to be 1,0,1,2 based on early works (Fowler, 1977; Saphon and Crofts, 1977; Forster and Junge, 1985), but this has recently been disputed. As far as the photosynthetic TL is concerned, however, the above classical pattern well explains varieties of oscillatory phenomena of photosynthetic TL so far known. Based upon this proton release pattern, the stable S-state species that carry a positive equivalent and contribute to photosynthetic TL are believed to be the and Beside Yz, another auxiliary secondary donor to P680 is known; the redox active Tyr 160 residue of D2 protein denoted as Its oxidized form known as EPR Signal IIs is stable in darkness at room temperature, and has been shown recently to participate in TL (Demeter et al.,

Yorinao Inoue

1993; Krieger et al., 1993; Johnson et al., 1994). It has not been clearly established, on the other hand, whether or not can be stabilized to participate in photosynthetic TL. Cytochrome b559 and some of the chlorophyll molecules surrounding the reaction center of PSII are known to be photo-oxidized under some conditions. However, their oxidation products do not seem to constitute the trap of positive charges for TL, although their formation by reaction center photochemistry can provide electrons to reduce (or instead of the Mn-cluster, and thereby indirectly modulate the TL phenomenon. 3. Emission of TL by Thermally Activated Recombination (Relationships with Delayed Fluorescence) From the first report of the phenomenon, TL and delayed fluorescence have been considered to share a common origin (Arnold, 1991), a radi-

Thermoluminescence, a specific probe of PSII

ative release of stored energy upon recombination between stabilized donor and acceptor at a photochemical reaction center. In PSII, the primary product of light-induced charge separation is a singlet radical pair, which rapidly proceeds to stabilization of positive and negative equivalents as forms of oxidized donor and reduced acceptor and regenerates a ground state reaction center. During this course, the major part of the energy captured by the light reaction is stored in the system in the form of redox potential difference, whereas a part of the energy is lost (stabilization energy), which results in trapping of the separated charge pair being provided with a barrier of activation energy against recombination. When this free energy loss is supplemented by thermal activation, the charge pair becomes capable of recombination to re-excite P680, resulting in photosynthetic TL emission from P680* or chlorophyll molecules present nearby (Fig. 2).

97

The process of recombination of is not always understood in detail, but it is generally considered to proceed through a series of equilibrations among various intermediate charge separated states (DeVault et al., 1983; DeVault and Govindjee, 1990) to finally generate radical pair in singlet or triplet configuration. Among the radical pairs with two different spin configurations, the triplet ones recombine nonradiatively to yield ground state P680, whereas the singlet ones are capable of generating singlet excited P680* (Van Gorkom, 1985), so that they can contribute to TL or delayed fluorescence. We may thus consider that delayed fluorescence (slow components) and TL are the two different expressions of the same process: the former is due to spontaneous recombination at a constant (usually low) frequency that is limited by the rate of thermal energy supply from the environment at a given temperature, while the latter is due to recombination at higher frequencies being accel-

98 erated with temperature rise during artificial heating. In fact, more or less the same period-four oscillation has been observed for both phenomena, the intensity of 30-s component of room temperature delayed fluorescence and the height of TL B-band, when leaves are exposed to a series of short flash illumination (Rutherford et al., 1984). This has also been confirmed by analyses of emission spectra of delayed fluorescence (Hideg et al., 1991) and TL (Sonoike et al., 1991). 4. Measurements and Analyses

a. TL Setup and Measurements Photosynthetic TL can be recorded with a simple setup. As opposed to the measurement of delayed fluorescence, it requires no timing control system, neither a photomultiplier gating nor a transient recorder. The most essential factor is cooling the photomultiplier in order to increase the signal-tonoise ratio by eliminating the dark current. Use of a photon counter, a digital thermometer equipped with a linearizing ROM system and a personal computer for data acquisition/analysis are preferable, but they are not always essential. A small sample holder having small heat capacity is recommended to afford quick cooling after flash illumination. A dark-relaxed leaf disc or a piece of filter paper soaked with an aliquot of dark-relaxed sample suspension (ca. chlorophyll) is fixed on a sample holder in which a small heater (ca. 50 W) is installed, illuminated with a short flash(es) of a few duration at a constant temperature (usually –40 to +25°C) and then quickly cooled by immersing the sample holder in liquid nitrogen. The cooled sample (with the holder) is then placed in a transparent Dewar flask, through which the image of the sample is focused onto the photomultiplier surface through a strong lens. The sample is then warmed gradually at a constant rate of 0.5 to by use of the heater installed in the sample holder, and the TL emission is measured with the photomultiplier while recording the sample temperature with a thermocouple. TL intensity vs. sample temperature is recorded on an X-Y recorder, or if necessary, through a computer. With a sample

Yorinao Inoue holder of small heat capacity, it is not easy to obtain a constant heating rate, due mainly to heat energy supply from the inner wall of the Dewar flask. It is preferable to keep the wall temperature constant by some means. Due to this difficulty and others (see Devault et al., 1983), the peak temperatures reported from various laboratories are variant. However, detection of a few degrees shift in peak temperature is not difficult if a series of measurements are done on the same setup whose operation is carefully controlled to retain reproducibility. Single flash illumination or continuous light illumination at a low temperature gives only one TL component, whereas continuous illumination during cooling the sample in liquid nitrogen gives several components. Sometimes two illuminations at two different temperatures (e.g. one at room temperature and another at 77 K) are useful in order to generate positive charges and negative charges separately (for this technique see sections II.B.3,7.).

b. Simulation of Glow Curves The kinetics of photosynthetic TL process has been described in various ways (Vass et al., 1981; DeVault et al., 1983; DeVault and Govindjee, 1990), basically according to the model conceived earlier for TL phenomena in solid state system. Among them, the one developed by Vass et al. (1981) is useful for practical applications, in which the complicated multiple equilibrations are treated as a single step first order process.

where T, k and B are measured parameters or a constant; TL intensity, absolute temperature, Boltzmann’s constant and heating rate, respectively. By changing c (proportionality constant), (activation energy) and s (frequency factor) as fitting parameters, we can simulate a glow curve with sufficient accuracy to resolve the overlapping TL bands and also to determine the activation energy for each TL components. An efficient personal computer software for graphical and numerical analysis of TL glow curves has been developed (Ducruet and Miranda, 1992). These glow curve simulations provide us with the

Thermoluminescence, a specific probe of PSII

99 quinone acceptors on the acceptor side of PSII, we can compare the effect of a treatment on the TL peak temperature between and charge pairs. If the same or similar extent of shift is detected for both TL bands, we may conclude that the shift is due to a change in the redox potential of the positively charged donor species with no or slight effects on their negative counterpart on the acceptor side (see section III).

parameters needed to calculate the half-life of a charge pair based on the equation below, which enables us to confirm the validity of these analyses (Table 1).

Among these parameters derived from TL measurements, the peak temperature gives us the most useful information. However, the peak temperature data so far reported for a TL component from various laboratories differ significantly. This is firstly due to a theoretical reason, the difference in heating rate used (DeVault et al., 1983), and secondly due to a technical reason, e.g. the difference in positioning of the thermocouple around the sample or sample holder. Despite the significant differences, a shift in peak temperature induced by treating the sample can easily and correctly be detected by comparing with an untreated control, when the same TL setup is used. An upshift in peak temperature of a TL band indicates that a larger activation energy is required for the charge pair to undergo recombination, indicative of induction of a deeper trap by the treatment, while a downshift indicates induction of a shallower trap (Fig. 2). If we assume that the treatment employed has a selective effect on either of the donor or acceptor side of PSII, a shift in peak temperature of TL can be directly interpreted as indicating a change in redox potential of the positively or negatively charged species, respectively: deeper and shallower traps on the donor side correspond to higher and lower redox potentials of the positively charged species, and those on acceptor side to lower and higher redox potentials, respectively. As a matter of fact, it is not always easy to restrict the effect of a treatment to one side of PSII. However, by taking advantage of the two

c. Analysis of Oscillations in TL Intensity In dark-adapted PSII, the redox pair is usually the most dominant species among the centers. When this center is exposed to a series of short flashes, the redox pairs expected after 1, 2, 3 and 4 flashes are and respectively. Since and are the positively charged species, as discussed in the previous section, the probability of positive charge stabilization on the donor side oscillates in a 1,1,0,0 pattern, whereas the probability of negative charge stabilization on the acceptor side oscillates in 1,0,1,0, a binary pattern, due to the two electron gate mechanism of the quinone. Thus the probability of stabilization of both positive and negative charges in one PSII reaction center, which is a prerequisite for emission of TL, oscillates in a 1,0,0,0 pattern. In actual PSII preparations, this pattern is perturbed by several factors: (i) initial distribution of (ii) initial distribution of (iii) the probability of misses and double hits and (iv) the ratio of TL yield between and charge recombinations. Of these factors, the factors (ii) to (iv) are mostly constant: misses and double hits are around 10% and 5%, respectively, and the TL yield ratio of In contrast, the initial ratio of varies depending on the sample and relaxation conditions: 50:50 in thylakoids after brief dark-relaxation or in well dark-adapted intact chloroplasts, whereas it is 25:75 in well darkrelaxed thylakoids or in PSII membrane fragments, so-called BBY-type particles. Table 2 describes the expected changes in the number of centers capable of emitting TL, when dark-adapted PSII preparations having two typical initial ratios are illuminated with a series of flashes. Distribution of to the centers

100

in and is assumed even, and misses and double hits are neglected. The number of TLemitting centers shows maxima after the 1st and 5th flashes if the initial or after the 2nd and 6th flashes if the initial When misses and double hits and the difference in luminescence yield are taken into account, the oscillatory patterns depicted in Fig. 3 are obtained. These patterns agree well with observed ones, supporting the mechanisms of TL oscillation and in turn the assignments of the B-bands (Inoue, 1983; Demeter and Vass, 1984). Upon inhibitory treatment, depletion of some essential cofactors or mutations in PSII, the oscillation deviates from these standard patterns, which gives information about the functioning of the S-state cycle and the electron gate in inhibited or modified PSII.

B. Assignments of TL bands to charge pairs The TL components are distinguished by their respective emission bands on the glow curve that exhibit characteristic temperatures for maximum emission. For several of these components, responsible charge pairs have been identified (Table

Yorinao Inoue

3). In the following we discuss information currently available about the properties of the respective TL bands. 1. The B-band The B-band that appears at around +30 °C is the best characterized TL component. This band is clearly correlated with the water-oxidizing enzyme (Inoue and Shibata, 1977a), in particular with the presence of functionally active Mn (Inoue, 1976; Rozsa and Demeter, 1982), and has been assigned to and charge recombination (Rutherford et al., 1982, 1985). When excited by a series of saturating flashes, the intensity of the B-band exhibits a period-four oscillation in thylakoids (Inoue and Shibata, 1977a; 1977b; Demeter and Vass, 1984; Demeter et al., 1984), intact leaves and PSII-enriched membrane fragments (Rutherford et al., 1984). Analysis of the oscillation pattern indicates that both the and are involved in generation of this TL component (Rutherford et al., 1982; Demeter and Vass, 1984). Participation of as the negative charge trap has been evidenced by modulation of the initial ratio by chemical or

Thermoluminescence, a specific probe of PSII

101 shape, but below pH 6.0, they exhibit two distinguishable components denoted and respectively, with a few degrees difference in peak temperature (Inoue, 1981). The TL yield from the latter recombination is higher than that from the former by a factor of 1.7–2.0 due to some unknown reasons (Rutherford et al., 1985; Demeter et al., 1985b). The emission spectrum of the B-band(s) has a maximum at around 690 nm (Sonoike et al., 1991) in accordance with the hypothesis that the recombination re-excites P680 as a reversal of charge separation.

2. The Q-band Upon treating PSII with DCMU or other PSII herbicides, the B-band is abolished with a concomitant appearance of a new TL band between 0 and +10 °C (Rubin and Venediktov, 1969). This band is called the Q-band and arises from charge recombination (Demeter and Vass, 1984; Rutherford et al., 1982). The conversion of the B-band to the Q-band is due to inhibition by DCMU of the electron transfer between and which enables stabilization of as a negative charge detectable by TL instead of The Q-band is often used to analyze the features of stabilization of the

3. The A-band

light pretreatment or by removal and reconstitution of (Rutherford et al., 1982; Demeter and Vass, 1984; Wydrzynski and Inoue, 1987). At or above pH 7.0-7.5, both and recombinations give rise to the same TL band with respect to the peak temperature and band

When PSII is preilluminated with two flashes at room temperature, then cooled to 77 K, and then further illuminated with continuous light at 77 K, a clear TL band denoted the A-band, appears at around –10°C (Läufer et al., 1978; Inoue, 1981; Demeter et al., 1985b). This band has been assigned to arise from charge recombination based on the following considerations: illumination with two flashes at room temperature generates an state which is incapable of TL due to the absence of negative charge (see Table 2), but the 77 K illumination oxidizes cytochromeb559 or chlorophyll instead of the Mn-cluster (in and generates which recombines with the previously formed by illumination with two flashes and stabilized by cooling and unaffected during the 77 K illumination (Koike et al., 1986).

102 4. The Tris-treated PSII depleted of the functional Mncluster emits a TL band at around –10°C (Inoue et al., 1977; Rosza and Demeter, 1982). This band is called the to distinguish it from the A-band having the same emission temperature (Koike et al., 1986). The origin of this component was not clear, but it was later proposed to arise from charge recombination between and a photooxidized histidine residue of a PSII reaction center protein by use of chemically modified PSII (Ono and Inoue, 1991a). This hypothesis was addressed in more detail by use of mutated PSII, and it was revealed that substitution of both His195 and His190 affects the emission of this TL component (Kramer et al., 1994). The positive equivalent on the photooxidized histidine exhibits a high affinity for exogenous and is proposed to be involved in photoligation of ions in the process of photoactivation of a latent oxygen-evolving system (Ono and Inoue, 1991b). 5. The A minor TL component denoted as the appears between –80 and –30°C. The emission temperature of this TL band varies depending on the excitation temperature (Ichikawa et al., 1975), being usually higher by 10 to 20 °C than the excitation temperature. This band is abolished by treatment with 1% ethanol and exhibits a periodfour oscillation dependent on S-state turnovers (Demeter et al., 1985b), although the band is reported to be emitted also from the D1/D2/cytochrome b-559 PSII reaction center preparation depleted of the Mn-cluster (Vass et al., 1989). Although it is proposed to arise from charge pair whose trapping depth is modulated by the conformation of the reaction center, its origin is not yet clear. The participation of as the negative counterpart has recently been confirmed by reconstitution of artificial in purified reaction center preparation by use of plastoquinone-9 (Chapman et al., 1991). 6. The C-band A TL component denoted as the C-band appears at around +50 °C on glow curves from DCMU-

Yorinao Inoue treated plant materials (Desai et al., 1975). The participation of as the negative counterpart is likely, judging from enhancement of its intensity in the presence of high concentration of DCMU. The C-band is reported to exhibit a period-four oscillation with maxima in the and states, suggesting that this band arises from charge recombination (Demeter et al., 1984). (Tyr 160 residue of D2 protein) has been proposed as a candidate for the positive equivalent in and Demeter et al. (1993), Krieger et al. (1993) and Johnson et al. (1994) reported that this component arises from charge recombination between and in normal and PSII as well.

7. The Z-band A strong and broad TL band appears at around 110K when various plant materials are illuminated at 77 K with continuous light. This component, denoted as the Z-band, exhibits an emission maximum at around 730 nm, in contrast to 690 nm of the B-band arising from PSII reaction center photochemistry, and is excited with blue light at a higher yield than with red light (Arnold and Azzi, 1968). Upon raising the excitation temperature above 77 K, only the higher temperature part of the Z-band appears, indicating that the trap for this component has a broad distribution of stabilization free energies. It turned out recently that LHCI and LHCII, that have no photochemical reaction center, or purified chlorophyll, particularly its aggregated form, emit this band more strongly than PSI or PSII (Sonoike et al., 1991), indicating that the Z-band is correlated with neither PSI nor PSII photochemistries. It has recently been found that this band is preferentially charged by blue light, and the trapped charges are detrapped by illumination with red light, due probably to a local heating effect (Hargen et al., 1994). This effect might be related with the mechanism of non-photochemical quenching which is postulated to result from conformational changes in LHC concomitant with the accumulation of zeaxanthin via the xanthophyll cycle.

Thermoluminescence, a specific probe of PSII

103

8. The TL Bands at Temperatures Below 77 K

the water-oxidizing enzyme, Y) effect of these chemicals affecting the properties of the state on donor side of PSII. Such effects on TL are used for characterizing the functioning of new herbicidal compounds (Asami et al., 1988; Koike et al., 1989). Herbicide resistant mutants exhibit a remarkably downshifted B-band. In triazine resistant mutants of Eligeron canadensis and Synechocystis PCC 6714, the emission temperature of the Bband is downshifted by 15 °C as compared with that of the wild types (Demeter et al., 1985a). This indicates that the redox potential difference between and is decreased from 70 to 30 mV due to a structural alteration of the site by the mutation. Analysis of cyanobacterial mutants reveals that the emission temperature of the B-band is strongly sensitive to the replacement of Ser264 of D1 by Ala or Gly that concomitantly induces strong resistance to triazines, but is rather immune to the replacement of Phe255 by Tyr that induces resistance to phenylureas, suggesting a more important role of Ser264 in as compared with Phe255 (Etienne et al., 1990; Gleiter et al., 1990, 1992).

Three new TL components emitting at around 20, 50 and 70 K have been resolved in the glow curve at temperatures below 77 K, and denoted as and respectively (Noguchi et al., 1992). The glow curve excited at liquid helium temperatures exhibits another component at around 90 K, but this turned out to be a different expression of the well-known Z-band that emits above 77 K. The properties of the three components are essentially the same as those of the Zband: being excited preferentially by blue light and emitted more strongly from LHCs and aggregated chlorophyll, suggesting that they originate from charge storage in a chlorophyll molecule interacting with another chlorophyll or solvent molecule(s). III. Application of TL as a Probe of PSII Photochemistry

A. Acceptor Side of PSII The redox couples and differ in redox potential by 50–70 mV, and this difference manifests on TL glow curves as a 25–30 °C difference in emission temperature between the Bband and Q-band, which arise from and charge pairs, respectively. The energetic stability of reduced quinones can easily be studied by monitoring TL (Demeter et al., 1985a). These properties enable one to use TL for studies of herbicide resistance (for other applications see Vass and Inoue, 1992). Upon treatment of PSII with herbicides the Bband is converted to the Q-band due to inhibition of the electron transfer between and Notably, however, the emission temperature of the resulting Q-band differs depending on the chemical structure of the herbicide employed: phenolic type herbicides induce a Q-band emitting between –15 and 0°C, whereas urea/triazine type herbicides induce a Q-band between 0 and +10°C (Vass and Demeter, 1982). The much lower Q-band peak temperature in the presence of phenolic type herbicides may be ascribed either to a lowered redox potential of or to the ADRY (acceleration of deactivation reactions in

B. Donor side of PSII The generation of the B-bands and the A-band involves the and so that the emission temperature and oscillatory behavior of those TL bands provide useful information regarding the properties of the S-states and the role of various cofactors involved in S-state transitions. These properties were first used to determine the temperature dependence of S-state transition. Upon lowering the ambient temperature, the Sstate transitions become inhibited. From the oscillation of the B-band under flash excitation, and transitions were shown to be blocked at –35 and –20 °C, respectively, while the transition was inhibited at –65 °C (Inoue and Shibata, 1977b). Further precise studies reveal that and transitions are completely blocked at –160, –65 and –40 °C, and half-blocked at –95, –45 and –23 °C, respectively (Demeter and Vass, 1984; Demeter et al., 1985b; Koike and Inoue, 1987). Through a sophisticated analysis of the tempera-

104 ture dependence by use of two different protocols for low temperature flash excitation, “last flash at low temperature” and “all flashes at low temperature”, a low-temperature sensitive intermediate state has been proposed to exist between and which might be a precursor in which proton release is not completed (Koike and Inoue, 1987). Another typical application of TL for studies of the donor side of PSII is the effect of removal of extrinsic proteins. Higher plants’ PSII contains a set of three extrinsic proteins of molecular masses of 33, 24 and 18 kDa, which are electrostatically associated with the lumenal surface of PSII and involved in regulating the requirement for inorganic cofactors and stability of the Mncluster. By a suitable salt washing followed by mixing with concentrated extract of the three extrinsic proteins, these proteins can be reversibly removed and reconstituted with concomitant inactivation and reactivation of oxygen-evolving activity. Upon removal of the set of proteins, oxygen evolution is almost lost whereas the TL Bband can still be fully induced. Although PSII depleted of the extrinsic proteins exhibits normal oscillation up to the the final transition to release molecular oxygen is blocked (Ono and Inoue, 1985). Upon removal of the extrinsic proteins, the peak position of the Bband remains at the normal temperature, suggesting no appreciable change in the stability of the charge pair. In contrast, however, the peak temperature of the Q-band is remarkably upshifted by 20–25 °C indicative of deeper stabilization of the state in the absence of the extrinsic proteins (Vass et al., 1987a). The different stabilization between and charge pairs may be due to a secondary but specific effect on the site, in addition to the major effect on the More or less the same TL data have been obtained for genetically engineered cyanobacterium mutant cells in which the psb0 gene encoding the 33 kDa extrinsic protein is deleted (Burnap et al., 1992), indicating that the TL methods can successfully characterize the properties of PSII donor side in mutant cells without isolating thylakoids. TL data implicating an incomplete association of the 33 kDa protein with PSII membranes have also been collected for cyanobactrial

Yorinao Inoue mutants in which a few amino acids on the lumenal loop of the CP47 protein are deleted (Gleiter et al., 1994). By analogy to these, TL has now been extensively applied in probing redox properties of the donor side of PSII (Ono and Inoue, 1989, 1990) or PSII (Vass et al., 1987b). These applications are discussed in more detail in Vass and Inoue (1992).

IV. Perspective: Merits and Demerits of TL Technique The first stage of studies on photosynthetic TL was the 10 years following the discovery of this phenomenon by Arnold and Sherwood in 1957, during which the basic phenomenology and fundamental concept about this phenomenon were established. The second stage would be the last 20 years, during which the origins of most of photosynthetic TL components were clarified subsequent to the finding that photosynthetic TL involves the so-called “charge accumulator” of the photosynthetic oxygen-evolving system. The second stage has now opened up a possibility to utilize photosynthetic TL as a simple but efficient probe for characterizing electron transport in PSII on both the donor and acceptor sides. Following are examples of promising areas to which photosynthetic TL can be efficiently applied: (i) Studies on structure and function of the Mncluster, particularly coupled with the use of genetically engineered mutants having some defects in water-oxidation activity. By simple TL analyses we will be able to characterize apparently small but functionally serious alterations occurring on the Mn-cluster of mutant cells, particulary those manifest only in higher S-states. (ii) Studies on structural determinants of the or herbicide-binding site, particularly coupled, also in this case, with genetically engineered mutants. Alterations in properties upon acquirement of herbicide resistance can easily be detected by the TL technique. (iii) Studies on environmental stresses including acclimation to low or high temperature and photoinhibition that affect the functioning of PSII. Taking all these into consideration, the advantages and disadvantages of the TL technique in photosynthetic studies can be summarized as fol-

Thermoluminescence, a specific probe of PSII lows: Advantages are: (i) Low cost setup, (ii) Simple operation requiring no experience, (iii) Simple sample preparation. Measurements with whole cells are sometimes essential in studies using algal mutants, (iv) Easy application for characterization of PSII functioning for physiological or environmental purposes, (v) Selective probing of PSII by ignoring PSI. Disadvantages are: (i) The phenomenon is restricted to PSII. It cannot be applied to PSI and bacterial reaction centers. The reason for this remains to be theoretically clarified, (ii) The TL data do not directly provide physical parameters. They are usually given as deviation from the control measurement, (iii) Alternative interpretations: e.g. a shift in peak temperature can be attributed to a change either on donor side or acceptor side.

Acknowledgements The author acknowledges the Science and Technology Agency of Japan (STA) which has been financially supporting the TL studies during the last 15 years under the title of Solar Energy Conversion by Means of Photosynthesis. The author is also grateful to Drs. Mamiko Kimimura-Taguchi and Han Kab-Cho for their technical assistance in preparing the manuscript, and also to Professor J. Amesz for his valuable suggestions.

References Arnold WA (1991) Experiments. Photosynth Res 27: 73–82 Arnold WA and Azzi JR (1968) Chlorophyll energy levels and electron flow in photosynthesis. Proc Natl Acad Sci USA 61: 29–35 Arnold WA and Sherwood HK (1957) Are chloroplasts semiconductors? Proc Natl Acad Sci USA 43: 105–114 Asami T, Koike H, Inoue Y, Takahashi N and Yoshida S (1988) Structure-activity relationships and physiological aspects of new photosynthetic electron transport inhibitors, 3-alkylamino-alkyliden-2H-pyran-2,4(3H)-diones (APs). Z Naturforsch 43c: 857–861 Burnap RL, Shen J-R, Jursinic PA, Inoue Y and Sherman LA (1992) Oxygen yield and thermoluminescence characteristics of a cyanobacterium lacking the manganese-stabilizing protein of photosystem II. Biochemistry 31: 7404– 7410 Chapman DJ, Vass I and Barber J (1991) Secondary electron transfer reactions of the isolated photosystem II reaction center after reconstitution with plastoquinone-9 and diacylglycerolipids. Biochim Biophys Acta 1057: 391–398 Demeter S and Vass I (1984) Charge accumulation and recom-

105 bination in photosystem II studied by thermoluminescence. I: Participation of the primary acceptor Q and secondary acceptor B in the generation of thermoluminescence of chloroplasts. Biochim Biophys Acta 764: 24–32 Demeter S, Vass I, Horvath G and Läufer A (1984) Charge accumulation and recombination in photosystem II studied by thermoluminescence. II: Oscillation of the C band by flash excitation. Biochim Biophys Acta 764: 33–39 Demeter S, Vass I, Hideg E and Sallai A (1985a) Comparative thermoluminescence study of triazine-resistant and -susceptible biotypes of Erigeron canadensis L. Biochim Biophys Acta 806: 16–24 Demeter S, Rozsa Z, Vass I and Hideg E (1985b) Thermoluminescence study of charge recombination in photosystem II at low temperature II: Oscillatory properties of the Z and A thermoluminescence bands in chloroplasts dark- adapted for various time periods. Biochim Biophys Acta 809: 379– 387 Demeter S, Goussias C, Bernet G, Kovacs L and Petrouleas V (1993) Participation of the g = 1.9 and g = 1.82 EPR forms of the semiquinone-ironcomplex, of photosystem II in the generation of the Q and C thermoluminescence bands, respectively. FEBS Lett 336: 352–356 Desai TS, Sane PV and Tatake VG (1975) Thermoluminescence studies on spinach leaves and Euglena. Photochem Photobiol 21: 345–350 DeVault D and Govindjee (1990) Photosynthetic glow peaks and their relationship with the free energy changes. Photosynth Res 24: 175–181 DeVault D, Govindjee and Arnold WA (1983) Energetics of photosynthetic glow peaks. Proc Natl Acad Sci USA 80: 983–987 Ducruet J-M and Miranda T (1992) Graphical and numerical analysis of thermoluminescence and fluorescence Fo emission in photosynthetic material. Photosynth Res 33: 15–27 Etienne A-L, Ducruet J-M, Ajlani G and Vernotte C (1990) Comparative studies on electron transfer in photosystem II of herbicide-resistant mutants from different organisms. Biochim Biophys Acta 1015: 435–440 Förster W and Junge W (1985) Stoichiometry and kinetics of proton release upon photosynthetic water oxidation. Photochem Photobiol 41: 183–190 Fowler CF (1977) Proton evolution from photosystem II. Stoichiometry and mechanistic considerations. Biochim Biophys Acta 462: 414–421 Gleiter H, Ohad N, Hirschberg J, Fromme R, Renger G, Koike H and Inoue Y (1990) An application of thermoluminescence to herbicide studies. Z Naturforsch 45c: 353–358 Gleiter H, Ohad N, Koike H, Hirschberg J, Renger G and Inoue, Y. (1992) Thermoluminescence and flash-induced oxygen yield in herbicide resistant mutants of the D1 protein in Synechococcus PCC7942. Biochim Biophys Acta 1140: 135–143 Gleiter H, Haag E, Shen J-R, Eaton-Rye J, Inoue Y, Vermaas WFJ and Renger G (1994) Functional characterization of mutant strains of the cyanobacterium Synechocystis sp. PCC6803 lacking short domain within the large, lumenexposed loop of the chlorophyll-protein CP47 in photosystem II. Biochemistry 33 12063–12071 Hagen C, Pascal A, Horton P and Inoue Y (1995) Influence

106 of changes in the photon protective energy dissipation on red light-induced detrapping of the thermoluminescence Zband. Photochem Photobiol 62: 514–521 Hideg E, Scott RQ and Inaba H (1991) Spectral resolution of long term (0.5–50 s) delayed fluorescence from spinach chloroplasts. Arch Biochem Biophys 285: 371–372 Ichikawa T, Inoue Y and Shibata K (1975) Characteristics of thermoluminescence bands of intact leaves and isolated chloroplasts in relation to the water-splitting activity in photosynthesis. Biochim Biophys Acta 408: 228–239 Inoue Y (1976) Manganese catalyst as a possible cation carrier in thermoluminescence. FEBS Lett 72: 279-282 Inoue Y (1981) Charging of the A band of thermoluminescence dependent on the in isolated chloroplasts. Biochim Biophys Acta 634: 309–320 Inoue Y (1983) Recent advances in the studies of thermoluminescence of photosystem II. In: Inoue Y, Crofts AR, Govindjee, Murata N, Renger G and Satoh K (eds) The Oxygen Evolving System of Photosynthesis, pp 439–450. Academic Press Japan, Tokyo Inoue Y and Shibata K (1977a) Thermoluminescence bands of chloroplasts as characterized by flash excitation. In: Hall DO, Coombs J and Goodwin TW (eds) Proc Fourth Intl Congr on Photosynthesis, pp 211–221.The Biochemical Society, London Inoue Y and Shibata K. (1977b) Oscillation of thermoluminescence at medium low temperature. FEBS Lett 85: 19– 197 Inoue Y and Shibata K (1982) Thermoluminescence from photosynthetic apparatus. In: Govindjee (ed) Photosynthesis; Energy Conversion by Plants and Bacteria, Vol. pp 507–533. Academic Press, New York Inoue Y, Ichikawa T and Shibata K (1976) Development of thermoluminescence bands during greening of wheat leaves under continuous and intermittent illumination. Photochem Photobiol 23: 125–130 Inoue Y, Yuashita T, Kobayashi Y and Shibata K (1977) Thermoluminescence changes during inactivation and reactivation of the oxygen-evolving system in isolated chloroplasts. FEBS Lett 82: 303–306 Johnson GN, Boussac A and Rutherford AW (1994) The origin of 40–50 °C thermoluminescence bands in photosystem II. Biochim Biophys Acta 1184: 85–92 Koike H and Inoue Y (1987) A low temperature sensitive intermediate state between and in photosynthetic water oxidation deduced by means of thermoluminescence measurement. Biochim Biophys Acta 894: 573–577 Koike H, Siderer Y, Ono T and Inoue Y (1986) Assignment of thermoluminescence A band to charge recombination: sequential stabilization of and by a two-step illumination at different temperatures. Biochim Biophys Acta 850: 80–89 Koike H, Asami T, Yoshida S, T Kahashi N and Inoue Y (1989) A new-type photosystem II inhibitor which blocks electron transport in water-oxidation system. Z Naturforsch 44c: 271–279 Kramer DM, Roffey RA, Govindjee and Sayre RT (1994) The AT thermoluminescence band from Chlamydomonas reinhardtii and the effects of mutagenesis of histidine resi-

Yorinao Inoue dues on the donor side of the photosystem II D1 polypeptide. Biochim Biophys Acta 1185: 228–237 Krieger A, Weis E and Demeter S (1993) Low-pH-induced ion release in the water-splitting system is accompanied by a shift in the midpoint redox potential of the primary quinone acceptor Biochim Biophys Acta 1144: 411–418 Läufer A, Inoue Y and Shibata K (1978) Enhanced charging of thermoluminescence A band by a combination of flash and continuous light excitation. In: Akoyunoglou (ed) Proc Intl Symp on Chloroplast Development, pp 379–387. Elsevier, Amsterdam Noguchi T, Inoue Y and Sonoike K (1992) Thermoluminescence emission at liquid helium temperature from photosynthetic apparatus and purified pigments. Biochim Biophys Acta 1141: 18–24 Ono T and Inoue Y (1985) S-state turnover in the system of photosystem II particles depleted of three peripheral proteins as measured by thermoluminescence: Removal of 33 kDa protein inhibits to transition. Biochim Biophys Acta 806: 331–340 Ono T and Inoue Y (1989) Removal of Ca by pH 3.0 treatment inhibits to transition in photosynthetic oxygen evolution system. Biochim Biophys Acta 973: 443–449 Ono T and Inoue Y (1990) Abnormal redox reactions in photosynthetic centers in NaCl/EDTA-washed PSII. A dark-stable EPR multiline signal and a new unknown positive charge accumulator. Biochim Biophys Acta 1020: 269–277. Ono T and Inoue Y (1991a) Biochemical evidence for histidine oxidation in photosystem II depleted of the Mn-cluster for FEBS Lett 278: 183–186 Ono T and Inoue Y (1991b) A possible role of redox active histidine in photoligation of Mn into photosynthetic evolving enzyme. Biochemistry 30: 6183–6188 Rozsa Z and Demeter S (1982) Effect of inactivation of the oxygen evolving system on the thermoluminescence of isolated chloroplasts. Photochem Photobiol 36: 705–708 Rubin AB and Venediktov PS (1969) Storage of the light energy by photosynthesizing organisms at low temperature. Biofizika 14: 105–109 Rutherford WA, Crofts AR and Inoue Y (1982) Thermoluminescence as a probe of photosystem II photochemistry; the origin of the flash-induced glow peaks. Biochim Biophys Acta 682: 457–465 Rutherford WA, Govindjee and Inoue Y (1984) Charge accumulation and photochemistry in leaves studied by thermoluminescence and delayed light emission. Proc Natl Acad Sci USA 81: 1107–1111 Rutherford WA, Renger G, Koike H and Inoue Y (1985) Thermoluminescence as a probe of PSII: The redox and protonation sate of the secondary acceptor quinone and the evolving enzyme. Biochim Biophys Acta 767: 548– 556 Sane PV and Rutherford WA (1986) Thermoluminescence from photosynthetic membranes. In: Govindjee, Ames J and Fork DC (eds) Light Emission by Plants and Bacteria, pp 329–360. Academic Press, Orlando Saphon S and Crofts AR (1977) Protolytic reactions in photosystem II. A new model for the release of protons ac-

Thermoluminescence, a specific probe of PSII companying the photooxidation of water. Z Naturforsch 32c: 617–626 Sonoike K, Koike H, Enami I and Inoue Y (1991) The emission spectra of thermoluminescence from photosynthetic apparatus. Biochim Biophys Acta 1058: 121–130 Strehler BL and Arnold WA (1951) Light production by green plants. J Gen Physiol 34: 59–73 Van Gorkom HJ (1985) Electron transfer in photosystem II. Photosynth Res 6: 97–1158 Vass I and Demeter S (1982) Classification of photosystem II inhibitors by thermodynamic characterization of the thermoluminescence of inhibitor-treated chloroplasts. Biochim Biophys Acta 682: 496–499 Vass I and Inoue Y (1992) The moluminescence in the study of photosystem II. In: Barber J (ed) The Photosystems: Structure, Function and Molecular Biology. Topics in Photosynthesis, Vol. 11 pp 259–294. Elsevier, Amsterdam Vass I, Horvath G, Herczeg T and Demeter S (1981) Photo-

107 synthetic energy conservation investigated by thermoluminescence. Activation energies and half-lives of thermoluminescence bands of chloroplasts determined by mathematical resolution of glow curves. Biochim Biophys Acta 634: 140–152 Vass I, Ono T and Inoue Y (1987a) Removal of 33 kDa extrinsic protein specifically stabilizes the charge pair in photosystem II. FEBS Lett 211: 71–76 Vass I, Ono T and Inoue Y (1987b) Stability and oscillation properties of thermoluminescence charge pairs in the evolving system depleted of or the 33 kDa extrinsic protein. Biochim Biophys Acta 892: 224–235 Vass I, Chapman DJ and Baber J (1989) Thermoluminescence properties of the isolated photosystem two reaction center. Photosynth Res 22: 295–301 Wydrzynski T and Inoue Y (1987) Modified photosystem II acceptor side properties upon replacement of the quinone at the site with 2,5-dimethyl-p-benzoquinone and phenyl-p-benzoquinone. Biochim Biophys Acta 893: 33–42

This page intentionally left blank

Chapter 7 Accumulated Photon Echo Measurements of Excited State Dynamics in Pigment–Protein Complexes Thijs J. Aartsma*, Robert J.W. Louwe and Peter Schellenberg Department of Biophysics, University of Leiden, P.O. Box 9504, 2300 RA Leiden, The Netherlands

Summary I. Introduction II. Homogeneous and Inhomogeneous Linewidths III. Photon Echo Phenomena A. Two-pulse Photon Echo B. Three-pulse Stimulated Photon Echo C. Accumulated Photon Echo IV. Accumulated Photon Echo: Experimental V. Energy Transfer VI. Photon Echo Experiments on Reaction Centers VII. Conclusion and Perspectives Acknowledgements References

109 109 110 111 111 111 113 114 116 118 119 120 120

Summary The homogeneous lineshape of optical transitions in molecular systems contains information about the intrinsic excited state dynamics. The lineshape parameters can be obtained from accumulated photon echo measurements. This technique has been applied to investigate the excited state dynamics of isolated photosynthetic pigment-protein complexes at low temperature. In antenna systems it is found that the coherent lifetimes of the initally excited exciton state can be hundreds of picoseconds. This means that the optical excitation is intrinsically delocalized, and that a hopping-model for energy transfer is less appropriate at low temperature. In reaction centers, the accumulated photon echo decay is determined by the rate of primary charge separation. The multiexponential decay of the echo-signal is most likely due to dispersive kinetics. Abbreviations: AOM – acousto-optic modulator; APE – accumulated photon echo; BChl – bacteriochlorophyll; EOM – electro-optic modulator; FMO – Fenna–Matthews–Olson; I – primary acceptor; P – primary donor; RC – reaction center; 2PE – two-pulse photon echo; 3PSE – three-pulse stimulated echo; Q – quinone

I. Introduction The primary function of photosynthetic pigmentprotein complexes is to collect energy by absorption of light and to make this energy available for biochemical metabolic processes. This occurs through a sequence of energy-stabilizing steps in-

*Correspondence: Fax: 31-71-5275819; E-mail:[email protected]

109 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 109–122. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

110

volving energy transfer and charge separation. These primary steps are extremely fast and compete very effectively with intrinsic molecular relaxation processes such as fluorescence, intersystem crossing and internal conversion. For a full understanding and description of the molecular mechanism, it is necessary to investigate in detail the excited state dynamics of the complexes involved. Such information will be essential to test models for energy transfer and charge separation, and for understanding the underlying intermolecular interactions in relation to the molecular structure. This chapter describes accumulated photon echo experiments as a means to investigate the excited state dynamics of photosynthetic systems at low temperature. In the first part we discuss the formation and decay of the accumulated photon echo in relation to the homogeneous linewidth, consider the conditions under which an accumulated photon echo can be observed, and describe a typical experimental setup for the measurements. In the second part we discuss the results which have been obtained by measurements on isolated photosynthetic pigment–protein complexes. It should be emphasized that the accumulated photon echo decay is determined by the loss of excited state coherence, and is not necessarily identical to results obtained by pump-probe measurements. Data from both types of measurements should be considered to be complementary, and as such will be important for the correct interpretation of the excited state dynamics involved. II. Homogeneous and Inhomogeneous Linewidths Optical transitions of a molecule are characterized by a finite bandwidth, usually referred to as the homogeneous linewidth, which is determined by the finite lifetime of the energy levels involved in the electronic transition. Considering the homogeneous broadening mechanism, it is customary to distinguish between population relaxation and pure optical dephasing, characterized by time constants and respectively. Population relaxation involves transitions between discrete energy levels within the molecule,

Thijs J. Aartsma et al. and determines the lower limit of the optical linewidth. Pure optical dephasing is due to a modulation of the optical transition frequency caused by guest–host interactions such as phonon-scattering and scattering on disorder modes. Thus the homogeneous linewidth carries information about the dynamics of the molecular system. The homogeneous line width is related to the inverse of the overall dephasing time

This equation for the homogeneous linewidth is essentially a manifestation of the Heisenberg-uncertainty principle. At sufficiently low temperatures, such that bath fluctuations are completely quenched, the homogeneous linewidth is determined by For an ‘isolated’ molecule in a host matrix, will in general be equal to the fluorescence lifetime. Intermolecular interactions between the guest-molecules may give rise to additional relaxation processes which contribute to population relaxation. This is typically the case in photosynthetic systems where energy transfer and charge separation play an essential role in the conversion of photons into useful chemical energy. Therefore, it may be expected that a study of the intrinsic homogeneous linewidth may be used to characterize such processes and the associated intermolecular interactions in more detail. The experimentally observed linewidth, however, is often much broader than the homogeneous linewidth. This is due to the effects of inhomogeneous broadening associated with the dependence of the transition frequency on the local electrostatic field. Because of disorder, the local field may differ considerably from one site to the other, with a corresponding change in the transition frequency through interaction of this field with the molecular polarizability and/or permanent dipole moment. To uncover the homogeneous linewidth parameters from the inhomogeneously broadened line, special methods are required. They can be divided in two main categories: those that operate in the frequency domain such as transient and permanent hole burning spectroscopy, and those that operate in the time domain. Hole burning spectroscopy is re-

Accumulated photon echo measurements viewed in chapter 8 by N.R.S. Reddy and G.J. Small. III. Photon Echo Phenomena In the time domain, is measured in a photon echo experiment (Abel et al., 1966; Aartsma and Wiersma, 1976; Hesslink and Wiersma, 1978) which involves coherent excitation of the sample with short laser pulses. As a result of coherent excitation, the oscillating electric dipoles of the excited molecules will have a well-defined phase relationship. At room temperature this phase relationship will be destroyed very rapidly because of thermally induced scattering processes, but at low temperature it may persist for considerable lengths of time. The coherence of the excited molecular system gives rise to a macroscopic polarization in the medium, which is oscillating at the driving optical frequency. The amplitude of this polarization is proportional to the number of excited molecules. Hence, the intensity of the radiation emitted by a coherent ensemble of dipole oscillators will be proportional to (Dicke, 1954), whereas in an incoherent emission process this intensity is proportional to However, even in the absence of scattering processes, the coherent superposition is lost very rapidly because of the dephasing which arises from the different resonance frequencies in the case of inhomogeneous broadening. A. Two-Pulse Photon Echo In the absence of phase-perturbing scattering processes, the time development of the relative phases of the oscillators in an inhomogeneously broadened system is well-defined despite the variation in resonance frequency. The magic of the two-pulse photon echo arises from the fact that this phase development can be reversed by applying a second pulse with the proper intensity at a suitable time delay (Abella et al., 1966). Since the rephasing of the oscillators proceeds at the same rate as the dephasing, determined by their individual detuning from resonance, the coherent superposition is restored at a time after the second excitation pulse. This state gives rise to an intense burst of coherent radiation, proportional to which is known as the photon

111

echo. The photon echo intensity will have a maximum when the pulse areas, defined as are equal to and for the first and second excitation pulse, respectively (Allen and Eberly, 1975). Here, is the electric field amplitude, and the pulse area is a measure for the interaction strength of the field with the transition moment The intensity of the two-pulse photon echo, will diminish as the time delay between the excitation pulses is increased because of the irreversible loss of coherence through excited state decay and dephasing. More explicitly,

Thus can be obtained by measuring the photon echo intensity as a function of the time delay between the excitation pulses. The dimensions of the sample are typically much larger than the wavelength of excitation. Therefore, constructive interference of the coherent radiation which gives rise to the photon echo, is only observed in a very specific direction which is defined by the phase-matching condition:

where are the wave vectors of the photon echo, and the second and first excitation pulses, respectively. For a more comprehensive description of the photon echo phenomenon we refer to Allen and Eberly (1975) and Levenson and Kano (1988). B. Three-Pulse Stimulated Photon Echo The time-development of the induced polarization can be manipulated by other pulse sequences (Allen and Eberly, 1975). The most familiar variant is the three-pulse or stimulated photon echo which involves the application of three excitation pulses. The formation and characteristics of the stimulated photon echo can be described by considering the effect of coherent optical excitation in the frequency domain. The power spectrum of a pair of identical, coherent light pulses propagating in the same direction at a relative time delay is characterized by a periodically varying amplitude with a fringe-spacing given by over a frequency range deter-

112

Thijs J. Aartsma et al.

mined by the optical bandwidth of the light pulses. If this pulse-pair excites an inhomogeneously broadened molecular system, this pattern gives rise to a similarly periodic variation of the population density as a function of transition frequency in the ground as well as in the excited state. This is referred to as a frequency grating (Hesselink and Wiersma, 1979; Duppen and Wiersma, 1986). The population density of the excited state will have a maximum, and the ground state population a minimum, at the peaks in the power spectrum. The frequency grating in the ground state will be complementary to that in the excited state. The modulation depth of the frequency grating is a measure of the homogeneous dephasing that has occurred during the interval This is equivalent to saying that the frequency grating can only be formed when the homogeneous linewidth is less than the fringe spacing in the power spectrum: The frequency grating is described by the diagonal elements of the density matrix. If we assume negligible population relaxation, these elements can be written as (Hesselink, 1980):

In this equation, and are the wave vectors and phases of the excitations, is the relative detuning of the optical excitation, and and are the pulse areas. The maximum modulation depth is achieved when For this condition the resulting modulation of the ground and excited states is depicted in Fig. 1A. A third pulse, at a time after the second, will induce the so-called stimulated echo at time The maximum amplitude is again obtained if this is a pulse. The effect of this third pulse is to transfer the phase information that was ‘stored’ in the population grating in frequency space to the off-diagonal elements of the density matrix (Duppen and Wiersma, 1986). As a result a rephasing process sets in, similar to that which gives rise to the two-pulse photon echo. At a time after the third pulse the macroscopic

polarization is reestablished, resulting in the emission of the stimulated echo. For a two-level system the stimulated echo intensity and direction are given by:

The first exponential factor is due to the homogeneous dephasing which occurs during the interval between the first two excitation pulses, while the second exponential factor reflects eradication of the frequency grating through depopulation of the excited state. It can be seen that by measuring the stimulated photon echo intensity as a function of the waiting time for a fixed it is possible to measure the lifetime of the excited state. On the other hand, can be measured by varying while keeping fixed. An excited molecular system will ultimately

Accumulated photon echo measurements decay to the ground state, either directly by fluorescence or radiationless relaxation, or via relaxation through some long-lived intermediate state, such as the lowest triplet state. In the former case, the frequency grating will disappear on a time scale of but in the latter case the frequency grating in the ground state may persist for a considerable length of time, determined by the lifetime of the intermediate state. This third level, if it is efficiently populated, serves as a bottleneck reservoir for the excited state population (Fig. 1). It is now possible to stimulate a long-lived or anomalous photon echo from the ground state grating at a time scale corresponding to the bottleneck lifetime (Morsink et al., 1979). In the limit that the lifetime of the bottleneck state is much larger than and the echo intensity from a three-level system excited by pulses can be written as (Morsink et al., 1982):

with rate constants as defined in Fig. 1B. The term in the square bracket is the contribution from the frequency grating in the ground state which persists on a time scale of The amplitude of this persistent grating, and therefore the intensity of the anomalous photon echo, is determined by the yield of the bottleneck state, given by the factor In the case of triplet states, this is the same as the intersystem crossing yield. C. Accumulated Photon Echo Two- and three-pulse photon echo measurements require relatively intense pulses so that it is necessary to amplify the output of the mode-locked dye-lasers typically used in these experiments. In addition, some form of time-gating is often used to selectively monitor the echo intensity, for example by frequency up-conversion (Hesselink and Wiersma, 1978; Meijers and Wiersma, 1992). Therefore, such experiments present a significant

113

experimental challenge. A much simpler variant of photon echo measurements is the accumulated photon echo (Hesselink and Wiersma, 1979) which can be observed if a long-lived bottleneck state exists in the molecular system (cf. Fig. 1B). Successive excitation of the sample with pulsepairs within the lifetime of the bottleneck state will result in the accumulation of a frequency grating. Such pulse-pairs are readily generated at a repetition rate of about 80 MHz, while the lifetime of the bottleneck state can easily be of the order of 1 ms or longer as is the case for a triplet state. The accumulation effect can extend over hundreds of excitation pulses, and it is possible even with weak excitation pulses to obtain a large modulation depth of the frequency grating. The contribution of each pulse pair to the frequency grating is given by Eq. (4). The generation of an accumulated grating has been verified experimentally in the case of a bottleneck state with infinite lifetime (Rebane et al., 1983), i.e. a photoinduced chemical transformation upon optical excitation. After the sample – a polymerized porphyrazine styrol solution – was excited with a train of pulse-pairs, the frequency grating could be observed by scanning the transmission of the sample as a function of wavelength. An essential requirement for accumulation of the grating is that the phase difference (cf. Eq. (4)) must be maintained for all pulse pairs, at least for the lifetime of the bottleneck state. This condition is not difficult to achieve for typical CW mode-locked dye lasers. The relationship between the phase difference and the frequency grating has been examined experimentally by Jefferson and Meixner (1992). Each of the pulses in the train of pulse pairs incident on the sample, while sustaining the accumulated frequency grating, simultaneously gives rise to a stimulated echo. The intensity of this echo is strongly enhanced by the accumulation effect, and the echo phenomenon under these conditions is referred to as the accumulated photon echo. Of course, the direction in which the accumulated photon echo is observed is determined by the phase-matching condition given in Eq. (6). Of particular interest is the phase-matching configuration given in Fig. 2A. Here, and are the wave vectors of the first and second

114

pulse (1 and 2 in Fig. 2B) in a pulse pair, which are separated by a time delay We examine the accumulated photon echo generated by the first pulse (3 in Fig. 2B) of a subsequent pulse pair with a wave vector According to the phase-matching condition, the accumulated photon echo is generated with a wave vector at a time delay of with respect to pulse 3. Note that the accumulated photon echo coincides in direction and time with the second pulse of the pulse pair, while it can be shown that they also have the same phase. Therefore, constructive interference occurs between the accumulated photon echo and this second pulse, giving rise to a homodyne component in the measured intensity of the transmitted beam in the direction of This component is measured as an increase of the transmitted intensity, and is proportional to the amplitude of the photon echo. This means that the homodyne detected accumulated photon echo will decay as instead of the factor in Eq. (5). More importantly, this detection scheme provides a greatly enhanced sensitivity.

Thijs J. Aartsma et al. So far we have assumed that the inhomogeneous broadening is static and much larger than the homogeneous broadening. However, the first assumption is generally not valid for disordered systems like glasses and proteins. In these systems, low energy disorder modes cause spectral diffusion on all time scales from picoseconds to infinity, even at very low temperatures (Breinl et al., 1984; Meijers and Wiersma, 1991; Saikan et al., 1992; Thorn-Leeson et al., 1994). As a consequence the distinction between homogeneous and inhomogeneous broadening becomes ill-defined. More importantly, the photon echo decay time and the linewidth as measured in hole burning spectroscopy depend on the time scale of the experiment (Narasimhan et al., 1990). In accumulated photon echo measurements, this time scale is of the order of microseconds to milliseconds, while in hole burning it extends from milliseconds to seconds or even minutes. The concept of the frequency grating has a strong analogy with hole burning spectroscopy. Indeed, it may be shown that the accumulated photon echo decay and the homogeneous line shape are related by a Fourier transformation (Saikan et al., 1988). Therefore, if effects of spectral diffusion can be neglected, the information derived from both types of experiments is expected to be the same. Finally we note that it is not absolutely necessary to use transform-limited pulses for the generation of accumulated photon echoes (Asaka et al., 1984). This can be easily understood if it is realized that the contribution to the buildup of the frequency grating is determined by the correlation function of the excitation pulses. Experimentally, stochastic pulses which are not fully transform-limited can be used to enhance the time resolution of the measurements (Meech et al., 1986). However, the enhanced time resolution is obtained at the expense of spectral resolution. IV. Accumulated PhotonEcho: Experimental In a typical accumulated photon echo experiment the sample is irradiated by two excitation beams which originate from a single mode-locked dye laser. The experimental set-up is schematically

Accumulated photon echo measurements

presented in Fig. 3. The two parallel beams are focussed into the sample, resulting in a noncollinear geometry as in Fig. 2. One of the beams is delayed with respect to the other by reflection from a retroreflector mounted on a variable translation stage. In this configuration, the sample is excited by a succession of pulse pairs at a repetition rate determined by the characteristics of the mode-locked laser, typically about 80 MHz. The accumulated photon echo is detected in the direction of the delayed beam. To enhance the signal-to-noise ratio, both beams are amplitude-modulated and the signal is measured by phase-synchronous detection. The first beam is modulated by an electro-optic modulator (EOM) at a frequency which is much higher than the inverse of the lifetime of the bottleneck state. The intensity of the accumulated photon echo will be modulated at the same frequency, while the build-up of the frequency grating is not significantly affected. The delayed beam is modulated by a mechanical chopper at a frequency which is (slightly) lower than the inverse of the bottleneck lifetime. The detection system (Fig. 4) consists of a photodiode in combination with a tuned amplifier, the frequency being determined by that applied to the EOM, which is 8 MHz. The output of the amplifier is demodulated by an 8 MHz amplitude demodulation circuit. This demodulated signal,

115

carrying the 2.5 kHz modulation frequency of the delayed beam, is fed into a lock-in amplifier, the output of which is proportional to the intensity of the accumulated photon echo. This double modulation scheme provides a high sensitivity for measurements of intensity changes. Equally important for accumulated photon echo measurements, however, is the rejection of the background to the accumulated photon echo signal. The background arises from the change in transmission of the delayed beam upon

116 population of the bottleneck state. There may also be an absorbance change due to the population of the excited state, but this is typically several orders of magnitude smaller than the intensity of the accumulated photon echo. Note that the diagram in Fig. 3 is actually identical to that of a pump-probe setup for measurements of absorbance changes (van Noort, 1994), but in the latter the EOM is replaced by an acousto-optic modulator (AOM). The diffracted beam of the AOM is subject to a small frequencyshift (equal to the acoustic frequency of the AOM) which effectively destroys the phase relationship between successive pulse pairs incident on the sample (de Boer, 1991). This prevents the accumulation of a frequency grating. V. Energy Transfer The accumulated photon echo technique provides a method to directly measure the coherent lifetime of the initially excited state in pigment–protein complexes. Therefore, such measurements can provide information about the mechanism of energy transfer in photosynthetic antenna systems at low temperatures, where pure optical dephasing can be neglected. Note, that in the weakcoupling limit excitations are localized on single pigment molecules, and these excitations jump from one molecule to another at a rate which is given by the well-known Förster-equation (Förster, 1948). In the strong coupling limit, the excited states of the system are more properly described in terms of exciton states. Such states are formally described by diagonalizing the Hamiltonian matrix including the dipolar coupling. Since the corresponding wavefunctions are linear combinations of the molecular wavefunctions of the interacting chromophores, exciton states are intrinsically delocalized. In this formalism, energy dissipation corresponds to (vibronically induced) transitions between exciton states. At room temperature, exciton scattering processes may be very fast, and under such conditions energy transfer is essentially an incoherent hopping process except, possibly, at very short times (tens of femtoseconds) after exitation. At sufficiently low temperature, however, one would expect a much longer lifetime of the exciton coherence, and a

Thijs J. Aartsma et al. corresponding difference in the nature of energy transfer. Accumulated photon echo experiments were performed on the water-soluble, BChl a containing Fenna–Matthews–Olson (FMO) complex from green sulfur bacteria (Louwe and Aartsma, 1994) in the wavelength range of 810 to 830 nm. This complex is well characterized in terms of structure, intermolecular interactions, and spectroscopic properties (Olson, 1978; Vasmel et al., 1983; Swarthoff et al., 1980; Swarthoff et al., 1981; van Mourik et al., 1994). The structure of the FMO complex has been elucidated by X-ray crystallography with near atomic resolution (Matthews and Fenna, 1980; Tronrud et al., 1986) and consists of three subunits in symmetry, each containing seven BChl a molecules. The nearestneighbour distance between the BChls within each subunit is around 12 Å and that between BChls belonging to different subunits is approximately 24 Å, yielding dipolar interactions between the BChls with a maximum of and respectively. The absorption and circular dichroism spectra at cryogenic temperatures can be simulated reasonably well using an exciton model including all 21 BChls (Pearlstein, 1992; Lu and Pearlstein, 1993). Time-resolved spectroscopy has mostly been applied at room temperature (Causgrove et al., 1988; Lyle and Struve, 1990; Savikhin et al., 1994; Gillbro, 1988; Zhou et al., 1994), and showed that localization of excitations occurs in less than 1 ps. Fluorescence lifetime measurements at 77 K (Zhou et al., 1994) revealed a dominant time constant of 2 ns. The experimental results of the photon echo measurements at 1.4 K show an increasing rate of dephasing as the wavelength of excitation decreases, with time constants varying from subpicoseconds at the shortest to hundreds of picosecond at the longest wavelengths (Louwe and Aartsma, 1994). At all wavelenghts the photon echo decay is strongly multiexponential. However, all decays between 816 and 830 nm can be fitted within experimental error with the same set of four time constants, given by 430, 140, 50 and 8 ps. The corresponding values of are obtained by multiplying the time constants with a factor of two. We assign these time constants to the dephasing of discrete electronic energy states of

Accumulated photon echo measurements the FMO complex. The spectral distribution of these four components form distinct bands with a width of approximately The transitions are centered around 826, 825, 822 and 818 nm, respectively. Experiments at 5 K showed a similar distribution, but with time constants of 250, 75, 12.5 and 8 ps. At wavelengths shorter than 816 nm even faster components appear, down to less than a few hundred fs at 790 nm. A straightforward explanation for the discrete energy levels observed in the decay associated spectra is that they correspond to the lowest energy states in the exciton manifold of the FMO complex. In total there are 21 exciton states, 14 of which are pair-wise degenerate due the symmetry of the complex. At very low temperature, where thermally induced processes are quenched, excited state decay can only occur to lower energy levels in the system. This involves not only decay to the ground state, but also transitions within the exciton manifold induced by some form of vibronic coupling. According to this picture, the accumulated photon echo decay at 1.4 K represents the loss of exciton coherence either by decay to the ground state or by downward exciton scattering. The dephasing time for a given wavelength of excitation is expected to decrease according to the number of lower lying states, in agreement with the experimental observation. In addition, the relaxation time will also be affected by the phonon density of states as a function of frequency. With increasing temperature, we expect that thermally activated transitions within the exciton manifold will occur. Fig. 5 shows the homogeneous linewidths of the lowest exciton states of the FMO complex calculated from the echo decay at 821 and 826 nm, as a function of temperature between 1.4 and 30 K. The temperature dependence of the two longest time constants for the dephasing could be fitted with a power law (Völker, 1989), combined with an exponential term (Jackson and Silbey, 1983) representing thermally activated population of higher exciton states (R.J.W. Louwe and T.J. Aartsma, to be published). Persistent hole burning experiments at 4.2 K (Johnson and Small, 1991) revealed eight transitions in the manifold. Relatively narrow holes width, limited by spectral resolution)

117

were only observed in the longest wavelength region of the absorption band of the FMO complex. At shorter wavelengths broad holes were obtained with a width of This large width was attributed to rapid energy relaxation, with a time constant of 0.1 ps, to the lowest energy state in the system. These results differ substantially from those obtained from the accumulated photon echo. At present, the origin of this discrepancy is not clear. Presumably, it is related either to the mechanism of hole burning in this system, or to the effect of spectral diffusion which would affect hole burning and accumulated photon echo measurements to a different degree considering the different experimental time scales involved. Accumulated photon echo experiments have also been performed on the light-harvesting complex LHCII of green plants, which has a absorption band extending from 640 to 680 nm. The structure of this complex has been resolved to 3.8 Å resolution, and contains chlorophyll a as well as chlorophyll b. The accumulated photon echo decays are very similar to those of the FMO complex in the sense that we obtain a set of time constants with a distinct spectral distribution, and

118 comparable in magnitude to those of the FMO complex (R.J.W. Louwe and T.J. Aartsma, to be published). Therefore, we believe that the results of the FMO-complex are representative of the excited state dynamics in antenna complexes at low temperature. VI. Photon Echo Experiments On Reaction Centers The kinetics of the primary electron transfer reactions that occur in photosynthetic reaction centers (RCs) have been the subject of numerous investigations (reviewed in Kirmaier and Holten, 1993; Zinth and Kaiser, 1993; Shuvalov, 1993). Most of these investigations involved measurements to provide information about the populations dynamics of the states involved. In the accumulated photon echo experiments described here, the optical dephasing of the excited state of the primary donor P at low temperature is measured directly. By a comparison of the dephasing rate with the population dynamics, further insight can be gained about the primary step of charge separation. The first photon echo experiments on RCs of purple bacteria have been performed by Meech et al. (1985). From their experiments, they concluded that the excited state of the primary donor decayed within less than 100 fs, supposedly to an intra-dimer charge transfer state. Furthermore, they concluded from hole burning experiments, that the absorption band of the primary donor is largely homogeneously broadened. Contrary to these results, Johnson et al. (1989) later observed holes with a FWHM of several wavenumbers in transient hole burning experiments, which is in contradiction to the abovementioned very fast electronic decay. The electron-phonon coupling in these systems is quite strong, in contrast to what is commonly observed in monomeric chlorophylls in organic glasses (van der Laan et al., 1991). The accumulated photon echo decay at 1.5 K in the reaction center of the photosynthetic bacterium Rb. sphaeroides R26 was recently reexamined, and extended to measurements on the mutant (M)Y210W (P. Schellenberg, R.J.W. Louwe, S. Shochat, P. Gast, A.J. Hoff and T.J. Aartsma, to be published). This mutant exhibits

Thijs J. Aartsma et al. an extraordinarily long electron transfer time (Shochat et al., 1994; Nagarajan et al., 1993). The experiments were performed under conditions that the triplet state of the special pair, with a lifetime of about (Chidsey et al., 1985), functions as the bottleneck state. Fig. 6A shows the echo amplitude decay of RCs of Rb. sphaeroides R26. The decay is characterized by a very sharp feature around t = 0, and a longer-lived component with a much smaller amplitude. This decay pattern is in good agreement with the shape of the homogeneous line obtained by hole burning measurements (Jankowiak and Small, 1993). Considering the Fourier-transform relationship, the sharp feature in the photon echo decay can be attributed to a contribution of the phonon sideband (Saikan et al., 1988) and reflects the extremely rapid dephasing which is associated with vibronic relaxation. The longer decay component is correlated with the width of the zero-phonon line, and is presumably determined by electron transfer. This component is only observed in the wavelength range of 908– 920 nm, the same region where the zero-phonon hole is found in hole burning experiments (Johnson et al., 1989). Closer inspection of the longer decay component shows that it is biexponential, with time constants of 1.2 and 8 ps. The biexponential kinetics has also been observed in pump-probe and fluorescence lifetime measurements at room temperature as well as at cryogenic temperatures (Kirmaier and Holten, 1993; Müller et al., 1992; Nagarajan et al., 1990). These low temperature data agree reasonably well with the values obtained from the accumulated photon echo measurements. Fig. 6B shows the photon echo decay of RCs from the (M)Y210W mutant of Rb. sphaeroides at 918 nm. The fast component can again be attributed to the phonon contribution to the electronic transition. Considering the longer lived components, it can be concluded that the primary charge transfer in this mutant at low temperature is much slower than in the native species, similar to results obtained at room temperature (Shochat et al., 1994; Nagarajan et al., 1993). However, whereas the rate of charge separation in the native strain increases towards low temperature

Accumulated photon echo measurements

(Lauterwasser et al., 1991), that of the mutant decreases significantly (van Noort, 1994). Thus, both the native and the mutant RCs show a biexponential accumulated photon echo decay which presumably corresponds to the intrinsic dynamics of the primary step of charge separation. Several explanations have been put forward to account for this nonexponential kinetics which is also observed in pump-probe and fluorescence decay experiments. From our measurements, the so-called parking state model (Müller et al., 1992; Hamm et al., 1993) can clearly be excluded, because the dephasing of the P* state would then be monoexponential. The biexponential decay is more consistent with the dispersion model (Kirmaier and Holten, 1990; Müller et al., 1992) which assumes that the free energy difference between P* and I being the primary acceptor, is characterized by a more or less continuous distribution. This is based on the conjecture that, in many respects, proteins behave like glasses and that the intrinsic disorder gives rise to a broad distribution of energy levels. In the dispersion model, the appropriate fit-function would be a stretched exponential, whereas in the substate model, a multiexponential function would have to be used. It is difficult to distinguish

119

between these two cases on the basis of the accumulated photon echo decay measurements. VII. Conclusion and Perspectives The accumulated photon echo technique has been used to study excited state dynamics of molecular systems, and its range of applications now includes photosynthetic pigment–protein complexes. A prerequisite for the observation of the accumulated photon echo is the efficient population of a long-lived intermediate state in the relaxation of the excited state to the ground state. Such an intermediate state functions as a bottleneck in the relaxation process. In photosynthetic antenna complexes the bottleneck consists of the triplet state, while in reaction center complexes either the triplet state or the charge-separated state may be involved depending on experimental conditions. At sufficiently low temperatures, pure optical dephasing is negligible, and the accumulated photon echo decay is dominated by population relaxation. In photosynthetic reaction centers this decay is determined by primary charge separation. Thus it has been possible to measure the charge separation rate at low temperature in the

120 reaction center of Rhodobacter sphaeroides R26 and in that of the (M)Y210W mutant of the same species. The results are consistent with pumpprobe measurements. Is is also concluded that there is no reason to assume an early intermediate in the primary charge transfer step, as was suggested by Meech et al. (1985). In isolated antenna complexes, the echo decay is strongly multiexponential and dependent on the wavelength of excitation. The different time constants appear to be associated with distinct bands in the absorption spectrum. These bands are attributed to transitions to discrete exciton states, which implies that the excitations are intrinsically delocalized. The coherent lifetimes of the exciton states are surprisingly long, and this would suggest that a hopping model for energy transfer at low temperature may not be appropriate. It may be concluded that accumulated photon echo measurements have proven to be very useful in the study of excited state dynamics in pigment– protein complexes, although the full potential of this technique has not been realized yet in this field of research. The results sofar provided a new perspective on the dynamics of exciton states in such complexes, and this warrants further investigations. Acknowledgements This work was supported by the Life Sciences Foundation (SLW), which is subsidized by the Netherlands Organization for Scientific Research (NWO) and by the European Union. (Contract ER8CHBGCT 930361). References Aartsma TJ and Wiersma DA (1976) Photon echo relaxation in molecular mixed crystals. Chem Phys Lett 42: 520–524. Abella ID, Kurnit and Hartmann (1966) Photon echoes. Phys Rev 141: 391–406. Allen L and Eberly JH (1975) Optical Resonance and TwoLevel Atoms. Wiley, New York. Asaka S, Nakatsuka H, Fujiwara M and Matsuoka M (1984) Accumulated photon echoes with incoherent light in doped silicate glass. Phys Rev A 29: 2286–2289. Breinl W, Friedrich J and Haarer D (1984) Spectral diffusion of a photochemical proton transfer system in an amorphous organic host: Quinizarin in alcohol glass. J Chem Phys 81: 3915–3921.

Thijs J. Aartsma et al. Causgrove TP, Yang S and Struve WS (1988) Polarized pumpprobe spectroscopy of exciton transport in bacteriochlorophyll a protein from Prosthecochloris aestuarii. J Phys Chem 92: 6790–6795. Chidsey CED, Takiff L, Goldstein RA and Boxer SG (1985) Effect of magnetic fields on the triplet state lifetime in photosynthetic reaction centers: Evidence for thermal repopulation of the initial radical pair. Proc Natl Acad Sci USA 82: 6850–6854. de Boer S (1991) Optical Dynamics of Molecular Aggregates. Doctoral thesis, University of Groningen, The Netherlands. Dicke RH (1954) Coherence in spontaneous radiation processes. Phys Rev 93: 99–110. Duppen K and Wiersma DA (1986) Picosecond multiple-pulse experiments involving spatial and frequency gratings: a unifying nonperturbational approach. J Opt Soc Am B 3: 614– 621. Förster T (1948) Zwischenmolekulare Energiewanderung und Fluoreszenz. Ann Phys 6: 55–75. Gillbro T (1988) Picosecond energy transfer kinetics in chlorosomes and bacteriochlorophyll a-proteins of Chlorobium limicola In: Olson JM, Ormerod JG, Amesz J, Stackebrand E and Truper HG (eds) Green Photosynthetic Bacteria, pp 91–96. Plenum Press, New York. Hamm P, Gray KA, Oesterhelt D, Feick R, Scheer H and Zinth W (1993) Subpicosecond emission studies of bacterial reaction centers. Biochim Biophys Acta 1142: 99– 105. Hesselink WH (1980) Picosecond Dephasing andRelaxation in Molecular Mixed Crystals. Doctoral thesis, University of Groningen, The Netherlands. Hesselink W and Wiersma DA (1978) Picosecond photon echoes detected by optical mixing. Chem Phys Lett 56: 227–230. Hesselink WH and Wiersma DA (1979) Picosecond photon echoes from an accumulated grating. Phys Rev Lett 43: 1991–1994. Jackson B and Silbey R (1983) Theoretical description of photochemical hole burning in soft glasses. Chem Phys Lett 99: 331–334. Jankowiak R and Small GJ (1993) Spectral hole burning: A window on excited state electronic structure, heterogeneity, electron-phonon coupling, and transport dynamics of photosynthetic units. In: Deisenhofer J and Norris JR (eds) The Photosynthetic Reaction Center Vol II, pp 133– 177. Academic Press, San Diego. Jefferson CM and Meixner AJ (1992) Frequency-domain measurements of spectral hole patterns burned with phasecoherent pulses. Chem Phys Lett 189: 60–66. Johnson SG and Small GJ (1991) Excited state structure and energy-transfer dynamics of the bacteriochlorophyll a antenna complex from Prosthecochloris aestuarii. J Phys Chem 95: 471–479. Johnson SG, Tang D, Jankowiak R, Hayes JM and Small GJ (1989) Structure and marker mode of the primary donor state absorption of photosynthetic bacteria: Hole-burned spectra. J Phys Chem 93: 5953–5957. Kirmaier C and Holten D (1990) Evidence that a distribution of bacterial reaction centers underlies the temperature and

Accumulated photon echo measurements detection-wavelength dependence of the rates of the primary electron-transfer reactions. Proc Natl Acad Sci USA 87; 3552–3556. Kirmaier C and Holten D (1993) Electron transfer and charge recombination reactions in wild-type and mutant bacterial reation centers. In: Deisenhofer J and Norris JR (eds) The Photosynthetic Reaction Center Vol II, pp 49–70. Academic Press, San Diego. Lauterwasser C, Finkele U, Scheer H and Zinth W (1991) Temperature dependence of the primary electron transfer in photosynthetic reaction centers from Rhodobacter sphaeroides. Chem Phys Lett 183: 471–477. Levenson MD and Kano SS (1988) Introduction to Nonlinear Spectroscopy. Academic Press, San Diego. Louwe RJW and Aartsma TJ (1994) Optical dephasing and excited state dynamics in photosynthetic pigment-protein complexes. J Luminescence 58: 154–157. Lu X and Pearlstein RM (1993) Simulations of Prosthecochloris bacteriochlorophyll a-protein optical spectra improved by parametric computer search. Photochem Photobiol 57: 86–91. Lyle PA and Struve WS (1990) Evidence for ultrafast localization in the band of bacteriochlorophyll a-protein from Prosthecochloris aestuarii. J Phys Chem 94: 7338–7339. Matthews B and Fenna R (1980) Structure of a green bacteriochlorophyll protein. Acc Chem Res 13: 309–317. Meech SR, Hoff AJ and Wiersma DA (1985) Evidence for a early intermediate in bacterial photosynthesis.A photonecho and hole-burning study of the primary donor band in Rhodopseudomonas sphaeroides. Chem Phys Lett 121: 287–292. Meech SR, Hoff AJ and Wiersma DA (1986) Role of chargetransfer states in bacterial photosynthesis. Proc Natl Acad Sci USA 83: 9464–9468. Meijers HC and Wiersma DA (1991) Spectral diffusion in glasses: a photon-echo study of zincporphin in ethanol. Chem Phys Lett 181: 312–318. Meijers H and Wiersma DA (1992) Glass dynamics probed by the long-lived stimulated photon echo. Phys Rev Lett 68: 381–384. Morsink JBW, Hesselink WH and Wiersma DA (1979) Photon echo stimulated from optically induced nuclear spin polarization. Chem Phys Lett 64: 1–4. Morsink JBW, Hesselink WH and Wiersma DA (1982) Photon echoes stimulated from long-lived ordered populations in multi-level systems. The effect of intersystem crossing and optical branching. Chem Phys 71: 289–294. Müller MG, Griebenow K and Holzwarth AR (1992) Primary processes in isolated bacterial reaction centers from Rhodobacter sphaeroides studied by picosecond fluorescence kinetics. Chem Phys Lett 199: 465–469. Nagarajan V, Parson WW, Gaul D and Schenk CC (1990) Effects of specific mutations of tyrosine-M210 on the primary photosynthtic electron-transfer process in Rhodobacter sphaeroides Proc Natl Acad Sci USA 87: 7888–7892. Nagarajan V, Parson WW, Davis D and Schenk CC (1993) Kinetics and free energy gaps of electron-transfer reactions in Rhodobacter sphaeroides reaction centers. Biochemistry 32: 12324–12336. Narasimhan LR, Littau KA, Pack DW, Bai YS, Elschner

121 A and Fayer MD (1990) Probing organic glasses at low temperature with variable time scale optical dephasing measurements. Chem Rev 90: 439–457. Olson JM (1978) Bacteriochlorophyll a-proteins from green bacteria. In: Clayton RK and Sistrom WF (eds) The Photosynthetic Bacteria, pp 161–178. Plenum Press, New York. Pearlstein RM (1992) Theory of the optical spectra of the bacteriochlorophyll a-antenna protein trimer from Prosthecochloris aestuarii, Photosynth Res 31: 213–226. Rebane A, Kaarli R, Saari P, Anijalg A and Timpmann K (1983) Photochemical time-domain holography of weak picosecond pulses. Opt Comm 47: 173–176. Saikan S, Nakabayashi T, Kanematsu Y and Tato N (1988) Fourier-transform spectroscopy in dye-doped polymers using the femtosecond accumulated photon echo. Phys Rev B 38: 7777–7781. Saikan S, Lin JW and Nemoto H (1992) Non-Markovian relaxation observed in photon echoes of iron-free myoglobin. Phys Rev B 46: 11125–11128. Savikhin S, Zhou W, Blankenship RE and Struve WS (1994) Femtosecond energy transfer and spectral equilibration in bacteriochlorophyll a-protein antenna trimers from the green bacterium Chlorobium tepidum. Biophys J 66: 110– 114. Shochat S, Arlt T, Francke C, Gast P, van Noort PI, Otte SCM, Schelvis HPM, Schmidt S, Vijgenboom E, Vrieze J, Zinth W and Hoff AJ (1994) Spectroscopic characterization of reaction centers of the (M)Y210W mutant of the photosynthetic bacterium Rhodobacter sphaeroides. Photosynth Res 40: 55–66. Shuvalov VA (1993) Time and frequency domain study of different electron transfer processes in bacterial reaction centers. In: Deisenhofer J and Norris JR (eds) The Photosynthetic Reaction Center Vol II, pp 89–103. Academic Press, San Diego. Swarthoff T, de Grooth BG, Meiburg RF, Rijgersberg CP and Amesz J (1980) Orientation of pigments and pigmentprotein complexes in the green photosynthetic bacterium Prosthecochloris aestuarii. Biochim Biophys Acta 593: 51– 59. Swarthoff T, Amesz J, Kramer HJM and Rijgersberg CP (1981) The reaction center and antenna pigments of green photosynthetic bacteria. Israel J Chem 21: 332–337. Thorn-Leeson D, Berg O and Wiersma DA (1994) Low-temperature protein dynamics studied by long-lived stimulated photon echo. J Phys Chem 98: 3913–3916. Tronrud DE, Schmidt MF and Matthews BW (1986) Structure and X-ray amino acid sequence of a bacteriochlorophyll a-protein from Prosthecochloris aestuarii refined at 1.9 Å resolution. J Mol Biol 188: 443–454. van der Laan H, Smorenburg HE, Schmidt T and Völker S (1991) Permanent hole burning with a diode laser: excitedstate dynamics of bacteriochlorophyll in glasses and micelles. J Opt Soc Am B 9: 931–940. van Mourik F, Verwijst RR, Mulder JM and van Grondelle R (1994) Singlet-triplet spectroscopy of the light-harvesting Bchl a complex of Prosthecochloris aestuarii. The nature of the low-energy 825 nm transition. J Phys Chem 98: 10307–10312. van Noort PI (1994) Energy Transfer and Primary Photochem-

122 istry in Photosynthetic Bacteria. A Picosecond Time-Resolved Study. Doctoral thesis, Leiden University, The Netherlands. Vasmel H, Swarthoff T, Kramer HJM and Amesz J (1983) Isolation and properties of a pigment-protein complex associated with the reaction center of the green photosynthetic sulfur bacterium Prosthecochloris aestuarii. Biochim Biophys Acta 725: 361–367. Völker S (1989) Spectral hole-burning in crystalline and amorphous organic solids. Optical relaxation processes at low temperature. In: Fünfschilling J (ed) Relaxation

Thijs J. Aartsma et al. Processes in Molecular Excited States, pp 113–242. Kluwer Academic Publishers, Dordrecht. Zhou W, LoBrutto R, Lin S and Blankenship RE (1994) Redox effects on the bacteriochlorophyll a-containing Fenna-Matthews-Olson protein from Chlorobium tepidum. Photosynth Res 41: 89–96. Zinth W and Kaiser W (1993) Time-resolved spectroscopy of the primary electron transfer in reaction centers of Rhodobacter sphaeroides and Rhodopseudomonas viridis In: Deisenhofer J and Norris JR (eds) The Photosynthetic Reaction Center Vol II, pp 71–88. Academic Press, San Diego.

Chapter 8 Spectral Hole Burning: Methods and Applications to Photosynthesis N. Raja S. Reddy* and Gerald J. Small Ames Laboratory, Department of Chemistry, Iowa State University, Ames IA 50011, USA

Summary I. Introduction II. Experimental Methods III. Applications A. Bacterial Reaction Centers B. Hole Burning at High Pressures C. Antenna Complexes D. Constant Fluence Hole Burning Spectroscopy Acknowledgements References

123 124 126 129 129 131 133 134 134 135

Summary Spectral hole burning Spectroscopy (SHB) is a high resolution technique that has been fruitfully applied to the study of photosynthetic protein complexes as well as electronic transitions in amorphous systems. It overcomes the limitations from inhomogeneous broadening and provides an improvement in spectral resolution of 2–3 orders of magnitude. The experimental aspects of SHB, including an apparatus for the hole burning studies at high pressures are discussed in considerable detail. For several reaction center and antenna complexes, the dependence of hole profiles on the burn wavelength has been used to identify the site inhomogeneous and homogeneous contributions to the -region absorption profiles. It has also been used to determine the excited state linear electron–phonon coupling parameters. Such data is crucial for understanding the energy and electron transfer dynamics. For example, the invariance of zero phonon hole width of P870* on burn wavelength strongly suggests that the recently observed non-single exponential decay kinetics in primary charge separation are not due to a distribution of values for electronic coupling matrix elements. The observation of nonphotochemical hole burning (NPHB) for reaction center of Rhodopseudomonos viridis has led to the identification for the first time, of all six states (including the special pair upper dimer component) of a bacterial RC. Recent results of hole burning in FMO antenna complex at high pressures have identified the need for modifying existing theories of pressure dependence of hole burned spectra. Absorption spectra for reaction center of Rhodopseudomonas viridis at high pressures are also presented. Novel constant fluence hole burning Spectroscopy used in the identification of BChl 870 and BChl 896 bands in the antenna complex of Rhodobacter sphaeroides as well as the lowest energy state of LHC II complex of Photosystem II is dismissed. Abbreviations: BChl – Bacteriochlorophyll; Chl – Chlorophyll; CP47 – A 47 kDa core antenna complex of Photosystem II; FMO complex – Fenna–Matthews–Olson complex of green sulfur bacteria; LH 1, LH 2 – Antenna complexes of purple bacteria; LHC II – Antenna complex of plants; NPHB – *Correspondence: Fax: 1-515-2941699; E-mail: [email protected]

123 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 123–136. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

124

N. Raja S. Reddy and Gerald J. Small

Nonphotochemical hole burning; P, P870, P960 – Primary electron donor; PHB – Photochemical hole burning; PSBH – Phonon sideband holes; Q – (First) acceptor quinone; RC – Reaction center; SHB – Spectral hole burning; THB – Transient hole burning; ZPH – Zero phonon hole

I. Introduction 1994 marked the twentieth anniversary of spectral hole burning (SHB) spectroscopies as developed by R.I. Personov and L.A. Rebane and their coworkers (Gorokhovskii et al., 1974; Kharlamov et al., 1974) for application to the electronic transitions of chromophores embedded in amorphous or crystalline hosts at low temperatures. For about ten years now, SHB has been fruitfully applied to photosynthetic protein complexes. There are several attributes of SHB which ensure that such an application will (or should) receive greater attention. Which of these are most important depends on the problem being addressed, be it the role of phonons and structural heterogeneity in energy or electron transfer, exciton level structure in complexes containing several strongly interacting chlorophylls (Chls) or the frequencies and Franck–Condon factors of the optically active modes of the cofactors. From an analytical chemistry or materials characterization point of view, SHB can also be a useful technique as illustrated, for example, by the work of Chang et al. (1994) on the CP47 antenna complex of Photosystem II. The attributes of SHB important to the study of the excited state electronic structure and dynamics of photosynthetic protein complexes have been the subject of reviews by Reddy et al. (1992a), Jankowiak and Small (1993) and Jankowiak et al. (1993). These reviews provide broad coverage of the complexes which have been studied and discussion of the importance of the results obtained. Also included are discussions of the basic principles of SHB and the theory of spectral hole profiles which one needs for a complete analysis of hole spectra. (The theory has been recently extended to arbitrary temperature by Hayes et al. (1994).) However, little attention in the above reviews is given to the experimental aspects of SHB. For this reason, these aspects are the primary focus of this chapter. The currently

available approaches to the recording (reading) of hole spectra are discussed along with their limitations in application to photosynthetic complexes. In addition, an apparatus for the study of the effects of high pressure (~1.5 GPa) on the excited state electronic structure and transport properties is described and unpublished results are presented. Other unpublished or recently published results which highlight the attributes of SHB are also given. We assume that the reader is familiar with the basic principles of SHB. For coverage of “non-photosynthetic” applications of SHB, both basic and technological, the reader is referred to the book edited by Moerner (1987) and a review article by Jankowiak et al. (1993). The latter also includes a section on photosynthetic complexes. Those interested in the connection (via of the nonlinear optical susceptibility) between SHB and photon-echo spectroscopies are referred to the review by M.D. Fayer and coworkers (Narasimhan et al., 1990) and a book by Mukamel (1995). Briefly, SHB involves the excitation of a narrow isochromat of chromophores within an inhomogeneously broadened absorption band. SHB occurs when the decay of the excited isochromat to the ground state is either blocked (persistent hole burning) or delayed (transient hole burning). Blocking of the decay to the ground state can be due to a photochemical reaction initiated in the excited state such as the inner proton-tautomerization of the porphyrins or due to a rearrangement of the host environment (typically a glass or protein). These two mechanisms are commonly referred to as photochemical (PHB) and nonphotochemical (NPHB) hole burning. In transient hole burning (THB), an intermediate state or a shortlived product (that reverts to the initial material) in a photochemical reaction is used to store (temporarily) the depleted ground state population. In addition to the zero-phonon holes (ZPH) at the burn laser frequency, all these mechanisms give rise, due

Spectral hole burning to coupling of the chromophore to the matrix phonons, to real- and pseudo-phonon sideband holes (PSBH) at energies higher and lower than the ZPH. In a similar manner, real- and pseudovibronic holes appear due to loss of absorption at the excited state vibrational frequencies of the chromophore. To end this introduction, we briefly discuss the types of information on protein complexes SHB can provide and why they are important. Because it is a line-narrowing technique, SHB eliminates the contribution of inhomogeneous broadening to the widths of origin absorption bands. The most important contribution (at least for a good glass forming solvent and mild detergent) to stems from the fact that the members of an ensemble of a given complex have slightly different structures (conformations), a natural consequence of the “glass-like “ nature of proteins. By determining the dependence of the hole structure (ZPH and its PSBH) on the burn frequency within the origin band, one can not only determine but also the center frequency of the inhomogeneous distribution of zerophonon line (ZPL) transition frequencies. It has been found that is depending on the complex. At the same time one obtains the frequency distribution of the low frequency protein phonons which couple (are Franck–Condon active) to the optical transition as well as their coupling strength S (Huang–Rhys factor). For all antenna protein complexes which have been studied it has been observed that (Gillie et al., 1989; Johnson and Small, 1991; Reddy et al., 1991) the mean phonon frequency, is ~20– the width of the one-phonon profile is and that the coupling is weak, S < ~0.6. The quantity can be viewed as the optical reorganization energy; would be the Stokes shift (when the state being probed is fluorescent). To illustrate the importance of such results for transport dynamics we consider Förster energy transfer from one Chl (2) to another Chl (1) under the condition that the mean electronic energy gap ~ ~ For this not uncommon situation, the dominant accepting modes are the protein phonons. Furthermore, the usual overlap between the donor (2) and the acceptor (1) fluorescence and absorption bands cannot be

125 used to calculate the transfer rates at temperatures much below room temperature (~200 K) because of the inhomogeneous broadening. Required for a correct calculation of the temperature dependence of the rate is the single complex spectrum, comprised of the ZPL and phonon sideband, in the low temperature limit. (The spectrum for higher temperature can then be calculated.) Also required is the distribution of values. Fortunately, it has been shown that, if and are the inhomogeneous broadenings, the width of the is Thus, all ingredients are at hand for a calculation of the temperature dependence of the rate and extent to which the kinetics are dispersive (non-exponential) (Hayes et al., 1994). We note that the physics just described has been used to investigate whether the non-single exponential decay of the primary electron donor state, P870*, of Rb. sphaeroides is due to dispersive kinetics stemming from a static distribution of energy gap values (Lyle et al., 1993). As reviewed by G.S. Small and coworkers (Reddy et al., 1992a; Jankowiak and Small, 1993; Jankowiak et al., 1993), the width of the ZPH burned into the origin band of a can also yield the lifetime of that state in the low temperature limit, where dephasing is dominated by population decay. In fact, the width (FWHM) of the ZPH in is given by For ps, the width is Although ultrafast time domain spectroscopy is generally a more powerful and versatile approach to transport dynamics, it should be noted that hole burning is truly state selective in that it measures decay from the total vibrational zero-point level. This turns out to be important in consideration of whether or not thermalization of chromophore and bath modes precedes ultra-fast electron and energy transfer as first discussed by Tang et al. (1989) and Johnson et al. (1990). Finally, we emphasize that the superior resolution afforded by SHB can often lead to resolution of closely spaced This has proven to be most useful for the problem of exciton level structure in antenna complexes which possess a structural unit with several strongly coupled Chl molecules (Zuber and Brunisholz, 1991; Lu and Pearlstein, 1993). A very nice example of this is

126

N. Raja S. Reddy and Gerald J. Small

from the work of Johnson and Small (1991) on the Fenna–Matthews–Olson BChl a complex of Prosthecochloris aestuarii. In this work it was also revealed that downward cascading between exciton levels occurs in ~100 fs at 4.2 K. This result was later confirmed by the femtosecond experiments. II. Experimental Methods The basic experimental setup for a hole burning is relatively simple. It consists of a laser (burn) for exciting a narrow isochromat of an inhomogeneously broadened absorption band and a means for probing the absorption prior to and after the burning. Modifications to this basic setup are dictated by the types of information sought and whether the type of hole burning is persistent or transient, vide infra. Absorption changes are monitored via either the transmission of a probe beam or fluorescence excitation (provided the sample is fluorescent) in any one of several ways. When one is concerned only with the ZPH, the same laser as used for burning (after suitable attenuation) can be tuned to measure the ZPH in the transmission mode. However, in the study of photosynthetic systems, the probe laser is often replaced by light from a monochromator and modulation techniques are used to improve the signal to noise ratio. A schematic diagram of an experimental set up in our laboratory for persistent hole burning is shown in Fig. 1. Both fluorescence excitation mode as well as in laser transmission mode can be employed. A Coherent ring dye laser (CR 699–29) pumped by an argon ion laser provides the required radiation for burning and scanning the holes. Sample S is located in a liquid helium cryostat with four optical windows. Two photomultipliers PMTS and PMTR provide the sample and reference signal. When used in transmission mode, the photomultiplier PMTS is moved to position T and absorbance is obtained from the signals from PMTS and PMTR. In the fluorescence excitation mode, the signal from PMTR is used to normalize the fluorescence signal from PMTS. For high resolution scans (<~20 MHz) of sufficiently narrow ZPH, the laser frequency is scanned by scanning the intracavity etalon assembly (ICA). Wider scans (~50 nm)

are achieved by removing the ICA and scanning the Lyot filter (via a stepper motor). The photon counting set up is useful when the sample is weakly fluorescent. Transient SHB can also be performed with this setup provided the bottleneck state has a sufficiently long lifetime (<~100 Wannemacher et al. (1993) developed the experimental setup shown in Fig. 2. It is based on burn-scan technique (Cone et al., 1984) and employs diode lasers. This setup can be used for both persistent and transient HB spectra and is capable of scanning wider frequency ranges at much faster rates The laser frequency is scanned either by adjusting the temperature of the diode or by modulating the current flow through it. A variable voltage source, triggered by a pulse generator, controls the laser current (hence the laser frequency). The pulse generator also provides synchronizing pul-

Spectral hole burning

ses for a pair of acousto-optic modulators which are used to provide light pulses of the required temporal width (for burning holes). A Fabry– Perot etalon allows for laser beam diagnostics and measures the laser frequency during a scan. Additional electronics, such as the optical shutter and oscilloscope, are also synchronized by the pulse generator. Holes are burned with a constant laser diode current (i.e. fixed burn frequency) and holes are scanned by ramping the voltage applied to the diode. Hole detection in the transmission mode (using a photomultiplier and signal averaging) is accomplished with a digital oscilloscope and a personal computer. The time between hole burning and detection can be as short as The above two setups are limited in their scan ranges to a few tens of nm which limits the usefulness of SHB in studies of energy/electron transfer and excited state structure in photosynthetic complexes. This problem can be overcome by employing a Fourier transform (FT) spectrometer that replaces everything after the light attenuating

127

filters NDF in Fig. 1. It is capable of fast (100 scans/minute), high resolution scans over a broad wavelength region (400–1100 nm). In another setup in our laboratory, an FT spectrometer operating in the visible and near-infrared region has been used in combination with a Coherent CR899–21 Ti:sapphire laser. Holographic detection of holes, based on laser induced gratings, utilizes two beams overlapped at the sample in the hole burning process (Renn et al., 1985). The spatially modulated light intensity at the crossing of the two beams forms a spatial grating in the medium. The spatial grating is due to spatially modulated absorption and refractive index coefficients. For readout, one tunable beam is used and diffraction of that beam by the spatial grating is detected. Holographic detection is a highly sensitive zero-background technique. Its application to the study of biological samples is however, limited since the diffraction efficiency is a function of read frequency and the stringent requirement that the optical quality of the sample be high.

128

For SHB studies in antenna and reaction center complexes at variable pressure and temperature (4.2–300 K), we have utilized a specially designed (Unipress Equipment, Poland) high pressure system (Buinowski et al., 1973). The high pressure setup consists of a three stage gas compressor and a demountable high pressure cell, containing the sample, that is located in a liquid helium cryostat Fig. 3(a). Helium gas is used as the pressure transmitting medium. High pressures generated at the compressor are transmitted to the pressure cell by a flexible beryllium–copper alloy capillary tube. The three stages of the compressor are oil/gas pressure intensifiers that are interconnected, with the first stage also connected to a helium gas cylinder. The gas flow between the compressor stages is regulated by use of needle valves. At the heart of the compressor operation is a pump capable of generating oil pressures of up to 80 MPa with manually con-

N. Raja S. Reddy and Gerald J. Small

trolled output. High initial compressibility of helium gas requires that the compression ratio in the first stage is close to unity. The maximum gas pressure that can be achieved in the first two stages is about 400 MPa. Gas compression in the final stage is different from that in the earlier stages because it occurs at constant mass, i.e. the amount of gas in the volume defined by the third stage (the capillary connector and the high pressure cell) is kept constant. The maximum pressure achievable in the third stage is about 1.5 GPa (~15,000 atm.). Fig. 3(b) shows a section of the cylindrical high pressure cell, made of beryllium–copper alloy, in the plane of the windows. Four (leuco)sapphire windows (shown in black) one of which can be replaced by a pressure gauge provide optical access to the sample. Indium-coated brass O-rings are placed at the end of the conical plugs (P). The conical plug is held in place by a annular

Spectral hole burning

brass plug (B). Tightening the brass plugs crushes the O-rings between the conical plug and the cell body, thus isolating the high pressure sample region from outside. The capillary tube connection to the compressor is made from the top of the high pressure cell. We conclude this section by noting that the recording or reading of hole spectra in the transmission mode has been, by far, the preferred method for photosynthetic protein complexes. (The advantages of using a FT spectrometer for reading have been emphasized.) There are several reasons for this, including simplicity and that for the maximum benefits from hole burning one needs to acquire the entire hole spectrum, not just the ZPH. III. Applications

A. Bacterial Reaction Centers As reviewed by Jankowiak et al. (1993) photochemical hole burning has been extensively applied to the RC complexes of Rhodobocter sphaeroides and Rhodopseudomonas viridis and, in particular, to their primary electron donor states P870* and P960*, respectively. Lyle et al. (1993) and Reddy et al. (1993b) have published PHB spectra with a vastly improved signal/noise ratio. Theoretical analyses of the burn frequency dependence of the hole spectra led to an improved set of values for the quantities that determine the P870 and P960 absorption profiles (Table I). Included in this Table are the values of inhomogeneous broadening and electron–phonon coupling parameters.

129

The coupling is dictated by phonons with a mean frequency of and a special pair “marker” mode with a frequency of 115 and for P870 and P960, respectively. The corresponding S-values establish that the electron–phonon coupling associated with the optical transition is strong in striking contrast with the of antenna protein complexes (cf. Introduction). Thus, P* is also “special” in that its electron–phonon coupling is strong. How this strong coupling is connected with the charge transfer character of P* is discussed in the review by Jankowiak and Small (1993). Interestingly, the coupling for the other of the RC is weak. Fig. 4 shows the burn frequency dependence of the P870 hole spectrum from Lyle et al. (1993) for the deuterated RC of Rhodobacter sphaeroides. The weakness of the ZPH coincident with the burn frequency is the result of strong electron–phonon coupling. Small et al. (1992) developed a theory for non adiabatic electron-transfer and used the hole burning and other results to argue that the intriguing, non-single exponential decays of P*, vide supra, is not due to a distribution of P*acceptor state values. The finding of Lyle et al. (1993) that the ZPH-width of P870* is invariant to the value of the burn frequency within the inhomogeneous distribution of ZPL transition frequencies strongly suggests that the non-single exponential decays are not due to distribution of values for the electronic coupling matrix elements associated with the initial phase of charge separation. At present, a convincing explanation for the non-single exponential decay of P870* does not exist.

130

DiMagno et al. (1992) reported on the nonsingle exponential decay kinetics for P870* of a series of mutants from Rhodobacter capsulatus. We have studied that mutant in which the phenylalanine and tyrosine residues at sites L181 and M208 respectively are both replaced by histidine. Both amino acid residues are in the near vicinity of the special pair and accessory BChl molecules. Jia et al. (1993) reported that the decay kinetics for this mutant can be fit reasonably well with a double exponential (1.14 ps (54.4%) and 10.3 ps (45.6%)). It was the long lifetime (relative to the wild type) that attracted our attention since, all other things being equal, the ZPH of this mutant should be more pronounced than in the wild-type RC, cf. Fig. 4. The burn frequency dependent spectra for the mutant are shown in Fig. 5. Surprisingly, the ZPH is not observed for any burn frequency. Analysis of the dependence of the hole profile on burn frequency led to the electron–phonon coupling parameter values given in Table I. The key finding (from simulations) is that is significantly larger than for P870 (wild type and R-26) and P960. This explains why the ZPH is not observed. Despite the stronger coupling and, by inference,

N. Raja S. Reddy and Gerald J. Small

greater charge-transfer character of P870* for the mutant, the primary charge separation kinetics are slower. Whether or not this finding, together with others from hole burning studies, are commensurate with the standard diabatic state model for charge separation is currently under investigation in our laboratory. A photochemical hole burned spectrum for the RC of Rhodopseudomonas viridis is shown as the middle spectrum of Fig. 6. Again the population bottleneck state for hole burning is In the region of P960 one observes a weak ZPH coincident with the burn frequency as well as the progression associated with the aforementioned special pair marker mode, The responses of the higher energy to formation of are more difficult to interpret although it has been long suggested that feature 1 is the hole associated with band A (of the absorption spectrum) which is the upper excitonic dimer component of the special pair. Reddy et al. (1993a) succeeded in obtaining nonphotochemical spectra for the RC, an example of which is shown in Fig. 6 (top spectrum) which provided definitive proof for this assignment (Reddy et al., 1993b). Remarkably, the anti-hole (solid arrow)

Spectral hole burning

131

B. Hole Burning at High Pressures

of the P960 hole is red-shifted by relative to the origin component of the P960 absorption band, i.e. the photoinduced structural transformation stabilizes P960* The largest response from hole burning directly into P960 is hole 1 associated with absorption band A (due to The anti-hole of hole 1 is indicated by the dashed arrow and is blue-shifted by Interestingly, the equal and opposite shifting of the anti-holes can be explained in terms of the simple excitonic dimer model by assuming a reduction of ~0.5 A in Mg … Mg distance. Reddy et al. (1993b) also concluded that absorption band B is predominantly due to the out-ofphase linear combination of the of the two accessory BChl molecules while band C is due mainly to the in-phase linear combination but with a significant contribution form the two monomers of the special pair. Their work identified, for the first time, all six of a bacterial RC (bands D and E are primarily associated with the two bacteriopheophytin molecules).

It has been known for some time that pressure is an important tool for studying the structure and dynamics of proteins (Frauenfelder et al., 1990). In recent years, hole burning at elevated pressures has been used with systems that exhibit very sharp ZPH to measure compressibilities, solvent shifts etc. in amorphous (protein) hosts (Zollfrank and Friedrich, 1992a, 1992b). Only modest values of high pressure (< ~10 MPa) were required in these studies because of the narrow ZPHs (due to long excited state lifetimes). To date applications of high pressures to the study of photosynthetic systems are limited (Foguel et al., 1991; Redline et al., 1991; Redline and Windsor, 1992). A simple theory, which predicted a pressure dependence of the frequency shift of the spectral holes and their broadening data was described by D. Haarer and coworkers (Sesselmann et al., 1987). A fully statistical microscopic theory of pressure effects on spectral holes and of the inhomogeneous line shape itself, has been developed by Laird and Skinner (1989). This theory assumes that a) the transition frequency of a solute molecule can be described by a sum of pairwise interactions with each solvent and b) each solvent molecule occupies a position that is independent of other solvent molecules. Their theory predicted that the hole shift with pressure is burn frequency dependent (color effect) i.e. it depends on where in the inhomogeneous line the hole is burned. This dependence is utilized to obtain the compressibility values of the host matrix. Similar frequency dependence was predicted for hole width. Comparison of the experimental results (Zollfrank and Friedrich, 1992a, 1992b) with theoretical calculations for hole shift and width demonstrates that the theory of Laird and Skinner is general enough to predict the results of the pressure tuning experiments, as well as the inhomogeneous line shape itself. We now briefly describe the results of our high pressure NPHB experiments on the FMO antenna complex for the green sulfur bacterium Chlorobium tepidum. Strong excitonic interactions exhibited by BChl a molecules in the FMO complex have been studied in detail by Johnson et al. (1989, 1991). The FMO complex comprises

132

of a trimer of subunits (each subunit consists 7 symmetry inequivalent BChl a molecules). Johnson et al. (1991) identified eight excitonic states establishing that inter-subunit interactions must be taken into account for a proper description of the spectroscopic data. The low energy band at 825 nm, in fact results from the two symmetry related lowest energy states at 824 nm and 827 nm. Narrow zero phonon holes can be burned only in the lowest energy state (Fig. 7). The pressure shift rates obtained using both broad absorption bands as well as the narrow zero phonon holes are similar to those observed in other proteins However, no color effect is observed, i.e. all holes burned in the 825 nm band shift, within experimental error, at the same rate. It should however be noted that in contrast to chromophores embedded in glassy hosts, a high degree of (spatial) correlation exists among the various BChl molecules in the antenna and RC complexes. This correlation violates the assumption of mutual solute position independence in Laird–Skinner theory. The theory by Laird and Skinner should, therefore, be modified

N. Raja S. Reddy and Gerald J. Small

to take into account the strong (excitonic) interactions present among the solute molecules. We have recently also obtained data on the pressure dependence of the absorption spectra of the reaction center of Rhodopseudomonas viridis in PVOH, at both 290 K (data not shown) and 4.2 K (Fig. 8). While the pressure shift rate for the accessory BChl band at ~835 nm changes from (290 K) to (4.2 K), the corresponding value for P960 is relatively unchanged to The pressure coefficient for the upper dimer state is only about one third of that for P960 That the pressure coefficients for the upper dimer component and P960 differ by a factor of 3 can be qualitatively understood by considering that the frequencies of the upper and lower dimer components are affected not only by the changes in excitonic interaction but also the solvent shift for the BChl molecules of the special pair. Using the above data, the pressure rate of increase in solvent shift is and that for excitonic splitting is However, another possible explanation is that charge-resonance states of the special pair, which contribute to P960* and the upper dimer component, are responsible for the asymmetry of the shifts. Freiberg et al. (1993) have determined the pressure

Spectral hole burning coefficients for the accessory BChl band (0.26 and the P870 band of the Rhodobacter sphaeroides reaction center which is structurally very similar to that of Rhodopseudomonas viridis. Increase in the shift rate for P870 is surprising because the average BChl macrocycle separation is larger in P870 (3.5 Å) than in P960 (3.3 Å). From our data it appears that changes in interactions with nearby amino acid residues is a more likely cause for the differences in shift rates. It should also be noted that the two weaker bands in Fig. 8 at 808 nm and 795 nm are due to the two bacteriopheophytin molecules of the reaction center and that they exhibit only a weak pressure dependence. These results are very encouraging in the sense that they establish that energy gaps between the six states of the reaction center exhibit a significant pressure dependence. Thus, one can expect that the energy and electron transfer dynamics should also be affected. We will be continuing our investigations on both P870 and P960 at high pressures. An important question in the study of the function of transport proteins (e.g. horse radish peroxidase and myoglobin) is whether or not there is correlation between the tautamer configuration and the structure of the apoprotein. i.e. can a small structural change of a prosthetic group trigger a rearrangement of protein structure. Recently, this aspect was investigated for mesoporphyrin substituted (for heme) horse radish peroxidase (Friedrich et al., 1994) using high pressure hole burning. Low temperature absorption spectrum shows three major bands which are shown to be associated with at least four tautomeric states. Compressibility values estimated using hole shift with pressure data for two of the tautomers differ by a factor of 3. This change in compressibility of apoprotein (binding the different tautomers) corresponds to a transition from solid state phase to an almost liquid like phase, i.e. large scale structural changes in proteins can in fact be induced by small changes in the prosthetic groups. Similar correlation was observed for protoporphyrin substituted myoglobin (Zollfrank et al., 1992).

C. Antenna Complexes Initial hole burning experiments (van der Laan et al., 1990) in B800 – 850 antenna complex of

133

Rhodobacter sphaeroides observed persistent narrow ZPH in the B800 band that were ascribed to B800 – B850 energy transfer. However, these studies that concentrated only on ZPH widths failed to detect any holes in the BChl 850 band. Reddy et al. (1991) using an FT spectrometer based set up showed that hole burning in BChl 850 is as facile as in BChl 800 band (Fig. 9). That narrow ZPHs can be burned in the BChl 800 band establishes that its width is due to inhomogeneous broadening. The situation however, is different for BChl 850 band The width and position of the broad holes was independent of the burn wavelength. In addition to the broad holes, higher energy satellite holes (one which is coincident with BChl 800) were observed. Analysis of this data showed that the BChl 850 band is predominantly homogeneously broadened and the mode was identified as the dominant mode for BChl 800–BChl 850 energy transfer. The large (compared to kT at room temperature) homogeneous width of BChl 850 band explained the weak temperature dependence of the BChl 800–BChl 850 energy transfer (Reddy et al., 1991).

134

D. Constant Fluence Hole Burning Spectroscopy Reddy et al. (1992b, 1993c, 1994) have shown that when a broad absorption band contains more than one underlying bands with differing hole burning properties, then hole burned spectra (burned with constant fluence) obtained as a function of burn wavelength can be used to resolve the underlying bands. This approach is especially useful in the study of excitonically coupled antenna systems (Reddy et al., 1992, 1994) where narrower ZPHs can be burned in the lowest excitonic state. The idea behind this approach is that if the hole burning efficiency is constant across a band, the depths of ZPHs reflect the profile of the absorption being burned into. The BChl 870 and BChl 896 bands in the antenna complex of Rhodobacter sphaeroides are examples where this approach has been used very successfully. Based on these results BChl 870 and BChl 896 states were identified as the “shuttle” states between LH2 and LH1 complexes and LH1 and the RC.

N. Raja S. Reddy and Gerald J. Small

Similar constant fluence spectra (Fig. 10) obtained for the LHC II complex of photosystem II (Reddy et al., 1994) were crucial in identifying the lowest energy state of this antenna complex at 680 nm, 4 nm below the most intense absorption band at 676 nm. That this state is the lowest state has been confirmed using fluorescence spectra and the linear electron phonon coupling parameters obtained using SHB. Acknowledgements Research at the Ames Laboratory was supported by the Division of Chemical Sciences, Office of Basic Energy Science, and the U.S. Department of Energy. Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under contract no. W-7405–Eng-82. We would also like to thank R. Jankowiak and H.-C. Chang for help with the high pressure experiments and Deborah Hanson for providing us the Rb. capsulatus samples.

Spectral hole burning References Buinowski W, Porowski A and Laisaar AI (1973) Installation for optical investigations under high pressure at liquidnitrogen temperatures. Instrum Exp Tech (USSR) 16:274 – 278 Chang HC, Jankowiak R, Reddy NRS, Yocum CF, Picorel R, Seibert M and Small GJ (1994) On the question of chlorophyll a content of the Photosystem II reaction center. J Phys Chem 98:7725 –7735 Cone RL, Harley RT and Leask MJM (1984) Nuclear quadrupole optical hole-burning in stoichiometric J Phys C (Solid State Physics) 17:3101–3111 DiMagno TJ, Rosenthal SJ, Xie X, Du M, Chan CK, Hanson DK, Schiffer M, Norris JR and Fleming GR (1992) Recent experimental results for the initial step of bacterial photosynthesis. In:Breton J and Vermeglio A (eds) Photosynthetic Bacterial Reaction Center II: Structure, Spectroscopy, and Dynamics, pp209–217, Plenum Press, New York Foguel D, Chaloub RM, Silva JL, Crofts AR and Weber G (1992) Pressure and low temperature effects on the fluorescence emission spectra and lifetimes of the photosynthetic components of cyanobacteria. Biophys J 63:1613– 1622 and references therein Frauenfelder H, Aldering NA, Ansari A, Braunstein D, Cowen BR, Hong MK, Iben IET, Johnson JB, Luck S, Marden MC, Mourant JR, Ormos P, Reimsch L, Scholl R, Schulte A, Shyamsunder E, Sorensen LB, Steinbach PJ, Xie A, Young R and Yue KT (1990) Proteins and pressure, J Phys Chem 94:1024–1037 Freiberg A, Ellervee A, Kukk P, Laisaar A, Tars M and Timpmann K (1993) Pressure effects on spectra of photosynthetic light-harvesting protein-pigment complexes. Chem Phys Lett 214:10 –16 Friedrich J, Gafert J, Zollfrank, Vanderkooi JM and J Fidy J (1994) Spectral hole burning and selection of conformational substates in chromoproteins. Proc Natl Acad Sci (USA) 91:1029–1033 Gillie JK, Small GJ and Golbeck JH (1989) Nonphotochemical hole burning of the native complex of Photosystem I (PS I–200). J Phys Chem 93:1620 –1627 Gorokhovskii AA, Kaarli RK and Rebane LA (1974) Hole burning in the contour of a pure electronic line in Shpolskii system. JETP Lett 20:216 – 220 Hayes JM, Lyle PA and Small GJ (1994) A theory for the temperature dependence of hole-burned spectra. J Phys Chem 98:7337–7341 Jankowiak R and Small GJ (1993) Spectra hole-burning: A window on excited state electronic structure, heterogeneity, electron–phonon coupling, and transport dynamic of photosynthetic units, In: Deisenhofer J and Norris J (eds) Photosynthetic Reaction Centers, pp133–177, Academic press, Boston Jankowiak R, Hayes JM and Small GJ (1993) Spectral holeburning spectroscopy in amorphous molecular solids and proteins. Chem Rev 93:1471–1502 Jia Y, DiMagno TJ, Chan C-K, Wang Z, Du M, Hanson DK, Schiffer M, Norris JR, Fleming GR and Popov MS (1993) Primary charge separation in mutant reaction centers of Rb. capsulatus. J Phys Chem 97:13180–13191

135 Johnson SG and Small GJ (1991) Excited state structure and energy transfer dynamics of the bacteriochlorophyll a antenna complex form Prosthecochloris aestuarii. J Phys Chem 95:471– 479 Johnson SG, Tang D, Jankowiak R, Hayes JM, Small GJ and Tiede DM (1990) Primary donor state mode structure and energy transfer in bacterial reaction centers, J Phys Chem 94:5849–5855 Kharlamov BM, Personov RI and Bykovskaya LA (1974) Stable gap in absorption spectra of solid solutions of organic molecules by laser irradiation. Opt Commun 12:191– 194 Laird B and Skinner JL (1989) Microscopic theory of reversible pressure broadening in hole-burning spectra of impurities in glasses. J Chem Phys. 90:3274–3281 Lu X and Pearlstein RM (1993). Simulations of Prosthecochloris bacteriochlorophyll a protein optical spectra improved by parametric computer search. Photochem Photobiol 57:86 – 91 Lyle PA, Kolaczkowski SV and Small GJ (1993) Photochemical hole-burned spectra of protonated and deuterated reaction centers of Rb. sphaeroldes. J Phys Chem 97:6924–6933 Moerner WE (ed.) (1987) Topics in Current Physics, Persistent Spectral Hole Burning: Science and Applications, Vol. 44, Springer-Verlag, New York Mukamel S (1995) Principles of Nonlinear Spectroscopy, Oxford University Press, New York Narasimhan LR, Littau KA, Pack DW, Bai YS, Elschner A and Fayer MD (1990) Probing organic glasses at low temperature with variable time scale optical dephasing experiments. Chem Rev 90, 439 – 457 Reddy NRS, Small GJ, Seibert M and Picorel R (1991) Energy transfer dynamics of the B800– 850 antenna complex of Rb. sphaeroides: A hole burning study. Chem Phys Lett 181:391–399 Reddy NRS, Lyle PA and Small GJ (1992a) Applications of spectral hole burning spectroscopies to antenna and reaction center complexes. Photosyn Res 31:167–194 Reddy NRS, Small GJ, Seibert M and Picorel R (1992b) B896 and B870 components of the Rb. sphaeroides antenna: A hole burning study. J Phys Chem 96:6458–6464 Reddy NRS, Kolaczkowski SV and Small GJ (1993a) A photoinduced persistent structural transformation of the special pair of a bacterial reaction center. Science 260:68–71 Reddy NRS, Kolaczkowski SV and Small GJ (1993b) Nonphotochemical hole burning of the reaction center of Rps. viridis. J Phys Chem 97:6934 – 6940 Reddy NRS, Cogdell RJ, Zhao L and Small GJ (1993c) Nonphotochemical hole burning of the B800 –B850 antenna complex of Rps. acidophila. Photochem. Photobiol 57:35– 39 Reddy NRS, van Amerongen H, Kwa SLS, van Grondelle R and Small GJ (1994) Low-energy exciton level structure and dynamics in Light Harvesting Complex II trimers from the Chl a/b complex of Photosystem II. J Phys Chem 98:4729 – 4735 Redline N and Windsor MW (1992) The effect of pressure on charge separation in photosynthetic bacterial reaction centers of Rps. viridis. Chem Phys Lett 198:334–340 Redline N, Windsor MW and Menzel R (1991) The effect of

136 pressure on the secondary (200 ps) charge transfer step in bacterial reaction centers of Rb. sphaeroides. Chem Phys Lett 186:204–209 Renn A, Meixner AJ, Wild UP, Burkhalter FA (1985) Holographic detection of photochemical holes. Chem Phys 93:157–162 Sesselmann Th., Richter W, Haarer D and Morawitz H (1987) Spectroscopic studies of guest–host interactions in dyedoped polymers: hydrostatic pressure effects versus temperature effects. Phys Rev B 36:7601–7611 Small GJ, Hayes JM and Silbey RJ (1992) The question of dispersive kinetics for the Initial phase of charge separation in bacterial reaction centers. J Phys Chem 96:7499–7501 Tang D, Jankowiak R, Small GJ and Tiede DM (1989) Structured hole burned spectra of the primary donor state absorption region of Rps. viridis Chem Phys 131:99–113 van der Laan H, Schmidt Th, Visschers RW, Visscher KJ, van Grondelle R and Volker S (1990) Energy transfer in

N. Raja S. Reddy and Gerald J. Small B800 – 850 complex of purple bacteria Rb. sphaeroides: a study by spectral holeburning Chem Phys Lett 170:231– 238 Wannemacher R, Koedijk and Völker S (1993) Spectra diffusion in organic glasses. Temperature dependence of permanent and transient holes. Chem Phys Lett 206:1–8 Zollfrank J and Friedrich J (1992a) Pressure shift and solvent shift: A hole-burning study of resorufin-doped glasses. J Phys Chem 96:7889–7895 Zollfrank J and Friedrich J (1992b) Spectral holes under pressure: Proteins and glasses. J Opt Soc. Am B 9:956 – 961 Zollfrank J, Friedrich J and Parak F (1992) Spectral hole burning study of protoporphyrin IX substituted myoglobin Biophysical J 61:716 – 724 Zuber H and Brunisholz RN (1991) Structure and function of antenna polypeptides and chlorophyll-protein complexes: Principles and variability. In: Scheer H (ed.) Chlorophylls, pp 627– 641, CRC Press, Boca Raton, FL

Chapter 9 Infrared and Fourier-Transform Infrared Spectroscopy Werner Mäntele Institut für Physikalische und Theoretische Chemie, Universität Erlangen-Nürnberg, Egerlandstrasse 3, 91058 Erlangen, Germany

Summary I. Introduction: Looking Back 100 Years II. What Can Infrared Spectroscopy Tell us about the Processes in Photosynthetic Membranes and Reaction Centers? III. From Bands to Bonds: Strategies for Band Assignments IV. Fourier-Transform Infrared (FTIR) Spectroscopy A. Basic Principles B. FTIR Spectroscopy used to Probe Protein Structures C. Techniques for Reaction-Modulated Difference Spectroscopy 1. Light-Induced Difference Spectroscopy 2. Electrochemically-lnduced Difference Spectroscopy 3. Photo-Chemo-Triggering: The Use of Caged Compounds D. Time-Resolved FTIR Spectroscopy 1. Rapid-Scan FTIR Spectroscopy 2. Stroboscope FTIR Spectroscopy 3. Step-Scan FTIR Spectroscopy V. Single Wavelength IR Techniques A. Dispersive Spectrometers B. Tunable IR Lasers C. Picosecond Pump-Probe Techniques VI. Sample Preparation for Infrared Spectroscopy VII. Conclusions and Outlook Acknowledgements References

137 138 139 139 141 141 144 145 145 147 148 148 149 151 151 152 153 153 154 155 157 157 157

Summary

During the last decade, infrared spectroscopic techniques have entered the field of photosynthesis and they are meanwhile established methods for probing chlorophyll or quinone molecules in their binding sites and for the investigation of molecular processes in the protein upon electron transfer or proton transfer. This advancement has mainly become possible with the rapid development and availability of Fourier transform infrared (FTIR) spectrophotometers, but also with the development of sensitive infrared semiconducor detectors and tunable IR lasers or picosecond IR laser systems. This chapter introduces the basic concepts for using IR Spectroscopy to study such complex systems as a reaction center or even complete photosynthetic membranes. The basic principles of FTIR spectroscopy for steady-state or time-resolved investigations will be described, and examples from the study of primary electron transfer in reaction centers will be given. Single-wavelength time-resolved techniques for the Correspondence: Fax: 49-9131-858307; E-mail: [email protected]

137 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 137–160. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

138

Werner Mäntele

millisecond-to-nanosecond and pump-probe techniques for the picosecond time domain are reported. A separate section deals with the strategies for the assignment of IR bands, an essential issue for the use of IR techniques in studies of macromolecules. Furthermore, techniques for the preparation of IR samples from photosynthetic membranes or reaction centers are described. Abbreviations: B – accessory bacteriochlorophyll; BChl – bacteriochlorophyll; CW – continuous wave; FTIR – Fourier transform infrared; – primary electron accepter (pheophytin); MCT – mercurycadmium-telleride; NIR – near infrared; P – primary electron demor; Q – acceptor quinone; – first and second acceptor quinones; RC – reaction center

I.

Introduction: Looking Back 100 Years

In the first volume of Physical Reviews in 1893, E.F. Nichols reported “a study of the transmission spectra of certain substances in the infrared”, using a spectrophotometer occupying two rooms, with a prism mounted in the wall between the two labs, and an assistant in the detector room taking the galvanometer reading with and without the substance in the beam (Nichols, 1893). Nichols noted that “the diathermancy of chlorophyll solutions depends on the type of solvent” he used, and thus was the first one to discover spectral shifts of chlorophyll absorption due to aggregation states. Although we would nowadays consider this as a near-infrared (NIR) experiment, it is the first study using long-wavelength radiation in photosynthesis. With increasing knowledge of the chemistry of chlorophylls and the advancement of infrared instrumentation, vibrational spectroscopy in the mid-infrared region became interesting for structural studies of the photosynthetic pigments, and a paper by Stair and Coblentz (1933) reported the infrared absorption spectra of some plant pigments. In the sixties and seventies, vibrational spectroscopic studies, mainly from the JJ Katz school, were extensively used for the investigation of chlorophyll aggregation and hydration (Katz et al., 1966, 1978; Ballschmiter and Katz, 1968, 1969). Although all these studies were on isolated pigments in solvents, it is obvious that they were undertaken in order to explain certain properties of the pigments in their native protein environment. As for the protein side, history goes back to the 50’s when Elliot and Ambrose (1950), in a

Nature article on the structure of synthetic polypeptides, pointed out the specific IR absorption of polypepide secondary structures, long before crystallography could provide us with atomic information on protein structures. This spectroscopic approach has become well-established with a theoretical background and numerous applications for the analysis of secondary structures of proteins (for a review, see Arrondo et al., 1993). In order to apply the vibrational spectroscopy of pigments together with those of polypeptides to the analysis of pigment-protein complexes, not necessarily in photosynthesis, technical developments had to be awaited. A breakthrough has been the development, the availability, and the “affordability” of Fourier transform IR spectrophotometers, which, for reasons given below, can provide the sensitivity needed to monitor vibrational modes in situ, either from a pigment in its native environment or from the protein. In this review, we shall briefly discuss the use of IR spectroscopy in the convenient sense for amide-I band analysis of photosynthetic pigmentprotein complexes. The major part will be dedicated to the discussion of FTIR techniques which provide the sensitivity to monitor changes at individual bonds of, say, a photosynthetic reaction center. The perturbation techniques needed to achieve this sensitivity lead to difference spectra, either between the stable states of a reaction or for metastable intermediates. We shall extend these FTIR techniques to time-resolved spectroscopy, although only very few applications have been attempted up to now in photosynthesis. Single-wavelength IR techniques will be discussed in more detail, in particular since recent develop-

Infrared and Fourier-transform infrared spectroscopy ments have led to vibrational IR studies with picosecond time resolution on photosynthetic reaction centers. A section on sample preparation for infrared spectroscopy and one on the sisyphean task of IR band assignment will provide the reader with the information that no photosynthetic protein is too complicated to be analyzed, and with the warning that to obtain nice spectra is relatively easy as compared with the task to interpret them. II. What Can Infrared Spectroscopy Tell us about the Processes in Photosynthetic Membranes and Reaction Centers? Infrared spectroscopy analyzes vibrational properties of a molecule and can thus address other properties of the RC than those monitored by optical spectroscopy in the UV, visible, and nearinfrared region. The vibrational spectrum of the RC is composed of individual and coupled modes of the numerous bonds of the protein, the pigments, the quinones, lipids, and water constituting the entire complex, and thus contains abundant information on structural details and functional properties. In this chapter we shall deal with the information which can be drawn from this “messy” overlap of multiple IR bands. This subject is also covered in reviews by Hoff (1992) and Mäntele (1993a,b, 1995). In infrared spectroscopy, photons with energies which match those of vibrational sublevels of the molecule (typically in the 0.5–0.05 eV range) are directly absorbed. These low energy photons do not cause photochemistry and thus can be used at high intensities. A large variety of techniques in infrared spectroscopy has emerged meanwhile, and most of them can be applied to obtain useful information on structural and functional properties of the RC. Infrared spectroscopy probes vibrational modes of the pigments, the quinones, the polypeptide backbone and amino acid side chains without any selectivity and may thus be used to address questions on pigment and quinone in situ properties and interactions, on the role of the protein mediating electron transfer and coupling proton transfer to electron transfer. This is complementary to Raman spectroscopy, where reson-

139

ance enhancement can yield spectra of individual pigments even in a complex photosynthetic membrane (Lutz and Mäntele, 1991). A problem of the non-selectivity of IR spectroscopy is the background absorption of the bulk part of the RC, of liquids, water, and buffer, which necessitates difference techniques. Since the size of the RC protein does not allow “classical” difference spectroscopies where two different samples are compared, “reaction-induced” IR difference techniques on a single sample have to be used. The “trigger” inducing the reaction should provide a minimum disturbance of the sample, but specifically and quantitatively start the desired reaction. The result are highly structured difference spectra which represent the sum of all molecular changes associated with the induced reaction, and which are selective for the functionally important parts rather for the global structure. At present, intrinsic photochemical reactions, electrochemical reactions in a spectroelectrochemical cell, and indirect photochemical reactions releasing an effector molecule which then starts the desired reaction, have been applied (for a review, see Mäntele, 1993a). Another possibibilty for a chemical triggering of a reaction using an ATR (attenuated total reflection) flow cell has been demonstrated (Baenziger et al., 1993). IR spectroscopy relies on physical processes with short intrinsic litetimes and thus allows very high time resolution for the study of intermediate states in electron transfer as well as of the dynamic processes in the protein. The past years have seen developments especially in time-resolved IR spectroscopy which have given access to all steps of electron transfer, from picoseconds to several seconds. To date, quite a few processes in RC have been detected in real-time in this entire time range which are functionally related to the electron transfer and proton transfer phenomena, but kinetically uncoupled from these. Vibrational spectroscopy can thus provide us with a detailed molecular view of the role of the protein in mediating electron transfer. III. From Bands to Bonds: Strategies for Band Assignments Ideally, the bands in an IR spectrum are assigned to the vibrational modes of the molecule by per-

140 forming a normal mode analysis and matching the force field parameters by replacing all atoms with stable isotopes (for the details of this procedure, see Colthup et al., 1990). Although normal mode analyses have been performed for small proteins, reliable calculations are far out of reach for the RC. Even without the limitation of computer capacity, a force field yielding consistent modes would be difficult to obtain. In the case of proteins, the concept of group frequencies, i.e. of vibrational modes uncoupled from others, has proven to be very useful. This concept assumes the vibrational spectrum of a protein to be composed of independent vibrational modes from the backbone, from amino acid side chain groups, from cofactors, lipids, detergent and buffer molecules, and water. Coupling of such modes is possible, but will only result in small shifts depending on the interactions. For each of these group frequencies, an empirical assignment is then attempted. One of the strategies for an empirical assignment consists of isotope labelling of particular bonds and comparison of the observed with the calculated frequency shifts. In the case of a C=O stretching mode, replacement of by should result in a ca. downshift of the frequency according to (F: force constant; reduced mass). The most simple isotope labelling experiment is substitution of by easily done by resuspending concentrated RC or membrane suspensions in a buffer and equilibrating. The resulting exchange of accessible N-, O-, or Sbound protons leads to shifts in the spectrum, which can be over for stretching modes. Exchange may happen at many sites and may thus lead to spectra in completely different from those in with the result that a clear assignment is not possible either. The probability for a given group to exchange depends on the global accessibility for water to the protein domain, on the local environment, and on the pK of the group. The extent of exchange can be monitored at the relative intensities of the amide II and the amide II' modes (the amide II mode arises from a coupling of the N–H and the C–N modes of the peptide bond, the corresponding mode in is termed amide II' ) modes. As a further complication, substitution of by

Werner Mäntele does not only act as simple mass label, but will also lead to changes of the pKs of groups and to kinetic isotope effects which can mess up an enzyme reaction. There is thus no guarantee that this isotope substitution leads to a conclusive assignment. Other mass labels, like or can be introduced biosynthetically by growing microorganisms on isotope-labeled nutrients, but will end unspecifically. Ideal, but still in its infancy, is site-directed isotope labeling, i.e. the placement of a non-natural amino acid in an expression system to a specific position in the polypeptide chain. Although the feasibility has been demonstrated for bacteriorhodopsin (Sonar et al., 1994), its application in photosynthesis still needs to be demonstrated. At present, isotope-labelled amino acids which are introduced into the RC via the nutrient, in combination with site-directed mutations, present a tedious yet sufficiently clear tool to identify and to assign the vibrational modes of a specific amino acid. In general, the remarkable potential of sitedirected mutagenesis of the RC is a tempting tool for IR assignments. However, any mutant resulting in severe blocking of the function renders reaction-induced IR difference spectroscopy useless. A further problem of site-directed mutagenesis, in the case of charged residues, is the perturbation of entire charge patterns, for which IR difference spectroscopy (which monitors the changes of dipole moments) is extremely sensitive. The balance seems to require generation of mutants with sufficiently small perturbations, but with retained general functional properties. In this line, natural variants can be considered of a similar relevance. In the RC, the exchange of cofactors by analogs is possible for the secondary and the primary quinone acceptor (Gunner and Dutton, 1989) as well as for the monomeric bacteriochlorophylls and the bacteriopheophytins (Struck and Scheer, 1990; Struck et al., 1990). The replacement of quinones by structurally different isotopically labelled analogs has become a useful tool for the assignment (Breton et al., 1994a–c; Brudler et al., 1994). In addition, the comparison of light or redox-induced FTIR difference spectra of reaction centers with the spectra of the isolated cofactors, i.e. bacteriochlorophylls, bacteriopheophy-

Infrared and Fourier-transform infrared spectroscopy tins, and quinones and all their numerous analogs, all in their neutral and relevant ion radical, dianion, or protonated forms, has created a basic data set of vibrational modes (Bauscher and Mäntele, 1992). Altogether, numerous vibrational modes from pigments, quinones, hemes, the polypeptide backbone and the amino acid side chain groups can be expected. Whether these contribute significantly to IR difference spectra depends on the (differential) extinction coefficients. In general, molar extinction coefficients of vibrational modes are at least an order of magnitude smaller than even moderate extinction coefficients of electronic transitions and may at most reach 2000– for bonds. For most of the modes contributing to the spectra of reaction centers and other photosynthetic proteins, smaller extinction coefficients are valid: around for a peptide C=O mode involved in an between 250 and for the COOH C=O mode of aspartic acid or the antisymmetric and symmetric modes of aspartates, and < 501 for the His mode. The water H–O–H bending mode, which peaks at ca. 1650 has only a small (extinction coefficient) of around However, since the concentration of water is 55 mol/l, even a layer will lead to an absorbance of l. In order to cope with this high background absorbance of water, concentrated protein solutions or suspensions or hydrated films have to be used. Table 1 summarizes the vibrational modes which can be expected to contribute to the infrared absorbance or difference spectra of a reaction center. Only the dominant modes in the to region are given, and we pretend, for the sake of simplicity, that they are all well localized group frequencies. Although many useful modes can be found at lower frequencies (i.e. below very little use has been made of these, since decreasing detector sensitivity and window transmission problems present experimental difficulties. On the other hand, the C–H, O–H, N–H or S–H frequencies above might well be used for diagnostic purposes, but this region is in general too congested for a detailed analysis.

141

IV. Fourier-Transform Infrared (FTIR) Spectroscopy

A.

Basic Principles

In a conventional infrared spectrophotometer, the infrared beam from a source, after passing the sample, is dispersed by a prism or a grating. A slit, which determines the width of the spectral element measured at a time is then used to block all other wavelengths, and the intensities of all spectral elements are successively measured by rotating the grating or the prism and correlating detector intensity with the position of the dispersive element. This results in slow acquisition of the spectrum since (1) all spectral elements are successively “scanned” and (2) the low energy in one spectral element necessitates damping with a high time constant. One possibility to reduce acquisition time would be the use of array detectors similar to an optical multichannel detecting system or a diode array, but unfortunately multiple IR detector elements from semiconductors (e.g., HgCdTe, InSb) are extremely costly and can hardly be fabricated with constant specifications as compared to silicon diodes. A Fourier transform spectrophotometer makes use of an interferometer, rather than scanning the individual wavelengths successively. An interferometer (typically of the Michelson type) produces an interference modulated beam which is passed through the sample and then reaches the detector. The detector intensity as a function of the interferometer position is called the interferogram. It is measured for a given mechanical mirror movement and digitally stored; its Fourier transform yields the IR spectrum. For the basic principles of Fourier transform infrared (FTIR) spectroscopy, the reader is referred to some reviews and textbooks (for example Griffiths and DeHaseth, 1986). A schematic representation of the optics used in an FTIR spectrophotometer is shown in fig. 1. The output of the detector–provided that no sample is present–is a function of the moving mirror position in the interferometer, but also of the spectral density of the source, of the transmission or reflection properties of all optical components, and of the spectral sensitivity of the detector. The exact position of the moving mirror is

142

determined by a He–Ne laser reference interferogram which is generated with the same interferometer and which controls, by detection of constructive and destructive interference pattern of the He–Ne laser, the timing of the data acquisition and digitization. The resolution of an FTIR instrument is determined by the distance d which the moving mirror travels from the point (“0”) where both branches

Werner Mäntele

of the interferometer exhibit the same optical pathlength. Typically, the mirror is moved from –d to +d, with the resolution being approximately 1/d wavenumbers. Each half interferogram (–d to 0 and 0 to +d) contains full spectral information, although both halves, even for a well-adjusted interferometer, are not necessarily symmetrical. Approximations that are made for the calcu-

Infrared and Fourier-transform infrared spectroscopy

lation of the spectrum from the interferogram include the limited movement of the mirror, which is not over an infinite distance as required for the procedure of Fourier transformation; thus, the interferogram is convoluted with an apodisation function. The details of data acquision, Fourier transformation, spectrum calculation, and some possible artefacts are discussed in a very useful paper by Gronholz and Herres (1985). FTIR spectroscopy results in a number of advantages over dispersive spectroscopy. The major improvement is the simultaneous recording of all spectral elements (“multiplex spectroscopy”) analogous to measurements in the visible using a diode array. This advantage is frequently termed Felgett’s advantage. Acquisition of an interferogram consisting of a few thousand data points (each digitized to at least 16 bit accuracy) takes less than a second even for a simple FTIR spectrophotometer, and only a few milliseconds for sophisticated “rapid scan” interferometers. We shall discuss below the relevance of this digitization time for time-resolved measurements in the rapid-scan mode. If the same total acquisition

143 time for a spectrum is allowed as for a dispersive instrument (i.e. at least several minutes), Felgett’s advantage can be used towards a gain in signal-tonoise-ratio by accumulating and averaging many interferometer scans before performing the Fourier transformation. On the other hand, in the rapid-scan mode, time-resolved studies in the msec domain are possible using series of single interferometer scans stored and transformed separately. The Felgett advantage of an FTIR instrument over a dispersive instrument depends on the width of the spectral range under consideration; as a rule of thumb, it is approximately given by the spectral range measured divided by the resolution, i.e. around 250 for the recording of the spectral region from 1000 to at resolution. A second advantage results from the absence of resolution-determining slits in an FTIR instrument. In a dispersive instrument, narrow rectangular slits are used to define resolution, resulting in a limited throughput even for moderate resolution. In an FTIR instrument, resolution is determined by the distance d the movable mirror travels, and only circular apertures are present to purposely limit energy flow and to collimate the beam. This higher energy throughput of an FTIR instrument, allowing higher sensitivity and the measurement of very low transmitting (“black”) samples, is often referred to as Jacquinot’s advantage. A third advantage of an FTIR instrument is the high wavelength accuracy because of the He– Ne laser reference interferometer present. Because of the known laser frequency, the wavelength accuracy may be as high as far more than needed for biological samples. This advantage is frequently referred to as Conne’s advantage. The usefulness of FTIR spectroscopy would be incomplete without the use of sensitive and rapid semiconductor detectors. For dispersive instruments, mostly thermocouples or the pneumatic Golay cell were used. FTIR instruments, in the standard versions, are mostly equipped with pyroelectric deuterated triglycine sulfate (DTGS) detectors, which have the advantage of a wide spectral sensitivity from the near IR down to a few hundred wavenumbers and which have a reasonably linear response. However, their responsivity

144 decreases with increasing modulation frequency, a fact which limits their application to rapid scan processes and fast data acquisition. Cooled (liquid semiconductor detectors from binary and tertiary semiconductors, either as photoconductors, as photovoltaic photodiodes or as biased photoconducting diodes, have limited spectral response, gradually increasing from short wavelengths to the maximum near the band edge. 1 Their detectivity may range up to Their response time can be in the order of nanoseconds, and even the standard mercury-cadmium-telluride (MCT) photoconducting detectors, selected by most FTIR users, have a response time in the domain. These detectors, in combination with fast digitizers (conversion rates around 100 to 200 kHz), allow to move the interferometer mirror at a high speed, close to the mechanical limits (see below in the section on rapid-scan FTIR spectroscopy). Apart from MCT detectors, which can be used for the to region, and which can be optimized for a frequency range of interest by varying the semiconductor composition, InSb detetors to or doped Ge detectors are occasionally used. A specific problem arising with semiconductor detectors is a nonlinear relation between the radiation intensity and the detector output at high intensities,one of the disadvantages of the multiplex detection. In order to avoid photometric inaccuracies or Fourier transform artefacts, it is recommended to use edge filters which cut out unnecessary frequency ranges. These filters mostly consist of coated Germanium; a long-wave pass filter mounted in front of the sample which cuts out, for example, the frequency range and passes the to range, will not only protect the sample from unnecessary heating, but also from photochemical reactions caused by the He-Ne beam which is coaxial to the IR beam, and will increase linearity as well as sensitivity (since the amplifier gain may be optimized). 1

The detectivity is a quantity defined from the responsivity (in Volts or amperes output per Watt of incoming radiation) and the noise voltage or current at a certain frequency and within a defined frequency interval, it allows to compare essentially different detector types and materials.

Werner Mäntele

B. FTIR Spectroscopy Used to Probe Protein Structures The IR spectrum of a protein is mainly characterized by vibrational modes from the peptide building units. A total of nine predominant modes of the CONH group (“amide modes”) are found in the region between and corresponding to in-plane and out-of plane modes with very different distribution of potential energy on the individual bonds (for details, see Arrondo et al., 1993). The highest frequency modes, “amide A” and “amide B”, at approx. and respectively, are almost purely NH stretching. In the region between and the protein IR spectrum is formed by the amide I mode between and (predominantly C=O stretching) and by the amide II mode around (a mixture of NH bending and CN stretching). The amide I absorption band is suitable for the analysis of protein structures, since its position depends on the type of secondary structure elements. Due to different Hbonding and environment in the various types of secondary structures, the amide I mode of an helix will absorb around , that of an antiparallel sheet around and of a parallel sheet around and of unordered structure around (for a discussion of the range of each mode, see a review by Arrondo et al., 1993). These numbers are based on one side on normal mode calculations, on the other side on the IR spectra of model peptides and small proteins with known structures. While a polypeptide of a single known secondary structure (assumed to be infinitely long in the case of helices and sheets) should give rise to a clear-cut band structure for the amide I mode, the inverse step–i.e. deduction of the secondary structure from the amide I band profile – is not always unequivocal for several reasons, even if the spectrum can be recorded at almost arbitrary precision using FTIR spectroscopy. First, the intrinsic bandwidth of the amide I band is typically higher than the separation of its constituents, assuming that different types of secondary structure lead to superposition of bands. Second, even if the decomposition of the

Infrared and Fourier-transform infrared spectroscopy amide I band is achieved, assignment of the type and extent of secondary structure to its constituents is not always unambiguous and needs additional support by other spectroscopic or biochemical data. Ideally, one would like to record a spectrum of the amide I region and obtain, via some kind of mathematical procedure, percentage numbers for helix, sheet contents, and so on. Unfortunately, the availability of corresponding software has considerably encouraged this type of “blackbox” use of IR spectroscopy, and occasional offthe-shelf assignments of secondary structures have discredited the substantial amount of serious spectroscopic work in this field (for a critical assessment of the method, see Surewiczs et al., 1993). Information on the type and composition of secondary structure and how it may change with some reaction is obtained by (1) recording highquality spectra of the sample in the range from approx. to (2) subtraction of the buffer, (3) application of resolution enhancement procedures such as derivation, deconvolution, Fourier self deconvolution or others, and (4) band fitting or band synthesis from possible constituents. It is obvious that either of these steps can bear a number of pitfalls: Photometric accuracy and detector nonlinearities enter in the first step, incorrect subtractions in the second, artefacts of resolution enhancements in the third and fitting ambiguities in the fourth. Finally, the contribution of amino acid side chain absorbance in the amide I region is substantial, but difficult to estimate. Another experimental difficulty is aggregation, possible orientation, and thus dichroism of the sample which leads to misjudgement of the intensities for secondary structure elements. Orientation of membranes and vesicles on an IR window tilted with respect to the measuring beam, on the other side, can be used to deduce the orientation of helices with respect to the membrane normal.

C. Techniques for Reaction-Modulated Difference Spectroscopy Although the sensitivity of FTIR spectroscopy is high enough to detect the changes of absorption brought above by alterations at one individual

145 bond even in a protein with 100 kDa and more, the straightforward approach to calculate a difference spectrum from two spectra of two different protein samples in two different states fails in most cases. The limits of precision for the adjustment of sample preparations and concentrations, cell pathlength, etc. leads to the frustrating insight that this approach is only appropriate for low-molecular weight proteins, for large spectral differences between the samples, or in spectral windows with low background absorbance such as the to range. Alternative techniques, in which the spectra of one single sample in two different states (for example, in the initial and final state of a reaction) are recorded in rapid succession, with a reaction induced in between, are more appropriate. A difference spectrum is thus obtained, which is selective for the functionally relevant parts of the protein rather than for the global structure. Since a suitable reaction is the trigger for the resulting difference spectra, the term reaction-induced or reaction-modulated IR difference spectroscopy has been coined (Mäntele, 1993a). The “trigger” employed should provide a minimum disturbance of the sample, but quantitatively and specifically start the desired reaction. Figure 2a illustrates the principle setup used for reaction-induced FTIR difference spectroscopy. Different triggers can act on the sample in this type of experiment and modify its state, which is then characterized by FTIR as well as by UV/VIS spectroscopy. 1. Light-Induced Difference Spectroscopy Light-minus-dark IR difference spectroscopy is the most straightforward version of reactionmodulated IR techniques, and can be used for many intrinsic photoreactions of photosynthetic pigment-protein complexes. FTIR difference spectra are obtained between dark and illuminated samples, either with the actinic light used to start the reaction and recording spectra before and after illumination, or under illumination creating different photostationary states. Excitation can be by continuous light or by a flash as outlined in Fig. 2a. The photostationary state of an RC sample in the FTIR spectrophotometer beam can be easily modified by additional bleaching light, which

146

Werner Mäntele

Infrared and Fourier-transform infrared spectroscopy does not perturb the measurement, since the spectral regions (VIS/NIR for excitation, mid-IR for detection) are well-separated. Light-induced FTIR difference spectra reflecting charge-separation between P and or are easily obtained at moderate bleaching intensities. The first spectra of this kind date back to 1985 (Mäntele et al. ,1985) and light-induced techniques of this kind have since then been used for the study of the molecular processes concomitant with primary charge separation for many different sample forms and RC and membrane preparations (for reviews, see Hoff (1992), Mäntele (1993a,b, 1995)). In the case of bacterial photosynthetic reaction centers, these spectra reflect the differences for the transition and are commonly termed or difference spectra, with the understanding that the bands of the disappearing neutral state, are negative, and those of the appearing chargeseparated state, are positive. Quite analogous to bacterial RCs, spectra reflecting charge separation in Photosystem I have been obtained (Nabedryk et al., 1989). However, care need to be taken with higher intensities of the bleaching light, and correct filtering is recommended, since even minor heating of the sample results in changes of hydration, leading to unspecific signals which can be much larger than those caused by light-induced charge separation. In general, it is advantageous to add a number of light-dark difference cycles rather than to accumulate a high number of scans for the dark-adapted or the lightadapted state. Needless to say that thorough thermostating of the samples is a prerequisite for high-quality spectra. Light-induced difference spectroscopy combined with the use of exogeneous donors can be used to obtain difference spectra between further states of the RC. Rapid rereduction of at low redox potential conditions leads to formation of a procedure which has been used for Rhodopseudomonas viridis RC and Chromatium vinosum membrane particles, and, recently, for Rhodobacter sphaeroides RCs (Nabedryk et al., 1986, 1995). At moderately reducing conditions, but in the presence of a redox mediator to shuttle electrons rapidly to the photooxidized reduction of or has been achieved without oxid-

147 izing a detectable fraction of P (Breton et al., 1991a,b). In RCs with bound cytochromes, playing with the redox potential for the IR samples has led to light-induced FTIR difference spectra where charge separation between the hemes and the quinone acceptors can be observed (Nabedryk et al., 1991). At low temperature, under reducing conditions, and at high-intensity illumination, the triplet difference spectrum has been obtained for Rb. sphaeroides RCs (Breton and Nabedryk, 1993). 2. Electrochemically-lnduced Difference Spectroscopy Steady-state FTIR difference spectra which have been obtained by use of the intrinsic photochemical reactions, however, present signals from a mixture of states of the donor side and the acceptor side. In the case of bacterial photosynthetic reaction centers, these are contributions from P, Q, and in a single spectrum reflecting charge separation. Difference spectra in which an exogeneous donor has been used seem to change only one cofactor’s redox state. However, the difference spectra due to the redox reaction of the exogeneous donor adds to that of the respective RC cofactor, although convincing evidence that the difference bands of the exogeneous donor are broad and featureless or simply in a different spectral region comes from control experiments. This limitation has been successfully overcome with the development of ultra-thin-layer electrochemical cells (Moss et al., 1990) which allow redox reactions in photosynthetic pigment-protein complexes to be triggered at a transparent electrode in combination with UV-VIS-NIR and FTIR spectroscopy. The design of this electrochemical cell is shown in Fig. 2b. Different types of working electrodes have been applied over the years, and the general idea is that a transparent layer of gold or platinum, a semiconductor electrode, or a gold grid electrode (see Fig. 2b) is chemically modified with bifunctional reagents following the idea of Armstrong et al. (1986). Ideally, this reagent adsorbs in a single layer on the electrode and presents to the solution a charge or a charge pattern which corresponds to that of the natural reaction partner of the protein

148 under investigation, thus allowing reversible docking of the protein close enough to the electrode to allow rapid electron transfer. Using this “direct” electron transfer, the addition of small redox-active molecules (“mediators”) can be avoided. Addition of mediators in micromolar concentrations with the appropriate redox midpoint potential, however, can considerably speed up equilibration. No spectral contributions from mediator redox reactions are observed for small enough concentration. Electron transfer at an electrode is more specific than light as a trigger, and different redoxactive cofactors present in a protein can be selected by the choice of the appropriate potential (“dial-a-cofactor”). These “redox-induced” difference spectra are as detailed and sensitive as are light-induced difference spectra; moreover, individual bands can be titrated and thus assigned to a specific redox transition. In the case of bacterial RCs, FTIR difference spectra of the donor side and of the acceptor side have been obtained (Leonhard and Mäntele, 1992; Mäntele et al., 1990a; Bauscher et al., 1993). The electrochemical oxidation of the electron donor of Photosystem I, P700, could be initiated electrochemically, and the hemes of the Rps. viridis and the Chloroflexus aurantiacus cytochrome subunit of the RC could be individually titrated (Hamacher et al., 1993; Fritz et al., 1992, 1993; Fritz, 1995). It should be mentioned that both redox-induced and lightinduced difference techniques can be combined in many cases on one sample to start photochemical reactions from a well-defined redox clamp. 3. Photo-chemo-triggering : The Use of Caged Compounds

Werner Mäntele “caged compounds”, has been reviewed by Corrie and Trentham (1993). For the inactivation of an effector molecule, chemical modification or sterical “caging” are conceivable. Figure 2c shows the structure and the net reaction of the archetype of “caged compounds”, caged ATP. With these “caged compounds”, time resolved studies of protein reactions are possible, with the time resolution only limited by the intrinsic reaction rates of the cage and diffusion to the protein. In this case, the experiment follows the scheme in Fig 2a with an intense UV flash initiating the release of the effector molecule. A general problem for the use of caged compounds for IR spectroscopy is the intrinsic IR difference spectrum from the triggering reaction itself, i.e. from the photolytic release of the cage, which comprises a primary phototransformation with subsequent dark reactions. However, most cages provide spectral “windows” where difference bands from the protein reaction are not masked by bands from the triggering reaction. These windows can be enlarged using isotopically-modified cages. There are many possible applications of “cages” for the study of photosynthetic proteins. Work performed up to now comprises a rapid and permanent change of pH induced by a “caged proton” (Fogel et al., 1993) and the release of from caged with subsequent binding to an ion-sensitive light-harvesting complex from Rb. sulfidophilus (Beck, 1992). It appears that further applications are only limited by the imaginativeness of the organic chemist who synthesizes the cage and combines biochemical requirements with requirements of vibrational spectroscopy.

D. Time-Resolved FTIR Spectroscopy The idea of reaction-modulated IR difference spectroscopy has been carried beyond the reactions which can be induced by intrinsic photochemical reactions or by electrochemical reactions. The rapid and uniform introduction of a substrate or an effector molecule by UV photolysis of an inactive but photolabile precursor molecule allows protein reactions to be triggered in an IR sample. The synthesis, the reactions, and the applications of such inactive photolabile precursor molecules, which have been termed

Actually, three variants of time-resolved FTIR spectroscopy have proven to be usetul to obtain time-resolved vibrational spectra of proteins, and (at least) two of them have been successfully applied to photosynthetic systems. For all three methods, the high energy throughput and the wavelength accuracy (termed Jacquinot and Conne advantage in section IV. A) can be fully exploited; however, they differ in the use of the Felgett (Multiplex) advantage. In addition, their applica-

Infrared and Fourier-transform infrared spectroscopy bility to reaction center studies stringently depends on the number of repeatedly excited processes needed for averaging to obtain a good signal-to-noise-ratio. Figure 3 illustrates the principles of time-resolved FTIR spectroscopy. 1. Rapid-Scan FTIR Spectroscopy In rapid-scan-FTIR spectroscopy illustrated in Fig. 3a (Braiman et al., 1987), the interferometer mirror is scanned back and forth in cycles as fast as possible and successive interferograms are stored in the computer memory (transformation of the interferograms can be performed later). A trigger, typically a flash or a short interval of illumination from a lamp defined by a shutter, is then fired between or during interferogram recording after sufficient interferograms have been stored to calculate a low-noise spectrum of the unperturbed sample (i.e. of the state before the flash). Difference spectra evolving with time are then calculated between the pre-flash spectrum and the spectra calculated from successive interferograms after the flash. A “blank” difference spectrum calculated from two sets of spectra of the unperturbed sample before the flash usually serves as a control for sample stability and defines a “level of confidence” for the later spectra. At first sight, mechanical problems in rapidly scanning a solid mirror driven by an electromagnetic linear motor on an air bearing at more than 20 cycles/second over approx. 1 cm amplitude – while interferometric accuracy of the mirror (i.e. its surface rectangular to the beam) must be maintained – seems limiting, but a serious bottleneck is given by the digitization of the detector signal, too. We keep in mind that 2000–8000 data points need to be digitized for the interferogram, each with at least 16 bit accuracy. Improved hardware and sophisticated software now allow to record interferograms during the forward and the reverse movement of the mirror (see Fig. 3a); these interferograms should in principle be identical, but in practice need a normalization procedure to overlap. Moreover, software procedures have been developed to use the four “half-interferograms” of a full mechanical forward-reverse interferometer cycle in order to provide more points on the time axis (see Fig. 3a). Principally, both “half-interferograms” (–d to 0 and 0 to +d)

149 contain full spectral information, and thus only need to be complemented by their mirror image to serve as a basis for the calculation of a spectrum. Again, only with an ideal interferometer the two halves of an interferogram are identical, and correction factors are needed to make identical spectra for the mirror moving from to –d to 0 and from 0 to +d. Using all these procedures, four spectra non-equidistant in time may be calculated from a single forward/reverse interferogram cycle. The duration of a cycle depends on the spectral resolution needed. After reversal of the movement at the extreme point of the mirror movement, a short way is needed before a uniform movement is reached and data can be recorded at –d or +d in order to avoid mirror vibrations. Rapid-scan FTIR spectroscopy is particularly suitable for slow reactions (msec to sec) or reactions which can only be triggered once (in case of irreversible bleaching of the sample or trapping of a state), or when back reactions are so slow that repetitive excitation becomes unattractive. It has been used successfully to obtain difference spectra between the and states of bacterial RC, by recording spectra immediately after a short illumination of RC containing approx. 50% (resulting in a mixture of and difference spectra) and succcessively after hundreds of milliseconds (resulting in spectra after full decay of the state. The normalized subtraction of both spectra gave a difference spectrum (Thibodeau et al., 1990a,b). Mechanical mirror movement and digitization of the interferogram presently set a limit for the time resolution, which is a compromise with the desired spectral resolution. At the spectral resolution needed for the studies of photosynthetic systems, which is mostly spectra are possible every 20–30 ms with a routine rapid-scan interferometer and software. In our laboratory, a setup making use of a Bruker Type 66 interferometer records spectra spaced on the average approx. 25ms for resolution, and approx. every 12 ms for resolution. Upon introduction of an appropriate delay between the start of the interferogram and the flash, the first spectrum can be recorded within 1–2 ms after the flash and thus be used for quite rapid reactions.

150

Werner Mäntele

Infrared and Fourier-transform infrared spectroscopy

The number of the spectra recorded this fast only depends on the computer memory, and typically coaddition of interferograms at later time intervals is used to improve signal-to-noise-ratio and to avoid extensive data floods. It can be foreseen that the time resolution for this version of rapid-scan FTIR spectroscopy can only marginally be improved by more rapid mechanical scanners and faster analog-to-digital converters. 2.

Stroboscope FTIR Spectroscopy

One of the bottlenecks of rapid-scan FTIR spectroscopy clearly is the limited time interval needed for the acquisition of a spectrum (in the order of some msec, see previous section). A transient spectral change with an intrinsic rise or decay time in the order of or faster than the acquisition time needed for the interferogram is thus convoluted in a complicated way over this time interval and a spectrum thus obtained will show a somehow averaged spectrum of the transient change. A clear attribution of band amplitudes to the transient product at a given time, however, is not easily possible. Nevertheless, with a known delay between the firing of the flash and the start of the interferogram, there is a clear relation between the amplitude at each interferogram data point and the delay time. “Stroboscope FTIR spectroscopy” (illustrated in Fig. 3b) makes use of this relation by recording many interferograms for an event which can be repeatedly excited and cycled rapidly. The actinic flash is fired at different delays with respect to the start of the data acqusition. If the time intervals are chosen sufficiently narrow, the interferogram amplitude for each optical delay between –d and +d can be obtained for each time delay. A complete “array of data” thus consists of several thousand interferograms recorded with different fixed delays. Data points from different optical delays corresponding to identical delay in time can now be reconstructed to interferograms each reflecting a fixed time interval after the flash. This set can be transformed to obtain spectra at fixed time intervals after the flash. “Stroboscope FTIR spectroscopy” does not need great effort on the hardware side of the FTIR spectrophotometer, but needs additional

151

software to rebuild interferogram data. The time resolution depends on the response time of the detector, the delay time spacing, the number of data points taken, and can be in the domain. Typically, for each time delay many interferograms are averaged to improve signal-to-noise ratio. In connection with the variation of time delay, recording of 10,000 to 100,000 interferograms may be necessary. In addition, the multiplex advantage is, at least partially, lost, since multiple recording of data points cannot be avoided. This technique is thus particularly suited for protein photoreactions which can be arbitrarily repeated, yielding identical response for each flash event. The need for 10,000 or much more repetitive cycles also requires rapidly cycling reactions. While the technique was successfully applied to bacteriorhodopsin (Braiman et al., 1991), application for photosynthetic reaction centers was not yet reported. An estimation of the IR signal amplitude expected for charge separation in reaction centers (on the order of of the absorbance) indicates that at least 50,000 interferograms would be needed for a full time-dependent series of spectra. Assuming that a dead time of at least one second is needed between two flashes to allow complete relaxation, this experiment will take about one day. It is clear that the same type of experiment for the study of charge separation will be extremely critical and take days to be completed, since a dead time of at least 10– 20 s between two flashes would be needed in order to allow full relaxation of the charge-separated state. Apart from the time needed for the experiment, it is questionable whether sample stability would be sufficient, i.e. whether the background absorbance could be kept constant (changing less than over this period, and whether photochemical deterioration of the sample (even if small) would not lead to artefacts because of the changing response of the protein to flash excitation. 3.

Step-Scan FTIR Spectroscopy

The major drawback of Stroboscope FTIR spectroscopy is the (partial) loss of multiplex advantage due to multiple recording of data points and the limitation of the time resolution by the

152 matched combination of optical delay/time delay. These drawbacks are avoided by the third timeresolved FTIR technique termed stop-and-go or step-scan FTIR spectroscopy (Uhmann et al., 1991), which is illustrated in Fig. 3c. With this technique, the interferometer path (–d to +d) is divided into many steps (in the order of 500 to 2000), and instead of moving continuously, the mirror is now moved stepwise. At each step, the moving mirror is stopped and the flash-induced signal is triggered. A transient absorbance change in the sample will lead to a change of intensity for this particular point of the interferogram, which is recorded in a transient recorder. Repetitive excitation may be used to improve the signal-to noise ratio for the time-dependence of the interferogram intensity. The mirror is then moved to the next step, and the full time-dependence of the interferogram is recorded by acquiring the transient changes at all steps. This “array of data” can be transformed to time-depedent spectra after rebuilding the time-dependent interferograms from the transient intensity changes. The step-scan technique, in contrast to the stroboscope technique, makes full use of the multiplex advantage, in that, except for averaging the transient changes of the interferogram intensity, no data point of the array (detector intensity as a function of time and optical delay) is recorded more than once. Thus, transient spectra can be recorded with much less repetition. However, the effort on the hardware side of the interferometer is considerably higher than with the stroboscope technique. The interferometer mirror is much less stable at a fixed position than it is during a uniform movement. In fact, if a step-scan spectrum is recorded at resolution, and the path of the interferometer is subdivided into 2000 intervals, the increment from step to step is in the order of one and the required minimum deviation from the parallel position corresponds to fluctuations in position of a few nanometer. An active stabilization of the movable mirror by mounting it onto piezo elements has been succesfully used. In this case, the piezo elements are driven by a control unit which uses interference signals from auxiliary lasers to maintain the lateral and the angular position of the mirror. Needless to say that a working step-scan interferometer is extremely sensitive to vibrations and thus needs

Werner Mäntele very stable optical benches and components. It is most successfully used in evacuated FTIR spectrophotometers, since purging of the instrument may already introduce too many vibrations. As for time-resolution, step-scan FTIR spectroscopy is only limited by the rise time of the detector and of the transient recorder, and nanosecond applications have been reported (for a review, see Palmer et al., 1993). The arguments on sample stability discussed above for stroboscope FTIR spectroscopy also hold for the step-scan technique: Although less flashes may be needed for a complete time-dependent spectrum because of the multiplex advantage, the response of the photoreactive protein to the trigger needs to be identical for each flash. It is probably for this reason that applications up to now were for bacteriorhodopsin (Weidlich and Siebert, 1993) and CO poisened myoglobin (Rödig et al., 1994), and only one application was reported for reaction centers of photosynthetic bacteria (for a review of other, nonbiological applications see: Palmer et al., 1993). Breton and coworkers (Burie et al., 1993) have used stepscan FTIR spectroscopy to investigate electron transfer in RCs at low temperature (90 K), collecting data at 457 mirror positions (averaging 20 laser flashes at each position). With 20 of these cycles averaged, around 180,000 flashes were needed. Burie et al. (1993) obtained difference spectra at after a laser flash which show the essential features of the steady-state difference spectra obtained under continuous illumination. Using RC poised at low redox potential, they were able to generate a flash-induced difference spectrum averaged over the time window between and Major bands of the difference spectrum were found to decay with a half-time of While this experiment took only about one day with 180,000 flashes fired at approx. 6 per second, it is clear that an investigation of a photoreaction which needs longer relaxation time is much more difficult to perform. V. Single Wavelength IR Techniques Single wavelength techniques can present an alternative to FTIR techniques whenever only a small spectral range is considered or when high time resolution is needed. In addition, cyclic reac-

Infrared and Fourier-transform infrared spectroscopy tions with long relaxation times (such as, for example, electron transfer from P to and further to could make a stroboscope or a step-scan FTIR experiment difficult if not impossible, but would still allow kinetic measurements in a limited range. Furthermore, kinetic measurements made “point by point” would even be possible for a less stable system which shows changes of kinetic parameters with time, while a stroboscope or step-scan experiment under these conditions would reflect these changes distributed throughout the spectrum. Single wavelength techniques either use light from thermal sources or from continuous wave or pulsed lasers. A variety of techniques have been developed, covering the mid-IR range and giving access to the time domain in milli- and microseconds as early as 1978 and picoseconds in the 90 s. In fact, the first IR difference spectrum of a photosynthetic membrane was obtained in 1984 using a modified IR spectrophotometer equipped with a fast chopper and a fast and sensitive MCT detector (Bartel et al., 1985). Time resolution was dominated by the duration of xenon flash to around

A. Dispersive Spectrometers A first version of a dispersive kinetic IR photometer was based on a conventional IR spectrophotometer (Perkin-Elmer Type 180) equipped with a fast MCT detector and a xenon flash for sample excitation (Mäntele, 1978; Siebert et al. 1980; Siebert et al., 1981; Mäntele et al., 1982). A fast chopper was used for modulation to reduce the low-frequency noise of the detector (1/f noise). With this setup, kinetic IR signals in the range to could be recorded at a time resolution from seconds to about 1 ms (chopped) and between 1 ms and some (unchopped). A similar setup using a gated detection for the MCT was described by Iwata and Hamaguchi (1989) and by Hauser (1994) with microsecond time resolution. These setups use a thermal source (“globar”: a silicon carbide rod heated to 600–800 °C) and a monochromator, which is typically mounted between the sample and the detector in order to reduce flash artefacts. Excitation of the sample by a light pulse and the subsequent photochemistry will result in some

153 fraction of the light energy deposited as heat, which will lead to broad signals (“heat signal”) overlapping the changes of intensities and of the frequency of individual modes. Placing the monochromator between sample and detector will reduce this artefact. The problem limiting the application of dispersive IR spectroscopy is the low spectral energy density. The half-width of IR bands and the overlap of IR signals necessitates a resolution of or better to resolve structures in the difference spectra. At this resolution, the radiation intensity in a spectral element at with a width of given by a typical thermal source is less than and cannot be increased easily. The low number of photons and the detector noise limit the time resolution to the ms domain or necessitate excessive averaging. On the other hand, the noise of semiconductor IR detectors is governed by intensity-independent components, and would allow to increase the intensity by orders of magnitude without increasing the noise. Only one application of dispersive IR spectroscopy in photosynthesis has been reported up to now, mainly because of the lack of sensitivity. Bartel et al. (1985) investigated the transient IR difference spectra of thylakoids and chromatophore membranes and found signals in the ms time domain which were insensitive to inhibitors of electron transfer. They ascribed them to energy dissipation processes at the pigments in the light-harvesting complexes. For bacterial membranes with functional reaction centers, slow additional components were found and assigned to the photooxidation of the primary electron donor bacteriochlorophylls in the reaction center.

B.

Tunable IR Lasers

An alternative to using thermal sources and monochromators is the use of infrared lasers, which need to be tunable in order to cover at least a small part of the difference spectrum in the to region. Ideally, highly stable continuous wave lasers should be used, but applications with pulsed lasers are also conceivable. For the spectral region between ca. and ca. CO lasers with high intensity can be used. They provide tunability

154 over at least with high-intensity modes every However, they show intensity fluctuations in the ms-to-ns time domain due to mode-hopping processes, and thus cannot be used for continuous wave applications in this time domain2. This restriction does not apply for CO lasers as a source of IR light for ultrashort pulse generation in upconversion experiments (see next section). Probably the most convenient tunable IR lasers for time-resolved IR spectroscopy are diode lasers. These diode lasers are made from lead salt semiconductors (europium doped lead selenide) and can be made for the frequency range between 2000 and Their frequency range can be chosen by varying the semiconductor composition, and, within that range, they can be tuned by varying the temperature of the semiconductor chip. This can be either reached by varying the heat sink temperature or by adjusting the laser current. Operating temperature is between a few K and 150 K at the highest, and stability of laser radiation requires temperature fluctuations to be smaller than approx. 1 mK. The tuning range can be up to for one laser, and modes are found within this range every with an output power of up to 1–2 mW. A kinetic IR photometer using these lasers has been described (Mäntele et al., 1990b) which uses several of these lasers in a cryostat, each at a different frequency range. The lasers, preadjusted in temperature, can be moved into the focus of an off-axis paraboloid which couples the beam into the photometer. A monochromator can be switched into the beam to determine the precise frequency, but also to cut out unwanted side modes. The IR probing beam is then focused onto the sample on a spot of around 1 mm 2 , which allows microsampling or lateral sample scanning for bleaching samples. The transmitted light is collected and focused on a MCT detector of selected sensitivity. The time resolution is only determined by the risetime of the detector, which can be 20ns for a photovoltaic MCT detector. With a laser output power of up to a few mW, 2

Preliminary experiments performed by the author in collaboration with H. Pascher, Department of Physics, University of Bayreuth, using an actively stabilized CO laser at cryogenic temperatures, indicate that the stability would be sufficient for its use as a source of cw measuring light for timeresolved IR experiments.

Werner Mäntele signal-to-noise is high enough to allow detection of absorbance changes in the order of to with a single flash experiment. This setup has been successfully used to monitor transient signals associated with electron transfer in photosynthetic reaction centers (Hienerwadel et al., 1992, 1995). With the time resolution available, electron transfer from to as well as all molecular relaxations concomitant with this step could be followed in real-time. Signals from quinone modes upon formation of the semiquinone anion could be identified, but also signals from the protein host site indicating microconformational changes. Most of these rearrangements of groups or polarization of dipoles occur in a concerted action and with the kinetic parameters of electron transfer. In the to range, however, signals with kinetic parameters uncoupled from electron transfer were observed, which correspond to proton uptake by Asp and Glu side chains. In particular, proton uptake by Glu L212 in the pocket could be demonstrated, and evidence was obtained for ionizable residues forming a cluster of partial charges around

C. Picoseond Pump-Probe Techniques Several laboratories have applied picosecond techniques for ultrashort IR studies on RC. Two basically different experimental approaches have been used. In the first one mainly put forward by R.M. Hochstrasser’s group in Philadelphia, a continuous wave CO laser (or a tunable diode laser) which defines the probing wavelength is mixed in a nonlinear optics experiment with an ultrashort light pulse in the visible, which is delayed with respect to the pump beam (Diller et al., 1991a,b; Hochstrasser et al., 1992; Maiti et al., 1993). The shifted frequency pulse resulting from the mixing of the IR light and the visible light pulse is detected in the visible region (“upconverted”), with the timing done by the ultrashort light pulse. This technique allows easier detection of the light pulse with high sensitivity in the visible; however, the signal-to-noise ratio depends on the stability of the continuous wave IR source. In the second approach put forward by W. Zinth’s group in Munich, an ultrashort IR pulse is

Infrared and Fourier-transform infrared spectroscopy generated from visible light pulses by subtractive mixing techniques (Hamm et al., 1995). This short pulse of IR light is then delayed with respect to the visible light pump pulse and sent to the sample as a probe pulse. The transmitted IR light is detected with a MCT IR detector. The authors elegantly used the spectral width of their IR pulse, which is approx for multiwavelength detection. The IR pulses were dispersed in a grating spectrometer and measured with a 10-element MCT IR detector with a spectral resolution of Apart from faster data acquisition, this simultaneous detection allowed a direct comparison of band intensities in this range. The authors obtained IR difference spectra from Rb. sphaeroides RCs at 1 psec and at 10 and 1000 ps. The latter spectrum, which represents the state of charge separation, closely corresponds to the FTIR difference spectrum obtained under steady-state illumination. The 10 ps difference spectrum mainly represents the charge separation, and shows essential features of FTIR difference spectra previously obtained for Rps. viridis (Nabedryk et al., 1986; Mäntele et al., 1988) and Rb. sphaeroides RC (Nabedryk et al., 1995). The 1 psec spectrum, however, corresponds to the transition, and shows a broad background of increased absorption in addition to changes of intensity and frequency of localized modes, such as the 9-keto mode of the primary donor bacteriochlorophylls. One of the expectations of picosecond spectroscopy was to obtain information on the contributions of the protein moiety to the primary steps of electron transfer, but unambiguous assignment of peptide C=O groups or amino acid side chains has not yet been possible. In addition, neither of the picosecond work provided clear evidence for or against electron transfer with an intermediate and isotope-labelled or chemically-modified bacteriochlorphyll incorporated at the monomer bacteriochlorophyll site will be needed for a clear decision. VI. Sample Preparation for Infrared Spectroscopy For the preparation of IR samples, the wavelength range to be investigated defines the primary strategy. As discussed above, of the fre-

155 quency range accessible with most FTIR spectrophotometers which spans from >4000 to the range is the most attractive one (and the most investigated), since the majority of bands with diagnostic relevance is located there. In this range, and overlapping most of the peptide C=O modes which can be used for structural studies, the strong water H–O–H bending mode is centered around Although its extinction coefficient is only ca. it necessitates thin layers for aqueous samples, since the absorbance from water at this frequency is around 1 for a pathlength of As an additional problem, the extinction coefficients for most of the protein modes (at most are one or two orders of magnitude smaller than the electronic transition moments of pigments, quinones, or hemes. RC samples for IR spectroscopy thus need to be of much higher concentration than samples used for optical transmission spectroscopies. Samples for spectroscopic investigations between and approx. form an exception, since this “IR window region” allows pathlengths up to or more. For the to range, the design of IR samples follows a compromise between the water content needed for unperturbed function, concentration, and stability. IR cell windows used for studies on photosynthetic protein complexes are almost always a compromise between transmission (from 190 nm to stability against aqueous solutions, and price. IR cells with windows allow spectroscopic investigations of the same samples from the UV to the mid-IR, or excitation in the UV/VIS/NIR range and detection in the mid-IR. As an alternative, or windows could be used: both offer slightly better access to the region below but exhibit less stability against aqueous solutions. Windows made from Ge or ZnSe crystals have only rarely been used, although they offer wide access to the low-frequency range. Their high refractive index, however, resulting in loss of light by reflection, and their cutoff at short wavelengths (Ge at ZnSe at 550 nm), limit their use for reaction-modulated IR difference techniques. Probably for funding limits, the routine use of diamond windows, which would be ideal for the visible and the IR, has not been reported.

156 For investigations of chlorophyll-protein complexes, the extinction coefficients for the pigments in the NIR (for example for the monomeric BChls of the RC) and for the RC protein (the amide I band) differ considerably and resp.). However, the high number of amide groups contributing to the amide I mode lead to comparable absorbance at 800 nm and at approx. for a RC sample. This facilitates excitation by visible or NIR light coaxially with the IR beam, or simultaneous VIS/NIR and mid-IR spectroscopic investigations such as described in a review by Mäntele (1993b) and shown in Fig. 2a. Two major types of samples have been used for FTIR and kinetic IR investigations. The first type are thin, semihydrated films dried on IR windows from solutions or suspensions of photosynthetic pigment-protein complexes in detergent, reconstituted in lipid vesicles, or from membraneous particles. For the preparation of these dried films, salt, buffer, and detergent concentrations in the starting suspensions need to be reduced considerably to avoid excessively high salt concentrations or detergent enrichment in the films upon drying. In order to maintain full activity of the pigment-protein complexes, controlled hydration of the films is essential. However, well-defined pH and ionic strength are almost impossible to maintain, and there has been a continuous concern that the properties of the protein in these films at reduced water content might not correspond to those of the protein dilute solutions. In addition, pigment-protein complexes can become oriented upon drying, leading to polarization effects even for samples perpendicular to the IR beam. In the case of films from photosynthetic RCs, careful control of the structural and functional integrity by measuring the electron transfer rates was performed, and has provided evidence that the electron transfer activity of the RC at high hydration (for example at equilibrium with 98% relative humidity) is indistinguishable from that in diluted solution. The second type of IR samples used are concentrated solutions or suspensions. Detergent-solubilized RCs can be concentrated to 1–2 mM in ultrafiltration cells. Of these concentrated protein solutions, pelleted membrane fractions, or pelleted lipid vesicles containing the protein only a

Werner Mäntele few are needed. Ionic strength can be easily controlled this way, and orientation does not occur. However, the pH may be offset toward the isoelectric point of the RC if the concentration of the buffer is too low with respect to that of the RC which has multiple buffering groups. This type of sample has been routinely used for electrochemical investigations (see section IV.C.2), where diffusibility of the RC is a prerequisite. IR microcells are typically formed from two IR windows(typically separated by an annular spacer which defines the pathlength. These cells, which are mostly of local design, are necessarily demountable in order to be filled with the concentrated solutions or suspensions and to be cleaned. A serious disadvantage of these cells is that capillary forces suck the water out of the central part (typically a few mm in diameter) lead to slow drying of the samples. We have solved this problem by cutting a circular groove in one of the windows around the sample area which, with a depth of at least 0.2 mm, breaks the capillary forces brought in by the spacer. A further improvement could be achieved by grinding a flat depression at the sample area which defines the pathlength (for a description of these cells, see Hienerwadel, 1993). These sample cuvettes are just formed by two windows (one with a depression and a circular groove around) and a flat cover window, with no spacer needed. Microcells with reproducible pathlengths between 4 and have been obtained this way which seal a sample for days. When concentrated solutions or suspensions are transferred to these microcells and the cover window is closed, excess material is pressed into the circular groove where it maintains humidity. Although the high protein concentration may seem frightening for biochemists, the quantities of protein needed for either sample type have frequently been overestimated (this may be a misjudgement from the “old days” of IR spectroscopy). Quantities of ca. of a 0.5–1 mM solution are sufficient for a film or concentrated suspension, and correspond to what is conveniently used at higher dilution for UV and VIS spectroscopy. Only slightly higher quantities (5– of a 0.5–1 mM solution) are needed to fill the spectroelectrochemical cell.

Infrared and Fourier-transform infrared spectroscopy VII. Conclusions and Outlook IR and FTIR spectroscopy for the investigation of photosynthetic pigment-protein complexes, in the past decade, has received increasing interest3 and is now routinely used by a growing number of research groups. It was accepted that IR spectroscopy can be used to probe the relaxation of the cofactor and of the protein in electron transfer processes (not only in photosynthetic RCs). Even within these ten years, the progress from easy-toobtain light-induced FTIR difference spectra to the more sophisticated IR difference spectra of single cofactors or to transient spectra showing slow or even ultrafast electron transfer steps in real-time was breathtaking (at least for me). The major lesson we have learnt from IR spectroscopy is that no large conformational changes occur with electron transfer. It taught us to look closer and search for microconformational changes in the vicinity of cofactors, which have been identified for some cofactors and are most pronounced (and best characterized) for the quinones. The IR signals characterized for and reduction illustrate why the protein is sometimes called an “optimized solvent”. The total of the changes in the vibrational spectrum of a cofactor upon its redox reaction presents a clue to internal and external reorganization energy. Ultrafast IR spectroscopy of RC is just at its beginnings, and yet the present data range from several hundred femtoseconds to nanoseconds. The possibility to follow the role of the protein in the primary steps of photosynthetic electron transfer is fascinating. The weight of these experiments is still on the technical highlights, but the search for protein signals in primary charge separation has started. The deficit of IR spectroscopy of photosynthetic pigment-protein complexes is in the part of the band assignment. It is clear that the combined effort of site-directed mutagenesis and isotope labelling can help to assign the detailed band structures in FTIR difference spectra. Whether 3 This is characterized by the way the techniques are classified on meetings: In 1985, FTIR spectroscopy of RC was listed among “. . . new and potentially interesting techniques”, while in 1995, a light-induced FTIR difference spectrum of RC was already presented as a “. . . . classical FTIR difference spectrum”.

157 site-directed isotope-labelling can become applicable in photosynthesis needs to be demonstrated. However, I doubt that we shall ever be able to fully exploit the wealth of information contained in an IR difference spectrum. Yet, focusing on “hot” spectral regions like the Asp and Glu side chain carboxyl group range between and is most promising, and I am optimistic that a full mapping of this range for the bacterial RC is within reach. Altogether, the use of IR spectroscopy on the RC over a wide wavelength range and over more than twelve orders of magnitude in the time domain will help us to understand the role of the protein matrix in charge separation, stabilization and photosynthetic energy conversion. Acknowledgements I thank my colleagues working in the field of IR and FTIR spectroscopy of proteins for valuable discussion and for providing information on unpublished work and on work in press. I appreciate the comments and suggestions from my coworkers, and thank S. Grzybek for the preparation of Fig 3. Our own work in this field has only been possible because of generous financial support from the Deutsche Forschungsgemeinschaft through several grants and a Heisenberg fellowship. References Armstrong FA, Hill HAO and Walton NJ (1986) Reactions of electron-transfer at electrodes. Quart Rev Biophys 18: 261–322. Arrondo JLR, Muga A, Castresana J and FM (1993) Quantitative studies of the structure of proteins in solution by Fourier-transform infrared spectroscopy. Prog Biophys Molec Biol 59: 23–56. Baenziger JE, Miller KW and Rothschild KJ (1993). Fourier transform infrared difference spectroscopy of the nicotinic acetylcholine receptor: Evidence for specific protein structural changes upon desensitization. Biochemistry 32: 5448– 5454. Ballschmiter K and Katz JJ (1968) Long wavelength forms of chlorophyll. Nature 220: 1231–1233. Ballschmiter K and Katz JJ (1969) An infrared study of chlorophyll-chlorophyll and chlorophyll-water interaction. J Am Chem Soc 91: 2661–2677. Bartel K, Mäntele W, Siebert F and Kreutz W (1985) Timeresolved infrared studies of light-induced processes in thylakoids and bacterial chromatophores. Evidence for the

158 function of water molecules and the polypeptides in energy dissipation. Biochim Biophys Acta 808: 300–315. Bauscher M and Mäntele W (1992). Electrochemical and infrared spectroscopic characterization of redox reactions of para-quinones. J Phys Chem 96: 11101–11108. Bauscher M, Leonhard M, Moss DA and Mäntele W (1993) Binding and interaction of the primary and the secondary electron acceptor quinones in bacterial photosynthesis. An infrared spectroelectrochemical study of Rhodobacter sphaeroides reaction centers. Biochim Biophys Acta 1183: 59–71. Beck M (1992) Spektroskopische Untersuchungen zur salzund pH-Abhängigkeit der spektralen Formen des Lichtsammelkomplexes B800–850 des photosynthetischen Bakteriums Rhodobacter sulfidophilus. Diplomarbeit, Department of Biology, University of Freiburg, Germany. Braiman MS, Ahl PL and Rothschild KJ (1987) Millisecond Fourier-transform infrared spectra of bacteriorhodopsin’s photoproduct. Proc Natl Acad Sci USA 84: 5221– 5225. Braiman MS, Bousché O and Rothschild KJ (1991) Protein dynamics in the bacteriorhodopsin photocycle: submillisecond Fourier transform infrared spectra of the L,M, and N intermediates. Proc Natl Acad Sci USA 88: 2388–2392. Breton J and Nabedryk E (1993) infrared difference spectrum of the triplet state of the primary electron donor in Rb. sphaeroides photosynthetic bacterial reaction centers. Chem Phys Lett 213: 571–575. Breton J, Thibodeau DL, Berthomieu C, Mäntele W, Verméglio A and Nabedryk E (1991a) Probing the primary quinone environment in photosynthetic bacterial reaction centers by light-induced FTIR difference spectroscopy. FEBS Lett 278: 257–260. Breton J, Berthomieu C, Thibodeau DL and Nabedryk E (1991b) Probing the secondary quinone environment in photosynthetic bacterial reaction centers by light-induced FTIR difference spectroscopy. FEBS Lett 288: 109– 113. Breton J, Burie JR, Berthomieu C, Berger G and Nabedryk E (1994a). The binding sites of quinones in photosynthetic bacterial reaction centers investigated by light-induced FTIR difference spectroscopy: Assignment of the vibrations in Rb. sphaeroides using or ubiquinones and vitamin K1. Biochemistry 33: 4953–4965. Breton J, Burie JR, Boullais C, Berger G and Nabedryk E (1994b). Binding sites of quinones in photosynthetic bacterial reaction centers investigated by light-induced FTIR difference spectroscopy: Binding of chainless symmetrical quinones to the site of Rhodobacter sphaeroides. Biochemistry 33: 12405–12415. Breton J, Boullais C, Burie JR, Nabedryk E and Mioskowski C (1994c). Binding sites of quinones in photosynthetic bacterial reaction centers investigated by light-induced FTIR difference spectroscopy: Assignment of the interactions of each carbonyl of in Rhodobacter sphaeroides using sitespecific ubiquinone. Biochemistry 33: 14378– 14386. Brudler R, de Groot HJM, Van Liempt WBS, Steggerda WF, Esmeijer R, Gast P, Hoff A J, Lugtenburg J. and Gerwert K (1994) Asymmetric binding of the 1- and 4-C=O groups

Werner Mäntele of Q(A) in Rhodobacter sphaeroides R26 reaction centres monitored by Fourier transform infra-red spectroscopy using site-specific isotopically labelled ubiquinone-10. EMBO J 13: 5523–5530. Burie JR, Leibl W, Nabedryk E and Breton J (1993) Stepscan FT-IR spectroscopy ef electron transfer in the photosynthetic bacterial reaction center. Appl Spectrosc 47: 1401–1404. Byler DM and Susi H (1986) Examination of the secondary structure of proteins by deconvolved FTIR spectra. Biopolymers 25: 469–487. Chirgadze YN and Brazhnikov EV (1974) Intensities and other spectral parameters of infrared amide bands of polypeptides in form. Biopolymers 13: 1701–1712. Chirgadze YN, Fedorov OV and Trushina NP (1975) Estimation of amino acid residue side chain absorption in the infrared spectra of protein solutions in heavy water. Biopolymers 14: 679–694. Colthup NB, Daly LH, and Wiberley SE (1990) Introduction to Infrared and Raman Spectroscopy. Academic Press, San Diego. Corrie JET and Trentham DR (1993) Caged nucleotides and neurotransmitters. In: Morrison H (ed) Biological Applications of Photochemical Switches; Bioorganic Photochemistry Series, Vol 2, pp 243–305 John Wiley and Sons, New York. Diller R, Iannone M, Bogomolni R and Hochstrasser RM (1991a) Ultrafast infrared spectroscopy of bacteriorhodopsin. Biophys J 60: 286–289. Diller R, Iannone M, Cowen BR, Maiti S, Bogomolni R and Hochstrasser RM (1991b) Picosecond dynamics of bacteriorhodopsin, probed by time-resolved infrared spectroscopy. Biochemistry 31: 5567–5572. Elliot A and Ambrose EJ (1950) Structure of synthetic polypeptides. Nature 165: 921–922. Fogel C, Petry F, Barth A and Mäntele W (1993) Titration of protonable residues in proteins by flash-induced release from “caged proton”: UV/VIS and IR studies. In Theophanides T, Anastassopoulou J and Fotopoulos N (eds.) Fifth International Conference on the Spectroscopy of Biological Molecules, pp 113–1140. Kluwer Academic Publishers, Dordrecht. Fritz F (1995) Elektrochemische und spektroskopische Untersuchungen der Multihämcytochrom-Untereinheiten von Chloroflexus aurantiacus und Rhodopseudomonas viridis. PhD Thesis, University of Freiburg, Germany. Fritz F, Moss DA and Mäntele W (1992) Electrochemical titration of the cytochrome hemes in the Rhodopseudomonas viridis reaction center. FEBS Lett. 297: 167–170. Fritz F, Griebenow K, Holzwarth A and Mäntele W (1993) Electrochemical and spectroscopic characterization of the multiheme cytochrome subunit of the reaction center of Chloroflexus aurantiacus. In Theophanides T, Anastassopoulou J and Fotopoulos N (eds) Fifth International Conference on Spectroscopy of Biological Molecules, pp 315– 316. Kluwer Academic Publishers, Dordrecht. Griffiths PR and De Haseth JA (1986). Fourier transform infrared spectrometry. Chem Anal Ser Monogr Anal Chem Appl Vol 83.

Infrared and Fourier-transform infrared spectroscopy Gronholz J and Herres W (1985) Understanding FT-IR data processing. I&C(Bruker Co) 3: 1–23. Gunner MR and Dutton PL (1989). Temperature and dependence of the electron transfer from to in reaction center protein from Rhodobacter sphaeroides with different quinones as J Am Chem Soc 111: 3400–3412. Hamacher E, Kruip J, Rögner M and Mäntele W (1993) Electrochemically-triggered infrared difference spectra of the primary donor Photosystem I from Synechocystis PCC6803. In Theophanides T, Anastassopoulou J and Fotopoulos N (eds) Fifth International Conference on the Spectroscopy of Biological Molecules, pp 163–164, Kluwer Academic Publishers, Dordrecht. Hamm P, Zurek M, Mäntele W, Meyer M, Scheer H and Zinth W (1995) Femtosecond infrared Spectroscopy of reaction centers from Rhodobacter sphaeroides between 1000 and Proc Natl Acad Sci USA 92: 1826–1830. Hauser K (1994) Zeitaufgelöste und statische Infrarotuntersuchungen an photolabilen ATP-Derivaten. Diplomarbeit, Department of Physics, University of Freiburg, Germany. Hienerwadel R (1993) Ladungstransportvorgänge in photosynthetischen Reaktionszentren. Untersuchungen mittels zeitaufgelöster und statischer Infrarot-Differenzspektroskopie. PhD Thesis, University of Freiburg, Germany. Hienerwadel R, Thibodeau DL, Lenz F, Nabedryk E, Breton J, Kreutz W and Mäntele W (1992) Time-resolved infrared Spectroscopy of electron transfer in bacterial photosynthetic reaction centers: Dynamics of binding and interaction upon and reduction. Biochemistry 31: 5799– 5808. Hienerwadel R, Grzybek S, Fogel C, Kreutz W, Okamura MY, Paddock ML, Breton J, Nabedryk E and Mäntele W (1995) Protonation of Glu L212 following formation in the photosynthetic reaction center of Rhodobacter sphaeroides: Evidence from time-resolved infrared spectroscopy. Biochemistry 34: 2832–2843. Hochstrasser RM, Diller R, Maiti S, Lian T, Locke B, Moser CC, Dutton PL, Cowen BR, and Walker GC (1992) Ultrafast infrared Spectroscopy of protein dynamics. In Martin L, Migus G, Mouru M and Zewail H (eds) Ultrafast Phenomena VIII, Springer-Verlag Berlin. Hoff AJ (1992) Optical Spectroscopy. Isr J Chem 32: 405– 412. Iwata K and Hamaguchi H (1989) Contraction of a versatile microsecond time-resolved infrared spectrometer. Appl Spectrosc 44: 1431–1437. Katz J J, Dougherty RC and Boucher LJ (1966) Infrared and nuclear magnetic resonance of chlorophyll. In: Vernon LP and Seely GR (eds) The Chlorophylls, pp. 185–251. Academic Press, New York. Katz JJ, Shipman LL, Cotton TM and Janson TR (1978) Chlorophyll aggregation: coordination interactions in chlorophyll monomers, dimers and oligomers. In Dolphin D (ed) The Porphyrins. Vol V, Part C, pp 401–458. Academic Press, New York. Leonhard M and Mäntele W (1993) Fourier transform infrared spectroscopy and electrochemistry of the primary electron donor in Rhodobacter sphaeroides and Rhodopseudomonas viridis reaction centers: Vibrational modes of the

159 pigments in situ and evidence for protein and water modes affected by formation. Biochemistry 32: 4532–4538. Lutz M and Mäntele W (1991) Vibrational spectroscopy of chlorophylls. In: Scheer H (ed) Chlorophylls, pp 855–902 CRC Press, Boca Raton, Florida. Mäntele W (1978) Aufbau einer Blitzlichtphotolyseapparatur mit infrarotem Messlicht für die kinetische Messung von Absorptionsanderungen im IR-Grundschwingungsbereich und Erprobung dieser Methode im Reaktionsmetabolismus des Rhodopsins. Diplomarbeit, University of Freiburg, Germany. Mäntele W (1993a) Reaction-induced infrared difference spectroscopy for the study of protein function and reaction mechanisms. Trends Biochem Sci 18: 197–202. Mäntele W (1993b) Infrared Vibrational spectroscopy of the photosynthetic reaction center. In: Deisenhofer J and Norris J (eds) Photosynthetic Bacterial Reaction Center, Vol. II, pp 239–283, Academic Press, New York. Mäntele W. (1995) Infrared vibrational spectroscopy of reaction centers. In: Blankenship RE, Madigan MT, and Bauer CE (eds) Anoxygenic Photosynthetic Bacteria, Chapter 28. pp 627–697, Kluwer Academic Publishers, Dordrecht. Mäntele W, Siebert F and Kreutz W (1982) Kinetic properties of rhodopsin and bacteriorhodopsin measured by kinetic infrared spectroscopy. Methods in Enzymology 88: 729– 740. Mäntele W, Wollenweber AM, Nabedryk E and Breton J (1988) Infrared spectroelectrochemistry of bacteriochlorophylls and bacteriopheophytins: Implications for the binding of the pigments in the reaction center from photosynthetic bacteria. Proc Natl Acad Sci USA 85: 8468–8472. Mäntele W, Leonhard M, Bauscher M, Nabedryk E, Breton, J and Moss DA (1990a) Infrared difference spectroscopy of electrochemically generated redox states in bacterial reaction centers. In: Michel-Beyerle ME (ed) Reaction Centers of Photosynthetic Bacteria, Structure and Dynamics, Springer Series in Biophysics, Vol 6, pp 31–44. SpringerVerlag, Berlin. Mäntele W, Hienerwadel R, Lenz F, Riedel J, Grisar R and Tacke M (1990b) Application of tunable infrared diode lasers for the study of biochemical reactions: Time-resolved spectroscopy of intermediates in the primary process of photosynthesis. Spectroscopy International 2: 29–35. Maiti S, Cowen BR, Diller R, Iannone M, Moser CC, Dutton PL and Hochstrasser RM (1993) Picosecond infrared studies on the dynamics of the photosynthetic reaction center. Proc Natl Acad Sci USA 90: 5247–5251. Moss DA, Nabedryk E, Breton J and Mäntele W (1990) Redox-linked conformational changes in proteins detected by a combination of infrared spectroscopy and electrochemistry: Evaluation of the technique with cytochrome c. Eur J Biochem 187: 565–572. Nabedryk E, Berthomieu C, Verméglio A and Breton J (1991) Photooxidation of the high-potential (c559, c556) and the low potential (c552) hemes in the cytochrome subunit of Rhodopseudomonas viridis. Characterization by FTIR spectroscopy. FEBS Lett 293: 53–58. Nabedryk E, Mäntele W, Tavitian BA and Breton J (1994) Light-induced Fourier transform infrared spectroscopic investigation of the intermediary electron acceptor reduction

160 in bacterial photosynthesis. Photochem Photobiol 43: 461– 465. Nabedryk E, Leonhard M, Mäntele W and Breton J (1994) Fourier transform infrared difference spectroscopy shows no evidence for an enolization of chlorophyll a upon cation formation either in vitro or during P700 photooxidation. Biochemistry 29: 3242–3247. Nabedryk E, Andrianambinintsoa S, Dejonghe D and Breton J (1995) FTIR spectroscopy of the photoreduction of the bacteriopheophytin electron acceptor in reaction centers of Rb. sphaeroides and Rps. viridis. Chem Phys 194: 371–378. Nichols EF (1893) A study of the transmission spectra of certain substances in the infra-red. Phys Rev 1: 1–18. Palmer RA, Chao JL, Dittmar RM, Gregorio VG and Plunkett SE (1993) Investigation of time-dependent phenomena by use of step-scan FT-IR. Appl Spectrosc 47: 1297–1310. Rödig C, Weidlich O and Siebert F (1994) Time-resolved FTIR spectroscopy of the rebinding of photodissociated carboxy-myoglobin. In: Lau A, Siebert F and Werncke W (eds) Time-Resolved Vibrational Spectroscopy VI, Springer Proceedings in Physics Vol. 74, pp 227–230. Springer Verlag, Berlin. Siebert F, Mäntele W and Kreutz W (1980) Flash-induced kinetic infrared spectroscopy applied to biochemical systems. Biophys Struct Mech 6: 139–146. Siebert F, Mäntele W and Kreutz W (1981) Biochemical applications of kinetic infrared spectroscopy. Canadian J Spectrosc 26:119–125. Sonar S, Lee CP, Coleman M, Patel N, Liu X, Marti T, Khorana GH, RajBhanday UL, and Rothschild KJ (1994) Site-directed isotope labeling and FTIR spectroscopy of bacteriorhodopsin. Nature Structural Biology 1: 512–517. Stair R and Coblentz WW (1933) Infrared absorption spectra of some plant pigments. J Res Natl Bur St 11: 703–711. Struck A and Scheer H (1990) Modified reaction centers from Rhodobacter sphaeroides R26: Exchange of monomeric bacteriochlorophyll with FEBS Lett. 261: 385–388. Struck A, Cmiel E, Katheder I and Scheer H (1990). Modified

Werner Mäntele reaction centers from Rhodobacter sphaeroides R26: 2. Bacteriochlorophylls with modified C-3 substituents at sites and FEBS Lett 268: 180–184. Surewicz WK, Mantsch HH and Chapman D (1993) Determination of protein secondary structure by Fourier transform infrared spectroscopy: A critical assessment. Biochemistry 32: 389–394. Susi H and Byler DM (1986) Resolution-enhanced Fourier transform infrared spectroscopy of enzymes. Methods in Enzymology 130: 290–311. Susi H and Byler DM (1987) Fourier transform infrared study of proteins with parallel Arch Biochem Biophys 2: 465–469. Susi H, Timasheff SN and Stevens L (1967) Infrared spectra and protein conformations in aqueous solutions. I. The amide I band in and solutions. J Biol Chem 242: 5460–5466. Thibodeau DL, Breton J, Berthomieu C, Mäntele W and Nabedryk E (1990a) Steady-state and time-resolved FTIR spectroscopy of quinone in bacterial reaction centers. In: Michel-Beyerle ME (ed.) Reaction Centers of Photosynthetic Bacteria, Structure and Dynamics. Springer Series in Biophysics, Vol 6, pp 87–98. Springer-Verlag, Berlin. Thibodeau DL, Nabedryk E, Hienerwadel R, Lenz F, Mäntele W and Breton J (1990b) Time-resolved FTIR spectroscopy of quinones in Rb. sphaeroides reaction centers. Biochim Biophys Acta 1020: 253–259. Uhmann W, Becker A, Taran C and Siebert F (1991) Timeresolved FT-IR absorption spectroscopy using a step-scan interferometer. Appl Spectrosc 45: 390–397. Venyaminov SY and Kalnin NN (1990) Quantitative IR spectrophotometry of peptide compounds in water Solutions. I. Spectral parameters of amino acid residue absorption bands. Biopolymers 30: 1243–1257. Weidlich O and Siebert F (1993) Time-resolved step-scan FTIR investigations of the transition from KL to L in the bacteriorhodopsin photocycle: Identification of chromophore twists by assigning hydrogen-out-of-plane (HOOP) bending vibrations. Appl Spectrosc 47: 1394–1400.

Chapter 10 Resonance Raman Studies in Photosynthesis – Chlorophyll and Carotenoid Molecules Bruno Robert Section de Biophysique des Protéines et des Membranes, DBCM / CEA and URA 1290/CNRS, CE. Saclay, 91191 Gif l Yvette, France

Summary I. Introduction II. Introduction to Raman and Resonance Raman Spectroscopy III. Resonance Raman Spectroscopy of Photosynthetic Pigments A. Chlorophyll and Related Molecules 1. Carbonyl Stretching Modes 2. Vinyl Stretching Modes and Other Contributions in the High Frequency Range 3. Marker Bands for Coordination State of the Central Mg Atom 4. Activity of the Different Modes According to the Excitation Wavelength B. Carotenoid Molecules C. Technical Aspects – Recent Advances IV. Resonance Raman Spectroscopy as Method of Chemical Analysis in Photosynthesis V. Resonance Raman Spectroscopy as a Probe for Molecular Conformation VI. Resonance Raman Spectroscopy as a Probe for Intermolecular Interactions A. The Binding Sites of the Primary Electron Acceptors in Different Photosystems B. The Molecular Structure of the Primary Electron Donor in Bacterial Reaction Centers C. The Amino Acids Interacting with the Long-Wavelength-Absorbing BChl Pair in Light-Harvesting Complexes from Purple Bacteria VII. Time-Resolved Resonance Raman Studies VIII. Resonance Raman Spectroscopy as a Probe for Studying the Nature of Electronic Transitions IX. Perspectives Acknowledgements References

161 162 162 163 163 164 164 165 166 166 166 167 168 169 169 169 170 171 172 173 174 174

Summary After a short introduction to the physical principles that govern resonance Raman spectroscopy, the various possibilities of this technique for studying (bacterio)chlorophyll and carotenoid molecules within photosynthetic systems are briefly discussed. Marker bands for coordination state of the central Mg atom and for the interaction state of the various carbonyl of (bacterio)chlorophyll molecules are described. A series of examples are given, illustrating how resonance Raman spectroscopy may help in: (1) assessing the molecular structure of a chlorin-type electron carrier; (2) precisely determining the configuration of a protein-bound carotenoid molecule; (3) measuring the interaction strength of the different carbonyl substituents of (bacterio)chlorophyll in a given proteic binding site; and (4) obtaining information on the electronic excited states involved in the photosynthetic process. Literature concerning the application of time-resolved resonance Raman to bacteriochlorophyll and carotenoid molecules in photosynthetic systems is also reviewed. Correspondence: Fax: 33-1-69084389; E-mail: [email protected]

161 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 161–176. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

162

Bruno Robert

Abbreviations: (B)Chl – (bacterio)chlorophyll; (B)Pheo – (bacterio)pheophytin; FT – Fourier-transform; THF – tetrahydrofuran; P – primary electron donor; PS – photosystem; RC – reaction center; RR –resonance Raman

I. Introduction For more than 20 years (Lutz, 1972), resonance Raman (RR) spectroscopy has developed in the field of photosynthesis, and its application has led to a number of significant contributions to this field. Although this spectroscopy is now applied on most of the molecular entities participating in the initial events of the photosynthetic process, in a wide variety of organisms, it is still routinely performed in few laboratories only in the world. This situation probably results from the many technical constraints and difficulties of the method, some of which have been solved recently by technological breakthroughs such as the introduction of Fourier-transform techniques to Raman spectrometers. The aim of this chapter is not to exhaustively review the contribution of resonance Raman spectroscopy to the understanding of the photosynthetic process, but rather to highlight its various possibilities through particular examples. Further, the limitations of this technique, either intrinsic, or experimental, will be discussed throughout the text, so as to illustrate when and why resonance Raman spectroscopy can (or cannot) yield useful information and the nature of this information. In order to keep this chapter as clear and concise as possible, I will focus mainly on the use of Raman spectroscopy for studying chlorophyll and carotenoid molecules. Although this choice might appear quite arbitrary, it can be justified as follows. The amount and nature of information obtained by a vibrational technique such as resonance Raman spectroscopy is strongly dependant on the nature of the molecules considered, and every example involving a study of a new molecule (such as e.g. a cytochrome or an iron–suphur protein) should be accompanied by a complete introduction of RR spectroscopy on this particular type of molecule. Moreover, and this is especially true for the two particular examples quoted above, the use of resonance Raman has developed for these molecules outside of the field

of photosynthesis, and a comprehensive introduction of their Raman spectra should mostly include references to these other fields.

II. Introduction to Raman and Resonance Raman Spectroscopy The Raman effect is the phenomenon of a change of frequency of the light, when it is scattered by polyatomic molecules. If the frequency of the incident light is and that of the scattered light is the frequency shift will correspond, in wavenumbers to the energy exchanged between the incident photon and the considered molecule during their inelastic collision. This energy, gained or lost by the photon, must be equal to that of a transition between molecular energy levels, in the following uniquely vibrational energy levels of the scattering molecules. Performing Raman spectroscopy thus essentially consists in measuring the vibrational energy levels of a given electronic state (the groundstate in classical Raman spectroscopy, but it can be an excited state if it is populated) by measuring the different Raman frequencies contained in the light scattered by a given type of molecule. From simple physical considerations (for a general introduction to the Raman effect, see Carey (1982), it can be deduced that the molecular vibrations which will be Raman-active will be those involving a change in molecular polarizability. One of the major consequence of this situation, at least for biochemical studies, is that the Raman spectrum of water is weak, and thus seldom interferes with that of the biological molecules, and this is an important advantage for the Raman technique over infra-red absorption spectroscopy. The vibrational levels of a particular molecule depends on its structure (i.e. the nature of its constituent atoms and the bonds between these atoms). Raman spectroscopy can thus be used as an analytical method for determining the chemical structure of the molecule. It also depends on the conformation of the molecule considered, as

Resonance Raman spectroscopy well as on the intra- and intermolecular interactions this molecule is involved in. These parameters may thus also be deduced from Raman spectroscopy. Moreover, although any vibrational mode formally involves the coordinates of every atom of the vibrating molecule, in the case of large molecules such as chlorophyll, some of these modes mainly involve the coordinates of particular chemical groups (such a vinyl or a carbonyl group). In this particular case, the other atoms of the molecule may be treated, to some extent, as a perturbation of this ‘localized’ vibrational mode, and the resolution of the method becomes submolecular. However, this information would be clearly difficult to extract from biological material as complex as the photosynthetic membranes if Raman spectroscopy did not exhibit a very specific, as well as convenient, property: a way of selecting the chemical nature of the scattering molecule, through the resonance Raman effect. What is called the resonance Raman effect is the enhancement of the Raman effect when the exciting frequency matches an electronic transition of the irradiated molecule. This effect, which may cause an enhancement by a factor of allows the selective observation of resonance Raman spectra of the absorbing molecule in a complex medium, provided that this molecule is the only one possessing an absorption transition. This resonance phenomenon gives to resonance Raman spectroscopy the ability to probe the interaction state and conformation of chromophores within proteins, and sometimes even when these remain embedded in a membrane, with little if any interference from signals due to the protein itself. In resonance conditions, the enhancement concerns a fraction only of the vibrational modes of the molecule. Most often, when only one electronic state is involved in the resonance phenomenon, the Raman bands arise from those modes in which variations of nuclei positions corresponds to distortions experienced by the molecule upon the electronic transition between the ground- and the excited state used for inducing the resonance. Because of this property, in resonance Raman, although the position of peaks depends solely on the electronic ground-state, their intensities yield information about the electronic excited state in-

163 volved in the resonance process, and, in particular, about the nature of the vibrational modes coupled to this electronic transition. III. Resonance Raman Spectroscopy of Photosynthetic Pigments

A.

Chlorophyll and Related Molecules

Detailed reviews on the nature of the modes active in resonance Raman spectroscopy of chlorophyll compounds have already been published (Lutz, 1984; Lutz and Robert, 1988; Lutz and Mäntele, 1991). Briefly, most of the bands observed in RR spectra of chlorophyll a and related molecules (i.e. ca. 30) appear to arise from inplane modes (Lutz, 1979). This can easily be understood as most of the electronic transitions of Chl a and related molecules are composed of transitions polarized parallel to the molecular plane. This still holds for vibrational modes involving coordinates from peripheral substituents, and thus the conformation of these groups may be evaluated on the basis of the RR activity of their stretching modes (Lutz, 1984). Most of these modes are complex, involving mixing of several internal coordinates. However, there is a small number of more localized modes, which have been assigned on the basis of isotopic substitutions, and comparisons between IR, Shpol’skii and RR spectra of various chlorophyll and porphyrin derivatives (Lutz, 1979, 1984). Indeed, only a few normal mode calculations have been performed for chlorophyll molecules, as these are particularly complex, due to the low molecular symmetry of these compounds (Boldt et al., 1987; Donohoe et al., 1988). For this reason, most of the conclusions drawn about biochemically-relevant samples are supported by comparisons with chemical models, from which the sensitivity of the different Raman bands is experimentally determined. It must be noted that resonance Raman spectroscopy will yield information on only those modes which are coupled with the electronic transitions of the (bacterio)chlorin molecules. It will thus yield information only on those chemical groups which may play a role in tuning the physicochemical properties involved in the energy/electron transfer between these molecules. In this

164 respect, the additional mode selection which occurs during the resonance process in Raman spectroscopy does not constitute a limitation, but rather an advantage for photosynthetic studies. 1. Carbonyl Stretching Modes In the high frequency range of(bacterio)chlorophyll RR spectra bands arise from near to pure stretching modes of the carbonyl groups of these molecules (i.e. 9-keto for Chl a and BChl c, 9-keto and 2-acetyl for BChl a and b, 9-keto and 3-formyl for Chl b and BChl e) (Lutz, 1972, 1974). The stretching frequency of the 9-keto carbonyl modes is observed at ca in the absence of intermolecular interactions, in an apolar environment. In (bacterio)pheophytins, their frequency is slightly higher and for Pheo a and BPheo a, respectively) (Mattioli et al., 1993). The frequencies of these modes is rather insensitive to the presence of another carbonyl group attached to the macrocycle as no more than a shift is observed between the frequency of the 9-keto stretching modes of Chl a and 2-acetyl Chl a (Feiler et al., 1994a). However, the 9-keto stretching mode of BChl c is observed at i.e. ca lower than in Chl a, although these molecules mainly differ by the presence of the residue at position 2 (a hydroxy and a vinyl group in BChl c and Chl a, respectively) (Feiler et al., 1994b). The stretching modes of the 2-acetyl and 3-formyl carbonyl groups are observed at in the absence of intermolecular interactions (Lutz, 1984). Schematically, the frequency of these modes is primarily sensitive to the intermolecular interactions in which they are involved, and, to a lesser extent to the permittivity of their environment. Upon entering intermolecular interactions, such as Hbonding, downshifts as large as are frequently observed, while the range of frequency variation due to solvent effect is ca (Lutz, 1984; Koyama et al., 1986). The free enthalpy of the H-bonds established between the (B)Chl carbonyl groups and their partner molecule may be evaluated, using a Badger-type calculation (Zadorozhnyi and Ishchenko, 1965). Considering the position of these substituents on the macrocycle rings, the stretching modes of

Bruno Robert the carbonyl in position 2 and 9 are expected to be more active in resonance with Y-polarized transitions. Consistently, excitation of BChl a in the transition did not yield observable band in the carbonyl frequency range (Lutz, 1984). By contrast, a formyl carbonyl at position 3 is expected to be more intense upon X-polarized excitations. These properties have recently been used to attribute the bands present in BChl e containing chlorosomes to either the 9-keto or the 3formyl stretching modes (Feiler et al., 1994c). In preresonance conditions with the transition of (bacterio)chlorin molecules, very weak contributions from the ester carbonyl can be observed at ca (Mattioli et al., 1991). Until now, these very weak bands have been seldom used for determining the interaction state of these groups. As these groups are expected to be poorly conjugated with the (B)Chl macrocycle, bands arising from their stretching modes should experience relatively smaller shifts upon H-bonding, and, moreover, these H-bonds should have limited influence on the physicochemical properties of the chlorophyll molecules. 2. Vinyl Stretching Modes and Other Contributions in the High Frequency Range As carbonyl stretching modes have been extensively used for assessing the interaction pattern of (bacterio)chlorophyll molecules in vitro as well as in vivo, it is of importance to identify the other possible contributions in the same spectral range. Early Raman studies of Chl a did not reveal any intense band in the region, which might primarily arise from the stretching mode of the vinyl C = C bond at position 2 (Lutz and Kleo, 1974). Similar conclusions had been reached by comparing RR spectra of nickel-methylpheophorbide and of nickel-mesopyropheophorbide (Boldt et al., 1987). By comparing the RR spectra of Chl a and 2-acetyl Chl a, or those of BChl a and 2-vinyl BChl a, Feiler et al. (1994b) observed the contribution of this mode at in vinyl-containing derivatives (Fig. 1). In Soret resonance, this contribution is very weak and partially hidden by those arising from the methine bridges stretching modes. From normal mode calculations, as well as from experiments performed on reaction center-

Resonance Raman spectroscopy

165

samples unambiguously indicates the presence of intermolecularly interacting 2-acetyl carbonyl groups.

3. Marker Bands for Coordination State of the Central Mg Atom

bound BChl a, it was proposed that CaCm and/or CbCb modes could contribute in the 1620– range (Boldt et al., 1987; Boldt et al., 1993). Experiments conducted on isolated BChl a in various conditions of excitation did not reveal the presence of any such modes (Lutz, 1984; Lutz and Robert, 1988; Mattioli et al., 1993). The same conclusion was reached for Chl a, after the replacement of the vinyl group at position 2 by an acetyl group (Fig. 1) (Feiler et al., 1993b). These predicted CaCm or CbCb modes must necessarily give rise to extremely weak bands, or should be slightly less up-shifted than predicted by the calculations (Feiler et al., 1993b). It can thus be safely concluded that the presence of a band in BChl a-containing

Bands arising from two types of modes have been used for diagnosing the coordination state of the central Mg atom of chlorophyll derivatives. First, the bands arising from modes involving this atom, in the frequency range (Lutz et al., 1975), constitute a reliable marker of the Mg atom coordination for Chl a, b and BChl a molecules (Lutz, 1977, 1984; Fujiwara et al., 1988). It must be noted that these bands do not involve the coordinates of the coordinating atom, i.e. that they cannot be used for. diagnosing the chemical nature of the axial ligand. They have been used in vivo for determining the interaction state of protein-bound Chl molecules (Lutz, 1977). They are particularly weak in BChl a RR spectra, and their use may prove difficult in biological studies. The band arising from the methine bridges stretching modes, at ca has also been used for diagnosing the coordination state of the central Mg atom, as well as, in Chl molecules, bands in the frequency range, arising from complex modes of the conjugated ring. These bands exhibit a definite, although indirect, sensitivity to the coordination state of the central Mg atom They constitute convenient markers because of their very high activity in resonance with the Soret electronic transition. The band arising from the stretching mode of the methine bridges is located at when the central Mg atom binds two external ligands, and at frequencies higher than when it interacts with one axial ligand only (Cotton and van Duyne, 1981). The frequency of this band also slightly depends on the temperature and on the excitation conditions (Mattioli et al., 1992). For Chl a molecules, a doublet of bands is observed at when the central Mg is fivecoordinated, and at when it is six-coordinated (Fujiwara and Tasumi, 1986; Tasumi and Fujiwara, 1987).

166

4. Activity of the Different Modes According to the Excitation Wavelength The activity of resonance Raman bands depends on the coupling between the vibrational modes and the electronic transition used for inducing the resonance effect. High activities of the methine bridge stretching modes may be achieved by exciting BChl a under Soret resonance conditions (Lutz, 1979, 1984). In these conditions, the band arising from these modes when the central Mg atom is five-coordinated) dominates the RR spectra, and it may partially hide the 2acetyl carbonyl stretching modes, which are usually weak at Soret resonance. Some carbonyl stretching modes have been reported missing in BChl spectra recorded in these conditions of excitation. This was the case in particular for the RR spectra of the primary donor in Rhodobacter sphaeroides reaction centers (Robert and Lutz, 1986; Mattioli et al., 1991), and for those of the 800 nm-absorbing BChl in peripheral light-harvesting complexes from Rhodopseudomonas palustris (Robert and Lutz, 1985; Sturgis et al., 1994). Excitations in preresonance with the transition of these molecules is definitively more favorable for the observation of bands arising from the carbonyl stretching modes. By contrast, inferring the number of axial ligands on the central Mg of these molecules is often easier in resonance conditions with the Soret electronic transition.

B. Carotenoid Molecules Carotenoid molecules are extremely efficient Raman scatterers and they have been the subject of RR studies as early as 1932 (Euler and Hellström, 1932). Their RR spectra mainly consist of four groups of bands at ca. 1530 1120– 1200 1000 and Systematic studies of different isomers of (Saito et al., 1983 Koyama et al., 1983), as well as normal mode calculations performed on these molecules in different configurations, have allowed the evaluation of the sensitivity of these different bands to the molecular conformation, as well as the precise attribution of each of these (Saito and Tasumi, 1983). The presence of particular bands could be safely connected to specific molecular

Bruno Robert structure, e.g. the presence of a band is diagnostic for a 15–15' cis conformation. Most of the different conformations of carotenoids may be distinguished on the basis of their Raman spectra (Koyama et al., 1988). As the signal arising from these molecules is particularly intense, they have been the subject of most of the time-resolved resonance Raman spectroscopy studies in photosynthesis (see below).

C. Technical Aspects – Recent Advances RR scattering is a low-probability, low yield process. It thus requires quite high irradiance on the sample, generally at wavelengths matching with the absorption transitions of the molecules being studied. Photodegradation of the pigments may be avoided either by cooling the samples to cryogenic temperatures or by using spinning cells or flowing samples. This crucial problem may be much more easily overcome by inducing the resonance phenomenon with the lowest energy transitions of the pigments, i.e. with red or infrared excitations. Up to five years ago, this was extremely difficult if not impossible, as under these conditions of excitation the intrinsic fluorescence of the pigments results in signal backgrounds which can be many orders of magnitude more intense than the Raman signal itself. New detectors, namely charge-coupled devices, have recently appeared, which allow the recording of RR signals of chlorophyll molecules excited in the lowest singlet transition. These detectors have practically no dark noise, a very high quantum efficiency and extended dynamics, which is absolutely necessary for recording the Raman signal over the intense fluorescence background. However, although their response in the 600–900 nm range is excellent, their sensitivity dies rapidly beyond 950 nm. Nevertheless, recording Raman spectra in these conditions must be performed with the most extreme care, in order to avoid artifactual signals. A second technological breakthrough has occurred with the development of Fourier-transform Raman spectrometers. In these spectrometers, the signal scattered by the sample is sent, after the filtering of the Rayleigh line to the interferometer of classical FT-IR spectrometers. At the moment, this technique allows only a single excitation wavelength, 1064 nm.

Resonance Raman spectroscopy

167

However, the recording of all spectral elements at one time counterbalances the fact that the Raman signal is much weaker in these conditions (the scattering effect depending on Although this technique is much more demanding in terms of sample concentration is often necessary) it allows the recording of RR spectra at room temperature and it possesses many different advantages which will be described amongst the examples chosen. IV. Resonance Raman Spectroscopy as Method of Chemical Analysis in Photosynthesis Resonance Raman spectroscopy may provide detailed information on the chemical nature of the molecules studied. This property, combined with the possibility of selective excitation of these molecules in a complex medium was recently used for determining the chemical nature of the primary electron acceptor in bacterial reaction centers of the green sulfur bacterium Chlorobium limicola. Several different pigments had been proposed for this particular cofactor, including BPheo c (van Bochove, 1984), BChl c-like pigment (Braumaun et al., 1986; Shuvalov et al., 1986) and Chl aisomer (van de Meent et al., 1992). Selective resonance Raman information on chlorin molecules in these photosystems was achieved by using a 441.6 nm excitation. From the absence of fingerprint contributions of BPheo at Feiler et al. (1994) concluded that the preparation was devoid of BPheo c. In BChl c the vinyl group in position of ring I of Chl a is replaced by a hydroxy-ethyl group. Accordingly, the band at arising from the vinyl stretching modes is absent from its RR spectra (Feiler et al., 1994a). Raman experiments performed on pigment extracts from Chlorobium reaction centers revealed the presence of this band, and it was concluded that the chlorin molecule present in Chlorobium RC contains a vinyl group i.e. that this pigment is a Chl a-type molecule (Feiler et al., 1994a). Resonance Raman spectroscopy may also be used for assessing more subtle structural features, such as the charge distribution over an assembly of (B)Chl molecules. Mattioli et al. (1991) showed that FT-Raman spectra excited at 1064

nm of reaction centers from purple bacteria contain primarily preresonant contributions from their primary electron donor, which is a dimer of BChl (P) (Fig. 2). When P is oxidised, exciting at the same wavelength results in resonance conditions with the weak 1250 nm electronic transitions of the state. FT-Raman spectra of purple bacterial reaction centers containing P in either its neutral or oxidized form are thus dramatically different from each other (Fig. 2). From the Raman study of the state, the charge distribution over each bacteriochlorophyll in the cation dimer could be addressed. In the FT-Raman spectra of a band is observed at which arises from the 9-keto carbonyl stretching mode of one BChl molecule, up-shifted by the presence of the positive charge. In the FT-Raman of P in its neutral state, this band contributes at (Fig. 2) (Mattioli et al,, 1991; Mattioli et al., 1994). Upon formation, the stretching mode of this 9-keto carbonyl stretching mode thus

168

Bruno Robert

experiences a upshift, as compared to the induced by one-electron oxidation of BChl in aprotic solvents (Mäntele et al., 1988). The charge on P is thus mainly borne by one of the BChl molecules. This BChl was identified as being the one located on the L side of the RC (Mattioli et al., 1991). The different factors influencing this charge distribution were studied by FT-Raman, and it was shown that it is influenced by the presence of H-bonds between the BChl constituting the primary donor and the protein (Wachtveitl et al., 1993; Mattioli et al., 1994) and not by site-selected mutations at the M210 locus (Jones et al., 1994). V. Resonance Raman Spectroscopy as a Probe for Molecular Conformation Although it would be certainly of interest, until now there have been no resonance Raman bands which could be precisely linked to particular conformations of the conjugated macrocycle. By contrast, RR Spectroscopy has brought a number of conclusions concerning the conformation and configuration of carotenoid molecules embedded in photosynthetic proteins. In reaction centers of purple bacteria, it was reported in 1976 that the spheroidene bound to the protein had an unusual RR signal (Lutz et al., 1976). The presence of a band at was considered as evidence of a 15–15' cis configuration for this molecule (Koyama et al., 1982). Nuclear magnetic resonance experiments performed on carotenoid extracted from the reaction centers of various bacteria confirmed this configuration (Lutz et al., 1987). However, it was shown by RR spectroscopy that, during the extraction procedure, the conformation of the molecule was modified (Lutz et al., 1987) and it was thus concluded that an additional twist of the polyene chain exists in the structure of the RC-bound carotenoid (Fig. 3). Relying on calculations performed by Saito and Tasumi (1983), Lutz et al. proposed that the twisted parts of the carotenoid bound to the reaction centers should be the and/or regions. These studies have been extended to the neurosporene conformation in reaction centers from Rb. sphaeroides, strain G1C, and

similar results were obtained (Koyama et al., 1988). In antenna complexes, it has been shown by RR spectroscopy that the bound carotenoid molecules are generally in all trans configuration (Robert, 1983; Robert and Lutz, 1985), although these molecules might be slightly distorted (Iwata et al., 1985).

Resonance Raman spectroscopy VI. Resonance Raman Spectroscopy as a Probe for Intermolecular Interactions

It is possible to determine, by RR spectroscopy, the strength of the interactions assumed by the conjugated carbonyl groups of the (bacterio)chlorin pigments, as well as the number of external ligands on their central Mg atom, when these molecules are embedded in a protein (Lutz, 1984; Lutz and Robert, 1988). This unique property of Raman spectroscopy has been extensively used in many different types of photosynthetic proteins. Three examples of such studies will be detailed in this section, so as to illustrate the diversity of results which can be obtained in this way.

A. The Binding Sites of the Primary Electron Acceptors in Different Photosystems In reaction centers from purple photosynthetic bacteria, the primary electron acceptor is the BPheo molecule mainly surrounded by the L subunit of the protein (Kirmaier and Holten, 1987). The electronic transition of this pigment, at low temperature, is located at 545 nm, whilst that of the inactive (or accessory) BPheo is located near 535 nm. Because of this small difference, selective Raman excitation of each of these pigments could be performed. In Rb. sphaeroides reaction centers, the stretching frequencies of the keto carbonyl groups of the acceptor and accessory BPheo are 1678 and respectively (Lutz, 1980, Robert, 1990). The 2-acetyl of the accessory BPheo is thus free, whilst that of the acceptor BPheo is involved in a ca. 4 kCal/M intermolecular interaction, likely with the Glu L 104 residue (Michel et al., 1986). These interactions are interspecifically conserved in purple bacteria (Zhou et al., 1987, 1989; Zhou, 1989). Michel and Deisenhofer (1988) proposed the structures of the purple bacterial reaction centers and PS II to be related, and according to this model, this particular Glu L 104 residue is aligned with the Glu D1 131 in the D1 subunit of PS II. In the latter photosystem, the primary acceptor is a pheophytin, and the Soret electronic transition is expected to be slightly blue-shifted relative to that of the various Chl molecules present in this type of preparation. Exciting D1D2 particles

169

(i.e. isolated reaction centers of PS II) at 406 or 413 nm indeed yielded Pheo contributions in the carbonyl stretching region of the RR spectra, located at 1680 and near (Moënne-Loccoz et al., 1989). Selective reduction of the acceptor Pheo molecule allowed the assignment of the band to the keto carbonyl group of this molecule. The strength of the interaction in which this group is involved is thus nearly identical to that observed in bacterial reaction centers (Moënne-Loccoz et al., 1989). It was thus concluded that the acceptor Pheo molecule in PS II particles interacts with the D1 131 Glu amino acid, and the geometry of this interaction is conserved between reaction centers from purple bacteria and PS II. In Photosystem I, the study of the specific interactions assumed by the primary electron acceptor may prove difficult, as the latter is a Chl a, and thus is not chemically different from the bulk pigments. From an extremely weak sequence homology, Robert and Möenne-Loccoz (1989, 1990) predicted that these interactions should be similar to those observed in PS II and bacterial reaction centers. Recently, Feiler et al. (1994a) studied the interactions assumed by the Chl a molecule which is the primary electron acceptor in the PSI-related reaction centers of C. limicola. The stretching mode of its keto carbonyl group is located at Relative to the frequency of this group when free from interactions corresponds to downshift of ca. . It was thus concluded that this group is involved in intermolecular interactions, the strength of which is very close to that measured in PS II and reaction centers from purple bacteria. These authors thus proposed that the binding site of the primary acceptors in all photo systems share common features, possibly inherited from a common ancestor (Feiler et al., 1994a).

B. The Molecular Structure of the Primary Electron Donor in Bacterial Reaction Centers In reaction centers from Rps. viridis, the primary electron donor consists of two BChl molecules in van der Waals contact at the level of the ring I, which bears the 2-acetyl carbonyl substituent

170

Bruno Robert

(Deisenhofer et al., 1984). It was initially proposed that the Mg atoms of these two molecules would each interact with the imidazole sidechain of neighboring histidine residues, and also with the 2-acetyl carbonyl group of the neighboring BChl molecule (Zinth et al., 1985). Selective observation of the contributions of the primary electron donor in protein could be obtained, under resonance conditions at 363.8 nm, by difference techniques. In these spectra, the methine bridge stretching band was located at It was thus concluded that both Mg atoms of the BChl constituting P were binding a single axial ligand. The stretching modes of the 2-acetyl carbonyl groups indicated that one of these group was intermolecularly bound, whilst the other one was free from interactions. A picture of the primary electron donor was thus drawn, different from the interpretation of the crystallographers (Robert and Lutz, 1986), in which the central Mg of each of the two BChl molecules was interacting with the neighboring histidine residues only. Upon the refinement of the structures derived from X-ray crystallography, it appeared that the structure of P was that deduced from the RR spectroscopy studies (Deisenhofer and Michel 1989). It is worth noting that, whereas most often RR spectroscopy was used for discarding structural models involving direct (B)Chl-(B)Chl interactions, it was also extensively used for demonstrating this type of interaction between BChl c molecules in chlorosomes from green photosynthetic bacteria (Lutz and van Brakel, 1988; Hildebrandt et al., 1991; Feiler et al., 1994d).

C. The Amino Acids Interacting with the Long-Wavelength-Absorbing BChl Pair in Light-Harvesting Complexes from Purple Bacteria Resonance Raman spectroscopy, when used in conjugation with site-selected mutagenesis, is a promising strategy for determining which amino acids are interacting with (bacterio)chlorophyll(s) in a proteic binding site, especially in the absence of structural results derived from crystallography. In RR spectra from the core antenna of purple bacteria, two bands are present at 1641 and in the carbonyl stretching frequency

range. The former was attributed to both the 2acetyl carbonyl stretching modes of the two BChl molecules present in these complexes, and the latter to both their two 9-keto carbonyl stretching modes (Robert and Lutz, 1985). In site-selected mutants from Rb. sphaeroides in which the tryptophane 43 of the subunit was replaced by a phenylalanine, the intensity of the band decreased by a factor of two, whilst that of the increased (Fig. 4) (Olson et al., 1994). This experiment shows that the Trp governs the H-bonding state of one of the 2-acetyl carbonyl groups in these complexes. When this residue was further mutated to a tyrosine, the

171

Resonance Raman spectroscopy

thus the function of these light-harvesting complexes (Fowler et al., 1994). VII. Time Resolved Resonance Raman Studies

band at in wild-type complexes splits into two components at 1633 and (Fig. 4). As is a frequency that has been observed for 2-acetyl carbonyl groups of BChl molecules interacting with the phenol sidechain of a tyrosine (Wachtveitl et al., 1993), this result led to the conclusion that the Trp is located in the binding site of one of the BChl interacting directly with its 2-acetyl carbonyl (Fig. 5). Similar experiments, performed on peripheral light-harvesting complexes from Rb. sphaeroides, showed that the tyrosines and were governing the H-bonding state of each of the two 2-acetyl carbonyl groups of the 850 nm-absorbing pair of BChl molecule. It was further suggested that these H-bonds finely tune the absorption, and

As the lifetime of the Raman process is extremely short s or shorter), Raman experiments may be performed with time resolution as short as the picosecond time scale. In photosynthesis, however, although much is anticipated from such experiments, only a few time-resolved resonance Raman experiments have been reported. This is due to two different types of technical difficulty, (i) (B)Chl Raman signals are weak, and care must be taken for limiting the instantaneous irradiance on the (B)Chl-containing sample so as not to destroy them, and (ii) the cation and/or anion and/or triplet states of (B)Chl pigments which are formed during or after the first steps of the electron transfer possess even smaller resonance Raman cross-sections than the neutral pigments. Only FT-Raman techniques, with excitation at 1064 nm, have yielded selective contributions of the cation state of the primary electron donor in RCs of purple bacteria (Mattioli et al., 1991), and of the green bacterium C. limicola (Feiler et al., 1994d). Most time-resolved studies on (B)Chl pigments have concerned the reaction centers of purple bacteria. To record resonance Raman spectra of transient states, pulsed lasers are often required. However, if one transient state has a lifetime longer than all others, then, when the number of photons per reaction center per second during the experiment is higher than the inverse of the lifetime of this transient state, it is possible to populate it sizeably, while the other transient states will be just populated according to the ratio between their lifetime and the lifetime of the longer excited state. By adjusting the intensity of the laser beam used for analysing the Raman signal, spectra of Rb. sphaeroides R 26 RCs containing small and large amounts of the transient state were recorded using Soret excitation condition (Robert and Lutz, 1988). The transfer of an electron from P to results in a local conformational change in the binding site of the accessory BChl located between P and the access-

172

ory BPheo and the 9-keto carbonyl group of this pigment becomes involved in a stronger H-bond with its environment than when the reaction centers are in their resting state (Robert and Lutz, 1988). It was, moreover, shown by chemically oxidizing P, that this change in the H-bond pattern of this group is directly correlated to the formation of Recent refinements of the Rps. viridis structure indicate that a water molecule is present near this keto carbonyl group, being Hbonded on one side to the imidazole ring of the histidine involved in liganding the central Mg ion of one of the two BChls constituting P (Deisenhofer and Michel, 1989) and on the other at a reasonable distance for interacting with the keto carbonyl group of the molecule. A small motion of this water molecule during P oxidation could modify its interactions with this carbonyl group. Recent molecular dynamics simulations indicate that, indeed, the distance between this water molecule and the 9-keto carbonyl of the accessory BChl nearby is sensitive to the positive charge borne by P (M. Marchi, unpublished data). Three groups have recorded RR spectra of reaction centers from purple bacteria in their transient state, by making use of nanosecondor picosecond-pulsed excitations (Mattioli et al., 1989; Atkinson et al. 1990; Koyama et al., 1992). RR spectra (excited at or near the maximum of the Soret transition of the neutral bacteriochlorin molecules) obtained by these three groups are quite similar to each other. The main change observed in these spectra is a net decrease of the neutral BPheo contributions. In addition to the disappearance of neutral BPheo contributions, the above-mentioned shift of a carbonyl vibrator from a BChl molecule could be present in these spectra, but this latter fact could not up to now be more clearly confirmed (Mattioli et al., 1989). Picosecond time-resolved experiments have been performed on both chromatophores from purple bacteria and chloroplasts from higher plants, under conditions of excitation favoring the contributions of the excited states of carotenoid molecules (Hayashi et al., 1991; Hashimoto and Koyama, 1990; Kuki et al., 1990). There seems to be a general consensus that the resonance properties of these molecules in their excited states reveal the existence of their long-postulated

Bruno Robert state, which has been proposed to be involved in the carotenoid to chlorophyll energy transfer. However, the effect of the proteic environment on the structure of this excited states is far from being clear. For example, the stretching mode in the state of spheroidene has been reported to shift down by in Rb. sphaeroides chromatophores, relative to isolated spheroidene (Kuki et al., 1990), while the same mode of the same state of spirilloxanthin has been observed, in chromatophores of Chromatium vinosum and in organic solvents, completely unshifted (Hayashi et al., 1991). Resonance Raman contributions arising from the triplet state of carotenoid molecules have also been observed in vivo, in chromatophores of Chromatium vinosum, using an apparatus with picosecond time resolution (Hayashi et al. 1991), and in the RC from Rb. sphaeroides, at microsecond resolution (Lutz et al., 1983). In the latter case, it was thus concluded that no cis–trans isomerisation occurs during the ground to triplet state transition. Furthermore, the presence of a very intense band in these spectra indicates that the molecule also retains the twisted conformation it possesses in its ground-state. Experiments conducted with a continuous pumpprobe excitation at 530 nm, have revealed the presence of a very active low frequency mode near in these spectra, which has not yet been assigned (Robert et al., 1985). VIII. Resonance Raman Spectroscopy as a Probe for Studying the Nature of Electronic Transitions While the position of the different resonance Raman bands yields information concerning the ground electronic state of the molecule studied, their intensities may provide information on the nature of the excited state used to induce the resonance phenomenon. During the photosynthetic process, some very important steps of the energy transduction (namely light-harvesting and primary charge separation) involve excited states of (bacterio)chlorophylls. A precise description of the structure of these excited states, which often concern (B)Chl dimers, is needed for our understanding of the physicochemistry of these reactions. Until now only very few groups have

Resonance Raman spectroscopy attempted to get this type of information from RR spectroscopy, mainly because the excited states involved in the energy conversion are the lowest (B)Chl singlet states, the strong fluorescence of which has for long precluded any RR measurement. Two groups have reported RR spectra of the primary electron donor in RC from purple bacteria, using near infra-red excitations provided by a Titanium-Sapphire continuous laser (Shreve et al., 1991; Palaniappan et al., 1992). Shreve et al. (1991) extracted the Raman contributions from the intense fluorescence background using shifted excitation difference. According to this technique, the Raman signal and the fluorescence background are recorded at two excitation wavelengths differing by a few wavenumbers. In the computed difference between these two excitation only Raman contributions are expected to shift according to the excitation used. Palaniappan et al. (1992) designed an experiment at low temperature in which RCs are mixed with ethylene glycol to decrease the level of the light scattered by the sample. Fluorescence backgrounds were subtracted after having been fitted by a polynomial curve. Results obtained by these teams are not fully compatible. Spectra obtained by Shreve et al. (1991) are dominated by a small number of relatively intense low-frequency modes, located between 38 and and three bands in the medium frequency range at 685, 730 and These features depend on the precise position of the excitation, and low-frequency modes typical from the primary electron donor were assigned. Spectra reported by Palaniappan et al. (1992) are composed of a very large number of bands of similar intensities from ca. to and few bands only appear at the same frequencies as those reported by Shreve et al. (1991). They do however exhibit a common feature, in the sense that they contain many intense bands in the low frequency region (30– which appear to be typical of the primary electron donor contributions. However, considering how much these two sets of results differ from each other, one may question whether the few similarities they exhibit have any significance. It would of course be of enormous interest, once the origin of the discrepancies between these

173

results have been understood, to know whether or not the modes observed by these teams, which must be coupled to the P to transition, are indeed coupled with the transition. Low frequency modes have been up to now poorly studied for BChl molecules. Careful comparison with spectra obtained from monomer and dimer model systems would be of great help in interpreting these spectra. It has for example been reported that, at some excitation conditions, the low frequency modes may be quite intense even for monomeric BChl (Lutz, 1979), and some of the frequencies observed are very near to those reported in the work of Shreve et al. (1990). In the absence of such work on model system, any conclusion has to be drawn with the most extreme care. However, it is clear that this type of experiments have opened a whole new field of application for RR spectroscopy in photosynthesis, and that they might give a unique opportunity to understand, in the long run, the physicochemistry of the charge separation mechanism. IX. Perspectives From this short overview of the different possibilities of resonance Raman spectroscopy, it is clear from the examples chosen that most of the work performed has concerned bacterial photosynthesis, although it is relatively more difficult to record RR spectra of BChl than those of Chl molecules. In this particular field of biology, RR spectroscopy is now a mature technique. Photosynthetic proteins from higher plants are more complex, and it is consequently more difficult to characterize the contributions of a selected pigment in their RR spectra. However, it is most likely that the forthcoming years should see an increasing number of contributions from resonance Raman in this particular field of photosynthesis. It is also clear that the full potential of resonance Raman has still not been explored, in particular with regard to the information which can be obtained by time-resolved resonance Raman and/or on the excited electronic states of (B)Chl molecules. On those two latter subjects, RR spectroscopy should develop from the preliminary, but particularly exciting data which have already been reported.

174

Acknowledgements The author thanks Delphine Albouy, James N. Sturgis, Tony A. Mattioli and Diane Spiedel for their skillful help and their many suggestions during the writing of this manuscript. References Atkinson GH, Hayashi H, Tasumi M and Kolaczkowski S (1990) Picosecond resonance Raman spectroscopy of Rhodobacter sphaeroides reaction centers. In: Michel-Beyerle ME (ed) Reaction Centers of Photosynthetic Bacteria, pp 140–146. Springer-Verlag, Berlin Boldt NJ, Donohoe RI, Birge RR and Bocian DF (1987) Chlorophyll model compounds: Effects of low symmetry on the resonance Raman spectra and normal mode descriptions of Nickel(II) porphyrins. J Am Chem Soc 109: 2284– 2298 Braumann T, Vasmel H, Grimme LH and Amesz J (1986) Pigment composition of the photosynthetic membrane and reaction centers of the green bacterium Prosthecochloris aestuarii. Biochim Biophys Acta 848: 83–91 Carey PR (1982) Biochemical Applications of Raman and Resonance Raman Spectroscopies. Academic Press, New York. Cotton TM and van Duyne RP (1981) Characterization of bacteriochlorophyll interactions in vitro by resonance Raman spectroscopy. J Am Chem Soc 103: 6020–6026 Deisenhofer J and Michel H (1989) The photosynthetic reaction center from the purple bacterium Rhodopseudomonas viridis. EMBO J. 8: 2149–2169 Deisenhofer J, Epp O, Miki K, Huber R and Michel H (1984) X-ray structure analysis of a membrane protein complex: electron density map at 3 Å resolution and a model of the chromophores of the photosynthetic reaction center from Rhodopseudomonas viridis. J Mol Biol 180: 385–398 Donohoe RI, Frank HA and Bocian DF (1988) Resonance Raman spectra and normal mode descriptions of a bacteriochlorophyll a model complex. Photochem Photobiol. 48 531–537 Feiler U, Mattioli TA, Katheder I, Scheer H, Lutz M and Robert B (1994a) Effects of vinyl substitutions on resonance Raman spectra of (bacterio)chlorophylls. J. Raman Spectrosc 25: 365–370 Feiler U, Albouy D, Pourcet C, Mattioli TA, Lutz M and Robert B (1994b) Structure and binding site of the primary electron acceptor in the reaction center of Chlorobium. Biochemistry 33: 7594–7599 Feiler U, Albouy D, Lutz M and Robert B (1994c) Pigment interactions in chlorosomes of various green bacteria: a resonance Raman study. Photosynth Res 41:175–180 Feiler U, Albouy D, Mattioli TA and Robert B (1994d) Resonance Raman studies of the primary electron donor in the reaction centre of Chlorobium limicola. Biochemistry (submitted) Fowler GJS, Sockalingum GD, Robert B and Hunter CN (1994) Blue shifts in bacteriochlorophyll absorbance corre-

Bruno Robert late with changed hydrogen bonding patterns in light-harvesting 2 mutants of Rhodobacter sphaeroides with alterations at Tyr 44 and 45 Biochem J 299: 695–700 Fujiwara M and Tasumi M (1986) Resonance Raman and infrared studies on axial coordination to chlorophylls a and b in vitro. J Phys Chem 90: 250–255 Fujiwara M, Hayashi H, and Tasumi M (1988) Low-frequency vibrational spectra of chlorophylls a and b in solution: effects of axial ligation. Croatica Chem Acta 61: 435–446 Hashimoto H and Koyama Y (1990) The state of a carotenoid bound to spinach chloroplast as revealed by picosecond transient Raman spectroscopy. Biochim Biophys Acta 1017: 181–186 Hayashi H, Noguchi T, Tasumi M and Atkinson GH (1991) Vibrational spectroscopy of excited electronic states in carotenoids in vivo. Biophys J 60: 252–260 Hildebrandt P, Griebenow K, Holzwarth A and Schaffner K (1991) Resonance Raman spectroscopic evidence for the identity of the bacteriochlorophyll c organization in protein-free and protein containing chlorosomes from Chloroflexus aurantiacus. Z Naturforsch 46c: 228–232 Iwata K, Hayashi H and Tasumi M (1985) Resonance Raman studies of the conformations of all-trans carotenoids in light-harvesting systems of photosynthetic bacteria. Biochim Biophys Acta 810: 269–273 Jones M, Dawson MH, Mattioli TA, Hunter CN and Robert B (1994) Mutations at the M 210 Tyr level in bacterial reaction centers do not affect charge localization on the primary donor. FEBS Lett 339:18–24 Kirmaier C and Holten D (1987) Primary photochemistry of reaction centers from the photosynthetic purple bacteria. Photosynth Res 13: 225–260 Koyama Y, Takii T, Saiki K, Tsukida K and Yamashita KJ (1982) Configuration of the carotenoid in the reaction centers of photosynthetic bacteria. Comparison of the resonance Raman spectrum of the reaction centers of Rhodopseudomas sphaeroides G1C with those of cis–trans isomers from Biochim Biophys Acta 680:109–118 Koyama Y, Takii T, Saiki K and Tsukida K (1983) Configuration of the carotenoid in the reaction centers of photosynthetic bacteria. 2) Comparison of the resonance Raman lines of the reaction centers with those of the 14 different cis–trans isomers of Photobiochem Photobiophys 5:139–150 Koyama Y, Umemoto Y and Akamatsu A (1986) Raman spectra of chlorophyll forms. J Molecular Struct 146: 273– 287 Koyama Y, Takatsuka I, Nakata M and Tasumi, M (1988) Raman and infra-red spectra of the all-trans, 7-cis, 9-cis, 13-cis and 15-cis isomers of key bands distinguishing stretched or terminal bent configurations from central-bent configurations. J Raman Spectrosc 19: 37–49 Kuki M, Hashimoto H and Koyama Y (1990) The state of a carotenoid bound to the chromatophore membrane of Rhodobacter sphaeroides 2.4.1. as revealed by transient resonance Raman spectroscopy. Chem Phys Lett 165: 417– 422 Lutz M, (1972) Resonance Raman spectroscopy of plant pigments in solution and included in chloroplast layers. Compt Rend Acad Sci Ser. B, 275: 97–504

Resonance Raman spectroscopy Lutz M (1974) Resonance Raman spectra of chlorophyll in solution. J. Raman Spectrosc 2: 497–516 Lutz M (1977) Antenna chlorophyll in photosynthetic membranes: a study by resonance Raman spectroscopy. Biochim Biophys Acta 460: 408–430 Lutz M (1979) Application de la diffusion Raman de résonance aux pigments chlorophylliens. Thesis, Université Pierre et Marie Curie, Paris Lutz M. (1980) Resonance Raman studies of the bacterial photosynthetic reaction center. In: Murphy WF (ed) Proc. VIIth Intern. Conf on Raman Spectroscopy, pp 520–521. North Holland, Amsterdam Lutz M (1984) Resonance Raman Studies in Photosynthesis. In: Clark RJH and Hester RE (eds) Advances in IR and Raman Spectroscopy. Vol 11, pp 21 1–300. John Wiley and Sons, New York Lutz M and Kléo J (1974) Diffusion Raman de resonance de la chlorophylle d. Compt Rend Acad Sci Paris 279:1413– 1416 Lutz M and Mäntele W (1991) Vibrational spectroscopy of chlorophylls. In: Scheer H (ed) The Chlorophylls, pp 855– 902. CRC press, Boca Raton, Florida Lutz M and Robert B (1988) Chlorophylls and the photosynthetic membrane. In: Spiro TG (ed) Biological Applications of Raman Spectroscopy. Vol 3, pp 47–411. John Wiley and Sons, New York Lutz M and van Brakel G (1988) Ground-state molecular interactions of bacteriochlorophyll c in chlorosomes of green bacteria and in model system: a resonance Raman study. In: Olson JM, Ormerod JG, Amesz J, Stackebrandt E and Trüper HG (eds) Green Photosynthetic Bacteria, pp 23–34, Plenum Press, New York and London Lutz M, Kléo J, Gilet R, Henry M, Plus R and Leicknam JP (1975) Vibrational spectra of chlorophyll a and b labelled with and In: Klein ER and Klein PD (eds) Proc 2nd International Conference on Stable Isotopes. OakBrook, Ill, pp 462–468. US Dept of Commerce, Springfield, Va. Lutz M, K1éo J and Reiss-Husson F (1976) Resonance Raman scattering of bacteriochlorophyll, bacteriopheophytin and spheroidene in reaction centers of Rhodopseudomonas spheroides. Biochem Biophys Res Comm 69: 711–717 Lutz M, Chinsky L and Turpin PY (1983) Triplet states of carotenoid bound to the reaction centers of photosynthetic bacteria. Time resolved resonance Raman spectroscopy, Photochem Photobiol 36: 503–513 Lutz M, Szponarski W, Berger G, Robert B and Neumann JM (1987) The stereoisomerism of bacterial, reaction centerbound carotenoids revisited: an electronic absorption, resonance Raman and 1H-NMR study. Biochim Biophys Acta 894: 423–433 Mäntele WG, Wollenweber AM, Nabedryk E and Breton J (1988) Infrared spectroelectrochemistry of bacteriochlorophylls and bacteriopheophytins: implications for the binding of the pigments in the reaction center from photosynthetic bacteria. Proc Natl Acad Sci USA 85: 8468–8472 Mattioli TA, Robert B and Lutz M (1989) Time-resolved resonance Raman spectroscopy of the in bacterial reaction centers. In: Bertoluzza A, Fagnano C

175 and Monti P (eds) Spectroscopy of Biological Molecules. State of the Art, pp 303–304) Esculapio, Bologna Mattioli TA, Hoffmann A, Robert B, Schrader B and Lutz M (1991) Primary donor structure and interactions in bacterial reaction centers from near-infrared Fourier-transform resonance Raman spectroscopy. Biochemistry 30: 4648–4654 Mattioli T, Sockalingum D, Lutz M, Robert B (1992) Low temperature Fourier-transform Raman studies on bacterial reaction centers. In: Murata N (ed) Research in Photosynthesis, Vol. 1, pp 403–408. Kluwer Academic Publishers, Dordrecht Mattioli TA, Hoffmann A, Sockalingum DO, Schrader B, Robert B, and Lutz M (1993) Application of near IR Fourier transform resonance Raman spectroscopy to the study of photosynthetic proteins. Spectrochim Acta 49A: 785– 799 Mattioli TA, Williams JA, Allen W and Robert B (1994) Changes in primary donor hydrogen bonding interactions in mutant reaction centers from Rhodobacter sphaeroides. Identifications of the Vibrational frequencies of all the conjugated carbonyl groups Biochemistry 33: 1636–1643 Michel H and Deisenhofer J (1988) Relevance of the photosynthetic reaction center from purple bacteria to the structure of Photosystem II. Biochemistry 27: 1–7 Michel H, Epp O and Deisenhofer J (1986) Pigment-protein interactions in the photosynthetic reaction center from Rhodopseudomonas viridis. EMBO J 5: 2445–2451 Möenne-Loccoz P, Robert B and Lutz M (1989) A resonance Raman characterization of the primary electron acceptor in Photosystem II. Biochemistry 28: 3641–3645 Möenne-Loccoz P, Robert B, Ikegami I and Lutz M (1990) Structure of the primary electron donor in Photosystem I: a resonance Raman study. Biochemistry 29: 4740–4746 Olsen JD, Sockalingum GD, Robert B and Hunter CN (1994) Modification of a hydrogen bond between a bound chromophore and the subunit of the light-harvesting I antenna of Rhodobacter sphaeroides. Proc Natl Acad Sci U.S.A. 91: 7124–7128 Palaniappan V, Aldema MA, Frank HA and Bocian DF (1992) Qy excitation resonance Raman scattering from the special pair in Rhodobacter sphaeroides reaction centers. Implication for primary charge separation. Biochemistry 31: 11050–11058 Palaniappan V, Martin PC, Chinwat V, Frank HA and Bocian DF (1993) Comprehensive resonance Raman study of photo synthetic reaction centers from Rhodobacter sphaeroides. Implication for pigment structure and pigment-protein interactions. J Am Chem Soc 115: 12035–12049 Robert B (1983) Etude par diffusion Raman de resonance de complexes proteine-pigment antennes des Rhodospirillales. Thesis, Université Pierre et Marie Curie, Paris Robert B (1990) Resonance Raman studies of bacterial reaction centers. Biochim Biophys Acta 1017: 99–111 Robert B and Lutz M (1985) Structures of antenna complexes of several Rhodospirillales from their resonance Raman spectra. Biochim Biophys Acta 807: 10–23 Robert B and Lutz M (1986) Structure of the primary donor of Rhodopseudomonas sphaeroides: difference resonance Raman spectroscopy of reaction centers. Biochemistry 25: 2303–2309

176 Robert B and Lutz M (1988) Proteic events following charge separation in the bacterial reaction center: resonance Raman spectroscopy. Biochemistry 27: 5108–5114 Robert B and Möenne-Loccoz P (1989) Un site possible pour l’accepteur primaire d’électrons du photosystème I. Compt Rend Acad. Sci. Paris Série III, 308: 407–409 Robert B and Möenne-Loccoz P (1990) Is there a proteic substructure common to all photosynthetic reaction centers? In: Baltcheffski M (ed) Current Research in Photosynthesis, Vol 1, pp 65–68. Kluwer, Dordrecht Robert B, Szponarski W and Lutz M (1985) Resonance Raman studies of transient states in bacterial reaction centers. Springer Proc Phys 4: 220–224 Saito S. and Tasumi M (1983) Normal-coordinate analysis of carotene isomers and assignments of the Raman and infrared bands. J. Raman Spectrosc 14: 310–321 Saito S, Tasumi M and Eugster CH (1983) Resonance Raman spectra of all-trans and 15-cis isomers of carotene in the solid state and in solution, measurements with various laser lines from ultraviolet to red. J Raman Spectrosc 14: 299–309 Shreve AP, Cherepy NJ, Franzen S, Boxer SG and Mathies RA (1991) Rapid-flow resonance Raman spectroscopy of bacterial photosynthetic centers. Proc Natl Acad Sci USA 88:11207–11211 Shuvalov VA, Amesz J and Duysens LNM (1986) Picosecond spectroscopy of isolated membranes of the photosynthetic green sulfur bacterium Prosthecochloris aestuarii upon selective excitation of the primary donor. Biochim Biophys Acta 851: 1–5 Sturgis JN, Cogdell RJ, Jirsakova W, Reiss-Husson F and Robert B (1994) The structure and properties of the bacteriochlorophyll binding site in peripheral light-harvesting complexes from purple bacteria, Biochemistry (submitted) Tasumi M and Fujiwara M (1987) Vibrational spectra of

Bruno Robert chlorophylls, in: Clark RJH and Hester RE (eds) Advances in Spectroscopy. Vol. 14, Ch 6, pp 407–429. Wiley, London Van Bochove AC, Swarthoff T, Kingma H, Hof RM, van Grondelle R, Duysens LNM and Amesz J (1984) A study of the primary charge separation in green bacteria by means of flash spectroscopy. Biochim Biophys Acta 764: 343–346 Van de Meent EJ, Kobayashi M, Erkelens C, van Veelen PA, Otte SCM, Inoue K, Watanabe T and Amesz J (1992) The nature of the primary electron acceptor in green sulfur bacteria. Biochim Biophys Acta 1102: 371–378 Wachtweitl J, Farchaus JW, Dos R, Lutz M, Robert B and Mattioli TA (1993) Structure, spectroscopic and redox properties of Rhodobacter sphaeroides reaction centers bearing point mutations near the primary electron donor. Biochemistry 32: 12875–12886 Zadorozhnyi BA, and Ishchenko IK (1965) Hydrogen bond energies and shifts of the stretching vibration bands of groups. Opt Spectrosc (Engl. transl. 19: 306–308 Zhou Q (1989) Relations structure-fonction au sein du centre réactionnel bactérien: études par spectrométrie Raman de resonance, Thesis, Université Pierre et Marie Curie, Paris Zhou Q, Robert B and Lutz M (1987) Intergeneric structural variability of the primary donor of photosynthetic bacteria: resonance Raman spectroscopy of reaction centers from two Rhodospirillum and Rhodobacter species. Biochim Biophys Acta 890: 368–376 Zhou Q, Robert B and Lutz M (1989) Protein-prosthetic group interactions in bacterial reaction centers: resonance Raman spectroscopy of the reaction center of Rhodopseudomonas viridis. Biochim Biophys Acta 977: 10–18 Zinth W, Knapp EW, Fischer SF, Kaiser W, Deisenhofer J and Michel H (1985) Correlation of structural and spectroscopic properties of a photosynthetic reaction center. Chem Phys Lett 119: 1–4

Chapter 11 Stark Spectroscopy of Photosynthetic Systems Steven G. Boxer Department of Chemistry, Stanford University, Stanford, CA 94305–5080, USA

Summary I. Introduction II. Methods A. Analytical Methods B. Sample Preparation and Dielectric Breakdown C. Light Sources, Detectors and Power Supplies III. Limitations and Conceptual Issues A. Experimental Limitations B. Analytical Limitations C. Local Field Correction IV. Examples of Recent Results for Photosynthetic Systems A. The Special Pair B. Vibrational Stark Spectroscopy C. Unidirectional Electron Transfer in the RC Acknowledgements References

177 177 178 178 179 180 181 181 182 183 184 184 185 186 188 188

Summary The effects of applied electric fields on the absorption or emission spectrum of a molecule is known as Stark Spectroscopy. This method probes the movement of charge associated with optical excitation, and is thus sensitive to features which are important for chromophore systems that carry out charge separation, such as photosynthetic systems. This chapter updates a review written about 2 years ago which also described electric field effects on reaction dynamics, notably electron transfer reactions (Boxer, 1993). Several recent extensions of the method such as higher order Stark Spectroscopy and vibrational Stark Spectroscopy are discussed along with examples. Experimental methods and methods of analyzing spectra are discussed in great detail, along with a discussion of experimental and conceptual issues which complicate the analysis of Stark spectra. Abbreviations: HOSS – Higher order Stark Spectroscopy; LDAO – Lauryldimethylamine-N-oxide; RC – Reaction center; VSE – Vibrational Stark Spectroscopy I.

Introduction

of which have been widely used in the literature. The basic approach goes back to early in this century when electric fields were first used to perturb the energy levels of atoms in the gas phase. Extensive work was published on simple molecules, often as guests in host crystals, primarily as a test of the predictions of molecular orbital calculations (Hochstrasser, 1973). Recent developments in this area were reviewed by me two

The effect of an applied electric field on an absorption or emission spectrum is known as the Stark effect. This term is used interchangeably with electroabsorption or electrochromism, both Correspondence: Fax: 1-415-7234817; E-mail: [email protected]

177 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 177–189. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

178 years ago with a specific emphasis on photosynthetic systems (Boxer, 1993). That review also described experiments on the use of applied electric fields to perturb reaction dynamics, notably the rate of electron transfer reactions in photosynthetic reaction centers (RCs), measured either directly (Lockhart et al., 1990; Franzen et al., 1990; Franzen and Boxer, 1993) or indirectly via the effect of the electric field on the fluorescence which competes with electron transfer (Lockhart and Boxer, 1988; Lockhart et al., 1988). The emphasis in the following is on recent developments of absorption Stark spectroscopy and its application to understanding the excited states of photosynthetic pigments. I have particularly stressed experimental and conceptual complications, as well as the successes. The change in transition frequency due to an externally applied field F, is given by: where and are the change in dipole moment and polarizability, respectively, between the states involved. The most desirable method for obtaining Stark effect spectra is for uniaxially oriented molecules. If the first term dominates, as is often the case, this interaction gives rise to a linear shift in the absorption as the applied field strength is increased (the so-called linear Stark effect); the second term depends on the square of the applied field (the so-called quadratic Stark effect). In principle, it is possible to obtain information on the components of the difference polarizability tensor by varying the orientation of the molecule in the applied electric field. Unfortunately, uniaxial orientation is only rarely achieved in practice for complex molecules, and I am not aware of an example of a biological molecule where the Stark spectrum has been analyzed quantitatively. For membrane proteins, the sample can be uniaxially oriented across a lipid bilayer (no orientation in the plane of the bilayer). It is then possible to create an electric field by varying the ionic strength on the two sides of the bilayer. This gives rise to spectral shifts, known historically as electrochromic shifts, and, at least at a qualitative level, these shifts are very widely used to probe changes in transmembrane potential using voltage sensitive dyes. A somewhat related effect in-

Steven G. Boxer volves creating the field internally, rather than externally. For example, if charge is separated in response to light, as in the RC, the electric field due to these charges causes shifts in both the electronic absorption and vibrational bands of other spectator chromophores within the RC. This internal, transient electric field has a fixed spatial relationship with the spectator chromophores, so the observed Stark or electrochromic bandshifts can be treated as if the sample were completely oriented. Of course in this case it is not possible to vary the applied electric field strength systematically. Another possibility is that the protein containing a chromophore can be biaxially oriented, e.g., the protein is oriented across the bilayer, but the orientation does not distinguish one side of the bilayer from the other. If the absorption spectrum of a chromophore in such a system were very narrow, then the application of a transmembrane electric field would split the absorption due to the interaction of the fixed field direction with the two antiparallel projections of the chromophore difference dipole moments on the membrane normal. In practice, the absorption spectrum of biological chromophores is invariably inhomogeneously broadened, and the linewidth is typically several hundred which is considerably larger than the interaction energy between and the applied field (for reasonable values of either). In this case, the effect of the field is to broaden the absorption spectrum, and, so long as the interaction energy is small relative to the inhomogeneous linewidth, the effect depends quadratically on applied electric field. In this case the distinction between effects due to and can not be made on the basis of the field dependence, but must rather be made by analyzing the lineshape of the Stark effect spectrum. The same situation applies to nonoriented, immobilized samples which are simple to prepare and are the only cases to be discussed further in the next sections.

II. Methods

A. Analytical Methods It is straightforward to show that for an isolated absorption band in a non-oriented, immobilized sample when is smaller than the inhomogene-

Stark spectroscopy

179

ous line width, the change in absorption upon application of a electric field, F, is given by (Liptay, 1974; Mathies, 1974):

where depends on the transition polarizability and hyperpolarizability; +

(neglecting the contribution from the cross term of the transition polarizability and and is the molecular angle between and the transition moment p. is obtained from the second derivative coefficients of the Stark spectra obtained at different values of the experimental angle between F and the electric vector of the linearly polarized light used to probe f is the local field correction which is discussed further below. In order to obtain information on the desired electro-optical properties, the coefficients A, B, and C in Eq. (2), the observed change in absorption is analyzed in terms of contributions from the zeroth, first and second derivatives of the absorption band. We have recently extended the conventional method to include the higher harmonic responses of the absorption to the applied AC field, a technique we have called higher-order Stark spectroscopy (HOSS) (Lao et al., 1995a). The field-induced change in absorbance by an externally applied sinusoidal electric field is given by:

Changes

in

absorbance, etc. are recorded using lock-in detection at the 2nd, 4th and 6thharmonic frequencies, respectively, of the field modulation frequency The nth-order spectrum depends on the nth power of the applied field, Like the conventional or Stark spectrum, the higher order Stark spectrum can be fit to sums of derivatives of the absorption lineshape. For example, in the case of the spectrum, is fit to the sum of up to the 4th derivative of the absorption lineshape. Additionally, for each nth-order Stark spectrum, the nth

derivative component depends only upon and Thus, if the nth-order Stark spectrum is dominated by the nth derivative of the absorption spectrum, it is immediately evident that dominates An even simpler diagnostic for this case is that the (n + 2)-order Stark spectrum is the second derivative of the nth-order spectrum. As seen below this simple diagnostic is quite powerful because it is possible to obtain Stark spectra with good signal-to-noise, whereas it is quite difficult to obtain good quality higher derivatives of the absorption spectrum.

B. Sample Preparation and Dielectric Breakdown The conventional Stark effect depends quadratically on applied field strength (Eq. 2) and the higher order Stark spectrum on higher even powers of the field (Eq. (3)). Because the higher order terms have coefficients that become (usually) much smaller with term order, it is very desirable to obtain the highest possible applied electric field strength. The ultimate limit is determined by dielectric breakdown. Several factors determine the field at which dielectric breakdown occurs. (1) Low temperatures are essential for achieving the highest applied field strengths. We typically perform our experiments with samples immersed in liquid nitrogen in a strain-free optical dewar bubbling is minimized by blowing a thin stream of gaseous helium over the surface). In other cases, we have cooled samples by blowing cooled gaseous nitrogen or helium over the sample in an optical dewar, or by contacting the sample with a cold-finger. Heat transport is less efficient in these cases than when the sample is immersed in liquid nitrogen, so it is difficult to obtain the highest fields. Recently, in collaboration with MMR Technologies, Inc. (Mountain View, CA), we have adapted a miniature Joule– Thompson refrigerator for electric field experiments (Stanley and Boxer, 1995). The MMR refrigerator is so small that it can easily be translated during an experiment, e.g., when the sample is degraded by prolonged exposure to a high peak power laser. In this device the sample is in contact with a cold finger which is cooled by the expansion of pure nitrogen through a network of capillaries. The problem with this system, and

180 many others that use refrigerators, is that the samples are often liquids at room temperature, the cold-fingers operate under vacuum, and the sample cell is not vacuum tight The entire MMR refrigerator can be put in a – 20°C or – 80°C freezer until the liquid sample (typically containing 50–75% glycerol) is very viscous. At that point the refrigerator can be evacuated without causing sample bubbling. Alternatively the same can be very rapidly cooled using methane. Obviously this problem is not present for polymer film samples (e.g. in polyvinylalchohol); however, we have found that much better results can be obtained with frozen glass samples because the inhomogeneous broadening is typically less, the sample is not dehydrated, and it is possible to work with extremely small quantities by avoiding the sample waste associated with the preparation of films. In some cases it is useful to work at pumped liquid helium temperatures because this produces the narrowest spectral lines. This is achieved using a standard helium cryostat. (2) By working with the thinnest possible sample it is possible to obtain the best glasses and the highest fields. Of course this requires that the sample be very concentrated in order to achieve sufficient optical density. There are various approaches to making thin samples. We typically use precision spacers made of Teflon which are available in thicknesses down to about 10 microns. (3) We have found that the highest fields can be obtained by reducing the concentration of detergents as much as possible. In the case of photosynthetic RCs, we typically work at the absolute minimum detergent concentration compatible with maintaining good solubility (about 0.01% LDAO). (4) We use a power supply which has a current fuse which quickly cuts off the applied field as soon as current flow is sensed. Dielectric breakdown is a cumulative effect, and the sample can be preserved if breakdown is halted as soon as it begins. This is also advantageous because valuable samples or samples in difficult conditions (e.g. pumped He) are not destroyed. It is also an important safety precaution. We can routinely obtain applied electric fields (AC modulation) of approximately 1 MV/cm on frozen aqueous glasses immersed in liquid nitrogen for 10–25 micron cells. This can be achieved even at room temperature in polymer films, and fields as high

Steven G. Boxer as 3 MV/cm have been achieved for very thin polymer films (several micron) using pulsed electric fields (Lao et al., 1993).

C. Light Sources, Detectors and Power Supplies For most photosynthetic pigments, the absorption of interest occurs in the visible and near-infrared regions of the spectrum. The change in absorption can be probed using a conventional high pressure xenon or tungsten-halogen lamp whose output is dispersed through a monochromator. Si photodiodes are excellent detectors from about 400 to 1100 nm and have extremely low noise. Because they are only moderately sensitive, relatively high probe light intensities are needed. This can create interesting complications for pigment systems like the RCs which undergo charge separation with high quantum efficiency. Thus, electric field effects on the kinetics (Lockhart et al., 1990; Franzen et al., 1990; Franzen et al., 1992; Franzen et al., 1993) and steady-state population of intermediates (Franzen and Boxer, 1993) may be combined or may compete with the Stark effect on absorption (Lao et al., 1995b) or emission (Lockhart et al., 1991) giving rise to lineshapes which bear little relationship with the sum-ofderivatives form of Eq. (2). Si avalanche photodiodes have the advantage of integrated amplification. Most Si detectors are overcoated to minimize degradation by UV light and cannot be used effectively in the near UV. We have found that a photomultiplier with good near-UV sensitivity, wired so that only a few stages of gain are used, offers a reasonable compromise between signal-to-noise and spectral sensitivity. This has recently been used in a detailed study of the Stark effect of tryptophan (Pierce and Boxer, 1995). The infra-red region has not been explored very much thus far. We have used a germanium photodiode for Stark effect measurements on weak mixed-valence transitions in the Creutz–Taube ion in the 1–2 micron range (Oh and Boxer, 1990; Oh et al., 1991) and on the 1250 nm absorption band which is characteristic of in Rb. sphaeroides RCs (Stocker et al., 1993). Recently, we have performed the first conventional Stark experiments in the 2–4 micron range on simple organic nitriles using an InSb detector (Chattopadhyay

Stark spectroscopy and Boxer, 1995). Vibrational Stark effects (VSE) have only rarely been measured in condensed phases; however, this should be a rich area of investigation in the future. These measurements have recently been extended into the mid-IR using a mercury–cadmium–telluride (MCT) detector (Chattopadhyay and Boxer, to be published). A Nernst glower source is adequate for vibrational Stark measurements. In most cases the signal from the detector is processed in a lock-in amplifier. Although any conventional lock-in works well for most cases, recently introduced digital lock-in amplifiers offer many advantages. In particular, these lock-ins have exceptional dynamic reserve so that very small signals can be reliably extracted. This is especially important for HOSS and VSE experiments. The power supply used to generate the applied field is a critical component in the setup. A number of suppliers manufacture voltage amplifiers which can be used with a standard waveform generator to produce the desired output. In collaboration with Joe Rolfe in the Stanford Chemistry Department electronics shop we have developed several generations of power supplies which offer many practical advantages (convenient variation of amplitude and modulation frequency, DC offset, fast shut-off if current flows, etc.). HOSS experiments are especially sensitive to harmonic distortion in the power supply output as this can introduce higherharmonic signals which are not related to the desired signals described in Eq. (3). We currently minimize this problem by using a digitally synthesized sine wave from the lockin amplifier which is then amplified. Detailed schematics and power supplies are available through the author. III. Limitations and Conceptual Issues

A. Experimental Limitations The sample thickness is one of the primary sources of uncertainty, as the field strength is obtained by measuring the thickness and the applied voltage (the latter can be obtained with high precision with a calibrated high voltage probe). There have been suggestions that the Stark spectrum itself is a useful method for calibrating the field. Of course this cannot be used as a primary

181 standard because it depends on knowing the field strength. Once the Stark effect has been measured using an independently calibrated field, then the Stark effect for a particular transition can be used as a secondary standard, so long as there are no other sources of uncertainty. We have found that small sample-to-sample variations in the linewidth (associated with different amounts of detergent, buffer, sample imperfections, etc.) can lead to variations in the magnitude of the change in absorption for a given field. Thus, it is hazardous to use the Stark effect itself as a field calibration. Another poor method is to use the sample capacitance to measure thickness. This not only depends on independent information about the dielectric properties of the sample (generally not known very accurately), but it is also a bulk measurement and can give deceptive results for samples with thickness variations. It is, however, very useful to measure the sample capacitance during an experiment as a monitor of sample integrity. For frozen glass samples, the cell thickness can usually be accurately measured at room temperature, either by measuring interference fringes or by using solutions of known concentration. However, because the Stark effect is usually measured at low temperature, the absorption spectra of standard solutions may change, the sample may contract, reducing the thickness, and the contraction may not be homogeneous across the portion of the sample which is probed, especially if thin windows are used to constuct the cell. Under the most favorable conditions, when the sample is thin and freezes homogeneously, interference fringes are observed at low temperature. Although these fringes interfere with the analysis of the absorption spectrum, they are entirely absent in the Stark spectrum which is obtained using lock-in detection at some multiple of the field modulation frequency. The sample absorption, needed for lineshape analysis, can be measured on a separate sample. It is possible to obtain very precise thickness measurements on very thin films which are prepared by spin-coating. We have used a device manufactured by Dektak which is commonly used in the semi-conductor industry for measuring the thickness of thin films. This device works by scratching the film through to the substrate and then passing a stylus across the

182 sample surface and into the scratch to determine the thickness. Using this device, thin films can be measured with an accuracy of about 1000 Å, and it is possible to obtain quantitative information on film flatness over large areas. Some groups have used polymer films pressed between glass plates. This is an especially unsatisfactory method in our experience because there are surface variations on the film, and it is difficult to avoid air gaps.

B. Analytical Limitations The absorption spectrum from which the derivatives are obtained must have extremely good signal-to-noise, especially in the wings of the absorption. Because the derivatives are typically much noisier than the Stark effect data, simply using the derivatives to fit the data introduces considerable uncertainty into the analysis. A better approach is to fit the absorption to an arbitrary lineshape, e.g. a sum of Gaussians (with no physical meaning to the individual components), and then obtain the analytical derivatives of this best fit. Although an improvement, the best-fit to the absorption tends to be least accurate in the wings of the absorption. This is not a problem for fitting the absorption; however, it is a problem for fitting the Stark data as the wings of the absorption are important in the derivatives. The best approach we have developed to date is to simultaneously fit the absorption and the Stark data (Middendorf et al., 1993). A fundamental assumption underlying the application of Eq. (2) is that the electro-optic parameters such as and are constant across the absorption band. If the Stark spectrum can not be fit to a sum of derivatives, it may be that the electro-optical parameters vary across the inhomogeneous band width. A case of great relevance to some photosynthetic pigments (e.g. carotenoids (Gottfried et al., 1991a, b) and the dimeric special pair primary electron donor (Middendorf et al., 1993)) is when the chromophore intrinsically has a small but a large When such a chromophore is inserted into an ordered environment, e.g. a protein matrix, then the internal matrix electric field due to the constellation of polar, charged and polarizable groups in the protein can induce a dipole moment in the chrom-

Steven G. Boxer ophore. There will always be small variations in these fields from protein to protein (inhomogeneous broadening), and thus a distribution of induced dipole moments. An interesting nonbiological example of this has been studied in detail by V.P. Wild and co-workers who examined the induced dipole moment in a centrosymmetric organic dye at different wavelengths within the inhomogeneous absorption by measuring the Stark effect on holes burned at different wavelengths at 1.5 K (Vauthey et al., 1994). They found that the value of the induced in a moderately polar polymer matrix varied by a factor of 3 across the absorption band. Stimulated by this work, we measured the conventional and higher order Stark effect spectra for the same dye in the same matrix (Moore, Bublitz and Boxer, unpublished results). As expected, it was not possible to fit the conventional Stark effect spectrum with any combination of derivatives of the absorption. Although such a failure is not automatically diagnostic that this is physically what is occuring, it is one possibility. The higher-order Stark spectra can provide further insight and constraint on the fitting process. Another problem is that even a simple electronic transition is almost always accompanied by some vibronic structure, often poorly resolved. Because the intensity of these vibronic features depends on coupling with other states whose electro-optic properties may be different, it is possible that the Stark effect for these features will be intrinsically different from that of the (typically) dominant 0–0 transition. This has not been examined systematically in any detail. In one case, the carotenoid spheroidene in the B800–850 antenna complex from Rb. sphaeroides, a prominent vibronic progression is observed. The entire progression could be quite well fit to the second derivative of the absorption band (Gottfried et al., 1991a,b). Multiple, overlapping electronic absorption bands, sometimes from the same molecular species, or, as is the case of RCs, from different chromophores are another difficulty. The obvious approach to this problem is to deconvolve the absorption features. Although this is a standard procedure for the analysis of complex absorption bands, the requirements for the analysis of the Stark spectrum are much more stringent because

Stark spectroscopy the electrooptic properties of the bands may be quite different from each other and the wing of the Stark effect of a transition with a large dipole moment may, for example, swamp the Stark effect for a nearby absorption with a small dipole moment. One of the most insidious examples arises when there are partially overlapping absorption features which have very different Stark effects, as in photosystem II RCs. Even in the purest preparations, there is substantial overlap among all the absorption bands of different chromophores in the region. In an early attempt to analyze this spectrum, Lösche et al. (1988) concluded that P680 has a small comparable to or smaller than a monomeric chlorophyll. This might indicate that P680 is not dimeric. We analyzed the Stark spectrum of PSII RCs, and attempted to fit the data more accurately. Our best fit suggested that for P680 was at least as large or larger than for a monomeric chlorophyll (Steffen, 1994); however, because the band overlap was so great, we never felt confident enough to draw any further conclusion. Finally, there is the possibility that the basic Liptay formalism underlying Eq. (2) is inappropriate. This may occur when the observed transition is strongly coupled to or degenerate with a dark state. General treatments have been developed by Reimers and Hush (1991) and Sacra et al. (1995), the latter for the special case of semi-conductor nanocrystals. To date, there is no evidence that a breakdown of the fundamental assumptions inherent in the Liptay formalism plays a significant role in the Stark spectra of photosynthetic pigments; however, this may well prove to be wrong, especially as the analyses become more refined.

C. Local Field Correction The local field correction accounts for the difference between the actual field felt at the position of the chromophore whose Stark spectrum is being probed and the precisely known applied electric field. Lack of knowledge of the local field correction is a long-standing limitation in the quantitative comparison between observed Stark effects and theory. The local field correction is discussed in depth in classical texts, such as Böttcher (1973). The local field correction is not

183 to be confused with the matrix electrostatic fields due to the protein. The latter are always present and may be very large in an ordered medium such as a protein, or they can change in time, as in the case of the electric field due the charges formed following light-induced electron transfer. The local field correction accounts for the difference in the field felt at the probe chromophore upon application of the external applied field. In general the local field correction is a tensor quantity. Taking the local field correction to be a scalar, The local field correction is not the dielectric constant. Various models for the local field correction, such as the Lorentz model (Böttcher, 1973), give expressions for f in terms of the dielectric constant, but the dependence is rather weak, e.g., for a spherical cavity the dependence is: Thus, for typical values of the dielectric constant of frozen solutions, f is not much greater than unity. Irrespective of the form chosen for the cavity, the value of f is greater than 1. Thus, the internal field actually felt at the position of the chromophores is greater than the externally applied field, though not by very much for typical dielectrics at low temperature. Expressions for the local field correction become still more complex for a chromophore inside a protein which is itself a dilute solute in a frozen glass (Lösche et al., 1988). However, because the great majority of the sample to which the external field is applied is the bulk frozen solvent and this is the same for different chromophores within a protein complex (e.g. the RC) or among different chromophore-protein complexes embedded in the same frozen solvent, differences in the local field correction for different chromophores in different local environments are likely to be small. Thus, although the absolute value of the electro-optic parameters may be systematically in error (likely by a small amount), the relative values are likely to be reliable. We have taken the approach of reporting electrooptic parameters in terms of the local field correction as a formality. Finally, the angle between the change in dipole moment and the transition dipole moment used to probe the Stark effect does not depend on the local field correction factor (see Eq. 2). This angle is often of greater

184 interest than the precise magnitude of the change in dipole moment itself. There are approaches to obtain quantitative information on the local field correction. For example, in the RC, the driving force for the recombination reaction is quite well known. Upon application of a large external electric field, it should be possible to perform the reverse reaction, namely charge separation without photoexcitation, as the energy of the state is tuned below that of ground state By measuring the concentration of at equilibrium in an accurately calibrated external electric field, and comparing this with the equilibrium concentration expected given the magnitude of the dipole moment (whose magnitude does not depend to any appreciable extent on the details of the locations of the charges on either radical ion), it should be possible to obtain the actual field felt by the dipole in the RC. We attempted this experiment, and at the highest fields available were unable to detect any appreciable formation in Rb. sphaeroides RCs in a PVA film. Based upon what is known about these RCs in the absence of an external field, we could conclude that f in this system could be no greater than 1.3, i.e., if it had been greater, the field external would have produced a measurable concentration of (Franzen and Boxer, 1993). A possibly related effect involving charge separation between P and cytochrome has been reported in Rps. viridis RCs (Alegria et al., 1993). Extensions of this approach to other systems where the driving force is smaller are in progress. IV. Examples of Recent Results for Photosynthetic Systems As mentioned at the outset, results until late 1992 were summarized in detail in an earlier review (Boxer, 1993) and these will not be repeated. A few recent results are discussed briefly in the following.

A. The Special Pair In the original papers describing the Stark effect for the special pair, it was obvious that for

Steven G. Boxer its transition was larger than for a monomeric BChl, either the B bands in the RC or the pure isolated monomeric chromophores (Lockhart and Boxer, 1987; 1988; Lösche et al., 1987, 1988a,b). The quantitative analysis assumed that only the second derivative contributed to the lineshape, although it was clear that this was not rigorously true. By working at 1.5 K in a frozen glass, where the special pair absorption is narrower and some substructure is evident, it was possible to test the lineshape analysis with much greater precision (Middendorf, 1992). The Stark lineshape at 1.5 K is not well fit by the second derivative of the absorption, rather there appears to be a significant contribution from the first derivative and some zeroth derivative. The interpretation of this result, no matter how good the data and the fit, is predicated on the validity of the Liptay formalism, which may break down for this system. Assuming the validity of the Liptay formalism, we find that for the special pair is huge compared to that of a monomer. The estimated magnitude of is little affected by this more refined analysis, though increases from about 38° to about 45°. Although it is still the case that is several times larger for the dimer than the monomer, the really striking finding is that is much larger for the dimer. This can be interpreted as being a natural consequence of mixing intra-dimer charge-transfer character into the lower exciton state of P (Middendorf et al., 1993). If this model is correct, then it might be expected that as a large field is applied, radically shifting the energies of the CT states, then the Stark lineshape should change. Model calculations demonstrate that this should be observed if the CT states are within about of the exciton states (the precise value is model dependent). We have measured the lineshape in PVA films for fields up to nearly 3 MV/cm and find an excellent fit to the expected quadratic field dependence of the amplitude, with no change in lineshape. Thus, these data suggest that the CT states are several thousand higher (or, in principle, lower) than the exciton states. The value of extracted from the lineshape analysis is even larger than for long polyenes which are known to be highly polarizable. Of course it is not possible to remove the special pair

Stark spectroscopy from the RC while maintaining its structure and study its electro-optic properties in a simpler matrix, as is possible for polyenes, though we have studied the Stark spectra of a number of covalently linked special pair model systems (Middendorf, 1992). If we make the assumption that the intrinsic (gas phase) of the special pair is small, and the observed large for the special pair is induced by interaction between the matrix (protein) electrostatic field in the RC and the large of the special pair, then we can estimate the matrix electric field to be at least several MV/cm. Given the absence of a dependence of the lineshape on applied field, this internal matrix field may be even larger. Furthermore, as discussed in detail in Middendorf et al. (1993), it is likely that is highly anisotropic, with the dominant component along the long axis of the special pair. If the matrix electric field pointed along this direction, then a much larger value of would be observed. Therefore, we concluded that the matrix field is large and points roughly perpendicular to the long axis of the special pair. This is roughly along the local axis of the RC. Some recent electrostatics calculations support this suggestion (M. Gunner, D. Chandler, personal communication). Higher-order Stark spectra have been obtained for monomeric BChl and BPheo, the special pair at 77 and 1.5K, the heterodimer mutant ((M)H202L), and for the carotenoid spheroidene both pure in an organic glass and in the B800– 850 antenna complex (Lao et al., 1995a). A key result is shown in Fig. 1 which compares the second and fourth order Stark spectra of the wildtype homodimer and mutant (M)H202L heterodimer. The absorption spectrum of the heterodimer consists of two features whose total oscillator strength is comparable to that of the homodimer (because the bands are so spread out, it appears to be weaker). In earlier work (Hammes et al., 1990), it was evident that there was a huge Stark effect for the lower energy band compared to that of the homodimer, and this is evident in Fig. 1. However, it proved difficult to analyze the Stark spectrum in terms of contributions of and because it is difficult to obtain precise information on the absorption lineshape of the transition corresponding to the large Stark

185

effect. As seen in Fig. 1, the fourth order Stark effect on this lowest energy transition is likewise very strong, and it is very close to the second derivative of the 2nd-order conventional Stark spectrum. Thus, for this transition, dominates The contrast between this result and the comparable data for the homodimer is striking. It is obvious by inspection that the fourth order Stark spectrum is not the second derivative of the 2nd order Stark spectrum for the native homodimer. This confirms the earlier conclusion that the Stark lineshape for the special pair is not dominated by but rather that makes the dominant contribution (Middendorf et al., 1993). A general theoretical treatment of the higher order lineshape is in hand, and we are currently attempting to extract further information on the components of the polarizability tensor from this data.

B. Vibrational Stark Spectroscopy Chromophores which absorb in the visible and near IR are not that common in most proteins; however, molecular vibrations are associated with every residue. Recently, with the advent of better FTIRs and time resolved IR methodology, there has been considerable interest in assigning some of the vibrational features of RCs (as well as many other proteins) and using changes in the vibrational frequencies or intensities to monitor the response of the protein to electron transfer events (e.g. Maiti et al., 1993). In order to evaluate these changes quantitatively, it would be useful to have an independent experimental calibration of the sensitivity of the vibrational frequencies to electric fields. Towards this end, we have performed the first measurements of the VSE for anisonitrile (chosen largely because it is a strong transition in a convenient spectral region) (Chattopadhyay and Boxer, 1995). The VSE spectrum is shown in Fig. 2. From this we obtain and i.e., is collinear with the transition moment, as expected. If an electric field were aligned with this bond direction, the Stark tuning rate is: Extensions to a wider variety of vibrational transitions are in progress. These data provide fundamental information on vibrational anharmonicity, as well as calibrating the sensitivity of vibrational spectra to local electric fields.

186

C. Unidirectional Electron Transfer in the RC One useful application of the electronic Stark effect is to provide a calibration for electrochromic shifts due to the transient internal electric field produced by charge separation (Steffen et al., 1994). In particular, when charge is separated between P and the resulting electric field is felt by the monomeric bacteriochlorophyll and bacteriopheophytin monomers (referred to as B and H, respectively). In order to calibrate these shifts, information on the sensitivity of monomeric pigments to electric fields is first obtained from conventional Stark spectroscopy. The B and H bands consist of overlapping absorptions from the monomeric BChls and BPheos, respectively, on the functional L and non-functional M sides. This limits the accuracy of the determination of

Steven G. Boxer

both the magnitude and direction of for these chromophores; however, from an analysis of the Stark spectrum the magnitude of the are roughly the same on the functional and non-functional sides. The direction of is more problematic because the measurement of the angle dependence is compromised when bands overlap. We have no reason to believe that the direction of for the monomeric chromophores is appreciably different within the RC than when the pigments are extracted from the RC and studied in a simple organic solvent. Unfortunately, the angle defines two projections of on the transition moment direction, i.e. two antiparallel cones. Upon formation of the B bands are observed to shift to higher energy and the H bands shift to lower energy. At a qualitative level this result immediately restricts the absolute direction of on each chromophore type to one of the

Stark spectroscopy

two cones. Thus, both the magnitude and absolute direction of for all four spectator chromophores are known reasonably well. The electrochromic bandshifts can be obtained by subtracting an absorption spectrum of a sample in the state from one in the ground state, normalized for the amount of bleach of P at its absorption maximum. The data at 1.5 K are shown in Fig. 3. Here too the overlap of the absorption bands causes some uncertainty; how-

187

ever, it is evident by inspection that and shift by approximately the same amount, shown expanded in Fig. 4. Because the Stark effect spectrum in an external field demonstrates that their absorption bands are comparably sensitive to an electric field, this result suggests that the field at the and sites due to the transient internal electric field is about the same. However, it is obvious by inspection of the X-ray structure that the internal field due to should be substantially larger at than at because is is much closer to the charges on and than is Thus, we are forced to conclude that the dielectric screening on the L-side is considerably greater than on the M side. A quantitative analysis suggests that it is approximately three times greater on the functional side (Steffen et

188

Steven G. Boxer band region of the spectrum, we believe that its intensity is quite small and the band is likely broad. These ideas are supported by more detailed analyses of the absorption spectra of modified RCs, including the Stark spectra (Moore and Boxer, to be published). Acknowledgements This work was supported in part by grants from the NSF Biophysics and Chemistry programs. The author gratefully acknowledges contributions to the work from his lab from Drs. David Lockhart, Dennis Oh, Thomas Middendorf, David Gottfried, Sharon Hammes, Martin Steffen, Steffen Franzen, Arun Chattopadhyay, and Kaiqin Lao, and from Laura Moore, Laura Mazzola, and Huilin Zhou. References

al., 1994). This dielectric asymmetry, probed functionally, may be a significant contributor to unidirectional electron transfer in the RC. Its molecular origin is not known; however, it is likely the result of long-range interactions from many amino acids. A similar analysis of the B bands gives nearly identical results for the dielectric asymmetry, even though some might question whether the B-band shifts upon are entirely electrochromic in origin (there is no evidence for appreciable exciton interactions between the P or B chromophores and the H chromophores). Our view is that the exciton interaction between P and B is considerably smaller than the shifts which occur upon formation. Furthermore, although the upper exciton band of P contributes to the absorption in the B-

Alegria G, Moser CC, and Dutton PL (1992) Pulsed electric field induced reversed electron transfer from ground state to the cytochrome c heme in Rps. viridis. In: Breton J and Verméglio A (eds) The Photosynthetic Bacterial Reaction Center II, pp 313–319. Plenum Press, New York. Böttcher CJF (1973) Theory of Dielectric Polarization, Vol 1, Elsevier, Amsterdam. Boxer SG (1993) Photosynthetic reaction center spectroscopy and electron transfer dynamics in applied electric fields. In: Deisenhofer J and Norris JR (eds) The Photosynthetic Reaction Center Vol 2 pp 179–220. Academic Press, New York. Boxer SG, Goldstein RA, Lockhart DJ, Middendorf TR and Takiff L (1989) Excited states, electron transfer reactions, and intermediates in bacterial photosynthesis. J Phys Chem 93: 8280–8294. Chattopadhyay A and Boxer SG (1995) Vibrational Stark spectroscopy. J Amer Chem Soc 117: 1449–1450. Franzen S and Boxer SG (1993) Temperature dependence of the electric field modulation of electron transfer rates: Charge recombination in photosynthetic reaction centers. J Phys Chem 97: 6304–6318. Franzen S, Goldstein R and Boxer SG (1990) Electric field modulation of electron transfer reaction rates in isotropic systems: Long-distance charge recombination in photosynthetic reaction centers. J Phys Chem 94: 5135–5149. Franzen S, Lao K and Boxer SG (1992) Electric field effects on kinetics of electron transfer reactions: Connection between experiment and theory. Chem Phys Lett 197: 380. Gottfried DS, Steffen MA and Boxer SG (1991a) Large protein-induced dipoles for a symmetric carotenoid in a photosynthetic antenna complex. Science 251: 662–665. Gottfried DS, Steffen MA and Boxer SG (1991b) Stark effect spectroscopy of carotenoids in reaction center and antenna

Stark spectroscopy complexes from photosynthetic bacteria. Biochim Biophys Acta 1059: 76–90. Hammes S, Mazzola L, Boxer SG, Gaul D. and Schenck C (1990) Stark Effect spectroscopy of the heterodimer mutant of Rb. sphaeroides. Proc Nat Acad-Sci 87: 5682–5686. Hochstrasser R (1973) Electric field effects on oriented molecules and molecular crystals. Acc Chemical Research 6: 263–269. Lao K, Franzen S, Stanley R, Lambright D and Boxer SG (1993) Effects of applied electric fields on the quantum yields of initial electron transfer steps in bacterial photosynthesis: I Quantum yield failure. J Phys Chem 97: 3165– 13171. Lao K, Moore LJ, Zhao H and Boxer SG (1995a) Higherorder Stark spectroscopy: Polarizability of photosynthetic pigments. A new method for probing the electronic properties of biological chromophores. J Phys Chem, 99: 496– 500. Lao K, Franzen S, Stanley R, Steffen MA, Lambright D and Boxer SG (1995b) Effects of applied electric fields on the quantum yields of initial electron transfer steps in bacterial photosynthesis: II Field induced absorption anisotropy and quantitative analysis. Chem Phys 197: 259–275. Liptay W (1974) in Excited states, EC Lim, (ed), Vol 1, pp 129–229, Academic Press, New York. Lockhart DJ and Boxer SG (1987) Magnitude and direction of the change in dipole moment associated with excitation of the primary electron donor in Rb. sphaeroides reaction centers. Biochemistry 26: 664–668. Lockhart DJ and Boxer SG (1988a) Stark effect spectroscopy of Rhodobacter sphaeroides and Rhodopseudomonas viridis reaction centers. Proc Natl Acad Sci USA 85: 107–111. Lockhart DJ and Boxer SG (1988b) Electric field modulation of the fluorescence spectrum in Rhodobacter sphaeroides reaction centers. Chem Phys Lett 144: 243–250. Lockhart DJ, Goldstein R and Boxer SG (1988) Structurebased analysis of the initial charge separation step in bacterial photosynthesis: Electric field induced fluorescence anisotropy. J Chem Phys 89: 1408–1415. Lockhart DJ, Kirmaier C, Holten D and Boxer SG (1990) Electric field effects on the initial electron transfer kinetics in bacterial photosynthetic reaction centers. J Phys Chem 94: 6987–6995. Lockhart DJ, Hammes SL, Franzen S and Boxer SG (1991) Electric field effect on emission lineshapes when electron transfer competes with emission: an example from photosynthetic reaction centers. J Phys Chem 95: 2217–2226. Lösche M, Feher G and Okamura MY (1987) The Stark effect in reaction centers from Rhodobacter spheroides R-26 and Rhodopseudomonas viridis. Proc Natl Acad Sci USA 84: 7537–41. Lösche M, Feher G and Okamura MY (1988) The Stark effect in photosynthetic reaction centers from Rhodobacter sphaeroides R-26, Rhodopseudomonas viridis and the complex of photosystem II from spinach. In: Breton J and Verméglio A (eds) The Photosynthetic Bacterial Reaction Center — Structure and Dynamics, pp 151–164, Plenum, New York.

189 Mathies RA (1974) Experimental and Theoretical Studies of the Excited Electronic States of Some Aromatic Hydrocarbons Through Electric Field Perturbation and Through Chemical Substituents. PhD Thesis, Cornell University, Ithaca, New York. Middendorf TR (1992) Photochemical Holeburning and Stark Effect Spectroscopy of Photosynthetic Reaction Centers and Model Compounds. PhD Thesis, Stanford University. Middendorf TR, Mazzola LT, Lao K, Steffen MA and Boxer SG (1993) Stark effect (electroabsorption) spectroscopy of photosynthetic reaction centers at 1.5K: Evidence that the special pair has a large excited-state polarizability. Biochim Biophys Acta 1144: 223–234. Oh D and Boxer SG (1990) Electrochromism of the near-IR absorption spectra of bridged ruthenium mixed valence complexes. J Amer Chem Soc 112: 8161–8162. Oh DH, Sano M and Boxer SG (1991) Electro-absorption (Stark effect) spectroscopy of mono and bi-ruthenium charge transfer complexes: Measurements of changes in dipole moments and other electro-optic properties. J Amer Chem Soc 113: 6880–6890 (1991). Pierce DW and Boxer SG (1995) Stark effect spectroscopy of tryptophan. Biophys J 68: 1583–1591. Reimers JR and Hush NS (1991) Electric field perturbation of electronic vibronic absorption envelopes: Application to characterization of mixed valence states In: Prassides K (ed) Mixed-valence systems: Applications in chemistry, physics and biology, pp 29–39, Kluwer Academic Publishers, Dordrecht. Sacra A, Norris DJ, Murray CB and Bawendi MG (1995) Stark spectroscopy of systems with overlapping interacting transitions: Application to CdSe nanocrystallites. J Chem Phys, 103: 5236–5245. Stanley RA and Boxer SG (1995) Oscillations in the spontaneous emission from photosynthetic reaction centers. J Phys Chem, 99: 859–863. Steffen MA, (1994) Electrostatic Interactions in Photosynthetic Reaction Centers. PhD Thesis, Stanford University, 1994. Steffen MA, Lao K and Boxer SG (1994) Dielectric asymmetry and the origin of unidirectional electron transfer in photosynthetic reaction centers. Science 264: 810–816. Stocker JW, Hug S and Boxer SG (1993) Stark effect spectroscopy of the 1250 nm band of Rb. sphaeroides reaction centers and related model compounds. Biochim Biophys Acta 1144: 325–330. Vauthey E, Voss J, deCaro C, Renn A and Wild UP (1994) Spectral hole-burning and Stark effect: frequency dependence of the induced dipole moment of a squaraine dye in polymers. Chem Phys 184:347–356. Maiti S, Cowen BR, Diller R, Iannone M, Moser CC, Dutton PL and Hochstrasser RM (1993) Picoseond infrared studies of the dynamics of the photosynthetic reaction center. Proc Natl Acad Sci USA 90: 5247–51. Walker GC, Maiti S, Cowen BR, Moser CC, Dutton PL and Hochstrasser RM (1994) Time resolution of electronic transitions of photosynthetic reaction centers in the infrared. J Phys Chem 98: 5778–5783

This page intentionally left blank

Chapter 12 The Photoacoustic Method in Photosynthesis – Monitoring and Analysis of Phenomena Which Lead to Pressure Changes Following Light Excitation Shmuel Malkin Biochemistry Department, The Weizmann Institute of Science, Rehovot, 76100, Israel

Summary I. Introduction — Historical Notes and Main Aspects II. Experiments and Results with the Gas-Phase Coupled Microphone A. Experimental Technicalities B. The Photothermal Signal – Observations, Properties and Interpretation 1. Measurement of Energy Storage (ES) 2. ES Dependence on Experimental Parameters Dependence on the Modulation Frequency f a. b. ES as a Function of the Wavelength c. ES as a Function of the Light Intensity I d. The Effect of Fluorescence – Determination of its Absolute Quantum Yield e. Time Domain Measurements f. Some Numerical Estimates 3. Acoustic Signals Obtained with Magnetic Field Modulation (Magnetophotoacoustic Effect) C. The Photobaric Signal – Observations, Properties and Interpretation 1. The Photothermal/Photobaric Composite Signal 2. Separation of the Photobaric Signal 3. Quantitation of the Photobaric Signal a. The Diffusion Model b. O/T as a Measure of the Quantum Yield 4. The Limiting Rate in the Oxygen Evolution Process 5. Uptake Signals III. Experiments and Results in the Time Domain with a Sample Coupled Piezoelectric Sensor IV. Applications to Physiological Studies References

191 192 194 194 195 195 196 196 197 197 197 198 198 199 199 199 200 201 201 201 202 202 202 204 204

Summary Following light excitation and relaxation of the excited state, pressure changes are developed in an investigated sample or in the gas phase around the sample. Commonly, these changes result from the conversion of the light energy to heat, leading to temperature rise and hence to pressure increase (photothermal effect). Although quite small and rapid, the pressure changes are detectable by suitable fast sensors. These are either piezoelectric transducers, used in a flash photolysis mode with time domain measurement or microphones, used commonly in conjunction with periodically modulated light excitation (yielding periodic pressure perturbations) but also in a flash photolysis mode. Correspondence: Fax: 972-8-9344118; E-mail: [email protected]

191 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 191–206. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

192

Shmuel Malkin

Photosynthetic samples are quite unique iª that their photochemical activities give rise to additional mechanisms which contribute appreciably to pressure changes. These are: (1) gas evolution and uptake, most notably oxygen evolution, detectable in leaves and lichens with a sensor put in the gas phase (photobaric effect), (2) molecular volume changes, resulting from photosynthetic charge separation and proton transfer to and from the buffered medium. The methodology which deals with the measurement and analysis of these pressure changes is commonly known as photoacoustics (or optoacoustics) and the resulting signals are referred to as photoacoustic (optoacoustic) signals. Although implied by the name, sound production as such, is important only at sufficiently high frequencies (or short times) where the sound wavelength is small relative to the sample dimensions. Otherwise, the pressure changes are practically uniform over the entire sample. This chapter outlines methods to measure, separate and analyze each of the contributions to the photoacoustic signals in photosynthetic samples and the value of the information obtained by this analysis. Broadly speaking, the photothermal contribution allows to deduce the magnitude of energy storage with the capability to specifically assign energy and decay kinetics to particular electron transfer, intermediates; the photobaric contribution serves as a sensitive analytical tool for photosystem II activity and electron transport in general, in complex structures such as leaves; molecular volume changes are related to molecular structural changes within the reaction center. The use of periodically modulated light offers various other advantages: high sensitivity, separation from dark processes and possibility of conditioning by background (non-modulated) lights without interfering with the measurements. Applications include straightforward analysis of the interaction between the two photosystems, characterization and dynamics of physiological changes and stress conditions, possible detection of cyclic electron flow in-vivo, among others. Abbreviation: ES – Energy storage; f – Modulation frequency; P870 – Primary electron donor (in purple bacteria); PA – Photoacoustic; PL – Photochemical loss; PS – Photosystem; – First and second acceptor quinones I. Introduction — Historical Notes and Main Aspects The fate of excited states is usually monitored by their decay, the formation of new state or chemical species and by light emission. A more specialized task is to measure the energy lost as heat during each step of the relaxation. This is because the involved temperature changes are very rapid and small, in the order of millidegree Celsius with a nearly saturating flash for ordinary concentrations of a photochemical substrate, and perhaps three orders of magnitude less for a photosynthetic sample that may contain hundreds of absorbing pigments per one reaction center. Nevertheless, it is possible, utilizing a number of consequent physical effects, to measure temperature changes with very sensitive sensors. One way, among others, is the photoacoustic (or optoacoustic) method (abbreviated PA), utilizing

microphones or piezoelectric transducers to measure the prompt pressure changes caused by the temperature changes (Callis et al., 1969; Rosencwaig, 1980; Patel and Tam, 1981; Tam 1986; Braslavsky and Heioff, 1989). PA studies concerning photophysical relaxation in the gas phase were already carried out in the forties, using periodically modulated light and a microphone which was inserted in the sample gas. The rate of heat release is modulated and is phase shifted, depending on the modulation frequency, due to the finite relaxation times. This results in corresponding phase shifts in the pressure modulation. From the frequency dependence of the amplitude and phase of the microphone signal (relative to a proper reference, where the relaxation to the ground state is sufficiently rapid and heat release is prompt) it was possible to deduce the energy and first order decay constants of the relaxation intermediates

The photoacoustic method (Hunter and Stock, 1974; Hunter et al., 1974, and references therein). Callis et al. (1969) constructed a flash photolysis microphone calorimeter for the liquid phase, using a special microphone that was inserted into the liquid. A flash produced an increase in the pressure following the increase in the temperature, which was recorded as a function of time. The excess pressure was released at longer times than the experiment time, by a constructed slow leak valve and by the slow heat release to the surrounding. Initial studies aimed at triplet states of organic dyes. In these experiments there was an initial “prompt” pressure increase (on the time scale of the experiment), due to the prompt heat release in the fast transitions from the excited singlet. This was followed by a slower measurable pressure increase due to the relaxation of the triplet, allowing to determine its yield, energy and relaxation kinetics. The same mode of PA measurement was later applied to photosynthetic bacterial samples: chromatophores (Callis et al. 1972) and reaction centers (Arata and Parson 1981). In these systems there was an additional contribution to the microphone response, which was assigned to a change in the molecular volume (at constant pressure). This was clearly demonstrated at ca. 4°C in the aqueous suspension (where the thermal expansion is zero) by the appreciable microphone response, which was obviously absent from a photochemically inert reference. Prompt molecular volume changes, assigned to the electron transfer from P870 to or were negative (i.e. signaling a contraction). A slower kinetic phase of expansion was assigned to molecular volume change associated with proton transfer. At a higher temperature, the photothermal contribution could be isolated, by subtraction of the molecular volume changes, resulting in an estimate for the energies of and Similar studies were conducted on bacteriorhodopsin containing membranes (Ort and Parson 1979 a–c). A different PA experimental approach, applicable for thin, optically dense, or strongly light scattering layered samples, uses a microphone in the gas phase surrounding the sample. With periodically modulated or pulsed light excitation, the resulting modulated or pulsed heat is first conduc-

193 ted to the surface, before the transduction to pressure changes. A thin gas layer adjacent to the surface, whose thickness is in the order of a “thermal diffusion length” (cf. below for a definition) undergoes expansion and relaxation cycles, acting as a piston on the bulk gas and resulting in bulk pressure changes (Rosencwaig and Gersho, 1978; Rosencwaig, 1980). The main limiting factor, in determining the strength of the signal, is the efficiency of the modulated heat diffusion to the sample’s surface. The oscillatory part in the temperature field within the sample is mathematically a superposition of waves, propagating from point sources where modulated heat is evolved, with amplitudes which are proportional to the extent of light conversion to heat. These thermal waves are however peculiar in that their amplitude decreases exponentially along the distance of propagation (a result of the heat conduction differential equation, being first order in the time). This property is characterized by a “thermal diffusion length” parameter, equal to the distance over which the amplitude is attenuated by a factor e, where f is the modulation frequency, is the heat diffusivity, the heat conductivity, the density and the specific heat (at constant pressure). Besides its significance as an attenuation factor, is also equal to the thermal wavelength. Roughly, is an average effective range for the thermal waves; only that part of the light absorbed along a distance of a very small number of thermal diffusion lengths below the surface will be effective in producing a measurable PA signal. For most non-conductors of heat is equal to just a few microns (e.g. ca. at 100 Hz for water), so that this mode of PA probes essentially the part of the sample adjacent to the surface. Layered suspensions of photosynthetic material (Carpentier et al., 1985, 1989; Owens et al., 1990) as well as water soaked marine algal thalli (Fork et al., 1991; Malkin et al., 1990) gave PA signals which represent photothermal conversion only and reflect significant photochemical energy storage. When PA measurements were carried out in leaves, the PA signals could not be interpreted by the photothermal mechanism alone. It was

194 concluded that a large fraction of the signal arises from modulated oxygen evolution (Bults et al., 1981), obviously arising from the modulated change in the number of moles released into a confined volume of the gas phase. This mechanism was tentatively named “photobaric”. Regarding the significance of assimilation and photorespiratory oxygen uptake in producing (uptake) PA signals, it is inferred that their modulation is heavily damped (at the range of frequencies used – higher than ca. 10 Hz) and therefore their contribution to the PA signal is negligible. An analogous transmission mechanism, as with thermal waves, probably exists also in the photobaric effect (Bults et al., 1981, Poulet et al., 1983 Korpiun et al., 1992). The modulated flux of the newly formed oxygen generates concentration waves in the aqueous phase, which propagate to the surface, where the gas is released. In complete analogy to thermal waves, the concentration waves are damped exponentially as they propagate, with a characteristic distance equal to (over which the wave amplitude decreases by a factor e), where is the oxygen diffusion constant. Typical values for at common frequencies (around 30 Hz) is roughly near which is also the closest distance from the chloroplasts to the inner air phase in leaves. The transduction of oxygen evolution to increase of pressure must therefore be located only at the inner air phase of the leaf. Therefore, the photobaric signal is observed in leaves, also in lichens (Canaani et al., 1984), less prominently in half dried layered microalgae (Canaani, 1986) but not from in suspensions of photosynthetic material, water soaked sea weeds, and water infiltrated leaves. The name photoacoustic implies sound. The original photoacoustic effect, discovered in the last century (Bell, 1881 cited in Rosencwaig, 1980) involved indeed the production of audible sound by the absorption of periodically modulated light. However, in the usual low frequency PA experiments, the dimensions of the sample compartment are smaller than the sound wavelength so that the pressure variations are actually uniform. Nevertheless, the name photoacoustics (or optoacoustics) is used for all cases.

Shmuel Malkin A more recent experimental mode, which uses a pulsed laser beam transversing the sample, is capable of measurement of high frequencies and involves true acoustic waves. Light induced thermal and molecular volume changes, which occur in the small confined region of the laser beam path, act as a piston on the adjacent fluid body to generate an acoustic wave, which propagates to the wall of the cuvette and then sensed by a piezoelectric transducer, attached to the wall from the outside (Patel and Tam, 1981; Tam, 1986; Braslavsky and Heioff, 1989). The intensity of the resulting signal is directly related to the volume change, whether thermal or molecular. However, the shape of the signal is a complicated function involving both the relaxation kinetics in the sample, the sound propagation and the response properties of the sensor, and it is not a trivial matter to extract exact relaxation information from the signal (Puchenkov, 1994). Various photochemical and photobiological systems were investigated by this PA mode (Peters and Synders, 1988; Braslavsky and Heioff, 1989, Churio and Braslavsky, 1994; Nitsch et al., 1988, 1989). In summary, PA signals in photosynthetic samples involve: (1) photothermal conversion (2) molecular volume changes, and (3) gas evolution. These effects are often superposed and procedures are sought to separate each, for its different information content and as a general or specific marker for activity. This will be the essence of this chapter. Due to the limited space most of the older literature, except for some leading papers, is not included in the references list. These are listed as references in the newer literature and are reviewed elsewhere (cf. Moore, 1983; Malkin, 1986; Buschman and Prehn, 1990; N’SoukpoéKossi and Leblanc, 1990; Fork and Herbert, 1993; Braslavsky and Heioff, 1989; Malkin and Canaani, 1994). II. Experiments and Results with the Gasphase Coupled Microphone

A. Experimental Technicalities Work with gas-phase coupled microphone is most simple and common (Moore, 1983), requiring a

The photoacoustic method miniature microphone, communicating with a small volume of a gas phase around a tightly enclosed flat thin sample: a leaf disc, a lichen piece, an algal thallus, a filter paper impregnated with photosynthetically active suspension. The apparatus must be well insulated from ground vibrations. Background PA signals, arising from other cell parts, are minimized by correct light collimation and use of good polished window (preferably quartz). The usable frequency range is usually below about 1000 Hz, as the PA signal strongly decreases as f increases and there is increased proportion of background signals from non-photosynthetic pigments (e.g. from the leaf epidermis, also cell parts). Modulated light is obtained by using either a chopper, electronic modulation (e.g. with light emitting diodes (Snel et al., 1990; Kolbowsky et al., 1990) or using flash sources (Mauzerall, 1990). Additional nonmodulated (termed “background”) lights, which obviously do not produce signal of their own, are used to “bias” the photosynthetic system into any desired physiological or electron transfer state. With periodic modulation the microphone output is analyzed by a lock-in amplifier, yielding either the amplitude and phase, or the two amplitudes of the sinusoidal (“inphase”) and cosinusoidal (“quadrature”) components, which define the signal as a mathematical two-dimensional vector. The separate contributions to the signal must obviously combine as vectors (Poulet et al., 1983). With flash excitation a transient signal is obtained, which is studied as a function of time (Mauzerall, 1990; Kolbowsky et al., 1990; Reising and Schreiber, 1992; Cha and Mauzeral, 1993). Apparently, a leaf disc sample confined to a small isolated volume is unsuitable for studying its photosynthesis, because is very rapidly depleted during illumination. replenishment by continuous gas streaming through the cell produces too much background noise. This problem is somewhat circumvented by a construction which allows gas streaming between consecutive measurements (Reising and Schreiber, 1992) or by diffusion through sintered glass (Herbert and Fork, 1992) or through a small orifice (T. Punnet – private communication). Without such gas control, photosynthesis is running at a compensation point, at which photorespiration may be dominant. However, this does not make much

195 difference to the interpretation of the PA measurements, which are related to electron transport, as such. Also, it is possible to work with fully expanded leaf, and bring only a small central section into the PA cell. In this case, there is very probably an equilibration by lateral diffusion of the inner air composition between the exposed and unexposed parts, shown by the immediate effect of increased concentration in the gas phase above the exposed part on the PA signal (S. Malkin, unpublished), which acts to inverse the signal sense by adding an uptake component (Reising and Schreiber, 1992, cf. below).

B. The Photothermal Signal – Observations, Properties and Interpretation A pure photothermal signal is observed when the modulated gas evolution and molecular volume changes are negligible. The second is always insignificant with the gas-phase microphone. The first is usually negligible in sufficiently thick aqueous samples and in leaves at sufficiently high modulation frequencies, because of the damping of the oxygen concentration waves. Suitable samples for study of photothermal conversion are algal and subcellular suspensions, thalli of marine algae and water infiltrated leaves. Ordinary leaves and lichens must be used at sufficiently high frequencies (typically above ca. 300 Hz for a leaf and 50 Hz for a lichen, depending on the species). 1. Measurement of Energy Storage (ES) Figure 1 represents a schematic typical experimental record for the photothermal PA signal, excited by periodically modulated light. The energy storage is evaluated by comparing the signal to that of a reference where all light energy is promptly converted to heat. Since the PA signal is extremely sensitive to the thermal/optical parameters of the sample, its surface and the surrounding gas phase (Rosencwaig and Gersho, 1977), the reference must have exactly the same parameters. A physical substitution of the sample with a reference, still keeping exactly the same parameters, is impractical. However, a reference state is easily and reversibly achieved in photosynthetic samples by adding sufficiently strong background light, which saturates photosynthesis

196

and drives its quantum yield, close to zero. This results in a reversible increase of the PA photothermal signal from a level S to the reference level R. The relative energy storage (i.e. the average fraction of the absorbed photon energy (hv) stored as chemical energy), neglecting fluorescence changes, is defined by ES = (R – S)/R. The increase from S to R is typical of viable photosynthetic system and is used as a marker of activity. 2. ES Dependence on Experimental Parameters Energy strategy (ES), is a function of the modulation frequency f, of the wavelength of the average light intensity I and of the sample’s condition. To see this we may write approximately, neglecting fluorescence:

Shmuel Malkin

where R is reference level, is the energy content of the formed intermediate(s) per successful photochemical event, hv is the photon energy and c is the light speed. depends on f and possibly on depends on I and the sample’s condition.

a. Dependence on the Modulation Frequency f The modulated signal reflects heat which is prompt, relative to the modulation cycle time (1/f). Slower heat evolution is not modulated. Hence is assigned to intermediate(s) formed in a time much shorter than 1/f and decay in a time much longer. Also, S and R have the same phase angle in those ranges of f where these conditions are satisfied (e.g. as in Fig. 1). Between

The photoacoustic method

these frequency regions there are transition segments where S is phase shifted with respect to R, with an in-phase amplitude (relative to R) which grows with f and connects between adjacent plateaus, and a quadrature amplitude (relative to R) which is maximized at half transition points (Fig. 2). The frequencies at the half transition points are equal to where is the lifetime of the shorter lifetime intermediate. In the above analysis (cf. Malkin and Cahen, 1979) S is always measured relative to R, to eliminate the strong dependence on f caused by the heat conduction mechanism and the transduction to pressure.

b. ES as a Function of the Wavelength ES depends explicitly on (Eq. (1)), but may vary too. Assuming that is constant, is proportional to and a scan through gives a relative quantum yield spectrum (Fig. 3). This procedure has the obvious advantage that light intensity or absorption measurements are unnecessary. However, one should consider that may vary, due to the presence of more than one photochemical pathways (e.g. a combination of linear electron flow and PS I cyclic electron transport – the last one favored in far-red wavelengths). Indeed, a comparison between energy storage activity and linear electron transport activity, led to a conclusion about the presence of

197

an in-vivo cyclic PS I electron flow at far-red wavelengths (Herbert et al., 1990), which later was characterized in more detail (Ravenel et al., 1994).

c. ES as a Function of the Light Intensity I decreases as I increases, finally tending to zero at light saturation. Normally, it is desirable to perform ES measurements at sufficiently low light intensities where is maximum. To study ES vs. I, which is a relative measure of vs. I, it is most convenient to add background light of varying intensities. ES vs. I dependence conforms often to the customary hyperbolic relation between rate and light intensity (Fig. 4), allowing to extrapolate and find the idealized maximum ES at zero intensity (Carpentier et al., 1985, 1989; Owens et al., 1990) and also to deduce the saturation behavior of the reaction rate.

d. The Effect of Fluorescence – Determination of its Absolute Quantum Yield If fluorescence is not neglected, Eq. (1) takes a more extended form:

where

and

are the mean electromagnetic

198 frequencies of the (modulated) exciting light and the fluorescence emission, respectively, is the fluorescence yield of the measured sample. – at sufficiently low light intensities, where the fluorescence yield attains its minimum is the maximum fluorescence yield, obtained at light saturation. Eq. (2) results from the obvious proportionality between R and and between S and The ratio (R – S)/R expresses the relative deficit of the signal due to both photochemistry and fluorescence and was termed “photochemical loss” (abbreviated PL). Eq. (2) allows to calculate ES more rigorously as:

Shmuel Malkin When a good reference state is not available for the PA measurement (e.g. in the absence of photochemistry), an alternative analysis then considers a comparison between the change in PA, normalized for light absorption, vs. the wavelength (called “thermal deactivation spectrum”) and the relative fluorescence yield vs. the exciting wavelength. From the variations in these parameters it may be possible to adjust these spectra to an absolute quantum yield scale. This is useful in analyzing energy transfer yields for a multichromophore antenna preparations and was utilized to study photosynthetic systems oriented in streched polyvinylalcohol films. et al., 1986, 1991; Wróbel and Hendrich, 1989).

e. Time Domain Measurements Inserting accepted estimates for the fluorescence yields (cf. also below), the use of Eq. (1) instead of the more complicated Eq. (3) leads to a small error only (a few per cent) in ES (Malkin and Canaani, 1994). The absolute fluorescence yield can be estimated from a combined PA and relative fluorescence measurements when there is no photochemistry, if both change as a function of some parameter 1990; Dau and Hansen, 1990). One might utilize the transient change in the relative fluorescence measured at light saturation, known as “non-photochemical quenching”, occuring in whole oxygenic organisms, most prominently in leaves (Krause and Weis, 1991), which consist of a drop in from its maximal initial value, to a significantly lower value The PA signal is expected to increase correspondingly from to since the only competing reactions under light saturation are heat emission and fluorescence. Writing and (where a is a proportionality coefficient), it follows that:

can be computed from and the ratio of the relative fluorescence values corresponding to and Preliminary estimations are available for leaves. From the data in Dau and Hansen (1990) and

A microphone transient signal was obtained for layered Chlorella cells, following brief flash excitation (Cha and Mauzerall, 1992). The signal rise and decay transient reflects heat release kinetics, heat diffusion and sensor time response. The overall signal from a sample is smaller than that of a reference, indicating energy storage, but the time characteristics are broadly similar, indicating that the heat release kinetics is not discerned and that the following heat conduction process limits the time response. Thus, ES measurement at a particular transient signal time point (7 ms in this work) is probably related to an earlier time in the true heat release kinetics.

f. Some Numerical Estimates Most optimal in vivo measurements of ES, near the absorption edge (ca. 680 m), in leaves and algae, center around 0.3, showing a weak or no f dependence in the amenable range (10–500 Hz) with no or small phase angle shift between S and R (Malkin et al., 1992, cf. Fig. 1). This indicates that not much energy dissipation occurs between about 0.3 – 20 ms after the photoact and with the conventional estimate it turns out that kCal per mole electron transfer, nearly corresponding to the combustion heat of 1/24 mole glucose (ca. 28 kCal). The flash experiments of Cha and Mauzerall (1993, cf. above) identified ES (at less than 7 ms, cf. above) for different partial reactions, that could be iso-

The photoacoustic method

199

lated in different conditions: PS I cyclic electron flow = 15 kCal per mole electron transfered at 695 m, assuming PS II + PS I whole electron flow and PS II electron transport from the Mn complex to 3. Acoustic Signals Obtained with Magnetic Field Modulation (Magnetophotoacoustic Effect) Following light induced primary electron transfer in quinone depleted bacterial reaction centers, the triplet state of the primary donor is obtained by charge recombination from the charged donor acceptor pair, which, during its life time, can cycle between singlet and triplet states. In the presence of a magnetic field (in a range below 1 kGauss) the probability to obtain the triplet decreases, due to the energy splitting of the triplet sub-levels, of which two become inaccessible (Hoff, 1981). Using constant illumination and periodically modulated magnetic field, the triplet state concentration is modulated at twice the frequency of the field modulation (since the magnetic field effect is direction independent), which is picked up by a number of methods (Hoff et al., 1993). Using a suitably constructed PA cell to work at cryogenic temperatures, pressure modulation signals were recently detected from such quinone depleted bacterial reaction centers when these were exposed to modulated magnetic field and illuminated. It was proved, by a number of criteria, that this effect reflects the modulated heat release, which accompanies the return of the triplet to the ground state. Calibration of this “magneto photoacoustic” signal was done, by comparison to an ordinary PA measurement, demonstrating the ability to estimate the energy or yield or lifetime of the triplet state when the two others are known (Malkin and Hoff, 1994).

C. The Photobaric Signal – Observations, Properties and Interpretation 1. The Photothermal / Photobaric Composite Signal When oxygenic photosynthetic membranes lie sufficiently close to a gaseous phase, the photothermal signal is usually accompanied by the pho-

tobaric one. This mechanism had to be invoked to explain the “abnormal” behavior of the PA signal in leaves at low modulation frequencies (less than about 100 Hz), which becomes very prominent at the lowest available f (e.g. 10 Hz): when an intense background light was added there was, instead of the typical increase, characteristic of a photothermal signal, a rather marked decrease in the signal amplitude, accompanied by a large phase shift (Fig. 5; cf. also Bults et al., 1981; Poulet et al., 1983). Such effect indicated an additional contribution to the PA signal, which was eliminated at light saturation. The modulated photothermal signal (which is always present) could be measured separately, by photothermal

200

Shmuel Malkin

radiometry, monitoring the modulated part in the thermal infra-red radiation emitted from the leaf. This disclosed a normal behavior of the photothermal conversion (Kanstad et al., 1983). It was clear that the additional contribution to the PA signal is not photothermal. Its assignment to oxygen evolution is supported by: (1) its elimination at light saturation (no modulation in oxygen evolution occurs then); (2) the absolute requirement of a gaseous phase in close proximity to the photosynthetic membranes (Bults et al., 1981; Malkin et al., 1992), indicating the involvement of gas evolution into the inner air phase; (3) its strong damping as f increases, consistent with a diffusion mechanism (cf. below). (4) the parallel behavior of the photoboric signal and the directly measured oxygen evolution, with regard to phenomena such as photosynthetic induction (Bults et al., 1981), notably the initial oxygen gush (Malkin, 1987), wavelength dependence (e.g. the “far-red drop”; Bults et al., 1981) and Emerson enhancement with addition of background far-red light (Canaani and Malkin, 1984). 2. Separation of the Photobaric Signal In a crude approximation, the PA level in the presence of a strong saturating background light serves as a base-line for the photobaric signal. However, since the photothermal signal itself increases by the background light, the true baseline is usually lower (Fig. 5). Its position may be found by an independent estimate of the PL (e.g. from high frequency measurements; infrared photothermal measurements; measurements on the same leaf after water infiltration). However, it is possible to obtain the photobaric signal alone as one of its vectorial components, which is perpendicular to the photothermal signal vector. For this, the phase of the lock-in amplifier is adjusted in the presence of a strong saturating background light until one signal component (say, the quadrature) is nullified. No photothermal contribution occurs in the quadrature channel and the quadrature signal that appears upon switching off the background light is purely photobaric (Fig. 5). Experience showed that the phase of the photothermal contribution is stable, so that a single phase adjustment is usually sufficient even for lengthy (hours) experiments. This photobaric

component, perpendicular to the photothermal signal vector, may alternatively be computed for any arbitrary setting of the phase angle, using simple vector analysis. It is equal to: where subscripts Q and I represent quadrature and inphase components of the signal, S, and superscripts(+) and (–) represent the presence and absence of background light, respectively. Fig. 6 exemplifies the time domain PA responses to exciting flashes, in absence and presence of background light. A particular nice time

The photoacoustic method separation between photothermal and photobaric signals is seen with weak light pulses (Kolbowsky et al., 1990; Reising and Schreiber, 1994). In very short and strong flashes the photobaric signal transient is saturated, limited by the internal electron carriers pool, while the photothermal signal increases approximately linearly with the light intensity. Under such conditions. the photobaric signal becomes small, relative to the photothermal signal, and the time discrimination between them is more diffuse (Canaani et al., 1988; Mauzerall, 1990). Still, it was possible to discriminate the photobaric signal by proper substraction and observe its dependence on the flash number and the intensity of the last flash, in a series of consecutive saturating flashes, which opens a way to study S-state phenomena in the oxygen evolution from leaves (Canaani et al., 1988). 3. Quantitation of the Photobaric Signal

a. The Diffusion Model As a first approach to the quantitative treatment of the photobaric signal, it is necessary to oversimplify the complex anatomical structure of the leaf. Since the mesophyll cell has its chloroplasts arranged on the periphery, close to the boundary (at an average distance L to the inner air phase), the simplest model is that of a single sheet of a photosynthetic membrane in an aqueous phase, parallel to the air boundary, at a fixed distance L. L is much smaller than the membrane surface dimensions, allowing a one-dimensional diffusion model, where oxygen concentration waves propagate in a direction perpendicular to the boundary. With the above model, it was possible to derive a simple relation between the ratio O/T, where O and T are the amplitudes of the photobaric and the maximum photothermal signals, respectively, and the modulation frequency, f, based on the attenuation of the concentration waves, as they propagate to the surface. This relation takes the following linear form:

is the limiting value for O/T as The plot of Ln (O/T) vs. was indeed linear in

201 many cases (Malkin et al., 1992). In assessing the effects of stress on leaf photosynthesis it was found that not only O/T at a particular f changed by the treatment but possibly also the slope of the above plot, indicating changes in the anatomical structure of the leaf (Havaux et al., 1986, 1987). Therefore, such effects should be better quantified by which is diffusion independent. For completion, the diffusion model should include also possible sinks, where oxygen may be absorbed (by respiration and photorespiration) on its way to the inner air phase. Probably there is no place for such sinks in the small space between the photosynthetic membrane and the cell boundary. However, in at least one isolated case (maple tree leaves) this possibility was raised to explain the slow transient decrease of the photobaric signal with time, at low light intensity, observed uniquely for this species (Charland et al., 1992).

b.

O/T as a Measure of the Quantum Yield

If the entire PA signal is generated in the inner air phase, the contributions from each source point inside the leaf to the photothermal and photobaric signals have the same ratio. The photobaric signal (O) is proportional to and the maximum photothermal signal (T) is proportional to hvI. Their ratio, O/T, is proportional to and is therefore a relative measure of Its scan over gives a quantum yield spectrum (with no need to measure light intensity or absorption). Keeping small amplitudes of the light intensity modulation, changing the average light intensity by adding background light at various intensities, O/T may be considered as proportional to the derivative of the rate vs. intensity – which one may call a “differential” quantum yield. By integration of O/T vs. I curves the relative rate vs. intensity dependence is reconstructed (Poulet et al., 1983). An absolute calibration for the oxygen evolution signal was suggested, but not yet implemented, by measuring the relative oxygen gush signal which occurs as a pulse in the beginning of the induction period, and comparing to the size of the plastoquinone pool (Malkin, 1987; Malkin and Canaani, 1994).

202 4. The Limiting Rate in the Oxygen Evolution Process The photobaric signal should be influenced also by the rates of the partial processes which lead to oxygen evolution, leading expectedly to a deviation from the linear plot of Eq. (5), at modulation frequencies comparable or higher than the rate limiting constant, k. While earlier work indicated such a deviation, corresponding to a reaction time of about a ms (Poulet et al., 1983; Canaani et al., 1984), a careful analysis indicated that this deviation was probably an artefact, caused by underestimation of the PL in the photothermal signal (Malkin et al., 1992). Thus, the limiting reaction time may even be smaller than a ms. This conflicts with conclusions from recent oxygen electrode measurements at low electrode polarization (Plijter et al., 1988), concluding that the reaction time is very much longer (cf. also H.J. van Gorkom, this volume). Since PA signals monitor the gas evolution directly, by the pressure variation, they must be regarded as very reliable informant compared to the oxygen electrode, which depends on the process of oxygen reduction by the polarized cathode, and is sensitive to imperfections in its surface. Time domain PA signal profiles (Mauzerall, 1990; Kolbowsky et al., 1990; Reising and Schreiber, 1994 – cf. Fig. 6) show even more clearly that oxygen evolution does not delay more than about a ms, probably just reflecting diffusion lag. 5. Uptake Signals While there is an increasing phase angle shift between the photobaric and the photothermal signals as f increases (Poulet et al., 1984) the two phase angles tend to approach each other as There are situations where at low frequencies the direction of the photobaric signal tends to be 180° out of phase, (i.e. with a negative sense, compared to the photothermal modulation), indicating uptake rather than evolution. In a time domain measurement this was seen clearly as a negatively going transient, when the direction of the photothermal signal is defined as positive. Uptake signals were seen after heat shock (Havaux et al., 1987b), after dark adaptation and during photosynthetic induction (Malkin, 1987;

Shmuel Malkin Mauzerall, 1990; Reising and Schreiber, 1992). Particularly noticeable are uptake signals obtained upon increase of the level (Reising and Schreiber, 1992). The simplest interpretation is that the uptake reflects oxygen photoreduction through PS I, as it is also induced by modulated far-red light, which normally does not produce a modulated oxygen evolution signal. However, Reising and Schreiber (1994) concluded, from the effect of a Carbonic anhydrase inhibitor, that at least the uptake signal seen at elevated levels arises from modulated solubilization of caused by the light induced modulation in the pH of the stroma. Accordingly, the uptake seen after dark adaptation was explained by the dark accumulation (by respiration) in the confined sample compartment (Reising and Schreiber, 1992). III. Experiments and Results in the Time Domain with a Sample Coupled Piezoelectric Sensor This experimental mode, involving laser excitation, was reviewed in detail (Tam, 1986; Braslavsky and Heioff, 1989) and is sometimes referred to as LIOAS (laser induced optoacoustic spectroscopy). As was mentioned in the Introduction, the signal time profile, (Fig. 7) is complex, requiring special methods of its diagnostics, hence for routine measurements only the amplitude of the first peak (or the difference between the first and second peaks) is often tentatively measured, and taken to indicate the heat release in a time scale shorter than the time scale of the appearance of the signal peaks (about a Measurements are performed in bulk aqueous suspensions (sometimes with the addition of glycerol, to increase the thermal expansion coefficient), through which a laser pulse is sent. Referencing for maximum heat production is here possible by replacing the sample with a photochemically inert reference, dissolved in the same buffer, keeping the same light absorption coefficient. The thermal and elastic properties are kept the same, being mainly due to the aqueous medium. Initial results were collected for PS I and PS II particles (Nitsch et al., 1988), purple bacterial cells (Nitsch et al., 1989) and a leaf submerged in water (Jabben and Schaffner, 1985). In these,

The photoacoustic method

203 mental proportionality coefficient. Eq. (6) can be used in a variety of ways. For example, an inert reference solution will yield at a given temperature a reference signal hence:

(subscript T1 refers to values of the parameters at is the prompt maximum heat, equal to the absorbed laser pulse energy). Inserting the values for 4°C and we obtain

no consideration was made on the presence of molecular volume changes, which were found in later experiments, and the signal was totally ascribed to photothermal conversion. Although the significance of the numbers obtained for energy storage is therefore indecisive, it was later argued (Malkin et al., 1994), that the use of 30% glycerol in the medium, which increases the thermal expansion coefficient significantly, makes the molecular volume changes less significant, relative to the photothermal ones. Recent experiments with purple bacterial reaction centers (Malkin et al., 1994), and with particles of PS I and PS II, chloroplasts and algal cells (Delosme et al., 1994) disclosed indeed significant molecular volume changes, observed separately at ca. 4°C (Fig. 7). The signal amplitude, S, at any other temperature is a composite of photothermal and molecular volume changes. Thus:

where C and are the thermal expansion coefficient, specific heat and specific density of the sample, respectively, the heat released and the molecular volume change. a is an instru-

If and are temperature independent (at constant laser energy) S/R would be a linear function of with a slope from which the energy storage can be found, and an intercept from which is calculated. Such plot was found to be indeed linear, which supports the above assumptions (Malkin et al., 1994). The results indicated significant energy storage, assigned to the state in contrast to the previous slower (ms) PA measurement, which yielded a very small energy storage, from which it was concluded that the free energy change was mainly entropic (Callis et al., 1972; Arata and Parson, 1981). Arata and Parson expressed their concern that something was wrong with their PA results, but could not trace the real reason. Perhaps there is a hidden relaxation process in a time scale between and ms, involving the entire protein moiety, which converts enthalpic free energy into entropic free energy. So far, this dilemma still waits for an answer. Delosme et al. (1994) used LIOAS measurements for spectral measurements of the quantum yield in oxygenic organisms, in which significant molecular volume changes were also observed and were used to calculate relative quantum yields. The quantum yield varied as a function of the wavelength in chloroplasts, but not in PS I or PS II particles. The quantum yield changes reflect probably the variations of the light distribution between the two photosystems.

204

IV. Applications to Physiological Studies PA measurements are particularly suited to get physiological information, being non invasive, sensitive, and with the capacity to reach various physiological states by background lights (Malkin and Canaani, 1994; Fork and Herbert, 1993). Examples are: quantitative information on light distribution between the two photosystems and its changes in different states (Canaani and its Malkin, 1984; Canaani, 1986; Malkin et al., 1990; Fork et al., 1991; Veeranjaneyulu et al., 1991a,b; Mullinaux et al., 1991; Bruce and Saleihan, 1992); the relation between the quantum yield of photochemistry and non-photochemical quenching of fluorescence (Snel et al., 1990a). Characterization of cyclic electron transport, by energy storage activity (Malkin et al., 1990; Herbert et al., 1991; Ravenel et al., 1994); characterization of photosynthesis induction (Malkin, 1987; Havaux 1988; Snel et al., 1990b; Reising and Schreiber, 1992); characterization of stress effects (Havaux et al., 1986, 1987a,b; Havaux, 1988). For a more complete list, see Fork and Herbert, 1993; Malkin and Canaani, 1994). The advantage of the PA measurements, for this analysis, is in that they are fast and that they measure true (gross) photosynthesis, compared to the customary measurements of net photosynthesis, which include all losses. In future work, particularly in PA work with leaves, it would be advantageous to take the example of Snel et al. (1992) and develop a PA system that can simultaneously measure at least at two frequencies and two wavelengths (by providing two measuring light channels and two lockin amplifiers), allowing simultaneous measurements of energy storage and oxygen evolution (low and high frequencies) and of separate measurement of PS I activity (far-red vs. shortwavelength modulated lights). This is necessary in view of the impossibility to repeat exactly the same experiment in a leaf. References Arata H and Parson WW (1981) Enthalpy and volume changes accompanying electron transfer from P-870 to quinones in Rps. spheroides reaction centers. Biochim Biophys Acta 636: 70–81.

Shmuel Malkin Braslavsky SE and Heihoff K (1989) Photothermal methods. In: Scaiano JC (Ed) CRC Handbook of Organic Photochemistry. Vol 1, pp 327–355, CRC Press, Boca Raton. Bruce D and Salehian O (1992) The efficiency of primary photosynthetic processes in state 1 and state 2. Biochim Biophys Acta 1100: 242–250. Bults G, Horwitz BA, Malkin S and Cahen D (1981) Photoacoustic measurements of photosynthetic activities in whole leaves – photochemistry and gas exchange. Biochim Biophys Acta 679: 452–465. Buschmann C and Prehn H (1990) Photoacoustic spectroscopy – photoacoustic and photothermal effects. In: Linskens H-F and Jackson JF (eds) Modern Methods in Plant Analysis Vol 11, pp 148–180, Springer-Verlag, Berlin. Callis JB, Gouterman M and Danielson JDS (1969) Flash calorimeter for measuring triplet yields. Rev Sci Instr 40: 1599–1605. Callis JB, Parson WW and Gouterman M (1972) Fast changes of enthalpy and volume on flash excitation of chromatium chromatophores. Biochim Biophys Acta 267: 348–362. Canaani O (1986) Photoacoustic detection of oxygen evolution and state 1 – state 2 transitions in cyanobacteria. Biochim Biophys Acta 852: 74–80. Canaani O and Malkin S (1984) Distribution of light excitation in an intact leaf between the two photosystems of photosynthesis – changes in absorption cross-sections following state 1 – state 2 transitions. Biochim Biophys Acta 766: 513–524. Canaani O, Ronen R, Garty J, Cahen D, Malkin, S and Galun M (1984) Photoacoustic study of the green alga Trebouxia in the lichen Ramalina duriaei in-vivo. Photosynth Res 5: 297–306. Canaani O, Malkin, S and Mauzerall D (1988) Pulsed photoacoustic detection of flash-induced oxygen evolution from intact leaves and its oscillation. Proc Nat Acad Sci (US) 85: 4725–29. Carpentier R, Nakatani H and Leblanc RM (1985) Photoacoustic detection of energy conversion in a Photosystem II submembrane preparation from spinach. Biochim Biophys Acta 808: 470–473. Carpentier R, Leblanc RM and Mimeault M (1989) Photoacoustic detection of photosynthetic energy storage in Photosystem II submembrane fractions. Biochim Biophys Acta 975: 370–376. Cha Y and Mauzerall D (1992) Energy storage of linear and cyclic electron flows in photosynthesis. Plant Physiol 100: 1869–1877. Charland M, Veeranjaneyulu K, Charlebois D and Leblanc RM (1992) Photoacoustic signal generation in leaves: Are O2–consuming processes involved? Biochim Biophys Acta 1098: 261–265. Churio MS, Angermund KP and Braslavsky SE (1994) Combination of laser induced optoacoustic spectroscopy and semiempirical calculations for the determination of molecular volume changes. The photoisomerization of carbocyamines. J Phys Chem 96: 1776–1782. Dau H and Hansen U-P (1990) A study on the energy-dependent quenching of chlorophyll fluorescence by means of photoacoustic measurements. Photosynth Res 25: 269– 278.

The photoacoustic method Delosme R, Béal D and Joliot P (1994) Photoacoustic detection of flash-induced charge separation in photosynthetic systems. Spectral dependence of the quantum yield. Biochim Biophys Acta 1185: 56–64. Fork DC and Herbert SK (1991) A gas-permeable photoacoustic cell. Photosynth Res 27: 151–156. Fork DC and Herbert SK (1993) The application of photoacoustic techniques to studies of photosynthesis. Photochem Photobiol 57: 207–220. Fork DC, Herbert SK and Malkin S (1991) Light energy distribution in the brown alga Macrocystis Pyrifera (Giant kelp). Plant Physiol 95: 731–739. D (1990) Joint applications of fluorescence and photoacoustic methods in photobiology. Appl Fluor Techn 2: 11–14. D, Erochina LG, Balter A, Lorrain L, Szurkowski J and Szych B (1986) Polarized absorption, fluorescence and photoacoustic spectra of phycobilisomes embedded in poly(vinylalcohol) films. Biochim Biophys Acta 851: 173– 180. D, Cegielski R and Abdurakhmanov IA (1991) Excitation energy transfer from carotenoids to bacteriochlorophyll in the bacterium Chromatium minutissimum. Photosynthetica 25: 621–629. Havaux M (1988) Induction of photosynthesis in intact leaves under normal and stressing conditions followed simultaneously by transients in chlorophyll fluorescence and photoacoustically monitored evolution. Plant Physiol Biochem 26: 695–704. Havaux M, Canaani O and Malkin S (1986) Photosynthetic responses of leaves to water stress, expressed by photoacoustics and related methods I. Probing the photoacoustic method as an indicator for water stress in vivo. II. The effect of rapid drought on the electron transport and the relative activities of the two photosystems. Plant Physiol 82: 827–839. Havaux M, Canaani O and Malkin S (1987a) Inhibition of photosynthetic activities under slow water stress measured in vivo by the photoacoustic method. Physiol Plantarum 70: 503–510. Havaux M, Canaani O and Malkin S (1987b) Oxygen uptake by tobacco leaves after heat shock. Plant Cell Environ 10: 677–683. Herbert SK, Fork DC and Malkin S (1990) Photoacoustic measurements in-vivo of energy storage by cyclic electron flow in algae and higher plants. Plant Physiol 94: 926–934. Hoff AJ (1981) Magnetic field effects on photosynthetic reactions. Quart Rev Biophys 14: 599–665. Hoff AJ, Gast P, van der Vos R, Vrieze J, Franken EM and Lous EY (1993) Magnetic field effects: MARY, MIMS and MODS. Z Physik Chem 180: 175–192. Hunter TF and Stock MG (1974) Photophysical processes in the vapour-phase measured by the optic-acoustic effect. Part 2 – Triplet state yield and life time in high pressure acetyl vapour. Part 3 – Radiactionless processes from the lowest singlet and triplet states of and benzene. J Chem Soc Faraday Trans II 70: 1022–1039. Hunter TF, Rumbles D and Stock MG (1974) Photophysical processes in the vapour-phase measured by the optic-acoustic effect. Part 1 – The model and apparatus for the study

205 of rediactionless processes. J Chem Soc Faraday Trans II 70: 1010–1021. Jabben M and Schaffner K (1985) Pulsed laser induced optoacoustic spectroscopy of intact leaves. Biochim Biophys Acta 809: 445–451. Kanstad SO, Cahen D and Malkin S (1983) Simultaneous detection of photosynthetic energy storage and oxygen evolution in leaves by photothermal radiometry and photoacoustics. Biochim Biophys Acta 722:182–189. Kolbowski J, Reising H and Schreiber U (1990) Computercontrolled pulse modulation system for analysis of photoacoustic signals in the time domain. Photosynth Res 25: 309– 316. Korpiun P, Malkin S and Cahen D (1992) Interpretation of the photoacoustic effect in leaves by evolution and transport of heat and oxygen. In: Bicanic D (Ed) Photoacoustic and photothermal phenomena, pp 59–61, Springer-Verlag, Berlin. Krause GH and Weis E (1991) Chlorophyll fluorescence and photosynthesis: the basics. Ann Rev Plant Physiol Plant Mol Biol 42: 413–449. Malkin S (1986) Photoacoustic probing of energy storage and gas exchange in the photosynthesis of leaves. J Chem Soc Farad Trans II 82: 2233–2235. Malkin S (1987) Fast photoacoustic transients from dark adapted intact leaves — I. Oxygen evolution and uptake pulses. A phenomenological record. Planta 171: 65–72. Malkin S and Cahen D (1979) Photoacoustic spectroscopy and radiant energy conversion: Theory of the effect with special emphasis on photosynthesis. Photochem Photobiol 29: 803–813. Malkin S and Canaani O (1994) The photoacoustic method and its characteristics in use for the study of photosynthesis. Ann Rev Plant Physiol Plant Mol Biol 45: 493–526. Malkin S and Hoff AJ (1994) Acoustic detection of triplet states formed by radical pair recombination in quinonedepleted photosynthetic reaction centers, by magnetic field modulation (magnetophotoacoustic effect) – a feasibility study. Photochem Photobiol 59: 670–676. Malkin S, Herbert SK and Fork DC (1990) Light distribution, transfer and utilization in the marine red alga Porphyra perforata from photoacoustic energy storage measurement. Biochim Biophys Acta 1016: 177–189. Malkin S, Charland M and Leblanc RM (1992) A photoacoustic study of water infiltrated leaves. Photosynth Res 33: 37–50. Malkin S, Churio M, Shochat S and Braslavsky S (1994) Photochmical energy storage and volume changes in the microsecond time range in bacterial photosynthesis – a laser induced optoacoustic study. J Photochem Photobiol B 23: 79–86. Mauzerall D (1990) Determination of oxygen emission and uptake in leaves by pulsed, time resolved, photoacoustics. Plant Physiol 94: 278–283. Moore TA (1983) Photoacoustic spectroscopy and related techniques applied to biological materials. In: Smith KC (ed) Photochemical and photobiological reviews Vol 7 (pp 187–221) Plenum, New York. Mullineaux CW, Griebenow S and Braslavsky SE (1991) Photosynthetic energy storage in cyanobacterial cells adapted

206 to light-states 1 and 2. A laser-induced optoacoustic study. Biochim Biophys Acta 1060: 315–318. Nitsch C, Braslavsky SE and Schatz GH (1988) Laser-induced optoacoustic calorimetry of primary processes in isolated photosystem I and photosystem II particles. Biochim Biophys Acta 934: 201–212. Nitsch C, Schatz GH and Braslavsky SE (1989) Laser-induced optoacoustic calorimetry of primary processes in cells of Rhodospirillum rubrum. Biochim Biophys Acta 975: 88– 95. N’Soukpoé-Kossi CN and Leblanc RM (1990) Application of photoacoustic spectroscopy in photosynthesis research. J Mol Structure 217:69–84. Ort DR and Parson WW (1979a) Flash-induced volume changes of bacteriorhodopsin. J Biol Chem 253:6158–6164. Ort DR and Parson WW (1979b) The quantum yield of flashinduced proton release by bacteriorhodopsin-containing membrane fragments. Biophys J 25: 341–354. Ort DR and Parson WW (1979c) Enthalpy changes during the photochemical cycle of bacteriorhodopsin. Biophys J 25: 355–364. Owens TG, Carpentier R and Leblanc RM (1990) Detection of photosynthetic energy storage in a photosystem I reaction center preparation by photoacoustic spectroscopy. Photosyn Res 24: 201–208. Patel CKN and Tam AC (1981) Pulsed optoacoustic spectroscopy of condesned matter. Rev. Mod Phys 53: 517–550. Peters KS and Synder GJ (1988) Time resolved photoacoustic calorimetry: probing the energetics and dynamics of fast chemical and biochemical reactions. Science 241: 1053– 1057. Plijter JJ, Aalbers SE, Barends JPF, Vos MH and Van Gorkom HJ (1988) Oxygen release may limit the rate of photosynthetic electron transport: the use of a weakly polarized cathode. Biochim Biophys Acta 935: 299–311. Poulet P, Cahen D and Malkin S (1983) Photoacoustic detection of photosynthetic oxygen evolution from leaves – Quantitative analysis by phase and amplitude measurements. Biochim Biophys Acta 724: 433–446. Puchenkov O (1994) Restoration of laser-induced process kinetics from photoacoustic measurements. Proc 8th Int Topical Meeting on Photoacoustic and Photothermal Phenomena pp 161–162.

Shmuel Malkin Ravenel J, Peltier G and Havaux M (1994) The cyclic electron pathways around photosystem I in Chlamydomonas reinhardtii as determined in vivo by photoacoustic measurements of energy storage. Planta 193:251–259. Reising H and Schreiber U (1992) Pulse–modulated photoacoustic measurements reveal strong gas-uptake component at high concentration. Photosynth Res 31:227–238. Reising H and Schreiber U (1994) Inhibition by ethoxyzolamide of a photoacoustic gas uptake: Evidence for carbonic anhydrase catalyzed Photosyn Res 42: 65–73. Rosencwaig A (1980) Photoacoustics and photoacoustic spectroscopy. Wiley, New York. Rosencwaig A and Gersho A (1977) Theory of the photoacoustic effect with solids. J Appl Phys 47: 64–69. Snel JFH, van Leperen W and Vredenberg WJ (1990a). Complete suppression of oxygen evolution in open PS 2 centers by non-photochemical fluorescence quenching? In: Baltschevsky M (ed) Current Research in Photosynthesis Vol 2, pp 911–914, Kluwer Academic Publishers, Dordrecht. Snel JFH, Kooijman M and Vredenberg WJ (1990b) Correlation between chlorophyll fluorescence and photoacoustic signal transients in spinach leaves. Photosyn Res 25: 259– 268. Snel JFH, Polm MW, Buurmeijer WF and Vredenberg WJ (1992) Deconvolution of photobaric and photothermal signals from spinach leaves. In: Bicanic D (ed) Photoacoustic and photothermal phenomena III. Springer series in optical sciences. Vol 69, pp 65–68, Springer-Verlag, Heidelberg. Tam A C (1986) Applications of photoacoustic sensing techniques. Rev Mod Phys 58: 381–431. Veeranjaneyulu K, Charland M, Charlebois DCN and Leblanc RM (1991a) Photosynthetic energy storage of photosystems I and II in the spectral range of photosynthetically active radiation in intact sugar maple leaves. Photosynth Res 30: 131–138. Veeranjaneyulu K, Charland M, Charlebois DCN and Leblanc RM (1991b) Photoacoustic study of changes in energy storage of PS I and PS II during state 1 – state 2 transition. Plant Physiol 97: 330–334. Wróbel D and Hendrich W (1989) Thermal deactivation and energy transfer in isolated photosystem II and light-harvesting complexes in polyvinyl alcohol film. J Photochem Photobiol B 3: 319–332

PART TWO

Magnetic resonance

This page intentionally left blank

Chapter 13 Magnetic Resonance: An Introduction Arnold J. Hoff Department of Biophysics, Huygens Laboratory, Leiden University, P.O. Box 9504, 2300 RA Leiden, The Netherlands

Abbreviations: CP-MAS – cross polarization-magic angle spinning; cw – continuous wave; ENDOR – electron-nuclear double resonance; EPR – electron paramagnetic resonance; ESE – electron spin echo; ESP – electron spin polarization; hfi – hyperfine interactions; NMR –nuclear magnetic resonance; ODMR – optically detected magnetic resonance; RC – reaction center

Photosynthetic charge separation results in two ionized cofactors: the primary donor cation and the primary acceptor anion Both are radicals possessing an unpaired spin, which are subsequently stabilized by dark electron transport. The stabilized radicals can be studied by electron paramagnetic resonance, EPR, and its multiple resonance extensions, as electron-nuclear double and triple resonance, ENDOR and TRIPLE. With EPR one looks at the valence electrons, in contrast to X-ray diffraction, which probes the core electrons. Thus, with EPR information can be obtained on the electronic wavefunctions directly involved in the electron transport properties of the cofactor system. The major sources of information offered by cw EPR spectra of doublet states (S = 1/2) are the g-factor (or g-matrix for anisotropic systems) and the hyperfine interactions (hfi). The former is an integral property of the radical, probing the average valence electron density distribution, and serves as a fingerprint for the identification of paramagnetic entities. The latter is a local probe that reflects the density distribution of the unpaired spin at individual nuclei of the radical and its immediate environment. The EPR signals of photosynthetic material are usually too little resolved to permit analysis of hyperfine splittings

and a broad, unstructured line shape results. This inhomogeneous broadening makes it difficult or impossible to measure the hfi directly. Here, multiple resonance techniques are necessary, and have been widely employed, first in frozen solution at cryogenic temperatures, later predominantly in liquid solution. In the past few years, single crystals (the dream of every EPR spectroscopist!) of bacterial RCs have become available, giving new life to solid state ENDOR. In the past, EPR of doublet states has contributed to the identification of numerous reactants in primary electron transport in bacterial and plant photosystems. This application is still important, especially in the study of the acceptors in lesser known bacterial systems, such as the green photosynthetic bacteria. There is a trend to enhance resolution by working at higher frequency, and we may expect mm EPR spectroscopy (at 95, 135 or even 230 GHz) to become increasingly more important. In addition to the above-described radical doublet states, a paramagnetic excited triplet state may be generated in RCs in which forward electron transport is blocked, e.g. by prereduction or removal of the secondary acceptor. The photoinduced radical pair state then must decay via recombination, which at cryogenic temperatures may yield for almost 100% the triplet state of the primary donor, EPR and optically detected magnetic resonance, ODMR, and

Correspondence: Fax 31-71-5275819; E-mail: [email protected]

209 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 209–210. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

210 ENDOR of this triplet state gives information on the molecular structure of the primary donor and its environment. When photolytically produced radicals are studied by EPR one often finds that the relative population of the spin levels is not in Boltzmann equilibrium. Thus, some hyperfine lines or parts of an inhomogeneously broadened line are in emission, or in enhanced absorption. Often both these effects are seen in a single EPR spectrum. This phenomenon is called Electron Spin Polarization, ESP. It is best seen employing time-resolved EPR techniques, but in some cases the lifetime of the polarization is sufficiently long to make it possible to observe ESP spectra using cw EPR. Analysis of the ESP spectrum gives information on the magnetic interactions between the cofactors, which is difficult to obtain in any other way. These interactions in turn report on the electronic interactions responsible for electron transport. In recent years two variants of time-resolved EPR have come to the fore: direct-detection EPR with a time resolution now approaching a few tens of ns, and pulsed EPR, or Electron Spin Echo (ESE) spectroscopy, with a time resolution of a few hundred ns. ESE spectroscopy offers in addition the possibility to determine small hyperfine and dipolar couplings and, in favorable cases, quadrupole couplings of and nuclei, through the so-called Electron Spin Echo Envelope Modulation (ESEEM) phenomenon. With the advent of commercially available instrumentation pulsed EPR is now within reach of a grow-

Arnold J. Hoff ing number of laboratories, and finds increasingly wider application in biophysical research in photosynthesis. Until recently applications of magnetic resonance concerned almost exclusively EPR and its offshoots, as ENDOR, ODMR and reactionyield detected magnetic resonance (RYDMR), with incidental use of NMR for the measurement of relaxation times. With the advent of selective spin-isotope labeling techniques, however, it is now also possible to carry out high resolution, solid state NMR of reaction centers. The application of cross polarization magic angle spinning (CP-MAS) solid state NMR to frozen bacterial RCs enriched selectively in and amino acids or cofactors promises to complement and extend structure determination with X-ray analysis. With the above-described array of magnetic resonance spectroscopies an impressive body of data and new insights in the fundamental mechanism of photosynthetic energy conversion has been accumulated, and magnetic resonance now rivals optical spectroscopy as a tool for penetrating the structure-function relationships of the photosynthetic reaction center. In the following chapters (14–18) compact treatments are presented of time-resolved EPR (Levanon), pulsed EPR (Britt), ENDOR (Lubitz), ODMR (Hoff), and solid-state NMR (De Groot). The presentations succintly cover theory and instrumentation, whereas the applications to research in photosynthesis are high-lighted with some recent results.

Chapter 14 Time-Resolved Electron Paramagnetic Resonance Spectroscopy – Principles and Applications Haim Levanon Department of Physical Chemistry and the Farkas Center for Light Induced Processes, the Hebrew University of Jerusalem, Jerusalem 91904, Israel

Summary I. Introduction II. Experimental A. TREPR Spectroscopy: General B. Methods of TREPR Detection 1. Light Modulation-Field Modulation (LFM) 2. Continuous Wave Direct-Detection (DDEPR) 3. Pulsed Microwave Detection, Fourier Transform EPR (FTEPR) 4. Comparison between Two TREPR Methods: DDEPR vs. FTEPR III. Results A. Introduction B. Triplet Detection in Anisotropic: Solid and Fluid Phases 1. Introduction 2. Liquid Crystalline Hosts 3. Examples a. Mesogenic Residues Attached to Porphyrins b. Sign Determination of the ZFS Parameter and Triplet Dynamics of Nonplanar Porphyrins C. Electron Transfer 1. Introduction 2. Examples a. Donors and Acceptors with Coulombic Interaction b. Multistep Electron Transfer in Covalently-Linked Assemblies c. Mixed Triplet- and Singlet-Initiated Intramolecular ET in LCs d. Why Liquid Crystals Attenuate Electron Transfer Rates e. Quantum Beats in Correlated Radical Pairs IV. Concluding Remarks Acknowledgements References

211 212 213 213 215 215 215 215 217 218 218 218 218 218 220 220 220 222 222 224 224 225 226 228 228 229 229 230

Summary The complexity of natural photosynthesis has prompted intensive studies of simpler model systems that might reproduce essential features of the in vivo apparatus. This is the so-called mechanistic approach, intended to solve some of the pertinent problems still open in primary photosynthesis. In this survey we concentrate on the application of time-resolved electron paramagnetic resonance (TREPR) Spectroscopy in studying photoinduced processes in model systems. The model systems consist of expanded electron systems as donors, and of covalently as well as non-covalently linked (electrostatically bound) Correspondence: Fax: 972-2-618033; E-mail: [email protected]

211 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 211–233. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

212

Haim Levanon

donor–acceptor systems. The required details of the electronic structure and interactions can be determined by advanced TREPR (and related spectroscopies described in this book) which combines high spectral resolution and time resolution in the 10 –100 ns domain. One of the most important features in TREPR is the electron spin polarization (ESP) effects associated with the reaction products of photoexcited donor–acceptor systems. Therefore, by controlling the environmental conditions, such as solvent polarity, viscosity, temperature, and medium anisotropy, the ESP pattern of such reactions is quite unambiguous. For example, the different line shape patterns reflect the variation of the molecular architecture, namely the relative orientation of the donor–acceptor and the nature of the spacer. The differences in molecular structures are manifested by the TREPR spectra through the magnitude of the spin–spin coupling (J) and the dipolar interaction (D), thus leading to different electron spin polarization mechanisms. In that respect, TREPR has a clear advantage over optical spectroscopy, which lacks adequate energy resolution. It is shown that TREPR applied to donor–acceptor systems, in isotropic and anisotropic liquid crystal (LC) media, allows a better understanding of the role of the microenvironment in long-range ET, a subject directly related to the in vivo protein–chromophore interaction. Abbreviations: CIDEP – chemically induced dynamic electron polarization; CRP – correlated radical pair; CW – continuous wave; DDEPR – direct detection EPR; diff – diffusion; DsA – donor–spacer– acceptor; EPR – electron paramagnetic resonance; ESP – electron spin polarization; ET – electron transfer; EnT – energy transfer; FTEPR – fourier transform EPR; ISC – intersystem crossing; LCs – liquid crystals; LFM – light modulation-field modulation; PSI – photosystem I; PcQ – porphyrincyclohexylene-quinone; PpQ – porphyrin-phenyl-quinone; RPM – radical pair mechanism; RCs – reaction centers; SO – spin orbit; THF – tetrahydrofuran; TM – triplet mechanism; TREPR – timeresolved EPR; ZQC – zero-quantum beats I. Introduction Fifty years have passed since Zavoisky (1945) came out with his first report on electron paramagnetic resonance (EPR) detection “of liquid solutions” (Zavoisky, 1945a, 1945b). Nevertheless, as Zavoisky pointed out, the first discovery of paramagnetic absorption in “solid solutions” (doped crystals of iron and chromium) by calorimetric methods was made by Gorter in 1936 (Gorter, 1936a, 1936b). In the past 50 years, EPR spectroscopy has been rapidly developing with emerging sub-fields of interest and thousands of publications. Obviously, this chapter is not intended to cover all areas of research associated with EPR spectroscopy. Here, we shall cover briefly the main features of EPR spectroscopy, focussing on the basic principles of time-resolved EPR (TREPR) and its application in photoinduced processes, of which photosynthesis is the most celebrated one. Other interdisciplinary areas are covered separately in this book and in recent reviews (Trifunac et al., 1986; Bixon et al., 1992).

Many photophysical and photochemical reactions in solids and liquids, which involve paramagnetic intermediates, exhibit electron spin polarization (ESP) effects, indicating deviation from thermal spin equilibrium. Departure from thermal equilibrium may be due to symmetry-driven molecular selection rules, such as selective singlet–triplet intersystem crossing (ISC) or other spin-state mixing mechanisms. Sufficiently high time-resolution, compared to spin relaxation rates, allows for the observation of ESP effects in three different modes: 1) Photoexcited triplet detection and ESP of chromophores (D) (Levanon and Norris, 1978, 1982; Budil and Thurnauer, 1991).

2) Doublet detection and ESP of uncorrelated donor (D) and acceptor (A) pairs, that are photolytically produced by the respective processes (Eqs. (2) and (3)), for singlet and triplet precursors. Chemically induced dynamic electron polarization (CIDEP) in terms of radical

Time-resolved EPR pair mechanism (RPM) or triplet mechanism (TM) is often associated with diffusing radicals (Salikhov et al., 1984; Trifunac et al., 1986).

or

3) Detection of correlated radical-ion pairs was first reported in photosynthetic RCs (Thurnauer and Norris, 1980), and was treated quantitatively in later studies (Buckley et al., 1987; Closs et al., 1987; Hore, 1989; Stehlik et al., 1989; Morris et al., 1990). Correlated radical pair (CRP) polarization, observed in radical pairs held together in micelles (DA) (Closs et al., 1987), or by Coulombic interactions (Hugerat et al., 1991; Zilber et al., 1992; Rozenshtein et al., 1993), or by a spacer s at a fixed distance, (DsA). Similar to the previous case, CRP can also be formed either via a singlet or a triplet precursor:

or

213

the spin-correlated biradical state, representing coupled donor–acceptor pairs (Eqs. (4), (5)). The chemical assemblies (representative structures are shown in Fig. 1), which are engaged in photoinduced energy- and electron-transfer (ET and EnT, respectively) reactions and which are of interest to model photosynthesis, can be grouped as: 1) traditional porphyrin and chlorophyll systems; 2) porphyrin-quinone systems (Fig. 1, 1– 3), which are linked together either covalently via a spacer, DsA (Bixon et al., 1992; Wasielewski, 1992), or via hydrogen bonds (Fig. 1, 4) (Sessler et al., 1993) or electrostatically bound porphyrins (Hugerat et al., 1991); 3) nonconventional porphyrinoids, e.g., the expanded porphyrins, such as sapphyrin (Bauer et al., 1983), the texaphyrins (Sessler et al., 1988, 1989), the porphycenes and stretched porphycenes (Fig. 1, 5) (Berman et al., 1993); and 4) non-porphyrinoid systems which consist of ion-pairs of doubly charged organic –systems and alkali metals (Zilber et al., 1992; Rozenshtein et al., 1993, 1994) and the novel carbon clusters, the fullerenes (Levanon et al., 1993, 1994; Regev et al., 1993; Michaeli et al., 1994). The fullerenes will not be covered here, and the reader is referred to a review (Levanon et al., 1994). Finally, specific environments in which these assemblies are embedded in will also be discussed; in particular, liquid crystal (LC) hosts, where the interacting species are confined to an anisotropic medium with reduced rotational and translational degrees of freedom (Levanon, 1987). II. Experimental

A. TREPR Spectroscopy: General In this chapter we confine ourselves to systems that follow reactions (1–5) and can be studied by either one, or two modes of TREPR, i.e., (CW) continuous wave and/or pulsed-TREPR. Since several electronic states are involved in photophysical and photochemical processes, the systems will be described in their different paramagnetic states. These include: 1) photoexcited triplet state of chromophores (Eq. (1)); 2) doublet-state radicals, representing in-cage and charge-separated intermediates (Eqs. (2) and (3)); and 3)

Studying photoinduced ET or EnT processes, requires time-resolved spectroscopic techniques. Traditionally, optical methods are used to study ET events, and new methods in the pico- and even femto-second time scale are one path toward understanding primary charge-separation process in photochemical events (Wasielewski, 1992; Jra et al., 1993; Martin, JL et al., 1993). Nevertheless, any increase in time resolution is accompanied by a decrease in reliable spectral res-

214

olution because of the uncertainty principle, leading to a loss of vibrational, fine, and hyperfine structures (Bixon et al., 1992). Magnetic resonance has the potential for much higher spectral resolution (KHz for EPR, and less than 1 Hz for NMR), which is compensated by a reduction in time resolution that is on the edge of the typical formation kinetics of a paramagnetic species. With ESP present, TREPR spectroscopy satisfies the need for accurate data of good sensitivity on kinetic rates and also provides spectra of the transient species (Levanon and Norris, 1978, 1982; McLauchlan, 1990; Budil and Thurnauer, 1991; Bixon et al., 1992). Moreover, developments in microwave technology resulted in the construction of high-frequency (high-field) EPR spectrometers. High-field studies, in particular combined with light excitation, are promising in

Haim Levanon

structure determination, a subject which is of prime interest in photosynthesis. Indeed, there are already reports at 250 GHz G for g = 2) (Budil et al., 1989), at 150 GHz (Lebedev, 1990), at 140 GHz (Prisner et al., 1993), at 94–95 GHz mm) (Burghaus et al., 1991, 1992; Klette et al., 1993; Van der Est et al., 1993; Wang,Wet al., 1994), or at 70 GHz (Box et al., 1979). In this chapter we confine ourselves to TREPR studies of photoinduced processes, with a few examples described, employing 3 basic methods of TREPR, in the X-band region (~9.5 GHz).

Time-resolved EPR

B. Methods of TREPR Detection 1.

Light Modulation-Field Modulation (LFM)

This earliest method of TREPR combines light modulation and magnetic field modulation (LFM) with phase-sensitive detection (Levanon and Weissman, 1971; Levanon, 1979), as depicted schematically in Fig. 2. The light sources in LFM experiments may consist of either continuous excitation being chopped by a mechanical sector, or a modulated light source (e.g., xenon arcs) with a varying frequency between 1–1000 Hz. This method suffers from relatively poor time resolution of about 20 dictated by the spectrometer’s band width of 10 KHz, governed by the field modulation frequency of 100 KHz, common to most commercial EPR spectrometers (Hoff et al., 1977; Poole, 1983). The relatively low time resolution of LFMEPR, together with the significant advancements in EPR detection (see below), makes this EPR detection method less common in time-resolved experiments. 2. Continuous Wave Direct-Detection (DDEPR) In this experimental setup (Fig. 2), referred to as CW direct-detection EPR (DDEPR), the timedependent signal is produced via wavelength selective laser excitation (Kim and Weissman, 1979; Furrer et al., 1981; Weissman, 1982; Gonen and Levanon, 1985; Budil and Thurnauer, 1991). The signal, generated under CW microwave excitation, is taken from the preamplifier which is connected directly to the microwave diode detector, with care to avoid any interference from the 100-KHz filters. This signal is then fed into an automatic back-off amplifier triggered prior to the laser pulse, ensuring the transient signal is on a dc level. The output signal is fed into a digitizer interfaced to a computer on-line with the experiment. Kinetic traces, that is the magnetization, for each field position are acquired and stored. A series of complete TREPR spectra, at different times after the laser pulse are reconstructed from these traces by slowly stepping the magnetic field (only one point in the EPR spectrum is followed as a function of time after each laser pulse). In this experimental setup the

215

earliest detectable signal is at about 50–150 ns after the laser pulse. These time resolutions allow the detection of Torrey oscillations (Torrey, 1949) as well as quantum beats (Kothe et al., 1994) (see below). 3. Pulsed Microwave Detection, Fourier Transform EPR (FTEPR) While pulsed techniques represent the bulk of modern NMR spectroscopy, technical limitations associated with the considerably faster time regime and high microwave power delayed a parallel development in EPR spectroscopy. These technological difficulties have been overcome within the last decade, leading the way to considerable progress which is described in recent books and review articles (Budil et al., 1989; Bowman, 1990; Schweiger, 1991; van Willigen et al., 1993). We will restrict the discussion here to the application of those techniques which are of relevance to photosynthesis. Pulsed EPR spectroscopy combined with laser excitation has been used with either electron spin echo (ESE) detection (Norris et al., 1978; Thurnauer et al., 1979) or Fourier transform (FTEPR) detection with time resolution of about 10 ns (Angerhofer et al., 1988a, 1988b; Bowman et al., 1988; Prisner et al., 1988; Bowman, 1990; van Willigen et al., 1993). In a typical FTEPR experiment (Fig. 2), the magnetization is created by a short laser pulse (the same as in DDEPR). At selected delay times after the laser pulse, a short and intense microwave pulse (typically, 16–24 ns, 1–2 KW) is applied to the spin system. The resulting free induction decay (FID) signal is digitized and Fourier transformed to obtain the spectrum. The spectrometer response time, called the dead time limits the time resolution and causes phase distortions in the spectrum. Linear prediction, (Barkhuijsen et al., 1985, 1986) may be used to reconstruct the FIDs within the dead time regime. In this way, both the spectral and time resolutions are improved, i.e., it becomes possible to gather information in the time slice where the laser and microwave pulses overlap with a ~ 10 ns time resolution between time slices. For sufficiently strong signals one can observe the first-order kinetics of the magnetization

216

Haim Levanon

Time-resolved EPR decay (via spin lattice and chemical processes) corresponding to half-lives of ns (Norris et al., 1980). 4. Comparison Between Two TREPR Methods: DDEPR Vs. FTEPR These two methods avoid the constraints of field modulation, resulting in time resolution enhancement (compared to LFM) by three orders of magnitude, down to less than 100 ns. The combination of laser excitation with either DDEPR or FTEPR provides a good match between time resolution and the time scale of electron- and energy-transfer reactions, with a positive identification of the species detected. Also, the two techniques are complementary to each other; DDEPR works well with broad EPR lines and large spectral widths, while FTEPR is best with narrow spectra having sharp lines. In that respect, the main advantage of using FTEPR is the possibility of detecting transient radicals within a few nanoseconds (~10 ns) after laser excitation, and conducting sophisticated pulse experiments (Schweiger, 1991) combined with light excitation. Thus, the time window between 10 and 200 ns becomes available by the employment of DDEPR and FTEPR. In typical experiments, the laser pulse initiates a reaction producing paramagnetic species, free radicals, radical pairs, or triplets. In most cases, the paramagnetic species having only magnetization, along the applied magnetic field, can be completely characterized by the populations in the different spin levels. The time dependence of the populations is sufficient for determining the reactions and rates involved. In FTEPR, a microwave pulse converts which is not readily measurable, into and which forms the observable FID signal. Any additional triplets or radicals, bom after the microwave pulse, generate new and do not contribute to the FID. Any chemical process destroying radicals or triplets produces line broadening and affects line shapes, but will not change the integrated intensity of the lines in the spectrum (Bartels, 1988; Bowman, 1990). Thus, FTEPR gives a spectrum of the originally present at the time of the microwave pulse with little complication from reactions occurring after the pulse.

217

In DDEPR, a weak CW microwave field continuously converts a small portion of into and which are then detected. The and persist for a time approximately equal to the inverse of the linewidth. Thus, the DDEPR signal represents the integrated over a time window of about A second feature of DDEPR is that the signal at early times is distorted by the finite response time of the spin system (Hasharoni et al., 1991; Levanon and Bowman, 1993). Regarding CW spectrometers, a serious drawback of DDEPR method may be associated with the possible misinterpretation of the spectra, taken immediately after the laser pulse, of small hyperfine splittings and narrow linewidth (Hore and McLauchlan, 1979; Basu et al., 1984; Hasharoni et al., 1991), due to the appearance of Torrey oscillations at off-resonance fields. At short times, the spins precess, as described by the Bloch equations, about the effective magnetic field. This precession produces transient signals for a time on either side of each hyperfine line. Thus, at times shorter than this effect produces spectra which have a very different appearance than spectra of the same radicals at longer times. We emphasize this point since the experimental results which exemplify the theoretical predictions are scarce although a relevant example has been discussed (Hasharoni et al., 1991). It is noteworthy that Torrey oscillations have been detected and analyzed while investigating the spin dynamics of triplet fullerenes (Regev et al., 1993). Within the context of photoinduced intermolecular ET reactions involving triplet porphyrins (or chlorophylls) as precursors (donors), and conventional quinones (acceptors), both method of detection are essential and complementary (Levanon and Bowman, 1993). In reactions with radical-ion products of small hyperfine coupling constants and narrow line widths (e.g., quinones), it has been demonstrated that the pulsed microwave method is more advantageous (Hasharoni et al., 1990, 1991; van Willigen et al., 1993). Nevertheless, one has to consider also that the excitation bandwidth of pulsed EPR is presently limited to about 100 MHz, insufficient for many EPR applications, particularly in the slow motion regime, typical for reaction centers (RCs) investigations. In summary, DDEPR and FTEPR applications

218

Haim Levanon

exist in most photosynthetic and model system studies. III. Results

A. Introduction In recent years, several groups have been involved in pursuing the mechanistic approach in model photosynthesis, that is relevant to primary photosynthesis on the molecular level (Bixon et al., 1992). This approach seeks to reproduce the states associated with basic features of energy transfer and charge separation in photosynthesis. While simple porphyrins fail to mimic natural photosynthesis, they are important in implementing novel techniques in time-resolved spectroscopy. In that respect, the covalentlylinked porphyrin-quinone systems are successful in many respects in the mechanistic approach (singlet-initiated ET and long-lived charge-separated states) but, nevertheless, also fail to mimic natural photosynthesis, because the in vivo chromophores are not covalently linked via spacer groups. The base-paired porphyrin-quinone system (4 in Fig. 1) are more closely related to the in vivo donor acceptor species (Sessler et al., 1993). Thus, with regard to the mechanistic approach, although the porphyrin moiety is the leading structure in these studies, we do not exclude other chemical systems which may be pertinent to the mechanistic point of view.

B. Triplet Detection in Anisotropic: Solid and Fluid Phases 1. Introduction In Fig. 3 we show the two mechanisms of triplet formation via ISC routes, i.e., spin-orbit (SO) vs. radical pair (RP). By differentiating these mechanisms, the triplet state becomes a powerful diagnostic probe for structure determination (Levanon and Norris, 1978, 1982; Budil and Thurnauer, 1991). In fact, the triplet state is appropriate for probing the photoexcited singlet state, which is not accessible to EPR techniques. Parameters that affect a triplet EPR spectrum are: 1) values and signs of the zero-field splitting (ZFS) parameters, D and E; 2) selective ISC rates

of triplet sub-level formation, that are related to ESP effects; and 3) anisotropy of the environment in which the triplet chromophores are embedded. The former two aspects were dealt extensively in the literature cited above. Here, we shall focus on LCs. Rigid isotropic matrices lack the free molecular motion essential for diffusion in energy and electron-transfer reactions, while LCs can maintain their anisotropic properties in the fluid phase in accommodating and orienting guest chromophores (Fessmann et al., 1988; Regev et al., 1990, 1991a; Van der Est et al., 1990). 2. Liquid Crystalline Hosts The utilization of LCs as host matrices for large molecules facilitates the triplet EPR spectral analysis due to the partial orientation of the guest solute in the LC (Levanon, 1987). In a uniaxial nematic phase characterized by a positive anisotropic diamagnetic susceptibility, the long axes of the LC molecules are aligned along a preferred orientation, defined by a macroscopic quantity called the director, L. A common LC is E-7, which is a mixture of biphenyls (Levanon, 1987), whose phase transitions are:

In such a uniaxial LC the partial orientation im-

Time-resolved EPR

poses a cylindrical distribution of the solute about L having the following effects on the EPR spectrum (Fig. 4). (a) case: since the chromophore planes are distributed axially about L, there is a vanishing probability for the out-of-plane axis to be parallel to B, leading to the total exclusion of the Z lines from the spectrum. (b) case: by rotating the frozen sample by in an axis perpendicular to B, there is a

219

non-vanishing probability of finding any one of the three canonical orientations parallel to B. The overall effect from such organization is an improved S/N ratio and suppression, or enhancement, of various line in the EPR spectrum due to alteration of the probability of certain molecular axis to be parallel to B. A different class of LCs (e.g., ZLI-1167 (Pohl and Eidenschink, 1978; Regev et al., 1991a)) are characterized by a negative anisotropic diamag-

Haim Levanon

220 netic susceptibility, i.e., A negative implies that all phases consist of domains having directors spanned in a plane normal to the external magnetic field (Fig. 4). In such a case the triplet EPR spectra exhibit spectral features which are reversed from those found with These properties are of extreme importance when EnT and ET are monitored by TREPR in the fluid phases of these LCs. While triplet EPR of aromatic with typical ZFS are difficult to detect in isotropic fluids (Weissman, 1958; McLauchlen et al., 1994), the reduction in rotational degrees of freedom in fluid LCs makes triplet EPR detection feasible (Fessmann et al., 1988; Regev et al., 1990, 1991a; Van der Est et al., 1990). Inhibition of spin-rotational correlation times in the organized photosynthetic apparatus is the reason why the “special pair” triplet is EPR detected at 296 K (Hoff and Proskuryakov, 1985). Hence, EPR spectroscopy may, in many respects, compete with optical methods for dynamic processes. This approach was implemented by studying the molecular dynamics of photoexcited triplet states of chlorophylls and related chromophores, incorporated into LC matrices, over a wide temperature range. Triplet EPR spectra of porphyrins, porphycenes, and texaphyrins, oriented in the frozen nematic phase of uniaxial symmetry, indicate that the director, L, is aligned in the plane of the porphyrin ring (Levanon, 1987) (see also Fig. 4). This yields unique probabilities, dictated by the LC symmetry, for each of the principal axes of the ZFS tensor to be parallel to the external magnetic field. The resulting EPR spectra are highly anisotropic showing the greatest differences when the director, L, is either parallel or perpendicular to the external magnetic field, B. The relation between order properties of the host matrix (LC) and the molecular structure of the guest-oriented chromophore is reflected in the triplet magnetic parameters (e.g., ZFS and spin dynamics) determined by line shape analysis. Such analysis allows to determine energytransfer processes in covalently linked dimers (Gonen and Levanon, 1986; Regev et al., 1986) and, most importantly, relates molecular structure to magnetic dipolar coordinate system (Gonen and Levanon, 1986; Regev et al., 1989). Let us examine a few examples.

3.

Examples

a. Mesogenic Residues Attached to Porphyrins Attaching mesogenic residues or “arms” orthogonal to the porphyrin ring (Fig. 5) yields porphyrins and ZnMesogenP) compatible with standard liquid crystal compounds (Neumann et al., 1991; Michaeli et al., 1992). This chemical modification is sufficient to change the orientation of the porphyrin ring from parallel to perpendicular to the director, L, as shown by the EPR line shape of the triplet spectrum. For the hypothetical case, where the mesogenic arms are infinitely long, the out-of-plane magnetic axis of the porphyrin is expected to be collinear with the director, L. For the other extreme case, where the mesogenic arms are infinitely short, i.e., conventional porphyrins, the LC director lies in the porphyrin plane. These two cases are illustrated in Fig. 5, where the triplet spectrum of enP is compared to that of clearly exhibiting out-of-phase relationship between the two triplets.

b. Sign Determination of the ZFS Parameter and Triplet Dynamics of Nonplanar Porphyrins It is well established that the ZFS parameter, D, that characterizes the photoexcited triplet, is an important diagnostic probe, in terms of electron distribution and spin alignment, as was demonstrated on sapphyrin dication in ethanol and the nematic E-7 (Levanon et al., 1990; Regev et al., 1991b). Although the absolute value of D is easily deduced from EPR spectra, an unambiguous sign determination at relatively high temperatures (~100 K) is not straightforward and thus requires experiments to be carried out at liquid helium temperatures (Hornig and Hyde, 1963) or by performing magnetophotoselection measurements (Thurnauer and Norris, 1976). This method has been implemented for the absolute sign determination of D in triplets of sapphyrin dication monomers and dimers (Regev et al., 1991b). The anisotropic LC properties and the triplet spectra associated with them can also be utilized by the sign determination of D, as demonstrated

Time-resolved EPR

221

222

with several chromophores (Levanon et al., 1990; Berman et al., 1993). The principles of sign determination are demonstrated below. and acetylene-cumulene porphycenes (“stretched porphycenes”) represent coherent with respect to molecular dimensions and symmetry (Jux et al., 1990; Vogel et al., 1990; Martire et al., 1993; Vogel, 1993) (5 in Fig. 1 and Fig. 6). TREPR experiments of their photoexcited triplet state were carried out in isotropic (toluene) and anisotropic (a nematic LC) matrices (Berman et al., 1993). While “stretched porphycenes” exhibit a red-shift in their absorption spectra, as a function of their molecular length, the magnitude of the ZFS parameter, D, does not exhibit such a dependence. This apparent anomaly was interpreted in terms of Eq. (6), where the sign and magnitude of the ZFS parameter, D, should be determined not only by the molecular length, but also by the triplet spin alignment, with respect to the molecular plane. From the point dipole– dipole approximation, the dipolar interaction is given by Levanon and Norris (1978, 1982). where and r are the relative orientation and distance between the dipoles, respectively. In the case of nearly square porphyrins and related chromophores, the side-by-side triplet spin alignment corresponds to i.e., D>0. The headto-tail spin alignment, which is the case in the “stretched porphycenes” (Fig. 6) and sapphyrin (Levanon et al., 1990), corresponds to i.e., D < 0. These two limiting cases can be easily differentiated by the EPR spectra taken in the LC matrix, at the parallel, and perpendicular, configurations (Berman et al., 1993). A related research area is associated with the triplet dynamics, as studied by TREPR, of nonplanar porphyrins. It is noteworthy that unlike traditional planar porphyrins (e.g., the triplet state can probe the consequences of nonplanarity on the photophysics of porphyrins. Such a study has been reported recently on the nonplanar free base and zincoctaethylporphyrins, and oriented in a nematic LC (Regev et al., 1994). Again, the use of the LC has a two-fold purpose: as demonstrated earlier, triplet spin dynamics can be studied over a wide range of temperatures not possible with

Haim Levanon conventional chromophores in isotropic solvents (Regev et al., 1990, 1991, 1993); and it serves as a primitive model of the effects organized protein environments may have on nonplanar chromophores in vivo. Analysis of the EPR spectra of and in the LC, as a function of temperature, suggests fast exchange processes between different conformers attributable to variable excursions from the canonical X-ray structures of the porphyrins in their ground state. These types of studies are relevant to the mechanistic approach, where TREPR spectroscopy allows to follow conformational changes of porphyrins a process that, most likely, occurs in the photosynthetic apparatus, i.e., the chromophores packing in the proteins.

C. Electron Transfer 1. Introduction Charge separation and ET processes are fundamentally important in photochemistry and photobiology with a substantial number of applications. Numerous donor–acceptor systems have been studied by fast and ultrafast optical methods as well as by fast time-domain EPR spectroscopy. The first observation by Tollin and Green (1962) of the EPR signal from the benzoquinone anion radical in illuminated solutions of chlorophyll and benzoquinone (Tollin and Green, 1962) started a series of TREPR studies of porphyrin-based chromophores. The first applications of FTEPR to light-driven intermolecular ET reactions were carried out by Bowman and Massoth (1987) (Massoth, 1987) followed by Prisner et al. (1988) and Angerhofer et al. (1988a). Prisner et al. (1988) reported first on the ET reaction between and duroquinone, exhibiting strong polarized EPR signals at very early times after the laser pulse in Fig. 2c) that decayed over the course of a few microseconds. Currently, many FTEPR experiments, utilizing photoexcited porphyrin or porphyrinlike donors and quinones acceptors, are well described in the literature (Levanon and Bowman, 1993; van Willigen et al., 1993). The ET reactions, as depicted by Eqs. (2)– (5), represent the different approach of utilizing covalently linked donor–spacer–acceptor rigid

Time-resolved EPR

systems. Covalently linked chromophores have been used by Closs and Miller (1988) to study the ET in system of well defined geometry and energy (Closs and Miller, 1988). These landmark studies have contributed significantly to the understanding of ET in general and thereby serve as a

223

basis for understanding many of the chemical events in natural and model photosynthesis. Although numerous rigid model systems exist in the literature, only a few studies report spin polarized TREPR-spectra following light excitation (Hasharoni et al., 1990, 1991, 1993; Wasie-

224

lewski et al., 1990; Lendzian and von Maltzan, 1991). The main reason for the scarce data is that charge separation and recombination are too fast (ps to a few ns) for EPR detection. Attempts to circumvent these constrains have been pursued by modifying the chemical architecture of the covalently linked compounds, and/or by controlling the environmental conditions, such as solvent polarity, viscosity and temperature. In those cases where these goals have been achieved, DD and FTEPR are well suited for characterizing the spin state and spin dynamics of charge-separated species. As expected, triplet initiated ET reactions are most commonly observed, whereas singlet initiated ET reactions are often too fast to be directly monitored by EPR. They can, however, be deduced from the spin dynamics of the transient radical pairs involved. 2. Examples

a. Donors and Acceptors with Coulombic Interaction While all the above-mentioned experiments involve neutral photoexcited states, which are involved in the ET reactions, only a few experimental (Depew and Wan, 1986) and theoretical (Pedersen and Freed, 1974) studies are related to ESP in systems with Coulombic interaction of charged species. In principle, an attractive interradical potential can strongly affect the lifetime of the RP in the cage and, thus, effectively diminish a rate of diffusion from the cage. It is expected that this type of ion-pairing should result in a reduction of diffusion rates, which can be manifested as specific CIDEP and CRP patterns. To achieve these purposes an classic system of basic importance (Paul et al., 1956; Hoijtink et al., 1961) was chosen to be investigated (Zilber et al., 1992; Rozenshtein et al., 1993). The system consists of ion-pairs of doubly charged pyrene (Py) and alkali metals (M = Li, Na, K, Rb, Cs) in tetrahydrofuran (THF) solutions, Py/M/THF. The basic experiments utilize FTEPR spectroscopy combined with pulsed laser excitation (Fig. 2c). The reduction of Py by alkali metals in THF predominantly results in the formation of two ion-paired complexes in equilibrium (Friedenberg and Levanon, 1976; Eliav et al., 1981):

Haim Levanon

Depending upon the nature of the metal cation, these ion-paired complexes either exist as solventseparated ion-pairs, or as contact ion-pairs. Counter ions with large atomic radii, like are likely to form contact ion pairs, where the average metal–metal distance in the complex is calculated to be approximately 12 Å. For metal ions with smaller radii, like and the metal–metal distances are assumed to be increased by the inclusion of two solvent molecules, one intervening between each metal atom and Py (Rozenshtein et al., 1993). The formation of solvent-separated ion-pairs can also be enforced with large metal ions like by the use of chelates, such as cryptands that bind selectively alkali-metal cations (Friedenberg and Levanon, 1976; Rozenshtein et al., 1993). Photoexcitation of results in the formation of an excited singlet state that decays by ISC to a longer-lived polarized triplet,

The polarized triplet state is the precursor for subsequent ET resulting in paramagnetic states that obey different ESP mechanisms (Zilber et al., 1992; Rozenshtein et al., 1993):

The RP-cage size, is in the order of a few Å, with a characteristic diffusion time of where D is the mutual diffusion coefficient. Thus, the experimental setup allows to investigate in-cage processes (e.g., CRP), which are characterized by Recalling that for normal liquids we expect that nanosecond EPR detection allows to observe directly the RP in fluids with nearly

Time-resolved EPR

normal viscosity. The observed spin dynamics in these systems covers different CIDEP and CRP mechanisms, and by proper tuning of the Coulombic-solvent interactions via specific alkali metal and solvation conditions, the various ESP effects can be controlled and differentiated. The resulting FTEPR spectra are displayed in Fig. 7. Except for the Py/K/THF system, all other systems exhibit practically the same CIDEP effects. The dynamic behavior of these systems are unique examples that combine together, within the same chemical system, properties that bear a direct relation to primary photosynthesis: 1) a fixed donor–acceptor distance due to the restricted diffusion, as implemented by the CRPM; and 2) time evolution of the RP constituents developing normal CIDEP effects. The remarkably similar

225

CIDEP effects observed with Li, Na, Rb, and Cs (triplet-initiated), and the exceptionally different effects observed with K (singlet-initiated), strongly suggest that these alkali-metal systems are new examples of the inverted region in the Marcus theory (Marcus, 1964; Marcus and Sutin, 1985).

b. Multistep Electron Transfer in Covalently-Linked Assemblies Time-resolved optical studies of a carotenoid-free base porphyrin-diquinone tetrad (1 in Fig. 1) and related compounds (Gust and Moore, 1989, 1991; Gust et al., 1990; Wasielewski, 1992) suggest that optical excitation of the porphyrin moiety produces a series of ET steps.

226

An electron is transferred from the porphyrin in its lowest excited singlet state to the adjacent naphthoquinone moiety, with a rate of to produce a radical-ion pair, This initial state is thought ultimately to produce on the basis of the characteristic optical absorption of the carotenoid radical cation. The quinone anion was not detected optically and no direct evidence of the spin multiplicity was obtained. TREPR in liquid solution should remove the ambiguity concerning which quinone the unpaired electron resides on, prior to the back reaction. For an unambiguous characterization of the radical pair, pulsed-and CWEPR studies were employed (Hasharoni et al., 1990, 1991). Indeed, EPR evidence (assignment of the hyperfine splittings) indicates that the charge-separated product involves the end-quinone The two detection methods confirm the experimental validity and the origin of the derivative-like spectra assigned to the end quinone radical interacting with the carotenoid cation radical This interaction produces the derivative-like spectra via a CRP mechanism. Although both methods detect and identify the radical anion in the FTEPR detects both end radicals and and, more importantly, confirms the lifetime of the state and its singlet-state genesis. Another example is the singlet-initiated charge separation in a rigid triad, TAPD–ZnP–NQ, where the porphyrin is situated between the diamine and naphthoquinone (Wasielewski et al., 1990). Light excitation of ZnP is followed by primary charge separation to singlet radical-pair with subsequent hole transfer on a sub-nanosecond time scale to form the CRP, i.e., with a center-to-center distance of 23 Å between TAPD and NQ, very similar to the reaction center case. The observed ESP pattern at X-band, as well as the temperature-independent EPR spectra (in the range of 5–77 K), resemble strikingly that of bacterial RC and PS I (Hoff et al., 1977; Petersen et al., 1987).

c. Mixed Triplet- and Singlet-Initiated Intramolecular ET in LCs The original study by TREPR (Lendzian and von Maltzan, 1991) showed that photoinduced ET,

Haim Levanon in a porphyrin linked to benzoquinone via an aromatic phenoxy spacer (diade, P–Q), is temperature and solvent viscosity dependent. In low viscous solvents at relatively high temperatures, singlet-initiated ET is the dominant mechanism to produce the charge-separated state, whereas by increasing the solvent viscosity and by lowering the temperature, triplet state population begins to compete with the singlet state. The energetics associated with the switched-ET processes is depicted in Fig. 3. At high temperatures, where the singlet precursor dominates, the reactions are too fast for EPR detection, while in the soft glass region the triplet channel is operative (i.e., ISC and subsequent triplet ET can compete with singlet ET), thus allowing EPR detection. The TREPR experiment, carried out in the soft glass, results in spectra exhibiting two features, the first is the broad triplet spectrum of the porphyrin moiety; and the other is a narrow signal at g ~ 2, which was identified as the triplet state spectrum of the RP. Replacing the aromatic spacer with an aliphatic one (cyclohexylene) (Lendzian et al., 1991) did not change significantly the interactions present within the RP. These observation were interpreted in terms of a cross-over of the ET from the nonadiabatic limit to the solvent controlled adiabatic limit. This is made possible by slowing down the dielectric relaxation time of the solvent via temperature decrease and viscosity increase. Evidently, EPR studies of ET in model systems embedded in isotropic media are limited to a small range of temperatures, in which both triplet and ET products can be observed concurrently. EPR triplet detection in fluid phases of LCs (Regev et al., 1990) was the impetus of employing these matrices in studying ET reactions, under the influence of: 1) the directionality of the molecular entity imposed by the LC director, with respect to the external magnetic field; and 2) the different dielectric properties of the LC matrix. This approach was practiced with different DsA molecules oriented in LC solutions (Hasharoni et al., 1995). These molecules are the cis and trans isomers of zincporphyrin-cyclohexylene-quinone (t-PcQ and c-PcQ, the cis isomer is shown in Fig. 1 (3) and the trans is shown in Fig. 8) (Hasharoni et al., 1993); the para and meta isomers of zincporphyrin-phenyl-quinone (p-PpQ and m-

Time-resolved EPR

PpQ), and zincporphyrin-amide-lumiflavin (3 in Fig. 1) (Hasharoni et al., 1995). The results obtained employing LCs are different from those in the isotropic ethanol, with respect to spin dynamics and solvent-solute interactions. In fact, the latter three compounds do not exhibit any TREPR spectra in ethanol. In Fig. 8 we present the time evolution of the EPR spectra at different temperatures. Below 200 K, only one spectrum is observed and is attributed to the triplet of the porphyrin, with no apparent mixing of additional spectra. Upon increasing the temperature a relatively narrow g ~ 2.00 signal starts to emerge and superimposes on the broad triplet spectrum of the porphyrin moiety. The narrow spectrum is that of and its time evolution corresponds to the disappearance of the triplet spectrum. This

227

behavior, where mixed spectra are detected, is noticed over a wide temperature range of 210– 330 K, covering the entire nematic (i.e., fluid) phase of the LC, for this particular LC. In the case of t-PcQ and p-PpQ the experiments clearly show that above 280 K the RP spectrum changes phase, implying a switch from triplet-initiated to singlet-initiated ET. The TREPR spectra describing this dramatic effect are shown in Fig. 8. Unfortunately, these model systems fail to mimic the photosynthetic reaction center with respect to temperature dependence of the ET rates. Nevertheless, despite this discrepancy the model systems are important, when considering the mechanistic point of view, as the spin dynamics of charge separation in the model and the in vivo cases are very close.

228

d. Why Liquid Crystals Attenuate Electron Transfer Rates Numerous intramolecular electron transfer (ET) reactions, in solutions at room temperature, proceed with rates of and are monitored by transient optical spectroscopies in the picosecond time scale (Wasielewski, 1992; Jra et al., 1993; Martin, JL et al., 1993). Although optical methods allow one to extract the actual rate constants of the process, their ability to identify unambiguously the constituents in a multicomponent ET process is rather limited. Contrary to optical spectroscopy, as was demonstrated above, TREPR is unique in its ability to identify transient radicals (Bixon et al., 1992; Levanon and Bowman, 1993). However, its time resolution is in the 50 –100 ns range, thus restricting its application in ultrafast photochemical ET processes. Evidently, should be reduced by 4–6 orders of magnitude so that EPR detection will be feasible. This is the main reason why TREPR results of ET in fluids are quite scarce. The inability of isotropic solvents to be employed in TREPR studies of intramolecular ET is removed by using LCs. The main reasons accounting for this are two fold: 1) the anisotropic properties of LCs; and 2) the temperature range of TREPR detection covers the whole nematic phase of the LC (solid and liquid), thus extending the ET detection to a temperature range that may exceed 100 degrees (see phase transition temperatures above). These advantageous properties of LCs, over isotropic solvents, was explained recently (K. Hasharoni and H. Levanon, 1995), by linking together the external field effect on the dielectric properties of LCs (Carr, 1962; Meier and Saupe, 1966; Martin, AJ et al., 1971), and the internal field change imposed by the ET process (Sumi and Marcus, 1986). Such a relationship can be understood in terms of the coupling between the molecular modes (the origin of the ET) and the solvent modes. During the ET process, a nonequilibrium charge distribution is formed around the molecule. To acquire a new charge distribution about the newly formed charge-separated state, the solvent dipoles have to change their orientation about the charge. This dielectric relaxation is characterized by the solvent longitudinal dielectric relaxation time, that is re-

Haim Levanon lated to the Debye relaxation time. In the solventcontrolled adiabatic limit of the ET, is the rate-determining step of the ET process. Thus, to detect a fast EPR response, the reaction dynamics should be brought into that limit which can be controlled by varying the viscosity and temperature. This is where isotropic solvents and LCs differ. In the former, acts as a rate-determining step only in a narrow temperature range (approximately 20–30 degrees). Beyond this range, the reaction becomes nonadiabatic, i.e., it depends only weakly on the solvent characteristics. With LCs, due to the existence of a potential barrier, the nematic potential hinders an isotropic rotation (Meier and Saupe, 1966; Martin, AJ et al., 1971). The variation of upon increasing of the temperature is very slow. When an electric field is applied to the LC, for example if an ET process occurs within the solute molecules, the change of is slow because the LC dipoles do not rotate freely as in the isotropic case. This is the origin of the rate attenuation in LCs. Eq. (13) shows how the Debye dielectric relaxation time in a LC, is reduced with respect to its value in an isotropic solvent, (Martin, AJ et al., 1971) (note that is closely related to via the dielectric constants):

where g is the retardation factor and q is the nematic potential, that depends on the LC order parameter. Fortunately, various donor–acceptor systems, DsA, obey this expectation as experimentally confirmed (Hasharoni et al., 1995; Berman et al., 1995). In all these systems, TREPR spectra were detected and analyzed in the entire soft-glass and nematic range.

e. Quantum Beats in Correlated Radical Pairs Spin-correlated radical pairs (CRP) can be described by four spin states: and Usually, only mixing is operative and it results in the formation of two new states, that are not natural eigenstates of the spin Hamiltonian. This gives rise to a coherent superposition of these two states, i.e., a zero-quantum coherence

Time-resolved EPR (ZQC) (Salikhov et al., 1990; Bittl and Kothe, 1991; Wang, Z et al., 1992; Zwanenburg and Hore, 1993). The oscillatory behavior of the probability of finding the system in one of the newly formed two eigenstates produces oscillations of the magnetization, namely, quantum beats. These can be manifested in the TREPR spectra, if the ZQC can be converted into a single-quantum coherence by the microwave field of the EPR experiment. This phenomenon is similar to the well-known Torrey oscillations (Torrey, 1949) The latter may appear concurrently with the quantum beats. However, in the Torrey oscillation case, the system is prepared with no coherence among its eigenstates, at t = 0, while in the quantum beats case the RP formation process creates the coherence at t = 0. Although it is not a trivial experiment, the detection and analysis of quantum beats allows to assign values to some spin Hamiltonian parameters and lifetimes, which cannot be determined unambiguously from the spectra because of limited spectral resolution (Bittl and Kothe, 1991) To detect quantum beats by EPR, a time-resolved experiment has to be set up. The transient CW-EPR method is preferred over pulse techniques because the dead time problem can be eliminated. However, EPR observations of quantum beats in CRPs are scarce (Kothe et al., 1991, 1994), due to the difficulty in achieving the high time resolution required (~10–20 ns) in the DDEPR experiment. The reaction so-far studied is the primary ET in plant photosystem I (Kothe et al., 1991, 1994):

229

beats because of hyperfine interactions. The oscillations vary as a function of both and The variations of the beat frequency with were attributed to the anisotropic nature of the transients. This anisotropy might be used to calculate orientational parameters of the RP under study (Bittl and Kothe, 1991; Kothe et al., 1991, 1994). From simulations and fitting of the signals, the lifetime of the secondary RP in Eq. (14) was found to be ~250 ns (Kothe et al., 1991, 1994). Reproducing the quantum beats by the simulations indicates that the second RP is formed in an initial pure singlet state, i.e., any triplet mixing present during the lifetime of the first RP can be neglected due to its short lifetime. This result strongly favors the ESP scheme where the lifetime of the first RP is too short to allow for any polarization to build. IV. Concluding Remarks TREPR spectroscopy, in its current time resolution, is an important method for the detailed study of photoinduced EnT and ET processes, which, within the context of this chapter, bears a direct relation to primary photosynthesis. In some studies, TREPR is employed as a better “detector” for conventional laser photolysis experiments, while implementation of this spectroscopy with other systems (e.g., the alkali-metal or the fullerenes) is unique in addressing questions that have no direct analog in other spectroscopic methods currently used. Acknowledgements

where

is the primary chlorophyll donor and and (FeS) are the electron acceptors. In this system, a spin-polarized EPR signal was detected and attributed to the radical pair The transverse magnetization of the various transitions in the spin-polarized EPR spectrum exhibit oscillatory behavior at short times (t < ~130 ns) following the laser pulse. This short time is due to rapid averaging of the quantum

I am indebted to my students Kobi Hasharoni and Assia Berman for their assistance, comments and helpful suggestions in preparing this manuscript. The inspiring and fruitful discussions with Prof. J. R. Norris are highly appreciated. I thank Dr. M. C. Thurnauer for her constant interest and to Dr. A. Shain for reading the manuscript and his helpful comments. The Farkas Center is supported by the Bundenministerium für Forschung und Technologie and the Minerva Geselschaft für die Forschung GmbH. The research described herein was supported by the U.S.– Israel BSF, the German–Israel Foundation,

230

G.I.F, the Israel Council for Research and Development, and by the DFG–Sfb 337. References Angerhofer A, Massoth RJ and Bowman MK (1988a) Fourier transform EPR measurements of homogeneous electron transfer rates. Isr J Chem 28: 227–238 Angerhofer A, Toporowicz M, Bowman MK, Norris JR and Levanon H (1988b) Effect of triplet spin dynamics memory in electron transfer reactions. Fourier transform-electron paramagnetic spectroscopy. J Phys Chem 92: 7164–7166 Barkhuijsen H, de Beer R, Bovée WMMJ and van Ormondt D (1985) Retrieval of frequencies, amplitudes, damping factors, and phases from time-domain signals using a linear least-squares procedures. J Magn Reson 61: 465–481 Barkhuijsen H, De Beer R and van Ormondt D (1986) Error theory for time-domain signal analysis with linear prediction and singular value decomposition. J Magn Reson 67: 371–375 Bartels DM (1988) “Passive” CIDEP and the effect of chemical decay on time-resolved EPR experiments. J Magn Reson 76: 341–346 Basu S, McLauchlan KA and Seely GR (1984) A microwaveswitched time integration (MISTI) method in the study of spin polarized radicals. A two dimensional electron spin resonance experiment. Mol Phys 52: 783–793 Bauer VJ, Clive DLJ, Dolphin D, Paine JB III, Harris FL, King MM, Loder J, Wang S-WC and Woodward RB (1983) Sapphyrins: novel aromatic pentapyrrolic macrocycles. J Am Chem Soc 105: 6429–6436 Berman A, Levanon H, Vogel E and Jux N (1993) Triplet Spin Alignment of “Stretched Porphycenes”. Chem Phys Lett 211: 549–554 Berman A, Izraeli ES, Levanon H, Wang B and Sessler JL (1995) Photoinduced intraensemble electron transfer in base-paired porphyrin-quinone system. Time-resolved epr spectroscopy. J Am Chem Soc 117: 8252–8257 Bittl R and Kothe G (1991) Transient EPR of radical pairs in photosynthetic reaction centers: prediction of quantum beats. Chem Phys Lett 177: 547–553 Bixon M, Fajer J, Feher G, Freed JH, Gamliel D, Hoff AJ, Levanon H, Möbius K, Nechushtai R, Norris JR, Scherz A, Sessler JL and Stehlik D (1992) Primary events in photosynthesis: problems, speculations, controversies, and future trends. Isr J Chem 32: 369–518 Bowman MK (1990) Fourier transform electron spin resonance. In: Kevan L and Bowman MK (eds) Modern Pulsed and Continuous-Wave Electron Spin Resonance, pp 1–42 Interscience, New York Bowman MK, Toporowicz M, Norris JR, Michalski TJ, Angerhofer A and Levanon H (1988) Fourier transform epr spectroscopy of electron transfer from excited states of chlorophyll and porphyrin to duroquinone. Isr J Chem 28: 215–222 Box HC, Budzinski EE, Freund HG and Potter WR (1979) Trapped electrons in irradiated single crystals of polyhydroxy compounds. J Chem Phys 70: 1320–1325 Buckley CD, Hunter DA, Hore PJ and McLauchlan KA

Haim Levanon (1987) Electron spin resonance of spin-correlated radical pairs, Chem Phys Lett 135: 307–312 Budil DE and Thurnauer MC (1991) The chlorophyll triplet state as a probe of structure and dynamics in photosynthesis. Biochim Biophys Acta 1057: 1–41 Budil DE, Earle KA, Lynch WB and Freed JH (1989) Electron paramagnetic resonance at 1 millimeter wavelengths. In: Hoff AJ (Ed) Advanced EPR. Application in Biology, and Biochemistry, pp 307–340, Elsevier, Amsterdam Burghaus O, Plato M, Bumann D, Neumann B, Lubitz W and Möbius K (1991) 3–mm EPR investigation of the primary donor cation radical in single crystals of rhodobacter sphaeroides R-26 reaction centers. Chem Phys Lett 185: 381–386 Burghaus O, Rohrer M, Götzinger T, Plato M and Möbius K (1992) A novel high field/high frequency EPR and ENDOR spectrometer operating at 3–mm wavelength. Meas Sci Technol 3: 765–774 Carr EF (1962) Influence of electric and magnetic fields on the dielectric constant and loss of the liquid crystal anisaldazine. J Chem Phys 38: 1536–1540 Closs GL and Miller JR (1988) Intramolecular long-distance electron transfer in organic molecules. Science 240: 440– 447 Closs GL, Forbes MDE and Norris JR (1987) Spin polarized electron paramagnetic resonance spectra of radical pairs in micelles. Observation of electron spin–spin interaction. J Phys Chem 91: 3592–3599 Depew MC and Wan JKS (1986) A time-resolved CIDEP study of the photogenerated camphoquinone radical anion: a case of dual singlet and triplet precursors. J Phys Chem 90: 6597–6600 Eliav U, Levanon H, Carton PM and Fessenden RW (1981) ESR line shape analysis of the photoelectron, and the loose ion-pair in THF. Chem Phys Lett 82: 365–371 Fessmann J, Rosch N, Ohmes E and Kothe G (1988) Molecular dynamic studied by transient EPR nutation spectroscopy of photoexcited triplet states. Chlorophyll a in liquid crystalline matrix. Chem Phys Lett 152: 491–496 Friedenberg A and Levanon H (1976) Polarized photoelectrons and monomer radicals of alkali metals in THF in the presence of dicyclohexyl 18–crown-6. Chem Phys Lett 41: 84–86 Furrer R, Fujara F, Lange C, Stehlik D, Vieth H-M and Vollman W (1981) Nanosecond time resolution in EPR transient nutation spectroscopy of triplet states. Chem Phys Lett 79: 28–33 Gonen O and Levanon H (1985) Time-resolved EPR spectroscopy of electron spin polarized ZnTPP triplets oriented in a liquid crystal. J Phys Chem 89: 1637–1643 Gonen O and Levanon H (1986) Energy transfer and fine structure axes determination in a hybrid porphyrin dimer oriented in a liquid crystal. Time-resolved triplet EPR spectroscopy. J Chem Phys 84: 4132–4141 Gorter C (1936a) Paramagnetic relaxation in a transversal magnetic field. Physica 3: 1006–1008 Goiter C (1936b) Paramagnetic relaxation. Physica 3: 503– 514.

Time-resolved EPR Gust D and Moore TA (1989) Mimicking photosynthesis. Science 244: 35–40 Gust D and Moore TA (1991) Photosynthetic model systems. Top Curr Chem 159: 103–151 Gust D, Moore TA, Moore AL, Lee S-J, Bittersman E, Luttrull DK, Rehms AA, DeGraziano JM, Ma XC, Gao F, Belford RE and Trier TT (1990) Efficient multistep photoinitiated electron transfer in a molecular pentad. Science 248: 199–201 Hasharoni K and Levanon H (1995) Attenuation of electron transfer rates in liquid crystals. J Phys Chem 99: 4875–4878 Hasharoni K, Levanon H, Tang J, Bowman MK, Norris JR, Gust D, Moore TA and Moore AL (1990) Singlet photochemistry in model photosynthesis: identification of chargeseparated intermediates by Fourier transform and cw EPR spectroscopies. J Am Chem Soc 112: 6477–6481 Hasharoni K, Levanon H, Bowman MK, Norris JR, Gust D, Moore TA and Moore AL (1991) analysis of time-resolved cw-EPR spectra of short-lived radicals at different times after laser excitation. Appl Magn Res 3: 357–368 Hasharoni K, Levanon H, von Gersdorff J, Kurreck H and Möbius K (1993) Covalently linked donor–acceptor assemblies in liquid crystals. Photoinduced electron transfer monitored by time-resolved EPR. J Chem Phys 98: 2916– 2926 Hasharoni K, Levanon H, Gätschmann J, Schubert H, Kurreck H and Möbius K (1995) Molecular architecture and environmental effects in intramolecular electron transfer. An electron paramagnetic resonance study. J Phys Chem 99: 7514–7521 Hoff AJ, Gast P and Romijn JC (1977) Time-resolved ESR and chemically induced dynamic electron polarisation of the primary reaction in a reaction center particale of Rhodopseudomonas sphaeroides wild type at low temperature. FEBS Lett 73: 185–190 Hoff AJ and Proskuryakov II (1985) Triplet EPR spectra of the primary electron donor in bacterial photosynthesis at temperatures between 15 and 296 K. Chem Phys Lett 115: 303–310 Hoijtink GJ, Townsend J and Weissman SI (1961) Spin density in pyrene negative ion. J Chem Phys 34: 507–513 Hore PJ (1989) Analysis of polarized EPR spectra. In: Hoff AJ (ed) Advanced EPR. Applications in Biology and Biochemistry, pp 405–440, Elsevier, Amsterdam Hore PJ and McLauchlan KA (1979) Chemically induced dynamic electron polarization (CIDEP) and spin relaxation measurements by flash-photolysis electron paramagnetic resonance methods. J Magn Reson 36: 129–134 Hornig AW and Hyde JS (1963) Paramagnetic resonance in triplet naphthalene at liquid helium temperatures. Mol Phys 6: 33–41 Hugerat M, Levanon H, Ojadi E, Biczok L and Linschitz H (1991) Multiple decay pathway and electron transfer in excited compound zinc–copper porphyrins: laser photolysis and time-resolved EPR spectroscopy. Chem Phys Lett 181: 400–406 Jra Y, DiMagno TJ, Chan C-K, Wang Z, Du M, Hanson DK, Schiffer M, Norris JR, Fleming GR and Popov MS (1993) Primary charge separation in mutant reaction centers of Rhodobacter capsulatus. J Phys Chem 97: 13180–13191

231 Jux N, Koch P, Schmickler H, Lex J and Vogel E (1990) Acetylene-cumulene porphyrinoids. Angew Chem Int Ed Engl 29: 1385–1387 Kim SS and Weissman SI (1979) Transient magnetization following photoexcitation. Rev Chem Intermed 3: 107–120 Klette R, Törring JT, Plato M, Möbius K, Bönigk B and Lubitz W (1993) determination of the g tensor of the primary donor cation radical in single crystals of Rhodobacter sphaeroides R-26 reaction centers by 3-mm high-field EPR. J Phys Chem 97: 2015–2020 Kothe G, Weber S, Bittl R, Ohmes E, Thurnauer MC and Norris JR (1991) Transient EPR of light-induced radical pairs in plant photosystem I: observation of quantum beats. Chem Phys Lett 186: 474–480 Kothe G, Weber S, Ohmes E, Thurnauer MC and Norris JR (1994) Transient EPR of light-induced spin correlated radical pairs: manifestation of zero-quantum coherence. J Phys Chem 98: 2706–2712 Lebedev YS (1990) High-frequency continuous wave electron spin resonance. In: Kevan L and Bowman MK (eds) Modern Pulsed and Continuous-Wave Electron Spin Resonance, pp 365–404, John Wiley and Sons, New York Lendzian F and von Maltzan B (1991) Triplet electron transfer and time-resolved EPR: detection of radical-pair state of a covalently linked porphyrin-quinone. Chem Phys Lett 180: 191–197 Lendzian F, Schlüpmann J, von Gersdorff J, Möbius K and Kurreck H (1991) Investigation of the light-induced charge transfer between covalently linked porphyrin and quinone units by time-resolved EPR spectroscopy. Angew Chem Int Ed Engl 30: 1461–1463 Levanon H (1979) Principles and application of optical perturbation-electron paramagnetic resonance (OPEPR) spectroscopy. In: Dorio M and Freed JH (eds) Multiple Electronic Resonance Spectroscopy, pp 437–474, Plenum, New York Levanon H (1987) Spin polarized Triplets Oriented in Liquid Crystals. Rev Chem Intermed 8: 287–320 Levanon H and Bowman MK (1993) Time-domain EPR spectroscopy of energy and electron transfer. In: Deisenhofer J and Norris JR (eds) Photosynthetic Reaction Centers, pp 387–418, Academic Press Inc, New York Levanon H and Norris JR (1978) The photoexcited Triplet State and Photosynthesis. Chem Rev 78: 185–198 Levanon H and Norris JR (1982) Triplet state and chlorophylls. In: Fong FK (ed) Light Reaction Path of Photosynthesis, pp 152–195, Springer-Verlag, Berlin Levanon H and Weissman SI (1971) Transient effects in excitation of triplet state. J Am Chem Soc 93: 4309–4310 Levanon H, Regev A, Michaeli S, Galili T, Cyr M and Sessler JL (1990) The photoexcited triplet state of sapphyrin dication. Unusual spin alignment in monomers and spin delocalizaion in dimers. Chem Phys Lett 174: 235–240 Levanon H, Meiklyar V, Michaeli S and Gamliel D (1993) Triplet state and dynamics of photoexcited C70. J Am Chem Soc 115: 8722–8727 Levanon H, Michaeli S and Meiklyar V (1994) Triplet dynamics of and and its participation in electron transfer reactions. Time-resolved electron paramagnetic resonance. In: Kadish K and Ruoff RS (eds) Recent Advances in the

232 Chemistry and Physics of Fullerenes and Related Materials, pp 894–908, The Electrochemical Society, Pennington, NJ Marcus RA (1964) Chemical and electron transfer theory. Annu Rev Phys Chem 15: 155–196 Marcus RA and Sutin N (1985) Electron Transfers in Chemistry and Biology. Biochim Biophys Acta 811: 265–322 Martin AJ, Meier G and Saupe A (1971) Extended debye thoery for the dielectric relaxation in nematic liquid crystals. Symp Faraday Soc 5: 119–133 Martin JL, Migus A, Mourou GA and Zewail AH (1993) Ultrafast Phenomena VIII, Springer-Verlag, Berlin Martire DO, Jux N, Aramendia PF, Negri M, Lex J, Braslavsky SE, Schaffner K and Vogel E (1993) Photophysics and photochemistry of and acetylene cumulene porphyrinoids. J Am Chem Soc 114: 9969–9978 Massoth RJ (1987): Time resolved Fourier transform EPR studies in solutions of vitamins and antioxidants. PhD Dissertation, Kansas State University McLauchlan KA (1990) Continuous-wave electron spin resonance. In: Kevan L and Bowman MK (eds) Modern Pulsed and Continuous-Wave Electron Spin Resonance, (pp 285– 363) John Wiley and Sons, New York McLauchlan KA, Shkrob IA and Yeung MT (1994) Roomtemperature observations of spin-polarized triplet anthracene. Chem Phys Lett 217: 157–162 Meier G and Saupe A (1966) Dielectric Relaxation in Nematic Liquid Crystals. Molecular Crystals 1: 515–525 Michaeli S, Hugerat M, Levanon H, Bernitz M, Natt A and Neumann R (1992) Control of porphyrin orientation in thermotropic liquid crystal by molecular design. A timeresolved EPR study. J Am Chem Soc 114: 3612–3618 Michaeli S, Meiklyar V, Schulz M, Möbius K and Levanon H (1994) Photoinduced electron transfer from triplet fullerene, to tetracyanoethylene. Fourier transform electron paramagnetic resonance study. J Phys Chem 98: 7444– 7447 Morris AL, Norris JR and Thurnauer MC (1990) Reaction centers of photosynthetic bacteria. In: Michel-Beyerle ME (ed) Biophysics. Vol 6, pp 423–434, Springer-Verlag, Berlin. Neumann R, Hugerat M, Michaeli S, Natt A, Bernitz M and Levanon H (1991) Controlled orientation of porphyrins by molecular design in thermotropic liquid crystals: a time resolved EPR study. Chem Phys Lett 182: 249–252 Norris JR, Thurnauer MC, Bowman MK and Trifunac AD (1978) Electron spin echo and photosynthesis. In: Dutton PL, Leigh JS and Scrapa A (eds) Frontiers of Biological Energetics. Vol 1, pp 581–591, Academic Press, New York Norris JR, Thurnauer MC and Bowman MK (1980) Electron spin echo spectroscopy and the study of biological structure and function. In: Lawrence JH, GofmanJW and Hayes TL (eds) Advances in Biological and Medical Physics. Vol 17, pp 365–416, Academic Press, New York Paul DE, Lipkin D and Weissman SI (1956) Reaction of sodium metal with aromatic hydrocarbons. J Am Chem Soc 78: 116–120 Pedersen JB and Freed JH (1974) Some theoretical aspects

Haim Levanon of chemically induced dynamic nuclear polarization. J Chem Phys 61: 1517–1525 Petersen J, Stehlik D, Gast P and Thurnauer MC (1987) Comparison of the electron spin polarized spectrum found in plant photosystem I and in iron depleted bacterial reaction centers with time-resolved K-band EPR; evidence the photosystem I acceptor is a quinone. Photosynth Res 14: 15–29 Pohl L and Eidenschink R (1978) High resolution nuclear magnetic resonance spectrometry in thermotropic nematic liquid crystals with negative diamagnetic anisotropy. Anal Chem 50: 1934–1935 Poole PP Jr (1983) Electron Spin Resonance, 2nd edition Wiley-Interscience, New York Prisner T, Dobbert O, Dinse KP and van Willigen H (1988) FT ESR study of photoinduced electron transfer. J Am Chem Soc 110: 1622–1623 Prisner TF, McDermott AE, Un S, Norris JR, Thurnauer MC and Griffin RG (1993) Measurements of the g-tensor of the signal from deuterated cyanobacterial photosystem I particles. Proc Natl Acad Sci USA 90: 9485–9488 Regev A, Galili T, Levanon H and Harriman A (1986) The effect of mutual orientation on the triplet state magnetization of covalently linked hybrid porphyrin dimers. Chem Phys Lett 131: 140–146 Regev A, Berman A, Levanon H, Murai T and Sessler JL (1989) Principal ZFS tensor axes versus optical transition measurements of novel expanded porphyrins (texaphyrins). Time-resolved triplet EPR spectroscopy. Chem Phys Lett 160: 401–406 Regev A, Levanon H, Murai T and Sessler JL (1990) Triplet EPR detection of porphyrinoids in fluid liquid crystals at elevated temperatures. J Chem Phys 92: 4718–4723 Regev A, Galili T and Levanon H (1991a) The photoexcited triplet state as a probe for dynamics and phase memory in a multiphase liquid crystal. Time-resolved EPR spectroscopy. J Chem Phys 95: 7907–7916 Regev A, Michaeli S, Levanon H, Cyr M and Sessler JL (1991b) Solvent effect in randomly and partially oriented triplets of the sapphyrin cation. Optical and fast EPRmagnetophotoselection measurements. J Phys Chem 95: 9121–9129 Regev A, Gamliel D, Meiklyar V, Michaeli S and Levanon H (1993) Dynamics of probed by electron paramagnetic resonance. Motional analysis in isotropic and liquid crystalline matrices. J Phys Chem 97: 3671–3679 Regev A, Galili T, Medforth CJ, Smith KM, Barkigia KM, Fajer J and Levanon H (1994) Triplet dynamics of conformationally distorted porphyrins: Time resolved electron paramagnetic resonance. J Phys Chem 98: 2520–2526 Rozenshtein V, Zilber G, Rabinovitz M and Levanon H (1993) Irregular photoinduced electron transfer between doubly-charged pyrene and alkali metal cations studied by Fourier-transform electron paramagnetic resonance. J Am Chem Soc 115: 5193–5203 Rozenshtein V, Zilber G and Levanon H (1994) Dynamics of spin polarized photoelectrons in rubidium-THF solutions. J Phys Chem 98: 4236–4242

Time-resolved EPR Salikhov KM, Molin YN, Segdeev RZ and Buchachenko AL (1984) Spin Polarization and Magnetic Effects in Radical Reactions, Elsevier, Amsterdam Salikhov KM, Bock LH and Stehlik D (1990) Time development of electron spin polarization in magnetically coupled, spin correlated radical pairs. Appl Magn Res 1. 195–211 Schweiger A (1991) Pulsed electron spin resonance spectroscopy: basic principles, techniques, and examples of applications. Angew. Chem Int Ed Engl 30: 265–292 Sessler JL, Murai T, Lynch V and Cyr M (1988) An “expanded porphyrin”: the synthesis and structure of a new aromatic pentadentate ligand. J Am Chem Soc 110: 5586– 5588 Sessler JL, Murai T and Lynch V (1989) Binding of pyridine and benzimidazole to a cadmium “expanded porphyrin”: solution and x-ray structural studies. Inorg Chem 28:1333– 1341 Sessler JL, Wang B and Harriman A (1993) Long-range photoinduced electron transfer in an associated but not covalently linked photosynthetic model system. J Am Chem Soc, 115: 10418–10419 Stehlik D, Bock CH and Petersen J (1989) Amsotropic electron spin polarization of correlated spin pairs in photosynthetic reaction centers. J Phys Chem 93: 1612–1619 Sumi H and Marcus RA (1986) Dielectric relaxation and intramolecular electron transfer. J Chem Phys 84: 4272–4276 Thurnauer MC and Norris JR (1976) Magnetophotoselection applied to the triplet state observed by EPR in photosynthetic bacteria. Biochem Biophys Res Commun 73: 501– 504 Thurnauer MC and Norris JR (1980) An electron spin echo phase shift observed in photosynthetic bacteria algae. Possible evidence for dynamic radical pair interactions. Chem Phys Lett 76: 557–561 Thurnauer MC, Bowman MK and Norris JR (1979) Time resolved electron spin echo spectroscopy applied to the study of photosynthesis. FEBS Lett 100: 309–312 Tollin G and Green G (1962) Light induced single electron transfer between chlorophyll a and quinones in solution. Biochim Biophys Acta 60: 524–538 Torrey H (1949) Transient nutations in nuclear magnetic resonance. Phys Rev 76: 1059–1068 Trifunac AD, Lawler RG, Bartels DM and Thurnauer MC (1986) Magnetic resonance studies of paramagnetic transients in liquids. Prog Reaction Kinetics 14: 43–156 Van der Est AJ, Steidl P, Schulz M, Bock CH, Vieth HM and Stehlik D (1990), The application of time–resolved

233 EPR to the study of photoreactive molecules in solid and liquid crystalline environments. Appl Magn Res 1: 54–69 van der Est A, Bittl R, Abresch EC, Lubitz W and Stehlik D (1993) transient EPR spectroscopy of perduterated Znsubstituted reaction centers of Rhodobacter sphaeroides R26. Chem Phys Lett 212: 561–568 van Willigen H, Levstein PR and Ebersole MH (1993) Application of Fourier transform electron papramagnetic resonance in the study of photochemical reactions. Chem Rev 93: 173–197 Vogel E (1993) The porphyrins from the “annulene chemist’s” perspective. Pure and Appl Chem 65: 143–152 Vogel E, Jux N, Rodriguez-Val E, Lex J and Schmickler H (1990) Porphyrin homologs. [22]porphyrin(2.2.2.2), an extended porphycene. Angew Chem Int Ed Engl 29:1387– 1390 Wang W, Belford RL, Clarkson PH, Forrer J, Nilges MJ, Timken MD, Walczak T, Thurnauer MC, Norris JR, Morris AL and Zhang Y (1994) Very high frequency EPR94 GHz instrument and applications to primary reaction centers from photosynthetic red bacteria and to other disordered systems. Appl Magn Res 6: 195–215 Wang Z, Tang J and Norris JR (1992) The time development of the magnetic moment of correlated radical pairs. J Magn Reson 97: 322–334 Wasielewski MR (1992) Photoinduced electron transfer in supramolecular systems for artificial photosynthesis. Chem Rev 92: 435–461 Wasielewski MR, Gaines GL, O’Neil MP, Svec WA and Niemczyk MP (1990) Spin-polarized radical ion pair formation in a fixed-distance photosynthetic model system. J Am Chem Soc 112: 4559–4565 Weissman SI (1958) On detection of triplet molecules in solution by electron spin resonance. J Chem Phys 29: 1189– 1190 Weissman SI (1982) Recent developments in electron paramagnetic resonance transient methods. Annu Rev Phys Chem 33: 301–318 Zavoisky E (1945a) Paramagnetic relaxation of liquid solutions for perpendicular fields. J Phys USRR, 9: 211–216. Zavoisky E (1945b) Spin-magnetic resonance in paramagnetics. J Phys USRR 9: 245–245 Zilber G, Rozenshtein V, Levanon H and Rabinovitz M (1992) Photoinduced electron transfer between doublycharged pyrene and alkali metals. Fourier transform EPR study. Chem Phys Lett 196: 255–260 Zwanenburg G and Hore PJ (1993) EPR of spin-correlated radical pairs. Analytical treatment of selective excitation including zero-quantum coherence. Chem Phys Lett 203: 65–74

This page intentionally left blank

Chapter 15 Electron Spin Echo Methods in Photosynthesis Research R. David Britt Department of Chemistry, University of California, Davis, Davis, CA 95616, USA Summary I. Introduction A. Inhomogeneous Broadening B. Electron Spin Echoes II. ESEEM A. 2-Pulse ESEEM B. 3-Pulse ESEEM III. ESE-ENDOR A. Davies ESE–ENDOR B. Mims ESE–ENDOR IV. Additional Examples of ESE Applications in Photosynthesis A. Exchange about the PS II Mn Cluster B. Davies ESE–ENDOR of C. Davies ESE–ENDOR of Mn Clusters V. Instrumentation Acknowledgements References

235 235 235 237 238 238 242 243 243 245 246 246 249 249 249 252 252

Summary Electron spin echo methods allow photosynthesis researchers to probe the local structure of paramagnetic transition metal and organic radical centers. In particular the electron spin echo methods of electron spin echo envelope modulation (ESEEM) and electron spin echo – electron nuclear double resonance (ESE–ENDOR) are used to measure nuclear spin transitions of magnetic nuclei of, and in the vicinity of, paramagnetic species. The resulting magnetic hyperfine and electric quadrupolar parameters can be interpreted to give insights into the structure of a complex and its interaction with the neighboring protein matrix and molecules bound as substrates or inhibitors. Abbreviations: CW – continuous wave; ENDOR – electron nuclear double resonance; EPR – electron paramagnetic resonance; ESE – electron spin echo; ESEEM – ESE envelope modulation; PSII – photosystem II; acceptor quinone; RF – radio frequency transfer systems, photosynthetic reaction centers contain a variety of organic cofactors and transition metal complexes that can be found or poised in paramagnetic states. Thus electron paramagnetic resonance (EPR) has been a very powerful tool in photosynthetic research. However conventional continuous-wave (CW) EPR spectroscopy has limitations that can be overcome by performing pulsed EPR experiments. This chapter fo-

I. Introduction

A. Inhomogeneous Broadening By their very nature of being light-driven electron

Correspondence: Fax: 1-916-7528995; E-mail: [email protected]

235 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 235–253. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

236

cuses on those pulsed EPR techniques based on multipulse electron spin echo (ESE) sequences. The ESE methods are useful because EPR lineshapes of photosynthetic paramagnetic entities are typically dominated by inhomogeneous broadening as illustrated in Fig. 1. The inhomogeneously broadened lineshape results from the overlap of resonances (shown as “spin packets”) resulting from a near continuum of magnetic environments. One such origin of inhomogeneous broadening is the unresolved overlap of hyperfine lines from many coupled nuclei, because the number of hyperfine lines increases multiplicatively with the number of classes of coupled nuclei. For example, for the case of a spin S = 1/2 complex coupled to k different classes of equivalent spin nuclei with coupling constants the number of EPR lines and the EPR spectral density are given by the following expressions (Kurreck et al., 1988):

R. David Britt

If the spectral density becomes sufficiently high, the spacing between individual hyperfine lines will become less than the intrinsic lifetime-broadened linewidth of each spin packet, and the EPR lineshape will become dominated by the resulting Gaussian lineshape. Other sources of inhomogeneous broadening include hyperfine and g anisotropies in non-crystalline samples and site-tosite “strain” of such parameters (Mims, 1972a).

Electron spin echo methods

B. Electron Spin Echoes Fortunately, the inhomogeneous broadening can be negated by the use of multipulse magnetic resonance sequences to generate spin echoes (Mims, 1972a; Mims and Peisach, 1981; Slichter, 1990). The sequence and underlying mechanism

237

of the simplest spin echo sequence, that of the 2pulse Hahn echo (Hahn, 1950), is illustrated in Fig. 2. At point (a), illustrated in the sequence and in the underlying rotating frame vector picture, a resonant field is applied along the xaxis to rotate magnetization from its initial equilibrium direction (along the z-axis coincident with

238

the static field).. By point (b) the magnetization has been rotated by onto the rotating frame y-axis and the field is removed. In the subsequent interval individual spin packets, of which a select set of five are displayed, will precess at different frequencies due to the effect of inhomogeneous broadening. Spin packet 0 has a resonance frequency equal to the rotating frame frequency relative to the laboratory frame, and therefore remains fixed along the y-axis. The remaining spin packets have higher or lower resonant frequencies, and therefore precess away from the y-axis during the free induction (c) following the pulse. At time (d) a pulse is applied to the spin system. The spin packets are rotated into the positions displayed at point (e). Subsequent to this second pulse, each spin continues to precess with its original sense and rate relative to the y-axis, and at a time after this pulse all spin packets refocus simultaneously onto the –y-axis. This evanescent magnetization coherence is referred to as a spin echo, and it manifests itself as a burst of radiation that can be detected with the spectrometer. The detected echo can be used to generate a field swept EPR spectrum analogous to the conventional CW EPR spectrum. Field modulation is not employed, so the spectrum displayed is a measure of the direct absorption rather than the field derivative. This can be beneficial for the detection of signals with broad, relatively featureless lineshapes where the field derivative is everywhere small (Nishi et al., 1980; Britt et al., 1992; Gilchrist et al., 1992; Zimmermann et al., 1993). Moreover since the spin echo experiment is intrinsically a time-domain measurement, its time resolved character can be fruitfully exploited. Thus ESE spectroscopy can be utilized to study kinetics of electron transfer and spin polarization (Thurnauer and Clark, 1984; Bosch et al., 1992; Moenne-Loccoz et al., 1994) as well as to accurately measure relaxation times of electron spin systems (Bosch et al., 1991; Lorigan and Britt, 1994). However the primary focus of this chapter is on applications of ESE spectroscopy designed to overcome the deleterious effects of inhomogeneous broadening obscuring hyperfine interactions with magnetic nuclei. The techniques of electron spin echo envelope modulation (ESEEM) and electron spin echo – electron nu-

R. David Britt clear double resonance (ESE-ENDOR) are utilized to detect the nuclear spin transitions of such nuclei that are in magnetic contact with electron spins. In these experiments the spin echo is the carrier onto which nuclear spin information is encoded, either by time domain interference (ESEEM) or through RF-driven magnetization transfer (ESE-ENDOR). Both techniques will be illustrated with experiments on a representative photosynthetic paramagnetic center: the semiquinone PS II radical generated by dithionite reduction in PS II particles where the non-heme Fe(II) has been converted to a diamagnetic lowspin state by treatment with cyanide (Sanakis et al., 1994). II. ESEEM In the ESEEM experiment, electron spin echoes are formed by the application of two or more resonant microwave pulses. In addition to inducing the electron spin transitions, the microwave pulses may also induce “semi-forbidden” transitions of nuclear spins magnetically coupled to the electron spins, resulting in quantum mechanical coherences in the nuclear spin sublevels associated with the electron spin levels. These coherences create interference effects which can be measured by varying the electron spin echo pulse timing. Fourier analysis of the resulting time-domain electron spin echo envelope modulation pattern reveals the frequencies of the nuclear spin transitions. The frequencies and amplitudes of the Fourier peaks can be interpreted to determine hyperfine and electric quadrupolar interactions of the coupled nuclei. Typical ESEEM experiments utilize two-pulse Hahn echo (Fig. 3a) or three-pulse ‘stimulated’ echo (Fig. 3b) sequences (Mims, 1972a, 1972b; Kevan, 1979; Mims and Peisach, 1981, 1989; Thomann and Bernardo, 1993).

A. 2-Pulse ESEEM In the two-pulse ESEEM experiment the electron spin echo amplitude is measured as a function of the interpulse spacing (Fig. 3a). A qualitative picture of how hyperfine information is encoded onto the echo amplitude as a function of is provided by Fig. 4, which parallels the final three

Electron spin echo methods

239

240

steps of Fig. 2. We focus on a single spin packet (1) starting at the onset of the pulse (d). Suppose that during the pulse there is a partially allowed spin transition for a class of I = 1/2 nuclei hyperfine-coupled to the electrons of the spin packet. Those electrons whose nuclear partners do not flip remain in spin packet 1 after the pulse (e). However those electrons whose coupled nuclei do flip are shifted into a different spin packet 1´ whose precession frequency differs from that of packet 1 by the hyperfine frequency The magnetization vectors of the two packets are coincident initially, but will beat in and out of phase due to the frequency difference between the two packets. Packet 1 converges with other packets on the –y-axis to form the echo at time after the pulse. The projection of the 1´ packet magnetization onto the –y-axis at time affects the amplitude of the echo. Thus the echo amplitude is modulated by the hyperfine frequency. A quantum mechanical treatment such as the density matrix approach used by Mims (1972a, 1972b) is required to properly describe the ESEEM phenomenon, including the amplitude of the modulation and the effects of other nuclear spin Hamiltonian terms such as the nuclear Zeeman and electric quadrupolar interactions. One specific result of the density matrix analysis is that in addition to the fundamental nuclear spin transition frequencies, sum and difference frequencies also appear in the 2-pulse ESEEM experiment. Figure 5 shows 2-pulse ESEEM results for the anion radical of PS II. The time domain modulation pattern is shown in Fig. 5a. The frequency domain spectrum shown in Fig. 5b is generated as the Fourier Transform of the time domain pattern. The peak at 14.1 MHz arises from weakly coupled protons. The strongly coupled protons of the radical do not produce measurable modulation at this magnetic field (3316 G), and these are generally best studied with ESE– ENDOR (vide infra). However there is much information in the lower frequency range of the spectrum. The pattern of three sharp low frequency peaks and a broader peak at higher frequency are characteristic of modulation when the hyperfine and external magnetic fields are of approximately the same amplitude (Mims and

R. David Britt Peisach, 1978; Flanagan and Singel, 1987). Under this “exact cancellation” condition the local magnetic field at the is essentially nulled for one electron spin orientation, and three sharp peaks arise because of the electric quadrupole interaction that splits the three levels of the I = 1 nucleus in the absence of a magnetic field (Das and Hahn, 1958). The higher frequency “doublequantum” peak arises from the transition between the outer energy levels of the for the other electron spin orientation where the hyperfine and external magnetic fields add. The corresponding transitions involving the inner level are typically too broad to observe in non-crystalline samples. Analysis of the frequencies and lineshapes of the quadrupolar peaks and the double quantum peak provides a full determination of the electric quadrupole and hyperfine interactions. As in other forms of magnetic resonance, isotopic substitution provides one of the most powerful assignment tools in pulsed EPR spectroscopy (for example: Hoff et al., 1985; De Groot et al., 1985; Evelo et al., 1989; Britt et al., 1989; Tang et al., 1994; Warncke et al., 1994). Figure 5c shows the frequency domain ESEEM results for in globally PS II particles. The low frequency region of the spectrum is completely altered, confirming our assignment of the previous transitions to a nucleus coupled to the radical. For the I = 1/2 nucleus we resolve a single transition at 2.78 MHz, a value very close to twice the Larmor frequency of 1.43 MHz at the applied field of 3316 G. This confirms that we are indeed in the “exact cancellation” limit for this nitrogen. We do not observe a low frequency peak from the nucleus for the electron spin orientation where the two fields cancel because it is of too low a frequency to observe with the 2-pulse ESEEM sequence. This explicitly brings up an important drawback to the 2-pulse sequence. The 2-pulse echo amplitude decreases rapidly with increased due to spin-spin relaxation processes. In the representation of Fig. 2 this corresponds to the magnitude of the magnetization vector of each spin packet decreasing over time due to random spin-spin dephasing. The effect of this limited “phase-memory” is dramatically seen in Fig. 5a, where the echo amplitude has dropped to almost nothing after 3

Electron spin echo methods

241

242

This loss of the echo carrier with time leads to a loss of resolution of low frequency modulation components. Fortunately, this problem can be circumvented by using the 3-pulse ESEEM method based on the stimulated echo sequence (Mims, 1972a).

B. 3-Pulse ESEEM The stimulated echo sequence can be seen as the first (solid) pulse train of the 3-pulse ESEEM diagram (Fig. 3b). The first pulse rotates the magnetization onto the y-axis. A second pulse, applied after the magnetization has dephased for time rotates this magnetization pattern into the xz-plane. A spin packet with a frequency that matches the rotating frame frequency will be shifted onto the –z-axis, as will spin packets that have rotated back onto the y-axis during the dephasing time On the other hand, spin packets which have rotated onto the –y-axis by time will be rotated onto the z-axis. This gives rise to a sinusoidal non-equilibrium magnetization pattern encoded along the z-axis as illustrated in Fig. 6. This magnetization pattern decays on the order of the spin-lattice relaxation time which is typically much longer than the phase memory time. A final pulse at time T

R. David Britt

generates the stimulated echo a time later. This stimulated echo is essentially the free induction decay of the sinusoidal magnetization pattern imposed by the first two pulses. In the 3-pulse ESEEM experiment, the stimulated echo is measured as a function of the time T. The great advantage is that the carrier echo now lasts a much longer time, allowing better spectral resolution of the modulation frequencies. A disadvantage is that the echo is only one half the initial intensity of the 2-pulse echo because magnetization components remaining along the ±x axis after the first two pulses are lost on the timescale of the phase memory. Another difference that appears from the density matrix theory is that only fundamental nuclear spin transition frequencies appear in the 3-pulse ESEEM experiment, although their amplitudes depend on the specific value of the the first interpulse time used in the experiment, and certain frequency components may be completely suppressed for certain values (Mims, 1972b, 1972c). An investigator typically performs the two-pulse ESEEM experiment along with a set of three-pulse ESEEM experiments with different values to completely characterize the ESEEM effects. Also, additional information correlating hyperfine transitions can be gained using the 2-dimensional ESEEM-derived method

Electron spin echo methods of HYSCORE (Hofer et al., 1986) which has recently been introduced to photosynthesis research by Käss et al. (1995). Figure 7a shows the longer time range that can be exploited in the 3-pulse ESEEM experiment for the radical. The carrier stimulated echoes show little diminution out to a time of 16 where the modulation is essentially damped by the intrinsic linewidth of the nuclear transitions. The frequency resolution in the Fourier Transform ESEEM spectrum (Fig. 7b) is much greater than in the corresponding 2-pulse spectrum (Fig. 5b). The weakly coupled proton modulation in these data sets is suppressed by working at a multiple of the proton Larmor frequency, and the frequency domain data are therefore plotted only to 6 MHz to expand the low frequency region of the spectra. The three peaks at 0.75, 2.06, and 2.85 MHz arise from the nucleus with electric quadrupole coupling parameters MHz and These quadrupole values are characteristic of a in a peptide bond (Edmonds, 1977), so we postulate that this modulation arises from a peptide nitrogen hydrogenbonded to the quinone. The 5.0 MHz frequency of the double quantum transition allows us to calculate a hyperfine coupling of 2.0 MHz to this nitrogen. The 3-pulse ESEEM spectrum of the sample (Fig. 7c) shows a very sharp low frequency at 0.268 MHz. The presence of such exceedingly sharp ESEEM lines is predicted by Lai et al. (1988) for I = 1/2 nuclei such as near the exact cancellation limit. A dipolar coupling of 0.3 MHz would give rise to such a line-narrowed peak under these experimental conditions. The 2.81 MHz transition seen in the 2-pulse spectrum (Fig. 5c) is suppressed at the value of 213 ns used in this data set. We also observe a contribution at the Larmor frequency of 1.43 MHz arising from additional weakly coupled nuclei. III. ESE ENDOR Unlike ESEEM, the ESE–ENDOR experiments do not rely on semi-forbidden nuclear spin transitions during the microwave pulses. Rather, the nuclear transitions are driven directly with separate high-power radio frequency (RF) pulses. In an ESE–ENDOR experiment, an alteration of

243

the initial electron spin magnetization is created by one or more high-power resonant microwave pulses. Application of the radio frequency pulse further perturbs the electron magnetization if the RF pulse induces spin transitions of nuclei magnetically coupled to the electron spins. The nuclear spin transition frequencies are measured by varying the radio frequency while monitoring the effect of the RF pulse on a subsequent electron spin echo (Hoffman et al., 1993; Thomann and Bernardo, 1993). Pulse sequences introduced by Davies (1974) and by Mims (1965) are typically utilized (Fig. 8).

A. Davies ESE–ENDOR The Davies ESE–ENDOR sequence is shown in Fig. 8a. The inverting microwave pulse inverts the electron magnetization in a narrow resonant bandwidth within the inhomogeneously broadened line (Fig. 9a). When resonant with a nuclear spin transition, the RF pulse transfers magnetization between an unperturbed region of the spectrum and the resulting “hole”, decreasing the extent of magnetization inversion in the hole. This results in an altered magnitude of the spin echo induced by a final two-pulse ESE sequence. The Davies sequence works best for relatively strongly coupled nuclei. For weakly coupled nuclei the nuclear spin flips simply transfer magnetization within the hole, giving a negligible change in echo amplitude. For nuclei with a modest coupling this effect can be mitigated by using long microwave pulses to burn narrow holes in the inhomogeneously broadened line. Figure 10a shows the Davies ESE–ENDOR of the sample. The broad overlapping powder patterns centered about the Larmor frequency of 15.4 MHz arise from protons of the radical and its immediate environment. The powder patterns are obtained with very good signal intensity. Detailed assignments have not been made. The 2.8 MHz transition is also observed with good intensity, though down appreciably from the proton signals due to the smaller gyromagnetic ratio for the nucleus. The Larmor frequency peak observed in the 3-pulse ESEEM is not detected. Parallel experiments show poor signals in the low frequency region (data not shown), most likely due to a

244

R. David Britt

Electron spin echo methods

combination of quadrupolar broadening and a smaller gyromagnetic ratio for the nucleus.

B. Mims ESE–ENDOR Figure 8b displays the Mims ESE–ENDOR sequence, in which the RF pulse is provided following the generation of the sinusoidal magnetization pattern produced by the first two pulses of a stimulated echo sequence (Fig. 6). The amplitude of this sinusoidal magnetization pattern is reduced by RF-induced nuclear spin transitions, resulting in reduced stimulated echoes when the radio frequency is resonant with nuclear spin transitions. The Mims ENDOR transitions are thus inverted; driving the nuclear resonance reduces the echo size. In principle, the Mims sequence gives larger ENDOR effects than the Davies sequence, but suffers from “blind-spots” for couplings at multiples of arising from the sinusoidal nature of the induced magnetization

245

pattern; transferring magnetization from one trough to another has no effect on the echo amplitude. However, for weakly coupled nuclei, i.e. nuclei with hyperfine interactions of a few MHz or less, this is not a problem as long as the value is kept small (minimum values of approximately 100 to 300 ns are typical depending on the ‘dead-time’ of the specific instrument). In fact, the value can be adjusted to give maximal sensitivity for a given hyperfine coupling value. Figure 10b shows the Mims ESE–ENDOR of the sample. The proton region was not scanned because we knew from experience that the broad proton powder patterns would be distorted because of the blind spots intrinsic to this method. In the nitrogen region of the spectrum the 2.8 MHz feature is clearly resolved. Unlike in the Davies ENDOR spectrum, there is also some intensity observed at the Larmor frequency of 1.56 MHz. To summarize, we have obtained pulsed EPR

246

results on the radical using both ESEEM and ESE–ENDOR methods. For the and nitrogen nuclei near the exact cancellation limit the ESEEM sequences give superb results. Measuring spin transitions of nitrogen nuclei in this regime is one of the classic applications of ESEEM (Mims and Peisach, 1981). Good results for are obtained with both Davies and Mims ESE– ENDOR sequences though the results are less favorable. For studying the broad proton powder patterns the Davies ESE–ENDOR method gives the best results.

R. David Britt

IV. Additional Examples of ESE Applications in Photosynthesis

A. Exchange about the PS II Mn Cluster In thepreceding experiments we observed the strength of ESEEM is measuring spin transitions of weakly coupled nitrogens. Another classic ESEEM application is measuring hyperfine interactions to weakly coupled exchangeable deuterons (Kevan, 1979). The modulation from the nucleus is approximately three times greater than from a proton in an equivalent site. Figure 11 shows the 2-pulse ESEEM results for the multiline signal of the oxygen evolv-

Electron spin echo methods

247

248

R. David Britt

Electron spin echo methods

ing complex (Dismukes and Siderer, 1981; Brudvig, 1989) following exchange. The timedomain pattern is simulated with contributions from proximal deuterons. The simulation is not unique, but we do require multiple deuterons at dipolar-coupled distances on the order of 2.5–2.7 Å to simulate the depth and rapid modulation damping of the time domain data. These close distances support water or hydroxo ligation to the Mn cluster by the of the Kok cycle. Further and ENDOR experiments will help to refine these results. B. Davies

ESE–ENDOR of

The tyrosine radical of Photosystem II (Barry and Babcock, 1987; Debus et al., 1988) has been studied with ESEEM in globally (Evelo et al., 1989) and selectively deuterated (Warncke et al., 1994) samples. ESEEM methodology for examining such radicals has recently been advanced (Warncke and McCracken, 1994). In our laboratory, Davies ENDOR has yielded very nice results on the and tyrosine radicals (Gilchrist et al., 1995). Figure 12 shows the Davies ESE–ENDOR and corresponding simulations of the radical. The ENDOR of the broad radical observed in depleted PS II (Boussac et al., 1990) and of the tyrosine are each very similar except for a slight increase in frequency of the more-strongly coupled proton, corresponding to a decrease in the dihedral angle between the CH bond and the tyrosine ring normal from 47 to 43° (data not shown). We consider the strong similarity of the and pleted radical ENDOR spectra to be strong evidence that the radical signal originates from C. Davies

ESE–ENDOR of Mn Clusters

The EPR spectrum of the PS II Mn cluster is complex because of the overlap of approximately 1300 hyperfine lines from four 100% natural abundance nuclei (Eq. (1)). We are using the Davies ESE–ENDOR method to measure the hyperfine and quadrupole couplings of these I = 5/2 nuclei. The couplings are quite large, so the Davies ESE–ENDOR sequence clearly is best for this application. The data analy-

249

sis for the tetranuclear cluster is complicated by the many overlapping EPR transitions that make up the overall signal, so we have started detailed analysis on a simpler system: an antiferromagnetically exchange-coupled Mn(III)Mn(IV) dinuclear cluster (Cooper et al., 1978). The EPR spectrum of the S = 1/2 ground state of this cluster shows 16 resolved lines resulting from the overlap of the 36 powder patterns resulting from the two inequivalent nuclei. The top trace of Fig. 13 shows the field-swept ESE spectrum of the first four lines and a simulation of the contributing hyperfine powder patterns. The first two lines arise from unique powder patterns, the second two from two overlapping powder patterns, and the central transitions (not shown) each result from 3 EPR powder patterns. Except when performing ENDOR in the outermost EPR lines, one must take in account multiple initial magnetic quantum numbers for the two nuclei when performing the simulations. The experimental ESE–ENDOR results obtained at a number of the EPR peaks across the spectrum are shown in the lower traces of Fig. 13 along with simulations. The simulations are seen to match reasonably well across the spectrum. Parameters used to simulate the ENDOR of the Mn(IV) ion are hyperfine parameters MHz and 211 MHz and quadrupole parameters 26.7 MHz and For the higher frequency transitions for the Mn(III) ion the employed parameters are MHz and MHz and quadrupole parameters and V. Instrumentation

The principal requirement for performing pulsed EPR experiments such as described in this chapter are high microwave pulse power (~1 kW) and electronics to provide for control of pulse timings on the nanosecond time scale. Traditionally, pulsed EPR instrumentation has been laboratorybuilt (Norris et al., 1980; Nishi et al., 1980; Britt et al., 1989; Sturgeon and Britt, 1992; McCracken et al., 1992). However, the commercial instrument provided by Bruker Instruments is now being used by a number of photosynthesis research groups (Astashkin et al., 1994; Käss et al.,

250

1995; Davis et al., 1993; Zimmermann et al., 1993). Over the past few years the advantage of performing ESEEM experiments over a wide range of magnetic field values has become clearly recognized, and for a fixed g-value, this translates to performing experiments over a wide microwave frequency range (Singel, 1989). For example, for nuclei with weak dipolar couplings, the modu-

R. David Britt

lation depth is inversely proportional to the square of the external field strength (Kevan, 1979). Moreover, as previously described, for nuclei with moderate hyperfine couplings, it is necessary for optimal results to be able to operate at external field values that cancel the external hyperfine fields (Flanagan and Singel, 1987; Lai et al., 1988). Therefore it is very important to have a spectrometer (or spectrometers) with a

Electron spin echo methods

251

252

wide frequency range. For example, the pulsed EPR instrument that we recently constructed operates over a 10 GHz frequency range from 8 to 18 GHz (Sturgeon and Britt, 1992). It is also very useful to have a wide range of frequencies available in the ESE–ENDOR experiment in order to track the movement of transition frequencies with field and to separate overlapping transitions from different nuclear species. In particular, at X band frequencies the ubiquitous proton signals often overlap with signals from other nuclei such as and If for some reason an ESE–ENDOR instrument were designed to work at a single microwave frequency, it would be preferable if the frequency were 20 GHz or higher in order to spread the ENDOR signals over a wider frequency range and to move the proton signals to a frequency range less populated with transitions from other nuclear species. One direction being actively pursued is in the area of ultra-high frequency pulsed EPR using superconducting magnets (Prisner et al., 1992, 1994; Weber et al., 1989; Disselhorst et al., 1995). Acknowledgements Research in my laboratory is supported by a grant from the National Institutes of Health, and by a NRICGP grant from the United States Department of Agriculture. I thank Professor William Armstrong, Dr. James A. Ball, Dr. Bruce A. Diner, M. Lane Gilchrist Jr., Gary A. Lorigan, Dr. Jeffrey M. Peloquin, David Randall, Dr. Bradley E. Sturgeon, and Dr. Xiao-Song Tang for contributing unpublished data for this chapter. References Astashkin AV, Kodera Y and Kawamori A (1994) Pulsed EPR study of manganese g = 4.1 signal in plant photosystem II. J Magn Reson 105: 113–119. Barry BA and Babcock GT (1987) Tyrosine radicals are involved in the photosynthetic oxygen-evolving system. Proc Natl Acad USA 84: 7099–7103. Bosch MK, Evelo RG, Styring S, Rutherford AW and Hoff AJ (1991) ESE relaxation measurements in the photosystem II. The influence of the reaction center non-heme iron on the spin-lattice relaxation of Tyr D. FEBS Lett 292: 279–283. Bosch MK, Gast P and Hoff AJ (1992) Applications of ESE spectroscopy in the study of electron spin polarization in bacterial photosynthesis. Pure Appl Chem 64: 847–857.

R. David Britt Britt RD. Zimmermann JL, Sauer K and Klein MP (1989) Ammonia binds to the catalytic Mn of the oxygen evolving complex of photosystem II: evidence by electron spin echo envelope modulation spectroscopy. J Am Chem Soc 111: 3522–3532. Britt RD, Lorigan GA, Sauer K, Klein MP and Zimmermann JL (1992) The g = 2 multiline EPR signal of the state of the photosynthetic oxygen-evolving complex originates from a ground spin state. Biochim Biophys Acta 1140: 95– 101. Brudvig GW (1989) EPR spectroscopy of manganese enzymes. In: Hoff AJ (ed) Advanced EPR: Applications In Biology and Biochemistry, pp 839–863. Elsevier, Amsterdam. Boussac A, Zimmermann JL, Rutherford AW and Lavergne J (1990) Histidine oxidation of the oxygen-evolving photosystem-II enzyme. Nature 347: 303–306. Cooper SR, Dismukes GC, Klein MP and Calvin M (1978) Mixed valence interactions in bridged manganese complexes. Electron paramagnetic resonance and magnetic susceptibility studies. J Am Chem Soc 100: 7248–7252. Das TP and Hahn EL (1958) Nuclear Quadrupole Resonance Spectroscopy. Academic Press, New York. Davies ER (1974) A new pulse ENDOR technique. Phys Lett 47A: 1–2. Davis IH, Heathcote P, MacLachlan DJ and Evans MCW (1993) Modulation analysis of the electron spin echo signals of in vivo oxidized primary donor nitrogen-15 chlorophyll centers in bacterial, P870 and P960, and plant photosystem I, P700, reaction centers. Biochim Biophys Acta 1143: 183– 189. Debus RJ, Barry BA, Babcock GT and McIntosh L (1988) Site-directed mutagenesis identifies a tyrosine radical involved in the photosynthetic oxygen-evolving system. Proc Natl Acad USA 85: 427–430. De Groot A, Evelo R, Hoff AJ, De Beer R and Scheer H (1985) Electron spin echo envelope modulation (ESEEM) spectroscopy of the triplet state of the primary donor of and bacterial photosynthetic reaction centers and of and bacteriochlorophyll a. Chem Phys Lett 118: 48–54. Dismukes CG and Siderer Y (1981) Intermediates of a polynuclear manganese center involved in photosynthetic oxidation of water. Proc Natl Acad USA 78: 274–278. Disselhorst JAJM, van der Meer H, Poluektov OG and Schmidt J (1995) J Magn Res (to be published). Edmonds DT (1977) Nuclear quadrupole double resonance. Phys Rep 4: 233–290. Evelo RG, Hoff AJ, Dikanov SA and Tyryshkin AM (1989) An ESEEM study of the oxidized electron donor of plant photosystem II: evidence that D is a neutral tyrosine radical. Chem Phys Lett 161: 479–484. Flanagan HL and Singel DJ (1987) Analysis of ESEEM patterns of randomly oriented solids. J Chem Phys 87: 5606–5616. Gilchrist ML, Lorigan GA and Britt RD (1992) Pulsed electron paramagnetic resonance studies of calcium-depleted photosystem II membranes. In: Murata N (ed) Research in Photosynthesis, pp 317–320. Kluwer, Dordrecht. Gilchrist ML, Ball JA, Randall DW and Britt RD (1995)

Electron spin echo methods Proximity of the manganese cluster of photosystem II to the redox active tyrosine Proc Natl Acad Sci USA 92: 9545–9549. Hahn EL (1950) Spin echoes. Phys Rev 80: 580–594. Hofer P, Grupp A, Nebenführ H and Mehring M (1986) Hyperfine sublevel correlation HYSCORE spectroscopy: a 2D ESR investigation of the squaric acid radical. Chem Phys Lett 132: 279–282. Hoff AJ, De Groot A, Dikanov SA, Astashkin AV and Tsvetkov YD (1985) Electron spin echo envelope modulation spectroscopy (ESEEM) of the radical cations of and bacteriochlorophyll a. Chem Phys Lett 118: 40–47. Hoffman BM, DeRose VJ, Doan PE, Gurbiel RJ, Houseman ALP and Telser J (1993) Metalloenzyme active-site structure and function through multifrequency CW and pulsed ENDOR. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, Vol 13, pp 151–218. Plenum, New York. Käss H, Rautter J, Bönigk B, Hofer P and Lubitz, W (1995) 2D ESEEM of the radical cations of bacteriochlorophyll a and of the primary donor in reaction centers of Rhodobacter sphaeroides. J Phys Chem 99: 436–448. Kevan L (1979) Modulation of electron spin-echo decay in solids. In: Kevan L and Schwartz RN (eds) Time Domain Electron Spin Resonance, pp 279–341. John Wiley and Sons, New York. Kurreck H, Kirset B and Lubitz W (1988) Electron Nuclear Double Resonance Spectroscopy of Radicals in Solution. VCH Verlag, Weinheim. Lai A, Flanagan HL and Singel DJ (1988) Multifrequency electron spin echo envelope modulation in S = 1/2, I = 1/2 systems: Analysis of the spectral amplitudes, line shapes, and linewidths. J Chem Phys 89: 7161–7166. Lorigan GA and Britt RD (1994) Temperature dependent pulsed EPR relaxation studies of the state multiline signal of the photosynthetic oxygen-evolving complex. Biochemistry 33: 12072–12076. McCracken J, Shin DH and Dye JL (1992) Pulsed EPR studies of polycrystalline cesium hexamethyl hexacyclen sodide. Appl Magn Reson 3: 305–316. Mims WB (1965) Pulsed ENDOR experiments. Proc Royal Soc London 283: 452–457. Mims WB (1972a) Electron spin echoes. In: Geschwind S (ed) Electron Paramagnetic Resonance, pp 263–351. Plenum Press, New York. Mims WB (1972b) Envelope modulation in spin-echo experiments. Phys Rev B 5: 2409–2419. Mims WB (1972c) Amplitudes of superhyperfine frequencies displayed in the electron-spin-echo envelope. Phys Rev B 6: 3543–3545. Mims WB and Peisach J (1978) The nuclear modulation effect in electron spin echoes for complexes of and imidazole with and J Chem Phys 69: 4921–4930. Mims WB and Peisach J (1981) Electron spin echo spectroscopy and the study of metalloproteins. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, Vol 3, pp. 213–263. Plenum, New York. Mims WB and Peisach J (1989) ESEEM and LEFE of metalloproteins and model compounds. In: Hoff AJ (ed) Advanced EPR: Applications in Biology and Biochemistry, pp 1–57. Elsevier, Amsterdam.

253 Moenne-Loccoz P, Heathcote P, Maclachlan DJ, Berry MC, Davis IH and Evans MCW. (1994) Path of electron transfer in photosystem 1: Direct evidence of forward electron transfer from to Biochemistry 33: 10037–10042. Nishi N, Hoff AJ and van der Waals JH (1980) Electron spin echo studies on chloroplasts. Spectral characteristics of electron transport components and light-induced transients. Biochim Biophys Acta 590: 74–88. Norris JR, Thurnauer MC and Bowman MK (1980) Advances in Biological and Medical Physics 17: 365–416. Prisner TF, Un S and Griffin RG (1992) Pulsed ESR at 140 GHz. Isr J Chem 32: 357–363. Prisner TF, Rohrer M and Mobius K (1994) Pulsed 95 GHz, high-field EPR heterodyne spectrometer with high spectral and time resolution. Appl Magn Reson 7: 167–183. Rigby SEJ, Nugent JHA and O’Malley PJ (1994) The dark stable tyrosine radical of photosystem 2 studied in three species using ENDOR and EPR spectroscopies. Biochemistry 33: 1734–1742. Sanakis Y, Petrouleas V and Diner BA (1994) Cyanide binding at the non-heme of the iron-quinone complex of photosystem II: at high concentrations, cyanide converts the from high (S = 2) to low (S = 0) spin. Biochemistry 33: 9922–9928. Singel DJ (1989) Multifrequency ESEEM. In: Hoff AJ (ed) Advanced EPR: Applications in Biology and Biochemistry, pp 119–133. Elsevier, Amsterdam. Slichter CP (1990) Principles of Magnetic Resonance, 3rd ed. Springer-Verlag, Berlin. Sturgeon BE and Britt RD (1992) Design of a sensitive pulsed EPR spectrometer with an 8 to 18 GHz frequency range. Rev Sci Instrum 63: 2187–2192. Tang XS, Diner BA, Larsen BS, Gilchrist ML, Lorigan GA, and Britt RD (1994) Identification of histidine at the catalytic site of the photosynthetic oxygen-evolving complex. Proc Natl Acad USA 91: 704–708. Thomann H and Bernardo M (1993) Pulsed electron nuclear double and multiple resonance spectroscopy of metals in proteins and enzymes. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, Vol 13, pp 275–322. Plenum, New York. Thurnauer MC and Clark C (1984) Electron spin echo envelope modulation of the transient EPR signals observed in photosynthetic algae and chloroplasts Photochem Photobiol 40: 381–386. Warncke K and McCracken J (1994) electron spin echo envelope modulation spectroscopy of strong, hyperfine coupling in randomly oriented paramagnetic systems. J Chem Phys 101: 1832–1841. Warncke K, McCracken J and Babcock GT (1994) Structure of the tyrosine radical in photosystem II as revealed by electron spin echo envelope modulation (ESEEM) spectroscopic analysis of hydrogen hyperfine interactions. J Am Chem Soc 116: 7332–7340. Weber RT, Disselhorst JAJM, Prevo LJ, Schmidt J and Wenkebach WT (1989) Electron spin echo spectroscopy at 95 GHz. J Magn Res 81: 129–144. Zimmermann JL, Boussac A and Rutherford AW (1993) The manganese center of oxygen-evolving and calcium-depleted photosystem II: a pulsed EPR spectroscopy study. Biochemistry 32: 4831–4841.

This page intentionally left blank

Chapter 16 ENDOR Spectroscopy Wolfgang Lubitz* and Friedhelm Lendzian Max-Volmer-lnstitut für Biophysikalische und Physikalische Chemie, Technische Universität Berlin, Straße des 17. Juni 135, D-10623 Berlin, Germany

Summary I. Introduction II. Principles of Electron–Nuclear Multiple Resonance Spectroscopy A. ENDOR in Liquid Solution 1. Spin Hamiltonian, Energies and Transition Frequencies 2. Phenomenological Description of the ENDOR Experiment 3. TRIPLE Resonance 4. ENDOR-lnduced EPR B. ENDOR in the Solid State 1. ENDOR in Frozen Solutions a. Isotropic g b. Anisotropic G-Tensor (Orientation Selection) 2. ENDOR in Single Crystals C. ENDOR of Triplet State Molecules D. New Techniques III. Selected Applications of ENDOR to Photosynthesis A. Pigment Radicals in Vitro B. Radical Ions in Photosynthetic Reaction Centers C. ENDOR of Triplet States IV. Concluding Remarks Acknowledgements References

255 256 258 258 258 259 260 262 262 263 263 263 265 265 267 268 268 268 271 272 272 272

Summary The basic principles of electron-nuclear multiple resonance techniques, like ENDOR and TRIPLE resonance, are described as applied to paramagnetic molecules in liquid and frozen solutions and in single crystals. The advantages of these techniques as compared with conventional EPR are discussed. ENDOR and TRIPLE resonance allows the detection of the electron-nuclear hyperfine coupling constants (hfcs) of large paramagnetic molecules in a complex surrounding, even in cases where the EPR spectrum is completely unresolved. From the assigned hfcs a map of the valence electron spin distribution over the molecule is obtained. When applied to solid state samples – and in particular to single crystals – the full hyperfine tensors of the various magnetic nuclei can be determined. Such measurements yield additional information about the spatial structure of a paramagnetic center. New developments with respect to very high frequencies (high field ENDOR), time resolution (pulsed EPR/ENDOR), and new detection schemes (stochastic ENDOR) are briefly described. Selected applications of ENDOR to photosynthesis are discussed, which focus on isolated pigment radicals and the radical ions and triplet states created during the charge separation process in bacterial and plant reaction centers. *Correspondence: Fax: 49-30-31421122; E-mail: lubitz @echo.chem.tu-berlin.de

255 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 255–275. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

256

Wolfgang Lubitz and Friedhelm Lendzian

Recent applications of the technique to reaction center single crystals and to the structural characterization of genetically modified reaction centers are highlighted. Abbreviations: (B)Chl – (bacterio)chlorophyll; (B)Ph – (bacterio)pheophytin; cw – continuous wave; ENDOR – electron nuclear double resonance; EPR – electron paramagnetic resonance; ESE – electron spin echo; ESEEM – electron spin echo envelope modulation; hfc – hyperfine coupling constant; hfs – hyperfine structure; HOMO – highest occupied molecular orbital; LUMO – lowest unoccupied molecular orbital; MO – molecular orbital; mw – microwave; P – primary donor (BChl-dimer); absorbing at 865 nm; of the dimer bound to the protein L-subunit; of the dimer bound to the protein M-subunit; RC – reaction center; rf – radio frequency; RHF-INDO/SP – restricted Hartree Fock-intermediate neglect of differential overlap/spin polarization; SCF – self consistent field; TRIPLE – electron nuclear nuclear triple resonance

I. Introduction In reaction centers of plants, algae and photosynthetic bacteria light-induced charge separation generates cation and anion radicals of the cofactors involved in this process. The method of choice to identify and characterize these species is electron paramagnetic resonance (EPR). In the past, this technique has been used extensively to identify various paramagnetic states in vivo simply by comparison of their spectral parameters (mainly g factors and linewidths) with those of model compounds in vitro (Norris et al., 1971; Feher and Okamura, 1978). The application of EPR techniques to bacterial reaction centers has been reviewed by Hoff (1993). EPR has also been used to study other paramagnetic species in photosynthetic systems like triplet states (Budil and Thurnauer, 1991) or transition metal complexes (Miller and Brudvig, 1991; Debus, 1992) and even protein-derived radicals (Barry, 1993; Hoganson and Babcock, 1992; Boussac et al., 1990). Many of the organic radicals can be classified as which have the unpaired electron delocalized in the of the conjugated system. The electron spin interacts with the various magnetic nuclei in the radical by magnetic dipole–dipole and contact interactions leading to the hyperfine structure (hfs) of the spectra. After spectroscopic assignment of the measured hyperfine coupling constants (hfcs) to the various nuclei a map of the electron spin density distribution over the molecule is obtained. This can then be compared with calculations of the spin density

distribution by advanced molecular orbital (MO) methods (Plato et al., 1991). These data are the basis for a profound understanding of the electron transfer processes and for the function of the reaction center complex in photosynthesis. However, EPR spectroscopy of biological systems has some serious problems: (i) very often the hfs patterns are not resolved due to interaction of the electron spin with too many different nuclei. From such spectra only the g-factor and the Gaussian envelope linewidth are available for the characterization of the radical. (ii) Several radical species are frequently present in the sample which lead to overlapping spectra. (iii) Serious line broadening occurs due to immobilization of the radical through binding to a large biopolymer. To overcome at least some of the aforementioned limitations, more advanced methods must be applied, in particular multiple resonance techniques. A typical example is Electron Nuclear DOuble Resonance (ENDOR), which was introduced by Feher (1956) in the solid state and almost a decade later extended to radicals in solution by Hyde and Maki (1964). In ENDOR spectroscopy the NMR transitions of magnetic nuclei in paramagnetic systems are detected via intensity changes of a simultaneously irradiated EPR line. The inherently higher resolution of ENDOR yields detailed information about the hyperfine interactions, even when the EPR spectrum is completely unresolved. An example is given in Fig. 1, which shows both EPR and ENDOR spectra of the bacteriochlorophyll a cat-

ENDOR spectroscopy

257

258 ion radical in isotropic solution. The hfs is not resolved in EPR, whereas the ENDOR spectrum directly yields the hfcs of the proton and nitrogen nuclei. The electron spin density distribution of a radical is most easily obtained in liquid solution via determination of the isotropic hfcs, i.e., by liquid state ENDOR. Unfortunately, many biological systems can only be studied in frozen solutions (powders or glasses), since the radical intermediates often have to be stabilized by freeze trapping. Furthermore, the systems are often so large that anisotropic interactions are not averaged out by tumbling in solution, which leads to powder spectra even at ambient temperature. In principle, powder ENDOR allows the determination of anisotropic and isotropic hfcs, although the spectral analysis is much more elaborate than in the liquid phase. If single crystals are available, the full hyperfine (hf) tensors of the various nuclei can be determined by ENDOR. ENDOR has been extended to electron–nuclear–nuclear TRIPLE resonance (Möbius and Biehl, 1979) in order to improve the sensitivity and resolution, and to measure the relative number of nuclei contributing to a resonance and the relative signs of the hfcs. In this chapter we want to describe continuous wave (cw) ENDOR and related techniques in some detail and illustrate the applications of the methods by selected examples from the field of photosynthesis. Here, we shall focus on the radical ions generated in the charge separation process. The basic principles of EPR spectroscopy which are not described here can be found in Chapter 14 (Levanon) and several monographs (Atherton, 1993; Weil et al., 1994, Poole, 1983). For a more elaborate description of ENDOR spectroscopies the reader is referred to some review articles (Möbius and Lubitz, 1987; Möbius et al., 1989; Hüttermann, 1993; Hoffmann et al., 1993) and books (Kevan and Kispert, 1976; Kurreck et al., 1988; Poole and Farach, 1994) which also present applications to other systems, e.g. to transition metal complexes in biological systems. Extensions of EPR and ENDOR to higher mwfrequency bands are reviewed by Möbius (1993). The important application of electron spin echo (ESE) techniques like ESEEM and Pulsed ENDOR to determine hyperfine interactions in

Wolfgang Lubitz and Friedhelm Lendzian paramagnetic species related to photosynthesis are described by Britt (Chapter 15) and other time-resolved EPR methods by Levanon (Chapter 14) in this monograph. II. Principles of Electron–Nuclear Multiple Resonance Spectroscopy

A. ENDOR in Liquid Solution 1. Spin Hamiltonian, Energies and Transition Frequencies For a paramagnetic molecule in fluid (isotropic) solution the static spin Hamiltonian contains the electron and nuclear Zeeman terms and the isotropic hyperfine coupling term. In a strong magnetic field, it is given to first order by:

where g is the electronic g-factor; the Bohr magneton; and are the nuclear g-factors and magneton, respectively; and are the zcomponents of the electron and nuclear spin operators along the direction of the magnetic field is the respective isotropic hyperfine coupling constant (in frequency units), and h is Planck’s constant. The sum i runs over all nuclei with in the radical. The corresponding energy eigenvalues to first order, classified by the electronic and nuclear magnetic quantum numbers, and are given by

where and –I + 1 , . . . + I ) . The resulting first order spin energy level system for the simplest case of one electron coupled to one nucleus is shown in Fig. 2. According to the selection rules and two EPR transitions are obtained at frequencies centered around the electron Zeeman frequency, In a conventional cw-EPR experiment, the microwave

ENDOR spectroscopy

frequency, conditions

is kept constant and the resonance

are adjusted by sweeping the magnetic field Consequently, the magnitudes of hyperfine coupling constants, a, obtained from EPR are usually given in magnetic field units (millitesla, mT), which may be converted to frequency units by use of the relation: where is the hyperfine coupling in mT. In an NMR experiment on this spin system (selection rule and also two transitions are obtained with frequencies which are now centered around the nuclear Zeeman frequency In this simple spin system of Fig. 2 no resolution enhancement of the NMR is obtained as compared with EPR. However, the number of allowed EPR transitions increases multiplicatively with the number of magnetically inequivalent nuclei coupled to the electron; whereas, each set of magnetically equivalent nuclei contributes only two lines to the NMR spectrum (Kurreck et al., 1988). Here, lines of different nuclei (e.g.

259

and are grouped around their respective nuclear Larmor frequencies, which further improves the resolution and helps with the assignment. In radicals derived from large biomolecules this leads to a significant resolution enhancement of NMR (or ENDOR) as compared with EPR. From the ENDOR spectrum of the BChl a cation radical (Fig. 1C) 12 proton hyperfine coupling constants (hfcs) and four hfcs can be obtained; the corresponding number of ENDOR lines is 24 for the and 8 for the From these hfcs more than EPR lines are calculated. The resulting EPR spectrum (Fig. 1A) therefore consists only of a Gaussian envelope from which the hyperfine information cannot be obtained. 2. Phenomenological Description of the ENDOR Experiment The sensitivity of conventional NMR experiments on paramagnetic molecules in fluid solution is rather low due to the small population differences of nuclear Zeeman levels and the large linewidth resulting from the interaction with the unpaired electron(s). In an ENDOR experiment the NMR transitions are therefore detected via intensity changes of a partially saturated EPR transition (e.g. see Fig. 2). The intensity

260 of this “pumped” transition is limited by all electron and nuclear relaxation rates which connect these two levels (directly or via levels 4 and 2, Fig. 2). If now one NMR transition (e.g. is pumped also with saturating power, the observed EPR transition is effectively desaturated, which leads to an increase of the EPR signal intensity. The magnitude of the observed ENDOR effect is typically only a few percent of the EPR signal intensity. ENDOR is, however, several orders of magnitude more sensitive than NMR on a paramagnetic system. The ENDOR response depends on the delicate interplay of the various electron and nuclear induced rates and relaxation rates. For the case of liquid solution, this has been extensively investigated for various magnetic nuclei both experimentally and theoretically, considering all relevant electron and nuclear relaxation mechanisms (Freed, 1979; Plato et al., 1981, see also Kurreck et al., 1988). In liquid solution, besides internal dynamics, Brownian rotational diffusion of the molecules modulates the various interactions in the radical and is, therefore, responsible for the relaxation processes. It enters into the equations for the relaxation rates in form of the rotational correlation time obtained from the Debye– Einstein relation in which is the effective molecular volume, and T are the viscosity and temperature of the solvent and k is the Boltzmann constant. Optimum ENDOR signals are obtained when no “relaxation bottleneck” exists in the spin system, i.e., (neglecting cross relaxation rates see Fig. 2). For organic radicals in liquid solution one often finds and (Plato et al., 1981). Therefore, the rates may be equalized by changing the temperature or viscosity of the solvent. However, the different nuclei in the molecule generally have different optimum values, since they have different isotropic and anisotropic hyperfine couplings which determine the relaxation rates (Plato et al., 1981; Kurreck et al., 1988). Therefore, the obtained ENDOR effect is different for each nucleus and does not reflect the number of contributing equivalent nuclei but rather reflects the different relaxation behavior.

Wolfgang Lubitz and Friedhelm Lendzian Isotropic EPR and ENDOR spectra are only obtained in the limit (Redfield, 1965) where denotes the amplitude of the timedependent anisotropic magnetic interaction (g or hyperfine anisotropy in energy units). For organic radicals in common solvents is in the order of to seconds and relation (7) is valid in most cases. However, for a large protein like the photosynthetic reaction center (RC) with an estimated molecular weight of daltons, the value is s at room temperature in water and only small hf anisotropies MHz) are effectively averaged out (Lendzian et al., 1981). For large proteins or highly aggregated species solid state EPR and ENDOR spectra are expected even in liquid aqueous solution at ambient temperature due to the slow tumbling motion. For further information and experimental details of ENDOR, the reader is referred to Kevan and Kispert (1976), Kurreck et al. (1988), Möbius et al. (1989), Atherton (1993), Poole and Farach (1994). 3. TRIPLE Resonance For biological samples, the relaxation rates often cannot be adjusted and leading to a very small ENDOR effect. In this situation, one can short-circuit the “nuclear relaxation bottleneck” by a second strong rf field. In this electron– nuclear–nuclear triple resonance experiment (called “Special TRIPLE”), the two rf fields are tuned to drive both transitions, and of the same nucleus simultaneously (see Fig. 2), thereby enhancing the efficiency of the relaxation bypass and hence the signal intensity (Möbius and Biehl, 1979). The second advantage of Special TRIPLE is that the EPR desaturation becomes independent of nuclear relaxation, i.e., the line intensities now reflect the number of nuclei involved in a transition. TRIPLE lines can therefore be assigned more easily than ENDOR lines. Special TRIPLE also has the advantage of somewhat higher resolution than ENDOR. In Fig. 3, a Special TRIPLE spectrum is compared schematically with an ENDOR spectrum. Both spectra are typically recorded in the first-

ENDOR spectroscopy

derivative mode using frequency modulation of the rf field (Kurreck et al., 1988). The Special TRIPLE looks like an “ENDOR half spectrum” with its origin at the free proton frequency, The intensity is, however, higher (the exact factor depends on experimental conditions) and the line intensities reflect approximately the relative number of contributing protons. Electron–nuclear–nuclear triple resonance can be generalized to include more than one nucleus, for example, two inequivalent protons (Fig. 3). In this “General TRIPLE” experiment, the ENDOR spectrum is swept with the first modu-

261

lated rf field while a second, fixed (unmodulated) rf field, is applied to pump a specific NMR transition. In general, all NMR transitions in the same electron spin manifold (characterized by the same value) as the additionally irradiated one will decrease in intensity, whereas those in the other manifold will increase. Since the assignment of a particular high- or low-frequency transition to an value is determined by the sign of the hfc, the observed intensity changes in the General TRIPLE spectrum can be used to determine the relative signs of the hfcs in the radical. In a heteronuclear TRIPLE experiment, the sign of

262 must be known (Kurreck et al., 1988). In Fig. 3 (top), such a General TRIPLE spectrum is displayed schematically in which line 2 is additionally pumped. This leads to an enhancement of the respective low-frequency line 2' (Special TRIPLE effect). For the line pair 1,1', the high-frequency line is increased and the low-frequency line is decreased in intensity. Thus, the respective hfcs and have opposite signs. Absolute signs can be obtained from theory or measured directly in NMR experiments (De Boer and MacLean, 1966; see also Lubitz, 1991). The signs of the hfcs are important for theoretical reasons and are helpful for assigning the couplings to molecular positions. A more thorough theoretical description of TRIPLE resonance can be found in (Möbius and Biehl, 1979; Möbius et al., 1989; Kurreck et al., 1988).

Wolfgang Lubitz and Friedhelm Lendzian ecules. In the case of transition metal ions cryogenic temperatures are usually required for ENDOR detection due to the short electron spin relaxation times. There exist several excellent recent review articles covering this field which demonstrate the ability of ENDOR to gain information in particular about the ligand structure (Schweiger, 1982; Hoffmann et al., 1993; Hüttermann, 1993; Thomann and Bernardo, 1993). Here, we concentrate on paramagnetic organic doublet state molecules in frozen solutions and in single crystals; triplet states are handled in a subsequent section. In the solid state, the static spin Hamiltonian of a radical contains, in addition to the terms in Eq. (1), all angular dependent traceless interactions and is given by:

4. ENDOR-lnduced EPR Occasionally different radical species may be present in the same sample. The EPR of such a mixture, especially when the g values are similar, often consists of a superposition of different spectra which are difficult or impossible to separate. Since the density of spectral lines is much smaller in ENDOR than in EPR, it is very likely that the different species have non-overlapping ENDOR lines. In this case the different EPR spectra can be recorded separately using ENDOR-induced EPR (EIE), first reported by Hyde (1965). In this experiment the intensity of an ENDOR line belonging to one species is monitored while sweeping the magnetic field over the EPR spectrum. Thereby, it is often necessary to sweep also the rf frequency, since the nuclear Zeeman frequency, is proportional to Some examples for EIE are given by Kurreck et al. (1988). A special application of EIE is in single crystals, where it offers the possibility to separate overlapping EPR spectra of different sites.

B. ENDOR in the Solid State In the solid state ENDOR spectroscopy has a much larger field of applications than in liquids. This includes practically all paramagnetic states ranging from transition metal complexes to immobilized organic radicals and triplet state mol-

Here, G is the g-tensor and is the hyperfine (hf) tensor of nucleus i containing the isotropic part (i = x,y,z) and the traceless dipolar part with elements and (i = x,y,z). is the quadrupole coupling tensor which describes the interaction of the nuclear electric quadrupole moment (for with the electric field gradient at the nucleus (Atherton, 1993). Even to first order, the expressions for the energy eigenvalues and EPR and NMR transition energies are rather complicated for the general case. This results mainly from the fact that the nuclear spin quantization is in the resultant of the applied and hyperfine magnetic field and is in general no longer a “good quantum number” for the various eigenstates. A detailed treatment of the angular dependent energy eigenvalues and the EPR and ENDOR transition frequencies is given by Atherton (1993). For the case of isotropic g-factor and no nuclear quadupole interaction, the first order energy eigenvalues for the spin system are given by:

ENDOR spectroscopy where A is the hyperfine coupling tensor (frequency units), and are the electron and nuclear spin quantum numbers and is the unit vector along the field, with components the direction cosines of in the reference axis system. The first order NMR frequencies for this case are given for by:

with the high-and low-frequency NMR transitions. These two NMR (ENDOR) frequencies are in general no longer symmetric about Only in the limit of isotropic A or for small anisotropy of A this symmetry is retained and Eq. (10) reduces to

It is important to note that this equation holds also exactly when the field is along one of the principal axes of A (Atherton, 1993). This means that the turning points in a powder ENDOR spectrum directly yield the true principal values of A, provided g is isotropic. This is a good approximation for most organic radicals. In general, however, g-anisotropy has to be considered. The NMR frequencies for are then given by (Atherton, 1993)

and the G-tensor principal values and eigenvectors have to be known for an analysis of solid state ENDOR spectra. Only for coaxial tensors G and A, and for along one of the principal axes of G and A, can the respective true principal values of A be obtained directly from the ENDOR spectrum.

263 neglecting nuclear quadrupole interaction. For more details the reader is referred to the literature (Atherton, 1993; Hoffmann et al., 1993; Hüttermann, 1993; Thomann and Bernardo, 1993).

a. Isotropic g This is usually a good approximation for organic radicals studied in X-band (microwave frequency 9.5 GHz). In large molecules the hyperfine interaction with many nuclei often leads to an unresolved Gaussian envelope EPR spectrum. This case is shown schematically in Fig. 4A. Molecules of all orientations, with respect to the magnetic field, contribute to the ENDOR spectrum obtained at the center of the EPR line. The turning points of this powder ENDOR spectrum yield directly the principal values of the respective nuclear hyperfine tensors (Fig. 4A, bottom). A special case is of purely dipolar hf tensors, which are often obtained from matrix protons of the surrounding medium, e.g. in hydrogen bonds. In the point dipole approximation the angular dependent dipolar hyperfine coupling is given by:

where is a constant depending on the respective nuclear magnetic moment, is the electron spin density at the contact position, r is the magnitude of the distance vector between electron and nuclear spin and is the angle between this vector and the direction of the magnetic field. From Eq. (13) an axially symmetric tensor is obtained with from which the distance r can be derived. For short distances and extended this simple approach is, however, insufficient and isotropic contributions to A have to be considered.

1. ENDOR in Frozen Solutions In frozen solutions all the angular dependent spectral contributions sum up to a so-called powder spectrum (Atherton, 1993; Weil et al., 1994). Here, we discuss qualitatively two limiting cases in order to demonstrate the resolution enhancement of ENDOR obtainable in frozen solutions. Thereby, we restrict ourselves to the case

b. Anisotropic G-Tensor (Orientation Selection) If the g-anisotropy is larger than the hyperfine couplings in the spin system, it will dominate the EPR spectrum and can be used to select specific orientations of the molecules with respect to the magnetic field. This is demonstrated sche-

264

matically in Fig. 4B for the case of an axially symmetric G tensor. If ENDOR spectra are recorded at the outer turning points of the EPR spectrum and “single crystal-like” spectra are obtained (Rist and Hyde, 1970). This leads to a greately enhanced spectral resolution and for the case of coaxial G and A tensors the separation of the ENDOR lines corresponds directly to and respectively (Fig. 4B, bottom). For any other position in the EPR spectrum usually many different orientations contribute to the ENDOR response and the spectra become

Wolfgang Lubitz and Friedhelm Lendzian

more complicated. With the help of simulation programs (Hüttermann, 1993; Hoffmann et al., 1993) it is, however, possible to elucidate principal values and relative orientations of principal axes of G and A from combined EPR and ENDOR epxeriments in frozen solutions. There are many reports in the literature in which, for example, the ligand structure of transition metal complexes and metalloenzymes has been determined in this way (Schweiger, 1982; Hoffmann et al., 1993; Hüttermann, 1993; Thomann and Bernardo, 1993).

ENDOR spectroscopy 2. ENDOR in Single Crystals The most detailed information about the hyperfine structure is obtained from single crystal studies. In the case of small molecular systems (organic radicals, transition metal complexes) it is usually necessary to incorporate the paramagnetic species in a diamagnetic host crystal in order to avoid spin–spin interactions between neigboring centers. In crystals of proteins or metal enzymes the paramagnetic centers are often far enough separated by the protein, and such interactions become sufficiently small to be neglected. In general, ENDOR spectra have to be obtained for rotation of the magnetic field in the crystallographic planes of the crystal. The “rotation diagrams” of the NMR transitions, and must be analyzed according to Eq. (12) using the g-tensor previously determined by EPR. The analysis yields the hyperfine tensor A in the crystal axis system. Diagonalization of A then gives the principal values and the orientations of the tensor axes in the crystal axis system. If G can be approximated to be isotropic and the hyperfine anisotropy is small Eq. (12) reduces to Eq. (11). Here, the two NMR transitions and are centered around for any values of (orientation of in the chosen axis system). In this case, the symmetric displacement of and about allows the application of Special TRIPLE resonance (like in the case of an isotropic hfc, a). The Special TRIPLE frequency (deviation of from is then simply given by

for rotation of in the crystal i–j plane. Such a case is presented below in Fig. 5 (Lendzian et al., 1993). Different sites of the paramagnetic centers in the crystal unit cell may complicate the ENDOR spectra (“site splitting”) and may lead in some cases to ambiguities for the orientations of the principal axes of A (Atherton, 1993). In general, EPR and ENDOR in single crystals yield the G and hyperfine tensors, for all nuclei i, including the respective eigenvectors (di-

265 rection cosines of the principal axes). These contain information about the spatial structure of the molecule which may then be compared with the structure obtained, for example, by X-ray analysis (Scholes et al., 1982; Hutchinson 1992).

C. ENDOR of Triplet State Molecules The spin Hamiltonian for an electronic triplet state including hyperfine interaction is given by (Atherton, 1993)

where D is the zero field splitting tensor. In organic triplet states this term describes the dipolar interaction between the two electron spins which leads to an angular dependent splitting of the two EPR transitions: (i) and (ii) (see Chapter 14). In the high field approximation the first order energy levels show hyperfine splitting only for the and levels. The level is only split by the nuclear Zeeman interaction. Usually, the elements of D are much larger than the hyperfine terms. This can be used to obtain orientation selection of the molecules in frozen disordered solution via EPR in the same way as described above for anisotropic g. In the case of isotropic g, which is a good approximation for most organic triplet states, the NMR frequencies are given by Eq. (12). Considering the simple case where D and A are coaxial and the magnetic field is parallel to the principal axis of the z component of the zero field splitting tensor D the NMR frequencies in the high-field appproximation are: (i) for the EPR transition and (ii) for the EPR transition The two EPR transitions are separated by where D is the zero field splitting constant (see Chapter 14). All NMR lines corresponding to positive appear at frequencies below whereas, lines arising from negative lie above for the EPR transition (i), and vice versa for EPR transition (ii). Hence, when the sign of D is

266

Wolfgang Lubitz and Friedhelm Lendzian

ENDOR spectroscopy known, the absolute signs of the hfcs are obtained directly from the ENDOR spectra. There is also an NMR transition at the Larmor frequency for both EPR transitions. This behavior is demonstrated in ENDOR spectra for naphthalene triplet states in frozen solution (Kirste and van Willigen, 1982) and enables the determination of the relative signs of D and

D. New Techniques A crucial factor which determines the magnitude of the observed ENDOR effect in continuous wave (cw) spectroscopy is the balance of electron and nuclear relaxation rates and the rates induced by the coherently irradiated mw and rf fields. This is a serious drawback for many spin systems, where, for example, the relaxation rates cannot be easily adjusted by variation of the temperature or mw and/or rf saturation is a problem. In these cases pulsed techniques are clearly superior to cw-spectroscopy. An alternative experimental technique for resolving the hyperfine structure of inhomogeneously broadened EPR spectra is pulsed EPR, especially electron spin echo envelope amplitude modulation (ESEEM), which is described in Chapter 15. The observed modulation effects result from coherent excitation of “allowed” and “forbidden” EPR transitions by a strong microwave pulse, and the observed echo amplitude modulations depend on off-diagonal elements of the respective hf tensors, see e.g. Dikanov and Tsvetkov (1992) and Schweiger (1990). Large effects are observed, if these off-diagonal elements are comparable in magnitude to the nuclear Zeeman splitting, whereas small or even no modulation effects are observed in the case of small values. Therefore, this technique is mostly applied to nuclei with small magnetic moments (e.g. which in turn are difficult to detect in cw-ENDOR (Käß et al., 1995). This restriction does not exist if pulsed EPR is extended to Pulsed ENDOR. Several pulse schemes have been employed in this technique. In all cases no match of electron and nuclear relaxation rates is required, since the whole pulse sequence is employed in a time shorter than the relaxation times. In general, one (or two) rf pulses are applied in the preparation or mixing pe-

267 riod between a series of microwave (mw) pulses and the amplitude of the finally detected electron spin echo is observed as a function of the rf frequency. The frequencies of the NMR (ENDOR) transitions are the same as those for cw-ENDOR. There are several recent reviews about techniques and applications of Pulsed ENDOR (Dinse, 1989; Grupp and Mehring, 1990; Schweiger, 1991; Thomann and Bernardo, 1993). A more detailed description is given in chapter 15. For the study of transition metal complexes and metal enzymes the case is often met even at low temperatures, i.e., the electron and nuclear relaxation rates cannot be matched leading to a strongly reduced ENDOR effect. In order to avoid fast passage effects, which lead to distorted spectra and strongly reduced ENDOR intensities (Kevan and Kispert, 1976), the rf modulation frequency, used for detection, has to be smaller than For rad/s this reduces additionally the apparent signal intensities due to the noise contribution of the microwave spectrometer (Poole, 1983). This disadvantage can be overcome, if the NMR frequency is not stepped sequentially through the spectrum, but if frequency values for each step are selected stochastically from the entire range of the ENDOR spectrum. Thereby discrete NMR frequencies are assigned each to a separate channel and the ENDOR signal is accumulated for each channel. In this “stochastic ENDOR” experiment the NMR frequency channels can be selected with a much higher rate than in a conventional cw-ENDOR experiment which, in general, leads to a signficantly increased sensitivity. Details of this technique and some applications are reviewed by Brüggemann and Niklas (1994). Recently, EPR- and also ENDOR-spectrometers operating at 95 GHz (W band) have been developed in a few laboratories. These spectrometers offer a ten times better resolution of G components compared with the standard X band (9 GHz). In addition to other advantages, e.g. for pulsed EPR, the operation at higher fields enables the detection of orientationally selected ENDOR spectra also from frozen solutions of organic radicals (Rohrer et al., 1995). An overview of recent developments and applications of

268

high-field EPR and ENDOR has been given by Möbius (1993). III. Selected Applications of ENDOR to Photosynthesis

A. Pigment Radicals in Vitro EPR, ENDOR and TRIPLE resonance has been extensively used to study the electron spin density distribution of the radical cations and radical anions of bacteriochlorophylls, bacteriopheophytins, chlorophylls and pheophytins, which occur as electron carriers in the reaction centers of photosynthetic bacteria and plants. An overview of the field has been given by Fajer and Davis (1979) and more recently by Lubitz (1991). These studies yielded detailed maps of the electron distribution in the frontier orbitals (HOMO and LUMO) of these species which can also be calculated quite reliably by all valence electron SCF methods like RHF-INDO/SP and related MO methods (for a review see Plato et al., 1991). Such studies on the isolated pigment radical ions are important for a comparison with the same species occurring in the reaction centers of plants and bacteria. As an example, Fig. 1 shows the EPR and ENDOR spectra of the BChl a cation radical in liquid solution, which was first studied by Borg et al. (1976). In this species the unpaired electron is delocalized in the (HOMO) and interacts with the various magnetic nuclei in the molecule etc., see Fig. 1B). This leads to a very complex EPR spectrum (A) with almost no resolved hfs. The ENDOR spectrum of this species (C) contains much fewer resonance lines and directly yields 12 hfcs and 3 hfcs. The ENDOR lines are grouped around (near 14 MHz) and (1 MHz), respectively. All 4 nitrogen hfcs could be clearly resolved in a BChl (Lubitz et al., 1984; Käß et al., 1995). The relative signs of all hfcs were obtained by General TRIPLE resonance. Assignments of the hfcs to their specific molecular positions were achieved by several experimental approaches, including a comparison with results from related NMR methods, structurally similar radicals and, in particular, partially deuterated species using biosynthetic labeling and H/D exchange experiments (Lubitz, 1991; and references therein). The

Wolfgang Lubitz and Friedhelm Lendzian ENDOR spectrum of selectively deuterated BChl that carries protons only at the methyl groups (position 1a, 5a and 2b) and in positions to the is shown as an example in Fig. 1D. The 3 line pairs remaining in the ENDOR were assigned accordingly. The observed ENDOR spectrum was also detected, and ENDOR lines could be discriminated by the different temperature dependence of their intensities. The optimum temperature for the deuterons was ~ 230K, whereas the nuclei require much higher temperatures for their detection (270K). By use of the measured and assigned hfcs the slightly structured EPR line of BChl in solution could be perfectly simulated (see Fig. 1A).

B. Radical Ions in Photosynthetic Reaction Centers The application of magnetic resonance to photosynthetic reaction centers has been reviewed by Hoff (1993). Here, we focus on the radical ions generated in the RCs as studied by cw EPR, ENDOR, and TRIPLE resonance techniques. The EPR signal detected in illuminated bacterial RCs near g = 2.0026 is due to the primary electron donor cation radical The narrowing of the EPR linewidth of as compared with that of monomeric BChl formed the basis of the “special pair” hypothesis for this species (Norris et al., 1971) which proved to be correct in the RC crystal structure analysis of two bacteria (Deisenhofer and Michel, 1989; Feher et al., 1989). The details of the electronic structure of were unraveled entirely by ENDOR techniques. Here, in particular Special TRIPLE resonance was used because of its inherently higher sensitivity and resolution. A typical liquid solution spectrum is shown in Fig. 5 (left, bottom), for which 10 hfcs were obtained by spectral deconvolution. These hfcs are smaller than those detected in monomeric BChl (Fig. 1C), indicating a delocalization of the unpaired electron in a dimeric species (supermolecule). Selective deuteration of the macrocycles was again instrumental in determining that in RCs of Rb. sphaeroides (Lendzian et al., 1992) and in RCs of Rps. viridis (Lendzian et al., 1988) is indeed a BChl dimer and that the unpaired electron is asymmetrically distri-

ENDOR spectroscopy buted over the dimer halves. It is interesting that a symmetric dimer has been reported very recently for the primary donor in the green bacterium Chlorobium limicola based on EPR/ENDOR measurements (Rigby et al., 1994b). In several purple bacteria containing a BChl adimer the asymmetry is, however, highly conserved (Rautter et al., 1994). For the primary donor cation radical in PS I, (Käß et al., 1994) and in PS II, (Rigby et al., 1994a), chlorophyll dimers have been recently postulated from ENDOR data with a strongly asymmetric spin density distribution over the dimer halves. ENDOR-in-solution on allows the sensitive detection of even small changes in the surrounding of the primary donor via changes of the spin density distribution of the dimer. An example is shown in Fig. 6, which compares Special TRIPLE spectra of in RCs of the wild type and of two mutants of Rb. sphaeroides. In these mutants hydrogen bonds are introduced either to the 9-keto group of the L-side of the dimer or to the related 9-keto group on the M-side see Fig. 6, top. The hydrogen bonds influence the orbital energies of the two halves of the dimer and thereby the distribution of the unpaired electron. This leads to a predominant localization of the spin either on the L or the M side which is clearly seen in the spectra (Fig. 6, bottom). The effect is even more pronounced in several double mutants of the same type (Rautter et al., 1995). ENDOR and TRIPLE resonance have become important tools in the characterization of the effects of such mutations on structure and function of the bacterial and plant RCs. More detailed information concerning the electronic structure of can be obtained by studying this species not only in liquid and frozen solutions but also in RC single crystals (Lendzian et al., 1993). Fig. 5 (left) shows such Special TRIPLE spectra of in Rb. sphaeroides obtained with the external magnetic field, oriented along either of the 3 principal crystallographic axes a, b, and c. In the given crystal system (space group all four in the four RCs of the unit cell are magnetically equivalent for these 3 orientations. If the crystal is rotated in one crystallographic plane, e.g. in the ac-plane, a splitting in 2 sites (I and II) is observed. For this plane the respective angular dependence of the hfcs is

269

shown in Fig. 5 (right). Experiments in the other two planes (ab, bc) finally yield a complete set of hf tensors for 8 nuclei from which not only the magnitude of the anisotropic hfcs are obtained but also the orientations of the tensor principal axes with respect to the crystal axes system. The latter contain detailed information about the spatial structure of the radical, since the tensor axes are related to the electron spin distribution and this is, in turn, related to the bonding structure of the molecule. In the axes information has been used advantageously to assign the hfcs to the various nuclei on the L and the M halves of the dimer (Lendzian et al., 1993). ENDOR studies on in single crystals of mutant RCs elucidate electronic and structural changes induced by specific mutations (Huber et al., 1994). For other systems with unknown spatial structure this method offers the possibility to determine the orientation of the spin carrying molecule in the protein. This has been shown for the primary donor cation radical in single crystals of photosystem I (Käß et al., 1994) and for the primary quinone acceptor, in Zn-substituted RC single crystals of Rb. sphaeroides (Isaacson et al., 1995a, 1995b). In addition to the primary donor and the stable quinone acceptor the anion radicals of the intermediate electron acceptors, have also been extensively studied by ENDOR in several bacteria and in plant photosystem II (reviewed in Lubitz, 1991). In these systems, the observed species is the anion radical of either bacteriopheophytin or pheophytin, respectively. In Fig. 7 the frozen solution ENDOR spectrum of trapped in Photosystem II is compared with that of the anion radical of pheophytin a (Lubitz et al., 1989). The intense and narrow lines are assigned to methyl groups in positions 1a (line pairs and and 5a (line pairs and The large intensity of these lines is attributed to the free rotation of the methyl groups even in frozen matrices and the small anisotropy of the hfcs. The strong line at the free proton frequency is mainly due to weakly dipolar coupled protons from the surrounding protein matrix. An analysis of these matrix ENDOR lines can yield information about distances and orientations of these protons in the environment of a paramagnetic center (Kevan and Kispert, 1976). In the ENDOR spec-

270

Wolfgang Lubitz and Friedhelm Lendzian

ENDOR spectroscopy

271

trum of H/D exchange experiments (Lubitz et al., 1989) revealed hfcs belonging to exchangeable protons which can be assigned to a hydrogen bond between an amino acid (glutamate) and the 9-keto group of Ph a (Fig. 7, top). A point-dipole analysis of the respective hf tensor (Eq. (13)) yielded an H-bond length of Such a strong H bond could be one of the possible reasons for the observed differences of the methyl hf tensors in and in Ph (see Fig. 7). Such effects have also been observed for RCs of Rb. sphaeroides and Rps. viridis (for a review see Lubitz, 1991). Fairly strong and asymmetrical H bonds have also been detected by ENDOR for in Zncontaining RCs of Rb. sphaeroides in frozen solution (Feher et al., 1985). The electron spin density distribution of this species deviates significantly from that of and of the ubiquinone anion radical in frozen alcoholic solution. ENDOR allows the sensitive detection of these differences in the electronic structure which are related to the different functions of the two quinones in the electron transfer process in the RC.

C. ENDOR of Triplet States Magnetic resonance experiments on triplet states are generally performed in the solid state (single crystals or frozen solutions) mostly at cryogenic temperatures because of the large zero field splittings giving rise to fast relaxation rates in liquid solutions. There are several transient EPR studies of photochemically generated triplet states of chlorophylls and porphyrins utilizing the large spin polarization effects usually generated by the intersystem crossing from the excited singlet state to the excited triplet state of the molecule (Levanon and Bowman, 1993). Triplet states of chlorophylls occur also in all bacterial photosystems and in plant PS I and PS II, if the secondary electron acceptors are pre-reduced before illumination (Budil and Thurnauer, 1991; Miller and Brudvig, 1991). These triplet states have been characterized by optically detected magnetic resonance (ODMR) yielding the zero field splitting parameters (Hoff, 1989). However, this technique is usually applied in zero field where hyperfine interactions (except the nuclear quadrupole interaction) occur only in second order (Chapter 17).

272 There are several cw-ENDOR studies of the photoexcited triplet state of porphyrins in frozen solutions performed in high field (X-band) demonstrating the ability of this method to resolve hyperfine structure using orientation selection via the large zero field splitting (see e.g. Hamacher et al., 1993). Triplet states of the primary donors in photosynthesis have been studied extensively by X-band EPR in single crystals yielding detailed information about the orientation of the zero field splitting tensor axes with respect to the molecular structure (see e.g. Norris et al., 1989). However, no hyperfine information was obtained. In the only ENDOR study reported so far (Lendzian et al., 1985b) several proton hfcs of in Rb. sphaeroides R-26 could be resolved, but no definite assignment to specific nuclei was made. For the same species the nitrogen hfcs have been determined by ESEEM on RCs (de Groot et al., 1985). In both papers it was concluded that the triplet excitation is delocalized over the two BChl a molecules constituting the primary donor Determination of the spin density distribution in the triplet state of the primary donor in all photosystems would yield important information about the electron distribution in the LUMO of P. This is important for understanding the forward electron transfer reactions in the RCs.

Wolfgang Lubitz and Friedhelm Lendzian relative orientations of ligand atoms with respect to a metal center. In organic radicals the orientations of principal axes of hf tensors are often related to directions of covalent bonds. Very detailed information is obtained from single crystals and this type of spectroscopy has been termed “ENDOR-crystallography” (Hutchinson, 1992). The field of cw-ENDOR, in particular on organic radicals in liquid solution, is well established, certainly also because of the commercial availability of cw-ENDOR spectrometers for almost two decades. New developments and applications are expected for the time domain ENDOR techniques. In particular, in combination with pulsed laser excitation, in which photoreactions create spin polarization, ENDOR offers the possibility to resolve the hyperfine structure of transient species. This has been demonstrated by transient cw ENDOR experiments on organic radicals in liquid solution and photoexcited triplet states (Lendzian et al., 1985a; Kay et al., 1995). In the solid state, pulsed ENDOR techniques are expected to be advantageous. Applied to photosystems, these techniques offer the possiblity to resolve the hyperfine structure of short-lived radical ions and triplet states on a microsecond time scale. Thereby light-induced structural changes and their influence on the process of charge separation could directly be followed.

IV. Concluding Remarks In this chapter, we have described the basic principles of ENDOR and some applications of this technique to photosynthesis. Thereby, we have concentrated on continuous wave (cw) spectroscopy. Basically, two different types of information can be obtained from the measured hyperfine interactions: (i) Isotropic hfcs obtained from ENDOR yield, after proper assignment to molecular positions, a map of the spin density distribution in the orbital(s) occupied by the unpaired electron(s). This offers experimental values for wave function coefficients at the position of specific nuclei in the molecule. These are important values, e.g. for calculating rates in electron transfer reactions. This information is most easily obtained for radicals in liquid solution. (ii) Anisotropic hf tensors obtained from ENDOR contain structural information, e.g. about distances and

Acknowledgements The authors are grateful to Prof. K. Möbius (Freie Universität Berlin) and Prof. G. Feher (UC San Diego) and their former and present coworkers for many helpful discussions and for their contributions to the work reviewed in this chapter. Furthermore, we want to thank all members of our group at the TU Berlin who were involved in the presented work and who are cited in the references. Financial support from the Deutsche Forschungsgemeinschaft (Sfb 312) and Fonds der Chemischen Industrie (W.L.) is gratefully acknowledged. References Allen JP, Feher G, Yeates TO, Komiya H and Rees DC (1987) Structure of the reaction center from Rb. sphaero-

ENDOR spectroscopy ides R-26: The cofactors. Proc Natl Acad Sci USA 84: 5730–5734. Atherton NM (1993) Principles of Electron Spin Resonance. Horwood, Chichester. Barry BA (1993) The role of redox-active amino acids in the photosynthetic water-oxidizing complex. Photochem Photobiol 57: 179–188. Borg DC, Forman A and Fajer J (1976) ESR and ENDOR studies of the cation radical of bacteriochlorophyll. J Am Chem Soc 98: 6889–6893. Boussac A, Zimmermann J-L, Rutherford AW and Lavergne J (1990) Histidine oxidation in the oxygen-evolving photosystem-II enzyme. Nature (London) 347: 303–306. Brüggemann W and Niklas JR (1994) Stochastic ENDOR. J Magn Reson Series A 108: 25–29. Budil DE and Thurnauer MC (1991) The chlorophyll triplet state as a probe of structure and function in photosynthesis. Biochim Biophys Acta 1057: 1–41. de Boer E and MacLean C (1966) NMR study of electron transfer rates and spin densities in p-xylene and p-diethylbenzene anions. J Chem Phys 44: 1334–1342. de Groot A, Evelo R, Hoff AJ, de Beer R and Scheer H (1985) Electron spin echo envelope modulation (ESEEM) spectroscopy of the triplet state of the primary donor of and bacterial photosynthetic reaction centers and of and bacteriochlorophyll a. Chem Phys Lett 118: 48–54. Debus RJ (1992) The manganese and calcium ions of photosynthetic oxygen evolution. Biochim Biophys Acta 1102: 269–352. Deisenhofer J and Michel H (1989) The photosynthetic reaction centre from the purple bacterium Rhodopseudomonas viridis. EMBO J 8: 2149–2170. Dikanov SA and Tsvetkov YD (1992) Electron Spin Echo Envelope Modulation (ESEEM) Spectroscopy. CRC Press, Boca Raton. Dinse K-P (1989) Pulsed ENDOR In: Hoff AJ (ed) Advanced EPR. Applications in Biology and Biochemistry, pp 615– 631. Elsevier, Amsterdam. Fajer J and Davis MS (1979) Electron spin resonance of porphyrin cations and anions. In: Dolphin D (ed) The Porphyrins. Vol IV, pp 197–256. Academic Press, New York. Feher G (1956) Observation of nuclear magnetic resonances via the electron spin resonance line. Phys Rev 103: 834– 835. Feher G and Okamura MY (1978) Chemical composition and properties of reaction centers. In: Clayton RK and Sistrom WR (eds) The Photosynthetic Bacteria, pp 349–386. Plenum Press, New York. Feher G, Isaacson RA, Okamura MY and Lubitz W (1985) ENDOR of semiquinones in RC’s from Rhodopseudomonas sphaeroides. In: Springer Series in Chemical Physics 42: 175–189. Springer-Verlag, Berlin. Feher G, Allen JP, Okamura MY and Rees DC (1989) Structure and function of bacterial photosynthetic reaction centers. Nature (London) 339: 111–116. Freed JH (1979) Theory of multiple resonance and ESR saturation in liquids and related media. In: Dorio MM and Freed JH (eds) Multiple Electron Spin Resonance Spectroscopy, pp 73–142. Plenum Press, New York.

273 Grupp A and Mehring M (1990) Pulsed ENDOR spectroscopy in solids. In: Kevan L and Bowman MK (eds) Modern Pulsed and Continuous-Wave Electron Spin Resonance, pp 195–229. John Wiley and Sons, New York. Hamacher V, Wrachtrap J, von Maltzan B, Plato M and Möbius K (1993) EPR and ENDOR study of porphyrins and their covalently linked dimers in the photoexcited triplet state. Appl Magn Reson 4: 297–319. Hoff AJ (1989) Optically detected magnetic resonance of triplet states. In: Hoff AJ (ed) Advanced EPR. Applications in Biology and Biochemistry, pp 633–684. Elsevier, Amsterdam. Hoff AJ (1993) Magnetic resonance of bacterial photosynthetic reaction centers. In: Deisenhofer J and Norris JR (eds) The Photosynthetic Reaction Center. Vol II, pp 331– 382. Academic Press, San Diego. Hoffmann BM, DeRose VJ, Doan PE, Gurbiel RJ, Houseman ALP and Telser J (1993) Metalloenzyme activesite structure and function through multifrequency cw and pulsed ENDOR. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, EMR of Paramagnetic Molecules. Vol 13, pp 151–214. Plenum Press, New York. Hoganson CW and Babcock GT (1992) Protein-tyrosyl radical interactions in photosystem II studied by electron spin resonance and electron nuclear double resonance spectroscopy: Comparison with ribonucleotide reductase and in vitro tyrosine. Biochemistry 31: 11874–11880. Huber M, Isaacson RA, Abresch EC, Gaul D, Schenck CC and Feher G (1996) Electronic structure of the oxidized primary electron donor of the HL(M202) and HL(L173) heterodimer mutants of the photosynthetic bacterium Rhodobacter sphaeroides: ENDOR on single crystals of reaction centers, Biochim Biophys. Acta, in press. Hutchinson CA (1992) The study of structure by electron magnetic resonance methods. A brief history. Appl Magn Res 3: 219–255. Hüttermann J (1993) ENDOR of randomly oriented mononuclear metalloproteins: toward structural determinations of the prosthetic group. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, EMR of Paramagnetic Molecules. Vol 13, pp 219–250. Plenum Press, New York. Hyde JS and Maki AH (1964) ENDOR of a free radical in solution. J Chem Phys 40: 3117–3118. Hyde JS (1965) ENDOR of free radicals in solution. J Chem Phys 43: 1806–1818. Isaacson, RA, Abresch EC, Feher G and Lubitz W (1995a) ENDOR studies of in single crystals of reaction centers from Rb. sphaeroides. Biophys J 68(2): A246. Isaacson RA, Lendzian F, Abresch EC, Lubitz W and Feher G (1995b) Electronic structure of in reaction centers from Rhodobacter sphaeroides. I. Electron Paramagnetic Resonance in single crystals. Biophys J 69: 311–322. Käß H, Fromme P, Witt HT and Lubitz W (1994) and ESEEM of in single crystals of photosystem I from Synechococcus elongatus. Biophys J 66(2): A228. Käß H, Rautter J, Bönigk B, Höfer P, and Lubitz W (1995) 2D ESEEM of the radical cations of bacteriochlorophyll a and of the primary donor in reaction centers of Rhodobacter sphaeroides. J Phys Chem 99: 436–448. Kay CWM, Di Valentin M and Möbius K (1995) A time-

274 resolved Electron Nuclear Double Resonance (ENDOR) study of the photoexcited triplet state of free-base tetraphenylporphyrin. Solar Energy Materials and Solar Cells 38: 111–118. Kevan L and Kispert LD (1976) Electron spin double resonance spectroscopy. John Wiley and Sons, New York. Kirste B and van Willigen H (1982) ENDOR on photoexcited triplets randomly oriented in solid solution. Chem Phys Lett 92: 339–342. Kurreck H, Kirste B and Lubitz W (1988) Electron nuclear double resonance spectroscopy of radicals in solution. Application to organic and biological chemistry. In: Marchand AP (ed) Methods in stereochemical analysis, VCH, Weinheim. Lendzian F, Lubitz W, Scheer H, Bubenzer C and Möbius K (1981) In vivo liquid solution ENDOR and TRIPLE resonance of bacterial photosynthetic reaction centers of Rhodopseudomonas sphaeroides R-26. J Am Chem Soc 103: 4635–4637. Lendzian F, Jaegermann P and Möbius K (1985a) Time-resolved CIDEP-enhanced ENDOR in short-lived radicals. Chem Phys Lett 120: 195–200. Lendzian F, van Willigen H, Sastry S, Möbius K, Scheer H and Feick R (1985b) Proton ENDOR study of the photoexcited triplet state in Rb. sphaeroides R-26 photosynthetic reaction centers. Chem Phys Lett 118: 145–150. Lendzian F, Lubitz W, Scheer H, Hoff AJ, Plato M, Tränkle E and Möbius K (1988) ESR, ENDOR and TRIPLE resonance studies of the primary donor cation radical in the photosynthetic bacterium Rps. viridis Chem Phys Lett 148: 377–385. Lendzian F, Geßner C, Bönigk B, Plato M, Möbius K and Lubitz W (1992) and resonance of the primary donor cation radical in isotopically labeled reaction centers of Rb. sphaeroides In: Murata N (ed) Research in Photosynthesis, Vol I, pp 433–436. Kluwer Academic Publishers, Dordrecht. Lendzian F, Huber M, Isaacson RA, Endeward B, Plato M, Bönigk B, Möbius K, Lubitz W and Feher G (1993) The electronic structure of the primary donor cation radical in Rb. sphaeroides R-26: ENDOR and TRIPLE resonance studies in single crystals of reaction centers. Biochim Biophys Acta 1183: 139–160. Levanon H and Bowman MK (1993) Time-domain EPR spectroscopy of energy and electron transfer. In: Deisenhofer J and Norris JR (eds) The Photosynthetic Reaction Center, Vol II, pp 387–418. Academic Press, San Diego. Lubitz W (1991) EPR and ENDOR studies of chlorophyll cation and anion radicals. In: Scheer H (ed) Chlorophylls, pp 903–944. CRC Press, Boca Raton. Lubitz W, Isaacson RA, Abresch EC and Feher G (1984) Electron Nuclear Double Resonance of the primary donor cation radical in reaction centers of Rhodopseudomonas sphaeroides: additional evidence for the dimer model. Proc Natl Acad Sci USA 81: 7792–7796. Lubitz W, Isaacson RA, Okamura MY, Abresch EC, Plato M and Feher G (1989) ENDOR studies of the intermediate electron acceptor radical anion in Photosystem II reaction centers. Biochim Biophys Acta 977: 227–232. Miller A-F and Brudvig GW (1991) A guide to electron para-

Wolfgang Lubitz and Friedhelm Lendzian magnetic resonance spectroscopy of Photosystem II membranes. Biochim Biophys Acta 1056: 1–18. Möbius K (1993) High-field EPR and ENDOR on bioorganic systems. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, EMR of Paramagnetic Molecules, Vol 13, pp 253–271. Plenum Press, New York. Möbius K and Biehl R (1979) Electron–nuclear–nuclear TRIPLE resonance of radicals in solutions. In: Dorio MM and Freed JH (eds) Multiple Electron Spin Resonance Spectroscopy, pp 475–507. Plenum Press, New York. Möbius K and Lubitz W (1987) ENDOR spectroscopy in photobiology and biochemistry. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, Vol 7, pp 129–247. Plenum Press, New York. Möbius K, Lubitz W and Plato M (1989) Liquid-state ENDOR and TRIPLE Resonance. In: Hoff AJ (ed) Advanced EPR, Applications in Biology and Biochemistry, pp 441–494. Elsevier, Amsterdam. Norris JR, Uphaus RA, Crespi HL and Katz JJ (1971) Electron spin resonance of chlorophyll and the origin of signal I in photosynthesis. Proc Natl Acad Sci (USA) 68: 625– 628. Norris JR, Budil DE, Gast P, Chang C-H, El-Kabbani O and Schiffer M (1989) Correlation of paramagnetic states and molecular structure in bacterial photosynthetic reaction centers: The symmetry of the primary electron donor in Rhodopseudomonas viridis and Rhodobacter sphaeroides R-26. Proc Natl Acad Sci USA 86: 4335–4339. Plato M, Lubitz W and Möbius K (1981) A solution ENDOR sensitivity study of various nuclei in organic radicals. J Phys Chem 85: 1202–1219. Plato M, Möbius K and Lubitz W (1991) Molecular orbital calculations on chlorophyll radical ions. In: Scheer H (ed) Chlorophylls, pp 1015–1046. CRC Press, Boca Raton. Poole CP (1983) Electron Spin Resonance. A Comprehensive Treatise on Experimental Techniques, John Wiley and Sons, New York. Poole CP, Farach HA (eds) (1994) Handbook of Electron Spin Resonance. Data Sources, Computer Technology, Relaxation, and ENDOR, AIP Press, New York. Rautter J, Lendzian F, Lubitz W, Wang S and Allen JP (1994) Comparative study of reaction centers from photosynthetic purple bacteria. Electron paramagnetic resonance and electron nuclear double resonance spectroscopy. Biochemistry 33: 12077–12084. Rautter J, Lendzian F, Schulz C, Kuhn M, Lin X, Williams JC, Allen JP and Lubitz W (1995) ENDOR-studies of the primary donor cation radical in mutant reaction centers of Rhodobacter sphaeroides with altered hydrogen-bond interactions. Biochemistry 34: 8130–8143. Redfield AG (1965) The theory of relaxation processes. In: Waugh JS (ed) Advances in Magnetic Resonance, Vol 1, pp 1–32. Academic Press, New York. Rigby SEJ, Nugent JHA and O’Malley PJ (1994a) ENDOR and Special Triple Resonance studies of chlorophyll cation radicals in photosystem 2. Biochemistry 33: 10043–10050. Rigby SEJ, Thapar R, Evans MCW and Heathcote P (1994b) The electronic structure of The primary donor of the Chlorobium limicola fsp thiosulphatophilum photosynthetic reaction center. FEBS Lett 350: 24–28.

ENDOR spectroscopy Rist GH and Hyde JS (1970) Ligand ENDOR of Metal Complexes in Powders. J Chem Phys 52: 4633–4643. Rohrer M, Plato M, MacMillan F, Grishin Y, Lubitz W and Möbius K (1995) Orientation selected 95 GHz high-field ENDOR spectroscopy of randomly oriented plastoquinone anion radicals. J Magn Res Series A, 116: 59–66. Scholes CP, Lapidot A, Mascarenhas R, Inubushi T, Isaacson RA and Feher G (1982) Electron Nuclear Double Resonance (ENDOR) from heme and histidine nitrogens in single crystals of aquometmyoglobin. J Am Chem Soc 104: 2724– 2735. Schweiger A (1982) Electron Nuclear Double Resonance of Transition Metal Complexes with Organic Ligands. Structure and Bonding, Vol 51, pp 1–121. Springer-Verlag, Berlin.

275 Schweiger A (1990) New trends in pulsed electron spin-resonance methodology. In: Kevan L and Bowman MK (eds) Modern Pulsed and Continuous-Wave Electron Spin Resonance, Chapter 2, pp 43–118. John Wiley and Sons, New York. Schweiger A (1991) Pulsed electron spin resonance spectroscopy: Basic principles, techniques, and examples of applications. Angew Chem Int Ed Engl 30: 265–292. Thomann H and Bernardo M (1993) Pulsed electron nuclear double and multiple resonance spectroscopy of metals in proteins and enzymes. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance, EMR of Paramagnetic Molecules. Vol 13, pp 275–320. Plenum Press, New York. Weil JA, Bolton JR and Wertz JE (1994) Electron Paramagnetic Resonance. Elementary Theory and Practical Applications. John Wiley and Sons, New York.

This page intentionally left blank

Chapter 17 Optically Detected Magnetic Resonance (ODMR) of Triplet States in Photosynthesis Arnold J. Hoff Department of Biophysics, Huygens Laboratory, Leiden University, P.O. Box 9504, 2300 RA Leiden, The Netherlands

Summary I. Introduction II. The Triplet Spin Hamiltonian in Zero Magnetic Field III. Optical Detection of Magnetic Resonance, ODMR A. Principles 1. Time-Resolved ODMR a. Pulsed Microwaves b. Determination of the Average Decay Rate B. Instrumentation IV. Double Resonance A. Triplet-Minus-Singlet (T–S) Absorbance Difference Spectra B. Linear Dichroic T–S Spectroscopy 1. Instrumentation for LD-(T – S) Spectroscopy V. ODMR in Photosynthesis A. ADMR Spectroscopy of Reaction Centers B. T–S Spectroscopy of Reaction Centers C. Linear-Dichroic T–S Spectroscopy 1. Bacterial Reaction Centers 2. Plant Reaction Centers a. Photosystem I b. Photosystem II VI. Concluding Remarks References

277 278 278 279 279 281 281 281 282 284 285 286 287 288 288 289 290 290 292 293 293 295 295

Summary The triplet state of aromatic molecules and of polyenes is a versatile probe of molecular structure and of the interactions with the environment, through the zero-field splitting (ZFS) parameters and the sublevel decay rates. These triplet properties can be determined accurately with magnetic resonance. Optical detection of magnetic resonance (ODMR) is often advantageous because it pairs the frequency resolution of magnetic resonance with the sensitivity of optical Spectroscopy. It is preferentially carried out in zero magnetic field, where magnetic-field dependent anisotropies are absent, thus considerably enhancing the resolution and sensitivity for disordered systems. In photosynthesis, the optical parameters used for the detection of ODMR are the fluorescence (FDMR) and the absorbance (ADMR). Especially the latter is of interest, because it can be used for samples with low fluorescence yield, and allows an optical-microwave double resonance technique, in which the wavelength of detection is scanned, keeping the microwaves at a resonant frequency. Thus, Correspondence: Fax: 31-71-5275819; E-mail: [email protected]

277 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 277–298. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

278

Arnold J. Hoff

triplet-singlet absorbance difference (T–S) spectra can be recorded with superior accuracy and resolution. In this Chapter, the physical background of ODMR is outlined, followed by a few highlights of its application to the study of photosynthetic reaction centers. Recent studies employing ADMR-recorded linear-dichroic T–S spectroscopy are emphasized. Key results are the determination of the orientation of several transition moments in reaction centers of several bacteria and of the two plant photosystems. This has allowed insight in the configuration of the various primary donors, and has aided in the interpretation of the reaction center absorbance spectrum. A further important finding is that isolated reaction centers generally exhibit a heterogeneity of the optical and magnetic resonance parameters of the primary donor, which is attributed to a distribution of conformations of the reaction center pigments. Abbreviations: ADMR – absorbance-detected magnetic resonance; BChl – bacteriochlorophyll; BPh – bacteriopheophytin; Chl –chlorophyll; CT – charge transfer; D – primary electron donor; monomeric BChl associated with the active and non-active electron transport chain, respectively; FDMR – fluorescence-detected magnetic resonance; LD – linear dichroic; MI – microwave-induced; MIA – microwave-induced absorbance; MIF – microwave-induced fluorescence; MIP – microwaveinduced phosphorescence; ODMR – optically detected magnetic resonance; PDMR – phosphorescencedetected magnetic resonance; Pheo – pheophytin; PS – photosystem; P680 – primary donor of Photosystem II; P700 – primary donor of Photosystem I; RC – reaction center; T–S – triplet-minus-singlet; ZFS – zero-field splitting I. Introduction Optical detection of magnetic resonance (ODMR) is often advantageous because it pairs the frequency resolution of magnetic resonance with the sensitivity of optical spectroscopy. It is primarily used for magnetic resonance of the triplet state, as the resonant transfer of population between the triplet sublevels generally gives rise to a concomitant change in the optical properties (phosphorescence, fluorescence, absorption) of the sample, because of the different decay characteristics of each sublevel. It is preferentially carried out in zero magnetic field, where the triplet wave functions are not mixed by the Zeeman interaction and the triplet sublevels are split by the dipolar interaction between the two unpaired electrons (and to a lesser extent by electron – nuclear hyperfine interactions with nuclei of This enhances considerably the resolution and sensitivity for disordered systems. In photosynthesis, phosphorescence detection of ODMR is not feasible because of the very low quantum yield of phosphorescence. Detection by the fluorescence and especially by the absorption, however, is possible with quite good sensitivity. The information obtained is twofold: Firstly, the

zero-field splitting (ZFS) parameters and the sublevel decay rates can be determined with much better accuracy than with regular EPR. These parameters reflect the structure of the primary donor and its environment. Secondly, keeping the microwave frequency constant at one of the resonant values (there are generally three ODMR resonances corresponding to the three ways one can connect two out of three sublevels with resonant microwaves), one can scan the wavelength at which the resonance is detected. This gives the microwave-induced fluorescence or absorption spectrum. The latter corresponds to the optical triplet–singlet (T–S) difference spectrum, which in principle carries much information on the configuration of the cofactors. In this Chapter, the physical background of ODMR will be sketched briefly, followed by a few highlights of its application to the study of photosynthetic reaction centers (RCs). II. The Triplet Spin Hamiltonian in Zero Magnetic Field The triplet spin Hamiltonian without external magnetic field comprises interactions involving the magnetic moment of the electrons. These are

ODMR of triplet states

two-fold: spin–spin coupling and spin–orbit coupling. The main contribution to the spin–spin coupling operator, is the classical magnetic dipole–dipole interaction between two electrons:

with g the electronic g-value, the electronic Bohr magneton, the magnetic moments of the two electrons, r their distance vector, the permeability in vacuum, D a tensor operator whose elements consist of integrals over the coordinates of the electrons, the total spin angular momentum operator X, Y and Z the principal values of D and (u = x, y, z) the components of along the principal axes of D. Often, these axes coincide with the molecular symmetry axes. In a two-electron approximation the triplet wave functions can be written in symmetry-adapted form

where and are the eigenfunctions of the component of the spin operator along the z direction, The functions are eigenfunctions of with eigenvalues X, Y and Z. The have the property

Thus, there is no net magnetic dipole moment associated with any of the triplet substates in zero magnetic field: but there is a transition dipole moment present between any two of the triplet substates where is the gyromagnetic ratio of the electron and is Planck’s constant divided by Thus, in

279

zero magnetic field, population can be transferred from one triplet sublevel to another by applying a resonant electromagnetic field. From Eq. (4b) it follows that the transition probability is proportional to where is the amplitude of the magnetic component of the driving field, From (5) it follows that the microwave transition (u,s = x,y,z) is polarized with transition moment along w = u × s. This allows us to perform microwave-selection spectroscopy, to which we will turn later. The resonance frequencies follow from the eigenenergies of namely X, Y and Z. Because X + Y + Z = 0 (the trace of D is zero), it is customary to express the energies in two independent parameters D and E, the zero-field splitting (ZFS) or fine-structure parameters: with by convention The zero-field splitting parameters represent averages over the spatial coordinates x', y', z' of the distance vector r of the two unpaired electrons, E being a measure of the deviation from axial symmetry about the z-axis. The relative order of the energy levels depends on the sign of D and E. For a flat molecule such as chlorophyll, one would expect D to be positive. (The z-axis is the axial symmetry axis and is perpendicular to the plane of the molecule, so that the z' component of r is on average much smaller than For a rod-like molecule, such as a bi-radical, D will be negative. III. Optical Detection of Magnetic Resonance, ODMR

A. Principles Continuous illumination will generate an equilibrium population of the triplet sublevels given for light levels that are not too high by is the probability to transit from the singlet excited state to the u-th triplet sublevel with the decay rate that governs de-

280

Arnold J. Hoff

excitation from the u-th sublevel back to the singlet ground state, K the overall rate of populating the triplet state, the population of its u-th sublevel, its total population and N the number of photoexcitable molecules. Both and are determined by molecular symmetry. When a microwave field of a frequency corresponding to a transition between the y- and the z-level is switched on, the field will transfer population from the heavily-populated, slowly decaying zlevel to the much less populated, fast decaying ylevel. A new equilibrium will be established that for a strong enough microwave field is given by with and If we take it is then immediately seen that because for Although the x-level will sense the new equilibrium via the photogeneration cycle of the triplet state, this is a second order effect. The change in triplet population will lead to a change in phosphorescence or fluorescence intensity, or in the singlet ground state absorbance. For saturating microwaves it is given by

In the above we have the essence of ODMR. A sample is continuously illuminated at liquid helium temperatures, preferably below 2.1 K, and simultaneously irradiated by microwaves of a frequency v not far from that corresponding to one of the triplet sublevel spacings: and (Fig. 1). The frequency v is slowly scanned across one of the frequencies while either the phosphorescence, (delayed) fluorescence or absorbance of the sample is monitored. When v is close to or precisely equal to the ensemble of triplet state is in resonance with the microwave field and the fluorescence (FDMR), phosphorescence (PDMR) or absorbance (ADMR) will be enhanced, or diminished, depending on the relative values of and Note that the FDMR or ADMR signal intensity, is proportional to the square of the incident light intensity because one photon is used for producing the triplet state and a second for

probing the microwave-induced change in the optical parameter. An important advantage of ODMR in zero magnetic field compared to conventional EPR in high field, where the absorption of microwaves is monitored, is that the optical probing occurs with quanta of much higher energy than the microwave quantum. This enhances detector sensitivity enormously. Secondly, since the modes of detection and excitation are decoupled, one is insensitive to noise sources due to the microwaves (as e.g. amplitude fluctuation), especially when the transition is saturated. (Note that the optical signal does not disappear upon saturation as does the microwave absorption in microwave detection; this is another advantage of optical detection.) Thirdly, compared to high-field EPR of triplet states, in zero-field resonance one has much narrower lines and a concomitant increase in sensitivity, since the anisotropy in resonance condition associated with an applied magnetic field is absent. Finally, the possibility to probe the resonance at various wavelengths gives especially for ADMR much new information.

ODMR of triplet states 1. Time-Resolved ODMR The time dependence of the ODMR signal in response to a change in the condition of microwave irradiation (switching them on or off, or applying pulses) is given by the solution of the differential equations describing the time development of the four-level system simplified by the neglect of the population. The general analytic solution is given by Hoff and Cornelissen (1982); for the present discussion it suffices to note that the response of the system on switching on or off resonant microwaves connecting or or both is given by the sum of three exponentials

where and are rather cumbrous functions of the rates of decay, spin-lattice relaxation W and microwave-induced transitions, and the rate K (proportional to the light flux) of populating the triplet state. For and and in the absence of microwaves, the triplet sublevels are uncoupled and each sublevel decays after some perturbation according to

a. Pulsed Microwaves If one perturbs the four-level system slightly by a pulse of microwaves resonant between two triplet sublevels, u and v, the return to equilibrium is governed by an equation similar to Eq. (9) (see Hoff and Cornelissen (1982) for explicit relations). In the absence of spin-lattice relaxation and for low light fluxes and a pulse duration that is much shorter than the fastest triplet sublevel decay time, Eq. (9) reduces to a good approximation to two exponentials with characteristic rates given by

and with amplitudes of opposite sign. Provided K

281

is low enough, the third sublevel is not perturbed. Note that the extrapolation for is nontrivial: Relations 11a,b are valid only for really low light intensities, for which the signal-to-noise ratio is poor, especially for fluorescence detection. They are therefore not suitable to determine these probabilities can be better determined by measuring the relative amplitude of the ODMR signal when connecting two of the three triplet sublevels by saturating microwaves (see e.g. Hoff, 1989). The amplitude of the pulse response to first order in K follows from Eq. (10) for the appropriate boundary condition: and where f is a parameter describing the effect of the pulse on the population of sublevels u and v: for saturation (high power and/or long pulse) ,f= 1 for inversion of and (the maximal effect). Normally, With the change and thus the signal amplitude S(FDMR,ADMR), is to first order in K given by

b. Determination of the Average Decay Rate When saturating microwaves are applied simultaneously to two of the three ODMR transitions, the triplet collapses to one level with average decay rate The set of differential equations describing the simplified four-level system then reduces to those for an equilibrium reaction, which immediately yields

for the triplet built-up after the onset of illumination at t = 0. Eq. (13) permits the evaluation of k by extrapolating to Obviously, the same equation applies when the triplet levels are coupled not by microwaves but by spin-lattice relaxation at higher temperatures. In fact, the latter method to determine k is to be preferred, since especially for randomly-oriented samples the condition of microwave saturation is difficult to obtain.

282

B. Instrumentation The minimal requirements for an ODMR experiment are a source of light, a source of microwaves, a cryostat and a detector. Up-to-date discussions of these parts and the relevant electronics are given by Maki (1984) and Hoff (1989,1990,1993). Here we will limit ourselves to a few notes and a discussion of special arrangements for ADMR and LD-ADMR. A schematic diagram of our present ADMR set-up is shown in Fig. 2a. With small modifications it can be used for PDMR and FDMR as well. Crucial for high-sensitivity ADMR is a highintensity, low-noise light source. To this end the

Arnold J. Hoff

high-power tungsten-halogen lamp powered by a current-stabilized DC power supply, introduced by den Blanken et al. (1982), proved very suitable, and is now an essential part of all ADMR spectrometers. Mercury or xenon lamps are more intense, certainly in the UV region, but suffer from instabilities in the intensity I, often to a level of or worse. The same holds for laser excitation. Adequate filtering of infra-red radiation should be provided to avoid heating of the sample. When the excitation beam is broadbanded, it may also serve as probe beam at various wavelengths. Microwaves are supplied preferably by a sweep unit (available with a range of 0.1–27 GHz) with

ODMR of triplet states

283

284 provisions to scan the frequency over the desired range and to amplitude- or frequency-modulate the output. Usually, the output (20–40 mW) is enough for regular ODMR at liquid helium temperatures without amplification. For measuring the ODMR lines the microwaves are fed into a broad-band resonator, e.g. a helix, that can admit the frequency range of interest. When very broad ODMR lines need to be scanned, attention should be paid to frequency-dependent reflections of the microwaves (due primarily to mismatch between helix and conductor), which may considerably affect the field intensity inside the helix, and thus the ODMR intensity. Also, without leveling provisions, the sweeper output may considerably depend on frequency, even in a relatively narrow range. When in doubt that certain features are artificial, another helix of different physical characteristics should be used for comparison. If one is interested in the (probe) wavelength dependence in MI spectroscopy, the microwave frequency is set exactly at resonance and a narrow-band cavity may be used. We have used a loop-gap resonator (Hardy and Whitehead, 1981) with slots for optical access (Fig. 2b). Such a cavity has a much higher Q-value (ratio of stored energy and loss per cycle), and consequently a much higher field at the same incident power, than the helix. This is an important advantage when measuring kinetics with pulsed ODMR, because then and thus the signal amplitude S(FDMR,ADMR), is much increased. Alternatively, a microwave power amplifier may be used. A further advantage of the loop-gap resonator is that it provides a well-polarized microwave field, of which good use is made in LD-ADMR (Section IV B 1). The cryostat is preferably a double-walled helium bath cryostat equipped with at least two windows. Helium boils at 4.2 K; scattering by the bubbles then prohibits optical detection. By lowering the pressure above the helium bath the temperature of the helium can be lowered to below 2.1 K, the so-called lambda point, below which helium is superfluid and bubbles disappear completely. Thus, adequate pumping facilities should be provided. For FDMR at 4.2 K a simple He-vessel may be used in which a light pipe with sample compartment at the end is lowered (van der Bent et al., 1976). Such a light pipe may also used in a bath cryostat (Chan, 1982).

Arnold J. Hoff The optical detector depends on the optical mode. For phosphorescence and fluorescence a photomultiplier is used. For ADMR, one may employ a strong probe beam (in fact the excitation beam may serve as such), providing a sufficiently high number of transmitted quanta that a photodiode may be used (up to 1100 nm, silicium; 1100 nm – germanium). To observe the resonance by fluorescence or phosphorescence, the wavelength of detection should be separated from the wavelength of excitation by adequate filtering. For ADMR, either a filter or, when the probe wavelength dependence is monitored, a monochromator is used. The sensitivity of the ODMR spectrometer for slow-passage experiments can be considerably enhanced by modulating the amplitude of the microwaves and applying frequency-selective amplification of the photodetector signal combined with lock-in detection. Noise is then reduced to that corresponding with the passband of the amplifierlock-in detector combination, leading to an increase in the signal-to-noise ratio of several orders of magnitude. For example, in our ADMR spectrometer a ratio of better than is routinely achieved. Obviously, the modulating frequency has to be less than the slowest sublevel decay rate. For slowly decaying triplets one may be better off by scanning the line with unmodulated microwaves and signal averaging. For kinetic measurements broad-banded amplification of the detector signal and signal averaging are used. Finally, as in all modern spectrometers the instrument is interfaced to a small, dedicated computer that handles monochromator setting, data collection and storage, and carries out simple operations as taking the ratio (important when recording the probe wavelength dependence). IV. Double Resonance Optical. Once the ODMR lines of a triplet have been determined, the resonance frequencies are known precisely, and one can investigate the dependence of the intensity of a particular resonance line on the probing wavelength. Thus, one irradiates the sample with (amplitude-modulated) resonant microwaves of sufficient, preferably saturating, intensity, and monitors the (lock-in detected) photodetector output as a function of the

ODMR of triplet states probe beam wavelength. The resulting spectra may be called microwave-induced phosphorescence, fluorescence or absorbance spectra, abbreviated as MIP, MIF and MIA spectra, respectively. For one particular triplet state, the shape of these spectra does not depend on the selection of the resonance frequency, i.e. or Obviously, if more than one triplet state is present, the MI spectra provide another means to sort out which resonances belong to the same triplet state. Conversely, MI spectroscopy allows the unraveling of complex optical spectra. MIA spectra are a case apart, since they provide much more information than the MIP or MIF spectra. As will be discussed in the next section, they represent the difference of the singlet ground state, ‘normal’, absorbance spectrum and the spectrum for the system when a triplet state is present. They have therefore been labeled triplet-minus-singlet absorbance difference (T–S) spectra, rather than MIA spectra (den Blanken and Hoff, 1982). Normally, the T–S spectrum is recorded with a single-beam optical spectrometer and the ADMR signal is normalized by the intensity of the incident light, I (den Blanken and Hoff, 1982). It can be shown that for small the normalized change in transmitted light intensity is proportional to the change in absorbance As mentioned above, so that for nonsaturating light intensities, the amplitude of the T–S spectrum is proportional to I. Microwaves. Another form of double resonance is irradiating the sample with microwaves of two different frequencies, one kept constant at a particular resonance line and one scanned over a certain frequency interval. There are two varieties of such a double-resonance experiment: either the microwaves of variable frequency or those at fixed frequency are amplitude-modulated; see e.g. Maki (1984) and Hoff (1989, 1990, 1993). If in the first case the microwaves are scanned over the resonance excited by the fixed frequency, then one observes the resonance line with a “hole”, centered at the “burn” frequency. Such a hole indicates that the resonance line is inhomogeneously broadened; its width recorded for low microwave power, and narrow-banded excitation and probing, is twice the homogeneous width of the individual resonances making up the broad line. Alternatively, one may scan a different resonance, for example the line. For

285 purple bacteria, this line is very weak owing to about equal steady-state population of the x- and y-sublevels connected by the microwaves. Burning at one of the transitions transfers population from the z-level to either the x- or the y-sublevel, so that their population difference is considerably enhanced and the line becomes easily observable as an “anti-hole”. For photosynthetic preparations, the two examples of microwave double resonance were demonstrated early-on (Hoff, 1976). In the second case, the normal resonance line is not observed, and only the hole shows up having a baseline at the intensity level of the ADMR signal corresponding to the modulated microwaves, dipping to a level close to zero intensity when the frequency of burn and probe microwaves coincide. This mode makes it somewhat easier to study variations in line shape as a function of burn power, etc. (see below). Both methods of carrying out a microwave double-resonance experiment are useful for assigning ODMR lines to a particular triplet state when several triplets contribute to the ODMR spectrum. If one of the lines is ‘tickled’ with microwaves, then from all other lines only those will be affected that represent either transitions within the triplet sublevel manifold to which the ‘tickled’ line belongs or are due to a triplet whose concentration depends on that of the triplet that is ‘tickled’, e.g. because it is populated through triplet energy transfer from the latter triplet state. Note that, because the effects measured are usually very small, adequate care must be taken that interference of the two microwave channels is avoided. One precaution is to insert directional couplers or circulators in the two channels (Fig. 2a). It is important to check before use the isolation provided by these devices, as storage on e.g. a metal-containing shelf may adversely affect the isolation.

A. Triplet-Minus-Singlet (T–S) Absorbance Difference Spectra When a triplet state is present, the absorbance spectrum contains the following contributions: 1. The unperturbed singlet ground state absorbance spectrum transitions) of all molecules that are not in the triplet state

286

Arnold J. Hoff T–S difference spectrum. On the other hand, recording T–S spectra for different ADMR resonance frequencies provides a means to discriminate the resonances belonging to one and the same triplet states, since in general contributions 2 and 3 will be different for different triplet states. The ADMR-monitored T – S spectrum is of much interest. For non-interacting triplet states it provides a very accurate triplet absorbance spectrum, since that is given by adding the properly normalized singlet ground state spectrum to the T–S spectrum. For interacting triplet states, for example present in a photosynthetic pigment-protein complex, it records these interactions in a very sensitive way and thus provides a unique means to study pigment configuration.

B. Linear Dichroic T–S Spectroscopy and that do not interact with the molecule that is in the triplet state. 2. The perturbed singlet ground state spectrum of those molecules in a molecular aggregate (including proteins) that are not in the triplet state but do interact with the triplet-carrying molecule. Generally this interaction will be different when this particular molecule is in the triplet state from that when the molecule is in the singlet ground state. 3. The absorbance spectrum of the triplet state itself, consisting of transitions. With square-wave, on-off amplitude-modulated microwaves, the T–S spectrum represents the difference in absorbance of the sample for microwaves on and microwaves off (Fig. 3). It can be shown (den Blanken et al., 1984) that this difference is proportional to the difference in absorbance with and without the triplet state present. In other words, the T–S spectrum represents the difference of the absorbance of the sample with all molecules in the singlet ground state and that when all molecules of one particular type are excited into the triplet state whose ODMR resonance is being monitored. It is important to note that other triplet states with different values of and and consequently different ODMR resonance frequencies may be present without showing up in the T–S spectrum. Their absorbance is not changed by the microwaves and therefore their contribution to the absorbance cancels in the ADMR-monitored

The microwave transitions between the u and v triplet sublevels are polarized along w = u × v. This is analogous to an optical transition, whose transition dipole moment usually has a well-defined direction in the molecular frame. Often, the direction of the triplet magnetic resonance transition moments are not as well-known. In chlorophylls, for example, one may be reasonably certain that the z-transition moment is perpendicular to the molecule and that the x- and y-transition moments lie in the plane of the macrocycle, but the precise direction in the plane of the latter was until recently not known. As will be shown below, LD-(T–S) spectroscopy provides a means to ascertain the directions of the magnetic transition moments. With this knowledge one may then derive from the LD-(T–S) spectra precise structural information on molecular aggregates. In optical spectroscopy, the transition probability for a transition with transition dipole p is proportional to where is the angle between p and the electric vector E of the (polarized) incident light. A similar relation holds for the magnetic microwave transitions between the triplet sublevels. Thus, for a microwave transition moment and an angle between and the magnetic vector of the (polarized) microwave field, we have a transition probability It follows that molecules oriented with more or less parallel to have a much higher transition probability than those

ODMR of triplet states oriented about perpendicular to (Of course, for the transition probability is exactly zero.) Hence, for random excitation to the triplet state, molecules oriented in an angular interval close to will experience a much higher change in their relative triplet concentration upon the application of (polarized) resonant microwaves than molecules in an interval close to Consequently, the distribution of triplet states, which was isotropic before the application of the microwaves, becomes axially anisotropic with the axis parallel to when resonant microwaves are switched on. Microwave-induced selection in the triplet state distribution is very similar to that of photoselection. We can therefore partake of the formalism derived for that technique (see e.g. Verméglio et al., 1978) to calculate the functional relationship between and the angle between p and It can be shown that the ratio R of the intensity of the LD-(T–S) spectrum (which represents the difference and the T–S spectrum is given by (den Blanken et al., 1984; Hoff, 1985)

Eq. (14) is plotted in Fig. 4, together with plots of the LD-(T–S) and T–S intensities versus It is seen that R is quite sensitive to so that with proper calibration of the T–S and LD-(T–S) spectra, can be determined quite accurately. Of course, all this applies rigourously only for single absorbance bands. If bands with different directions of p vis-à-vis overlap, R will have some intermediate value and one will have to simulate the complete LD-(T–S) and T–S spectra to obtain values for the various The angle in Eq. (14) refers to one particular say that corresponding to the x-polarized y transition at frequency Tuning the microwaves to the frequency, that is the or y-polarized transition, allows the recording of LD-(T–S) spectra and the determination of R for a different viz. the angle between p and y. When p, x and y lie in one plane, since the triplet spin axes x, y and z span a cartesian coordinate frame. If p, x and y are not coplanar, the two measurements

287

uniquely define the orientation of p in the x,y,z coordinate frame. This is a great advantage over the ordinary photoselection experiment, where one determines just one angle between two transition moments, which leaves one with a conical ambiguity. Note that it suffices to record LD(T–S) spectra for just two of the three possible ADMR transition frequencies. Since the orientations of all values of p are determined in one and the same coordinate frame, their mutual angular dependence immediately follows. 1. Instrumentation for LD-(T–S) Spectroscopy The instrumentation for LD-(T–S) spectroscopy is quite similar to that of isotropic T–S spectroscopy and we can refer to Fig. 2a for a description. First of all we need polarized microwaves, therefore a simple helix is not suitable. We use a splitring or loop-gap cavity as described by Hardy and Whitehead (1981). This design has the advantage that for the rather long wavelength of the microwaves (between 50 and 100cm) one still has a cavity of manageable dimensions, which fits easily into a four-window liquid helium cryostat. The field is polarized along the vertical, and there-

288 fore it is perpendicular to the horizontal light beam. The probe light is unpolarized. The field induces an ellipticity in the transmitted light T, which is detected via a photoelastic modulator (PEM) after the sample followed by a polarizer. The PEM rotates the ellipse spanned by the unequally-transmitted light vectors and (with respect to by 180° at a frequency of 50 kHz. The analyzer converts this polarization modulation into an amplitude modulation at 100 kHz that is proportional to (T, transmission; A, absorbance) for small differences. The microwaves are as in T–S spectroscopy modulated at low frequency (say 315 Hz), so that the light intensity falling on the photodiode is doubly modulated at 100 kHz and 315 Hz. Demodulation at 315 Hz combined with suitable electronic filtering gives the normal T–S signal. Double demodulation at 100 kHz and 315 Hz gives the difference signal. This procedure is illustrated in Fig. 2c. The signal-to-noise ratio is enhanced by inserting selective amplifiers in both modulation channels. Scanning the monochromator yields simultaneously the T–S and the LD-(T–S) spectra. As before, the signals are divided by the intensity I to correct for changes in lamp output, monochromator sensitivity, etc. as a function of wavelength. To correctly evaluate the ratio R (Eq. (14)) one must mutually calibrate the T–S and LD-(T–S) signals. This calibration includes amplification factors, lock-in sensitivities, etc. It is best done by simulating the transmitted modulated light by a modulated light-emitting diode (LED). The amplitude of the 100 kHz modulation at specific instrument settings can then be accurately compared with that of the 315 Hz modulation. Care has to be taken to avoid as much as possible ellipticities induced by extraneous sources, such as the lamp, cryostat windows, sample cell, etc. In our set-up these extraneous ellipticities amounted to less than 1% of the LD-(T–S) signal. V. ODMR in Photosynthesis The use of ODMR in photosynthesis research has proved to be a particularly fruitful field of its application in biology. By far the most important triplet state in photosynthetic membranes is that

Arnold J. Hoff generated on the primary electron donor by radical recombination under conditions that forward electron transport is blocked by the prereduction of one of the acceptors (or by its physical deletion):

Because is prereduced and cannot normally accept two electrons, the radical pair recombines in about 20 to 50 ns to either the singlet excited or ground state of D, or with almost 100% yield at cryogenic temperatures to its triplet state, (Dutton et al., 1972). Applications of ODMR have included fluorescence detection (reviewed in Hoff, 1982, 1989) and absorbance detection. Because the latter technique now dominates the field we will mostly discuss applications of ADMR to various photosynthetic preparations.

A. ADMR Spectroscopy of Reaction Centers The great advantage of the ADMR variant above fluorescence or phosphorescence detection is that it can always be applied, regardless of quantum yields of emission, provided the lifetime of the triplet state is not too short (this holds for all cw ODMR techniques) and a sufficient optical density can be attained gives maximal signal). Both conditions hold for the photosynthetic triplets, and in the first application of ADMR to isolated bacterial reaction centers (den Blanken et al., 1982; den Blanken and Hoff, 1982) it was shown that the sensitivity of ADMR was several orders higher than that of FDMR on the same material. The high sensitivity of the ADMR method opened the way to studies of numerous isolated reaction centers, pigment solutions, etc., both of bacterial and plant origin. Accurate values of and the were determined (den Blanken et al., 1982,1983a,b; den Blanken and Hoff, 1982,1983a,b,c; Hoff and Cornelissen, 1982; Vasmel et al., 1984; Ulrich et al. 1987), as well as approximate values of the populating probabilities (Angerhofer et al., 1991) (earlier determined with FDMR by Hoff and De Vries, 1978). The values of for the primary donor triplet in RCs

ODMR of triplet states of purple bacteria are lower by 20–30% compared to that of the isolated bacteriochlorophyll (BChl) pigment (den Blanken and Hoff, 1983a). From recent EPR data on the reaction center triplet state in single crystals (Norris et al., 1987) and from ADMR spectroscopy (Lous and Hoff, 1987) it was concluded that, at least in Rhodopseudomonas (Rps.) viridis, the triplet state is largely localized on one of the dimer BChls, namely The difference between the in vivo and in vitro value of was ascribed to admixture of charge transfer (CT) states of the form to the monomeric state (Norris et al., 1987). The temperature dependence of the zero-field splitting parameters was investigated with ADMR by Ullrich et al. (1987) and Aust et al. (1990), and yielded further proof that it cannot be explained as a thermally activated population of one of the accessory BChls, in accordance with temperature-dependent EPR measurements (Hoff and Proskuryakov, 1985). Hole burning double-resonance ODMR spectroscopy has provided additional resolution. The first such experiment on photosynthetic RCs was carried out with FDMR (Hoff, 1976). A narrow hole of about 1.2 MHz FWHH was observed, indicating that the FDMR line itself is inhomogeneously broadened because of of site-heterogeneity. This result was confirmed with ADMR hole burning on protonated and deuterated RCs of Rhodobacter (Rb.) sphaeroides R26 and protonated RCs of Rps. viridis (Greis et al., 1994), with observed hole width of < 0.25, 1.0, and 2.0 MHz, respectively. The observed hole width in principle yield the transversal relaxation time giving for protonated RCs of Rb. sphaeroides R26. This value differs from the observed with spin-echo pulsed ADMR of 1.16 ±0.05 (Lous and Hoff, 1987b). Careful analysis of the physical cause of the hole width resolved this discrepancy (Greis et al., 1994). It was shown that the hole width results from unresolved second-order hyperflne interactions. In a hole burning experiment, many such interactions are time-averaged because of fast nuclear spin flip-flops. In time-resolved spin-echo ADMR spectroscopy, however, a limited distribution of flip-flop processes is sampled, leading to an apparently longer than calculated from the hole width. Measured hole widths were well-

289 reproduced when it was assumed that the triplet state of Rb, sphaeroides R26 is fully delocalized, that of Rps. viridis localized on one of the dimer BChls. A further interesting result was the observation of satellite holes when the ADMR transition was driven with high microwave powers. Then, normally forbidden quadrupole transitions are induced because of simultaneous electron and nuclear spin flips, and every nucleus connected to the unpaired electrons of the triplet state gives rise to three lines at both sides of the hole. Experimentally, instead of the expected 8 × 6 = 48 lines, only four satellites are observed, suggesting that all eight nitrogens of the primary donor dimer experience about the same electric field gradient, with two quadrupole transitions overlapping. ADMR of triplets of carotenoids present in reaction center and antenna complexes of various photosynthetic bacteria yield accurate and values (Ulrich et al., 1989; Aust et al., 1991). It is even possible to distinguish different naturally ocurring carotenoids in the same antenna preparation (Aust et al., 1991). In plant RCs the values of and the are practically the same as those of monomeric chlorophyll in solution (den Blanken and Hoff, 1983b; den Blanken et al., 1983a). This can be explained by either one of three possibilities: i) the primary donor of both plant photosystems I and II is a monomeric Chl a molecule, ii) the primary donor is a plane-parallel sandwich (Chl dimer with a fully delocalized triplet state (strong exciton coupling), iii) the primary donor is a dimer on which the triplet state is fully localized on one of the monomeric constituents and does not have CT admixture. In the latter case the term dimer in the sense of two interacting molecules obviously applies only to the singlet and possibly oxidized states of the primary donor. More on this in Section V.C.

B. T–S Spectroscopy of Reaction Centers ADMR was first used to record T–S spectra by den Blanken et al. (1982). It was immediately clear that this technique permits the recording of low temperature T–S spectra far more accurately than was possible with conventional flash tech-

290 niques. T–S spectra related to the primary donor triplet have now been recorded for many bacterial photosystems and for PS I and PS II of plants (reviewed by Hoff, 1989,1990,1993). The interpretation of some of these spectra wil be discussed in the next section. T–S spectra of carotenoids in various bacterial reaction centers and antenna complexes were reported by Lous and Hoff (1989), Ulrich et al. (1989) and Aust et al. (1991), of plant antenna complexes by van der Vos et al. (1991) and Carbonera et al. (1992).

C. Linear-Dichroic T–S Spectroscopy As explained in Section IV.B LD-(T–S) spectroscopy is similar to that of photoselection, in which an oriented distribution of e.g. photo-oxidized primary donors is produced by exciting the immobilized unoriented sample with a beam of linearlypolarized light of a wavelength corresponding to a particular absorption band, e.g. that of D. It has the advantage that, in contrast to photoselection, the dichroic spectrum can be recorded with respect to two axes of reference that are perpendicular to each other. This reduces considerably the ambiguity of unraveling the orientation of the various transition moments in the T–S spectrum. Complex T–S spectra such as those of the photosynthetic RC, where many overlapping features are present such as appearing bands, bleachings, band shifts to the red and to the blue, can only be interpreted with certainty if the corresponding LD-(T–S) spectra are available (Hoff et al., 1985). The experimental T–S and LD-(T–S) spectra can be compared with spectra simulated with exciton theory, which describes the interaction between pigment molecules brought about by the electrostatic dipole–dipole coupling between their electronic transition moments. For two identical molecules this interaction causes a shift and a splitting of the excited state, which is now composed of two levels with (apart from the shift) energies given by the original energy plus or minus the interaction energy. The theory is easily extended to n pigments, each pigment interacting differently with the other T–S pigments (Davidov, 1981; Kasha et al., 1965; Pearlstein, 1982). For a known pigment configuration the absorption spectrum of the excitonically coupled spec-

Arnold J. Hoff trum is then readily obtained. Generally the resulting bands are mixtures of the original uncoupled bands, with transition moments that are vectorial combinations of the original transition moments. Thus, in general, introducing a triplet state in a coupled pigment system, with corresponding changes in the dipole–dipole couplings, will lead to changes in the position, intensity and the polarization of individual absorption bands. Especially the change in polarization is useful for drawing conclusions on the extent of dipolar coupling between the cofactors in photosynthetic reaction centers. 1. Bacterial Reaction Centers Exciton calculations based on the crystal structure of the RC of Rps. viridis (Deisenhofer et al., 1985) yield satisfactory simulations of its ADMRmonitored T–S and LD-(T–S) spectra (Lous and Hoff, 1987a; Knapp et al., 1986; Scherer and Fischer, 1987a) (Fig. 5). The spectral fits allowed to conclude that the triplet is localized on one of the BChl components of D, The bleaching at 990 nm is due to the disappearance of the longwavelength band of D that is shifted to the red because of excitonic coupling between the two BChls of D. Exciton coupling is much weaker in the triplet state of so that in a localized state part of the BChl absorption is bleached and part shifts back to the ‘normal’ wavelength of BChl absorption in a protein matrix. The large positive band at about 830 nm and the two smaller features at the long- and short-wavelength side of the positive band were attributed to band shifts of the two accessory BChls induced by the change (den Blanken and Hoff, 1982). The small positive band at 872 nm was attributed to a triplet–triplet absorption of whose transition moment is oriented along the axis in the BChl macrocycle. The absence of significant bands in the 790 nm region indicates that the two bacteriopheophytins, and are only weakly coupled to D. With appropriate modifications, a similar picture holds for chemically-modified RC of Rb. sphaeroides R-26 (Scherer and Fischer, 1987b; Beese et al., 1988; Angerhofer et al., 1989). The above interpretation has stood the test of time rather well, with the exception of the

ODMR of triplet states

assignment of the band at 872 nm. Similar work was recently carried out on RCs of Rb. sphaeroides R26 and its LH(L131), LH(M160), HL(L173) and HL(M202) mutants, and on BChl a in vitro (Vrieze and Hoff, 1995a,b; Vrieze et al., 1995a,b). The orientation of the moment of the primary donor relative to the triplet x- and y-axes was found to be 85 ± 5° and 18 ± 3°, respectively. The absorption region of the accessory BChls was deconvoluted in two bleached bands, corresponding to the two bands of the low-temperature absorption spectrum, and three appearing bands, which included the transition at 818 nm previously attributed to a triplet– triplet transition by analogy to the interpretation of the 872 nm band for Rps. viridis. This previous

291

assignment could now be excluded on the basis of the measured polarization and the polarization of the in vitro triplet–triplet absorption band (Vrieze and Hoff, 1995a). With the aid of the results for two mutants that were changed in the composition of the primary donor and its environment, the two bleachings and two of the appearing bands were assigned to band shifts resulting from the introduction of the triplet state. The third band was attributed to mostly a primary donor contribution, likely the “monomer” transition, although it should be realized that all bands reflect mixed transitions (Fig. 6). Surprisingly, the orientations of the bleached bands and of the corresponding appearing bands were practically the same. Because all reasonable exciton models

292

Arnold J. Hoff

major difference being that one of the accessory BChls is replaced by a BPh (Vasmel, 1986; Vasmel et al., 1987). The T–S spectra of green sulfur bacteria are more complex and have only been tentatively interpreted (den Blanken et al., 1983b; Hoff et al., 1988). ADMR and T–S spectra of membranes of Heliobacterium chlorum, a representative of the Heliobacteriaceae containing BChl g, were recently recorded by Vrieze et al. (Vrieze et al., 1995c). A variety of triplet states were detected, which with the aid of their corresponding T–S spectrum were assigned to the primary donor and several distinct antenna pigments. 2. Plant Reaction Centers predict considerable changes in the orientation, it was concluded that exciton interaction was not the single determining coupling between the pigments in the reaction center. With minor differences the qualitative aspects of the above picture hold for all purple bacteria investigated (Dijkman et al., 1988). The same applies to the green filamentous bacterium Chloroflexus aurantiacus (Vasmel et al., 1984), whose RC is very similar to that of purple bacteria, the

The first highly-resolved ADMR-recorded T–S spectra of the plant photosystems were obtained already a decade ago (den Blanken and Hoff, 1983b; den Blanken et al., 1983a). They are characterized by a strong bleaching of donor bands at 703 and 682 nm for PS I and II, respectively, and a positive band at the blue side of this bleaching, at ~ 674 nm, close to the wavelength of the absorption of Chl a in vitro (Fig. 7). This

ODMR of triplet states

band, which is more pronounced in intact membrane fragments than in core or RC particles (den Blanken et al., 1983a; van der Vos et al., 1992; Angerhofer et al., 1993; Carbonera et al., 1994a), was tentatively attributed to the appearing absorption of a monomeric Chl a molecule belonging to a primary donor dimer on which the triplet state was localized on the second Chl a (den Blanken and Hoff, 1983b), supporting the notion that the primary donor of the two photosystems is a dimeric Chl a complex. Recent work employing LD-(T–S) spectroscopy, however, could not confirm this assignment (see below). The similarity of the T–S spectra of intact PS I and PS II suggests that the composition and pigment environment of the primary donor of PS I and PS II are similar.

a. Photosystem I The orientations of the of P700 and of the triplet–triplet transition of in the x,yplane of the latter as determined with LD-(T– S) spectroscopy are shown in Fig. 8a. The two orientations are practically identical; both lie approximately in the triplet x,y-plane and are rotated by about 13° with respect to the corresponding transition moments of Chl a in vitro (Fig. 8b,c). This co-rotation strongly suggests that is localized, as for a delocalized triplet the of P700 is necessarily a delocalized exciton transition, and must either coincide with one of the in-plane triplet axes, or for a parallel

293

dimer, be aligned along the monomer tion, in order to produce a geometry for which the ZFS parameters are identical to those of monomeric Chl a. Neither of these cases obtains for Note that a similar argument holds for the orientation of the of P680 (see below). The orientation of the 674 nm band deduced from LD-(T–S) spectroscopy precludes its assignment to a ‘monomer’ transition of it is likely due to a band shift of an accessory Chl a located close to P700. Small bands at 663 and 670 nm were assigned to the ‘active’ and ‘inactive’ acceptor Chl a. Lastly, the band at 687 nm was attributed to a Chl a chromophore adjacent to P700, and coupled to the primary acceptor (Vrieze et al., 1995d). No evidence was found for a higher-energy exciton component of an excitonically-coupled P700 dimer.

b. Photosystem II The question of the configuration of the primary donor in PS II (Chl a monomer or dimer) was addressed by T–S and LD-(T–S) spectroscopy (van der Vos et al., 1992; van der Vos, 1994; Vrieze and Hoff, 1995b) of the so-called D1/D2– cyt b-559 RC particle (Nanba and Satoh, 1987) and of Chl a in methanol. It was found that the ADMR signals and corresponding primary donor bleaching in the T–S spectrum spectrum are quite heterogeneous, being composed of five distin-

294

guishable bands, all related to active P680. These bands are also present in FDMR signals of the D1/D2 particle (Carbonera et al., 1994b), but largely coalesce for intact PS II membranes (Carbonera et al., 1994a), suggesting that the optical and ODMR heterogeneity of the Dl/D2–cyt b559 RC is introduced by the isolation procedure. Note that a similar heterogeneity of the ADMR and optical properties of the triplet state of the primary donor was previously found for bacterial RCs and, to lesser extent, chromatophores (den Blanken and Hoff, 1983c). The angles between the transition moments of the major components of the P680 bleaching and the triplet x,y-axes were determined with LD-(T– S) spectroscopy and proved to be practically the same, ruling out the possibility that one would represent a higher exciton component (van der Vos et al., 1992). The moment of P680 lies approximately in the plane spanned by the triplet x,y-axes. The – x,y angles with their margin of error are displayed in Fig. 9, which also shows the corresponding angles for Chl a in methanol. Two possibilities are considered: (i) is localized on one Chl a of an excitonically coupled dimer (Fig. 9b), (ii) is localized on one non-excitonically coupled, monomeric Chl a (Fig. 9c). The small rotation (10–20°) observed in Fig. 9c of the

Arnold J. Hoff

moment of P680 compared to the of monomeric Chl a in methanol can easily result from differences in the environment of the triplet state (Vrieze and Hoff, 1995b), and therefore does not necessarily falsify assumption (ii) that P680 is a non-coupled Chl a monomer. Thus, the ADMR data is consistent with, but does not definitely prove, that P680 is a (weakly) excitonically-coupled Chl a dimer. The directional information obtained by LDADMR can be used to extract the orientation of the moment of P680 relative to the membrane with the aid of the orientation of the triplet axes with respect to the membrane plane obtained from EPR on oriented membranes by van Mieghem et al. (1991). For the primary donor pigments angles result of 15.5° and 12.5°, for the accesssory BChls, 21° and 23° (van der Vos et al., 1992). None of these angles are close to the experimental angle, so that the orientation of P680 in the membrane must differ from that of the primary donor of the purple bacteria. The high resolving power of ADMR is nicely demonstrated by the observation of a pheophytin (Pheo) triplet absorbing at 681 nm (Fig. 10) (van der Vos et al., 1992). Because has quite characteristic ZFS parameters, the microwaveoptical double-resonance experiment of Fig. 10

ODMR of triplet states unequivocally demonstrates that at least one Pheo in the D1D2–cyt b-559 complex absorbs at 681 nm and overlaps with the P680 nm absorption. A microwave double-resonance experiment suggested that is populated indirectly through energy transfer from or enhanced intersystem crossing in RCs ‘closed’ by the formation of (van der Vos et al., 1992). It was found in later work, however, that the doubleresonance experiment had suffered from interference effects (van der Vos and Hoff, 1995), and that is formed in a one-photon process not correlated with the presence of in ‘degraded’ RCs, possibly representing a conformationally changed state in which for example the distance between P680 and Pheo is increased (Angerhofer et al., 1994). VI. Concluding Remarks The triplet state is a versatile probe of structure and function in photosynthesis on a molecular level. ODMR, and especially ADMR, in zeromagnetic field of triplet states is a powerful tool for determining triplet properties that give direct information on the structure and interactions of pigments in photosynthetic pigment-protein complexes. Accurate values have been determined of the zero-field splitting parameters D and E and the sublevel decay rates, of the triplet state of the primary donor, and of chlorophyll and carotenoid antenna chromophores, for a variety of photosynthetic preparations, including the two plant RCs, a number of bacterial RCs, and several antenna pigment-protein complexes. ADMR has further allowed the recording of accurate, high-resolution T–S and LD-(T–S) spectra, which have offered insight in the relative orientation of, and the interactions between, reaction center pigments, and in the “multiplicity” of the primary donor. An important finding is that isolated reaction centers generally exhibit a heterogeneity of the optical and magnetic resonance parameters of the primary donor, which is attributed to somewhat different conformations of the RC pigments. This observation relates to the many experiments carried out on such RCs, for example by fast laser flash photolysis and ENDOR and (high-field) EPR spectroscopy, which normally lack the resolution to discriminate between the various confor-

295 mations of the RC. Consequently, the results of these experiments represent an average over the conformations, static at low temperatures, likely dynamic at higher temperatures, and it may be difficult to compare the results meaningfully with properties calculated from the crystal structure, which presumably is much better defined than the “glassy” matrix of the RC protein in solid and liquid solution. A key result of ADMR-recorded LD-(T–S) spectroscopy is the determination of the orientation of several transition moments. This has recently allowed much better insight in the composition of the absorbance spectrum, which is of crucial importance in the interpretation of ultrafast pump-probe laser flash spectroscopy, and therefore for our understanding of the details of primary electron transfer. Future applications of ODMR will focus on comparative studies of mutant and chemicallymodified reaction centers, and on RC heterogeneity. Pulsed ADMR will become increasingly important, allowing studies of the temperature dependence of the various triplet parameters, and of dynamical aspects of line-broadening, triplet– triplet transfer, etc. Results on the orientation of the various absorption bands in the T–S absorbance difference spectrum call for renewed effort in theoretically interpreting the optical properties of the RC pigment complex.

References Angerhofer A, Beese D, Hoff AJ, Lous EJ and Scheer H (1989) Linear dichroic triplet-minus-singlet absorbance difference spectra of borohydride treated reaction centres of Rhodobacter sphaeroides R26. In: Singhal G (ed) Applications of Molecular Biology in Bioenergetics of Photosynthesis, pp 197–203. Narosa Publishing House, New Delhi. Angerhofer A, Speer R, Ullrich J, von Schütz JU and Wolf HC (1991) Time resolved ADMR applied to the triplet state of the primary donor of bacterial photosynthetic reaction centers. Appl Magn Reson 2: 203–216. Angerhofer A, Bernlocher D and Robert B (1993) Absorption detected magnetic resonance of D-1/D-2 complexes from Pisum sativum. In: Steiner U (ed) Magnetic Field and Spin Effects in Chemistry, pp 167–180. Oldenburg Verlag, München. Angerhofer A, Friso F, Giacometti GM, Carbonera D and Giacometti G (1994) Optically detected magnetic resonance study on the origin of the pheophytin triplet state in

296 D-1/D-2 cytochrome b-559 complexes. Biochim Biophys Acta 1188: 35–45. Aust V, Angerhofer A, Parot PH, Violette CA and Frank HA (1990) Temperature-dependent ADMR on borohydridetreated reaction centers of Rhodobacter sphaeroides R26. Chem Phys Lett 173: 439–442. Aust V, Angerhofer A, Ullrich J, von Schlitz JU, Wolf HC and Cogdell RJ (1991) ADMR of carotenoid triplet states in bacterial photosynthetic antenna and reaction center complexes. Chem Phys Lett 181: 213–221. Beese D, Steiner R, Scheer H, Robert B, Lutz M and Angerhofer A (1988) Chemically modified photosynthetic bacterial reaction centers: Circular dichroism, Raman resonance, low temperature absorption, fluorescence and ODMR spectra and polypeptide composition of borohydride treated reaction centers from Rb. sphaeroides R26. Photochem Photobiol 47: 293–304. Carbonera D, di Valentin M, Giacometti G and Agostini G (1992) FDMR of chlorophyll triplets in integrated particles and isolated reaction centers of Photosystem II. Identification of P680 triplet. Biochim Biophys Acta 1185: 167–176. Carbonera D, Giacometti G, Agostini C, Angerhofer A and Aust V (1994a) ODMR of carotenoid and chlorophyll triplets in CP 47 complexes of spinach. Chem Phys Lett 194: 273–281. Carbonera D, Giacometti G and Agostini G (1994b) A well resolved ODMR triplet minus singlet spectrum of P680 from PS II particles. FEBS Lett 343: 200–204. Chan IY (1982) Zero-field ODMR techniques - phosphorescence detection. In: Clarke RH (ed) Triplet State ODMR Spectroscopy, pp 1–24. Wiley-Interscience, New York. Davidov AS (1981) Theory of Molecular Excitons, Plenum Press, New York. Deisenhofer J, Epp O, Miki K, Huber R and Michel H (1985) Structure of the protein subunits in the photosynthetic reaction centre of Rhodopseudomonas viridis at 3Å resolution. Nature 318: 618–624. den Blanken HJ and Hoff AJ (1982) High-resolution optical absorption-difference spectra of the triplet state of the primary donor in isolated reaction centers of the photosynthetic bacteria Rhodopseudomonas sphaeroides R26 and Rhodopseudomonas viridis measured with optically detected magnetic resonance at 1.2 K. Biochim Biophys Acta 681: 365–374. den Blanken HJ and Hoff AJ (1983a) Sublevel decay kinetics of the triplet state of bacteriochlorophyll a and b in methyltetrahydrofuran at 1.2 K. Chem Phys Lett 96: 343–347. den Blanken HJ and Hoff AJ (1983b) High-resolution absorbance-difference spectra of the triplet state of the primary donor P700 in Photosystem I subchloroplast particles measured with absorbance-detected magnetic resonance at 1.2 K. Evidence that P700 is a dimeric chlorophyll complex. Biochim Biophys Acta 724: 52-61. den Blanken HJ and Hoff AJ (1983c) Resolution enhancement of the triplet–singlet absorbance-difference spectrum and the triplet-ESR spectrum in zero field by the selection of sites. An application to photosynthetic reaction centers. Chem Phys Lett 98: 255–262. den Blanken HJ, van der Zwet GP and Hoff AJ (1982) ESR in zero field of the photoinduced triplet state in isolated

Arnold J. Hoff reaction centers of Rhodopseudomonas sphaeroides R26 detected by the singlet ground state absorbance. Chem Phys Lett 85: 335–338. den Blanken HJ, Hoff AJ, Jongenelis APJM and Diner BA (1983a) High-resolution triplet-minus-singlet absorbance difference spectrum of photosystem II particles. FEBS Lett 157: 21–27. den Blanken HJ, Vasmel H, Jongenelis APJM, Hoff AJ and Amesz J (1983b) The triplet state of the primary donor of the green photosynthetic bacterium Chloroflexus aurantiacus. FEBS Lett 161: 185–189. den Blanken HJ, Meiburg RF and Hoff AJ (1984) Polarized triplet-minus-singlet absorbance difference spectra measured by absorbance detected magnetic resonance. An application to photosynthetic reaction centres. Chem Phys Lett 105: 336–342. Dijkman JA, den Blanken HJ and Hoff AJ (1988) Towards a new taxonomy of photosynthetic bacteria: ADMR-monitored triplet difference spectroscopy of reaction center pigment-protein complexes. Isr J Chem 28: 141–148. Dutton PL, Leigh JS and Seibert M (1972) Primary processes in photosynthesis: in situ ESR studies on the light induced oxidized and triplet state of reaction center bacteriochlorophyll. Biochem Biophys Res Commun 46: 406–413. Greis JW, Angerhofer A, Norris JR, Scheer H, Struck A and von Schütz JU (1994) Spectral diffusion and quadrupole splittings in absorption detected magnetic resonance hole burning spectra of photosynthetic reaction centers. J Chem Phys 100: 4820–4827. Hardy WN and Whitehead LA (1981) Split-ring resonator for use in magnetic resonance from 200–2000 MHz. Rev Sci Instrum 57: 213–216. Hoff AJ (1976) Kinetics of populating and depopulating of the components of the photoinduced triplet state of the photosynthetic bacteria Rhodospirillum rubrum, Rhodopseudomonas spheroides (wild type), and its mutant R26. Biochim Biophys Acta 440: 765–771. Hoff AJ (1982) ODMR spectroscopy in photosynthesis II. The reaction center triplet in bacterial photosynthesis. In: Clarke RH (ed) Triplet State ODMR Spectroscopy, pp 367–425. John Wiley and Sons, New York. Hoff AJ (1985) Triplet-minus-singlet absorbance difference spectroscopy of photosynthetic reaction centers by absorbance-detected magnetic resonance. In: Michel-Beyerle ME (ed) Antennas and Reaction Centers of Photosynthetic Bacteria. Structure Interaction and Dynamics, pp 150–163. Springer-Verlag, Berlin. Hoff AJ (1989) Optically-detected magnetic resonance of triplet states. In: Hoff AJ (Ed) Applications in Biology and Biochemistry, pp 633–684. Elsevier, Amsterdam. Hoff AJ (1990) Triplet states in photosynthesis: linear dichroic optical difference spectra via magnetic resonance. In: Linskins HF and Jackson JF (eds) Physical Methods in Plant Sciences Vol II, Modern Methods of Plant Analysis, pp 23–57. Springer-Verlag, Berlin. Hoff AJ (1993) Optically detected magnetic resonance of triplet states in proteins. In: Riordan JF and Vallee BL (eds) Metallobiochemistry Part D Physical and Spectroscopic Methods for Probing Metal Ion Environments in Metallop-

ODMR of triplet states roteins. Methods in Enzymology Vol 227, pp 290–330. Academic Press, San Diego. Hoff AJ and Cornelissen B (1982) Microwave power dependence of triplet state kinetics as measured with fluorescence detected magnetic resonance in zero field. An application to the reaction centre bacteriochlorophyll triplet in bacterial photosynthesis. Mol Phys 45: 413–425, Hoff AJ and de Vries HG (1978) Electron spin resonance in zero magnetic field of the reaction center triplet of photosynthetic bacteria. Biochim Biophys Acta 503: 94–106. Hoff AJ and Proskuryakov II (1985) Triplet EPR spectra of the primary electron donor in bacterial photosynthesis at temperatures between 15 and 296 K. Chem Phys Lett 115: 303–310. Hoff AJ and van der Waals JH (1976) Zero field resonance and spin alignment of the triplet state of chloroplasts at 2 K. Biochim Biophys Acta 423: 615–620. Hoff AJ, Govindjee and Romijn JC (1977) Electron spin resonance in zero magnetic field of triplet states of chloroplasts and subchloroplast particles. FEBS Lett 73: 191– 196. Hoff AJ, den Blanken HJ, Vasmel H and Meiburg RF (1985) Linear-dichroic triplet-minus-singlet absorbance difference spectra of reaction centers of the photosynthetic bacteria Chromatium vinosum, Rhodopseudomonas sphaeroides R26 and Rhodospirillum rubrum S1. Biochim Biophys Acta 806: 389–397. Hoff AJ, Vasmel H, Lous EJ and Amesz J (1988) Tripletminus-singlet optical difference spectroscopy of some green photosynthetic bacteria. In: Olson JM, Ormerod JG, Amesz J, Stackebrandt, E and Trüper HG (eds) Green Photosynthetic Bacteria, pp 119–126. Plenum Press, New York. Kasha M, Rawls HS and El-Bayoumi MA (1965) The exciton model in molecular spectroscopy. Pure Appl Chem 11: 371–392. Knapp EW, Scherer POJ and Fischer SF (1986) Model studies of low-temperature optical transitions of photosynthetic reaction centers. A-, LD-, CD-, ADMR and LD-ADMRspectra for Rhodopseudomonas viridis. Biochim Biophys Acta 852: 295–305. Lous EJ (1988) Interactions between Pigments in Photosynthetic Protein Complexes. An Optically-detected Magnetic Resonance and Magnetic Field Effect Study. Doctoral Thesis, Leiden. Lous EJ and Hoff AJ (1987a) Exciton interactions in reaction centers of the photosynthetic bacterium Rhodopseudomonas viridis probed by optical triplet-minus-singlet polarization spectroscopy at 1.2 K monitored through absorbance-detected magnetic resonance. Proc Natl Acad Sci USA 84: 6147–6151. Lous EJ and Hoff AJ (1987b) Absorbance detected electron spin echo spectroscopy of non-radiative states in zero field. An application to the primary donor of the photosynthetic bacterium Rhodobacter sphaeroides R26. Chem Phys Lett 140: 620–625. Lous EJ and Hoff AJ (1989) Isotropic and linear dichroic triplet-minus-singlet absorbance difference spectra of two carotenoid containing bacterial photosynthetic reaction centers in the temperature range 10–288 K. A model for

297 bacteriochlorophyll-carotenoid triplet transfer. Biochim Biophys Acta 974: 88–103. Maki AH (1984) Techniques, theory, and biological applications of optically-detected magnetic resonance (ODMR). In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance Vol 6, pp 187–294. Plenum Press, New York. Nanba O and Satoh K (1987) Isolation of a Photosystem II reaction center consisting of D-1 and D-2 polypeptides and cytochrome b 559. Proc Natl Acad Sci USA 84: 109–112. Norris JR, Lin CP and Budil DE (1987) Magnetic resonance of ultrafast chemical reactions. Examples from photosynthesis. J Chem Soc Faraday Trans 1 83: 13–27. Pearlstein RM (1982) Chlorophyll singlet excitons. In: Govindjee (ed) Energy Conversion in Plants and Bacteria. Photosynthesis Vol I, pp 293–329. Academic Press, New York. Scherer POJ and Fischer SF (1987a) Model studies to lowtemperature optical transitions of photosynthetic reaction centers of Rhodobacter sphaeroides and Chloroflexus aurantiacus. Biochim Biophys Acta 891: 157–164. Scherer POJ and Fischer SF (1987b) Application of exciton theory to optical spectra of sodium borohydride treated reaction centers from Rhodobacter sphaeroides R26. Chem Phys Lett 137: 32–36. Ullrich J, Angerhofer A, von Schütz JU and Wolf HC (1987) Zero-field absorption ODMR of reaction centers of Rhodobacter sphaeroides at temperatures between 4.2 and 75 K. Chem Phys Lett 140: 416–420. Ullrich J, Speer R, Greis J, von Schütz JU, Wolf HC and Cogdell RJ (1989) Carotenoid triplet states in pigmentprotein complexes from photosynthetic bacteria: Absorption-detected magnetic resonance from 4.2–225 K. Chem Phys Lett 155: 363–370. van der Bent SJ, de Jager PA and Schaafsma TJ (1976) Optical detection and electronic simulation of magnetic resonance in zero magnetic field of dihydroporphin free base. Rev Sci Instrum 47: 117–212. van der Vos R (1994) Antenna and Reaction Center Complexes in Photosynthesis. An Absorbance-detected Magnetic Resonance and Magnetic Field Effect Study. Doctoral Thesis, Leiden. van der Vos R and Hoff AJ (1995) Optically–detected magnetic field effects on the D1–D2 cyt b-559 complex of Photosystem II. Temperature dependence of kinetics and structure. Biochim Biophys Acta 1228: 73–85. van der Vos R, Carbonera D and Hoff AJ (1991) Microwave and optical spectroscopy of carotenoid triplets in lightharvesting complex LHCII of spinach by absorbance-detected magnetic resonance. Appl Magn Reson 2: 179–202. van der Vos R, van Leeuwen PJ, Braun P and Hoff AJ (1992) Analysis of the optical absorbance spectra of D1–D2– cytochrome b-559 complexes by absorbance-detected magnetic resonance. Structural properties of P680. Biochim Biophys Acta 1140: 184–196. van Mieghem FJE, Satoh K and Rutherford AW (1991) A chlorophyll tilted 30° relative to the membrane in the Photosystem-II reaction centre. Biochim Biophys Acta 1058: 379–385. Vasmel H (1986) The Photosynthetic Membrane of Green Bacteria. Doctoral Thesis, Leiden. Vasmel H, den Blanken HJ, Dijkman JA, Hoff AJ and Amesz

298 J (1984) Triplet-minus-singlet absorbance difference spectra of reaction centers and antenna pigments of the green photosynthetic bacterium Prosthecochloris aestuarii. Biochim Biophys Acta 767: 200–208. Vasmel H, Meiburg RF, Amesz J and Hoff AJ (1987) Optical properties of the reaction center of Chloroflexus aurantiacus at low temperature. Analysis by exciton theory. In: Biggins J (ed) Progress in Photosynthesis Research Vol 1, pp 403–406. Martinus Nijhoff, Dordrecht. Verméglio A, Breton J, Paillotin G and Cogdell R (1978) Orientation of chromophores in reaction centers of Rhodopseudomonas sphaeroides: A photoselection study. Biochim Biophys Acta 501: 514–530. Vrieze J, Gast P and Hoff AJ (1992) The structure of the reaction center of Photosystem I investigated with linear dichroic absorbance-detected magnetic resonance at 1.2 K. In: Murata N (ed) Research in Photosynthesis Vol I, pp 553–556. Kluwer Academic Publishers, Dordrecht. Vrieze J and Hoff AJ (1995a) Interactions between chromophores in reaction centers of purple bacteria. A reinterpretation of the triplet-minus-singlet spectra of Rb. sphaero-

Arnold J. Hoff ides R26 and Rps. viridis. Biochim Biophys Acta, submitted. Vrieze J and Hoff AJ (1995b) The orientation of the triplet axes with respect to the optical transition moments in (bacterio)chlorophylls. Chem Phys Lett 237: 493–501. Vrieze J, Williams JC, Allen J and Hoff AJ (1995a) A LDADMR study on reaction centers of the LH(L131) and LH(M160) mutants of Rb. sphaeroides. Biochim Biophys Acta, submitted. Vrieze J, Schenck CC and Hoff AJ (1995b) The triplet state of the primary donor in reaction centers of the HL(L173) and HL(M202) heterodimer mutants of Rb. sphaeroides. Biochim Biophys Acta, submitted. Vrieze J, van de Meent EJ and Hoff AJ (1995c) Bacteriochlorophyll g triplet states of the primary donor and antenna in membranes of Heliobacterium chlorum. Biochemistry, submitted. Vrieze J, Gast P and Hoff AJ (1995d) The structure of the reaction center of Photosystem I of plants. An investigation with linear-dichroic absorbance-detected magnetic resonance. J Phys Chem, submitted.

Chapter 18 Magic Angle Spinning Nuclear Magnetic Resonance of Photosynthetic Components Huub J.M. de Groot Leiden Institute of Chemistry, Gorlaeus Laboratories, P.O. Box 9502, 2300 RA Leiden, The Netherlands

Summary I. Introduction II. Magic Angle Spinning NMR Spectroscopy III. Probing the Local Environment of M(Y)210 in Rb. sphaeroides R26 RC with MAS IV. The Configuration of the Spheroidene in the Rb. sphaeroides RC V. The Asymmetric Binding in Rb. sphaeroides R26 VI. New Developments. CIDNP and Correlation Spectroscopy VII. Concluding Remarks Acknowledgements References

299 300 300 302 305 306 309 312 312 312

Summary This chapter is the first extensive review of magic angle spinning (MAS) NMR studies of photosynthetic components. We describe how the chemical environment of the (M)Y210 in Rhodobacter sphaeroides R26 reaction centers was investigated and how labeled Rb. sphaeroides 2.4.1 (M)Y210W mutant reaction centers were used to assign the response of (M)Y210 in R26 to a narrow signal at 152.2 ppm. According to the MAS NMR, the Y(M)210 is in a homogeneously ordered region of the complex, and probably contributes to the fine-tuning of the energy levels of prosthetic groups involved in electron transfer. In another MAS NMR investigation, analysis of the chemical shifts of labels in the carotenoid supports a configuration for the spheroidene in the Rb. sphaeroides RC. This is a detail of the structure that can not yet be resolved unambiguously with modern X-ray techniques. A temperature-dependent asymmetry was reported for since the CP/MAS signal of carbonyl 4 is not observable at temperatures K, contrary to the signals from the opposite carbonyl 1. Together, these studies represent the first systematic MAS investigation of a membrane protein complex. Finally, some new technical developments are discussed. A novel example of MAS photo-CIDNP was discovered recently, and yields strong emissive signals for depleted, or pre-reduced, labeled RC. Using MAS dipolar correlation Spectroscopy on uniformly enriched chlorophyll a/water aggretates, a complete assignment for resonances was obtained and intermolecular transfer of coherence was observed. This development is an important step forward towards structure determination in multispin labeled proteins and complexes that are not accessible to diffraction methods.

Correspondence: Fax: 31-71-5274537; E-mail: [email protected]

299 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 299–313. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

300

Huub J.M. de Groot

Abbreviations: BChl – bacteriochlorophyll; Chl – chlorophyll; CIDNP – chemically induced dynamic nuclear polarization; CP – cross polarization; FID – free induction decay; LCAO – linear combination of atomic orbitals; MAS – magic angle spinning; MO – molecular orbital; P – special pair; – primary quinone acceptor; – accessory bacteriochlorophyll; RC – reaction center; – ubiquinone-10; – bacteriopheophytin

I. Introduction In this chapter the recent application of the CP/MAS NMR technique for the investigation of photosynthetic pigments, in particular the bacterial photosynthetic reaction center of Rb. sphaeroides, is reviewed. MAS NMR, in conjunction with selective isotope enrichment, is the method of choice for NMR investigations of membrane protein complexes and other large macromolecular entities. Some years ago, we started, in an intensive collaboration with several other groups, a multidisciplinary effort to investigate the photosynthetic reaction center of Rb. sphaeroides with MAS NMR and site- specific isotope enrichment (de Groot et al., 1990; 1992). Very recently, the group of A.E. McDermott has reported on a first observation of MAS photoCIDNP data of reaction centers (Zysmilich and McDermott, 1994). Altogether, these studies now represent the first systematic investigation of a membrane protein complex with MAS NMR techniques, and they have already provided information that is not otherwise accessible. MAS NMR is a technique for obtaining highresolution NMR data from solids. A brief introduction to the technique is presented in section II. Subsequently the results obtained thus far for bacterial photosynthetic reaction centers will be discussed. Section III deals with the MAS NMR characterization of the side chain of the (M)Y210 amino acid, which is highly conserved between different species and has attracted considerable attention over the past few years, since it is thought to be of importance for the fast rate of primary electron transfer. In section IV we describe how MAS NMR was used to address one specific structural aspect to atomic resolution, the configuration of the spheroidene carotenoid around the central 15 = 15' bond, a detail that could not be unambiguously resolved with X-ray

crystallography and is crucial for understanding the protective triplet quenching mechanism. The investigation of protein-cofactor interactions of importance for the specific one- electron redox properties of the primary quinone are discussed in section V. Finally, we discuss in section VI new developments concerning photo-CIDNP of RC and MAS dipolar correlation spectroscopy of chlorophyll a model systems. II. Magic Angle Spinning NMR Spectroscopy The standard CP/MAS experiment and a resulting FID are shown schematically in Fig. 1. Following a pulse on the ensemble of abundant nuclear spins, symbolized by I, the net proton magnetization is rotated to the plane perpendicular to the direction of the magnetic field Subsequently the phase of the proton transmitter is changed by resulting in a spin-lock pulse, that can be interpreted as a small rf field along the magnetization vector, while simultaneously another rf field in the same coil is applied at the frequency of the less-abundant nucleus S, typically with

Magic angle spinning NMR

301

the Hartmann–Hahn condition. Since the splitting of the and levels are now the same in the rotating frame, cross polarization, i.e. transfer of coherence between the two spin species, can take place via the heteronuclear dipolar interactions. In proteins cross polarization times are typically 1–5 ms, and to a reasonable approximation signal intensities vary with according to the double exponential function (see e.g. Mehring, 1983)

In this expression the sets the amplitude, is the characteristic polarization transfer time associated with the dipolar coupling and is the longitudinal proton relaxation time in the rotating frame, i.e. under spin-lock conditions. The theoretical maximum for the transfer of polarization of the protons into S-spin polarization is

with the ratio of the number of lessabundant nuclei and abundant protons As normally the and for this represents an enhancement of with respect to thermal equilibrium. For membrane protein studies this enhancement factor is convenient, but not crucial. Since the NMR experiment involves the acquisition of 20,000–50,000 scans, the total time required for obtaining a spectrum is in practice determined by the repetition time between individual scans, which is of the order of the longitudinal relaxation time in membrane proteins can be very long for nuclei located inside a membrane protein complex and direct acquisition of signals is practically impossible. In contrast, longitudinal relaxation proceeds much faster, as the excess energy can rapidly transfer into the lattice via the homonuclear dipolar interactions through e.g. rotating methyl groups with an in-

trinsically short Since in the CP/MAS experiment the is polarized from the spin system, the repetition time between scans is limited by the short This represents a significant gain in overall efficiency, which is in fact crucial. Since strong continuous wave decoupling is applied during acquisition in the CP/MAS experiment, the main linebroadening mechanism for randomly oriented molecules in a powder is due to the anisotropy of the chemical shift,

The effect is illustrated in Fig. 2A, which shows the CP response of a powder sample of [1– The chemical shift is represented by a real symmetric second rank tensor In its principal axis frame the is diagonal. In the discussion of MAS results obtained on membrane proteins, two representations of the are commonly used. Most often

302

Huub J.M. de Groot

with and the principal components, arranged in such a way that is the most upfield component in the spectrum and is the most downfield component. Alternatively, the isotropic and anisotropic part may be separated according to

with the isotropic shift, and 0. By convention and the difference between the two representations is a reordering of the tensor principal components. Since the anisotropic part is traceless, it can be characterized with two parameters, the anisotropy and the asymmetry The chemical shift broadening is effectively suppressed by macroscopic sample rotation around an axis at the magic angle A detailed treatment of the MAS averaging is beyond the scope of this chapter, and the reader is referred to the existing literature (see e.g. Mehring, 1983). Briefly, during MAS individual molecules are subject to physical sample rotation and the chemical shift of every molecule varies periodically with the rotation rate. Since many different chemical shift trajectories will be possible, the macroscopic nuclear spin polarization collapses in a very short time, but refocuses at After every full rotor cycle the signal is recovered, yielding a rotational echo train, modulated by the overall precession frequency corresponding to the resonance offset (Fig. 1). Fourier transformation then gives the the effect of the sample spinning illustrated in Fig. 2B–D. In the frequency domain, MAS generates an infinite number of sidebands at integral multiples of the spinning speed with respect to the average shift experienced by every crystallite during the sample rotation. When the sideband intensities are determined by the the can be extracted from a MAS pattern by the comparison of the experimental data with theoretical simulations (see e.g. de Groot et al., 1991). This can provide invaluable structural information, with respect to conformation and configuration, and provides in-

sight into electrostatic polarization effects at the molecular and atomic level. III. Probing the Local Environment of M(Y)210 in Rb. Sphaeroides R26 RC with MAS The initial electron transfer time in the photosynthetic reaction center of Rb. sphaeroides is sensitive to the replacement of one particular tyrosine, M210, which is highly conserved between different species, and is located close to P, and the (Fig. 3; Allen et al., 1987; Chang et al., 1991; Ermler et al., 1994). It has been suggested from electrostatic calculations that the presence of (M)Y210 lowers the energy of which affects the mechanism of electron transfer through the A branch (Parson et al., 1990). Indeed it was found that the electron transfer kinetics from P to and the redox potential of P are altered when this tyrosine, close to P, and is substituted by other amino acids (Dimagno et al., 1992). CP/MAS NMR was used to study structural and electrostatic heterogeneities of the protein matrix in the immediate environment of (M)Y210 (de Groot et al., 1990; Fischer et al., 1992; Shochat et al., 1995). Fig. 4a shows the response

Magic angle spinning NMR

of the Rb, sphaeroides R26 sample, recorded with a spinning speed at T = 230K from frozen RC solutions. The strong signals around 205, 155, and 105 ppm are from the isotope labels. A natural abundance spectrum was collected at the same spinning speed for an unlabeled R26 RC sample (Fig. 4b).

303

The difference spectrum, containing essentially only the response from the labels, is shown in Fig.4c. The label signals in Fig. 4 are from 28 different tyrosines. An assignment of the (M)Y210 signal was obtained by comparing with data for labeled (M)Y210W mutant RC, with

304

M210 changed into a tryptophan (Fig. 5). In order to emphasize signals from individual tyrosines in the protein interior the second derivatives of the data are shown. Taking the second derivative is not a commonly accepted procedure in the NMR field. However, in this case it yields surprisingly good results that reproduce extremely well. For instance, in both spectra, which are from different species, a small doublet remains in the residuals after deconvolution, around 154.5 ppm, as indicated by the circles in Fig. 5. It is now obvious that the signal at 152.2 ppm for R26 is associated with (M)Y210. The observation that only one narrow MAS NMR signal is

Huub J.M. de Groot suppressed by the mutation implies that the direct environment of M210 label is unique. Mattioli et al. (1991) reported that (M)Y210 is probably not hydrogen bonded. This is expected to produce a resonance frequency several ppm upfield from the hydrogen bonded side chains, in agreement with the NMR results. Normally, resonances of proteins, including membrane proteins, are subject to some observable inhomogeneous broadening under MAS conditions when the temperature is lowered. The signal at 152.2 ppm from (M)Y210 is very narrow and its line width is independent of temperature. Therefore this tyrosine must be located in a well defined and rigid protein environment. Its linewidth of only 34 Hz reveals that the phenolic side chain can be considered static with respect to rotational diffusion on time scales as long as (Fischer et al., 1992). Since the 152.2 ppm resonance is the only signal that is suppressed by the mutation, the MAS data provide strong evidence against structural or electrostatic (functional) heterogeneity in the chemical environment of (M)Y210 in the wild type, a possibility that has been suggested to explain the nonexponentiality of the primary charge transfer (Dimagno et al., 1991). On the scale of the MAS NMR, the protein environment of the M210 label is structurally and electrostatically (and in this sense functionally) homogeneous. For the Rb. sphaeroides RC the changes in electron transfer kinetics upon M210 mutation are quite spectacular when the tyrosine is replaced by a tryptophan, since the primary charge separation rate is slowed down by more than a factor of 10 (Shochat et al. 1994). This could be related to unanticipated structural changes or increased heterogeneity associated with the mutation. From the decon volution of the data in Fig. 5 it appears that the remaining part of the signals in the carotenoid-less R26 is almost identical in the (M)Y210W species, which contains a spheroidenone carotenoid. Notably, the differences in chemical shift for the individual components vary less than 0.4 ppm between R26 and (M)Y210W mutant (Shochat et al., 1995). Since 160 ppm corresponds to 1 unit of charge, this translates into variations of electrostatic polarization on individual residues in the order of electronic equivalents. This is very small,

Magic angle spinning NMR

305

considering that the 4'-position in the phenolic side chain of the tyrosine is probing widely different aspects of the microscopic chemical environment, for instance, differences in hydrogen bonding environment, polarization by nearby polar residues or differences in conformation. Using the label signals as internal markers, we conclude that there is very little effect of both the mutation and the presence of the carotenoid on the overall structure and electrostatic properties of the frozen protein complex. In addition, the NMR data effectively exclude that a > 25% fast electron transfer component could be due to partial back-mutation (Shochat et al. 1994). Since the signal at 152.2 ppm in Figs. 5 and 6 is completely suppressed, it can be concluded that back-mutation in the (M)Y210 is less than 5%. IV. The Configuration of the Spheroidene in the Rb. Sphaeroides RC The photosynthetic reaction center of Rb. sphaeroides 2.4.1 contains one carotenoid, spheroidene, which protects the protein complex against photodestruction, probably by quenching chlorophyll triplet states and preventing the chlorophyll-sensitized formation of singlet state oxygen, a major oxidizing agent. The carotenoid structure deduced from the X-ray diffraction studies suffers from many ambiguities and uncertainties (Ermler et al., 1994). Based on a comparison of the Raman spectra of model compounds with the Raman spectrum of the RC, Koyama et al. (1983) proposed a central cis double bond for the spheroidene in the RC. The same conclusion was reached by Lutz et al. (1987), who extracted the spheroidene from the RC under low-light conditions followed by NMR analysis of the solution. The structure around the central (15–15' ) double bond of the bound spheroidene carotenoid was also investigated in situ with low-temperature MAS NMR and site- specific isotope-labeling (Fig. 6). CP/MAS NMR spectra were obtained of R26 RCs reconstituted with spheroidene specifically enriched at the C-14' or C-15' position (de Groot et al., 1992). The analysis of the configuration around the 15–15' double bond is built upon a characterization of the shifts of the resonances in spheroidene upon isomerization.

Since the 15–15'- cis spheroidene is highly unstable and isomerizes to the all-trans form, alltrans and 15–15'- cis were used as models. It appears that both in solution and in the solid state, the shifts upon isomerization are less than one ppm, except for the two carbons situated in the 15–15' double bond, and the C-14 and C-14' on either side of it, which shift by 4–6 ppm. This phenomenon may be used to analyze the configuration of the 15-15' double bond of spheroidene in situ by using specific labelling at C-14' and C- 15' in conjunction with CP/MAS NMR. In Fig. 7 the spectrum of spheroidene R26 RCs in frozen solution at T = 220 K and is shown. In this sample the signal of the C-15' is increased by a factor 90 because of the 99% enrichment. A tiny sharp resonance from the label at 126.5 ppm is indicated by an arrow. In Table 1 the protein data for a C-14' and a C-15' label are compared with solution data for spheroidene and the two isomers. It appears that the signals from the labels of the spheroidene incorporated in the RC are both shifted upfield with respect to the all-trans molecule in solution, as is to be expected for the 15– 15' cis form. In addition, the magnitudes of the shifts compare well with the values observed for the isomerization of It is unlikely that other mechanisms than isomerization could be responsible for the upfield shifts upon incorporation of the molecule in the RC. The Raman studies by Lutz et al. (1987)

306

Huub J.M. de Groot capabilities of modern X-ray techniques (Ermler et al., 1994).

V. The Asymmetric Sphaeroides R26

have provided evidence for a planar center of the RC-bound spheroidene and the close correspondence between solution and solid state chemical shifts argues against an explanation in terms of small conformational distortions. According to the X-ray data, the labels are probably more than 0.6 nm away from the plane of the accessory bacteriochlorophyll monomer (B). This is the nearest macroaromatic cycle, and any ring current shift should be less than 0.5 ppm. Although shifts in the range of 4–6 ppm may in principle be caused by the presence of a polar group in the protein, such an explanation would be in contradiction with the X-ray structure, which reveals an apolar binding site. Hence the NMR data of the specifically enriched, reconstituted RC complexes provide convincing evidence for a 15–15'-cis conformation of the carotenoid. Determining this detail of the structure is at present still beyond the

Binding in Rb.

A first set of CP/MAS NMR results on Rb. sphaeroides R26 RCs reconstituted at the site with selectively labeled was obtained by van Liemt et al. (1995). Using site-specific labeling of positions 1, 2, 3, and 4 forming a path in the quinone ring between the two carbonyls, the protein-cofactor interactions affecting the quinone ring system can be investigated. The MAS results confirm asymmetric binding of by the protein. The most remarkable difference in NMR properties between the two carbonyl positions in is the temperature dependence of the signal at position 4, which is not detected at temperatures (Fig. 8). This temperature effect is now observed routinely in our laboratories and its origin is currently being investigated in detail. The differences in isotropic shifts between crystalline and are only 0.2–1.1 ppm for the labelled positions. This contrasts with the ground-state shift of for a mode dominated by the vibration detected by FTIR (Brudler et al., 1994, Breton et al., 1994). However, even without substantial changes of the isotropic shift, variations of principal components of the chemical shift tensor can be quite large. A deconvolution analysis using computer-generated MAS patterns of the carbonyl data in Fig. 9 indicates that the chemical shift asymmetry for the is somewhat smaller than for the 4– position in crystalline (Table 2). In a highly simplified formalism the chemical shielding of atom A can be expressed as the sum of a paramagnetic and a diamagnetic term

Magic angle spinning NMR

307

The diamagnetic shift depends on the relative placements of the atoms and the associated electron densities, while in this description the paramagnetic term takes into account the hybridization effects. In an average energy approximation, using a simple LCAO molecular orbital scheme (Pople, 1962),

308

with e and m the charge and mass of the electron, Planck’s constant divided by c the velocity of light and the average excitation energy of the excited electronic states in the absence of a magnetic field. The summation runs over neighbours B of A, the atom with the nucleus of interest and is proportional to the average value for the 2p orbitals. According to Pople (1962),

and Similar expressions exist for and The first part is expressed in terms of the generalized charge densities while the nearest neighbour bond effect is expressed in terms of the generalized bond orders Because of the ordering of the MO coefficients in the and the transitions are not contributing to the in this model. However, bond order changes may be expected to affect the principal components of the chemical shift tensor considerably through the terms that depend on From experimental data for the relation between carbonyl frequencies and bond orders (Josien et al., 1953), it can be inferred that the shift of the mode dominated by the stretch represents a variation of the bond order of approximately 0.1. A crude estimate based on trends put forward in a recent summary of characteristic shielding anisotropies and approximate calculation of shielding tensors using Eqs. (8–10), suggests that contributions to the ten-

Huub J.M. de Groot

sor elements could give rise to changes in the 10 ppm range for a bond order change of ~0.1 (Zilm and Duchamp, 1992). In addition, the can vary depending on the surrounding structure. However, with experimental errors in the range of 10–20 ppm, for the principal components of the chemical shift tensor for the carbonyls listed in Table 2, it was concluded that there is at present no conflict between the NMR shift anisotropy analysis and the quite dramatic 60 shift of the stretch mode observed with infrared spectroscopy (van Liemt et al., 1995). Apparently the two spectroscopic techniques are complementary, since the NMR is most suitable to detect charge densities, while the FTIR is much more sensitive to rehybridization effects. In line with the FTIR results, the estimates of the shift anisotropy indicate an effect at C-4 and not at C-1. The fact that a strong and narrow response for the can be detected at ambient temperature provides evidence that this carbonyl is static on the scale of the NMR and in a well-defined part of the binding pocket. On the other hand, the loss of signal for should be related to dynamic destructive interference of motion with the NMR signal. The CP/MAS experiment is sensitive to dynamic perturbations with correlation times in the 0.01–1 ms range, and in particular CP and decoupling are very sensitive to local dynamics. Recent additional experiments support these inferences and suggest that the temperature dependence of the signal is related to a shortening of the at temperatures approaching 255 K (B.J. van Rossum et al., 1996).

Magic angle spinning NMR

309

Several possible mechanisms for the temperature effect in the NMR response have now been put forward in the literature (Brudler et al., 1994; Breton et al., 1994). It could be related to dynamic hydrogen bonding interactions on the scale of the NMR, 0.01–1 ms, for instance switching between the two possible H-bond donors. In such a scheme the dynamics would freeze at lower temperatures, explaining why for K the 4 carbonyl signal can be observed. Kleinfeld et al. (1984) inferred that under physiological conditions transitions between different structural states occur on a time scale affecting the distance by However, a definite explanation in terms of a precise molecular mechanism cannot be given as yet, and further studies are necessary. VI. New Developments. CIDNP and Correlation Spectroscopy Zysmilich and McDermott (1994) reported the observation of MAS photo-CIDNP under continuous illumination of enriched RC with forward electron transfer blocked by depletion or by pre-reduction of the This is the first observation of CIDNP in the solid state with the chromophores in the protein driving the nuclear spin selection mechanism. According to the discoverers of the effect, it relies on a partial interconversion between and followed by charge recombination, nuclear relaxation and ground state decay in this channel. An example of the phenomenon is shown in Fig. 10, for depleted RC in the dark and under continuous illumination. The data under illumination reveal strong emissive signals that are characteristic for the CIDNP effect, with an estimated enhancement of ~300 over the natural abundance response. The CIDNP response is probably associated with the nuclei of P and the signal strengths are essentially independent of proton decoupling power level. When the precise mechanism of the enhancement is fully understood, this new CIDNP approach could provide information about the electron density in the transient species as opposed to the stable oxidized radical cation extensively

studied by EPR. Moreover, the chemical shifts of the nuclei in the special pair can be characterized. Another recent technical development concerns MAS dipolar correlation Spectroscopy of uniformly enriched chlorophyll a (Chl a), the ubiquitous chromophore involved in light-harvesting and photochemistry in the photosynthetic energy conversion processes of green plants and related organisms (Fig. 11). When exposed to Chl a can aggregate and it is thought that water molecules form intermolecular hydrogenbonded bridges from the Mg to the carbonyls and (Katz et al., 1991). The CP/MAS NMR spectrum of the enriched Chl-a/water aggregate is shown in Fig. 11 (Boender et al., 1995). The Chl- a/water micelle and similar aggregates, for instance the BChl c entities in chlorosome antennae, provide typical examples of systems where high-resolution solid state NMR spectroscopy will be indispensable to provide in-

310

formation at the atomic level about the precise structure. Recently, important progress towards a general approach for MAS NMR structure refinement using 2-D dipolar correlation spectroscopy in uniformly enriched moderately sized molecules was reported by Boender et al. (1995). In their study, pure-phase MAS NMR dipolar correlation spectra of the enriched Chla/water aggregates were obtained with the pulse sequence of Fig. 12. An integral multiple n of XY-8 phase alternated rotor-synchronized trains of pulses was used to promote exchange of coherence through dipolar interactions during the mixing period (Bennet et al., 1992). In the average Hamiltonian approximation for a nuclear spin I = 1/2 coupled to another spin S = 1/2 the density matrix is:

Huub J.M. de Groot

Magic angle spinning NMR

at the start of detection time Therefore polarization transfer between and can take place during mixing. In Fig. 13 an example of a 2-D correlation spectrum of the enriched Chl

311

aggregate, collected with an 10,000 ± 5 Hz and ± is shown, with strong 2–D cross-peaks revealing transfer of coherence. The horizontal and vertical lines illustrate how the molecular framework gives rise to a correlation network, similar to the assignment procedures in solution NMR. A complete assignment of the resonances obtained from the analysis of 2-D correlation spectra is compared with data for the molecule in solution. It appears that the two methyl groups and are shifted upfield by more than 3 ppm in the aggregate with respect to their chemical shifts in solution. According to Boender et al. (1995), this is probably induced by the ring-current of the of the macroaromatic cycle

312 of a neighbouring molecule, and would be in agreement with current models for the aggregate structure, in which the edges of the macrocycles of neighbouring molecules are above one another, and the atoms are shifted by the macroarornatic cycle of a neighbouring molecule. Moreover, carbon 2 appears to correlate with carbon 14, carbon has a correlation with carbon 20 and carbon 4 as a correlation with carbon 12. The carbon atoms 12, and 14 are on the opposite side of the Chl a ring with respect to carbons 2, 4 and 20. Such cross-peaks are therefore most likely associated with short intermolecular distances. This indicates that the stacking of the Chl a rings proceeds roughly in agreement with the model of Chow et al. (1975), which was based on the X-ray structure of ethyl chlorophyllide a, although the presence of the correlation between and 20 suggests that the rings are rotated slightly along an axis perpendicular to the plane of the molecule, generating a curvature of the array of stacked rings which could explain the formation of tubular micelles in the Chl-a/water aggregates. Dipolar correlation spectroscopy of multispin systems as a generally applicable research tool is thus forthcoming, but it should be mentioned that it still requires the development of a new generation of high-field wide-bore solid NMR instruments (750 MHz) with high-speed (>20 kHz) MAS probes and excellent performance with respect to long-term stability. In addition, further development of NMR-methods and advanced partial labeling schemes of larger systems will be required. Nevertheless, the experiment of Fig. 13 represents an important step in the development of MAS NMR as a tool for comprehensive structure determination at the atomic level in solids that are neither accessible to X-ray nor to solution NMR methods. Many biological subcellular lifesustaining systems fall into this category. VII. Concluding Remarks In this chapter, we have summarized how MAS NMR was used to investigate structure-function properties of Rb. sphaeroides photosynthetic reaction centers and their significance to fundamental molecular mechanisms of photochemical energy conversion. MAS NMR in conjunction with

Huub J.M. de Groot site-directed specific isotope labeling is at present a technological growth area in the field of membrane protein research. It is rapidly developing into a versatile technique suitable for obtaining information about microstructure and structurefunction relations in essentially unperturbed biologically ordered solid-type systems, even when there is no translation symmetry and X-ray or solution NMR are not applicable. Since in principle atomic selectivity can be obtained in systems with molecular weights counting several megadalton, many of the components of the photosynthetic apparatus are already within reach of the technique, and more MAS NMR applications in photosynthetic systems are forthcoming. Acknowledgements The contributions of Ir. G. Boender, Dr. M. Fischer, Prof. H. Frank, Dr. P. Gast, Dr. R. Gebhard, Drs. K. v.d. Hoef, Dr. W. van Liemt, Mrs. S. Shochat, Dr. C. Violette and Dr. C. Winkel to the MAS explorations of the bacterial RC are gratefully acknowledged. Special thanks are accorded to Prof. A.J. Hoff, Dr. J. Raap and Prof. J. Lugtenburg, for continuing collaborations and for making the initial MAS studies of photosynthetic reaction centers possible. This work was supported by the Royal Netherlands Academy of Arts and Sciences (KNAW), the Foundations for Biophysical and Chemical Research (financed by The Netherlands Foundation for Scientific Research), and the Commission of the European Communities. References Allen JP, Feher G, Yeates TO, Komiya H and Rees DC (1987) Structure of the reaction center from Rhodobacter sphaeroides R26: The cofactors. Proc Natl Acad Sci USA 84: 5730-5734. Bennet AE, Ok JH, Griffin RG and Vega S (1992) Chemical shift correlation spectroscopy in rotational solids: Radio frequency- driven dipolar recoupling and longitudinal exchange. J Chem Phys 96: 8624–8627. Boender GJ, Raap J, Prytulla S, Oschkinat H and de Groot HJM (1995) MAS NMR structure refinement of uniformly enriched chlorophyll-a/water aggregates with 2-D dipolar correlation spectroscopy. Chem Phys Lett 237: 502– 508. Breton J, Boullais C, Burie J-R, Nabedryk E and Mioskowski C (1994) Binding sites of quinones in photosynthetic bac-

Magic angle spinning NMR terial reaction centers investigated by light-induced FTIR difference spectroscopy: Assignment of the interactions of each carbonyl of in Rhodobacter sphaeroides using sitespecific ubiquinone. Biochemistry 33: 14378– 14386. Brudler R, de Groot HJM, van Liemt WBS, Steggerda WF, Esmeijer R, Gast P, Hoff AJ, Lugtenburg J, and Gerwert K (1994) Asymmetric binding of the 1– and groups of in Rhodobacter sphaeroides R26 reaction centres monitored by Fourier transform infra-red spectroscopy using site-specific isotopically labelled ubiquinone-10. EMBO J 13: 5523–5530. Chang CH, E-Kabbani O, Tiede D, Norris J and Schiffer M (1991) Structure of the membrane-bound protein photosynthetic reaction center from Rhodobacter sphaeroides. Biochemistry 30: 5352–5360. Chow HC, R. Serlin R and Strouse CE (1975) The crystal and molecular structure and absolute configuration of ethyl chlorophyllide a dihydrate. A model for the different spectral forms of chlorophyll a. J Am Chem Soc 97: 7230–7242. de Groot HJM, Raap J., Winkel C., Hoff AJ and Lugtenburg J (1990) Magic-angle-spinning NMR with atomic resolution of a photosynthetic reaction center enriched in Chem Phys Lett 169: 307–310. de Groot HJM, Smith SO, Kolbert AC, Courtin JML, Winkel C, Lugtenburg J, Herzfeld J and Griffin RG (1991) Iterative fitting of magic-angle-spinning NMR spectra. J Magn Res 77: 251–257. de Groot HJM, Gebhard G, vd Hoef I, Hoff AJ, Lugtenburg J, Violette CA and Frank HA (1992) magic angle spinning NMR evidence for a configuration of the spheroidene in the Rhodobacter sphaeroides photosynthetic reaction center. Biochemistry 31: 12446–12450. DiMagno TJ, Rosenthal SJ, Xie X, Du CK, Chan CK, Hanson D, Schiffer M, Norris J and Fleming GR (1992) Recent experimental results for the initial step of bacterial photosynthesis. In: Breton J and Verméglio A (ed) The Photosynthetic Bacterial Reaction Center II, pp 209–217. Plenum Press, New York. Ermler U, Fritzsch G, Buchanan SK and Michel HM (1994) Structure of the photosynthetic reaction centre from Rhodobacter sphaeroides at 2.65 Å resolution: cofactors and protein-cofactor interactions. Structure 2: 925–935. Fischer MR, de Groot HJM, Raap J., Winkel C, Hoff AJ and Lugtenburg J (1992) magic angle spinning study of the light-induced and temperature-dependent changes in Rhodobacter sphaeroides R26 reaction centers enriched in tyrosine. Biochemistry 31: 11038–11049. Josien ML, Fuson N, Lebas JM and Gregory TM (1953) An infrared spectroscopic study of the carbonyl stretching frequency in a group of ortho and para quinones. J Chem Phys 21: 331–340. Katz JJ, Bowman MK, Michalski TJ and Worcester DL (1991) Chlorophyll aggregation: Chlorophyll/water micelles as models for in vivo long-wavelength chlorophyll. In: Scheer H (ed) Chlorophylls, pp 211–235. CRC Press, Boca Raton. Kleinfeld D, Okamura MY, and Feher G (1984) Electrontransfer kinetics in photosynthetic reaction centers cooled to cryogenic temperatures in the charge-separated state:

313 Evidence for light- induced structural changes. Biochemistry 23: 5780–5786. Koyama Y, Kito M, Takii T, Saiki K, Tsukida K and Yamashita J (1983) Configurations of neurosporene isomers isolated from the reaction center and the light-harvesting complex of Rhodobacter sphaeroides G1C. A resonance Raman, electronic absorption, and study. Photochem Photobiol 48: 107–114. Lutz M, Szaponarski W, Berger G, Robert B and Neumann J (1987) The stereoisomerism of bacterial, reaction centerbound carotenoids revisited: An electronic absorption, resonance Raman and study. Biochim Biophys Acta 894: 423–433. Mattioli TA, Gray KA, Lutz M, Oesterhelt D and Robert B (1991) Resonance Raman characterization of Rhodobacter sphaeroides reaction centers bearing site-directed mutations at tyrosine M210. Biochemistry 30: 1715–1722. Mehring, M (1983) High Resolution NMR in Solids. SpringerVerlag, Berlin. Parson WW, Chu ZT and Warshel A (1990) Electrostatic control of charge separation in bacterial photosynthesis. Biochim Biophys Acta 1017: 251–272. Pople JA (1962) Molecular-orbital theory of diamagnetism. I. An approximate LCAO scheme. J Chem Phys 37: 53–59. Shochat S, Arlt T, Francke C, Gast P, van Noort PI, Otte SCM, Schelvis HPM, Schmidt S, Vijgenboom E, Vrieze J, Zinth W and Hoff AJ (1994) Spectroscopic characterization of reaction centers of the (M)Y210W mutant of the photosynthetic bacterium Rhodobacter sphaeroides. Photosynth Res 40: 55–66. Shochat S, Gast P, Hoff AJ, Boender GJ, van Leeuwen S, van Liemt WBS, Vijgenboom E, Raap J, Lugtenburg J and de Groot HJM (1995) MAS NMR evidence for a homogeneously ordered environment of tyrosine M210 in reaction centers of Rhodobacter sphaeroides, Spectrochim Acta 51A: 135–144. van Liemt WBS, Boender GJ, Gast P, Hoff AJ, Lugtenburg J and de Groot HJM (1995) magic angle spinning NMR characterization of the functionally asymmetric binding in Rhodobacter sphaeroides R26 photosynthetic reaction centers using site-specific ubiquinone-10, Biochemistry 34: 10229–10236. van Rossum B-J, van Liemt WBS, Boender GJ, Gast P, Hoff AJ, Lugtenburg J and de Groot HJM (1996) MAS NMR relaxation study of the binding in Rhodobacter sphaeroides-R26 reaction centers. In P. Mathis (ed) Photosynthesis: From Light to Biosphere Vol. I, pp. 899–902. Kluwer Academic Publisher, Dordrecht. 1995. Zilm KW and Duchamp JC (1992) Comparisons of shielding anisotropies for different nuclei and other insights into shielding from an experimentalist’s viewpoint. In: Tossel JA (ed) Nuclear Magnetic Shieldings and Molecular Structure, NATO ASI Series, Series C: Mathematical and Physical Sciences, pp. 315–334. Kluwer Academic Publishers, Dordrecht. Zysmilich M and McDermott AE (1994) Photo-chemically induced dynamic nuclear polarization in the solid state spectra of reaction centers from photosynthetic bacteria Rhodobacter sphaeroides R-26. J Am Chem Soc 116: 8362– 8366.

This page intentionally left blank

PART THREE

Structure and oxygen

This page intentionally left blank

Chapter 19 Structure Determination of Proteins by X-Ray Diffraction† Marianne Schiffer* Center for Mechanistic Biology and Biotechnology, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439–4833, USA

Summary I. Introduction II. Theory, Equations, and Some of the Terms Used in X-Ray Structure Determination III. Determination of Protein Structure A. Characterization of the Crystals B. Data Collection C. Determination of Preliminary Structure (or Phases) D. Refinement of Structure IV. Quality of the Structure V. Comparison with Structural Information Obtained with Other Techniques Acknowledgements References

317 317 318 318 318 319 319 321 323 323 323 324

Summary The arrangement of atoms within protein molecules, as determined by X-ray diffraction of single crystals, forms some of the basic data for many spectroscopic, protein engineering, and computation studies. In this chapter, the terminology used in protein crystallography is explained and a brief description is given of how a protein structure is determined. Further, some ideas are discussed on how to judge X-ray data and the resulting protein structure and how to compare the data derived by X-ray diffraction with data obtained by other methods. I.

Introduction

from the Protein Data Bank at Brookhaven National Laboratory.‡ The number of protein structures is increasing at a very high rate; in 1989, there were 400 structures while the January 1994 release of the Data Bank contained 2143. In the last ten years, protein crystallography has become a significant component of photosynthesis research. It is the aim of this chapter to make the protein crystallography literature more accessible and comprehensible to the typical spectroscopist and biologist. Since there are several very good textbooks (e.g., Blundell and Johnson, 1976; McPherson, 1982; McRee, 1993), two volumes in the Methods of Enzymology (Wyckoff et al., 1985a,b), and an excellent mini-review (Eisenberg and Hill, 1989) that describe the determi-

The arrangement of atoms within protein molecules, as determined by X-ray diffraction of single crystals, forms some of the basic data for many spectroscopic, protein engineering, and computational studies. Depending on the specific application, it is important to know how to evaluate the atomic coordinates that can be obtained †

The U.S. Government’s right to retain a non-exclusive, royalty-free licence in and to any copyright is acknowledged. *Correspondence: Fax: 1-708-2525517; E-mail: [email protected]. ‡ Coordinates can be obtained by FTP from pdb.pdb.bnl.gov (Internet address 13.199.144.1).

317 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 317–324. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

318 nation of protein structures by X-ray diffraction, I will concentrate on defining the terminology used in protein crystallography, provide a brief description of how a protein structure is determined, and give some ideas and personal views on how to judge X-ray data and the resulting protein structure. The emphasis in this chapter is on the methods of structure determination when a suitable protein crystal is already in hand.

II. Theory, Equations, and Some of the Terms Used in X-Ray Structure Determination Electrons within a crystal scatter X-rays in directions that are determined by the geometry of the unit cell of the crystal and its orientation relative to the X-ray beam. The X-rays can be considered as “reflected” from a plane or family of planes of the crystal. Each plane is defined by three indices: h, k, and l. The perpendicular distance between these planes is The Bragg equation relates d with the scattering angle and the X-ray wavelength in angstroms.

The quantity (in angstroms) is quoted in general as the resolution of the dataset. It defines the largest angle of scattering the crystal is capable of, or, in other words, the crystalline order within the crystal. The more ordered the crystal is, the more higher-resolution observations can be made and the more accurately the structure can be determined. The higher the resolution of the data, the lower the Data at 2.5 Å resolution is considered “high” resolution and good for a protein crystal; 1.8 Å resolution is excellent, and 5–6 Å is low resolution. The intensity (I) of the reflected beam is the experimental quantity that is measured for all possible reflections of the crystal. The intensity (I) is proportional to the square of the structure factor (F). If the crystal structure is known, the structure factor can be calculated according to the following formula:

Marianne Schiffer

where is the atomic scattering factor or form factor that depends on the electron distribution in the nth atom (carbon, oxygen, etc.); are the coordinates of the atom and B is the atomic temperature factor. where is the square of the average displacement of the atom. The structure factor of each reflection (hkl) is a vector that consists of an amplitude F(hkl) and a phase angle or phase Only the amplitude of the reflection can be measured directly. The phases of the reflections have to be determined using other methods. When both the amplitudes and phases are known, the three-dimensional electron density distribution of the crystal can be calculated. Electron density at point xyz in the unit cell is defined by a Fourier series in three dimensions. An electron density map or Fourier map is calculated according to:

The summation is over the three indices, hkl; V is the volume of the unit cell. III. Determination of Protein Structure The determination of a crystal structure has several steps: characterization of the crystal, collection of diffraction data, determination of preliminary phases (and structure), refinement of phases (and structure), and description of the structure. A review paper by Brünger and Nilges (1993) describes the computational aspects of structure determination and refinement. A review by Finzel (1993) summarizes the most frequently used computer program packages.

A.

Characterization of the Crystals

Unit cell dimensions, three axes (a, b, and c), and three angles and define the basic repeat

X-ray diffraction unit of the crystal. The cell dimensions are determined from the geometry of the diffraction pattern, obtained from precession photographs or area detector data. The space group is determined from the symmetry of the diffraction pattern and from reflections that are systematically absent. The space group defines the symmetry (symmetry operators) within the crystal. Possible space groups and their properties are listed in the International Tables for Crystallography, Volume A (Hahn, 1983). The space group symmetry operates on the asymmetric unit of the crystal. The asymmetric unit usually contains one molecule, but it can include more than one molecule or part of the molecule. If it contains more than one molecule (e.g., two molecules), the structure that must be determined is two times larger than the single molecule and, therefore, the unique set of reflections defining the structure is also two times larger. If a partial molecule is the asymmetric unit, the symmetry element of the unit cell relates two halves of the molecule; e.g., a twofold axis and the number of unique observations defining the molecule is also halved. In other words, the amount of diffraction data required to “solve” a structure is determined not only by the size of the protein itself, but also by the size of the asymmetric unit. The unit cell volume (calculated from the unit cell dimensions) and the space group determine the numbers of molecules in the unit cell and in the asymmetric unit. Volume/dalton values for soluble proteins range from 1.7 to with an average value of (Matthews, 1968). For membrane proteins, because of the presence of detergent in the crystal, is larger and can vary from 3.5 to The number of molecules in the asymmetric unit can be determined experimentally by measuring the density of the crystal, but more usually it is done by calculating the values and choosing a number that will result in a value that falls in the expected range.

B. Data Collection There are several different methods for data collection. The method of choice may be determined by crystal cell dimensions and how well the crystal diffracts X-rays. Diffractometers measure one re-

319 flection at a time, and each reflection can be very accurately determined; diffractometers are now mostly used for small molecules. Area detectors can simultaneously measure a plane of reflections (possibly several hundred of them) at one orientation of the crystal; such detectors include electronic area detectors, X-ray film, and imaging plate detectors (for comparison of these detectors, see Eikenberry et al., 1992). Protein crystals deteriorate (the intensity of the reflections “decays”) in the X-ray beam as a function of time. Although time-dependent decay correction can be applied to the diffraction data, after a certain point a fresh crystal is used. Corrections for the asymmetric absorption of X-rays by the crystal and surrounding mounting material are made using a predetermined function for diffractometer data. For area detector data, the same reflection is measured several times at different crystal settings. Scale factors applied to the data collected at different crystal settings usually correct for absorption and crystal decay. To assess the quality of the data, R-sym, the agreement between equivalent reflections measured at different geometries, have to be examined. R-sym is generally better for reflections with higher intensities and, therefore, it gets worse with increasing resolution where the reflections get weaker. Weak, low-intensity reflections are also strongly affected by the background correction. Of course, R-sym also depends on whether the absorption and decay corrections were effective. A good dataset will have an overall R-sym of less than 0.1. As in other measurements, maximizing the peak signal-to-background ratio is important. This is usually achieved by using crystals with as large a volume as possible or with intense Xray beams such as those provided by synchrotron radiation sources.

C. Determination of Preliminary Structure (or Phases) The most generally used method for determining the structure of large protein molecules is multiple isomorphous replacement or MIR (Dickerson et al., 1961). If a structure of a homologous protein molecule is known, then molecular replacement (Rossmann, 1972,1990) can be applied. For example, the MIR method was used to determine

320 the structure of the reaction center from Rps. viridis, (Deisenhofer et al., 1984) while molecular replacement was used to determine the Rb. sphaeroides reaction center structure; this determination exploited the availability of the coordinates from Rps. viridis as the search structure (Chang et al,. 1986; Allen et al., 1986). These molecules have similar structures but crystallized in different unit cells and had different space groups. In multiple isomorphous replacement, heavy atoms (in the form of salts of heavy atoms; e.g., or organic derivatives, e.g., para-chloroHg-phenol) are introduced into a crystal generally by soaking the “native” crystal in a solution containing the heavy atom (HA). In special cases, the HA can be covalently bound to the protein or involve the replacement of a metal atom in the structure with a heavier one. For this method to work, the HA contribution (1) has to be significant, (2) its introduction does not change the unit cell of the protein (the two crystals are isomorphous), and (3) only few atoms or molecules are introduced per molecule of protein, so their position can be determined. As a first step, the HA is located in the unit cell (by Patterson or sometimes by direct methods, that are generally used to determine small molecule structures). The Patterson function or map is a Fourier series calculated with squared structure factors. The peaks in a Patterson map correspond to vectors between atoms. The Patterson map calculated with the difference between the derivative and native structure factors squared will give the vectors between the heavy atoms. When there are relatively few HAs per protein molecule, their locations can be determined from the Patterson map. Direct methods (Karle, 1989; Hauptman, 1989), based on the relationships between structure factors, are used when Patterson methods fail to locate the HA positions. The positions of the HAs are then refined using the quantity Several HA “derivatives” are prepared and analyzed, and together they are used to determine the phases for the protein (“native” crystals). The quality of the resultant phases depends on the quality of the HA and native datasets, how isomorphous the derivative is, the interpretation of the HA atom data (where all HA sites are found), and the size

Marianne Schiffer of the contribution of the HA to the native data. If the quality of the phases (and therefore the resulting Fourier or electron-density map) is good, a crystal structure can be interpreted even at 4 Å resolution. Certainly at 3.5 Å resolution, the polypeptide chains can be traced and side chains can be identified. In molecular replacement, a known homologous structure, the search molecule, is located in the new unit cell by first finding its orientation (the rotation problem) and then its position (the translation problem). In this method, vectors between objects in the unit cell are analyzed. Several methods and program packages are available for these searches (e.g., Fitzgerald, 1988; Brünger, 1990). In general, the signal-to-noise ratio for the rotation problem is better than that for the translation problem. Most often, low resolution data (between 20 Å and 6–4 Å resolution) are used for the searches. After the molecule location is found, its position and orientation are refined with rigid body refinement. It is very important to check that the resulting molecular positions in the unit cell make sense; that is to say, two molecules don’t interfere with each other and don’t occupy the same space. Another way to check the molecular replacement solution is to see if there is density for parts of the molecule that were not part of the search molecule. For example, if the search molecule had an Ala in a certain position while the molecule had a Trp, then if the solution is correct, electron density should be present for Trp (see Chang et al., 1986, and Fig. 1). This should also work for longer segments, but very often loops on the surface of the molecule only in contact with the solvent don’t have density. The lack of density could signal physical disorder (more than one conformation). It could also result from a problem with molecular replacement method, because for the same structure, electron density for the fragment can be found when using MIR methods and not with molecular replacement method. After preliminary phases are obtained by any method, the molecule is built “into” the electron density map using the amino acid sequence of the protein. The features of the electron density are interpreted in terms of molecular structure. At low resolution, observable features include the

X-ray diffraction

molecular envelope and At medium resolution, the protein backbone and side chains become apparent. Protein crystals don’t diffract highly enough to see individual atoms. “Mini maps”, or small-scale maps, are helpful for tracing the polypeptide chain and locating the alpha carbon positions. Further building of the protein structure is carried out with computer graphics programs like FRODO (Jones, 1978) or O (Jones et al., 1991). The construction of the molecular model takes into account the geometry and contacts that are expected from small molecule and other protein structures. Of course, if the structure was solved by molecular replacement, the coordinates of the search molecule can be used as the starting point of rebuilding.

D.

Refinement of Structure

The refinement of the “structure” consists of improving the atomic positions in the model and improving the temperature factors so that the structure factors calculated or using the improved structure will more closely approxi-

321

mate the observed structure factors or The quality of the result is expressed by the residual, or R-factor summed over all reflections, where

Once a preliminary structure is built according to the electron density map, refinement programs can be used. These programs first “regularize” the structure so it will have the expected bond angles, bond lengths, and nonbonded contacts. Then they refine the atomic positions against the “observed” structure factors. One of the methods uses least squares refinement; e.g., programs PROLSQ (Hendrickson, 1985) and TNT (Tronrud et al., 1987). In least squares refinement, because the ratio of observations to atomic parameters is relatively low, restraints on the structure are used as additional observations. The restraints are based on the known rules of the stereo chemistry of the molecule, and they keep the molecule close to the ideal geometry. (For small

322 molecular structures that diffract to atomic resolution, there are about ten times as many observations as parameters being refined; therefore, the least squares refinement is well behaved.) During the refinement, different weights can be given to the observations and the restraints. Depending on the relative weights, good geometry can be maintained, or a better fit to the structure factors, as reflected by the R-factor, can be attained. Ideally, the aim is to get a low Rfactor with good geometry. Besides the three coordinates of each atom, overall scale factors and temperature factors are optimized in the beginning of the refinement process. As the refinement progresses, individual atomic temperature factors are introduced. The refinement technique known as simulated annealing (SA) has a larger radius of convergence than least squares refinement. With SA refinement, atoms can move over 2 Å to their correct positions. This method is based on molecular dynamics and was introduced by Brünger et al. (1987). In X-PLOR, the total energy of the system (the crystal) is refined. In addition to bonded and nonbonded interactions, the experimental diffraction data is also considered as an energy term. After the relative weights are established for the structural and diffraction data, the energy of the system is minimized. This is followed by molecular dynamics calculations that allow the protein to move over energy barriers. After molecular dynamics calculations at different temperatures, using different protocols, the energy of the system is further minimized. The method is very computer-intensive, but because it can refine structures that are farther from the correct one, less manual rebuilding is needed, and generally the overall refinement process is faster. The SA refinement generates an ensemble of structures that agree with the X-ray data; small outside loops can take different conformations (Xu et al., 1990) and have to be examined carefully. Cycles of automated refinement are interspersed with “manual” refinement that involves examination of the electron density map and making required adjustments by rebuilding the structure using a computer graphics program (e.g., FRODO or O). During the refinement of the structure, the calculated phases also improve and get closer to the correct phases; there-

Marianne Schiffer fore, electron density maps calculated with the improved and will show an improved structure. The refinement process alternates between improving the coordinates and examining the fit of the coordinates to the calculated electron density map. (A map calculated with and coefficients reproduces the structure at that point of the refinement. A map calculated with and has the correct measured amplitude for the structure factor; therefore, it will be closer to the real structure.) Difference maps calculated with coefficients and areused to identify errors because they give more information than the maps calculated with In theory, when an atom is incorrectly positioned, it should be observed at half height in a map, while another half-height peak should show up in the correct position. In an map, there should be a half- height positive peak in the correct position and a half-height negative peak in the position where it was introduced by mistake. Therefore, the combination of these two maps should have a positive peak in the correct position and a negative one in the incorrect one. Difference maps calculated with both and coefficients are used later in the refinement process to locate cofactors, ions, detergent, and water molecules. To identify these additional small molecules, the shape of the “extra” electron density and the hydrogen bonding potential have to be taken into account. The newly rebuilt protein molecule and the additional atoms then are refined further to see if the fit improved in the refinement and whether the new features improve in the subsequent maps. In general, when an atom is introduced, it will appear in the maps, even when its position is incorrect; therefore, it is very hard sometimes to eliminate incorrect features of the molecule that were introduced at some point in the modelbuilding process. If that atom or segment has been part of the molecule for many refinement cycles, removing it from the phase calculation is not sufficient. The structure has to be refined without the segment in question, and it is also helpful to remove some of the surrounding residues as well before new refinement cycles are carried out and new maps are built. Such considerations led to the recent introduction of the

X-ray diffraction free R-factor (Brünger, 1993), which is a good monitor of the progress of the refinement. To determine the free R-factor, a fraction of the reflections are not refined. If the structure improved during the refinement of the majority of the reflections, this should improve the R-factor also for the ones that were left out. IV. Quality of the Structure How can you tell how good a structure is? This question has been the topic of discussions (see Bränden and Jones, 1990). The R-factor is one indication. A well refined high resolution 2–2.5 Å structure that includes solvent (water) molecules will have an R-factor of 0.15–0.20. Medium resolution structure of about 3 Å that does not include solvent molecules will have an Rfactor of 0.20–0.25. But even with such good Rfactors, segments of the structure can be incorrect. The accuracy of the structure depends on the goodness of the diffraction data, correctness of phase angles, the resolution of the data, and how carefully the structure was determined. For a de novo structure, the ø, angles (Ramachandran plot) reflecting the nonbonded contacts, distances between neighboring protein molecules, and in general parameters that were not explicitly restrained can be good indicators of structure quality; they should have values expected from studies of small molecules. The “correctness” of the structure is also a good indicator; i.e., (most) hydrophobic residues are on the inside while hydrophilic residues are on the outside (Lüthy et al., 1992). Judging a solution obtained by molecular replacement is more difficult because this “correctness” was already met in the search molecule. Here it is important to examine intermolecular contacts; they are expected to make chemical sense. The course of the refinement is also an important indicator. Did the refinement proceed smoothly and did the free R-factor improve upon introduction of new features? For example, lower R-factors can be achieved by introducing water molecules, but they do not necessarily improve the structure.

323 V. Comparison with Structural Information Obtained with other Techniques The refinement statistics give overall values of bond angles, bond lengths, etc. Luzzatti plots (Luzzatti, 1952) give average positional errors in the atomic coordinates. The positional error can vary from 0.2 Å to 0.5 Å, for a refined 2 Å structure to a refined 3 Å structure. Because these are average values, the errors in the specific residues or chain segments can be much larger; as with all statistical values, some can lie outside the limit. Protein crystal structures are time-averaged structures; examination of the atomic temperature (B) factors (part of the data deposited in the Protein Data Bank) gives an indication of how well the positions of the chain fragments or residues are known. The B values are generally low in the middle of the molecule and increase for outside loops or sidechains on the surface of the molecule. High temperature factors can be caused by movement of the residue or chain segment, or by static disorder. High B-factor also could mean that part of the molecule is incorrect. Contacts of protein molecules within the crystal (lattice contacts) can influence both the structure and the B-factor at or near the contact point. To compare data obtained by X-ray diffraction with data obtained by other spectroscopic techniques, several facts have to be remembered. Proteins in the crystals are at very “high” concentrations and (sometimes) in the presence of high salt or other precipitants, whereas spectroscopy is usually carried out in solutions which are more diluted. The pH and the composition of the buffer used to obtain the crystal and to carry out the spectroscopic measurement are also likely to be different. Acknowledgements I thank Dr. Fred J. Stevens for helpful discussions, Dr. Rosemarie Raffen and George Johnson for reading the initial manuscript, Dr. Phani Pokkuluri for making the figure, and David E. Nadziejka for technical editing assistance. Supported by the U.S. Department of Energy, Office of Health and Environmental Research, under

324 Contract No. W-31–109–ENG-38; also supported by Public Health Service Grant GM36598.

References Allen JP, Feher G, Yeates TO, Rees DC, Deisenhofer J, Michel H and Huber R (1986) Structural homology of reaction centers from Rhodopseudomonas sphaeroides and Rhodopseudomonas viridis as determined by X-ray diffraction. Proc Natl Acad Sci USA 83: 8589–8593 Blundell TL and Johnson LN (1976) Protein Crystallography. Academic Press, London Bränden CI and Jones A (1990) Between objectivity and subjectivity. Nature 343: 687–689 Brünger AT (1990) Extension of molecular replacement: a new search strategy based on Patterson correlation refinement. Acta Cryst A46: 46–57 Brünger AT (1993) Assessment of phase accuracy by cross validation: the free R value. Methods and applications. Acta Crystallogr D49: 24–36 Brünger, AT and Nilges M (1993) Computational challenges for macromolecular structure determination by X-ray crystallography and solution NMR-spectroscopy. Quart Rev Biophys 26: 49–125 Brünger AT, Kuriyan J. and Karplus M (1987) Crystallographic R factor refinement by molecular dynamics. Science 235: 458–460 Chang CH, Tiede D, Tang J, Smith U, Norris J and Schiffer M (1986) Structure of Rhodopseudomonas sphaeroides R26 reaction center. FEBS Lett 205: 82–86 Deisenhofer J, Epp O, Miki K, Huber R and Michel H (1984) X-ray structure analysis of a membrane protein complex. J Mol Biol 180: 385–398 Dickerson RE, Kendrew JC and Strandberg BE (1961) The crystal structure of myoglobin: phase determination to a resolution of 2 Å by the method of isomorphous replacement. Acta Cryst 14: 1188–1195 Eikenberry EF, Tate MW, Bilderback DH and Gruner SM (1992) X-ray detectors: comparison of film, storage phosphors and CCD detectors. In: Photoelectronic Image Devices 1991, Bristol: Institute of Physics 2: 273–280 Eisenberg D and Hill CP (1989) Protein crystallography: more surprises ahead. Trends Biochem Sci 14: 260–264 Finzel BC (1993) Software for macromolecular crystallography: a user’s overview. Current Opinion in Structural Biology 3: 741–747 Fitzgerald PMD (1988) MERLOT, an integrated package of

Marianne Schiffer computer programs for determination of crystal structures by molecular replacement. J Appl Crystallogr 21: 273–278 Hahn T (1983) International Tables for Crystallography, Volume A, Space-group symmetry. D. Reidel Publishing, Dordrecht Hauptman HA (1989) The phase problem of X-ray crystallography. Physics Today November 24–29 Hendrickson WA (1985) Sterochemically restrained refinement of macromolecular structures. Methods Enzymol 115: 252–270 Jones TA (1978) A graphics model building and refinement system for macromolecules. J Appl Cryst 11: 268–272 Jones TA, Zou J-Y, Cowan SW and Kjeldgaard M (1991) Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Cryst 47: 110–119 Karle J (1989) Direct methods in protein crystallography. Acta Cryst A45: 765–781 Lüthy R, Bowie JU and Eisenberg D (1992) Assessment of protein models with three-dimensional profiles. Nature 356: 83–85 Luzzatti V (1952) Traitement statistique des erreurs dans la détérmination des structures cristallines. Acta Cryst 5: 802– 810 Matthews BW (1968) Solvent content of protein crystals. J Mol Biol 33: 491–497 McPherson A (1982) Preparation and Analysis of Protein Crystals. John Wiley and Sons, New York McRee DE (1993) Practical Protein Crystallography. Academic Press, New York Rossmann MG (1972) The Molecular Replacement Method: A Collection of Papers on the Use of Non-Crystallographic Symmetry. Gordon and Breach Science Publishers, New York Rossmann MG (1990) The molecular replacement method. Acta Cryst A46: 73–82 Tronrud DE, Ten Eyck LF and Matthews BW (1987) An efficient general-purpose least-squares refinement program for macromolecular structures. Acta Cryst A43: 489–501 Wyckoff HW, Hirs CHW and Timasheff SN (1985a) Diffraction Methods for Biological Macromolecules, Part A, Methods in Enzymology. Volume 114, Academic Press, New York Wyckoff HW, Hirs CHW and Timasheff SN (1985b) Diffraction Methods for Biological Macromolecules, Part B, Methods in Enzymology. Volume 115, Academic Press, New York Xu Z-B, Chang C-H and Schiffer M (1990) Testing the procedure of simulated annealing by refining homologous immunoglobulin light-chain dimers. Protein Engineering 3: 583–589

Chapter 20 Electron Microscopy Egbert J. Boekema* and Matthias Rögner1 Bioson Research Institute, Biophysical Chemistry, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands 1 Institute of Botany, University of Münster, Schlossgarten 3, D-48149 Münster, Germany Summary I. Principles A. Introduction B. The Electron Microscope C. Specimen Preparation D. Averaging Methods E. Possibilities of EM in Terms of Resolution and in Relation to Object Size and Specimen Preparation II. Periodic Averaging A. Two-Dimensional Crystallization B. Periodic Averaging by Fourier Methods C. Averaging of Photosystem I Crystals D. High-Resolution EM E. Light-Harvesting Complex II III. Single Particle Averaging A. Method 1. Alignment by Correction Methods 2. Multivariate Statistical Analysis 3. Classification Step B. Analysis of Photosystem I Trimers from Cyanobacteria C. Analysis of Photosystem II Dinners from Cyanobacteria IV. Concluding Remarks Acknowledgements References

325 326 326 326 327 328 329 330 330 330 330 331 331 332 332 332 332 333 333 333 335 335 335

Summary Electron microscopy (EM) in combination with image analysis is a powerful technique to study protein structure at low- and high resolution. Since electron micrographs of biological objects are very noisy, substantial improvement of image quality can be obtained by averaging of individual projections. Averaging procedures can be divided into crystallographic and non-crystallographic methods and both will be described. Crystallographic averaging, based on two-dimensional crystals of rather small proteins, has the potential of solving a structure to atomic resolution just as the more common techniques of Xray diffraction and NMR. Single particle analysis is an alternative method for large proteins, viruses and all non-crystallizable proteins. It is a fast method to reveal the low-resolution structure with details in the range of maximally 10–15 Å. Results of EM on Light-harvesting complex II (LHC-II) and Photosystem I will be presented as examples for the crystallographic averaging. Trimeric Photosystem I complexes and dimeric Photosystem II complexes will be discussed as examples for the potential of single-particle averaging. *Correspondence: Fax: 31-50-3634800; E-mail: [email protected]

325 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 325–336. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

326

Egbert J. Boekema and Matthias Rögner

I. Principles

A.

Introduction

Direct information about the three-dimensional (3D) structure of a protein is essential for understanding its functional organization. At present electron microscopy (EM) is a widely applied technique for studying the structure of proteins and membranes; however, it is still less common than X-ray diffraction, where solving the 3D structure of proteins becomes almost routine, once suitable crystals have been obtained. On the other hand, X-ray diffraction has two disadvantages in comparison to EM. First, X-rays cannot be focussed and only diffraction patterns are obtained, whereas EM results in direct information in the form of images. Second, the interaction of X-rays with material is a factor of about 10,000 weaker than that of electrons. This makes EM a useful technique as images of single protein molecules or one-layer thick crystals can be obtained, whereas X-ray diffraction needs much thicker specimens. In this chapter on EM, the first sections will briefly introduce: the electron microscope with some instrumental aspects (I.B); specimen preparation (I.C); image analysis averaging techniques (I.D) and finally the possible resolution of EM with respect to object size and specimen preparation (I.E). Sections II.A and III.A will focus on image analysis of periodic and non-periodic objects including some examples. For more details concerning theory of the electron microscope, techniques for recording the signal and image analysis, we refer to the book by Hawkes and Valdrè (1990), which was written for the field of protein structure determination.

B.

The Electron Microscope

The resolution of a light microscope, which is about 2000 Å, is mainly limited by the wavelength of the light. Improving the resolution of a microscope is only possible by exploiting waves with a much shorter wavelength. According to the well known formula of de Broglie, accelerated particles such as electrons also have a wave character. At an acceleration voltage of 100,000

V the wavelength of the electron beam is 0.037 Å. This is certainly sufficient to enable microscopy at atomic resolution. Shortly after de Broglie had described the wave character of particles, it was discovered, and put into practice, that electrons can be focussed by electric and magnetic fields with axial symmetry. Based on these principles, Ernst Ruska and Max Knoll constructed the first simple electron microscope in 1931, a tube under vacuum with an electron source and several lenses. Improvements in the following years enhanced the resolution to 100 Å, already much better than the resolution of a light microscope. Since the early days the electron microscope has been gradually but substantially improved to an instrument which can now routinely achieve a resolution of about 2 Å. This resolution is limited by lens geometries and reflects compromises between several optical parameters, such as minimizing the spherical aberration, which is a kind of lens error. Overall, much further improvement in resolution cannot be expected, but 2 Å resolution is sufficient to solve a protein structure at the atomic level. To minimize the lens aberrations, the lenses in the electron microscope have a small opening: the holes in lens apertures are less than 0.1 mm in diameter. This results in a large “depth of field” and “depth of focus” at the object plane and image plane, respectively. As a consequence, EM gives two-dimensional (2D) projections in which the upper- and lower side of a thin object (up to a few 1000 Å) are seen superimposed with the same “sharpness”. As a result, a simple focussing on selected levels in an object, as is possible with a light microscope, cannot be done with the electron microscope. To get information about the 3D shape of a protein, the specimen must be tilted in the microscope and the various projections must be compared or combined into a 3D reconstruction. Imaging by EM of, for example, a thin metal foil or a gold cluster will easily provide projections with atomic resolution, but obtaining structures of proteins at high resolution is much harder work. Why? The contrast in the electron microscope is caused by scattering. Electrons are deflected at atomic nuclei through large angles and by other electrons through small angles. Since the

Electron microscopy scattering by nuclei is proportional to the atomic number, biological material, containing only the lighter elements, will give images with very low contrast. Besides, radiation damage by an electron beam easily destroys biological samples. Radiation damage cannot be avoided, but only minimized by cooling the specimen to liquid nitrogen or liquid helium temperature and by minimizing the electron dose. As a result, electron micrographs are noisy and objects are hardly visible. Therefore, image analysis techniques have been developed to improve the signal recorded in the noisy EM pictures.

C. Specimen Preparation Since modern electron microscopes have enough resolving power for structural studies of macromolecules, other factors than instrumental ones are of importance. The specimen preparation method is one of these factors and it strongly determines the ultimate resolution that can be achieved. In the negative staining technique the contrast is enhanced by embedding proteins or protein crystals in a heavy metal salt solution. Upon drying, the metal salts fill spacings and cavities around the molecules, but do not penetrate the protein interior. As a result, negatively stained specimens show protein envelopes with good contrast, but with a resolution that usually does not exceed 15 Å, due to the graininess of the contrasting agent. Because of its simplicity, negative staining has been widely applied. Single particles as well as crystals of photosynthetic membrane proteins, such as Photosystem I and II, have been successfully prepared by negative staining (Boekema, 1991; Boekema et al., 1994). As an alternative for negative staining, Unwin and Henderson (1975) pioneered the embedding of proteins in other media, such as glucose. It was demonstrated for crystals that images with a resolution of better than 10 Å could be achieved. This opened the way to high-resolution EM for biological macromolecules. A second important discovery in this context was the reduction of radiation damage, leading to a better signal. Specimen holders were developed that could be cooled down to liquid nitrogen-or liquid helium temperature. It was found for organic- and pro-

327 tein crystals that at the temperature of liquid nitrogen radiation damage is roughly reduced by a factor of 3–5 and by a factor of 10–20 at liquid helium temperature (Zemlin, 1992). Presently, the instrumental difficulties of practical cryo-EM have been mostly solved and the performance of the newest microscopes at low temperature is almost as good as at room temperature. Cryo-EM was further stimulated by the discovery of vitrification of protein solutions (Adrian et al., 1984). By rapidly cooling a protein solution, the formation of ice crystals can be avoided and proteins embedded in a thin layer of amorphous ice are obtained. Contrast is caused by the difference in density between amorphous ice (0.93 and protein and is rather low in comparison to negative staining. However, there are several advantages of cryo-EM of vitrified specimens: Specimen flattening and other drying artifacts are circumvented. Moreover, cryo-images better reflect the true density of a protein, because the contrast directly originates from scattering by the protein rather than from the surrounding stain (Fig. 1). Also, negative stain interaction with the protein is often quite complex. In thinner stain layers, the upper part of the protein could easily be less well embedded in the stain layer, as pointed out in Fig. 1. This means that the contribution of the upper- and lower half of a protein in the final recorded image do not have the same weighting. Other techniques are less important for high resolution structural work. Freeze-fracture techniques have been widely applied in research on photosynthesis. Cell or membranes are rapidly frozen, cleaved and replicated. The replicas give useful information about the overall size and distribution of the complexes embedded in these membranes (Staehelin, 1988). But the resolution is rather limited. Only particle diameters and the overall shape of membrane protein can be obtained. The main value of these techniques lies in the imaging of the in vivo situation of the membranes. It can reveal crystalline packing of photosystems and sometimes the multimeric state of large protein complexes. Evidence for the existence of a dimeric organization of PS II in vivo was obtained by freeze-fracturing (Mörschel and Schatz, 1987).

328

D. Averaging Methods In EM image analysis, improving the signal of an object recorded in a noisy electron micrograph is performed by averaging. By adding hundreds or, if possible, thousands of projections the signal improves substantially and trustworthy electron density maps are obtained. There are two general methods for averaging of 2D projections, depending on the object. One method is based on

Egbert J. Boekema and Matthias Rögner filtering images of periodic objects, which are usually 2D crystals. The other one deals with singleparticle projections. Periodic averaging takes advantage of the fact that in crystals, protein molecules are arranged in a regular packing. This means that neighborto-neighbor distances have a fixed value. In other words: the precise position of the molecules can be easily determined, even if the crystal is recorded with a low electron dose to prevent radiation damage and the molecules can be barely seen. As a result, higher resolution can be obtained. If 2D crystals with a diameter of at least several can be grown, EM can be performed under cryo-EM conditions at high resolution. For some small membrane proteins with a mass of 20–40 kDa, averaging over very large areas resulted in projection structures with a resolution better than 5 Å. For bacteriorhodopsin, the three-dimensional structure could be determined entirely from EM data by fitting the amino acid chain into the electron density map (Henderson et al., 1990). Another example is in the field of photosynthesis, where a second high resolution structure determination, that of the light-harvesting complex II (LHC-II) from pea was recently completed (Kühlbrandt et al., 1994). The crystallographic method, which is based on Fourier methods, is further described in section II, where results on LHC-II will be discussed in more detail. Projections of single particles can be averaged after they have been brought into equivalent positions by shifting them rotationally and translationally. This a-periodic averaging technique or single particle analysis is able to reveal the predominant projections of protein molecules (Frank et al., 1988). The fact that crystallization of the protein is not required is an advantage of this method. A disadvantage is that the maximal possible resolution by single-particle analysis is restricted to about 10–25 Å. This limit is set by the signal-to-noise ratio, which is related to the size of the object and has a relative low value for small objects. The resolution is also limited by the fact that the particles are not fixed in a definite position, as in a crystal. Small tilts from a common predominant position cause slight differences between similar-looking projections, resulting in an ensemble of projections that are all

Electron microscopy

slightly different. Averaging all of them would give a sum in which the finest details would be blurred out.

E. Possibilities of EM in Terms of Resolution and in Relation to Object Size and Specimen Preparation With the present state of the art in EM, as described in the previous sections, we can give an overview of the possibilities of EM in the field of proteins. Fig. 2 describes the potential of 5 types of EM approaches in the field of proteins. The thickness of the lines in Fig. 2 indicates the suitability and the attainable resolution in relation to the molecular mass of a protein. Approach 1. Single-particle averaging of negatively stained specimens is able to resolve the structure up to 15 Å in favorable cases. This resolution is sufficient for the localization of subunit positions in projections. Examples will be given for Photosystem I and II. Approach 2. If the negative staining method is replaced by cryo-EM of vitrified solutions, single particle averaging can be applied to objects with a mass of at least a few hundred kDa. From smaller proteins, especially those with a mass of 100 kDa or lower, the projections from single molecules as recorded by EM are too noisy for accurate averaging. Approach 2 works better than ap-

329

proach 1 for larger objects, for reasons mentioned before, such as removing the flattening upon drying. Another reason is the lower contrast of cryoEM, which becomes relatively disadvantageous for small molecules, as we will briefly explain. The contrast in biological material is largely caused by the scattering of the electron cloud of the C, N and O elements. This contrast is called phase contrast. The contrast originating from interaction with atomic nuclei, the scattering contrast, is relatively unimportant. Enhancement of phase contrast is possible by a stronger defocussing of the objective lens. But this coincides with a degradation and loss of fine details in the image. The smaller the object, the larger the defocussing needed to see the object at all. Approach 3. Periodic averaging of stained objects. The advantage of a periodic object is its regular packing. This enables an accurate determination of the positions of the molecules and thus an accurate averaging of projections. For crystals of about in size, a resolution of about 15–25 Å is usual. Small proteins (10– 50 kDa) are advantageous, because they allow averaging over a larger number of molecules per area. Approach 4. Cryo-EM of unstained, periodic objects. As for approach 2, the contrast comes largely from phase contrast. But crystals can be considered as a “big object” and can be recorded with small defocus values. Although the signal of one individual molecule is almost buried in the noise, it will still be interpretable, because it is averaged periodically. Small crystals of organic molecules and small proteins of a few in size can be analyzed to high resolution. For such small macromolecules, EM provides the same quality as X-ray diffraction, although the image analysis is not as straightforward. Approach 5. Based on perfect, large 2D crystals atomic resolution is possible for proteins up to at least 50 kDa. For larger proteins, the resolution will gradually decrease. For proteins with a mass between 300 and 600 kDa, like Photosystem I and II and ATP synthase, solving the structure at high-resolution by EM would be difficult. The preferential size of 2D crystals of such objects is at least Such large crystals are difficult to grow and have not yet been obtained.

330

Egbert J. Boekema and Matthias Rögner

II. Periodic Averaging

A. Two-Dimensional Crystallization As indicated in Fig. 2, small proteins should be crystallized into 2D crystals, to obtain the best structural information. The best 2D crystals produced in vitro have been grown from detergentsolubilized, purified material. Starting from highly purified protein, crystallization conditions can be controlled more easily and results are more reproducible (Kühlbrandt, 1992; Jap et al., 1992; Engel et al., 1992). Reconstitution of membrane proteins into lipid bilayers by detergent removal is currently the most universal method of 2D crystallization. A suspension of a lipiddetergent mixture is usually added to a detergentsolubilized protein preparation. Crystallization of the protein into sheets or vesicle crystals is then induced by removal of detergent by means of dialysis or absorption.

B.

Periodic Averaging by Fourier Methods

Usually, images of 2D crystals are recorded on electron micrographs. Fourier analysis has been proven to be very valuable in the processing of micrographs. The Fourier transform is a frequency decomposition in reciprocal space. A 2D Fourier transform calculated from an image gives a 2D diffraction pattern. If the image shows a good 2D crystal, its transform will show a pattern with many spots laying on a regular pattern, as in X-ray diffraction patterns. These spots represent the structure factor amplitudes and have an amplitude (peak height) and a phase. But the noise in the images will also show up in the diffraction pattern. An illustration of the Fourier techniques will be given in the next section; it shows an effective way to get rid of the noise.

C. Averaging of Photosystem I Crystals At medium resolution no complicated strategy is necessary since correcting for image aberrations is only necessary for high-resolution EM (see II.D). We will show by a simple example the basics of periodic averaging. Monomeric Photosystem I (PS I) has been crystallized into two-dimensional arrays from the cyanobacterium Synechococcus

elongatus (Böttcher et al., 1992) by removal of detergent (Fig. 3A). In a 2D Fourier transform of the PS I crystal (Fig. 3B), we see spots at regular distances. They tell us about the lattice parameters and about resolution. In this case, the lattice of the crystal is rectangular. Based on the Fourier transform of the image, a filtering is performed. Computationally, a mask is constructed that is superimposed on the Fourier transform (Fig. 3C). The mask has holes that neatly surround the peaks, which contain the crystalline information. The mask shields the space between the peaks, which represents noise in the crystal image. With a reverse Fourier-transformation (Fig. 3D) a real image is generated again, in which much of the noise is removed. This procedure is called Fourier-peak filtering and the comparison between Figs. 3E and F illus-

Electron microscopy trates the considerable improvement in signal-tonoise ratio. Further improvement in filtering is also possible. As crystals never have a perfect lattice, a small bending in the plane results in molecules being slightly displaced from their ideal position. Corrections can be made by “cutting” the crystal into pieces containing one or several molecular projections and shifting them into their ideal position. This is done by correlation methods, which have also been used in single particle averaging methods (Frank, 1982). An application of this method is the analysis of PS I crystals (Böttcher et al., 1992).

D. High-Resolution EM While a low-resolution structure by single particle averaging can be obtained within weeks, obtaining a high-resolution structure from EM data may take years. One of the limits of a structure determination at high resolution is merely the production of well-ordered, large crystals. Once this prerequisite is fulfilled, a resolution better than 10 Å is possible. However, recording the highest quality images and extensive processing are also time consuming. The signal recorded by EM suffers from aberrations, which increase in severity as the required resolution becomes higher (Henderson et al., 1986). Images providing structural information beyond a resolution of about 5 Å need extensive processing, which is carried out in Fourier space on the raw image amplitudes and phases. The phases are corrected for the effects of the contrast transfer function, beam tilt and phase origin (Henderson et al., 1986). A further treatment and discussion of these image distortions and their analysis and correction is beyond the scope of this contribution. For the interested reader we refer again to the book of Hawkes and Valdrè (1990), as a useful source of further information, and to the paper of Henderson et al. (1990). The microscope can be configured in two different ways: a) image mode and b) diffraction mode, i.e. recording the image formed in the back-focal plane of the objective lens. Electron diffraction (ED) patterns of crystals contain many spots, as in X-ray diffraction patterns and also represent the structure factors without the phase

331

information that is present in calculated Fourier transforms. ED patterns usually have much stronger spots than those from calculated transforms. The reason possibly lies in the specimen movement during image recording. Fourier transformation and diffraction in an optical system have the property that they are “translation invariant”. This means that movements during the recording of an ED pattern are less dramatic than during the recording of a real image, because they do not result in blurring the ED. Therefore spots in ED are much stronger and this makes recording of only ED for high resolution work tempting. But then a “phase problem”, as in Xray diffraction, is created and phases need to be generated. Isomorphous replacement, as used in X-ray diffraction, is not a useful method for phasing the ED data from 2D protein crystals because the scattering contrast of heavy atoms is low for electrons and the noise level in the patterns is relatively high. To compute a protein map at high-resolution by Fourier methods, images, are recorded as well. They are necessary to extract the phase from each of the spots. The phases (from the images) and amplitudes (from ED) are finally combined and corrected for some image errors, briefly mentioned in section I.D. This is called a “Fourier synthesis”. An example of the Fourier techniques for high-resolution structure determination will be given in the next section.

E. Light-Harvesting Complex II The light-harvesting chlorophyll a/b protein complex associated with photosystem II (LHC-II) harvests light energy and is capable of passing it to photosystem II. Detergent-solubilized, purified LHC-II forms large, highly ordered 2D crystals, which are ideal objects for recording high-resolution EM. Based on the averaging methods developed by Henderson et al. (1990), Kühlbrandt et al. (1994) were able to calculate a 3D structure at 3.4 Å resolution. Images were recorded with an electron microscope capable of resolving 2 Å object features at a specimen temperature of 4.2 K. Tilted and untilted images were collected for a 3D reconstruction. Since it is difficult to record images at tilt angles above 60°, there was a region in the Fourier space where amplitudes and phases could not be measured. This caused a reduction

332

Egbert J. Boekema and Matthias Rögner III. Single-Particle Averaging

A.

Method

Isolated proteins prepared on a carbon support film exhibit a full range of rotational and translational orientations in the plane of the support film. As a consequence, the projections of the proteins have a random orientation within this plane and computer averaging of such projections can only be achieved after an alignment procedure. Also, proteins often are attached in various ways to the support film (Fig. 1) and this results in a further variety in the obtained projections. To separate the various predominant projections, multivariate statistical analysis together with automated classification was introduced (Van Heel and Frank, 1981). The alignment procedure, multivariate statistical analysis and classification form the main steps in the single particle averaging procedure (Frank et al, 1988). These steps will be briefly discussed. 1. Alignment by Correlation Methods

in the resolution in the direction perpendicular to the membrane plane to between 4.6 and 4.7 Å. But due to the high quality of the phases it was possible to trace the polypeptide chain in the 3D electron density map. About 80% of the amino acids could be fitted, as well as the tetrapyrrole rings of 12 chlorophylls and two carotenoids (Fig. 4). LHC-II is the second protein solved by EM to atomic resolution. No doubt, the work on LHCII is a major step forward both for EM and for photosynthesis. To stress the similarities with Xray crystallography, the term electron crystallography has been introduced for high-resolution protein determination by EM.

The first step in comparing images of projections of biomolecules is to bring them into register in the plane: the alignment process. A reference image is taken and each image in the data set is compared with this reference; rotational and translational cross-correlation functions are computed to determine the best angular and translational shifts to bring the image into a position most similar to the reference image. The alignment procedure is an iterative process because sums of projections align the data set better than does a noisy projection of just one molecule. In practice, several references are used and results are combined to circumvent a possibly wrong choice of the first reference. 2. Multivariate Statistical Analysis Once a data set of typically hundreds of projections has been aligned, they can be compared in some numerical way. In particular, correspondence analysis, a special form of principal component analysis, is used to extract relevant feature information from the data set (Van Heel and Harauz, 1988). Each image of × pixels can be

Electron microscopy

333

represented as a point in an × dimensional “hyper”space, and the entire set of images forms a cloud in this space. Correspondence analysis determines a new, rotated-coordinate system in which the first axis represents the direction of the greatest inter-image variance, the second axis represents the direction of the largest remaining inter-image variance, and so on. The cloud of images can now be described with respect to this new coordinate system. By describing each image with respect to only the first few components, the images can be considered as points in a much smaller (than n × n) dimensional space. In this way a very large reduction in the amount of data to be analyzed is achieved. 3. Classification Step After the data compression by correspondence analysis, the grouping together of those images that are most similar is achieved by automatic classification schemes. Output of the classification are “classes” of groups of projections that are most similar. In the classification process, projections are shifted between the classes to optimize the variance between the classes and to minimize the variance of the members belonging to the classes. The number of classes to be chosen is somewhat arbitrary, but usually a number of 6– 12 are chosen. The differences between classes may represent real structural features (as will be illustrated by two examples) or are merely noiserelated when there is not more than one type of projection present in the data set.

B. Analysis of Photosystem I Trimers from Cyanobacteria A first practical example of single particle analysis concerns Photosystem I. In cyanobacteria, such as the thermophilic Synechococcus elongatus and the mesophilic Synechocystis PCC 6803, PS I is arranged as a trimeric complex with a diameter of about 200 Å in the membrane (Boekema et al., 1989; Kruip et al., 1993). From electron micrographs of PS I from Synechocystis PCC 6803 (Kruip et al., 1993), top view projections (Fig. 5) were extracted, aligned, treated by correspondence analysis and classified. The analysis resulted in the separation of the top view projec-

tions into two types, which are mirror-related (Fig. 6). These two types are generated because particles are asymmetric and can be attached to the carbon support film in two ways (upside-up and upside-down or “flip” and “flop”). It should be noted that it is not easy to judge by eye whether the PS I trimer projections (Fig. 5) belong to the flip or flop-type, but after a 3–fold averaging most of them (but not all) can be classified by eye.

C. Analysis of Photosystem II Dimers from Cyanobacteria Photosystem II is another membrane protein with a size large enough to enable single particle analysis. Dimeric particles were purified from the Synechococcus elongatus (Rögner et al., 1987; Dekker

334

et al., 1988). We visualized these particles in the presence of the detergent dodecyl maltoside by negative staining with uranyl acetate (Boekema et al., 1994). Averaged top views of dimeric PS II show that the dimers are built up from two monomers which are arranged in an anti-parallel way (Fig. 7A). In contrast to the example of PS I top view projections, there was only one predominant top view present in the data set. The dimer has dimensions of 120 × 155 Å in the membrane (corrected for detergent). The side views, however, show more variations. Original images clearly show protrusions in the side-view position (see Fig. 9 in Boekema et al., 1994). Often two protrusions can be seen on one dimer, but sometimes only one is visible. The protrusions

Egbert J. Boekema and Matthias Rögner

are thought to represent the 33 kDa oxygenevolving subunit resulting in a maximal height of 90 Å for the particles. Image analysis of side views gives another example of the usefulness of single particle analysis. By correspondence analysis and classification, a discrimination between different views was possible, which we interpret as overlap- and non-overlap views. In the nonoverlap views two protrusions can be seen (Fig. 7D, E), whereas the overlap view shows one protrusion (Fig. 7F). Note that the non-overlap view shows a longer projection than the overlap views, indicating a different position of the respective dimers on the carbon support.

Electron microscopy IV. Concluding Remarks In conclusion, EM techniques are available to study isolated (membrane) protein complexes from very small to very large size. In combination with techniques used in cell biology for studying whole cells or cell fragments, EM has the possibility and the potential to supply the field of photosynthesis with a detailed picture of the photosynthetic membrane and its interacting proteins. For structure determination, crystals are more useful than single particles. But not in all cases will it be possible to grow good crystals in a short term for such complicated structures as PS I, PS II and the synthase complex including their interacting donor-, acceptor- and regulatory molecules and proteins. In the meantime a combination of low-resolution EM reconstructions of these photosynthetic membrane complexes with atomic structures of their individual components, determined by EM, NMR or X-ray diffraction, will be useful for understanding the structure and function of these proteins. Acknowledgements We are grateful to Dr. W. Keegstra for his help with computer image analysis, Dr. G. Perkins for discussion and Mr. K. Gilissen for photography. Work has been supported by the Netherlands Foundation for Chemical Research (SON) with financial aid from the Netherlands Organization for Scientific Research (NWO), by a grant from the European Union BIO2CT-930078 (EJB), the Deutsche Forschungsgemeinschaft (MR) and a grant from NEDO/RITE, Japan (MR). References Adrian M, Dubochet J, Lepault J, McDowall AW (1984) Cryoelectron microscopy of viruses. Nature 308: 32–36 Boekema EJ (1991) Negative staining of integral membrane proteins. Micron and Microsc Acta 22: 361–369 Boekema EJ, Dekker JP, Rögner M, Witt I, Witt HT and van Heel, MG (1989) Refined analysis of the trimeric structure of the isolated Photosystem I complex from the thermophilic cyanobacterium Synechococcus sp. Biochim Biophys Acta 974: 81–87 Boekema EJ, Boonstra AF, Dekker JP and Rögner (1994) Electron microscopic structural analysis of Photosystem I, Photosystem II, and the cytochrome b6/f complex from

335 green plants and cyanobacteria. J Bioenerg Biomembr 26: 17–29 Böttcher B, Gräber P and Boekema EJ (1992) The structure of Photosystem I from the thermophilic cyanobacterium Synechococcus sp. determined by electron microscopy of two-dimensional crystals. Biochim Biophys Acta 1100: 125–136 Dekker JP, Boekema EJ, Witt HT Rögner M (1988) Refined purification and further characterization of oxygen-evolving and Tris-treated Photosystem II particles from the thermophilic cyanobacterium Synechococcus sp. Biochim Biophys Acta 936: 307–318 Engel A, Hoenger A, Hefti A, Henn C, Ford RC, Kistler J and Zulauf M (1992) Assembly of 2–D membrane protein crystals: dynamics, crystal order, and fidelity of structure analysis by electron microscopy. J Struct Biol 109: 219–234 Frank J (1982) New methods for averaging non-periodic objects and distorted crystals in biologic electric microscopy. Optik 63: 67–89 Frank J, Radermacher M, Wagenknecht T and Verschoor A (1988) Studying ribosome structure by electron microscopy and computer-image processing. Methods in Enzymology 164: 3–35 Hawkes PW and Valdrè U (1990) Biophysical Electron Microscopy. Basic concepts and modern techniques. Academic Press, London Henderson R, Baldwin JM, Downing KH, Lepault J and Zemlin F (1986) Structure of purple membrane from Halobacterium halobium: recording, measurement and evaluation of electron micrographs at 3.5 Å resolution. Ultramicroscopy 19: 147–178 Henderson R, Baldwin JM, Ceska TA, Zemlin F, Beckmann E and Downing KH (1990) Model for the structure of bacteriorhodopsin based on high-resolution electron cryomicroscopy. J Mol Biol 213: 899–920. Jap BK, Zulauf M, Scheybani T, Hefti A, Baumeister W, Aebi U and Engel A (1992) 2D crystallization: from art to science. Ultramicroscopy 46: 45–84 Kruip J, Boekema EJ, Bald D, Boonstra AF and Rögner M (1993) Isolation and structural characterization of monomeric and trimeric Photosystem I complexes and from the cyanobacterium Synechocystis PCC 6803. J Biol Chem 268: 23353–23360 Kühlbrandt W (1992) Two-dimensional crystallization of membrane proteins. Quaterly Rev of Biophys 25: 1–49 Kühlbrandt W, Wang DN and Fujiyoshi Y (1994) Atomic model of plant light-harvesting complex by electron crystallography. Nature 367: 614–621 Mörschel E and Schatz GH (1987) Correlation of photosystem-II complexes with exoplasmic freeze-fracture particles of thylakoids of the cyanobacterium Synechococcus sp. Planta 172: 145–154 Rögner M, Dekker JP, Boekema EJ and Witt HT (1987) Size, shape and mass of the oxygen-evolving photosystem II complex from the thermophilic cyanobacterium Synechococcus sp. FEBS Lett 219: 207–211 Staehelin LA (1988) Chloroplast structure and supramolecular organization of photosynthetic membranes. In: Staehelin LA and Arntzen CJ (eds.) Photosynthesis III, pp. 1–84, Springer-Verlag, Berlin

336 Unwin PNT and Henderson R (1975) Molecular structure determination by electron microscopy of unstained crystalline specimens. J Mol Biol 94: 425–440 Van Heel M and Frank J (1981) Use of multivariate statistics in analysing the images of biological macromolecules. Ultramicroscopy 6: 187–194

Egbert J. Boekema and Matthias Rögner Van Heel H and Harauz G (1988) Biological macromolecules explored by pattern recognition. Scanning Microscopy, Supplement 2: 295–301 Zemlin F (1992) Desired features of a cryoelectron microscope for the electron crystallography of biological material. Ultramicroscopy 46: 25–32

Chapter 21 X-Ray Absorption Spectroscopy: Determination of Transition Metal Site Structures in Photosynthesis Vittal K. Yachandra* and Melvin P. Klein Structural Biology Division, Lawrence Berkeley National Laboratory, University of California, Berkeley, CA 94720, USA

Summary I. Introduction II. X-ray Absorption Spectroscopy (XAS) A. X-ray Absorption Near Edge Structure (XANES) B. Extended X-ray Absorption Fine Structure (EXAFS) 1. Basic Theory of EXAFS a. Energy of the Photoelectron b. Definition of EXAFS c. The EXAFS Equation C. Experimental Methodology D. Advantages and Limitations of XAS 1. Advantages 2. Limitations III. Applications of XANES and EXAFS in Photosynthesis A. Fe–S Proteins 1. Soluble Plant Ferredoxin and in Photosystem I 2. Fe–SAcceptors 3. Rieske Fe–S Clusters B. Manganese Oxygen Evolving Complex in Photosystem II IV. Future Directions Acknowledgements References

337 338 338 338 338 339 339 340 341 343 343 343 344 345 345 345 346 348 348 350 351 352

Summary This chapter gives a brief description of the theory and experimental methodology involved in X-ray absorption Spectroscopy, both X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS). The advantages and limitations of the methods are discussed. Several examples of applications of X-ray absorption Spectroscopy in the field of photosynthesis are presented. Abbreviations: EPR – Electron Paramagnetic Resonance; EXAFS – Extended X-ray Absorption Fine Structure; cyt – cytochrome; OEC – Oxygen Evolving Complex; PS I – Photosystem I; PS II – Photosystem II; XANES – X-ray Absorption Near Edge Structure; XAS – X-ray Absorption Spectroscopy *Correspondence: Fax: 1-510-4866059; E-mail: [email protected]

337 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 337–354. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

338 I. Introduction X-ray absorption spectroscopy (XAS) is the measurement of transitions from core electronic states of the element to the excited electronic states or continuum states, which is known as X-ray absorption near edge structure (XANES), and the study of the fine structure in the absorption cross section at energies greater than the threshold for electron release, which is known as extended X-ray absorption fine structure (EXAFS). These two methods give complementary structural information, the edge-spectra reporting oxidation state and symmetry, and the EXAFS reporting numbers, types and distances to ligands and neighboring atoms from the absorbing atom. The major application of XAS in biology and in photosynthesis has been in the study of the structure of the metal sites in metallo-proteins and metallo-enzymes (reviewed in Yachandra, 1995; Cramer, 1988; Scott, 1984; Powers, 1982). This was primarily due to the availability of Xray beam lines at synchrotron sources optimized in the X-ray energy region of metal K-edges (from Ca to Mo) and also the ease of making measurements at such X-ray energies, without interference from the protein matrix, water or air. Secondly, advances being made in the field of bioinorganic chemistry were raising important questions of correlation between structure and function of the metal sites in metallo-proteins which were amenable to XAS studies. II. X-Ray Absorption Spectroscopy (XAS)

A. X-Ray Absorption Near Edge Structure (XANES) X-ray absorption spectra of any material, whether it is atomic or molecular in nature, is characterized by sharp increases in absorption at specific X-ray photon energies, which are characteristic of the absorbing element. These sudden increases in absorption are called absorption edges, and correspond to the energy required to eject a core electron into the continuum thus producing a photoelectron. The absorption discontinuity is known as the K-edge, when the photoelectron originates from a 1s core level, and an L-edge

V.K. Yachandra and M.P. Klein when the ionization is from a 2s or 2p level. Fig. 1 shows a typical energy level diagram. The physics of the processes involved as it pertains to the K-edge XANES of transition metals has not been satisfactorily explained, despite advances in theory (Stöhr, 1992; Durham, 1988; Bianconi, 1988). There are two main approaches to XANES analysis. The first method developed by Shulman et al. (1976) in their work with the highly ionic metal fluorides involved transitions to atomic states localized on the metal, and the spectra were rationalized on the basis of ligand field perturbations of the localized atomic orbitals. For complexes that contain significant metal-ligand interactions, such as is commonly found in biological materials, this simple approach is not valid; and quantitative treatment probably requires a knowledge of the low lying unoccupied molecular orbitals of the complex. The effect of core holes on the outer lying states is also important in the consideration of the L-edges. The second method has used multiple scattering of the photoelectron from the neighboring atoms to explain the structure seen on the edges (Kutzler et al., 1980; Pendry 1983; Rehr et al., 1992). The most fruitful use of K-edge spectra in biology has been in determining the oxidation state of the metal in the active site of metallo-proteins. The oxidation state and the coordination sphere of the metal largely determines the position and shape of the X-ray absorption K-edge. In complexes with similar coordination, as the oxidation state of the metal increases the K-edge shifts to higher energy, because as the effective positive charge increases, it becomes correspondingly more difficult to remove an electron from the metal. However, it is not only the oxidation state of the metal which controls the effective positive charge of the metal, but also the nature and number of the ligands of the metal (Kirby et al., 1981, Conradson et al., 1985).

B. Extended X-Ray Absorption Fine Structure (EXAFS) At higher energies above the edge, the absorption decreases in accordance with the theory of the photoelectric effect, where the excess energy is transferred to the photoelectron as kinetic en-

X-ray absorption spectroscopy

ergy. However, there is a crucial difference in this falloff of absorption between atomic or isolated atoms, and molecular or condensed systems. For the latter, superimposed on the gradual decrease in absorption is a periodic modulation in absorption, starting immediately past an absorption edge and extending to about 1 keV above the edge. These oscillations in the X-ray absorption spectra are known as Extended X-ray Absorption Fine Structure or EXAFS. A typical X-ray absorption spectrum at the Mn K-edge is shown in Fig. 2. The EXAFS region starts about 50 eV above the edge, and the region before that is the XANES region. EXAFS oscillations occur only in molecular or condensed systems. For all systems there is no requirement for long range order as exists in crystalline materials. EXAFS contains information about the local environment around the absorbing atom (reviewed in Eisenberger and Kincaid, 1978). In fact, the EXAFS oscillations result from the interference between the outgoing photoelectron wave and components of this wave backscattered from neighboring atoms in the molecule. The importance of the EXAFS technique depends directly on the fact that the EXAFS modulations contain information about the distance between the absorbing and backscattering atoms within a distance of about 5–6 Å, as well as the identity and number of the backscattering atoms.

339

Essentially, EXAFS analysis is used to determine the radial distribution of atoms around a particular absorbing atom, thus providing a probe for the local structure in the vicinity of the absorbing atom. 1.

Basic Theory of EXAFS

a.

Energy of the Photoelectron

The EXAFS modulations shown in Fig. 2 are a direct consequence of the wave-nature of the photoelectron whose wavelength is given by the de Broglie relation

where h is Planck’s constant, the electron mass, and v is the velocity of the photoelectron, which is the velocity imparted to the photoelectron by the energy of the absorbed X-ray photon which is in excess of the binding or threshold energy for the electron. The kinetic energy of the photoelectron is given by the following relation:

where E is the X-ray photon energy, and is the ionization or threshold energy for the electron. The EXAFS modulations are better expressed

V.K. Yachandra and M.P. Klein

340

as a function of the photoelectron wave-vector k, which is related to the de Broglie wavelength described above as follows:

which now can be expressed by the substitution of Eqs. (1) and (2) as follows:

or

where E and are expressed in electron volts (eV) and k has the units of inverse angstroms

b.

Definition of EXAFS

The general definition for the EXAFS phenomenon which is the oscillatory portion of the absorption coefficient, is the difference between the observed absorption coefficient and the free atom absorption coefficient normalized by the free atom contribution:

The absorption coefficient is proportional to the square of the electric dipole transition moment which is given quantum mechanically by a matrix element between the initial and final states of the photoelectron: In the case of K-edge absorption the initial state wave function is that of the 1s state in the

X-ray absorption spectroscopy

341

atomic core and the final state wave function is that of the photoelectron. In an isolated atom the final state wave function is the wave function of the outgoing photoelectron In a molecular or condensed system the final state wave function of the photoelectron consists of both the outgoing wave and the backscattered wave from the neighboring j atoms and the total final wave function is given by the sum of the outgoing and the scattered wave functions from each of the j backscattering atoms. The transition dipole moment can now be expressed as follows:

and it is the interference between and that lead to the EXAFS modulations, and substitution in Eq. (6) gives,

The electric-dipole moment is non-zero only in the region where the initial state is nonzero, which is near the center of the absorbing atom. So one needs to know only how the surrounding atoms perturb the outgoing wave and what effect this has at the center of the absorbing atom. One can envision this process more clearly by the help of a schematic of the outgoing and backscattered waves as shown in Fig. 3. As the energy of the photoelectron changes so does the wavelength, of the photoelectron. At a particular energy the outgoing and the backscattered waves are in phase and constructively interfere, thus increasing the probability of Xray absorption or in other words increase the absorption coefficient. At a different energy the outgoing and backscattered wave are out-of phase and destructively interfere, decreasing the absorption coefficient. This modulation of the absorption coefficient by the backscattered wave from neighboring atoms is essentially the basic phenomenon of EXAFS.

c.

The EXAFS Equation

A quantitative description of the EXAFS modulation, depends on an evaluation of the final state wave function, Many excellent

sources exist for a rigorous derivation of the EXAFS equation (Sayers et al., 1971; reviewed in Stern, 1988). For our purposes can be expressed as follows:

where is the number of equivalent backscattering atoms j at a distance from the absorbing atom, is the backscattering amplitude which is a function of the atomic number of the backscattering element j, and includes the phase shift from the central atom absorber as well as the backscattering element j. The phase shift occurs due to the presence of atomic potentials that the photoelectron experiences as it traverses the potential of the absorber atom, the potential

342

of the backscattering atom, and then back through the potential of the absorber atom. There is inherent static disorder due to a distribution of distances and dynamic disorder due to thermal vibrations of the absorbing and scattering atoms. Eq. (10) is modified to include this disorder term or the Debye–Waller factor where is the root-mean-square deviation to give the following equation

The loss of photoelectrons to inelastic scattering processes can be accounted for by including a term, where is the mean free path of the photoelectron. The EXAFS contribution from each backscattering atom j is a damped sine wave in k-space, with an amplitude, and a phase, which are both dependent on k. EXAFS spectra contain a significant amount of information about atomic surroundings of the central absorbing atom in the system being studied and have been used in numerous applications (see for example, Bertagnolli and Ertel, 1994). The contribution and importance of each of the terms in the EXAFS equation to the EXAFS spectrum, and the analysis procedures used to extract the information contained in the EXAFS equation are described in detail by Teo (1986), and Sayers and Bunker (1988). From the phase of each sine wave the absorber-backscatterer distance can be determined if the phase shift is known. The phase shift is obtained either from theoretical calculations (Rehr et al., 1991; McKale et al., 1988) or empirically from compounds characterized by crystallography with the specific absorber-backscatterer pair of atoms. The phase shift, depends on both the absorber and scatterer atoms. Because one knows the absorbing atom in an EXAFS experiment, an estimation of the phase shift can be used in identifying the scattering atom. The amplitude function contains the Debye– Waller factor and the number of backscatterers at These two parameters are highly correlated and make the determination of difficult at best. The backscattering amplitude function,

V.K. Yachandra and M.P. Klein depends on the atomic number of the scattering atom and, in principle, can be used to identify the scattering atoms. In practice, however, the phase shift and backscattering amplitude function, both of which are dependent on the identity of the backscattering atom, can be used only to identify scattering atoms that are well separated by atomic number. The sinusoidal nature of the backscattering contributions makes Fourier analysis particularly appropriate for analyzing EXAFS spectra. A Fourier transform of the EXAFS data in k-space can be employed to separate the different scattering distances into unique peaks in the conjugate R-space. The Fourier transform amplitude peaks at the characteristic distances where is directly related to the phase shift in Eq. (11), and it is a powerful visual tool in providing a simple physical picture of the local structure of the metal site. The Fourier transform provides a spectrum which is similar to a radial distribution function with peaks at R', which are shifted from R by but which correspond to the shells of backscatterers around the central absorbing atom. Quantitative information is obtained from curve fitting to however, fits are rarely performed on obtained from raw data, especially for biological systems because of the signalto-noise limitations. It is desirable to separate the EXAFS contributions into their component shells and to filter out the high frequency noise. The technique employed to achieve this is called Fourier filtering and consists of Fourier transforming the data into R-space, next selecting the Fourier peak of interest by applying an appropriate window function, and then back-transforming into k-space. Thus one can isolate the contribution from one shell of backscatterers which can then be fit to the EXAFS equation to obtain the structural parameters for just that shell. This isolation technique works well if the backscattering atoms are well separated from each other in radial distance from the absorbing atom (~ 1 Å). In its most general form curve fitting consists of (1) proposing a chemically feasible model based on information from other studies and by inspection of the properties of the Fourier transform, and then simulating the for such a model using the EXAFS equation; (2) comparing

X-ray absorption spectroscopy

343

the calculated EXAFS with that from experiment and (3) varying the parameters N, and the Debye–Waller parameter, to minimize the difference between the calculated and observed spectrum using a non-linear least squares minimization procedure. This protocol is followed for each of the Fourier filtered shells individually, hence simplifying the fitting procedure. However, that does not mean that each Fourier peak corresponds to only one kind of atom at one distance. In fact such a case would be an exception in most biological systems. In most cases, curve fitting is a trial and error enterprise consisting of several iterations in the choice of the backscattering atoms and distances and in the number of sub-shells that constitute each Fourier peak. Physical reasonableness and chemical feasibility of the fit parameters always help direct the search for the right fit. There are four variables for each shell of atoms, N, and hence, including another shell doubles the number of variables and most often leads to a better fit. However, it is important to consider the statistical criteria for including another shell, by the degree of improvement in the goodnessof-fit parameter. The statistical rule-of-thumb for the number of independently variable parameters is given by where is the width of the window used for Fourier filtering, and is the length of the data set, For a data set that extends from 3.5–12 and a Fourier window width of 1.5 Å, the maximum number of parameters that can be varied in a multi-shell fit is 8. Data analysis is usually performed from about 3.5 because of problems associated with multiple scattering effects below that energy.

or fluorescent X-rays. Tunable monochromatic X-rays are usually obtained by the use of Bragg diffraction of X-rays by crystals. A detailed account of the design and use of synchrotron radiation for an XAS experiment is presented by Heald (1988a, b). The sensitivity of the technique is extended several orders of magnitude by fluorescence detection of the absorption spectrum (Jaklevic et al., 1977). This enhanced sensitivity is essential to the measurement of X-ray absorption spectra of biological systems. In biological systems one is usually studying a dilute metal embedded in a matrix of protein, lipids and/or nucleic acids, which all scatter X-rays. So ideally one needs to discriminate the fluorescence photons (mostly, 2p to 1s for K-edge measurements) of the element of interest from the large scattering observed from the matrix. At present the detectors best suited for biological spectroscopy are the Si or Ge solid state detector which is placed at right angles to the incident beam (Cramer at al., 1988). In addition a third X-ray detector is used downstream from which is used to measure simultaneously the spectrum of a reference compound, which provides an energy reference. The essential components described above are shown in Fig. 4. The source used for most biological XAS experiments is the synchrotron radiation produced at electron storage rings. The intense flux generated by this means is significantly higher than that from any other X-ray source, and this technology makes X-ray measurements of dilute systems, which is typical of most biological materials, feasible.

C. Experimental Methodology

D. Advantages and Limitations of XAS

The X-ray absorption measurements are analogous to optical absorption and fluorescence excitation measurements. The spectrometer consists of a source of X-rays, a monochromator, an Xray detector which monitors the incident flux, a sample compartment, and a second detector which monitors the flux of X-rays transmitted through the sample, The most common form of collecting X-ray spectra consists of scanning through a range of wavelengths and measuring the flux of the incident and the flux of transmitted

1. Advantages 1) XAS is element specific, so one can focus on one element without interference from other elements present in the sample. In a system which has more than one metal, like the photosynthetic apparatus, it is possible to study selectively the structural environment of each metal atom. The specificity of the technique also makes it suitable for studying biological systems which contain several polypeptides, lipid molecules, pigment mol-

344

ecules which is typical of the photosynthetic systems (bacterial reaction center, PS I and PS II). The specificity and the fact that it is always possible to obtain an X-ray spectrum of an element, provided it is present in sufficient concentrations, also means that one ‘sees’ all of the metal of interest that is present in the sample. This makes it imperative that one is sure of the biochemical homogeneity of the sample and, if there is more than one site for the same metal, to resolve the structural parameters of the different sites. 2) Another important advantage of XAS is that the metal of interest is never ‘silent’ with respect to X-ray absorption spectra. The system could be ‘silent’ with respect to EPR, optical or other spectroscopic methods, but one can always probe the metal site structure by XAS. 3) XAS is not limited by the state of the sample, because it is sensitive only to the local metal site structure. The sample can be prepared as a

V.K. Yachandra and M.P. Klein

powder, a solution or, as is done most often, as a frozen solution for biological samples. The advantages of this are many-fold. It is not necessary to obtain single crystals of the material to examine the local structure of the metal. Obtaining single crystals can often be difficult and some samples do not crystallize. The more important aspect is that one can either trap intermediates in the enzymatic cycle or modify the site by the addition of inhibitors or substrate or generate other chemical modifications. Such samples can be made as frozen solutions, avoiding the problems of trying to obtain single crystals. 2. Limitations 1) Damage to biological samples by X-rays is cause for serious concern for XAS experiments. It is absolutely essential that the samples are checked for their biochemical integrity before and

X-ray absorption spectroscopy

after exposure to X-rays by an independent method, whenever possible. The use of a liquid He flow cryostat, where the samples are at ambient pressure in a He atmosphere has greatly reduced the risk of sample damage by X-rays. 2) It is also important to realize the intrinsic limitations of EXAFS, beyond those of a purely experimental nature (Lee et al., 1981). A frequent problem is the inability to distinguish between scattering atoms with little difference in Z, the atomic number (C, N, O or S, Cl, or Mn, Fe). Care must also be exercised when deciding between atoms that are well separated in Z, because frequently, it is possible to obtain equally good fits using backscattering atoms which are very different in Z (for example, Mn, or Cl) but which are at different distances from the absorbing atom. This is more acute when dealing with Fourier peaks at greater distances. In bridged multinuclear centers, it is not always possible to unequivocally assign the Fourier peaks at > 3 Å (Scott and Eidsness, 1988), The peaks at > 3 Å could arise from scattering from second or third shell scattering from ligands like the imidazole ring of histidine or the pyrrole ring of hemes, or from backscattering from a bridged metal atom. 3) Distances are usually the most reliably determined structural parameters from EXAFS. But the range of data that can be collected, oftentimes due to practical reasons like the presence of the K-edge of another metal, limits the resolution of distance determinations to between 0.1 to 0.2 Å. 4) Determination of coordination numbers or number of backscatterers is fraught with difficulties. The Debye-Waller factor is strongly correlated with the coordination number and one must have recourse to other information, like comparison to inorganic model complexes, to narrow the range that is possible from curve fitting analysis alone. The most important point in the analysis is to differentiate between fit parameters which are required and others which are merely consistent with the data. The EXAFS method is most useful when delineating all the structural alternatives based on required fit parameters, or addressing the question of subtle structural changes in systems well characterized by other techniques like X-ray crystallography.

345 III. Applications of XANES and EXAFS in Photosynthesis

In the two decades since the XAS technique has become practical, it has become a standard method for probing the metal site structure in metallo-proteins. The technique has been applied to virtually every metallo-protein that has been isolated containing Fe, Mo, V, Cu, Co, Mn, Zn, Ni, Ca and other metals. In this section applications of XAS to Fe-S centers and the Mn oxygen-evolving complex in photosynthesis are described. Other components of the photosynthetic apparatus, plastocyanin (reviewed in Blackburn, 1990), the Fe-quinone acceptor complex in bacterial reaction centers (Bunker et al., 1982; Eisenberger et al., 1982), and the complex of ATPase (Carmelli et al., 1986) have also been studied using XAS.

A. Fe–S Proteins 1. Soluble Plant Ferredoxin Among the first metallo-proteins studied by XAS are the class of non-heme iron sulfur proteins which are found in most redox-mediated pathways in biology. Initially, three classes of Fe–S structures containing 1Fe, 2Fe and 4Fe metal sites were recognized. This list has expanded to include 3Fe sites and higher nuclearity structures containing 7 and 8 Fe atoms in the M and P clusters of nitrogenase. The most common structural units are in 1Fe containing Fe– S protein like rubredoxin, in 2Fe– 2S proteins like soluble plant ferredoxins, and in 4Fe–4S proteins like ‘high potential iron proteins’ (HIPIP) and other ferredoxins generally referred to as bacterial ferredoxins. The Fe EXAFS of these iron-sulfur proteins is dominated by backscattering from sulfur and iron atoms in the active site. In rubredoxin one Fe atom is ligated to four thiolate derived S atoms in near tetrahedral symmetry and it is a good example of a single-backscattering environment. The EXAFS of rubredoxin (see Fig. 5) exhibits a single wave and the Fourier transform shows one distinct peak, indicative of a single-shell system with one type of Fe–S distance of 2.26 Å (Teo and Shulman, 1982).

346

V.K. Yachandra and M.P. Klein

In contrast to the EXAFS from rubredoxin, the EXAFS of the 2Fe–2S soluble plant ferredoxin and the 4Fe–4S bacterial ferredoxin exhibits (see Fig. 5) a beat pattern indicating the presence of at least two sine waves. The respective Fourier transforms show two peaks. The first peak is due to backscattering from S ligand atoms at ~ 2.23–2.25 Å, and the second peak is due to backscattering from the Fe atoms at 2.73 Å. These distances compare well with distances derived from X-ray crystallography of the 2Fe–2S proteins from Spirulina platensis (Tsukihara et al., 1978), Anabaena 7120 (Rypniewski et al., 1991), 4Fe–4S proteins as well as synthetic analogs (reviewed in Spiro, 1982; Teo and Shulman, 1982). 2. Fe–S Acceptors Photosystem I

and

in

The electron transfer reactions from the primary donor of PS I to the substrates are initially

mediated by a number of PS I bound electron acceptors, labeled and which are Fe– S clusters with redox potentials of –705, –590, and –530 mV, respectively. XANES and EXAFS studies at the Fe K-edge have been used to study these Fe-S clusters. In the Fe K-edge XANES study of PS I the transition to bound 3d states was used to derive information about the symmetry of the Fe–S acceptor complexes. The 1s to 3d pre-edge transition is electric-dipole forbidden, but it is often observed in the edge spectra and is commonly attributed to d-p mixing. In centrosymmetric complexes d–p mixing is symmetry unallowed and the 1s to 3d transition is weak. In non-centrosymmetric complexes d–p mixing is allowed and the 1s to 3d transition is more intense. The intensity of the 1s to 3d pre-edge transition can be used as a marker to differentiate between octahedral and tetrahedral complexes (Shulman et al., 1976; Roe et al., 1984). Fig. 6 shows the Fe K-edge spectrum of Fe–S

X-ray absorption spectroscopy

centers and The centers are compared to a 4Fe–4S model compound and to a centrosymmetric hexacoordinate Fe complex. There is a decrease in the intensity of the 1s to 3d transition, and there is a change in the shape of the spectrum between tetrahedral Fe–S centers and a centrosymmetric hexacoordinate system. The intensity of the pre-edge 1s to 3d transition shows that the Fe–S acceptors and are in tetrahedral Fe–S complexes (McDermott et al., 1988a, 1989). Fe K-edge EXAFS studies of PS I preparations from spinach and the thermophilic cyanobacterium Synechococcus sp. showed that the spectra were similar to those from Fe–S clusters, with a Fourier peak corresponding to S backscattering and another peak corresponding to backscattering from Fe. The results from PS I preparations containing and showed that the data could be simulated with either three 4Fe–4S clusters or two 4Fe–4S and one 2Fe–2S cluster. This was due to the complexity of dealing with PS

347

I preparations which contained about 11–14 Fe atoms (McDermott et al., 1988a). A subsequent EXAFS study using PS I preparations which contained only clearly showed that it was also a 4Fe–4S cluster (McDermott et al., 1989). The recently determined crystal structure of PS I has confirmed that all three acceptors are 4Fe–4S clusters (Krauss et al., 1993). 3. Rieske Fe–S Clusters

A Rieske Fe–S cluster is part of the cyt complex in higher plants which mediates electron transfer between PS II and PS I. A Rieske Fe–S center is also present in the cyt complex of photosynthetic bacteria which is involved in cyclic electron transport. The Rieske Fe–S centers are characterized by an EPR signal and redox properties (150–350 mV) which are very different from the other more ubiquitous 2Fe–2S and 4Fe–4S proteins. Analogous Rieske centers are part of

348 the mitochondrial electron transport scheme. Although there are no studies from plant sources, EXAFS studies of Rieske Fe–S protein from the cyt complex from mitochondria (Powers et al., 1989) and phthalate dioxygenase from Pseudomonas cepacia (Tsang et al., 1989) have been reported. Curve fitting results from both studies showed that better fits were obtained by including a mixed S and N coordination to Fe, but the quantitation of the number of nitrogens per Fe was not unequivocal. These studies are in accord with the ENDOR (Gurbiel et al., 1991) and ESEEM (Britt et al., 1991) results which showed there may be two terminal histidine ligands per 2Fe–2S cluster. The EXAFS studies found that: the Fe–S binding and terminal distances and the Fe–Fe distance were similar to the regular 2Fe–2S and 4Fe–4S clusters providing evidence that the unique properties of the Rieske centers are due to the N ligation or due to factors other than differences in Fe–S and Fe–Fe distances.

B. Manganese Oxygen Evolving Complex in Photosystem II Most of the oxygen in the atmosphere which supports life on earth is generated by plants by the photo-induced oxidation of water to dioxygen. The reaction shown in Eq. (12) is catalyzed by a tetranuclear Mn complex, which sequentially stores four oxidizing equivalents that are used to oxidize two molecules of water to molecular oxygen. The Mn complex is part of a multiprotein assembly called Photosystem II, which contains the reaction center involved in photosynthetic charge separation and an antenna complex of chlorophyll molecules. The complex also contains cyt and a Fe-quinone electron acceptor complex (reviewed in Debus 1992; Rutherford et al., 1992). Owing to the complexity of the system and the presence of so many pigment and other components, study of the Mn complex by optical and other spectroscopic methods can be difficult. EXAFS is ideally suited for the study of the structure of the Mn complex because the specificity of the technique allows us to look at the Mn without interference from the pigment molecules, or the

V.K. Yachandra and M.P. Klein protein and membrane matrix, or other metals like Ca, Mg, Cu and Fe which are also present in active preparations. An active oxygen-evolving complex has not yet been crystallized, but EXAFS does not need single crystals; the structural studies can be performed on frozen solutions. Also, several of the intermediate states mentioned above have been stabilized as frozen solutions and studied by EXAFS. We and others have studied the structure of Mn in the OEC using XAS. An earlier series of papers from our group reported Mn–Mn interactions at 2.7 Å and Mn–O interactions at 1.75 Å, and an additional highly disordered shell of light atom (O, N) scatterers at ~ 2 Å. Similar results were found for both PS II-enriched membrane preparations from spinach chloroplasts and for detergent-solubilized OEC preparations from the thermophilic cyanobacterium Synechococcus sp. (McDermott et al., 1988b). From these results on samples poised in both the and states, it was predicted that the OEC contains binuclear dibridged Mn units whose structures remain largely unchanged upon the advance to (reviewed in Sauer et al., 1992; Klein et al., 1993). More recent EXAFS studies of Mn in the OEC, at substantially lower sample temperatures and with improved signal to noise ratio, have provided evidence for scatterers at > 3 Å in addition to the interaction at 2.7 Å. These experiments include various combinations of oxygenevolving preparations and EXAFS analysis techniques. George et al. (1989) report Mn scatterers at distances of 2.7 and 3.3 Å in angle-dependent EXAFS studies of whole oriented spinach chloroplasts. It was found that the scatterers at these distances exhibited significant dichroism. PennerHahn et al. (1990) also reported the 2.7 Å shell and at least one and possibly two shells of scatterers at > 3 Å in EXAFS experiments using core preparations from spinach. Both these studies reported difficulties in detecting the 1.8 Å shell; this was probably due to Mn(II) contamination in their experiments. MacLachlan et al. (1992) reproduced the 1.8 Å shell along with the 2.7 Å scatterers, but differed from the above experiments by predicting a shell of Ca at a significantly longer distance of 3.7 Å. These results all have different implications for a structural model of the OEC.

X-ray absorption spectroscopy

Fig. 7 shows a typical Fourier transform of Mn in PS II from our data. The first Fourier peak requires two different distances to be fit adequately, one at ~ 1.8 Å and the other 1.95–2.15 Å to light elements like C, N, or O. The second peak on the other hand is best fit by a heavier atom and, as noted above, probably a Mn atom at 2.7 Å; the amplitude is best fit to an average of one such interaction per Mn atom in the complex. Fitting the third peak, at a longer distance is more difficult. It can be fit to C, or Mn or Ca, or to a combination of these entities at 3 Å. The quality of fits is always better when Mn is included, but it is not clear if there is only Mn or also Ca atoms (Latimer et al., 1995). The presence of both the Mn–O (C,N) distances at ~ 1.8 Å and the Mn– Mn distances at 2.7 Å in all known bridged multinuclear Mn complexes, and the presence of a 3.3 Å Mn–Mn separation in monobridged Mn complexes leads to the conclusion that both of these structural motifs are present in the Mn complex of PS II (Wieghardt 1989; Pecoraro 1992). The number of such scattering interactions leads to the conclusion (remembering that there are four Mn atoms) that there are at least two Mn–Mn bridged

349

units and at least one Mn–Mn bridged unit (DeRose et al., 1994). Figure 8 shows the Fourier transform of the same system in oriented samples (Mukerji et al., 1994). It is evident that the 2.7 and 3.3 Å Mn– Mn vectors (the second and third peak in the Fourier transform) are oriented differently. The 2.7 Å vector is more parallel to the membrane normal and the 3.3 Å is more perpendicular. Detailed analysis of the dichroism can be used to determine the angles more accurately. The dichroism requires that the complex be asymmetric. One of the many possible structures (DeRose et al., 1994) consistent with data shown in Fig. 7 and the dichroism measurements is shown in Fig. 9 and provides us with a working model for the Mn cluster (Yachandra et al., 1993) Figure 10 shows the Mn K-edge spectrum from complexes in oxidation states (II), (III) and (IV). The inflection point shifts to higher energy as the oxidation state increases. There is also a dramatic change in the general shape of the edge as shown by the changes in the second derivatives. The shape of the edges is also an important indicator of oxidation state, as seen in Fig. 10 (Yachandra et al., 1993). A combination of the position and the shape were used to assign the oxidation states

350

V.K. Yachandra and M.P. Klein in the and states to and respectively. Fig. 11 shows the Mn K-edge from preparations in the and states of the photosystem II complex, and it shows that the Mn K-edge shifts to higher energy on advancing from to and from to indicating oxidation of Mn (Goodin et al., 1984; McDermott et al., 1988b; Guiles et al., 1990a). However no such shift was evidenced in samples prepared in the state by a cryogenic double turnover protocol, which leads to the conclusion that Mn is not oxidized in the to transition. It was proposed that the oxidation equivalent is stored on a ligand or amino acid residue (Guiles et al., 1990b). Ono et al. (1992) have presented Mn XANES of PS II samples that advanced progressively from by one through five laser flashes through the Kok cycle. Their data were interpreted to provide evidence for Mn oxidation from to and to and then reduction on going from to It is important to point out that Ono et al. have assumed that their samples were advanced on each flash but have presented no substantiating data, such as the pattern of multiline EPR signal intensity vs. flash. We have recently completed such a study (Roelofs et al., 1995, Andrews et al., 1994), and our data support our earlier results that showed Mn is oxidized during the to and to transition but that it may not be oxidized on the to advance.

IV. Future Directions The future holds much promise for the use of XAS in photosynthesis, and bio-inorganic chemistry in general. The availability of new storage rings and beamlines at these storage rings dedicated to research in biology is increasing. There has been considerable improvement in the brightness of the X-ray sources and also in detector technology. This makes the measurement of Xray absorption spectra of dilute species easier. The high brightness and the increased sensitivity of the technique are being used to couple microscopy with spectroscopy. XAS in the soft X-ray region, which has been mostly used in materials and surface science studies, is increasingly being applied to biological problems. Some of the edges of interest in photosynthesis include the K-edges of P, S and Cl and also C, N, and O, and the L-edges of transition

X-ray absorption spectroscopy

metals. L-edges, as shown in Fig. 1, are 2p to 3d transitions, and in contrast to the 1s–3d transitions in K-edge spectra are electric dipole allowed and hence are intense. The natural linewidths of K and L-edge transitions, for example, in Mn are 1.12 and 0.32 eV, respectively, making L-edges considerably more sensitive to factors such as symmetry that influence the d orbital splittings and population. Differential absorption of circularly polarized X-rays in the presence of a magnetic field (XMCD, X-ray magnetic circular dichroism) is another method that is being applied to biological systems. X-ray magnetic circular dichroism with its selectivity for paramagnetic species and the relative magnetic orientation of different species in a mulicenter system promises to be a powerful tool for studying metal centers in the photosynthetic apparatus.

351

Acknowledgements We thank Dr. Matthew Latimer and Dr. Annette Rompel for a critical reading of the manuscript. We are grateful to Prof. Kenneth Sauer for his suggestions. The work from our laboratory presented in this article was supported by the National Science Foundation grant DMB91– 0414, and by the Director, Division of Energy Biosciences, Office of Basic Energy Sciences, Department of Energy (DOE) under contract DEAC03–76SF00098. Synchrotron radiation facilities were provided by the Stanford Synchrotron Radiation Laboratory (SSRL) and the National Synchrotron Light Source (NSLS), both supported by DOE. The Biotechnology Laboratory at SSRL and Beam Line X9–A at NSLS are supported by the National Center for Research Resources of the National Institutes of Health.

352 References Andrews JC, Cinco R, Dau H, Latimer MJ, Liang W, Roelofs TA, Rompel A, Sauer K, Yachandra VK and Klein MP (1994) Photosynthetic water oxidation: Structural insights to the catalytic manganese complex. Physica B 208: 657– 659. Bertagnolli H and Ertel TS (1994) X-ray absorption spectroscopy of amorphous solids, liquids, and catalytic and biochemical systems - capabilities and limitations. Angew Chem Int Ed Engl 33: 45–66. Bianconi A (1988) XANES spectroscopy. In: Koningsberger DC and Prins R (ed) X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, pp 573–662 John Wiley, New York. Blackburn NJ (1990) EXAFS studies of copper proteins. In: Hasnain SS (ed) Synchrotron Radiation in Biophysics, pp 63–103. Ellis Horwood, Chichester, UK. Britt RD, Sauer K, Klein MP, Knaff DB, Kriauciunas A, Yu C-A, Yu L and Malkin R (1991) Electron spin echo envelope modulation spectroscopy supports the suggested coordination of two histidine ligands to the Rieske Fe-S centers of the cytochrome complex of spinach and cytochrome complexes of Rhodospirillum rubrum, Rhodobacter sphaeroides R-26, and bovine heart mitochondria. Biochemistry 30: 1892–1901. Bunker G, Stern EA, Blankenship RE and Parson WW (1982) An X-ray absorption study of the iron site in bacterial photosynthetic reaction centers. Biophys J 37: 539–551. Carmelli C, Huang JY, Mills DM, Jagendorf AT and Lewis A (1986) Extended X-ray absorption fine structure of and complex bound to coupling factor 1 of the from chloroplasts J Biol Chem 261: 16969–16975. Conradson SD, Burgess BK, Newton WE, Hodgson KO, McDonald JW, Rubinson JF, Gheller SF, Mortenson LE, Adams MWW, Mascharak PK, Armstrong WA and Holm RH (1985) Structural insights from the Mo K-edge X-ray absorption near edge structure of the iron-molybdenum protein of nitrogenase and its iron-molybdenum cofactor by comparison with synthetic Fe–Mo–S clusters. J Am Chem Soc 107: 7935–7940. Cramer SP (1988) Biological applications of XAS, In: Koningsberger DC and Prins R (eds) X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, pp 257– 320. John Wiley and Sons, New York. Cramer SP, Tench O, Yocum M and George GN (1988) A 13–element Ge detector for fluorescence EXAFS. Nucl Instrum Methods A266; 586–591. Debus R (1992) The manganese and calcium ions of photosynthetic oxygen evolution. Biochim Biophys Acta 110: 269– 352. DeRose VJ, Mukerji I, Latimer MJ, Yachandra VK, Sauer K and Klein MP (1994) Comparison of the manganese oxygen-evolving complex in photosystem II of spinach and Synechococcus sp. with multinuclear manganese model compounds by X-ray absorption spectroscopy. J Am Chem Soc 116: 5239–5249. Durham PJ (1988) Theory of XANES. In: Koningsberger DC

V.K. Yachandra and M.P. Klein and Prins R (eds) X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, pp 53–84. John Wiley and Sons, New York. Eisenberger P and Kincaid BM (1978) EXAFS: New horizons in structure determinations. Science 200: 1441–1447. Eisenberger P, Okamura MY and Feher G (1982) The electronic structure of in reaction centers from Rhodopseudomonas sphaeroides II. Extended X-ray absorption fine structure. Biophys J 37: 523–538. George GN, Prince RC and Cramer SP (1989) The manganese site of the photosynthetic water-splitting enzyme. Science 243: 789–791. Goodin DB (1983) The local structure of manganese in the photosynthetic apparatus and superoxide dismutase: An Xray absorption study. Ph D Dissertation. Universtity of California, Berkeley, CA, USA Lawrence Berkeley Laboratory Report, LBL–16901. Goodin DB, Yachandra VK, Britt RD, Sauer K and Klein MP (1984) State of manganese in the photosynthetic apparatus. 3 Light-induced changes in X-ray absorption (K-edge) energies of manganese in photosynthetic membranes. Biochim Biophys Acta 767: 209–216. Guiles RD (1988) Structure and Function of the Manganese Complex Involved in Photosynthetic Oxygen Evolution Determined by X-ray Absorption Spectroscopy and Electron Paramagnetic Resonance Spectroscopy. Ph D Dissertation University of California, Berkeley, CA, USA, Lawrence Berkeley Laboratory Report, LBL-25186. Guiles RD, Yachandra VK, McDermott AE, Cole JL, Dexheimer SL, Britt RD, Sauer K and Klein MP (1990a) The state of photosystem II induced by hydroxylamine: differences between the structure of the manganese complex in the and states determined by X-ray absorption spectroscopy. Biochemistry 29: 486–496. Guiles RD, Zimmermann J-L, McDermott AE, Yachandra VK, Cole J, Dexheimer SL, Britt RD, Wieghardt K, Bossek U, Sauer K and Klein MP (1990b) The state of photosystem II: differences between the structure of the manganese complex in the and states determined by X-ray absorption spectroscopy. Biochemistry 29: 471–485. Gurbiel RJ, Ohnishi T, Robertson DE, Daldal F and Hoffman BM (1991) Q-band ENDOR spectra of the Rieske protein from Rhodobactor capsulatus ubiquinol-cytochrome c oxidoreductase show two histidines coordinated to the [2Fe– 2S] cluster. Biochemistry 30: 11579–11584. Heald SM (1988a) Design of an EXAFS experiment. In: Koningsberger DC and Prins R (eds) X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, pp 87– 118. John Wiley and Sons, New York. Heald SM (1988b) EXAFS with synchrotron radiation. In: Koningsberger DC and Prins R (eds) X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, pp 119– 161. John Wiley and Sons, New York. Jaklevic J, Kirby JA, Klein MP, Robertson AS, Brown GS and Eisenberger P (1977) Fluorescence detection of EXAFS-sensitivity enhancement for dilute species and thin films. Solid State Commun 23: 679–682. Kirby JA, Goodin DB, Wydrzynski T, Robertson AC and Klein MP (1981) State of manganese in the photosynthetic

X-ray absorption spectroscopy apparatus. 2 X-ray absorption edge studies on manganese in photosynthetic membranes. J Am Chem Soc 103: 5537– 5542. Klein MP, Sauer K and Yachandra VK (1993) Perspectives on the structure of the photosynthetic oxygen evolving manganese complex and its relation to the Kok cycle. Photosynth Res 38: 265–277. Krauss N, Hinrichs W, Witt I, Fromme P, Pritzkow W, Dauter Z, Betzel C, Wilson KS, Witt H and Saenger W (1993) 3–Dimensional structure of system-I of photosynthesis at 6 angstrom resolution. Nature (London) 361: 326– 331. Kutzler FW, Natoli CR, Misemer DK, Doniach S and Hodgson KO (1980) Use of one-electron theory for the interpretation of near edge structure in K-shell X-ray absorption spectra of transition metal complexes. J Chem Phys 73: 3274–3288. Latimer MJ, DeRose VJ, Mukerji I, Yachandra VK, Sauer K and Klein MP (1995) Evidence for the proximity of calcium to the manganese cluster of photosystem II: Determination by X-ray absorption spectroscopy. Biochemistry 34: 10898–10909. Lee PA, Citrin PH, Eisenberger P and Kincaid BM (1981) Extended X-ray absorption fine structure - Its strengths and limitations as a structural tool. Rev Mod Phys 53: 769– 806. MacLachlan DJ, Hallahan BJ, Ruffle SV, Nugent JHA, Evans MCW, Strange RW and Hasnain SS (1992) A EXAFS study of the manganese complex in purified photosystem II membrane fractions. Biochem J 285: 569– 576. McDermott AE, Yachandra VK, Guiles RD, Britt RD, Dexheimer SL, Sauer K and Klein MP (1988a) Low-potential iron–sulfur centers in photosystem I: An X-ray absorption spectroscopy study. Biochemistry 27: 4013–4020. McDermott AE, Yachandra VK, Guiles RD, Cole JL, Britt RD, Dexheimer SL, Sauer K and Klein MP (1988b) Characterization of the manganese complex and the iron-quinone acceptor complex in photosystem II from a thermophilic cyanobacterium by electron paramagnetic resonance and X-ray absorption spectroscopy. Biochemistry 27: 4021–4031. McDermott AE, Yachandra VK, Guiles RD, Sauer K, MP Klein, Parrett KG and Golbeck J (1989) EXAFS structural study of the low potential Fe–S center in photosystem I Biochemistry 28: 8056–8059. McKale AG, Veal BW, Paulikas AP, Chan SK and Knapp GS (1988) Improved ab initio calculations of amplitude and phase functions for extended X-ray absorption fine structure spectroscopy. J Am Chem Soc 110: 3763–3768. Mukerji I, Andrews JC, DeRose VJ, Latimer MJ, Yachandra VK, Sauer K and Klein MP (1994) Orientation of the oxygen-evolving manganese complex in a photosystem II membrane preparation: An X-ray absorption spectroscopy study. Biochemistry 33: 9712–9721. Ono T-A, Noguchi T, Inoue Y, Kusunoki M, Matsushita T and Oyanagi H (1992) X-ray detection of the period-four cycling of the manganese cluster in the photosynthetic water oxidizing enzyme. Science 258:1335–1337. Pecoraro VL (1992) Structurally diverse manganese coordi-

353 nation compounds. In: Pecoraro VL (ed) Manganese Redox Enzymes, pp 197–231, VCH Publishers, New York. Pendry JB (1983) The transition between XANES and EXAFS. In: Bianconi A, Incoccia L and Stipcich S (eds) EXAFS and Near Edge Structure, p 4. Springer-Verlag, Berlin. Penner-Hahn JE, Fronko RM, Pecoraro VL, Yocum CF, Betts SD and Bowlby NR (1990) Structural characterization of the manganese sites in the photosynthetic oxygenevolving complex using X-ray absorption spectroscopy. J Am Chem Soc 112: 2549–2557. Powers L (1982) X-ray absorption spectroscopy: applications to biological molecules. Biochim Biophys Acta 683: 1– 38. Powers L, Schägger H, von Jagow G, Smith J, Chance B and Ohnishi T (1989) EXAFS studies of the isolated bovine heart Rieske cluster. Biochim Biophys Acta 975: 293–298. Rehr JJ, de Leon JM, Zabinsky SI and Albers RC (1991) Theoretical X-ray absorption fine structure standards. J Am Chem Soc 113: 5135–5140. Rehr JJ, Albers RC and Zabinsky SI (1992) High-order multiple-scattering calculations of X-ray absorption fine structure. Phys Rev Lett 69: 3397–3400. Roe AL, Schneider DJ, Mayer RJ, Pyrz JW, Widom J and Que L (1984) X-ray absorption spectroscopy of iron-tyrosinate proteins. J Am Chem Soc 106: 1676–1681. Roelofs TA, Liang W, Latimer MJ, Cinco R, Rompel A, Andrews JC, Yachandra VK, Sauer K and Klein MP (1995) Manganese oxidation states of the flash-induced S-states of photosystem II. In: Mathis P (ed) Photosynthesis from Light to Biosphere, Vol. II, pp 459–462. Kluwer Academic Publishers, Dordrecht. Rutherford AW, Zimmermann, J-L and Boussac A (1992) Oxygen evolution. In: Barber J (ed) The Photosystems: Structure, Function and Molecular Biology, pp 179–229. Elsevier, Amsterdam. Rypniewski WR, Breiter DR, Benning MM, Wesenberg G, Oh B-H, Markley JL, Rayment I and Holden HM (1991) Crystallization and structure determination to 2.5 Å resolution of the oxidized [2Fe–2S] ferredoxin isolated from Anabaena 7120 Biochemistry 30: 4126–4131 Sauer K, Yachandra VK, Britt RD and Klein MP (1992) The photosynthetic water oxidation complex studied by EPR and X-ray absorption spectroscopy. In: Pecoraro VL (ed) Manganese Redox Enzymes, pp 141–175. VCH Publishers, New York. Sayers DE and Bunker BA (1988) Data Analysis. In: Koningsberger DC and Prins R (eds) X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, pp 211– 253. John Wiley and Sons, New York. Sayers DE, Stern EA and Lytle FW (1971) New technique for investigating noncrystalline structures: Fourier analysis of the extended X-ray absorption fine structure. Phys Rev Lett 27: 1204–1207. Scott RA (1984) X-ray absorption spectroscopy. In: Rousseau RL (ed) Structural and Resonance Techniques in Biological Research, pp 295–362. Academic Press, Orlando, FL. Scott RA and Eidsness MK (1988) The use of X-ray absorption spectroscopy for detection of metal–metal interac-

354 tions. Applications to copper-containing enzymes. Comments Inorg Chem 7: 235–267. Shulman RG, Yafet Y, Eisenberger P and Blumberg, WE (1976) Observation and interpretation of X-ray absorption edges in iron compounds and proteins. Proc Natl Acad Sci USA, 73: 1384–1388. Spiro TG (1982) Iron–Sulfur Proteins. John Wiley, New York Stern E (1988) Theory of EXAFS. In: Koningsberger DC and Prins R (ed) X-ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, pp 3– 51. John Wiley and Sons, New York. Stöhr J (1992) NEXAFS Spectroscopy. Springer-Verlag, Berlin. Teo B-K (1986) EXAFS: Basic Principles and Data Analysis. Springer-Verlag, New York. Teo B-K and Shulman RG (1982) X-ray absorption studies of iron–sulfur proteins and related compounds. In: Spiro TG (ed,) pp 343–366. John Wiley and Sons, New York. Tsang H-T, Batie CJ, Ballou DP and Penner-Hahn JE (1989)

V.K. Yachandra and M.P. Klein X-ray absorption spectroscopy of the [2Fe–2S] Rieske cluster in Pseudomonas cepacia phthalate dioxygenase. Determination of core dimensions and iron ligation. Biochemistry 28: 7233–7240. Tsukihara T, Fukuyama K, Tahara H, Katsube Y, Matsuura Y, Tanaka N, Kakudo M, Wada K, and Matsubara H (1978) X-ray analysis of ferredoxin from Spirulina platensis. II Chelate structure of active center. J Biochem (Tokyo) 84: 1645–1647. Wieghardt K (1989) The active site in manganese-containing metalloproteins and inorganic model compounds. Angew Chem Int Ed Engl 28: 1153–1172. Yachandra, VK (1995) X-ray absorption spectroscopy and applications in structural biology. In: Sauer K (ed) Methods in Enzymology. Biochemical Spectroscopy Vol 246. pp 638–675. Academic Press, Orlando, FL. Yachandra VK, DeRose VJ, Latimer MJ, Mukerji I, Sauer K and Klein MP (1993) Where plants make oxygen: A structural model for the photosynthetic oxygen evolving manganese cluster. Science 260: 675–679.

Chapter 22 Mössbauer Spectroscopy Peter G. Debrunner Physics Department, University of Illinois at Urbana-Champaign, 1110 W. Green Street, Urbana, IL 61801–3080, USA

Summary I. Introduction II. Mössbauer Spectroscopy: Physics and Formalism A. Basic Features B. Hyperfine Interactions C. Electronic States of the Iron D. Spin Coupling E. Dynamical Aspects F. Calculation of Mössbauer Spectra G. Experimental Considerations III. Applications A. Iron–quinone Complex B. Iron–sulfur Proteins C. Cytochromes References

355 356 357 357 357 359 361 362 363 364 365 365 368 371 371

Summary spectroscopy (MS) has played an important role in the elucidation of the iron centers in the photosynthetic apparatus. In 1975, G. Feher and collaborators demonstrated that the single iron of bacterial reaction centers (RC) was high-spin ferrous irrespective of the state of Moreover, they showed that reduction of broadened the Mössbauer lines at 4.2K, indicative of spin coupling between the semiquinone and the iron, related to the broadening observed in the EPR signal of the semiquinone (Debrunner et al., 1975). Photosystem I (PSI) of green plants and algae contains three iron–sulfur centers labeled and that have originally been identified by EPR. Evans et al. (1977, 1979, 1981) showed that the Mössbauer spectra of PSI were practically identical with those of the well understood bacterial 4Fe–4S centers. The low-potential center remained controversial, however, as others suggested a 2Fe–2S center instead. The controversy was resolved by Petrouleas et al. (1989), who studied a mutant lacking centers and and found that had indeed all the properties of a 4Fe–4S center. The more difficult task of analyzing Photosystem II (PSII) of green plants was undertaken by Petrouleas and Diner (1982, 1986, 1990), who identified the redox center known as with the iron– quinone complex and showed the iron to be redox active in contrast to the Fe(II) of the bacterial RC. The same group demonstrated, by MS, that formate affected the iron–quinone complex (Diner and Petrouleas, 1987), and finally that the Fe(II) formed an NO derivative.

Correspondence: Fax: 1-217-3339819; E-mail: [email protected]

355 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 355–373. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

356

Peter G. Debrunner

After a review of the methodology, the various iron sites of bacterial RC, of PSI and PSII will be discussed in detail to illustrate the application of the method. Abbreviations: EFG – electric-field gradient; EPR – electron paramagnetic resonance; LDAO – lauryldimethylamine N-oxide; MS – Mössbauer spectroscopy; PSI – Photosystem I; PSII – Photosystem II; RC – (bacterial) reaction center

I. Introduction This chapter attempts to acquaint the reader with the basic concepts of spectroscopy (MS) and its applications in photosynthesis research. is the most important Mössbauer isotope, and since the photosynthetic apparatus contains several prominent iron complexes, MS can be utilized to study the properties of these complexes under various experimental conditions. For more general and in-depth treatments of MS the reader is referred to the literature (Greenwood and Gibb, 1971; Gonser, 1975; Gibb, 1976; Gütlich et al., 1978; Huynh and Kent, 1984; Cranshaw et al., 1985; Dickson and Berry, 1986; Debrunner, 1993). Since MS involves the resonant absorption of gamma rays by nuclei, it has several features that set it apart from other types of spectroscopy. First of all, the quantum energy of the transition is quite large, in the case of yet the intrinsic linewidth, is sufficiently small to resolve the hyperfine interactions of the nucleus with its electron shell. The applications of MS discussed here indeed use the known electric and magnetic moments of the to probe the internal electric and magnetic fields produced by the electrons and to gather information, in the process, about the state of the iron. Secondly, the 14.4 keV radiation to be resonantly absorbed by the nuclei is emitted by a source of radioactive which decays to stable via the 14.4 keV excited state. The source thus emits 14.4 keV gamma rays of intrinsic line width (plus other radiations contributing to the non-resonant background). Given the discrete energy, keV, of the radiation emitted by the source, the spectrum of the absorber can be scanned by mov-

ing the source with velocity v relative to the absorber, imparting thereby a Doppler shift to the gamma rays,

where c is the speed of light. It is customary in MS to express all energies in terms of the measured velocities v. To convert to more familiar energy units the following conversion factors are useful:

Thirdly, it follows from energy and momentum conservation that Mössbauer transitions of intrinsic linewidth can only occur in solids and only in a fraction f < 1 of the cases as will be discussed in Section II.E. A fourth unique feature of MS is its sensitivity to a single isotope, here Since the natural abundance of is only 2.2%, biological samples need to be enriched in to improve the signal-to-noise ratio. As will be discussed in Section II, different oxidation and spin states produce distinctly different spectral contributions, and with a few empirical rules it is in principle possible to resolve a composite spectrum into its components and to deduce the oxidation and spin state of each iron site. To illustrate this process, Section III discusses the analysis and interpretation of spectra taken on bacterial reaction centers, PSI and PSII. It will be noted that the appropriate level of sophistication in the data analysis depends on the quality of the data. At the very least, MS can assess the purity and composition of a sample as far as its iron content is concerned. Once the quality and reproducibility of the samples are established, the spectral components can be assigned to the various iron species present, and the assignments can be verified by titr-

Mössbauer spectroscopy ation, by the expected temperature and field dependences, etc. Finally, it may be possible to characterize the iron environment(s) further by fitting spectra, taken as a function of temperature and field, to a spin Hamiltonian or other theoretical model.

357 of an nucleus with transition encan be written as where depends on the transition probability and other factors. The intensity transmitted through an absorber with n nuclei per unit area then is

ergy

II. Mössbauer Spectroscopy: Physics and Formalism

A. Basic Features The stable ground state of has nuclear spin I = 1/2 and connects via magnetic dipole transition to the first excited state at 14.4 keV with spin I* = 3/2 and mean life The Heisenberg principle then predicts a Lorentzian energy distribution of the emission or absorption line with intrinsic width where is Planck’s constant. Most experiments are done in a transmission geometry, whereby the radiation emitted by a single-line source impinges on the absorber to be investigated, and the transmitted 14.4 keV gammas are counted in a detector, e.g. a proportional counter. As the Doppler-shifted energy of the incident beam matches the transition energy of some nuclei in the absorber, these nuclei absorb the resonant gammas in proportion to predictable transition probabilities, and the count rate of the detector decreases. Doppler shifts are varied periodically between and data are typically accumulated for many hours. The final spectrum is a plot of the total number of counts, N(v), recorded as a function of Doppler velocity v. To find a quantitative expression for a transmission spectrum let the intensity distribution of the incident beam be I(E, v)dE, where is the center of the Lorentzian line emitted by the source. With the substitutions can be written as which integrates to Here and (below) are the recoilless fractions of the source and absorber, respectively, and will be discussed further in Section II.E. Similarly, with the absorption cross section

with thin-absorber

The which approximates by is valid for most biological samples. To of Eq. (4) one has to add an energy-independent background B. The transmission spectrum therefore consists of a (negative) Lorentzian of full width at halfheight and relative height The maximum resonance cross section of with no hyperfine splitting is which is much larger than the (non-resonant) absorption cross section of of iron at 14.4 keV. In a more general case several types of iron environments will be found in an absorber, and each is subject to hyperfine interactions, which lift the two- and four-fold degeneracies of the ground and excited states, respectively. The factor n in Eq. (4) then refers to the of a given species, and as well as depends on the nuclear eigenstates involved in the transition as described below. limit,

B. Hyperfine Interactions This section discusses how the electron shell and/or external fields affect the nuclear energy levels and thereby the Mössbauer spectra, and how the electronic properties of the iron can be deduced from the spectra. In addition to the spin angular momenta I and I* of the ground and excited state already mentioned, the nucleus has magnetic moments Î and respectively, with and an electric quadrupole moment

358 where all starred quantities refer to the I* = 3/2 excited state, the ground state of spin I = 1/2 having no measureable quadrupole moment. In the expressions for and above is the nuclear magneton. The difference in mean-square charge radii, moreover, gives rise to an electric monopole interaction that shifts the transition energy proportional to the s-electron density, at the nucleus and manifests itself in a shift of the whole spectrum without affecting its shape. This so-called chemical or isomer shift, is given by the expression

where the subscripts A and S in the last term refer to absorber and source, respectively. In order to define the isomer shift independently of the particular source used, is usually quoted as relative to the centroid of the spectrum of metallic iron taken at 300K. It should be noted that the overall shift has a dynamical contribution as well which will be discussed in Section II.E. The electric quadrupole interaction, another correction to the Coulomb energy of the atom, accounts for the non-spherical charge distribution of the nucleus in the excited state. It splits the Mössbauer transition into two lines separated by the quadrupole splitting and occurs whenever an electric-field gradient (EFG) is present at the nucleus. An EFG arises from the valence electrons of the iron and from surrounding charges unless excluded by symmetry. In iron proteins high local symmetry with no EFG such as tetrahedral, octahedral, etc., is rare, and a measurable quadrupole splitting is therefore the rule. Since a single 3d-electron can produce a splitting of the valence contribution typically exceeds the lattice contribution which arises from charges outside the electron shell. The EFG or its negative, x,y,z, is a symmetric, traceless second rank tensor. It is readily calculated from the charge density of the valence electrons and from any external charges. For the covalent, low-symmetry iron sites of interest here a proper treatment requires a molecular orbital approach, but simpler crystal-

Peter G. Debrunner field approximations have been used in the past. Model calculations are complicated by the distortion of the inner electron shells as parametrized by the Sternheimer factors. Several equivalent expressions for the electric quadrupoleinteractions are in use (Abragam and Bleaney (1990))

In the last two expressions are the principal-axes components of and the are the components of the nuclear spin operator, Î, along these axes. The numerical values in Eq. (6) are those for the I* = 3/2 excited state. The last expression uses the condition and the definition of the asymmetry parameter

With the convention, is limited to the range In the absence of magnetic interactions splits the I* = 3/2 excited state into two doublets with a quadrupole splitting of

Since the ground state does not split, the quadrupole interaction leads to the characteristic twoline pattern of the Mössbauer spectrum, both lines being Lorentzians of minimum linewidth and equal areas for randomly oriented samples. The magnetic dipole interaction, finally, lifts the degeneracy of the nuclear levels completely, and the magnetic dipole selection rules allow six transitions between the two and the four equidistant levels of the ground and excited state, respectively. In the presence of electric quadrupole interaction, the I* = 3/2 eigenstates are linear combinations of the states, and up to eight transitions are possible. An external magnetic field B leads to the nuclear Zeeman interatction for the ground state,

where the last expression assumes that B defines

Mössbauer spectroscopy the direction of z. An analogous expression holds for the excited state. The magnetic hyperfine interaction couples the nuclear spin operators Î or Î* with the electron spin operators and is given by (Abragam and Bleaney 1970)

Here, the sum is over all unpaired electrons with position orbital angular momentum and spin and is a numerical factor of roughly 0.35. An analogous expression applies for the excited state. The three contributions in Eq. 10 are the orbital, the traceless spin dipolar and the isotropic Fermi contact term, respectively, where the last one generally dominates. As stated in Eq. 10 the hyperfine tensor à has units of energy and can be compared directly with EPR/ENDOR data. It is convenient, however, to divide à by so that

represents an internal field that can be substituted for, or added to, in Eq. 9. Note that Eq. 11 contains the expection value of which is a classical vector rather than an operator so that the left-hand side can be equated to a field. Whenever a sufficiently large external field is aplied to a sample of spin such that the Zeeman interaction far exceeds the magnetic hyperfine interaction, typically then can be calculated from the electron spin Hamiltonian alone, Eq. 15, as will be discussed later in Section II.C. In field units the Fermi contact term for six-coordinate iron with oxygen/ nitrogen ligands has the value which decreases to for the highly covalent coordination of iron-sulfur proteins. In a general case the nuclear Hamiltonian is given by the sum

If the internal field approach, Eq. 11, is valid, simplifies to

which depends on nuclear spin operators only.

359 Since refers to a molecular frame while is defined in the laboratory, each molecule in a randomly oriented sample will have a different nuclear Hamiltonian and therefore different energy levels and transition probabilities. As a result, the Mössbauer spectrum no longer consists of a set of discrete lines but rather of a continuous distribution, and any quantitative analysis requires computer programs. The information available from such an analysis can be substantial even for the simplest case of a diamagnetic compound with Here, the applied field broadens the quadrupole doublet, and the lineshape allows one to deduce as well as the sign of two parameters that characterize the symmetry of the valence electrons in more detail than the quadrupole splitting alone.

C. Electronic States of the Iron Ferric iron has five 3d-electrons with spins arranged either parallel, resulting in a orbital singlet of total spin S = 5/2, or with four spins paired leaving a single unpaired spin, S = 1/2, in an orbital triplet The former, high-spin state is found in more ionic, weak-field compounds, whereas the latter, low-spin state is found in highly covalent, strong-field compounds, e.g., the ferricytochromes. The spin state of ferric compounds is readily recognizable from MS, and although the same can be said of EPR, which is vastly more sensitive, Mössbauer data can provide additional information not available otherwise. The combination of the two methods is clearly most powerful. We begin with a discussion of high-spin Fe(III). The sixfold spin degeneracy of the state is partially lifted by the zero-field splitting which is given to lowest order by the expression

By convention, the rhombic term E in Eq. 13 is limited to in a ‘proper’ coordinate system. The axial term D is typically small for S = 5/2 iron, but both signs of D are possible. The eigenstates of are three Kramers doublets; for E = 0 these doublets are the

360 states, while they are linear combinations of the states for The remaining twofold degeneracy is lifted by the Zeeman interaction where is the Bohr magneton, and is the g-tensor, which is expected to have the spin-only value of since the state allows no orbital contribution. The total electronic spin Hamiltonian is then the sum of Eq. 10 should in principle be added to Eq. 15 since it depends on the electron spin operator Inclusion of is warranted for vanishing only since requires an expansion of the basis set to (2I + 1)(2S + 1), etc., and complicates the solution of Eq. 15 considerably. For is a small perturbation that does not affect appreciably and can therefore be neglected in Eq. 15. Diagonalization of yields the six eigenstates and their energies and allows one to calculate EPR transitions, the spin expectation values of each state, etc. In the presence of an applied field mT the internal field formalism, Eq. 11, is valid, and the nuclear eigenstates and energies can be found from Eq. Each spin eigenstate produces a different internal field, however, and for each molecular orientation six different component spectra with different Boltzmann factors have to be added. In practice, spectra of high-spin Fe(III) typically show well resolved bands with overall splittings of roughly 16 for The orbital singlet state, allows no orbital and spin dipolar contribution to Ã, and any deviations reported from isotropy are indeed small. The nominal singlet also predicts zero valence contribution to the quadrupole interaction, but does not rule out lattice contributions or effects due to anisotropic electron delocalization. Typical splittings are of the order but exceptional cases with twice that value are known. Low-spin Fe(III), a orbital triplet, is best understood as a single-hole state in an otherwise filled subshell. The magnetic properties, in particular the g- and A-tensors, are reasonably

Peter G. Debrunner well reproduced by a crystal-field model proposed by Griffith (1957). According to this model the threefold orbital degeneracy obtained in strong octahedral field is lifted by crystal field components of lower symmetry, namely an axial and a rhombic component. If the Hamiltonian consisting of the electrostatic potential energy and the known spin-orbit interaction is solved, three Kramers doublets are obtained as linear combinations of orbitals with appropriate spin functions. The splittings are such that only the ground doublet is populated at any accessible temperature, and once the wavefunction of this ground doublet is known, all observables, in particular the tensors Ã, and can be calculated. In practice, the axial and rhombic crystal field components are adjusted to match the measured gvalues, and Taylor (1977) has given a particularly simple algorithm for this purpose. The g-values may deviate substantially from the spin-only value due to orbital contributions, and the Atensors typically are quite anisotropic for the same reason and because of spin dipolar contributions. The quadrupole splitting reflects the combination of orbitals that make up the ground state wavefunction: if a single orbital dominates, may be as large as 3 if all orbitals contribute equally, may be close to zero. In low-spin heme compounds the lattice or covalency contributions to are substantial (Debrunner, 1989). Ferrous iron has six 3d-electrons, i.e. one more than Fe(III), and both high- and low-spin states are found in weak- and strong- field compounds, respectively. Low-spin Fe(II) has a filled subshell and is therefore diamagnetic as exemplified by ferrocytochromes. The valence contribution to the quadrupole splitting is zero, but the lattice and covalency contributions lead to in the cytochromes. Isomer shifts of the latter are in the range of High-spin Fe(II) is characterized by the largest isomer shifts, 0.9 where the low value applies to the more covalent, 5-coordinate heme proteins, the high value to the most ionic, 6-coordinate compounds. The exceptionally covalent coordination of the ironsulfur proteins leads to Another characteristic of high-spin Fe(II) is the large quadrupole splitting of

361

Mössbauer spectroscopy

which has a noticeable temperature dependence as illustrated in Fig. 1 and explained below. The extra 3d-electron of high-spin Fe(II), which causes the large leads to a or state in octahedral or tetrahedral symmetry, respectively. As illustrated for the case of in the upper inset of Fig. 1, crystal field components of axial and/or rhombic symmetry lift the orbital degeneracy, and the thermal population of the higher orbital state(s) with different quadrupole interactions averages out the observable and hence makes it temperature dependent. A proper treatment of has to include the spin-orbit coupling as well as the axial and rhombic crystalfield terms (Ingalls, 1964). The magnetic properties of high-spin Fe(II) are usually described by the spin-Hamiltonian, Eq. 15. The zero-field splittings are typically several i.e. larger than in high-spin ferric compounds, and the eigenstates of Eq. 13 are singlets for rather than Kramers doublets. The last point is an important, general distinction between integer-spin and half-integer spin systems: For integer-spin, non-Kramers systems the spin expectation values vanish in zero field, and no spontaneous magnetic hyperfine splitting is observed.

The Mössbauer spectra of high-spin Fe(II) therefore show unbroadened quadrupole doublets in zero field as illustrated in Fig. 2 (top). The only known exceptions are magnetically ordered materials or the unusual case that is comparable to the splitting between the eigenstates of Eq. 13. Strong external fields will induce internal fields described by Eq. 11 and thus allow one to deduce the hyperfine tensor à as illustrated in Fig. 2 (bottom), which will be discussed further in Section III.A. It should be kept in mind that the parameters of Eq. 15 are constants only as long as the next higher orbital level is far removed in energy from the ground level, a condition that is certainly not satisfied if is strongly temperature dependent.

D. Spin Coupling The formalism of Section II.C. refers to isolated iron sites. In many biological systems including the photosynthetic apparatus, however, the iron

362 is either found in clusters as in the 4Fe–4S proteins or near a radical as in the ferroquinone complex. In both cases the component spins couple, changing the properties of the coupled system drastically. This subsection therefore attempts to describe the coupled system in terms of the properties of its parts. The simplest type of coupling between two spin operators and is given by the Heisenberg-Van Vleck expression for isotropic exchange, which is the dominant term when the wavefunctions of the two centers overlap. has eigenstates of spin and energy For J > 0 the state of lowest energy is thus the one with the smallest spin S, and the opposite is true for J < 0. The situation is simple as long as is much larger than any other term in the spin Hamiltonian, Eq. 15, of the spins and This is the case for the 2Fe–2S proteins, which exist in the oxidation states Fe(III)–Fe(III) and Fe(III)– Fe(II) and have strong antiferromagnetic coupling so their net spins are S = 0 and S = 1/2, resp. The situation is more complicated in the case of the 4Fe – 4S proteins, which exist in an oxidized and a reduced state. The problem is that there are six interacting iron pairs that can not simultaneously align their spin antiparallel. The energy of the system can be lowered, however, by double exchange (Münck et al., 1988), whereby the extra electron of Fe(II) delocalizes onto Fe(III) so that the average charge per iron is 2.5 as judged by the isomer shift. The spins of the delocalized pair align parallel to a spin of 9/2. Oxidized has two such pairs lining up antiparallel to a cluster spin of S = 0. In reduced there is a delocalized, S' = 9/2 pair and a diferrous pair with aligned spin, S'' = 4 and the cluster spin is S = 1/2. For the system spin is a good quantum number, and using spin algebra the expectation value of any operator involving or can be calculated in any eigenstate of To illustrate the process, consider Eq. (10), written here for the intrinsic spin In

Peter G. Debrunner terms of the system spin the Hamiltonian becomes where the A-tensor in the S representation is given by If is small or comparable to as is the case in the ferroquinone complex, the full Hamiltonian has to be diagonalized.

E. Dynamical Aspects The vibrational motion of the iron affects the Mössbauer spectra in three significant ways, viz. (i) the recoilless fraction f, (ii) the thermal redshift which adds to the isomer shift, and (iii) the spin-lattice coupling, which controls the spin dynamics. We will discuss these effects in this order. As stated in the Introduction, the Mössbauer effect is observable in solids only. The reason is that energy and momentum have to be conserved in the emission and absorption processes, and this is possible without recoil energy loss only if the emitting or absorbing atom is part of a quantized vibrational system with discrete energy levels. When a bound iron absorbs a photon, the lattice will on average increase its energy by the recoil energy and it does so by emitting n phonons, n = 0,1,2,..., where the inclusion of n = 0 is crucial as it implies a zero-phonon, recoilless Mössbauer transition. The larger the phonon energy as compared to the recoil energy, the larger the recoilless fraction f will be, where f is the fraction of n = 0 events out of the total. For a solid with harmonic forces the recoilless fraction is given by where is the mean-square displacement of the iron in the direction of the gamma ray and pm is the wavelength of the 14.4 keV radiation. Since is a monotonically increasing function of temperature, T, the recoilless fraction decreases rapidly with increasing T. The zero-point vibrations of iron proteins are comparable to those of other iron compounds, and recoilless fractions of are typical at 4.2 K. For iron proteins, the linear increase of for 50K < T < 150K is larger, however, and for T > 150K non-vibrational motions contribute to increasing its rise with T further.

Mössbauer spectroscopy

363 ues in Eq. 11 can be replaced by their thermal average where i refers to the (2S + 1) eigenstates of in Eq. 15. For Fe(III) in weak fields the magnetic hyperfine splitting collapses completely in the fast-fluctuation limit since approaches zero, thus the spectra simplify and the whole intensity concentrates in the emerging quadrupole doublet. While Mössbauer spectra can be simulated for any spin fluctuation rate and the temperature dependence has been modeled successfully (Schulz et al., 1988), it is not always practical to reach the simple limiting cases, and the data analysis then remains difficult.

F. Calculation of Mössbauer Spectra

For T > 230K limited diffusion sets in, which results in line broadening, motional decrease of etc. (Keller et al., 1980; Aleksandrov et al., 1987). The thermal redshift or second-order Doppler shift is given by

where is the mean-square velocity of the iron, which stays close to the minimum value given by zero-point vibrations for T < 50K and approaches the classical limit of 3kT/M without reaching it at high temperatures. The thermal redshift adds to the isomer shift and is illustrated in Fig. 3. The last dynamical feature to be discussed is the spin–phonon coupling that causes transitions between the eigenstates of the spin Hamiltonian Eq. 15, and is related to the spin relaxation rate in EPR. In the presence of magnetic hyperfine interactions spin state fluctuations affect the shape of the spectra profoundly. So far it has been tacitly assumed that the spin states are stationary on the Mössbauer time scale, a case that typically applies near 4.2K. In the opposite limit of fast spin state fluctuations, which is approached at higher temperatures, the spin expectation val-

A variety of computer programs are available for the quantitative analysis of Mössbauer spectra, and only the general principles will be discussed here. In the simplest case, the spectrum consists of a number of Lorentzians on a constant background, each characterized by its center, width and height, and a least-squares routine will find an optimum parameter set with standard deviations and correlation coefficients. The results will then have to be interpreted in terms of quadrupole doublets and/or single lines attributable to different spectral components. Alternatively, constraints can be incorporated in the fitting routine, e.g. requiring identical shapes or areas for both lines of a quadrupole doublet. To allow for inhomogeneity in the iron environment a Gaussian distribution of Lorentzians or Voigt line shape can be used with the Gaussian width as a parameter. In an external field each molecule has a different nuclear Hamiltonian, Eq. 12 or 12' and the spectra are calculated by adding contributions from an appropriate sample of molecular orientations, typically more than 100. For paramagnetic centers the spin expectation values or needed in Eq. 12 or 12' will also depend on the molecular orientation, thus both and have to be diagonalized for each orientation. Since and depend on a large number of parameters, a successful analysis is possible only with data of high quality, but the resulting parameter set, in particular Ã, and the relative orientation of these tensors, will characterize the iron environment in great detail.

364

Peter G. Debrunner

G. Experimental Considerations Most Mössbauer measurements are done in transmission, and count rates upward of are obtained with sources of roughly 50 mCi. If N is the total number of counts per data point, the mean square deviation of N equals N, and It therefore takes counts per data point to measure a 1% dip to a standard deviation of 10% and counts to measure a 0.1% dip to the same relative accuracy, etc. According to Section II.A the total area under the absorption spectrum is proportional to the recoilless fraction times the number of atoms per and for a given spectral area the peak absorption is obviously largest if the number of lines and the widths are minimal. Although MS can provide unique information about the environment of the iron in the photosynthetic apparatus, its application has been limited by the need for quantities of per sample. Enrichment in is therefore absolutely mandatory, and as long as the iron cannot be exchanged, growth of the photosynthetic organisms, either bacteria, cyanobacteria or algae, on enriched media has been the only practical approach. As pointed out by Evans et al. (1977), an excess of iron in the growth medium should be avoided. In R. sphaeroides with one iron per RC, samples containing up to of have been obtained, resulting in spectra with a total area of at 4.2K as shown in Fig. 2. Based on figures reproduced here, comparable estimates of in PSII preparations are 5.5 and in Fig. 4b and Fig. 6a,b, respectively. With two irons per PSII, these numbers imply a roughly tenfold lower concentration of RCs than in R. sphaeroides. To achieve equivalent signalto-noise ratios in the two cases, the number of counts accumulated would have to be times larger in the case of PSII. Any increase in the concentration of the iron-bearing proteins as well as any reduction in the number of species present in a sample, e.g. by genetic engineering or chemical means, will greatly facilitate the Mössbauer experiments. The problems are well illustrated by Fig. 4, which compares the spectra of membranes from cyanobacteria in (a) with oxygen-

evolving core complexes in (b). The total areas of the spectra are 1.8 and respectively, indicating an estimated amount of 15 and of The dominant doublet in Fig. 4a arises mainly from iron–sulfur proteins and cytochromes in PSI and the cytochrome complex; it hides the PSII spectrum of Fig. 4b, which actually contains a cytochrome impurity exceeding the amount of cytochrome of PSII (Picorel et al., 1994). PSI contains three 4Fe–4S clusters, and Figs. 10 and 11 indicate that samples of adequate signal-to-noise ratio can be prepared. Temperature is an important variable in any Mössbauer experiment since most samples are measured as frozen solutions, since the recoilless fraction is largest at helium temperatures, and since the spin fluctuation rates are strong functions of T. Accordingly, a variable temperature cryostat is essential. For samples of adequate sig-

Mössbauer spectroscopy nal-to-noise ratio external fields are useful also as illustrated in Figs. 2 and 11. III. Applications This section applies the basic concepts and formalism developed so far to the three types of iron sites found in the membrane-bound photosynthetic apparatus of bacteria, cyanobacteria and algae. Among the latter the experiments done on Chlamydomonas reinhardtii stand out. The three types of iron sites are the iron–quinone complex of the bacterial reaction center (RC) and PSII of the higher organisms, the iron–sulfur clusters of PSI, and the cytochrome of PSII. Here, the goal is to illustrate the arguments leading from experimental data to final conclusions and to point out strengths and weaknesses of the method.

A. Iron–quinone Complex Early Mössbauer experiments on bacterial RCs conclusively showed the existence of a ferroquinone complex which is now known in atomic detail from x-ray diffraction (Deisenhofer et al., 1985; Feher et al., 1989), although the function is still elusive (Debus et al., 1985). Equally significant was the demonstration of an analogous iron site in PSII, which was not only redox-active in contrast to the bacterial one, but was sensitive to the presence of bicarbonate/formate and able to form an Fe(II)NO adduct of spin S = 3/2. Since the iron quinone complex has no precedent, it will be discussed in some detail starting with the bacterial RC of Rhodobacter sphaeroides, Figs. 1–3 (Debrunner et al., 1975; Boso et al., 1981). A first step in the Mössbauer study of any new sample is to check its iron content, its purity, homogeneity and reproducibility. As shown in Fig. 2 (top), the spectrum of native RC consists of a single quadrupole doublet, i.e. the sample appears to be pure. Based on a calibration with a known absorber, the total area of matches within 10% the of estimated from the optical absorption with the assumption that 90% of the iron comes from the enriched iron in the growth medium. The linewidth, is larger than the minimum experimental linewidth, but is smaller than that observed in ferrous heme

365

proteins. The homogeneity of the iron site is therefore quite high. The combination of and finally, leaves no doubt that the iron is high spin ferrous. The good match of the isomer shift with the values found in compounds of spin S = 2 suggested mixed N- and O-ligands as borne out by x-ray diffraction. Figure 1 illustrates the typical temperature dependence of this high-spin ferrous complex; the lines (A), (B), and (C) represent three attempts to fit using a crystal field model. The upper inset defines the axial and rhombic crystal field terms Dl and D2 and sketches the effect of the spin–orbit coupling which is given in terms of the standard value and an adjustable covalency factor Curves (A) and (B) further assume a lattice or covalency contribution to the EFG, quantified by whereas curve (C) assumes different delocalization for the spin and the charge densities. Obviously, these four-parameter models fit the data reasonably well, yet the strong covalency of the 3d-electrons that all of them require implies that the crystal field approach is a poor approximation. Figure 2 (bottom) shows the magnetic hyperfine broadening brought about by a strong field and a simulation of the spectrum based on eqs. and 15. At a temperature of 156K the limit of fast spin fluctuation rates applies, and the thermal average, of the spin expectation can therefore be used in Eq. 11. The spectrum is from a series of measurements taken at different temperatures and fields and illustrates the type of information that can be extracted, which is summarized by the parameters given in the caption. The main conclusion from the data is that the iron site definitely has low symmetry. The hyperfine tensor is highly anisotropic, and its average value of is well below the Fermi contact term of indicating substantial orbital and spin dipolar contributions. Moreover, the quadrupole tensor with asymmetry is rotated relative to the zero-field splitting which was assumed to be coaxial with Ã. As mentioned earlier, the Hamiltonian parameters must be thermal averages rather than constants since Fig. 1 clearly indicates that higher orbital

366

states are being populated over the temperature range of interest. Figure 3, finally, compares the temperature dependences of the isomer shifts of six RC samples with nominally zero, one or two quinones, prepared with different detergents, with and o-phenanthroline, or with chemically reduced as verified by EPR. By and large is identical for all samples, and the temperature dependence is given by the same thermal redshift, Eq. 18. From this result and from the observation of nearly identical (not shown) it clearly follows that the ligands of the iron are the same in all samples, hence neither quinone nor o-phenanthroline bind to the iron. The only discrepancy in Fig. 3 is the 4.2K data point of the semiquinone sample (6). At higher temperatures this sample has 5–10% larger linewidths than all the others, but at 4.2K the quadrupole lines broaden asymmetrically by roughly a factor 1.7, a clear indication of unresolved magnetic hyperfine interaction. The explanation is obviously the magnetic coupling of the iron spin, S = 2, to the spin S = 1/2 of the semiquinone, a coupling that can be modeled by Eq. 16. The interaction of the two spins manifests itself not only in spontaneous broadening of the Mössbauer lines in zero field, but also in the large, temperature-dependent broadening of the EPR spectrum. The latter has been modeled quantitatively by Butler et al. (1984) using the spin Hamiltonian, Eq. 15 of the iron. Next, we turn to the work of Petrouleas and Diner (1982,1986,1990) on PSII of Chlamydomonas reinhardtii which established the existence of an iron–quinone complex in PSII of green plants. The complex appears to be structurally related to the bacterial one (Michel and Deisenhofer, 1988). Fig. 5 compares the Mössbauer spectra of PSII particles taken at 130K (a) and 4.2K (b,c). Obviously, the absorption is much smaller than in Fig. 2, the spectrum is more complex, and one component, an asymmetric doublet labeled (II) in the top trace, apparently vanishes at 4.2K. The major persistent component (I) has parameters characteristic of high-spin Fe(II), and the comparison in Table 1 suggests its analogy with the iron–quinone complex of the bacterial RC. Component (III) may arise from air-oxidized

Peter G. Debrunner

iron–quinone or some unidentified impurity. The properties of doublet (II), which are listed in Table 3, indicate that it is due to ferricytochrome: Magnetic hyperfine splitting broadens it beyond the detection limit at 4.2K, but partial collapse of the magnetic splitting leaves an asymmetric doublet with at 130K. This assignment is confirmed by optical difference

Mössbauer spectroscopy

measurements between reduced and oxidized samples and by EPR. In two subsequent papers, Petrouleas and Diner showed that the Fe(II) of the iron–quinone complex can be reversibly oxidized and identified the Fe(III)/Fe(II) couple with the high-potential electron acceptor of PSII first described by Ikegami and Katoh (1973). Here, only the Mössbauer data will be discussed while all the supporting evidence will be skipped. Figure 6 shows spectra taken at different redox potentials of membranes from a mutant lacking PSI and the cytochrome complex (Diner and Wollman, 1980). Spectrum (a) of the untreated sample is comparable to that of Fig. 5a with the exception of a narrow quadrupole doublet assigned to ferrocytochrome. The spectrum of Fig. 6b was obtained after oxidation of the sample with ferricyanide to a potential of 450 mV. The high-spin Fe(II) doublet of the iron–quinone complex has disappeared completely, and a broad, ill-defined absorption in the range of is seen, comprising contributions from ferricytochrome, ferro-ferricyanide and possibly from highspin Fe(III) with partially collapsed magnetic hyperfine splitting. On reduction of this sample with ascorbate to a potential of 300 ± 20 mV, Fig. 6c, the high-spin Fe(II) doublet of the iron–quinone complex reappears, but the spectrum differs from the original one because of the peak near due to ferrocyanide. Similar results were obtained by Picorel et al. (1994) with PSII isolated from the cyanobacterium Phormidium laminosum. Figure 7 shows Mössbauer spectra obtained from oxygen-evolving core complexes at 77K (a) as isolated, (b)

367

after oxidation with ferricyanide, and (c) after removal of the ferricyanide and reduction by dithionite. Spectra (a) and (c) are indistinguishable and show two quadrupole doublets, one from the high-spin ferrous iron–quinone complex, the other from the cytochromes and where the latter is an impurity. The rather symmetrical spectrum obtained on oxidation is a superposition of the ferricytochrome doublet and a broad highspin Fe(III) line indicative of relatively fast spin fluctuations. Another important finding of Diner and Petrouleas (1987) was the observation that the quadrupole splitting of the iron–quinone complex changes when bicarbonate is replaced by formate and vice versa as illustrated in Fig. 8. Semin et al. (1990) reported analogous results for PSII particles from Synechococcus elongatus. There must be a reversible change in the iron environment, but it is not clear yet whether either one of the compounds binds directly to the iron. Fig. 9, finally, shows the effect of NO binding on the Mössbauer spectrum of the iron–quinone complex (Petrouleas and Diner, 1990; Diner and Petrouleas, 1990). The intensity of the S = 2 doublet decreases on treatment with NO, and the lines at and grow instead. The authors do not provide a quantitative analysis, but it is clear that the isomer shift and the quadrupole splitting of the adduct are substantially smaller than for the S = 2 state. EPR shows that the Fe(II) NO complex has spin S = 3/2, and Mössbauer as well as EPR data are therefore analogous to those obtained for NO complexes of Fe(II) EDTA and dioxygenases (Arciero et al., 1983).

368

Peter G. Debrunner

B. Iron–sulfur Proteins EPR and MS have long been the major spectroscopic probes for the study of iron–sulfur proteins (Cammack et al., 1977), and PSI is no exception. EPR titrations identified three distinct but overlapping signals and that were assigned to iron–sulfur centers in PSI (Malkin and Bearden 1971). Evans et al. (1977, 1979, 1981) did the first Mössbauer studies on oxidized and reduced PSI samples and concluded that all three centers were 4Fe–4S clusters. Since the low-potential center remained controversial, having been assigned to a 2Fe–2S cluster as well (Bertrand et al., 1988), Petrouleas et al. (1989) examined PSI lacking and (Parrett et al., 1989) and confirmed it to be a 4Fe–4S cluster. The iron–sulfur clusters of iron–sulfur proteins are built from the same structural unit, namely a high-spin iron coordinated to four sulfurs, where the sulfurs are either cysteine or bridging

Mössbauer spectroscopy

369

depending on how many of these units are joined together. The high covalency of the complexes leads to characteristically small isomer shifts and A-tensors that set them apart from other iron environments. As discussed in Section II.D, the magnetic properties of larger clusters are dominated by the strong exchange interactions, Eq. 16, and for clusters with more than two irons by double exchange leading to delocalized valence pairs. All these properties have been studied extensively in small, well defined iron proteins as well as in model complexes, and the Mössbauer studies of PSI have revealed no new features apart from the unusually low redox potential of the center. Accordingly, the discussion that follows will be kept brief. To consider the simplest system first, it begins with the

370

mutant containing only and presents the complete PSI later. Figure 10 shows the Mössbauer spectra obtained by Petrouleas et al., (1989) from PSI core protein of Synechococcus 6301, a mutant of this cyanobacterium that lacks the two centers and but still contains The two spectra of the oxidized cluster on the right consist of an asymmetric, broad doublet with at 77K, indicating that the four iron sites have similar, indistinguishable Mössbauer parameters. On cooling to 4.2K the spectrum just moves slightly to more positive velocities as expected from the second-order Doppler shift, Eq. 18, but it does not broaden magnetically. As shown in Table 2, the Mössbauer parameters match those of other clusters of spin S = 0. The oxidized protein is known to be EPR inactive, while the reduced protein with S = 1/2 shows the EPR signal that led to its discovery.

Peter G. Debrunner

The Mössbauer spectrum of the reduced protein on the left is marked ‘D’ after correction for a 15% admixture of the oxidized spectrum. It is still a broad doublet at 77K although with different parameters, At 4.2K, on the other hand, the spectrum broadens, extending from to and showing the features typical of magnetic hyperfine splitting in as will be discussed below. Figure 11, finally, shows the 4.2K-Mössbauer spectra of PSI from the cyanobacterium Chlorogloea fritschii (Evans et al., 1981). These samples contain all three iron–sulfur centers and were partially reduced to show the EPR spectrum of only in (a), of and in (b), and of

Mössbauer spectroscopy

and in (c). The resolution of the spectra into oxidized and reduced material is shown, based on the spin Hamiltonian parameters deduced by Middleton et al. (1978) for the 4Feferredoxin from Bacillus stearothermophilus. These parameters clearly fit all three spectra quite well, and there is every reason to believe that the spectral contributions of and are indistinguishable, i.e. that all three are generic 4Fe–4S centers. It should be recalled that the nominal Fe(II)Fe(III) pair in has spin S' = 9/2 and delocalized valence while the Fe(II)Fe(II) pair has spin S'' = 4 to give a net spin of S = 1/2. The two pairs have different hyperfine tensors Ã' and Ã'' and therefore contribute distinct Mössbauer spectra as seen in Fig. 11.

C. Cytochromes Cytochrome is an intrinsic part of PSII and therefore appears in the spectra of Figs. 4–9, generally as an undesirable component that overlaps the low-energy line of the iron-quinone doublet. The few Mössbauer parameters reported are summarized in Table 3. As a low-spin heme protein, reduced cytochrome is diamagnetic and shows a sharp quadrupole doublet in the

371

Mössbauer spectra at all temperatures. Oxidized on the other hand, has spin S = 1/2, is EPRactive and is expected to have a Mössbauer spectrum with wide magnetic hyperfine splitting at 4.2K, a spectrum that has collapsed near 80K to a quadrupole doublet with broad, asymmetric lines. Although cytochrome is of considerable interest, especially since it can exist in a highand a low-potential form associated, presumably, with a change in orientation of the axial histidine ligands, any meaningful Mössbauer studies would have to be done on isolated cytochrome rather than on PSII preparations to achieve adequate signal-to-noise ratio. Since most of the work has been done by optical techniques and EPR, which are more sensitive (e.g. Babcock et al., 1985), the interested reader is referred to a Mössbauer study of a cytochrome b model (Walker et al., 1986) that deals with the question of the axial ligand orientation. References Abragam A and Bleaney B (1970) Electron Paramagnetic Resonance of Transition Ions. Oxford University Press. Aleksandrov AY, Novakova AA and Semin BK (1987) Mössbauer spectroscopy study of the conformational dynamics of native membrane proteins. Phys Lett 123: 151– 154.

372 Arciero DM, Lipscomb JD, Huynh BH, Kent T and Münck E (1983) EPR and Mössbauer studies of protocatechuate 4,5–dioxygenase. Characterization of a new environment. J Biol Chem 258: 14981–14991. Babcock GT, Widger WR, Cramer WA, Oertling WA and Metz JG (1985) Axial ligands of chloroplast cytochrome Identification and requirement of a heme-cross-linked polypeptide structure. Biochemistry 24: 3638–3645. Bertrand P, Guigliarelli B, Gayda J-P, Sétif P and Mathis P (1988) An interpretation of the peculiar magnetic properties of center X in Photosystem I in terms of a 2Fe–2S cluster. Biochim Biophys Acta 933: 393–397. Boso B, Debrunner P, Okamura MY, and Feher G (1981) Mössbauer spectroscopy studies of photosynthetic reaction centers from Rhodopseudomonas sphaeroides R-26. Biochim Biophys Acta 638: 173–177. Butler WF, Calvo R, Fredkin DR, Isaacson RA, Okamura MY and Feher G (1984) The electronic structure of in reaction centers from Rhodopseudomonas sphaeroides. III. EPR measurements of the reduced acceptor complex. Biophys J 45: 947–973. Cammack R, Dickson DPE and Johnson CE (1977) Evidence from Mössbauer spectroscopy and magnetic resonance on the active centers of the iron-sulfur proteins. In: Lovenberg W (ed) Iron–Sulfur Proteins Vol. III, pp. 283–330. Academic Press, New York. Cranshaw TE, Dale BW, Longworth GO and Johnson CE (1985) Mössbauer spectroscopy and its applications. Cambridge University Press, Cambridge. Debrunner PG (1989) Mössbauer Spectroscopy of Iron Porphyrins. In: Lever ABP and Gray HB (eds.) Physical Bioinorganic Chemistry Series. Iron Porphyrins Vol 3, pp. 137–234. VCH Publishers. Debrunner PG (1993) Mössbauer spectroscopy of iron proteins. In: Berliner LJ and Reuben J (eds) Biological Magnetic Resonance Vol. 13, pp. 59–101. Plenum Press, New York and London. Debrunner PG, Schulz CE, Feher G and Okamura MY (1975) Mössbauer study of reaction centers from R. sphaeroides . Biophys J 15: 226a. Debus RJ, Okamura MY, and Feher G (1985) Reconstitution of iron-depleted reaction centers from Rhodopseudomonas sphaeroides R-26 with Fe, Mn, Cu and Zn. Biophys J 47: 3a. Deisenhofer J, Epp O, Miki K, Huber R and Michel H (1985) Structure of the protein subunits in the photosynthetic reaction centre of Rhodopseudomonas viridis at 3Å resolution. Nature (London) 318: 618–624. Dickson DPE and Berry FJ (1986) Mössbauer spectroscopy. Cambridge University Press, Cambridge. Diner BA and Petrouleas V (1987) the non-heme iron of the Photosystem II iron–quinone complex. A spectroscopic probe of quinone and inhibitor binding to the reaction center. Biochim Biophys Acta 895: 107–125. Diner BA and Petrouleas V (1990) Formation by NO of nitrosyl adducts of redox components of the Photosystem II reaction center. II. Evidence that binds to the acceptor-side non-heme iron. Biochim Biophys Acta 1015: 141–149. Diner BA and Wollman F-A (1980) Isolation of highly active

Peter G. Debrunner Photosystem II particles from a mutant of Chlamydomonas reinhardtii. Eur J Biochem 110: 521–526. Evans EH, Carr NA, Rush JD and Johnson CE (1977) Identification of a non-magnetic iron centre and an iron-storage or transport material in blue-green algal membranes by Mössbauer spectroscopy. Biochem J 166: 547–551. Evans EH, Rush JD Johnson CE and Evans MCW (1979) Mössbauer spectra of Photosystem I reaction centres from the blue-green alga Chlorogloea fritschii. Biochem J 182: 861–865. Evans EH, Dickson PE, Johnson CE, Rush JD and Evans MCW (1981) Mössbauer spectroscopic studies of the nature of centre X of Photosystem I reaction centres from the cyanobacterium Chlorogloea fritschii. Eur J Biochem 118: 81–84. Feher G, Allen JP, Okamura MY and Rees DC (1989) Structure and function of bacterial photosynthetic reaction centres. Nature (London) 339: 111–116. Gibb TC (1976) Principles of Mössbauer spectroscopy. Chapman and Hall, London. Gonser U (1975) Mössbauer Spectroscopy. Springer-Verlag, New York. Greenwood NN and Gibb TC (1971) Mössbauer Spectroscopy. Chapman and Hall, London. Griffith JS (1957) Theory of electron resonance in ferrihaemoglobin azide. Nature (London) 180: 30–31. Gütlich P, Link R and Trautwein A (1978) Mössbauer Spectroscopy and Transition Metal Chemistry. Springer-Verlag, New York. Huynh BH and Kent TA (1984) Mössbauer studies of iron proteins. In: Eichhorn GL and Marzili LG (eds.) Advances in Inorganic Biochemistry, Vol. 6, pp. 163–223). Elsevier, Amsterdam. Ikegami I and Katoh S (1973) Studies on chlorophyll fluorescence in chloroplasts II. Effect of ferricyanide on the induction of fluorescence in the presence of 3–(3,4–dichlorophenyl)-1,1–dimethylurea. Plant Cell Physiol 14: 829– 836. Ingalls R (1964) Electric-field gradient tensor in ferrous compounds. Phys Rev 133: A781–A795. Keller H and Debrunner PG (1980) Evidence for coformational and diffusional mean square displacements in frozen aqueous solution of oxymyoglobin. Phys Rev Lett 45: 68– 71. Malkin R and Bearden AJ (1971) Primary reactions of photosynthesis. Photoreduction of a bound chloroplast ferredoxin at low temperature as detected by EPR spectroscopy. Proc Natl Acad Sci USA 68: 16–19. Michel H and Deisenhofer J (1988) Relevance of the photosynthetic reaction center from the purple bacteria to the structure of Photosystem II. Biochemistry 27: 1–7. Middleton P, Dickson DPE, Johnson CE and Rush JD (1978) Interpretation of the Mössbauer spectra of the four-iron ferredoxin from Bacillus stearothermophilus. Eur J Biochem 88: 135–141. Münck E, Papaefthymiou V, Surerus KK and Girerd JJ (1988) Double exchange in reduced clusters and novel clusters with In Metal Clusters in Proteins, Que L (ed.), ACS Symposium Series, Vol. 372, pp. 302–325, Am Chem Soc, Washington, D.C.

Mössbauer spectroscopy Parrett KG, Mehari T, Warren PG and Golbeck JH (1989) Purification and properties of the intact P-700 and taining Photosystem I core protein. Biochim Biophys Acta 973: 324–332. Petrouleas V and Diner BA (1982) Investigation of the iron components in photosystem II by Mössbauer spectroscopy. FEBS Lett 147: 111–114. Petrouleas V and Diner BA (1986) Identification of a high-potential electron acceptor of Photosystem II, with the iron of the quinone–acceptor complex. Biochim Biophys Acta 849: 264–275. Petrouleas V and Diner BA (1990) Formation by NO of nitrosyl adducts of redox components of the Photosystem II reaction center. I NO binds to the acceptor-side nonheme iron. Biochim Biophys Acta 1015: 131–140. Petrouleas V, Brand JJ, Parrett KG, and Golbeck JH (1989) A Mössbauer analysis of the low-potential iron–sulfur center in Photosystem I: Spectroscopic evidence that is a [4Fe–4S] cluster. Biochemistry 28: 8980–8983.

373 Picorel R, Williamson DL, Yruela I and Seibert M (1994) The state of iron in the oxygen-evolving core complex of the cyanobacterium Phormidium laminosum: Mössbauer spectroscopy. Biochim Biophys Acta 1184: 171–177. Schulz CE, Nyman P and Debrunner PG (1987) Spin fluctuations of paramagnetic iron centers in proteins and model complexes: Mössbauer and EPR results. J Chem Phys 87: 5077–5091. Semin BK, Loviagina ER, Aleksandrov AY, Kaurov YN and Novakova AA (1990) Effect of formate on Mössbauer parameters of the non-heme iron of PSII particles of cyanobacteria. FEBS Lett 270: 184–186. Taylor CPS (1977) The EPR of low spin heme complexes. Relation of the hole model to the directional properties of the g-tensor, and a new method for calculating the ligand field parameters. Biochim Biophys Acta 491: 137–149. Walker FA, Huynh BH, Scheidt WR and Osvath SR (1986) Models of the cytochrome b. Effect of axial ligand plane orientation on the EPR and Mössbauer spectra of low-spin ferrihemes. J Am Chem Soc 108: 5288–5297.

This page intentionally left blank

Chapter 23 Characterization of Photosynthetic Supramolecular Assemblies Using Small Angle Neutron Scattering† David 2M. Tiede1,* and P. Thiyagarajan2

1

Chemistry Division D-200 and Intense Pulsed Neutron Source, Argonne National Laboratory, Argonne, IL 60439, USA

Summary I. Introduction II. Small Angle Neutron Scattering A. Analysis of Scattering in the Very Small q Domain B. Scattering in the Intermediate q Domain III. SANS Studies of Photosynthetic Complexes A. Crystallization of Photosynthetic Proteins 1. Detergent Micelle Structures in Crystallization Conditions 2. Reaction Center Aggregation States B. Structural Characterization of Photosynthetic Supramolecular Assemblies C. Internal Structure in Supramolecular Assemblies D. New Results IV. Concluding Remarks Acknowledgements References

375 376 377 378 379 379 379 380 382 385 386 388 388 388 389

Summary

Small angle neutron scattering (SANS) offers opportunities for resolution of structure in molecular assemblies that can not be readily accessed by crystallography, such as inherently disordered assemblies, like micelles or vesicles, or multiple protein component complexes that can not be easily crystallized, like RC-cyt complexes or RC-antenna complexes. Scattering measurements on solution samples allow direct correlations to be made between structural features of Supramolecular assemblies and their spectroscopically determined function. Parameters which can be resolved by SANS include the size, shape, molecular weight, volume of macromolecules, internal packing for multiple component protein complexes. This information can be used to discriminate between possible molecular models for supramolecular structures. This chapter surveys possibilities for the application of this technique for the characterization of supramolecular assemblies in photosynthesis. SANS has been used to characterize the effect of ionic strength and detergents on reaction center aggregation. These measurements are being used to examine the pathways for reaction center crystallization. Applications are also presented for structural characterization of light-harvesting antenna complexes. Abbreviations: CMC – Critical micelle concentration; cyt – Cytochrome c; HT – Heptane-1,2,3–triol; LDAO – Lauryldimethylamine-N-oxide; LH – Light-harvesting complex; PEG – Polyethyleneglycol; RC – Reaction center; SANS – Small angle neutron scattering †

The US government’s right to retain a non-exclusive, royalty-free licence in and to any copyright is acknowledged. *Correspondence: Fax: 1-708-2529289, E-mail: [email protected]

375 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 375–390. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

376

I. Introduction Small angle neutron scattering (SANS) offers opportunities for the determination of solution structures of macromolecules with sizes in the range of 10Å to 500 Å which complements structural determination by other techniques. A distinguishing feature of neutron scattering compared to x-ray, electron and light scattering is that neutrons interact with atomic nuclei instead of electrons. This property allows neutron scattering to be sensitive to both light and heavy atoms in a structure, including protons which are typically not observed in x-ray and electron scattering. Table 1 shows coherent neutron scattering lengths, and incoherent scattering cross sections, for selected elements compared to their x-ray scattering amplitudes in the forward direction, f(0). The magnitudes of the coherent neutron scattering lengths for H and D are seen to be nearly comparable to those for other atoms found in biological molecules, while large differences are seen for x-ray scattering amplitudes which increase with increasing atomic number. This implies that H and D will make significant contributions to neutron scattering for biological molecules, while they will make relatively weak contributions for x-ray scattering signals. Another significant feature of neutron scattering that makes it suitable for probing biological structures is the relatively low energy of the neutron beam. At wavelengths compatible for resolving structure on the 10 Å to 500 Å scale, the

David M. Tiede and P. Thiyagarajan millivolt energy of the neutron beam is a million fold less than the kilovolt energy used in comparable x-ray scattering experiments, resulting in several orders of magnitude reduction in radiation damage. Furthermore, the penetration depth of the neutron beam allows neutron scattering analysis to be made on aqueous samples with 1 to 5 millimeter path lengths. This penetration depth allows SANS measurements to be made with samples similar to those encountered for the characterization of photosynthetic proteins by optical spectroscopies. This concurrence in sample constraints will permit the same sample to be analyzed by optical spectroscopy and SANS, enabling direct comparisons between photosynthetic function and structure. SANS is a widely applied scattering technique, and several reviews cover the application to biological systems (Jacrot, 1976; Stuhrmann and Miller, 1978; Chen, 1986; Feigin and Svergun, 1987). This technique allows extraction of form factors, which are descriptions of macromolecular size and shape, and particle–particle structure factors in solutions. This technique has also been demonstrated to provide structural information on supramolecular assemblies in their native state in aqueous solutions (Ramakrishnan et al., 1984). While form factors are a relatively low resolution image of a macromolecular assembly, this information can precisely define the distribution of aggregation states for proteins in solution, characterize protein packing in aggregates, and charac-

Small angle neutron scattering terize the dimensions of multiple component supramolecular assemblies such as protein-detergent complexes, or multiple component protein complexes. Unique opportunities also exist for the resolution of structure factors for selected components within complex mixtures by combining SANS measurements with isotopic substitution. This chapter will examine selected applications of SANS for resolution of molecular structures in photosynthesis.

377

characterization of the intraparticle structure factor. The intraparticle structure factor describes the scattering due to the atom position vectors within a particle: F(q) is the particle form factor, defined in terms of the atomic scattering lengths, and the atomic position vectors, with respect to the particle center of mass:

II. Small Angle Neutron Scattering Neutron scattering for proteins or other particles in solution is determined by the atomic composition and time-averaged position vectors of all nuclei in the sample. Neutron scattering is measured as a function of the scattering vector where is the neutron wavelength and is the angle of scattering. Persistent distance correlations between atoms in the sample produce variations in the scattered intensity due to interference effects of the scattered neutrons. Distance correlations between atoms will arise from the fixed atom positions within a particle, as well as due to the distribution of particles within the sample volume. As a result, neutron scattering, measured as the differential scattering cross section, can be written as the product of the number density of particles, n, the intraparticle structure factor, P(q) and the interparticle structure factor, S(q): S(q) makes a significant contribution to scattering under conditions in which the location of one particle affects the distribution of other particles around it. This correlation occurs in concentrated samples and in solutions of particles with surface charge (Chen, 1986; Chen et al., 1988). This effect can be exploited to investigate the nature of particle–particle interactions using SANS. In dilute solutions, or solutions of non-interacting particles, particle locations are not correlated, and S(q) approaches 1. Under these conditions, neutron scattering is determined exclusively by the number density of particles and their structure. Analysis of neutron scattering under these conditions yields the most accurate

The brackets in Eq. (2) indicate the average over all particle orientations. The scattered intensity can be written: where and are the position vectors of the i th and jth atoms within the particle, and and are the atomic scattering lengths. In general the scattering length of an atom, depends on the isotope and its spin states. The examples of scattering lengths listed in Table 1 are averages of scattering lengths taken over all spin states for specific isotopes. The distribution of spin states for a stable nucleus gives rise to two components to neutron scattering of a given atom, which are the coherent and incoherent scattering cross sections. Both contribute to scattering signals. For example H has a large incoherent cross section while D has very little. Hydrogen rich biological molecules produce significant amounts of incoherent scattering; however, if they are deuterated, the incoherent scattering cross sections can be substantially reduced. The differential scattering cross section of a particle in Eq. (4) thus has two terms, where

The incoherent scattering component does not depend on the position vectors of the atoms. Experimentally it is observed as a flat background signal uncorrelated to structure. The coherent sig-

378

nal, on the other hand, depends on the position correlations of the atoms in the particle and it is the basis for obtaining structural information from SANS. Equation (4) describes neutron scattering due to intraparticle atom distance correlations in vacuum, but does not take into account the effect of placing the particle in a solvent. In the small and intermediate angle scattering domains, atom– atom correlations are measured over distances that are large compared to individual atom bond lengths. In these domains, the solvent acts as a continuum with an average scattering length density, defined as the sum of the scattering lengths for solvent atoms contained within a unit volume. Similarly, the particle can be considered as composed of volume elements whose scattering length densities are determined by the atomic composition within each unit volume: This allows the scattering due to the atom–atom correlations with the particle to be expressed in terms of scattering length density of individual volume elements: The scattering from the particle is detected with respect to the scattering length density of the solvent:

David M. Tiede and P. Thiyagarajan ing sections, analysis of these signals permits determination of size, shape and structural organization of the molecular assemblies. Table 2 lists mean scattering length densities for molecules relevant to the solubilization and crystallization of isolated photosynthetic proteins. These scattering length densities are seen to fall between the scattering length densities of and This allows scattering from selected components within these mixtures to be minimized by adjusting the aqueous ratio so that the mean scattering length density of the solvent matches that of the selected component. Table 2 also shows the effectiveness of deuteration for the enhancement of scattering length densities of biological molecules. For example the average scattering length density of proteins typically increases from to upon complete deuteration. The scattering characteristics of a subset of components within a macromolecular assembly can be resolved if they can be selectively deuterated, and by recording the difference in scattering between unlabelled and labelled material. Elegant demonstrations of this technique include the resolution of phosphatidylcholine and bilesalt organization in rod-like mixed micelles (Hjelm et al., 1992), the resolution of subunit organizations in the 30S (Capel et al., 1987), and 50S (Nowotny et al., 1989) ribosomes and a RNA polymerase (Lederer et al., 1991).

A. Analysis of Scattering in the Very Small q Domain Hence, the difference in the scattering length densities of the particle and the solvent, termed contrast, dictates the intensity of the SANS signal. Equation 10 indicates that neutron scattering contrast can be varied either by changing the isotopic composition of the particle or of the solvent. This allows scattering due to selected volume elements to be minimized by matching their scattering length densities to that of the solvent, while offering a mechanism for maximizing the contribution of other volume elements by increasing their contrast. Thus, isotopic substitution, in combination with contrast matching, is a powerful approach for the identification of scattering for selected components within complex macromolecular assemblies, while minimizing chemical perturbation of the system. As outlined in the follow-

At sufficiently low q, Eq. 10 can be shown to reduce to a particularly simple form, termed the Guinier equation (Guinier and Fournet, 1955): The Guinier approximation is valid for the scattering region where the condition is met. From a plot of ln[I(q)] vs. the size parameter, and the forward scattering cross section, I(0), can be obtained. is the radius of gyration, which is defined as the root mean squared distance of all of the atoms to the centroid of the scattering length distribution. While does not directly resolve the macromolecular shape, the determination of by Guinier analysis is a sensitive index for monitoring relative macromolecular

379

Small angle neutron scattering

size and aggregation state. I(0) is the amplitude of the scattering measured at q = 0:

where n is the number of macromolecules per unit volume, and, and V are the average scattering length density and volume of the macromolecule respectively. If SANS data are obtained on an absolute scale, then I(0) can be used for obtaining the molecular weight and volume of the particles, provided the concentration of the particle and the scattering length densities of the solvent and the particles are known.

B. Scattering in the Intermediate q Domain Since the determination of by Guinier analysis does not give information on the particle shape, this information has to be extracted by fitting I(q) data with model form factors. One procedure is to fit experimental scattering data with I(q) profiles calculated using atomic coordinates based upon crystal structures (Feigin and Svergun, 1987; Glatter, 1991). This approach has been used extensively with x-ray scattering. For example, this procedure has been used to determine the extent to which crystal structures of proteins are compatible with their structure in solution (Fedorov and Denesyuk, 1978; Heidorn and Trewhella, 1988; Hubbard et al., 1988), and for the determination of the structure of protein aggregates in solution (Grossmann et al., 1993). For molecular assemblies for which there are no molecular models, such as with proteins or protein complexes that have not been crystallized, or with inherently disordered structures

like micelles or membrane structures, a second approach has been to fit experimental data with form factors calculated from geometric shapes. Methods have been developed which range from fitting with simple geometric shapes of constant scattering length density, to fitting with multipole expansions of a set of spherically harmonic shape functions (Guinier and Fournet, 1955; Stuhrmann and Miller, 1978; Feigin and Svergun, 1987; Glatter, 1991; Svergun, 1991). These procedures provide tools for analysis of dimensions, shapes, and internal structures of supramolecular assemblies based on SANS measurements. However, without the input of other structural information, SANS cannot generally determine a supramolecular form factor unambiguously. Instead, a set of distinct form factors, representing different supramolecular shapes and dimensions, may typically be found to reasonably fit experimental data recorded over a restricted q-range. Often this information can be combined with other physical or chemical data to resolve the most likely molecular structure. The following sections provide examples that illustrate how the SANS technique can contribute towards an understanding of the structural basis for function in photosynthetic supramolecular assemblies.

III. SANS Studies of Photosynthetic Complexes

A. Crystallization of Photosynthetic Proteins The crystallization of photosynthetic proteins is of crucial importance for investigation of photo-

380 synthetic mechanisms. However, there has only been limited success in the production of high quality crystals of photosynthetic membrane proteins. The reasons for success or failure in crystallization are not known, nor are the mechanisms for crystallization. Bacterial photosynthetic reaction centers provide a useful model for examining mechanisms for membrane protein crystallization. Reaction centers from two different species, Rhodopseudomonas viridis and Rhodobacter sphaeroides, have been successfully crystallized from detergent mixtures, and their molecular structures have been determined by x-ray diffraction (Miki et al., 1986; Allen et al., 1987; Yeates et al., 1987; Deisenhofer and Michel, 1989; Chang et al., 1991). So far, only two detergents have been found to yield crystals suitable for high resolution structural analysis. LDAO has been used successfully to produce high quality crystals of reaction centers, but only in conjunction with the addition of amphiphiles such as HT (Michel, 1982; Allen and Feher, 1990; Buchanan et al., 1993). Alternatively, OG has been used successfully in the absence of additional amphiphiles (Allen and Feher, 1984; Chang et al., 1985; Ducruix and Reiss-Husson, 1987). Crystallization in the presence of LDAO was accomplished with a variety of precipitants, including ammonium sulfate (Michel, 1982), potassium phosphate (Buchanan et al., 1993), and polyethylene glycol, PEG/sodium chloride mixtures (Allen and Feher, 1984; Allen and Feher, 1990). In contrast, successful crystallization in the presence of OG has only been reported using PEG/sodium chloride mixtures (Allen and Feher, 1984; Chang et al., 1985; Ducruix and Reiss-Husson, 1987; Franck et al., 1987). These crystallization studies suggest that LDAO, unlike OG, requires the addition of an amphiphile to permit crystallization, while LDAO is less sensitive to the chemical nature of the precipitant than is OG. The unique suitability of PEG/sodium chloride as a precipitant in the presence of OG is also suggested by a comparison of the crystallization of 21 water-soluble proteins in the presence of OG (McPherson et al., 1986). OG was found to be generally beneficial for crystallization when PEG/sodium chloride mixtures were used as the pre-

David M. Tiede and P. Thiyagarajan cipitant, but not with ammonium sulfate as a precipitant (McPherson et al., 1986). These crystallization studies indicate that the conditions required for reaction center crystallization are significantly regulated by detergent properties. Causes for the different requirements for crystallization in the presence of OG and LDAO have been suggested from SANS (Thiyagarajan and Tiede, 1994). These studies illustrate some of the capabilities of the SANS technique, and they are summarized below. 1. Detergent Micelle Structures in Crystallization Conditions Detergent micelle size, number density, and nature of inter-micelle interaction are all likely to be critical in determining micelle compatibility for crystallization. These properties can be directly accessed by SANS. SANS studies have identified clear differences in micelle structure and micelle– micelle interactions for the detergents OG and LDAO under conditions used for crystallization. For example, Fig. 1a and 1b show Guinier plots for 1% solutions of LDAO and OG respectively, as a function of NaCl concentration. For LDAO in the absence of NaCl, scattering intensity is seen to fall-off too quickly at low q. This is a clear indication of repulsive inter-micelle interactions (Chen, 1986; Timmins et al., 1991; Thiyagarajan and Tiede, 1994). Much stronger deviations of this type are seen for anionic detergents (Chen, 1986). For LDAO, this presumably reflects electrostatic interactions between micelles arising from the zwitterionic character of the molecule. Fig. 1a also shows that upon addition of 1M NaCl, the deviation from linearity is nearly completely removed, presumably due to dielectric screening. This data shows that in the absence of NaCl, LDAO micelles will experience appreciable electrostatic repulsion. This behavior can be contrasted with the nonionic OG micelle. In the absence of salt, the micelles were found to be non-interacting, as reflected by the linear plot throughout the range. In the case of OG solutions, Fig. 1 b, the addition of 1M NaCl caused a vertical shift of the plot, along with an upward deviation of the Guinier plot at low The upward deviation is a clear indication of a salt-induced aggregation of the

Small angle neutron scattering

OG micelle. This effect is the opposite of that seen with LDAO. The vertical shift is of interest since it reflects an increase in micelle number density. This data shows that the CMC of OG is much more strongly influenced by ionic strength than that of LDAO. The relative insensitivity of the observed slopes to ionic strength demonstrates that LDAO and OG micelle sizes are not strongly affected by ionic strength. However, these SANS data have shown that these detergents differ significantly in the nature of inter-micelle interaction and response to salt additions. Fitting of SANS profiles to model form factors has also resolved differences in the size and shape of LDAO and OG micelles. The LDAO micelle was found to be larger, and best fit with an ellipsoidal shape, while a spherical shape was found for the OG micelle (Thiyagarajan and Tiede, 1994). Using these procedures, a comparative study of OG and LDAO micelles under conditions used for reaction center crystallization has shown that the successful crystallization methods can be rationalized in terms of an optimization of micelle size, number density, and suppression of intermicelle interactions (Thiyagarajan and Tiede, 1994). LDAO and OG micelle characteristics in different solution conditions are schematically summarized in Figs. 2 and 3 respectively. These studies showed that the requirement for

381

HT for crystallization with LDAO as the solubilizing detergent can be understood from the beneficial effects that this reagent had on LDAO micelle size and inter-micelle interactions. As indicated in the top panel in Fig. 2, the LDAO micelle in the presence of NaCl and PEG is noninteracting, but retains the relatively large ellipsoidal shape that presumably interferes with crystallization. The addition of HT converts these micelles into smaller, spherical mixed-micelles. The bottom panel illustrates the finding that LDAO micelles are strongly associating in the presence of and also retain the ellipsoidal shape. Under these conditions, the addition of HT was found to have the remarkable effect of dispersing the LDAO aggregates into smaller, non-interacting, spherical mixed-micelles. The use of HT in crystallization mixtures with either PEG/NaCl mixtures or as protein precipitants result in the formation of HTLDAO mixed-micelles having similar characteristics. This property is consistent with the observed flexibility in the choice of precipitant in crystallization schemes employing HT-LDAO micelles. In the case of OG, both NaCl and additions were found to cause aggregation of micelles, as indicated in the lower panel in Fig. 3. Uniquely, the addition of PEG to OG in NaCl solutions dissociated these aggregates, resulting

382

in the formation of relatively small, spherical, non-interacting micelles that are compatible with crystallization. The effect of PEG appeared to be due to a direct molecular interaction between PEG and OG, as the presence of PEG was found to raise the CMC of OG. These SANS studies have shown that LDAO and OG micelles are fundamentally different in terms of size, shape, and nature of inter-micelle interactions in non-crystallizing conditions. However, these differences were found to be minimized under conditions used for protein crystallization. Under crystallization conditions both LDAO-HT mixed micelles and OG micelles were found to be spherical, possibly reflecting a flexible radius of curvature; they were small relative to the size of the protein (micelle radius 17 Å–23

David M. Tiede and P. Thiyagarajan

Å, reaction center dimensions 74 Å × 70 Å × 40 Å), and non-interacting. These results suggest that these shared micelle characteristics are necessary features required to permit successful protein crystallization. Interestingly, these studies also showed that even smaller spherical micelles could be produced by the formation HT-OG mixed micelles in PEG/NaCl mixtures, as indicated in the far-right panel in Fig. 3. These results suggest that conditions may be searched to further minimize possible micelle–micelle contacts in the crystalline lattice. 2. Reaction Center Aggregation States The physical characteristics of isolated detergent micelles can be expected to influence the solubility and aggregation behavior of corresponding protein-detergent complexes. We have used SANS to study the aggregation states and interparticle behavior for the RC solubilized by OG and LDAO as a function of NaCl. Striking dependencies of RC aggregation state on detergent and ionic strength were found. The aggregation behavior of the isolated RC can be understood from

Small angle neutron scattering

the ionic strength dependencies for inter-micelle interactions for these detergents. This work demonstrates the importance of the detergent micelle characteristics in determining the solubility of the corresponding detergent-protein complex. As an example, Fig. 4 shows neutron scattering intensity, I(q), as a function of q for two fully deuterated RC samples in 0.03% LDAO. The samples are matched in RC and detergent concentrations, but differ by the addition or absence of 2 M NaCl. The ratio was adjusted to 5% to contrast match the detergent. Under these conditions, the scattering can unambiguously be assigned as solely due to the RC. The figure shows that the scattering intensity is markedly enhanced for LDAO solubilized RCs in the absence of NaCl compared to that in the presence of 2 M NaCl. The solid lines are fits using the procedure described below. Qualitatively, the marked difference in scattering intensity between the two samples identifies a higher aggregation state for the reaction center in the low-ionicstrength solution. Calculations based upon the reaction center crystal structure showed that the scattering profile for reaction centers in 2 M NaCl, 0.03% LDAO

383

can be fit by a monomeric state of the reaction center. This sample also allowed a test of the fitting of SANS data with low-resolution geometrical form factors. Fig. 5a shows a fit to the scattering profile in 2 M NaCl using a cylindrical form factor that was determined from a maximum entropy search method (Hjelm et al., 1990; Morrison et al., 1992). The algorithm fits the scattering data by searching a predefined dimension space using a distribution of particles of all possible sizes. This method has proven to be particularly effective in fitting polydisperse systems. The output shows that the reaction center sample at high ionic strength in LDAO can be fit by a monodispersed distribution of cylindrical particles centered about a diameter of 68 Å and length of 60 Å. These dimensions correlate nicely with the average monomeric dimensions of 70 Å × 74 Å × 40 Å determined from the crystal structure. In contrast to a monomeric state of reaction centers in 2 M NaCl, 0.03% LDAO, fits to the SANS profile at low ionic strength requires a distribution of particles in two size ranges. A minor component is seen with dimensions consistent with monodispersed RCs. The major portion of the scattering is fit to a distribution of RC aggregates having a length of more than 500 Å, which falls outside the size-domain for the experimental q-range. The increase in the size of the aggregate along a single dimension suggests a linear aggregate. Similar measurements were also done with RCs in solutions of OG and 17% This ratio removes the scattering from OG micelles. Significantly, the aggregation state of the RC was also found to be ionic strength dependent with this detergent, but the ionic-strength effect was opposite to that measured in LDAO. In OG, RCs were found to be predominately monomeric at low ionic strength, but a linear aggregate was found as the predominant form at high ionic strength. By comparison to the ionic strength dependencies for LDAO and OG micelles described above, the SANS results for RCs demonstrate that a monodispersed state for the RC is only found under conditions in which detergent micelle–micelle interactions are removed. This result establishes the importance of micelle characteristics in

384

determining the solution behavior of the solubilized protein-detergent complex. This approach can be extended to examine the RC aggregation states in crystallization mixtures as the sample proceeds through the pre-saturated, metastable and labile saturated states. This investigation of structural intermediates in crystallization mixtures can lead to an identification of crystallization mechanism. We have begun such a characterization of the OG/PEG/NaCl crystallization method. As a first step, we have characterized the effect of the addition of PEG4000 at concentrations below the precipitation threshold for RCs solubilized by OG. Table 2 shows that the scattering length densities for OG and PEG are nearly equivalent, and that both are far from that of a fully deuterated protein. SANS data for the deuterated RCs were collected in 17% which provided a contrast match for OG and PEG4000. Residual scattering due to PEG4000 could be eliminated by subtraction of scattering profiles collected for appropriate reference samples in the absence of RCs. This method of data acquisition allowed SANS profiles for RCs to be exclusively detected without interference from either OG or PEG4000. Fig. 6a shows RC SANS profiles in the presence and absence of 10% PEG4000. In the absence of PEG4000 the scattering profile follows that ex-

David M. Tiede and P. Thiyagarajan

pected for monodispersed RCs. The scattering profile in the presence of 10% PEG4000 shows additional scattering at low q, corresponding to RC–RC correlations in aggregated states. A fit to the scattering profile with PEG4000 using the maximum entropy method is shown in Fig. 6b. The fit shows that the RCs are distributed between the monomeric state and a series of linear aggregated states. The fact that aggregation of OG micelles did not occur in the presence of PEG4000 shows that this effect is a property of the RC protein and not the detergent annulus. RC aggregation is observed with PEG4000 concentrations far from those required for RC precipitation and crystallization. It can be expected that as crystallization mixtures progressively decrease the solubility of the RC either through increasing the ionic strength with fixed PEG concentration, or by simultaneously increasing PEG and salt concentrations, the observed equilibrium between monomeric and aggregated RC states will be shifted towards the aggregate. The existence of this equilibrium has direct implications for the mechanism of RC crystallization. A prevalent view of protein crystallization assumes that protein monomers serve as intermediates in crystallization (Feher and Kam, 1985; Durbin and Feher, 1986; Durbin and Feher, 1991). This view assumes that crystal growth oc-

Small angle neutron scattering

curs by the addition of protein monomers to the crystalline array, and that nucleation arises from an equilibrium between protein monomers and a crystalline aggregate. In this mechanism, the observed RC aggregation will act as a competing pathway, and optimization of crystallization must minimize the participation of non-crystalline aggregation paths. However, an alternative mechanism is possible in which the observed aggregates themselves function as intermediates in crystallization. In this mechanism, the key step in crystallization is the conversion of structurally compatible, non-crystalline aggregates into crystalline ones for nucleation, and the incorporation of compatible aggregates into a crystalline lattice during crystal growth. The participation of aggregates as intermediates in crystallization has also been suggested from light scattering (Skouri et al., 1992; Thibault et al., 1992; Forsythe and Pusey, 1994) and SANS (Boue et al., 1993) studies of crystallization mixtures for the water-soluble protein lysozyme. The similarity between these results and those from the reaction center SANS studies raises the possibility that non-crystalline aggregates may be a general feautre of protein crystallization pathways.

385

B. Structural Characterization of Photosynthetic Supramolecular Assemblies In the SANS studies of RCs in solution, a matching of experimental scattering with that calculated from the RC crystal structure unambiguously identified the scattering profile associated with the monodispersed RC state. Knowledge of the crystallographic dimensions of the RC also permitted identification of the most appropriate fitting of the RC with simple geometric form factors. The most definitive characterization of a supramolecular structure by SANS is achieved by fitting experimental data with scattering profiles calculated for molecular assemblies built from the crystal structures (Grossmann et al., 1993). Ultimately this technique will be used to accurately characterize the structure of the RC aggregates and complexes of RC with other proteins. In the case of supramolecular assemblies for which there is no data on the structure of the constituents, the strength of SANS analysis lies in providing a test for plausible structural models. This capability can be illustrated by a SANS characterization of the light-harvesting antenna complexes of photosynthetic bacteria. The antenna of the purple bacteria consists of

386 two complexes, LH1 and LH2; for reviews see Hunter et al., (1989) and Zuber and Brunisholz, (1991). Each complex is composed of two short proteins, and containing approximately 50 amino acid residues. In LH1 each pair is associated with two bacteriochlorophyll molecules, while each pair is associated with three bacteriochlorophylls in LH2. Functional energy delocalization occurs throughout an array of these building blocks, involving up to 200 or more chlorophylls (Holzwarth, 1991; Pullerits et al., 1994). There is no definitive characterization of the structure of the antenna assemblies either in the natural membrane or in detergent-isolated states. The fundamental building block of the LH1 and LH2 complexes has been proposed to be either or units, with cylindrically symmetric complexes (Hunter et al., 1989; Zuber and Brunisholz, 1991) or an complex (Kleinekofort et al., 1992) functioning as the minimal functional unit. SANS studies can discriminate between these possible models without detailed fitting of scattering profiles to molecular models. For example, the two models for the minimal functional unit for LH2, and differ in molecular mass. Peptide and pigment composition analyses for the LH2 complex from Rb. sphaeroides (Zuber and Brunisholz, 1991) indicates that the molecular masses for these models are 88 kDa and 118 kDa respectively. Figure 7 shows a comparison of scattering profiles for deuterated Rb. sphaeroides LH2 and RC. Contributions from the solubilizing detergent, OG, were eliminated by contrast matching through solubilizing both samples in 17% solutions. Scattering profiles for LH2 were indistinguishable with either LDAO solutions or OG as the solubilizing detergent, and using 5% and 17% respectively to contrast match the detergent. Model fitting and Guinier analysis show that both samples are monodispersed. The scattering profiles in Fig. 7 were normalized at low q. This normalization illustrates that the falloff in scattering intensity starts at a lower q for LH2 than for RC. This unambiguously establishes that the LH2 assembly has a structure with longer atom-atom distance correlations than does the RC. This is also indicated by a comparison of Guinier analysis which yields an of 31±1.5

David M. Tiede and P. Thiyagarajan

Å and 38 ± 1.3 Å for the RC and LH2 complex respectively. These observations require that the LH2 complex have slightly larger size than the RC. Maximum entropy fitting procedures found that the scattering profiles for LH2 can be fit using a solid, cylindrical form factor having a particle of 68 Å diameter and length of 64 Å The dimensions and molecular mass of the proposed models (Hunter et al., 1989; Zuber and Brunisholz, 1991) are smaller than those for the RC. Thus, the SANS data rule out the complex as a possible structure for the LH2 in these samples. The molecular mass of the model is 1.17 fold larger than that of the RC. Form factors consistent with this molecular mass can be fit to the LH2 scattering data, and hence the model is more consistent with the SANS results. This example demonstrates how SANS data that resolves only the overall size and mass density of a supramolecular assembly can be used for identification of possible structural models.

C. Internal Structure in Supramolecular Assemblies In addition to a resolution of particle size and mass density, SANS measurements offer opportunities for resolution of internal structures in

Small angle neutron scattering

supramolecular assemblies. In the q-range up to this can be illustrated with SANS measurements for the LH1 complex. Figure 8 shows SANS profiles for the Rb. sphaeroides LH1 at two different concentrations. Maximum entropy analysis required dispersions of LH1 particle sizes to fit both profiles. Both samples were found to be composed of different mixtures of particles with diameters of approximately 60 Å and 130 Å. A dispersion of particle sizes, which varied as a function of LH1 concentration, was also found by size-exclusion chromatography and electron microscopy of LH1 (Boonstra et al., 1993). This behavior is reflected in the SANS data. The shoulder seen in the scattering profile for the more concentrated sample in Fig. 8 at approximately is noteworthy, since features of this kind have not been seen in any of the monodispered or aggregated RC samples, nor in the LH2 samples. Additionally, this feature was not fit using solid cylindrical form factors in the polydispersity fitting analysis. One intriguing possibility is that this feature arises from the internal structure, or packing of subunits within the larger LH1 particles, as suggested from the following modelling studies. Figure 9 shows scattering profiles calculated

387

for two model structures. The first was a cylindrically symmetric ring composed of six cylinders, having a 50 Å diameter and 60 Å length. Each cylinder represents a geometrical form factor approximation to the unit. This model has a hole in the middle with a size sufficient to contain a seventh cylinder. The scattering profile calculated for this model is shown by the solid line in Fig. 9. One noticeable feature is the shoulder in the q region to that arises from intra-particle interference effects. The shape of this feature is similar to that seen in the experimental data. The sensitivity of the shoulder to the presence of the hole in the model structure is illustrated by the scattering profile marked by the dashed line in Fig. 9, which was calculated for a model consisting of the same 6 cylinder ring, but that additionally contained a seventh cylinder in the center. This filled ring model removed the shoulder seen in the scattering profile in the q region to Fits to the experimental data will require optimization of the model as well as inclusion of effects due to aggregate size dispersity. However, these calculations demonstrate the plausibility of a ring-like assembly to give rise to the observed scattering fine structure for LH1. Hence, the SANS data support structural models for the LH1 at high concentration that incorporate ring-like structures. The ability to detect internal structure within

388 the LH1 assembly at high concentration arises because of the large size of the aggregated assembly, with corresponding large structural repeats and scattering length density discontinuities. These results offer encouragement for the application of SANS measurements at higher q to probe structural repeats of smaller dimension, such as the internal packing of protein in smaller assemblies such as the LH2 and RC complexes.

D. New Results Since the first draft of this chapter, a high resolution, R = 2.5 Å, structure has been determined for the LH2 complex from Rps. acidophila by Xray diffraction on single crystals (McDermott et al., 1995), and a low-resolution, R = 8.5 Å, structure has been determined for the LH1 from Rhodospirillum by electron microscopy on two-dimensional crystals (Karrasch et al., 1995). The crystallography of LH2 showed the complex to be composed of 9 pairs, arranged symmetrically in a barrel-shaped bundle, with a 36 Å diameter hole in the center. The overall dimensions of the LH2 complex were 68 Å diameter by 70 Å length (McDermott et al., 1995). These dimensions are in approximate aggreement with the 68 Å by 64 Å dimensions determined by our fitting of the SANS data for the Rb. sphaeroides LH2 complex in solution. It is likely that the differences in measured dimensions arise from the use of cylindrical form factors as approximations of the protein shape for fitting the SANS data. Å more exact comparison of the crystal and solution structures for LH2 will require comparisons of experimental SANS profiles with those calculated from atomic crystal coordinates. As discussed for the LH1 complex, a hollow or ringlike structure will give rise to characteristic undulations in SANS profiles due to the cylindrical symmetry. Undulations due to a 36 Å diameter hole in the LH2 structure would not be seen in the present SANS profiles recorded in the q-range These features would be expected to appear at higher q-values. The two-dimensional crystals of the R. rubrum LH1 showed the complex to be composed of 16 pairs arranged, on average, in a 116 Å ring with a 68 Å hole in the middle (Karrasch et al.,

David M. Tiede and P. Thiyagarajan 1995). A ring-like structure was also anticipated from SANS data for the Rb. sphaeroides LH1 in solution, (Fig. 8), which has a secondary maximum at Preliminary modelling suggests that the ring-like structure described by Karrasch et al. can account for the shoulder seen in the scattering profile for the Rb. sphaeroides LH1. However, a detailed modelling of the solution scattering profiles will require that particle size dispersity be taken into account. The electron microscopy of the two-dimensional LH1 arrays showed that there were small ellipitical distortions in the ring shape in different crystals, and that there were different crystalline forms composed of LH1 particles with different dimensions (Karrasch, S. et al., 1995). A detailed evaluation of the solution SANS data in light of the new structured information is underway. IV. Concluding Remarks SANS studies on photosynthetic complexes demonstrate the capability of this technique for resolving inter-particle interactions, size, shape and internal order or packing of supramolecular assemblies pertinent to photosynthesis. Opportunities for definitive assignment of supramolecular structures are possible by comparison of experimental scattering data with scattering profiles calculated for molecular models built from crystal structures of the composite proteins. Photosynthesis ultimately relies upon a hierarchy of structures and intermolecular interactions. SANS provides a complement to crystallographic studies by providing a technique for assessing structure in functional, supramolecular assemblies that can not be examined by crystallography. Acknowledgements The authors thank Dr. Rex Hjelm (Los Alamos National Laboratory) and Dr. D. S. Sivia (ISIS) for use of the maximum entropy fitting program used to characterize the size distributions of photosynthetic complexes, and Dr. Stephen Henderson (Oak Ridge National Laboratory) for the use of his program BIOMOD for the calculation of scattering profiles for model structures in Fig. 9. This work was supported by the U.S Department of Energy, Office of Basic Energy Sciences, Divi-

Small angle neutron scattering sion of Chemical Sciences and the Division of Material Sciences, under Contract W-31–109– Eng-38 and P.T. was additionally supported in part, by a NASA Microgravity Biotechnology Program Grant M951–ES-3–004–2511. References Allen JP and Feher G (1984) Crystallization of reaction center from Rhodopseudomonas sphaeroides: preliminary characterization. Proc Natl Acad Sci USA 81: 4795–4799. Allen JP and Feher G (1990) Crystallization of reaction centers from Rhodobacter sphaeroides. In: Michel H (ed) Crystallization of Membrane Proteins, pp 137–154. CRC Press, Boca Raton. Allen JP, Feher G, Yeates TO, Komiya H and Rees DC (1987) Structure of the reaction center from Rhodobacter sphaeroides R-26: The protein subunits. Proc Natl Acad Sci. USA 84: 6162–6166. Boonstra AF, Visschers RW, Calkoen F, van Grondelle R, van Bruggen EFJ and Boekema EJ (1993) Structural characterization of the B800–850 and B875 light-harvesting antenna complexes from Rhodobacter sphaeroides by electron microscopy. Biochim. Biophys Acta 1142: 181–188. Boue F, Lefaucheux F, Robert MC and Rosenman I (1993) Small angle neutron scattering study of lysozyme solutions. J Crystal Growth 133: 246–254. Buchanan SK, Fritzsch G, Ermler U and Michel H (1993) New crystal form of the photosynthetic reaction centre from Rhodobacter sphaeroides of improved diffraction quality. J Mol Biol 230: 1311–1314. Capel MS, Engelman DM, Freeborn BR, Kjeldgaard M, Langer JA, Ramakrishnan V, Schindler DG, Schneider DK, Schoenborn BP, Sillers I-Y, Yabuki S and Moore PB (1987) A complete mapping of the proteins in the small ribosomal subunit of Escherichia coli. Science 238: 1403– 1406. Chang C-H, Schiffer M, Tiede DM, Smith U and Norris JR (1985) Structure of the membrane-bound protein photosynthetic reaction center from Rhodopseudomonas sphaeroides R-26 by x-ray diffraction. J. Mol. Biol. 186: 201–203. Chang C-H, El-Kabbani O, Tiede DM, Norris J and Schiffer M (1991) Structure of the membrane-bound protein photosynthetic reaction center from Rhodobacter sphaeroides. Biochemistry 30: 5352-5360. Chen SH (1986) Small angle neutron scattering studies of the structure and interaction in micellar and microemulsion systems. Annu Rev Phys Chem 37: 351–399. Chen SH, Sheu EY, Kalus J and Hoffman H (1988) Smallangle neutron scattering investigation of correlations in charged macromolecular and supramolecular solutions. J Appl Cryst 21: 751-769. Deisenhofer J and Michel H (1989) The photosynthetic reaction center from the purple bacterium Rhodopseudomonas viridis (Noble Lecture). Angew Chem Int Ed Engl 28: 829– 968. Ducruix A and Reiss-Husson F (1987) Preliminary characterization by x-ray diffraction of crystals of photochemical

389 reaction centres from wild-type Rhodopseudomonas sphaeroides. J Mol Biol 193: 419–421. Durbin SD and Feher G (1986) Crystal growth studies of lysozyme as a model for protein crystallization. J Crystal Growth 76: 583–592. Durbin SD and Feher G (1991) Simulation of lysozyme crystal growth by the Monte Carlo method. J Crystal Growth 110: 41–51. Fedorov BA and Denesyuk AI (1978) Large-angle x-ray diffuse scattering, a new method for investigating changes in the conformation of globular proteins in solution. J Appl Cryst 11: 473-477. Feher G and Kam Z (1985) Nucleation and growth of protein crystals: general principles and assays. Methods Enzymology 114: 77–112. Feigin LA and Svergun DI (1987) Structure Analysis by Small Angle X-ray and Neutron Scattering 1–335. Forsythe E and Pusey ML (1994) The effects of temperature and NaCl concentration on tetragonal lysozyme face growth rates. J Crystal Growth 139: 89–94. Franck HA, Taremi SS and Knox JR (1987) Crystallization and preliminary x-ray and optical spectroscopic characterization of the photochemical reaction center from Rhodopseudomonas sphaeroides strain 2.4.1. J Mol Biol 198:139– 141. Glatter O (1991) Small-angle scattering and light scattering. In: Lindner P and Zemb T (ed) Neutron, X-ray and Light Scattering, pp 33–82. Elsevier, Amsterdam. Grossmann JG, Abraham ZHL, Adman ET, Neu M, Eady RR, Smith BE and Hasnain SS (1993) X-ray scattering using synchrotron radiation shows nitrite reductase from Achromobacter xylosoxidans to be a trimer in solution. Biochemistry 32: 7360–7366. Guinier A and Fournet G (1955) Small Angle Scattering. John Wiley and Sons, New York. Heidorn DB and Trewhella J (1988) Comparison of the crystal and solution structures of calmodulin and troponin c. Biochemistry 27: 909–915. Hjelm RP, Thiyagarajan P, Sivia DS, Lindner P, Alkan HA and Schwahn D (1990) Small-angle neutron scattering from aqueous mixed colloids of lecithin and bile salts. Prog Colloid Polym Sci 81: 225–321. Hjelm RP, Thiyagarajan P and Alkan-Onyuksel H (1992) Organization of phosphatidylcholine and bile salt in rodlike mixed micelles. J Phys Chem 96: 8653–8661. Holzwarth AR (1991) Excited state kinetics in chlorophyll systems and its relationship to the functional organization of the photosystems. In: Scheer H (ed) Chlorophylls, pp 1125–152. CRC Press, Boca Raton. Hubbard ST, Hodgson KO and Doniach S (1988) Small-angle x-ray scattering investigation of the solution structure of troponin c. J Biol Chem 263: 4151–4158. Hunter CN, van Grondelle R and Olsen JD (1989) Photosynthetic antenna proteins: 100 ps before photochemistry starts. Trends Bioch Sci 14: 72–76. Jacrot B (1976) The study of biological structures by neutron scattering from solution. Rep Prog Phys 39: 911–953. Karrasch S, Bullough PA and Ghosh R (1995) The 8.5 Å projection map of the light-harvesting complex I from Rho-

390 dospirillum rubrum reveals a ring composed of 16 subunits. EMBO J 14: 631–638. Kleinekofort W, Germeroth L, van der Broek JA, Schubert D and Michel H (1992) The light-harvesting complex II (B800/850) from Rhodospirillum molischianum is an octamer. Biochim Biophys Acta 1140: 102–104. Lederer H, Mortensen K, May RP, Baer G, Crespi H, Dersch D and Heumann H (1991) Spatial arrangement of and core enzyme of Escherichia coli RNA polymerase: A neutron solution scattering study. J Mol Biol 219: 747–755. McDermott G, Prince SM, Freer AA, Hawthornthwaith-Lawless AM, Papiz MZ, Cogdell RJ and Isaacs NW (1995) Crystal structure of an integral membrane light-harvesting complex from photosynthetic bacteria. Nature 374: 517– 521. McPherson A, Koszelak S, Axelrod H, Day J, Williams R, Robinson L, McGrath M and Cascio D (1986) An experiment regarding crystallization of soluble proteins in the presence of J Biol Chem 261: 1969–1975. Michel H (1982) Three-dimensional crystals of a membrane protein complex. The photosynthetic reaction centre from Rhodopseudomonas viridis. J Mol Biol 158: 567–572. Miki K, Saeda M, Masaki K, Kasai N, Miki M and Hayashi K (1986) Crystallization and preliminary x-ray diffraction study of ferrocytochrome from Rhodopseudomonas viridis. J Mol Biol 191: 579–580. Morrison JD, Corcoran JD and Lewis KE (1992) The determination of particle size distributions in small-angle scattering using the maximum-entropy method. J Appl Cryst 25: 504– 513. Nowotny V, Nowotny P, Voss H, Nierhaus KH and May RP (1989) The quaternary structure of the ribosome from Escherichia coli- A neutron small-angle scattering study. Physica B 156: 499–501. Pullerits T, Visscher KJ, Hess S, Sundstrom V, Freiberg A, Timpmann K and R. vG (1994) Energy transfer in the inhomogeneously broadened core antenna of purple bacteria: A simultaneous fit of low-intensity picosecond absorption and fluorescence kinetics. Biophys J 66: 236–248.

David M. Tiede and P. Thiyagarajan Ramakrishnan VR, Capel M, Kjeldgaard M, Engleman DM and Moore PB (1984) Position of protein S14, protein S18 and protein S20 in the 30S ribosomal subunit of Escherichia coli. J Mol Biol 174: 265–284. Sears VF (1986) Neutron scattering lengths and cross sections. In: Sköld K and Price DL (ed) Neutron Scattering. Methods of experimental physics, Celotta R and Levine J (series eds) 23A, pp 521–549. Academic Press, New York. Skouri M, Munch J-P, Lorber B, Giege R and Candau S (1992) Interactions between lysozyme molecules under precrystallization conditions studied by light scattering. J Crystal Growth 122: 14–20. Stuhrmann HB and Miller A (1978) Small-angle scattering of biological structures. J Appl Cryst 11: 325–345. Svergun DI (1991) General theorems of small-angle scattering by disperse systems. In: Lindner P and Zemb T (ed) Neutron, X-ray and Light Scattering, pp 83–98. North-Holland, Amsterdam. Thibault F, Langowski J and Leberman R (1992) Optimizing protein crystallization by aggregate size distribution analysis using dynamic light scattering. J Crystal Growth 122: 50–59. Thiyagarajan P and Tiede DM (1994) Detergent micelle structure and micelle-micelle interactions determined by small angle neutron scattering in solution conditions used for membrane protein crystallization. J Phys Chem 98: 10343– 10351. Timmins PA, Hauk J, Wacker T and Welte W (1991) The influence of heptane-1,2,3–triol on the size and shape of LDAO micelles. FEBS Letts 280: 115–120. Yeates TO, Komiya H, Rees DC, Allen JP and Feher G (1987) Structure of the reaction center from Rhodobacter sphaeroides R-26: Membrane-protein interactions. Proc Natl Acad Sci USA 84: 6438–6442. Zuber H and Brunisholz RA (1991) Structure and function of antenna polypeptides and chlorophyll-protein complexes: Principles and variability. In: Scheer H (ed) Chlorophylls, pp 627–703. CRC Press, Boca Raton.

Chapter 24 Measurement of Photosynthetic Oxygen Evolution Hans J. van Gorkom* and Peter Gast Department of Biophysics, Huygens Laboratory of the State University, P.O.Box 9504, 2300 RA Leiden, The Netherlands

Summary I. Introduction II. Polarography A. The Clark Electrode B. Unwanted Chemistry at the Bare Cathode C. Rate Electrodes D. Concentration Electrodes E. Polarograms F. Flash-Induced Kinetics III. EPR Oximetry A. EPR Line Broadening by Oxygen B. Oxygen Probes for EPR C. Sensitivity 1. Sensitivity in Direct Linewidth Broadening Measurement 2. Sensitivity in Amplitude Measurement D. Applications in Photosynthesis Research IV. Mass Spectrometry V. Photoacoustic Spectroscopy VI. Galvanic Sensors VII. Prospects Acknowledgements References

391 392 392 392 393 394 395 395 397 398 398 398 399 399 399 401 401 402 402 402 403 403

Summary An overview is presented of methods that have been used to measure photosynthetic oxygen evolution over the past 25 years. Oxygen polarography in its many versions is treated in some detail, the complications caused by a large bare cathode are discussed and an interpretation of the current/voltage characteristic (polarogram) of flash-induced oxygen signals is proposed. The recent controversy on the interpretation of the kinetics of such signals is briefly summarized. The discovery of a new class of spinprobes for EPR oximetry has greatly enhanced its possibilities. The sensitivity of the method is evaluated, consequences of its nonlinearity in time-resolved measurements are indicated, and the first reports on its use in photosynthesis research are summarized. The use of mass spectrometry, photoacoustic spectroscopy and galvanic sensors to measure photosynthetic oxygen evolution is briefly reviewed.

*Correspondence: Fax: 31-71-5275819; E-mail: [email protected]

391 J. Amesz and A. J. Hoff (eds.), Biophysical Techniques in Photosynthesis, pp. 391–405. © 1996 Kluwer Academic Publishers. Printed in the Netherlands.

392

Hans J. van Gorkom and Peter Gast

I. Introduction

II.

Polarography

Oxygen evolution was the first photosynthetic reaction discovered and two centuries of research have left an impressive track record of methods applied to measure it (Rabinowitch, 1945, 1951, 1956; Burr and Mauzerall, 1968; Joliot, 1993). It is useful to keep old methods in mind. A striking example was the recent application of the classical work of T. Engelmann in the 1880's, who used bacteria with oxotropic motility to locate the source of photosynthetic oxygen at the microscopic level: this method allowed V. Zimmermann's group to select the photosynthetically competent cells among hybrids produced by electrofusion (Hampp et al., 1986). Due to its important medical and industrial applications, the measurement of oxygen has received much attention. A wide range of methods has been reviewed in a symposium dedicated to the subject (Degn et al., 1976). In biological and medical research, polarography has become the predominant technique, on which extensive documentation is available (Fatt, 1976; Hitchman, 1978; Gnaiger and Forstner, 1983). Also in photosynthesis research polarography has replaced the traditional Warburg manometry, a variety of electrochemical cells has been designed and other ways of measuring oxygen have been explored. Here we present a survey of such methods as published over the past 25 years or so, after the discovery of the period four oscillation of the oxygen yield with flash number upon illumination of Photosystem II with a series of short saturating flashes (Joliot et al., 1969). This finding marked a shift of scientific interest from rate measurements in continuous light to the measurement of oxygen yields of individual flashes in a series. In recent years also the kinetics of oxygen release after a flash have been at the focus of attention, requiring measurements with even higher time resolution. In this chapter we concentrate on methods meeting these requirements and, due to our own involvement, pay some extra attention to the interpretation of time-resolved polarographic signals and to a promising new class of spin-probes for EPR-oximetry.

A. The Clark Electrode Routine measurements of photosynthetic oxygen evolution are now carried out with commercially available equipment based on the Clark electrode (Clark, 1956). This is a ‘bipolar' electrochemical oxygen sensor: the cathode and anode and the connecting electrolyte solution are separated from the sample by a teflon membrane, which is permeable to oxygen but not to water and ions. The platinum cathode reduces oxygen to hydroxyl ions. At the silver anode initially AgCl is formed and later, as accumulates, also AgOH. Apart from its small-oxygen consumption it is essentially a closed system, isolated from the sample by the teflon membrane, and can be used to measure oxygen in the gas phase (Delieu and Walker, 1983) as well as in solution. The use of a small cathode and stirring of the sample (if in solution) ensure that the oxygen concentration gradient is restricted to the membrane and the use of a sufficiently low cathode potential, about –0.7 V relative to the anode, ensures that this gradient stays at its maximum value because the cathode surface is kept anaerobic. The measured current is simply proportional to the oxygen tension (partial pressure) in the sample and is easily calibrated by comparison to the signal in an airsaturated solution at the same temperature. The oxygen concentration then follows from the oxygen solubility in the medium used (which can be much reduced at high salt concentrations). The properties and theoretical background of the Clark electrode are described most comprehensively in the monograph by Hitchman (1978), which also contains oxygen solubility data. In photosynthesis research, situations often arise where the unmodified Clark electrode is too insensitive, too slow, or both. In particular, it does not allow resolution of the individual oxygen yields of successive flashes in a series, which is required to study the 4-flash redox cycle of the oxygen evolving complex. Sufficient sensitivity for this purpose may be reached by a substantial increase of the cathode surface area (Velthuys and Kok, 1978; Lübbers et al., 1993), maintaining the advantages of the physical separation by the teflon membrane between sample and electro-

Oxygen evolution chemistry. However, even with vigorous stirring it will be difficult to prevent the oxygen diffusion gradient from spreading into the the sample and the advantage of easy calibration is probably lost. With a thin membrane, the time-resolution may be good enough for single flash resolution at a usual 1 Hz flash frequency. A further enhancement of sensitivity and time-resolution can only be obtained by removing the membrane, sacrificing all advantages of the Clark electrode concept.

B. Unwanted Chemistry at the Bare Cathode When a bare cathode is used, the sample must contain the electrolyte connecting the cathode and anode and a buffer to avoid excessive pH increase due to production at the cathode, and should preferably not contain any substance that could mediate electrochemical reduction of redox centers in the sample. A polarized bare platinum electrode covered by a thin sample layer is the standard tool for electrochemical titrations and can readily impose its potential on all redox couples in the sample if adequate mediators are present. Fortunately, most redox centers in Photosystem II are highly inaccessible and the ‘substrate binding sites’, where plastoquinone reduction and water oxidation take place, seem to allow reaction with a very limited variety of molecules only. Even the mobile plastoquinone pool is not rapidly reduced by the cathode if no mediators are added. However, the cathode itself may produce redox mediators. Hydrogen peroxide, formed at the cathode surface as an intermediate in the reduction of to can interact with the oxygen evolving complex, especially in Photosystem II preparations where the protective shield of extrinsic polypeptides has been damaged (Schröder and Åkerlund, 1986). Addition of catalase might help to avoid this complication, but one should keep in mind that there is also evidence for the production of hydrogen peroxide by Photosystem II itself under some circumstances (Wydrzynski et al., 1989). The inhibition observed by Plijter et al. (1988) at cathode potentials lower than –0.5 V vs SHE (Standard Hydrogen Electrode) may be due to hydrogen produced at the cathode surface. As the inhibition appears to be an irreversible

393 all-or-none effect (the shape of the period four oscillation in a flash series remains the same), this might be due to over-reduction and subsequent dissociation from the reaction center of or manganese. This problem can be avoided by weaker polarization, and depends on pH and cathode material, Pt being a much more efficient catalyst for hydrogen production than Au. Last but not least, the cathode removes oxygen and oxygen usually is involved in poising the redox potential in the sample. There have been reports that oxygen evolution requires oxygen (Bader and Schmid, 1988). The oxygen concentration near the surface of a strongly polarized cathode surface can become quite low (Baumgärtl et al., 1974) and it is primarily the oxygen evolution in this region that one measures. The modification of the chemical conditions in the sample by the cathode is normally limited by using a thin sample layer exposed on the other side to a continuous flow of conditioning medium (rate electrodes). It can be avoided altogether by a continuous, very fast sample replacement (Etienne, 1968) or minimized by using a very weak polarization (Plijter et al., 1988) (concentration electrodes). Moreover, the experiment may require the addition of chemicals, such as an artificial electron acceptor when isolated Photosystem II particles are used. Joliot et al. (1966) introduced modulated illumination and lock-in detection to measure selectively the electrode current resulting from photosynthetic activity (see also Joliot, 1972). Particularly troublesome, however, are substances which produce spurious signals upon flash illumination due to photochemistry in the sample or at the cathode surface. In some cases such signals can be avoided by applying a thin film of collodion on the cathode. With a flat disk electrode this can be done conveniently by letting a drop of collodion solution fall on the cathode. By choice of solvent, collodion concentration, drop size and speed, one can obtain reproducible films of the right diameter to cover the cathode and thin enough not to affect the kinetics of the flash induced oxygen signals. In one respect the measurement of oxygen evolution should be simplified by the use of a bare cathode. The cathode current is proportional to the oxygen concentration in a boundary layer adjacent to the cathode surface. With a bare cath-

394 ode there is no phase separation between this layer and the sample and one measures concentration rather than partial pressure, independent of the oxygen solubility in the medium used. Unfortunately, this potential advantage is more than offset by other difficulties involved in quantitative calibration of signals obtained with a bare cathode. In fact no such calibration seems to have been achieved, except for the turbulent flow – concentration electrode used by Etienne (1968), and one has to resort to comparison of the signal to that obtained with a more easily calibrated method.

C. Rate Electrodes Until about 25 years ago, photosynthesis researchers measuring oxygen evolution were often primarily interested in highly precise measurements of the rate of photosynthesis to study induction phenomena and action spectra and sophisticated electrochemical cells had been developed for this purpose, as reviewed by Fork (1972). The design of these cells was based on the pioneering work of L. Blinks and coworkers. Blinks and Skow (1938) already used photosynthetic material appressed to a large surface platinum cathode. Haxo and Blinks (1950) introduced the ‘rate electrode’, using a semi-permeable membrane to keep the sample appressed to the cathode and immersing the ensemble in a large volume of stirred or flowing medium to keep the oxygen concentration and other chemical conditions constant. The concentration gradient is now in the thin sample layer and the idea of F. Haxo and L. Blinks was that the rate of oxygen evolution there would simply add up to the background current due to oxygen flow from the medium, through the sample, towards the cathode. In their conditions most of the photosynthetic oxygen may actually be lost to the medium, but anyway the amplitude of the light-induced increase of the signal is proportional to the rate of oxygen evolution. Later rate electrode designs mostly used a shallow sample compartment in which unicellular algae or isolated chloroplasts were allowed to settle in a thin layer on the cathode. Many variations have been published. A typical system may consist of (from bottom up): Pt cathode, about

Hans J. van Gorkom and Peter Gast sample sediment, about stationary supernatant medium, dialysis membrane, flowing medium of constant composition, and window for illumination. The Ag/AgCl anode, which much be shielded from illumination, may be a ring surrounding the cathode, to reduce electromagnetic interference. The positive spike at the moment of the flash, often seen in reported measurements of flash-induced oxygen evolution, is an artifact caused either by insufficient optical shielding of the Ag/AgCl anode from the flash light, or by insufficient electromagnetical shielding of the electrochemical cell as a whole from the flash tube discharge. The anode can be placed in the sample compartment, in the flowing medium compartment, or in a third compartment separated from the conditioning medium by a second dialysis menbrane and itself being part of a second flow system. The latter arrangement was introduced by Joliot and Joliot (1968) to avoid chemical interference of the conditioning medium with the anode. It should also allow the use of different electrolytes in the sample and anode compartments (e.g. to study chloride depletion), but this facility appears not to have been used. Pickett (1966) placed the anode with the cathode under a teflon membrane, combining the Haxo and Blinks electrode and Clark electrode concepts (used e.g. by Weiss and Sauer, 1970). Den Haan et al. (1976) used a sample compartment of adjustable depth: the stationary supernatant medium was pushed out mechanically through the membrane. Diner and Mauzerall (1973) used a flowing gas instead of a solution as the conditioning medium. This may be the most efficient way to supply oxygen and, if necessary, to remove hydrogen, but it does not remove and leaves little room for electrolyte to connect anode and cathode electrically. A too large electrical resistance leads to an electrical potential gradient in the electrolyte, which corresponds to using a weaker polarization, but may also result in a slow response time if the circuit contains a large capacitance (Meunier and Popovic, 1988). Rate electrodes can be very sensitive. The modulated rate measurement as described by Joliot (1972) may be the most sensitive way to measure the rate of photosynthesis: an oxygen detection limit of has been reported.

Oxygen evolution

D. Concentration Electrodes Going back to the setup of Blinks and Skow (1938), one might also try to optimize the system for measurement of the oxygen concentration rather than of its rate of change. To avoid oxygen loss, the sample should be bound by an oxygenimpermeable wall, or at least be a thick layer on the cathode. The problem is that the large bare cathode will quickly modify the chemical conditions near its surface. Vigorous stirring (Joliot, 1965) helps to spread the cathode-induced changes in chemical conditions rapidly over the whole sample but does not prevent them. Etienne (1968) used continuous sample replacement. A small cathode was placed at the outlet of a capillary tube in which a fast, turbulent sample flow was maintained. By varying the distance between the cathode and a small, brightly illuminated spot in the capillary, the flash induced kinetics could be determined in steady state measurements, allowing a good signal/noise ratio in spite of the small size of the cathode. The system has limited applicability, but it does prevent all influence of the cathode on the sample. Plijter et al. (1988) used a stationary sample suspension in a closed cuvette and minimized the influence of the cathode by using an extremely weak polarization ( – 0.1 V vs. SHE), compensating the reduction in cathode efficiency by a large cathode surface area Ideally, the cathode should function as a non-disturbing probe and should not create substantial diffusion gradients in the sample. The measured signals were indeed shown to be independent of viscosity (van Gorkom et al., 1989). Electrochemists probably would not call this method polarography at all. The electrochemical cell described by Plijter et al. (1988) was complicated by the requirement of simultaneous UV absorbance measurements (van Leeuwen et al., 1990) via a high transmittance grid cathode as used by Marsho and Hommersand (1975), but we have obtained similar results with a very simple system consisting of an 0.5mm thick Pt-plastic-Ag/AgCl sandwich inserted in a standard 1 or 2 mm path length spectrophotometer cuvette, the Pt side facing the flash lamp (H.J. van Gorkom and M.A. van Dijk, unpublished). The detection limit of the Plijter electrode, with 10 ms response time and no aver-

395 aging, is an oxygen concentration change of about M. For reasons yet unknown, mutually exclusive results were obtained by Etienne (1968) and Plijter et al. (1988), so these methods should be used with caution.

E. Polarograms A polarogram is a plot of the measured current as a function of the polarization voltage, as in Fig. 1a. As usual, the polarization voltage is indicated by the potential difference applied between the Pt cathode and Ag/AgCl anode, but what happens to oxygen when it hits the cathode depends only on the electrochemical potential of the cathode, which should preferably be indicated relative to the Standard Hydrogen Electrode. If the cathode potential is not measured or fixed relative to a separate reference electrode by a potentiostat circuit, it may be estimated on the basis of the equilibrium potential of the Ag/AgCl couple at the chloride concentration used, but the actual anode potential will be higher. The deviation depends on the current density at the anode and can safely be neglected only if the anode is many times larger than the cathode. The polarogram shows an exponentially rising curve (the cathode potential affects the activation energy of the reaction) until diffusion of oxygen towards the cathode becomes rate limiting: a plateau sets in at the voltage where oxygen in a boundary layer on the cathode surface is reduced more rapidly than it can be replenished from the solution. The steady state concentration in the boundary layer can only approach zero. The maximum concentration gradient in the stationary layer between cathode and stirred solution, and hence the rate of oxygen transport towards the cathode and the electrode current, is therefore limited by the concentration in the bulk solution. This is what makes the current in a Clark electrode voltage independent. It does not mean that every oxygen molecule hitting the cathode is reduced. Stronger polarization may still allow an exponentially faster oxygen reduction and decrease the concentration in the boundary layer manifold, but if that concentration was negligible already the concomitant increase of the diffusion

396

gradient and of the steady state electrode current will be negligible, too. With a rate electrode the background current due to oxygen diffusing from the flowing conditioning medium to the cathode also shows a polarogram with a more or less well-defined plateau; only the current increase due to hydrogen evolution at strong polarization starts earlier due to the lower pH (Fig. 1b, open symbols). Haxo and Blinks (1950) used a polarization voltage in the middle of this plateau. With a rate electrode, however, one measures oxygen evolution in the boundary layer: the sample is on the electrode side of the diffusion gradient and the background current resulting from this gradient is in fact irrelevant. As just stated, the rate of oxygen reduction in the boundary layer should increase exponentially with the voltage, if not limited by the electronics, until every oxygen molecule hitting the cathode is reduced. The polarization voltage at which that happens can be estimated from the

Hans J. van Gorkom and Peter Gast

kinetics of the flash induced signal if the kinetics without disturbance by the measurement are known. The data and calculations of Plijter et al. (1988) indicate that 100% efficiency of the cathode reaction is approached at a cathode potential of about – 0.7 V vs. SHE (or – 1.0 V vs. Ag/AgCl at At this potential hydrogen production at a Pt cathode is already causing a considerable background current and all reported oxygen measurements with rate electrodes were carried out at much weaker polarization, mostly around – 0.4 V vs. SHE, where oxygen reduction by the cathode still depends exponentially on the cathode potential and the cathode is not at all the efficient oxygen trap it is usually thought to be. Polarograms of flash-induced oxygen evolution measured with rate electrodes show an exponential dependence of the current on the polarization voltage only to about – 0.3 V vs. SHE and a less steep increase of the current at stronger polariz-

Oxygen evolution ation (Fig. 1b, solid symbols). This has sometimes been attributed to limitation by O 2 diffusion from – 0.3 V followed by a gradual change from 2electron reduction formation) to 4-electron reduction formation) at – 0.7 V, but there is little evidence for that. It could be due instead to the convolution of oxygen production and its consumption by the electrode, and saturation of the electrode efficiency could explain the maximum near – 0.7 V vs. SHE. Myers and Graham (1963) already noted the absence of a flat plateau and emphasized the need to stabilize the cathode potential and minimize the resistance in the circuit. A potentiostat circuit using a separate reference electrode and/or current to voltage conversion to reduce the resistance are now generally used.

F. Flash-Induced Kinetics The kinetics of the current transient upon flashinduced oxygen evolution in a rate electrode setup is not fully understood. It consists of a sigmoidal rise in a few ms, followed by a decay in tens of ms. It is agreed that the rise (or its Fourier transform, as measured by Joliot et ah., 1966) is caused by the to transition’, the reduction of the oxygen evolving complex by water, which normally proceeds with a time constant of 2 ms. The generally accepted view is that this coincides with the release of an oxygen molecule, and hence that the rise of the signal shows the release of the oxygen. According to Plijter et ah. (1988), however, a good rate electrode measures the rate of oxygen evolution, so the kinetics of the flashinduced current transient is the first derivative of the oxygen concentration, in agreement with their findings with a concentration electrode. The rise of the signal from a rate electrode in that case means a delay preceding oxygen release, attributed to water oxidation by the oxygen evolving complex, and the time constant of the release process itself is reflected in the decay of the signal. No valid criticism or feasible alternative explanation of the data or reasoning in the Plijter et ah. paper has been published to date. Also no one has reported an attempt to reproduce the measurements, but the method has been applied ever since in this laboratory, using various elec-

397 trode arrangements and electronics, and we have found nothing wrong with the measurements. The study by Plijter et al. (1988) was prompted by the paradox that after removal of the extrinsic 33 kDa protein the UV absorbance change due to reduction of the oxygen evolving complex, and hence presumably water oxidation, was much slower than oxygen release as measured by the rise of the signal from a thin sample layer centrifuged onto a strongly polarized electrode (Miyao et al., 1986). Plijter et al. (1988) showed both theoretically and experimentally that at 100% cathode efficiency a first order oxygen release process in a stationary sample suspension in a closed cuvette produces a transient current increase with a rise time 7 times shorter than the time constant of the release process, and the situation of a very thin sample sediment on a very inefficient cathode under a stationary supernatant is mathematically the same. The combination of a thin layer, strong polarization, and oxygen removal by a flowing medium can only lead to a further acceleration of the signal. It seems inevitable, therefore, that oxygen release is at least 7 times slower than the rise of the signal one would expect to measure with a rate electrode. However, the observed rise kinetics of the signal is not understood. It is clearly sigmoidal and the initial delay is not accounted for in the model of Plijter et al. It might be due to the chloridedependent artifact described by these authors. A good fit can be obtained by assuming that all oxygen sources are confined to a plane at a distance close to or even exceeding the thickness of the sample sediment and release oxygen with a time constant several fold less than that of water oxidation (Lupatov, 1979; Lavorel, 1992), but this is clearly unrealistic. More likely values of these parameters do not allow an acceptable fit of the initial delay in the signal and the steep decline after its maximum (M.H. Vos, unpublished). Many authors have claimed inconsistency between results obtained with the method of Plijter et al. and those obtained with rate electrodes and other methods (Lavergne, 1989; Mauzerall, 1990; Strzalka et al., 1990; Jursinic and Dennenberg, 1990; Schulder et al., 1990, 1992; Meunier and Popovic, 1991; Tang et al., 1991; Lavorel, 1992; Ichimura et al., 1992; Joliot et al., 1992), but in

398 no case this was demonstrated for the same material in the same conditions. It is often overlooked that the release times given in Plijter et al. refer to a temperature of 5°C and decrease with increasing temperature by a factor of 2 per 20°C, and that shorter signal rise times were observed in samples which exhibit flash-induced oxygen uptake as well. It is clear from the previous section, however, that Plijter et al. overemphasized the importance of weak polarization. They actually found similar signal rise times at cathode potentials down to – 0.45 V (– 0.73 V vs. Ag/AgCl at which is well in the range of values commonly used but still implies a low cathode efficiency. The different signal shape obtained with rate electrodes must be primarily due to diffusion of oxygen out of the thin sample layer, away from the cathode. III. EPR Oximetry

A. EPR Line Broadening by Oxygen EPR oximetry has first been applied by the group of Y.N. Molin (Backer et al., 1977). This technique is based on the fact that molecular oxygen is paramagnetic. As a consequence, collisions between oxygen and a properly chosen spin probe, will increase spin-spin and spin lattice relaxation of this probe through the so-called Heisenberg exchange mechanism (Windrem and Plachy, 1980; Subczynski and Hyde, 1981; Popp and Hyde, 1981). For homogeneously broadened lines, this will result in broadening of the spectrum and changes in its microwave power saturation behavior. Under non-saturating conditions, at low microwave power, the EPR line will broaden and thus the amplitude of the first derivative signal, which is inversely proportional to the square of its linewidth, will decrease. In first approximation, the line broadening is often linearly dependent on the oxygen concentration (Windrem and Pachy, 1980; Glockner and Swartz, 1992). Measuring of oxygen levels by linewidth broadening can be done in two ways: a) direct measurement of the linewidth of the EPR signal, by recording of the EPR spectrum, and b) indirect measurement of the change in linewidth by recording the change in amplitude of the first derivative EPR spectrum. In the latter method,

Hans J. van Gorkom and Peter Gast the magnetic field sweep is turned off and the magnetic field is set at the maximum of the EPR signal. By comparing the amplitude changes at low and high microwave power distinction can be made between changes due to oxygen concentration changes and those caused by disappearance of the radical through chemical reaction (Strzalka et al., 1986). Under saturating conditions, at high microwave power, an increase of the signal intensity may be observed under certain conditions, due to de-saturation at elevated oxygen levels, and recently this effect has also been used to measure oxygen concentration changes (Ligeza et al., 1994). EPR oximetry is non-invasive, does not consume oxygen or produce harmful reaction products and does not require stirring of the sample. It can be used to measure the oxygen concentration in very small systems, like cells. In such experiments a spin probe is added that penetrates the cell and the extracellular nitroxide signal is quenched by adding paramagnetic salts like ferricyanide (Salikhov et al.,1971; Swartz, 1978), which do not enter the cell. The EPR signal of the extracellular nitroxide is broadened to such an extent that it virtually disappears and only the signal from intracellular nitroxides remains. Spinlabels can also be distributed homogeneously throughout an organ or whole body, and 2D and 3D EPR oximetry imaging is possible (Bacic et al.,1988; Demsar et al., 1988; Swartz and Glockner, 1989). For a review on EPR oximetry and its many uses, see Swartz and Glockner (1989).

B.

Oxygen Probes for EPR

A prerequisite for using EPR in oximetry is the presence of a (stable) radical in the system under study. Free radicals are rare in biological systems, and stable radicals are almost absent (an exception to this is melanin, which can also be used as oxygen probe (Sarna et al., 1980). Therefore these radicals have to be added. For this purpose nitroxide radicals have been widely used and characterized in spin-label experiments as ‘reporter’ molecules. A major disadvantage of EPR oximetry has been the fact that many spin labels and especially the nitroxides are rather toxic. Another complica-

Oxygen evolution tion is that there are many parameters other than oxygen that will broaden the EPR line of nitroxide radicals, like the presence of paramagnetic ions (Salikhov et al.,1971), nitroxide concentration, viscosity, microwave power, pH and temperature. Therefore it is necessary to calibrate the oxygen effect for each system and to keep parameters like temperature, microwave power and nitroxide concentration constant. Furthermore, EPR oximetry is not very sensitive, especially at high oxygen concentrations (see below). The new oxygen-sensitive probes that have been introduced by the group of H. Swartz in recent years (Swartz et al. 1991; Glockner and Swartz, 1992; Liu et al., 1993) seem to overcome the above mentioned shortcomings of the traditional, nitroxide-based EPR oximetry. They are fusinite (a certain type of coal), certain chars, soot, lithiumphthalocyanine crystals (PcLi) (Turek et al., 1987) and even Indian Ink. What these new spin-probes have in common is that they are all highly insoluble in most solvents, very sensitive to oxygen, non-toxic, virtually insensitive to the environment and extremely stable (although PcLi in water may be less stable than previously thought, M. Moussavi, personal communication). They are solid particles and measure the oxygen partial pressure, whereas the soluble nitroxides measure concentration (Povitch, 1975) unless enclosed in lipid droplets (Ligeza et al., 1994). The sensitivity and nontoxicity of the new probes has been demonstrated by measuring changes in oxygen concentration in the brain of a living mouse, in living cells, liver, and heart; their stability was convincingly shown by recording oxygen levels in a rat’s leg muscle over a period of more than 150 days (Glockner and Swartz, 1992; Liu et al., 1993).

C. Sensitivity When calculating the sensitivity of an EPR oxygen-probe, distinction should be made between the two ways of measurement: a) direct linewidth broadening measurement and b) amplitude measurement. The amplitude measurement is more sensitive, but cannot always be used since in many systems the nitroxides are rapidly bio-reduced to EPR silent hydroxylamines (Swartz et al., 1986).

399

This means that the spin-concentration is not constant. 1. Sensitivity in Direct Linewidth Broadening Measurement If the broadening is proportional to the oxygen concentration (Windrem and Pachy, 1980; Glockner and Swartz, 1992) the linewidth can be written as: The minimal change in linewidth that can be measured will depend on the linewidth itself. This is the reason why often deuterated nitroxides are used because of their reduced linewidth. Based on experimental facts, it is not unreasonable to state that for an EPR signal with reasonable signal-to-noise ratio (S/N) a change of 5% in linewidth of an EPR signal can be observed. Therefore:

and If the proportionality constant c is determined from the linewidths at 0 and at the latter being about the concentration in air-saturated water at 20°C (Hitchman, 1978), we get: For the deuterated nitroxide TEMPONE, with and (Swartz and Pals, 1989) the minimally detectable change in oxygen concentration at is about and at this value has increased to 20 The most sensitive probe, PcLi, with a 14 mG and will have a minimally detectable change at of about 200 nM and at of 15 2. Sensitivity in Amplitude Measurement This method is in principle much more sensitive than the linewidth measurement, since changes in amplitude are more easily measured than changes in linewidth and because the amplitude of the first derivative of an EPR line is proportional to the square of the inverse of the line-

400

Hans J. van Gorkom and Peter Gast

width. The minimally detectable oxygen-change with this method will depend on the S/N of the probe signal. Although there are limitations to the maximum concentration for spin-labels (due to line broadening at high concentration), it is not unreasonable to assume that the S/N at can be 1000 or more for many systems (taking the number of spins/g in PcLi as (Glockner and Swartz, 1992) and the sensitivity of a commercial X-band EPR spectrometer as spins/Gauss, 10 nanogram will be sufficient to achieve this S/N under anaerobic conditions). If we further assume that the minimal detectable amplitude change is a change with S/N = 1 we obtain: which is 100 times less than that obtained in a linewidth measurement. For d-TEMPONE the detection limit becomes at and 75 nM at 0 for PcLi it becomes 150 nM at 300 and 2 nM at 0 The last value is probably comparable to the minimum flash yield detectable with a polarographic rate electrode. It should be noted, however, that these numbers apply in rather ideal situations. For instance, when EPR oximetry is used in a whole-body system using a low-Q surface probe instead of a standard high-Q cavity, and using low microwave frequency to increase microwave penetration (Nilges et al., 1989; Bacic et al., 1989), or in timeresolved EPR oximetry, the sensitivity may be greatly reduced. If one can exclude or correct for spin-label reduction, both methods will give identical results, taking the quadratic decrease of the amplitude with increasing linewidth into account. However, in time-resolved studies where the magnetic field sweep is turned off and the field is set at the maximum of the EPR signal, deviating results may be obtained when the changes in oxygen concentration are large. This is due to the fact that when the oxygen concentration during the measurement is increased or decreased, the linewidth will broaden or narrow, respectively, and as a result, the magnetic field position will ‘slide off the peak of the first derivative spectrum. This is shown in Fig. 2. Here the calculated response is depicted of the EPR amplitude, set at the maxi-

mum of the EPR line at t = 0, to an oxygen concentration increase with a time constant of 200 ms. When the change is small, the signal amplitude reflects the oxygen evolution kinetics. However, when the change in oxygen concentration is large, a more rapid and non-exponential curve is observed. As expected, the deviation becomes significant at much smaller oxygen concentration changes with PcLi than with d-TEMPONE, since PcLi has a much sharper EPR line. In photosynthetic material, such large changes may occur locally for a short time. For instance, a chloroplast suspension of a chlorophyll concentration of 1 mM may produce more than oxygen on a flash and all that oxygen may initially be confined to a small fraction of the volume: the chloroplasts occupy about 1/40 and the thylakoids only 1/300

Oxygen evolution of the suspension volume (Heldt et al., 1973). On a time scale of some ms, oxygen diffusion processes and the non-linearity of the EPR amplitude may complicate the observed kinetics, depending on the microscopic distribution of the oxygen and that of the spin probes used. On the other hand, this might become a unique tool to obtain time-resolved information on the microscopic distribution of oxygen.

D. Applications in Photosynthesis Research Up to now only a few reports have appeared on EPR oximetry using photosynthetic material. Strzalka et al. (1986) used the spin label TEMPONE to measure oxygen evolution in thylakoid membranes of spinach. They measured amplitude changes and used the difference in microwave power saturation behavior to distinguish between production of oxygen and photoreduction of the probe, which could be controlled by adding sufficient amounts of an electron acceptor (p-benzoquinone). Belkin et al. (1987) have used EPR oximetry to measure light induced oxygen production in whole cells of cyanobacteria under continuous illumination. Direct linewidth as well as amplitude measurements were used in this study. By the paramagnetic agent the EPR signal from the extracellular oxygen probe was broadened beyond detection and the intracellular oxygen production and photo-inhibition was measured selectively. In a second report from Strzalka et al. (1990) the more sensitive probe perdeuterated TEMPONE was used to measure the oxygen release time in thylakoids after a short flash by the change in EPR amplitude. This was found to be 0.4–0.5 ms, which is much shorter than the 2 ms time constant of water oxidation and may perhaps have been caused by the non-linearity discussed above. Perhaps contrary to the soluble nitroxides, the new, solid spin probes should allow unambiguous distinction between free oxygen and oxygen still associated with the photosynthetic system, if the slow release process postulated by Plijter et al.(1988) exists. The only report so far on the use of PcLi crystals as an oxygen probe in photosynthetic material is from Tang et al. (1991), who found by amplitude measurements an oxygen release time in PSII membranes of 1–2 ms. In a recent article by Dis-

401 mukes et al. (1994) TEMPONE oximetry was used to demonstrate hydrogen peroxide production in photosystem II membranes inactivated by depletion: oxygen production was restored not only by readdition of but also by addition of catalase. With oxygen polarography, such experiments are complicated by the production of hydrogen peroxide at the cathode. Ligeza et al. (1994) used the oxygen-dependent desaturation at high microwave power to measure oxygen evolution in leaves, after injection of an emulsion containing a nitroxide enclosed in droplets of paraffin oil covered with serum albumin. This system largely prevents photoreduction of the nitroxide. A detection limit of oxygen was reported.

IV. Mass Spectrometry Unlike most processes discussed in this volume, photosynthetic oxygen evolution involves the rearrangement of nuclei to form one kind of molecule from another and can be studied by traditional biochemical tools using labeling of the substrate by nuclear isotopes and measuring the isotope composition of the product. The usefulness of mass spectrometry in photosynthesis research was established by Hoch and Kok (1963), who replaced the sample injection port of a conventional mass spectrometer by a vessel containing a stirred suspension of photosynthetic material separated only by a teflon membrane from the vacuum system. Later authors have mostly used a thin layer of photosynthetic material sedimented on the teflon membrane. The selective permeability of the membrane allows direct measurement of the gases dissolved in the sample solution. Sufficient sensitivity and time resolution can be reached to allow individual flash yields in a series to be analysed. The method has been reviewed by Radmer and Ollinger (1980). It has more recently been used e.g. to measure oxygen uptake kinetics during the induction of oxygen evolution upon continuous illumination (Peltier and Ravenel, 1987), to distinguish oxygen evolution from water and that from hydrogen peroxide (Mano et al., 1987), and to prove that water oxidized on the to transition is still exchangeable in (Radmer and Ollinger, 1986; Bader et al., 1987), the kinetics of which have

402 now been measured with 30 ms time resolution (Messinger et al., 1995). Photosynthetic oxygen evolution is almost always accompanied by light-dependent oxygen uptake processes and an unambiguous distinction between the two can normally not be made by other methods, which measure only the net change in oxygen concentration. But also mass spectrometry may not help. It should be kept in mind that these uptake processes take place at short distance and may selectively consume the oxygen just produced photosynthetically. In fact that can be the cause of their light-dependence, as was strikingly illustrated by the measurement of a period four oscillation in the oxidation of mitochondrial cytochrome c upon flash illumination of a green alga (Lavergne, 1989). Perhaps a quantitative trapping of photosynthetic oxygen before it can leave the cell may explain the apparent oxygen dependence of oxygen evolution reported for a cyanobacterium (Bader and Schmid, 1988).

Hans J. van Gorkom and Peter Gast VI. Galvanic Sensors Galvanic sensors to detect oxygen in the gas phase have been developed for technical applications mainly (Kleitz and Fouletier, 1976; Heyne, 1976), but there are a few reports of the use of zirconium oxide detectors in photosynthesis research (Björkman and Gaul, 1970; Greenbaum and Mauzerall, 1976; Meyer et al., 1989). These devices are based on the phenomenon that a mixed crystal of conducts at high temperatures (around 800°C) and can be used as a solid electrolyte between two Pt electrodes exposed to the gas to be measured and to a reference gas, respectively. Due to the required transport and drying of the gas the time-resolution is poor and the flash number dependence of the oxygen yield can be determined only by the cumulative method (measuring the total yield of a series of 1, 2, 3, etc. flashes). The main advantages of the technique are that it combines single flash sensitivity with easy quantitative calibration, and that there is no chemical interference between sample and detector.

V. Photoacoustic Spectroscopy VII. Prospects The traditional volumetric measurement of photosynthetic oxygen evolution by the Warburg technique may now be obsolete, but it has a modern pendant: the pressure increase in the gas phase due to oxygen evolution causes a ‘photobaric’ contribution to photoacoustic signals (Bults et al., 1982). Photoacoustic spectroscopy is reviewed elsewhere in this volume by S. Malkin (Chapter 12). The photobaric oxygen signal can be measured with high sensitivity and time resolution. It shows the characteristic period four oscillation in a flash series (Canaani et al., 1988) and the flash-induced kinetics have been studied with ms time resolution (Mauzerall, 1990). The concomitant photothermal signal, oxygen uptake phenomena in the leaf discs used, and the use of an open microphone and a.c. amplification led to convoluted signal kinetics, but their analysis according to Mauzerall suggested that oxygen release was fast. The use of a very thin layer of a PS II preparation which does not show oxygen uptake, and a pressure transducer as detector, might indeed open the road to obtain conclusive evidence on the oxygen release time.

There is no single ‘best’ way to measure photosynthetic oxygen evolution. Each of the techniques described above has its own specific possibilities and drawbacks, and different experimental situations may call for different methods of measurement. Moreover, all these methods still hold potential for further development. Polarography, being relatively cheap and easy to implement, continues to be a rewarding playground for inventive experimenters. The application of EPR oximetry in photosynthesis research is in an early, exploratory phase and is just beginning to prove its value; mass spectrometry will remain indispensable to unravel the interplay between oxygen evolution and oxygen consuming processes in photosynthetic organisms. Methods which have so far only been studied for their own interest, but also phenomena now considered as troublesome complications of established methods, may suddenly develop into valuable tools to solve a particular problem. Despite its long history, research on the mechanism of photosynthetic oxygen evolution is

Oxygen evolution still in full swing and the oscillatory phenomena caused by the 4–flash redox cycle of the oxygen evolving complex have not lost their charm. It is to be expected that oxygen measurements will continue to play an essential role in their study. Acknowledgements We are grateful to Dr. M.H. Vos for advice and extensive fitting of time-resolved polarographic data kindly supplied by Dr. J. Lavorel, and to Dr. G.H. Schmid for pointing out the benefits of collodion films and how to apply them. Research in this laboratory was supported by the Netherlands Foundation for Chemical Research (SON), financed by the Netherlands Organization for Scientific Research (NWO). P.G. is a research fellow of the Royal Netherlands Academy of Arts and Sciences (KNAW). References Bacic G, Demsar F, Zolnai Z and Swartz HM (1988) Contrast enhancement in ESR imaging: role of oxygen. Magn Res Med Biol 1: 54–65. Bacic G, Nilges MJN, Magin RL, Walczak T and Swartz HM (1989) In vivo localized ESR spectroscopy reflecting metabolism. Magn Res Med 10: 266–272. Backer JM, Budker VG, Eremenko SI and Molin YN (1977) Detection of the kinetics of biochemical reactions with oxygen using exchange broadening in the ESR spectra of nitroxide radicals. Biochim Biophys Acta 460: 152–156. Bader KP and Schmid GH (1988) Mass spectrometric analysis of a photosystem II mediated oxygen uptake phenomenon in the filamentous cyanobacterium Oscillatoria chalybea. Biochim Biophys Acta 936: 179–186. Bader KP, Thibault P and Schmid GH (1987) Study on the properties of the by mass spectrometry in the filamentous cyanobacterium Oscillatoria chalybea. Biochim Biophys Acta 893: 564–571. Baumgärtl H, Grunewald W and Lübbers DW (1974) Polarographic determination of the oxygen partial pressure field by Pt microelectrodes using the field in front of a Pt macroelectrode as a model. Pflügers Arch 347: 49–61. Belkin S, Mehlhorn RJ and Packer L (1987) Determination of dissolved oxygen in photosynthetic systems by nitroxide spin-probe broadening. Arch Biochem Biophys 252: 487– 495. Björkman O and Gaul E (1970) Use of the zirconium oxide ceramic cell for measurement of photosynthetic oxygen evolution by intact leaves. Photosynthetica 4: 123–128. Blinks LR and Skow RK (1938) The time course of photosynthesis as shown by a rapid electrode method for oxygen. Proc Natl Acad Sci USA 24: 420–427. Bulls G, Horwitz BA, Malkin S and Cahen D (1982) Photoac-

403 oustic measurements of photosynthetic activities in whole leaves. Photochemistry and gas exchange. Biochim Biophys Acta 679: 452–465. Burr A and Mauzerall D (1968) The oxygen luminometer. An apparatus to determine small amounts of oxygen, and application to photosynthesis. Biochim Biophys Acta 153: 614–624. Canaani O, Malkin S and Mauzerall D (1988) Pulsed photoacoustic detection of flash-induced oxygen evolution from intact leaves and its oscillations. Proc Natl Acad Sci USA 85: 4725–4729. Clark LC (1956) Monitor and control of blood and tissue oxygen tensions. Trans Am Soc Art Int Organs 2: 41–45. Degn H, Balslev I and Brook R (eds) (1976) Measurement of Oxygen. Elsevier, Amsterdam. Delieu TJ and Walker DA (1983) Simultaneous measurement of oxygen evolution and chlorophyll fluorescence from leaf pieces. Plant Physiol 73: 534–541. Demsar F, Walczak T, Morse PD, Bacic G, Zolnai Z and Swartz HM (1988) Detection of diffusion and distribution of oxygen by fast-scan EPR imaging. J Magn Res 76: 224– 231. Den Haan GA (1976) Chlorophyll a fluorescence as a monitor for rapid reactions in system II of photosynthesis. Doctoral Thesis, University of Leiden. Den Haan GA, Gorter de Vries H and Duysens LNM (1976) Correlation between flashinduced oxygen evolution and fluorescence yield kinetics in the 0 to 16 range in Chlorella pyrenoidosa during incubation with hydroxylamine. Biochim Biophys Acta 430: 265–281. Diner B and Mauzerall D (1973) Feedback controlling oxygen production in a cross-reaction between two photosystems in photosynthesis. Biochim Biophys Acta 305: 329–352. Dismukes GC, Zheng M, Hutchens R and Philo JS (1994) The inorganic biochemistry of photosynthetic water oxidation. Biochem Soc Trans (London) 22: 323–327. Etienne AL (1968) Etude de l' étape thermique de l' émission d' oxygéne par une méthode d' écoulement. Biochim Biophys Acta 153: 895–897. Fatt I (1976) Polarographic Oxygen Sensors. CRC Press, Cleveland OH. Fork DC (1972) Oxygen electrode. In: San Pietro A (ed) Methods in Enzymology Vol 24 Photosynthesis and Nitrogen Fixation part B, pp 113–122. Academic Press, New York. Glockner JF and Swartz HM (1992) In vivo EPR oximetry using two novel probes: fusinite and lithiumphthalocyanine. In: Erdman W and Bruley DF (eds) Oxygen transport to tissue XIV. Plenum Press, New York. Gnaiger E and Forstner H (eds) (1983) Polarographic Oxygen Sensors. Springer-Verlag, Berlin. Greenbaum E and Mauzerall DC (1976) Oxygen yield per flash of Chlorella coupled to chemical oxidants under anaerobic conditions. Photochem Photobiol 23: 369–372. Hampp R, Mehrle W and Zimmermann U (1986) Assay of photosynthetic oxygen evolution from single protoplasts. Plant Physiol 81: 854–858. Haxo FT and Blinks LR (1950) Photosynthetic action spectra of marine algae. J Gen Physiol 33: 389–422. Heldt HW, Werdan K, Milovancev M and Geller G (1973)

404 Alkalization of the chloroplast stroma caused by light-dependent proton flux into the thylakoid space. Biochim Biophys Acta 314: 224–241. Heyne L (1976) Some properties and applications of zirconiabased solid electrolyte cells. In: Degn H, Balslev I and Brook R (eds) Measurement of Oxygen. Elsevier, Amsterdam. Hitchman ML (1978) Measurement of dissolved oxygen. In: Elving PJ, Winefordner JD and Kolthoff IM (eds) Chemical Analysis Vol 49. John Wiley and Sons, New York. Hoch G and Kok B (1963) A mass spectrometer inlet system for sampling gases dissolved in liquid phases. Arch Biochem Biophys 101: 160–170. Ichimura T, Miyairi S, Satoh K and Katoh S (1992) Preparation and characterization of heatstable and very active oxygen evolving photosystem II particles from the thermophilic cyanobacterium Synechococcus elongatus. Plant Cell Physiol 33: 299–305. Joliot P (1965) Cinétiques des réactions liés à l' émission d' oxygène photosynthétique. Biochim Biophys Acta 102: 116– 134. Joliot P (1972) Modulated light source use with the oxygen electrode. In: San Pietro A (ed) Methods in Enzymology Vol 24. Photosynthesis and Nitrogen Fixation part B, pp 123–134. Academic Press, New York. Joliot P (1993) Earlier researches on the mechanism of oxygen evolution: a personal account. Photosynth Res 38: 214– 223. Joliot P and Joliot A (1968) A polarographic method for detection of oxygen production and reduction of Hill reagent by isolated chloroplasts. Biochim Biophys Acta 153: 625–634. Joliot P, Hofnung M and Chabaud R (1966) Etude de l' émission d' oxygéne par des algues soumises à un éclairement modulé sinusoïdalement. J Chim Phys 10: 1423–1441. Joliot P, Barbieri G and Chabaud R (1969) Un nouveau modèle des centres photochimiques du système II. Photochem Photobiol 10: 309–331. Joliot P, Lavergne J and Beal D (1992) Plastoquinone compartmentation in chloroplasts. I. Evidence for domains with rather different rates of photoreduction. Biochim Biophys Acta 1101: 1–12. Jursinic PA and Dennenberg RJ (1990) Oxygen release in leaf discs and thylakoids of peas and photosystem II fragments of spinach. Biochim Biophys Acta 1020: 195–206. Kleitz M and Fouletier J (1976) Technological improvements and accuracy of potentiometric measurements with zirconia gauges. In: Degn H, Balslev I and Brook R (eds) Measurement of Oxygen, pp 103–122. Elsevier, Amsterdam. Lavergne J (1989) Mitochondrial responses to intracellular pulses of photosynthetic oxygen. Proc Natl Acad Sci USA 86: 8768–9772. Lavorel J (1992) Determination of the photosynthetic oxygen release time by amperometry. Biochim Biophys Acta 1101: 33–40. Ligeza A, Wisniewska A, Subczynski WK and Tikhonov AN (1994) Oxygen production and consumption by chloroplasts in situ and in vitro as studied with microscopic spin label probes. Biochim Biophys Acta 1186: 201–208. Liu KJ, Gast P, Moussavi M, Norby SW, Vahidi N, Walczak

Hans J. van Gorkom and Peter Gast T, Wu M and Swartz HM (1993) Lithiumphthalocyanine, a probe for electron paramagnetic resonance oximetry in viable biological systems. Proc Natl Acad Sci USA 90: 5438–5442. Lübbers K, Haumann M and Junge W (1993) Photosynthetic water oxidation under flashing light. Oxygen release, proton release and absorption transients in the near ultraviolet – a comparison between thylakoids and a reaction centre core preparation. Biochim Biophys Acta 1183: 210– 214. Lupatov VM (1979) Investigation of the rate of the final stage of the formation of oxygen during photosynthesis. Biophysics 23: 466–471. Transl. from Biofizika 23 (1978): 462–466. Mano J, Takahashi M and Asada K (1987) Oxygen evolution from hydrogen peroxide in photosystem II: flash-induced catalatic activity of water-oxidizing photosystem II membranes. Biochemistry 26: 2495–2501. Marsho TV and Hommersand MH (1975) Slow 514 nm absorption phases and oxygen exchange transients in Ulva. Biochim Biophys Acta 376: 354–365. Mauzerall DC (1990) Determination of oxygen emission and uptake in leaves by pulses, time resolved photoacoustics. Plant Physiol 94: 278–283. Messinger J, Badger M and Wydrzynski T (1995) Detection of one slowly exchanging substrate water molecule in the state of photosystem II. Proc Natl Acad Sci USA 92: 3209–3213. Meunier PC and Popovic R (1988) Optimization of the bare platinum electrode as an oxygen measurement system in photosynthesis. Photosynth Res 15: 271–279. Meunier PC and Popovic R (1991) The time for oxygen release in photosynthesis: reconciliation of flash polarography with other measurement techniques. Photosynth Res 28: 33–39. Meyer B, Schlodder E, Dekker JP and Witt HT (1989) evolution and nanosecond reduction kinetics in single flashes as a function of pH. Biochim Biophys Acta 974: 36–43. Miyao M, Boussac A, Plijter JJ, Vernotte C, Lavorel J and Etienne AL (1986) Influence de la proteine membranaire extrinsèque de 33 kDa sur la cinétique de dégagement d' oxygène. In: Colloque de Photosynthèse. Institut National des Sciences et Techniques Nucléaires, Saclay. Myers JM and Graham JR (1963) Futher improvements in stationary platinum electrode of Haxo and Blinks. Plant Physiol 38: 1–5. Nilges MJ, Walczak T and Swartz HM (1989) 1 GHz in vivo ESR spectrometer operating with a surface probe. Phys Med 2: 195–201. Peltier G and Ravenel J (1987) Oxygen photoreduction and variable fluorescence during a dark-to-light transition in Chlorella pyrenoidosa. Biochim Biophys Acta 894: 543– 551. Pickett JM (1966) An oxygen electrode for measuring rates of photosynthesis on aliquots of a pretreated suspension. Carnegie Institution of Washington Yearbook 65: 487–492. Plachy W and Kivelson D (1967) Spin exchange in solutions of d-tertiairy-butyl nitroxide. J Chem Phys 47: 3312–3318. Plijter JJ, Aalbers SE, Barends JPF, Vos MH and van Gorkom HJ (1988) Oxygen release may limit the rate of photo-

Oxygen evolution synthetic electron transport; the use of a weakly polarized oxygen cathode. Biochim Biophys Acta 935: 299–311. Popp CA and Hyde JS (1981) Effects of oxygen on EPR spectra of nitroxide spin-label probes of model membranes. J Magn Res 43: 249–258. Povich MJ (1975) Measurement of dissolved oxygen concentrations and diffusion coefficients by electron spin resonance. Anal Chem 47: 346–347. Rabinowitch EI (1945) Photosynthesis and Related Processes Vol I Chemistry of Photosynthesis, Chemosynthesis and Related Processes in Vitro and in Vivo. Interscience Publishers, New York. Rabinowitch EI (1951) Photosynthesis and Related Processes Vol II, Part 1 Spectroscopy and Fluorescence of Photosynthetic Pigments; Kinetics of Photosynthesis. Interscience Publishers, New York. Rabinowitch EI (1956) Photosynthesis and Related Processes Vol II, Part 2 Kinetics of Photosynthesis (Continued); Addenda to Vol I and Vol II, Part 1. Interscience Publishers, New York. Radmer R and Ollinger O (1980) Measurement of the oxygen cycle: the mass spectrometric analysis of gasses dissolved in a liquid phase. In: San Pietro A (ed) Methods in Enzymology Vol 69 Photosynthesis and Nitrogen Fixation, part C, pp 547-560. Academic Press, New York. Radmer R and Ollinger O (1986) Do the higher oxidation states of the photosynthetic evolving system contain bound FEBS Lett 195: 285–289. Salikhov KM, Doctorov AB, Molin YN and Zamaraev KI (1971) Exchange broadening of ESR lines for solutions of free radicals and transition metal complexes. J Magn Res 5: 189–205. Sarna T, Duleba A, Korgtowski W and Swartz HM (1980) Interaction of melanin with oxygen. Arch Biochem Biophys 200: 140–148. Schröder WP and Åkerlund HE (1986) accessibility to the photosystem II donor side in protein-depleted insideout thylakoids measured as flash-induced oxygen production. Biochim Biophys Acta 848: 359–363. Schulder R, Bader KP and Schmid GH (1990) An amperometric study on the time constants of oxygen release in thylakoids of Nicotiniana tabacum and Oscillatoria chalybea. Z Naturforsch 45c: 1117–1126. Schulder R, Burda K, Strzalka K, Bader KP and Schmid GH (1992) Study of the parameters affecting oxygen release time measurements by amperometry. Z Naturforsch 47c: 465–473. Strzalka K, Sarna T and Hyde JS (1986) ESR oxymetry: measurement of photosynthetic oxygen evolution by spinprobe technique. Photobiochem Photobiophys 12: 67–71. Strzalka K, Walczak T, Sarna T and Swartz HM (1990) Measurement of time resolved oxygen concentration changes in

405 photosynthetic systems by nitroxide-based EPR oximetry. Arch Biochem Biophys 281: 312–318. Subczynski WK and Hyde JS (1981) The diffusion-concentration product of oxygen in lipid bilayers using spin-label method. Biochim Biophys Acta 643: 283–291. Swartz HM (1978) Use of nitroxides to measure redox metabolism in cells and tissues. J Chem Soc, Faraday Trans I 83: 191–202. Swartz HM and Glockner JF (1989) Measurement of concentration of oxygen in biological systems using EPR techniques. In: Hoff AJ (ed) Advanced EPR; Application in Biology and Biochemistry, pp 753–782. Elsevier, Amsterdam. Swartz HM and Pals MA (1989) Measurement of intercellular oxygen. In: Miquel J, Weber H and Quintanilha A (eds) CRC Handbook of Free Radicals and Antioxidants in Biomedicine 3, pp 141–151. CRC Press, Cleveland OH. Swartz HM, Sentjurc M and Morse II PD (1986) Cellular metabolism of water soluble nitroxides: effect on rate of reduction of cell/nitroxide ratio, oxygen concentration and permeability of nitroxides. Biochim Biophys Acta 880: 82– 90. Swartz HM, Boyer S, Gast P, Glockner JF, Hu H, Liu KJ, Moussavi M, Norby SW, Vahidi N, Walczak T, Wu M and Clarkson RB (1991) Measurement of pertinent concentrations of oxygen in vivo. Magn Res Med 20: 333–339. Tang XS, Moussavi M and Dismukes GC (1991) Monitoring oxygen in solution by ESR oximetry using lithium phthalocyanine: application to photosynthesis. J Am Chem Soc 113: 5914–5915. Turek P, Girandeau A and Simon J (1987) Preparation and study of lithiumphthalocyanine radical: optical and magnetic properties. Chem Phys Lett 134: 471–476. Van Gorkom HJ, van Leeuwen PJ, Vos MH and Barends JPF (1989) Oxygen evolution: UV spectroscopy and polarography. In: Baltscheffsky M (ed) Current Research in Photosynthesis Vol I, pp 693–700. Kluwer Academic Publishers, Dordrecht. Van Leeuwen PJ, Vos MH and van Gorkom HJ (1990) Photosynthetic oxygen evolution. Biochim Biophys Acta 1018: 173–176. Velthuys BR and Kok B (1978) Photosynthetic oxygen evolution from hydrogen peroxide. Biochim Biophys Acta 502: 211–221. Weiss C and Sauer K (1970) Activation kinetics of photosynthetic oxygen evolution under 20–40 nanosecond laser flashes. Photochem Photobiol 11: 495–501. Windrem DA and Plachy WZ (1980) The diffusion solubility of oxygen in lipid bilayers. Biochim Biophys Acta 600: 655–665. Wydrzynski T, Ångström J and Vänngård T (1989) formation by photosystem II. Biochim Biophys Acta 973: 23–28.

This page intentionally left blank

Index Since almost all chapters deal with chlorophyll or bacteriochlorophyll, these words are not used as entry in this index.

A absorbance-detected magnetic resonance, see ADMR, ODMR absorption spectrum 4-6 derivative of 4 dichroic 13, 16, 18, 28, 34 difference, light-induced 4-6 difference, reaction modulated infrared 153 difference, redox-induced 148 difference, electric field-induced 177-188 FTIR 137-157 hole burning 123-134 line width 43, 110, 125 transient 70, 71 X-ray 337-352 accumulated photon echo spectroscopy 109-120 ADMR 277-295 amorphous ice 327 anisotropy of absorbance 14 of emission 26 function 20 of time-resolved fluorescence 57 of transition dipoles 24 antenna, see LH 386 atomic scattering factor 318 temperature factor 318 averaging, a-periodic 328

B bacteriopheophytin 164, 169, 185, 186, 289 bottleneck state 113

C C-phycocyanin 57 windows 155, 156 caged compounds 148 carotenoids 27, 161, 166, 168, 185, 289, 290, 295, 305 charge pairs 97 formation in PSII 95 half lives of 99 recombination of 91 stabilization of 95 chemical shift tensor 306 chemically induced dynamic spin polarization 212 chirality 24, 25, 27, 30 chirally induced differential scattering (CIDS) 30-32 Chlamydomonas reinhardtii 69, 365 Chlorobium limicola 167, 171, 269 tepidum 131 Chloroflexus aurantiacus 29, 148, 292 Chlorogloea fritschii 370, 371 Chromatium vinosum 147, 172 chlorosome 6, 30

circular dichroism 24-34 excitonic 28 fluorescence-detected 13, 26, 34 intrinsic 28 long-range interaction 26 orientation dependence 26 psi-type 27, 31, 33 vibrational 26 Clark electrode 392 concentration electrode 395 Conne’s advantage 143 correlation spectroscopy 309 Cotton effect 25 cross-polarisation 301 cryo-electron microscopy 327, 328 crystal structure, determination of 317-323 cytochrome 359, 365, 375 cytochrome 371

D data fitting 78 dephasing 112, 118 dichroic ratio 18 dichroism, reduced 18 dielectric asymmetry 188 screening 187 difference spectrum, see absorption spectrum differential polarization images 22 microscopy 13 scattering 25 dipolar interactions 301 dipole moment 178 strength 15 dispersion model 119 double exchange 369 double resonance 277, 284, 285, 294, 295

E electric field 51, 177, 178 gradient 358 orientation 24 electroabsorption 177 electrochemical cell 147 oxidation 148 electrochromism, see Stark spectroscopy electrode, Clark 392 concentration 395 electron transfer at 148 gold grid 147 rate 394 electron crystallography 332 density map 320, 322 microscopy 326, 331

Index

408 nuclear double resonance, see ENDOR paramagnetic resonance, see EPR spin echo, see ESE spin echo envelope modulation, see ESEEM electron-phonon coupling 118, 130 Eligeron canadensis 103 emission, see fluorescence, phosphorescenc ENDOR 255-272 bacteriochlorophyll 256 Davies ESE 243 electron acceptors 271 general TRIPLE 261 in frozen solution 263 in liquid solution 258 in single crystals 265 Mims ESE 245 of 269 stochastic 267 TRIPLE 260 triplet states 265, 271 energy storage 195-199 transfer 6, 44, 87, 116, 125, 133, 213, 217, 226 EPR 211-229, 258, 278, 280, 289, 294, 295 direct detection 215 ENDOR-induced 262 Fourier transform 215-217, 226 multiline signal 248, 350 oximetry 398 pulsed 249 time-resolved 211-229 error analysis 85 ESE 235-252 ESEEM 238 EXAFS 338 definition 340 equation 341 Fe 346 Mn 349 theory 339 excitation transfer, see energy transfer excited state, see exciton, energy transfer exciton, annihilation 54 bands 26 circular dichroism 28 coherence 116 coupling 25, 29 dynamics 109, 110 interaction 29, 30, 33 manifold 117 scattering 116 states 116 extended X-ray absorption fine structure, see EXAFS

F FDMR 277, 278, 280-284, 288, 289 Fe centers 348, 355 K-edge spectra 347 quinone complex 365, 366 Fe (II) high-spin 361

Fe (III) high-spin 360 Fe (II) low spin 360 Fe(II) NO complex 367 Fe-S acceptors in photosystem I 346 Fe-S proteins 345, 346, 362, 368 Felgett’s advantage 143 Fenna-Matthews-Olson, see FMO film-stretching 21 flattening effect 4, 26 fluorescence, depolarization of 44, 68 emission 6, 18, 47 spectrum 6, 7, 46 lifetime 44 microwave-induced 278, 285 polarization of 18 relaxation 45 self absorption 48 single photon counting 52 spectrophotometer 45 time-resolved 52, 55 upconversion 53, 64 yield 6, 45, 197, 198 fluorescence-detected magnetic resonance, see FDMR, ODMR FMO complex 7, 116, 117, 131 force field 140 formate 367 Fourier analysis 330 transform infrared, see FTIR transform EPR 215-217, 226 Fourier-peak filtering 330 Franck-Condon principle 43 frequency grating 112 FTIR 137-137, see also infrared amide I mode 144 amide II mode 144 difference spectra 140, 145, 147, 148 protein modes 155 spectrophotometer 141, 145, 153, 154 spectroscopy 137-157, 306 techniques 138 time resolved 148 fusinite 399

G g-factor 258 galvanic sensors 402 gel-squeezing 19, 21 glow curves 94, 98 gold grid electrode 147 group frequencies 140, 141

H Hartmann-Hahn condition 301 Heliobacterium chlorum 292 heterogeneity 278, 289. 294, 295 high pressure hole burning 128 hole burning 117 double resonance ODMR 289

Index high-pressure 128 non-photochemical 124 photochemical 124 spectroscopy 114, 123-134 transient 124 homodyne detection 114 homogeneous linewidth 110 Huang-Rhys factor 125 hydrogen peroxide 393 hyperfine interaction 249, 258, 356, 357

I infrared 137-157 cell 155 detector 137, 141, 143, 144 difference spectroscopy 140, 145, 148 dispersive spectroscopy 153 lasers 153 photometer 153, 154 picosecond spectroscopy 137, 154, 155 spectroscopy 137-157 windows 156 inhomogeneous broadening 43, 125 linewidth 43, 110 interferogram 149 interferometer 141, 142, 149, 152 internal conversion 44 intersystem crossing 44 isomorphous replacement 319 isotope labelling 140, 312

J Jacquinot’s advantage 143

K K-edge spectra 338, 347, see also EXAFS, XANES Krönig-Kramers transforms 25

L Langmuir Blodgett film 16 laser, argon ion 64, 126 diode 127 dye 64, 164 Nd:YAG 165 picosecond infrared 137 regenerative amplifier 65 spectroscopy 63-72, 154, 155 Ti:Sapphire 64-66 tunable infrared 137, 154 LH1 67, 87, 170, 386 LH2 386, 388 LHC II 27, 29-33, 69, 117, 328, 331, 332 light-harvesting, see LH1, LH2, LHCII light scattering 4, 33

409 light-induced difference spectroscopy 147 linear dichroic T-S 286, 278, 290 linear dichroism 13, 16, 18. 34, see also orientation liquid crystal 218, 228 lithium phthalocyanine 399

M macro-organization of chromophores 26, 27, 30, 32 magnetic circular dichroism (MCD) 26, 34 mass spectrometry 401 metallo-proteins 338 microwave-induced absorption (MIA) 278-285 fluorescence (MIF) 278-285 phosphorescence (MIP) 278-285 micelles 380 Mn, see also oxygen evolving complex Mn cluster 246, 249 EXAFS 349 K-edge spectra 350 molecular dynamics calculation 322 molecular replacement 319 Mössbauer spectroscopy 355-372 Mueller images 30 matrix 14, 34 multiplex advantage 143, 151

N negative staining 327, 328, 333 nitroxide radicals 398 NMR 335 chemical shift tensor 306 CIDNP 309 CP/MAS 299, 300 cross polarization 301 Hartmann-Hahn 301 solid state 299-313 two-dimensional 311 non-photochemical hole burning 124 normal mode analysis 140 nuclear magnetic resonance, see NMR Hamiltonian 359

O ODMR 277-295 time-resolved 280 optical rotary dispersion (ORD) 25 optically detected magnetic resonance, see ODMR orientation, angle 17, 21, 22, 28 by electric field 20 by film-stretching 21 by flow 19 by gel-squeezing 19, 21 by magnetic field 20 mechanical 19 selection 263 orientation-dependent circular dichroism 26

410 oxygen diffusion 201 oxygen evolution 391 by photoacoustics 200-202 limiting rate 201-202 S-states 95, 200, 201 oxygen evolving complex 95, 99, 246, 334, see also Mn double hits 100 misses 95 Mn complex 95, 348, 350 S-states 95, 200, 201

P PDMR 278, 280, 282 periodic averaging 328, 330 phase contrast 329 phonon sideband holes 124 Phormidium laminosum 364, 367 phosphorescence 45, 278, 280, 284, 286, 288 microwave-induced 278, 285 phosphorescence-detected magnetic resonance, see PDMR, ODMR photoacoustic spectroscopy 34, 191-204 photobaric oxygen signal 402 photochemical hole burning 124 photon echo 111, see also accumulated photon echo photoselection 18, 20, 287. 290 photosystem I 7, 69, 87, 94, 169, 249, 290, 293, 327, 330, 333, 346-348, 355, 367-371 photosystem II 7, 94-105, 169, 249, 290, 293. 327, 333, 346348, 355, 366, 371 polarizability 178 polarization microscopy 24 polarized IR spectroscopy 19 light 13, 25 polarography 392, 395 Prochlorothrix hollandica 30 Prosthecochloris aestuarii 29, 126 protein crystallography 318 proton release pattern 96 transfer 137 psi-type aggregates 27, 30-33 pulsed microwaves 281 pump-probe techniques 138

Q quadrupole coupling 262 interaction 358 quinone 147, 157, 220, 365, 366 replacement 140

R R-factor 321 radical-pair mechanism 224 Raman spectroscopy 161-173 rate electrode 394 reaction center 118, 129-131, 152, 166-170, 268-271, 288-

Index 296, 302-309, 321, 365, 375, 383 regenerative amplifier 65 resonance Raman spectroscopy 161-173 Rhodobacter capsulatus 68, 71, 135 sphaeroides 68, 69, 71, 118, 125, 129, 133, 147, 155, 166, 169, 170-172, 180, 182, 184, 269, 271, 289, 291, 300, 303-309, 321, 364, 365, 388 Rhodopseudomonas acidophila 388 marina 32 palustris 166 viridis 8, 29, 129, 130, 133, 147, 155, 166, 169, 172, 271, 289, 291, 303-309, 321 Rhodospirillum 388 Rieske Fe-S clusters 348 rotational correlation time 260

S secondary structure 32, 144 sieve effect 4, 26 simulated annealing 322 single-photon counting 52 single-particle averaging 332 small-angle neutron scattering 375-389 space group 319 spectroelectrochemical cell 156 spin coupling 361 Hamiltonian 258, 357, 360, 361 label 398 polarization 212 Stark spectroscopy 177-188 step-scan interferometer 152 Stokes parameters 14, 15 shift 47 Synechococcus elongatus 330, 332, 367 Synechocystis 103, 333 structure factor 318 subtractive mixing techniques 155 supramolecular assemblies 375 synchrotron radiation 343

T T-S spectrum 278, 285-293, 295 TEMPONE 399 thermal deactivation spectrum 198 thermoluminescence 93-104 activation energy 97 application 103 assignments 100 components 101 origin 94 oscillation 99 peak temperature 96 time-averaged structures 323 transient absorption 70, 71 transition dipole moment, electric 15, 16, 278, 279, 286, 287, 290-295 magnetic 25

Index TRIPLE resonance, see ENDOR triplet state 277-298 magnetic field effect on 199 mechanism 213, 218 reaction center 289 spin Hamiltonian 278 triplet-minus-singlet absorbance difference, see T-S two-dimensional crystallization 330 tyrosine 71, 249, 303

U unit cell dimensions 318

V vibrational modes 141 spectroscopy 138 wavepackets 71

W water-oxidation system, see oxygen evolving system

411 X X-ray absorption near edge structure, see XANES absorption spectroscopy 337-352 diffraction 317-323, 326, 335 linear dichroism 16 XANES 338,346 K-edge spectra 338, 346

Z Zeeman frequency 258 zero-quantum beats 228--29 zero-field splitting parameters 277, 278-280, 289, 293-295, 359, 402 zero-phonon hole 124 line 118 ZFS, see zero-field splitting zirconium oxide detector 402