90 Structure and Bonding
Metal Sites in Protein and Models Redox Centres
Volume Editors: H. A. O. Hill, P. J. Sadler, A. J. Thomson
Springer
Berlin Heidelberg New York
The series Structure and Bonding publishes critical reviews on topics of research concerned with chemical structure and bonding. The scope of the series spans the entire Periodic Table. It focuses attention on new and developing areas of modern structural and theoretical chemistry such as nanostructures, molecular electronics, designed molecular solids, surfaces, metal clusters and supramolecular structures. Physical and spectroscopic techniques used to determine, examine and model structures fall within the purview of Structure and Bonding to the extent that the focus is on the scientific results obtained and not on specialist information concerning the techniques themselves. Issues associated with the development of bonding models and generalizations that illuminate the reactivity pathways and rates of chemical processes are also relevant. As a rule, contributions are specially commissioned. The editors and publishers will, however, always be pleased to receive suggestions and supplementary information. Papers are accepted for Structure and Bonding in English. In references Structure and Bonding is abbreviated Struct Bond and is cited as a journal.
Springer WWW home page: http://www.springeronline.com Visit the SB content at http://www.springerlink.com
ISSN 0081-5993 (Print) ISSN 1616-8550 (Online) ISBN-13 978-3-540-62888-0 DOI 10.1007/3-540-62888-6 Springer-Verlag Berlin Heidelberg 1998 Printed in Germany
Preface
Controlling the movement of electrons in time and space is vital to biology. How do proteins achieve this? In this volume (the third in a three-part special series on Metal Sites in Proteins and Models, volumes 88, 89 and 90) metals from all three transition metal series are highlighted: V, Mn, Fe, Ni and Cu from the first, Mo from the second and W from the third series. Without manganese and photosynthesis, and the conversion of water into dioxygen, there would not be abundant chemical energy! and Penner-Hahn critically reviews the evidence for Mn clusters and cofactors in the oxygenevolving complex of Photosystem II - does it contain one or two clusters.~ what are the Mn oxidation states? do we need to invoke the presence of additional metals.~ Copper, too, is sometimes found in clusters in proteins, for example the purple dicopper in cytochrome o oxidase, and trinuclear sites of some other copper oxidases; Messerschmidt shows how the structures of copper centres are related to their properties and to an overall classification of copper sites. He also discusses the first structure of a vanadium-containing protein, the enzyme chloroperoxidase, and its unexpected coordination geometry. (The wider chemistry and biochemistry of vanadium is reviewed in detail in Volume 89) An important role for molybdenum involves the catalysis of oxygen atom transfer reactions coupled to electron transfer between substrate and cofactors such as flavins, Fe/S centres and hemes. About 70 molybdenum oxotransferase enzymes are known, which are widely distributed amongst eukaryotes, prokaryotes, and archaea. They belong to the families: xanthine oxidase, sulfite oxidases and assimilatory nitrate reductases, and DMSO reductases. The first representative structure of a member of the xanthine oxidase family (aldehyde oxido-reductase from Desulfovibrio gigas) is described here by Romeo and Huber; it contains a catalytically essential Mo = S group, and a molybdopterin cofactor, both in close proximity to Fe/S centres, but curiously no direct Moprotein bonds. Hydrogenases (which catalyze the two-electron oxidation of H 2) are enzymes of enormous biotechnological interest, providing potential sources of clean energy, and being vital to a wide variety of bacteria. Frey describes the first 3D structure of a Ni-Fe hydrogenase with its unusual Fe site and three bound diatomic ligands, and a Ni site which can bind hydride. Capozzi, Ciurli and Luchinat analyze in detail how protein-induced modulations of metal-metal interactions allow the redox potentials of iron-sulfur centres to span a wide range of redox potentials, and elegantly demonstrate how individual iron
VIII
Preface
atoms in clusters can be distinguished. The remarkable versatility of iron sites is also addressed in Volume 88 of this series. Finally, Hagen and Arendsen argue that tungsten has a widespread catalytic role in nature. There is no doubt that is proving to be the case in certain bacteria, but will it also be true for eukaryotes.~ And if Mo and W have such roles, what about Cr? That topic will have to wait for a future volume! We hope you will enjoy reading these articles and find them as stimulating as we have. H. Allen O. Hill, Peter ]. Sadler and Andrew J. Thomson
Contents
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex ]. E. Penner-Hahn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidases and Vanadium-Containing Enzymes A. Messerschmidt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
Structure and Function of the Xanthine-Oxidase Family of Molybdenum Enzymes M. ]. Romeo, R. Huber . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
Nickel-Iron Hydrogenases: Structural and Functional Properties M. Frey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
97
Coordination Sphere Versus Protein Environment as Determinants of Electronic and Functional Properties of Iron-Sulfur Proteins F. Capozzi, S. Ciurli, C. Luchinat . . . . . . . . . . . . . . . . . . . . . . .
127
The Bio-Inorganic Chemistry of Tungsten W.R. Hagen, A. E Arendsen . . . . . . . . . . . . . . . . . . . . . . . . . .
161
Author Index Volumes 1 - 90 . . . . . . . . . . . . . . . . . . . . . . . . .
193
Contents of Volume 88 Metal Sites in Proteins and Models Iron Centres Volume Editors: H. A. O. Hill, P. ]. Sadler, A.]. T h o m s o n
Polyiron Oxides, Oxyhydroxides and Hydroxides as Models for Biomineralisation Processes A. K. Powell Heme: The Most Versatile Redox Centre in Biology.~ S. K. Chapman, S. Daff, A.W. Munro Rationalisation of Metal-Binding to Transferrin: Prediction of Metal-Protein Stability Constants H. Sun, M. C. Cox, H. Li, P. ]. Sadler Metal Centres of Bacterioferritins or Non-Heam-Iron-Containing Cytochromes b557 N. E. Le Brun, A. ]. Thomson, G. R. Moore Ribonucleotide Reductases - A Group of Enzymes with Different Metallosites and a Similar Reaction Mechanism B.-M. Sj6berg Protein Engineering of Cytochrome P450cam L.-L. Wong, A. C. G. Westlake, D. P. Nickerson
Contents of Volume 89 Metal Sites in Proteins and Models Phosphatases, LewisAcidsand Vanadium Volume Editors: H. A. O. Hill, P.J. Sadler, A. J. T h o m s o n
Advances in Zinc Enzyme Models by Small, Mononudear Zinc(II) Complexes E. Kimura, T. Koike, M. Shionoya Zinc Catalysis in Metalloproteases D. S. Auld Modeling the Biological Chemistry of Vanadium: Structural and Reactivity Studies Elucidating Biological Function C. Slebodnick, B. J. Hamstra, V. L. Pecoraro Vanadium Bromperoxidase and Functional Mimics A. Butler, A. H. Baldwin Metal Ions in the Mechanism of Enzyme Catalysed Phosphate Monoester Hydrolyses D. Gani, J. Wilkie The Dimetal Center in Purple Acid Phosphatases % Klabunde, B. Krebs
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex James E. P e n n e r - H a h n Department of Chemistry, 930 N. University Avenue, Ann Arbor, Michigan 48109-1055, USA
E-mail:
[email protected]
The photosynthetic conversion of solar to chemical energy is based on light-driven charge separation in a chlorophyll-based pigment. In higher-plants, the electrons required for this process are extracted from H20, ultimately producing 02 as a waste by-product of photosynthesis. The photosynthetic oxidation of water takes place at the oxygen evolving complex (OEC) on the donor (lumenal) side of Photosystem II. The OEC contains four Mn ions, together with calcium and chloride as essential inorganic cofactors. The techniques which have proven most useful in characterizing the nature of the OEC are X-ray absorption spectroscopy and EPR. Recent results from both techniques are reviewed. Key Words: Photosynthesis; oxygen evolution; manganese; X-ray absorption spectroscopy (EXAFS, XANES); electron paramagnetic resonance (EPR)
1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
1.1 1.2 1.3 1.4
Importance of Photosynthesis ................... Reactions in P h o t o s y s t e m II . . . . . . . . . . . . . . . . . . . . The Kok Cycle . . . . . . . . . . . . . . . . . . . . . Basic Elements o f P h o t o s y s t e m II . . . . . . . . . . . . . . . . .
2 3 4 5
2
Physical M e t h o d s for C h a r a c t e r i z i n g t h e OEC . . . . . . . . . . .
5
2.1 2.2 2.3
X-Ray A b s o r p t i o n S p e c t r o s c o p y . . . . . . . . . . . . . . . . . . X-Ray A b s o r p t i o n Near Edge S t r u c t u r e . . . . . . . . . . . . . . E x t e n d e d X-Ray A b s o r p t i o n Fine S t r u c t u r e . . . . . . . . . . . .
6 7 8
3
Structural C h a r a c t e r i z a t i o n o f t h e OEC . . . . . . . . . . . . . .
9
3.1 3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 3.2 3.3 3.3.1 3.3.2
Magnetic Properties . . . . . . . . . . . . . . . . . . . . . D i m e r I n t e r p r e t a t i o n s of the Multiline Signal . . . . . . . . . . . T e t r a m e r I n t e r p r e t a t i o n s o f the Multiline Signal . . . . . . . . . I n t e r p r e t a t i o n s o f the g ~ 4.1 Signal . . . . . . . . . . . . . . . . The $1 State EPR Signal . . . . . . . . . . . . . . . . . . . The "$3" State EPR Signals . . . . . . . . . . . . . . . . . . O x i d a t i o n State A s s i g n m e n t s . . . . . . . . . . . . . . . . . . . . Atomic Arrangements . . . . . . . . . . . . . . . . . . . . Mn-Nearest Neighbor Interactions ................ Mn-Mn Interactions . . . . . . . . . . . . . . . . . . . . .
......
. . .
. . . . . . . . . . . .
9 10 10 12 14 15 16 16 17 18
Structure and Bonding,Vo].90 © SpringerVerlag BerlinHeidelberg1998
2
James E. Penner-Hahn
3.3.3 3.4 3.4.1 3.4.2 3.4.3 3.4.4 3.4.5
Outer Shell Interactions . . . . . . . . . . . . . . . . . . . . . . S-State Dependence of the OEC . . . . . . . . . . . . . . . . . . TheS2 State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The $3 State - XANES . . . . . . . . . . . . . . . . . . . . . . . The $3 State - other measures of the oxidation state The $3 State - EXAFS . . . . . . . . . . . . . . . . . . . . . . . The So State . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4
Chemical Perturbations of the OEC . . . . . . . . . . . . . . . .
27
4.1 4.2 4.3
Inhibitors of Oxygen Evolution . . . . . . . . . . . . . . . . . . Modifications at the Chloride Site . . . . . . . . . . . . . . . . . Reduced Derivatives of the OEC . . . . . . . . . . . . . . . . . .
27 28 28
5
Structural Possibilities . . . . . . . . . . . . . . . . . . . . . . .
30
5.1 5.2 5.3
One vs. Two Clusters . . . . . . . . . . . . . . . . . . . . . . . . Arrangements of Mn Ions . . . . . . . . . . . . . . . . . . . . . Mn oxidation States . . . . . . . . . . . . . . . . . . . . . . . .
31 32 32
Conclusions and Future Prospects References
.......
................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19 21 21 22 25 26 27
33 33
Symbols and Abbreviations OEC XAS EXAFS XANES EPR ESEEM ESE-ENDOR PSII
oxygen evolving complex X-ray absorption spectroscopy extended X-ray absorption fine structure X-ray absorption near edge structure electron paramagnetic resonance electron spin echo envelope modulation electron spin echo detected electron nuclear double resonance Photosystem II
1 Introduction 1.1
Importance of Photosynthesis The ukimate source of most of the chemical energy used in life is solar energy, which is captured and converted to a chemically useful form by photosynthetic organisms. The high-energy compounds that are produced in this process are typically reduced hydrocarbons, with the required reducing equivalents coming, in the case of plants and cyanobacteria, from water (Eq. 1). The waste product of this reaction is oxidized water, i.e. 02. Oxygenic photosynthesis produc-
StructuralCharacterizationof the Mn Site in the PhotosyntheticOxygen-EvolvingComplex
3
ed most of the oxygen in the atmosphere, and in the process has dramatically akered the earth's environment. 2 H20 + light --> 02 + 4 H + + 4 e-
(1)
This chapter describes recent progress in characterizing the photosynthetic oxygen evolving complex (OEC). There have been numerous recent reviews of various aspects of the OEC (see for example [1], [2], [3], and [4]). The purpose of the present review is not to provide a repetition of earlier reviews but rather to provide a critical assessment of the available structural data, particularly recent results, and to use this to assess possible models for the oxygen evolving complex. 1.2 Reactions in Photosystem II
Oxygenic photosynthesis involves reactions at two separate photosystems, with oxygen evolution taking place at Photosystem II (PSII). The net reaction catalyzed by PSII is the transfer of electrons from water to plastoquinone. A schematic illustration of the principal polypeptides involved in Photosystem II and the routes of electron flow through the complex are illustrated in Fig. 1 (for details see [2]). The photochemical reaction at the heart of Fig. 1 is the charge separation that takes place at a chlorophyll pigment known as P680. The acceptor side of PSII is believed to be similar to that in the reaction centers of purple non-sulfur bacteria [5], with the electron flowing through a pheophytin to the plastoquinone QA and then eventually to a second exchangeable plastoquinone Qb. A second charge separation event then reduces Qb to the hydroquinone
(28
"/
CP4;
1
zl''N 4H÷+
Fig. 1. Schematicillustration of the polypeptides of PhotosystemII. Arrows indicate the electron flowfrom (top half) and to (bottom half) P680followingexcitation
4
]ames E. Penner-Hahn
form, which exchanges with the oxidized plastoquinone pool. On the donor side of P680,tyrosine Yz reduces P~80,resetting the system for another charge separation. Yz" is, in turn, reduced with an electron from water, via the Mn in the OEC. From Fig. 1 it is apparent that a basic problem in oxygenic photosynthesis is how to couple the one-electron charge separation reactions of P680to the fourelectron oxidation of water. 1.3 The Kok Cycle
The broad outlines of the solution to this problem were first recognized with the demonstration by ]oliot and co-workers [6] that photosynthetic membranes which are illuminated with short (ca. 5 ps) but saturating flashes of light evolve oxygen maximally every fourth flash, starting with the third flash. This demonstrated that each reaction center acts independently to acquire four oxidizing equivalents in a sequential manner and led Kok and co-workers to propose [7] the model shown in Fig. 2. This model, which has guided nearly thirty years of research into photosynthesis, is based on four kinetically resolvable intermediates known as Sn states, where the subscript refers to the number of stored oxidizing equivalents. The dark stable state of the OEC is the S1 state. Three flashes oxidize this to the transient $4 state, which evolves 02, regenerating So and beginning a new period 4 oscillation. Although $4 is postulated, only So through $3 have been observed spectroscopically. After a number of flashes, the period 4 oscillations are gradually damped. This damping is generally attributed to a combination of double hits (advance of two rather than one S state) and misses (no productive photooxidation). Recently, Shinkarev and Wraight [8, 9] have noted that a complete discussion of charge separation, and in particular the miss fraction, must consider both the donor side (the OEC Sn states) and the acceptor side (the quinones). The involvement of both a non-exchangeable quinone (QA) and an exchangeable quinone (Qb) (see Fig. 1) means that there will be a period two oscillation on the acceptor side, with different probabilities of charge separation depending on whether the acceptor is oxidized or reduced. This two-fold oscillation is superimposed on the four-fold oscillation resulting from the donor side (S-state) cycling. The im-
[S4]
4H+~
t
+j
I
% Sl
;;rP68°'~--'/
} --" 2H20
",,.. s 2/
3 "~,,,.,..,.~..,,,~ S
Fig. 2.
Kok cycle for oxygenic photosynthesis. The S1 state is the dark-stable state. The S 4 state is unstable with respect to production of 02 and the SOstate. Solid lines show changes in the OEC, dashed lines indicate electron flow
StructuralCharacterizationof the Mn Sitein the PhotosyntheticOxygen-EvolvingComplex
5
portant consequence of the acceptor side cycling is that the probability of S state advance (i.e. the miss fraction) shows a pronounced 2-fold periodicity. When all of the kinetic and thermodynamic factors are considered [8], the apparent miss fraction, particularly on the $2-~ $3 and $3 ~ So transitions, can be quite large. 1.4 Basic elements of Photosystem II
Initial work on the OEC used thylakoid membranes. An important advance came with the development of the so-called BBY preparation [10, 11], which removed most of the contaminating polypeptides. The essential elements of PSII are illustrated in Fig. 1. Peptides D1 and D2 are homologous with the L and M subunits of bacterial reactions centers [2]. These bind the primary redox cofactors P680,QA, Qb, and Yz, together with a non-heme iron bound between QA and Qb and a second redox-active tyrosine YD. In addition, D1 and D2 probably provide some of the ligands to the Mn and/or Ca ions. The CP43 and CP47 intrinsic membrane polypeptides are part of the light-harvesting antennae, peptides a and fl bind cytochrome b559, and the 28 kDa peptide plays a role in Qb binding [2]. In addition to these intrinsic transmembrane polypeptides, there are three extrinsic polypeptides, with molecular masses of approximately 17, 24, and 33 kDa. The 33 kDa polypeptide helps to stabilize the Mn, while the two lighter extrinsic polypeptides enhance the binding of Ca 2+ and C1-.A more highly purified preparation of the OEC, known as the reaction center complex (RCC) [12], lacks the 17 and 24 kDa extrinsic polypeptides. This preparation requires added Ca 2+ and C1- for activity, but appears to be spectroscopically similar or identical to BBY preparations. An important advantage of RCC preparations is that they have an approximately 3-fold higher Mn concentration than BBY preparations, and consequently give spectra with a better signal/noise ratio. Although there have been other suggestions [13], it is generally accepted that the OEC contains four Mn ions [2]. Full activity also requires at least one Ca2+ and one C1- ion per reaction center. Among the outstanding questions regarding the OEC are the nuclearity of the Mn cluster; the geometry of the cluster; the ligands to the Mn and Ca ions; the role of the inorganic cofactors; and the oxidation state of the Mn. As discussed below, the answers to many of these questions remain in dispute, although in some cases a consensus is beginning to emerge. 2 Physical Methods for Characterizing the OEC
Characterization of the OEC is inherently difficult since the portions of interest (Mn, C1, Ca) are elements that are generally hard to study. Most of the techniques imaginable have been turned to OEC at one time or the other. Some are indirect (e. g. thermoluminescence), some are difficult to attribute directly to the Mn/Ca/C1 cluster (FTIR) and some are extremely insensitive
6
James E. Penner-Hahn
(UV-vis). When they have been used, many of these techniques have been (and in some cases continue to be) controversial, with very different conclusions drawn from similar data. In many cases, a generally accepted consensus of which results and which interpretations are reliable has not yet been established. This review focuses on the application of magnetic resonance and Xray absorption spectroscopy to characterization of the OEC. The attraction of these methods is that they are highly specific either to paramagnetic centers or to a specific element (e.g. Mn). Most of the recent structural models that have been proposed for the OEC have been based on results from magnetic resonance and X-ray absorption. Before reviewing these results it is worth reviewing the limitations of these techniques. For magnetic resonance, the limitations are straightforward - it is only possible to obtain information about paramagnetic centers, and, perhaps more importantly, not all paramagnetic centers give rise to detectable magnetic resonance signals. With X-ray absorption, the limitations, most of which are related to the fact that X-ray spectroscopy is really a spectroscopically detected scattering method as opposed to a more conventional spectroscopy, are somewhat more subtle and warrant additional discussion.
2.1 X-Ray Absorption Spectroscopy X-ray absorption spectroscopy (XAS) refers to the structured absorption on the high-energy side of an X-ray absorption "edge" (the abrupt increase in absorption resulting from excitation of a core electron) [14-16]. For Mn, the "K" (ls initial state) edge occurs at ca. 6500 eV ~ 2 ~. X-ray absorption spectra are frequently divided into the X-ray absorption near edge structure (XANES) region, for structures within ca. 50 eV of the edge, and the extended X-ray absorption fine structure (EXAFS) region, for structures from -50 eV to 1000 eV above the edge. The structure in X-ray absorption spectra is due to the modulations of X-ray photoabsorption cross-section that occur as a consequence of interference between the X-ray excited and the back-scattered photoelectron waves. The principal attraction of XAS, particularly in the context of the OEC, is that it is able to provide element-specific structural information about non-crystalline systems. Unlike magnetic resonance, XAS spectra are always detectable, regardless of the spin-state of the sample. This is, however, also a potential weakness. All of the metal in a sample (e.g. all of the Mn in the OEC) contributes to the observed XAS spectrum. This makes it difficult, if not impossible, to determine the structure of each of the metal sites in a protein. Instead, only the average structure can be determined. The physical principles responsible for XAS structure are the same regardless of whether one is concerned with the XANES or the EXAFS regions. Although this equivalence has been recognized for some time, it has continued to be useful to maintain a distinction between XANES and EXAFS based on the fact that the lower-energy XANES structure is more sensitive to the geometry and the long-range structure of the absorbing site. This is a conse-
StructuralCharacterizationof the Mn Sitein the PhotosyntheticOxygen-EvolvingComplex
7
quence of the longer mean-free-path of the X-ray-excited photoelectron at low photoelectron kinetic energies. In contrast, the EXAFS region can typically be modeled by considering only the simpler, short-range structure. With modern ab initio theory, EXAFS spectra can be modeled to a fairly high degree of precision [17, 18]; similar treatments of the XANES region are not yet possible. 2.2 X-Ray Absorption Near Edge Structure
It is generally accepted that the energy at which an X-ray absorption edge occurs is correlated with the oxidation state of the absorbing atom, with higher oxidation states giving higher energy edge thresholds [19-21]. The physical basis of this effect is open to debate, but probably involves both electrostatic effects (edge energy increases with increasing nuclear charge as it becomes more difficult to ionize the core electron) [19] and continuum resonance effects (edge energy.~ 1/R2 where R = absorber-scatterer distance) [22-24]. It is extremely difficult to separate these effects, since higher oxidation state complexes typically have shorter bond lengths, and are thus expected to have higher energy edges according to either interpretation. The significance of the fact that bond length can affect edge energy is that this has the potential to confound oxidation state assignments: If a complex has unusually short bond lengths, it might have a higher edge energy than would otherwise be expected. Despite the uncertainty in interpretation, and the potential for misleading results, there is ample empirical data which demonstrate an excellent correlation between XANES energy and oxidation state [19, 20, 25]. One of the difficulties with XANES analyses is that it is hard to define exactly what is meant by"the" edge energy. Most XANES spectra consist of numerous poorly resolved transitions and cannot really be said to have a single unique edge energy. A variety of definitions have been used, including the energy at half height [26, 27] and the energy of the first inflection point, as defined either by the maximum [28] in the first derivative or the zero-crossing [29] in the second derivative of the XANES spectrum. All three of these definitions suffer from the fact that they are extremely sensitive to minor variations in the shape and/or resolution of the XANES spectrum. This is illustrated by the simulation in Fig. 3. The spectra in Fig. 3 are the XANES for a Mn(III) model compound subjected to various amounts of broadening. This could represent either instrumental (monochromator) resolution or broadening due to chemical heterogeneity. It is apparent that relatively small changes in resolution can result in large changes in the energy of the first inflection point. An alternative approach to XANES analysis is to fit a spectrum using linear combinations of spectra drawn from a library of model compounds of known oxidation state and structure [30]. Such analyses are substantially less sensitive to minor variations in shape, but suffer the disadvantage of being dependent on the library-of model compounds that are used to fit the unknown spectrum.
8
James E. Penner-Hahn 4020-
-1200 -
1000
-800~>
c~
-20~ ~'~-40!
600~ -400 ~' •200 •
-60i -801
...........................
6530
6550
6 5 7 0 6590 eV Fig. 3. Effect of spectral resolution on the XANES for an Mn model complex. XANES data are in bold (right hand scale), second derivative of the XANES is in light (right hand scale). The edge energy is the negative-going zero crossing in the second derivative at ca. 6554 eV. As resolution improves, new first inflection points are resolved. Even if the same inflection points are followed, the energy shifts by ca. 0.5 eV with resolution
2.3 Extended X-Ray Absorption Fine Structure EXAFS spectra contain oscillations from all of the neighbors around the absorbing atom, with each neighbor contributing oscillations of a characteristic frequency 2R when spectra are plotted in k space (k = 2~-mme (E-Eo)/t~ 2 is the photoelectron wave vector, Eo is the threshold energy for photoelectron excitation). Spectra are often shown as their Fourier transform, so that each "shell" of scatterers shows a peak at R + a, where a is a phase-shift due to the photoelectron scattering. Here a "shell" refers to a group of atoms of similar atomic type (e.g. N or O) at the same distance from the absorbing atom. It is generally accepted that EXAFS can be used to determine bond lengths with an accuracy of 0.02 ,~, coordination numbers to one atom in four or five, and atomic numbers to within one row of the periodic table [15, 16], at least for analyses of firstcoordination shells. Somewhat lower accuracy is generally obtained for more distant atoms. An important limitation of EXAFS concerns the resolution. Like all scattering methods, the ability of EXAFS to resolve two shells is limited by the range of the data. This limitation is particularly severe for the Mn edge, since there are at least two endogenous Fe atoms in the OEC (see Fig. 1). The presence of the Fe K absorption edge (7100 eV) limits the accessible Mn EXAFS range to kmax < _ ~ 11.9 Jr-.1 This means that, under the best conditions, the closest distance spacing that can be resolved is AR/2Ak ~ 0.14 A. In practice, the true resolution limit is generally somewhat worse than this [31]. A second consequence of the limited range of OEC EXAFS spectra is that only a limited amount of information is available. Because EXAFS spectra are band-limited, the amount of information contained in a spectrum is determined by the range of the spectrum (Ak) and by the range of distances that contribute to the spectrum (AR) [32, 33]. The number of independent data points, Nidp = 2 AkAR/n is typically <, 15. This n u m b e r can drop dramatically for fits to
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex
9
Fourier"filtered" data, where a back Fourier transform is used to isolate a single peak from the overall radial distribution [31 ].
3 Structural Characterization of the OEC 3.1 Magnetic Properties A key breakthrough in understanding the OEC was the discovery that the S2 state is EPR-active [34]. Illumination at 200 K, which limits the OEC to a single S-state advance, gives a multiline EPR signal centered at g ~-2 with 18-20 resolved lines spread over ca. 1900 G (for reviews, see [35, 36]). This signal is generally similar to those seen for mixed-valence Mn dimers, although the signal for the OEC is considerably broader. It was subsequently found that different conditions (e.g. low-temperature illumination or the presence of inhibitors such as fluoride or ammonia) gave a broad EPR signal at ca. g = 4.1 rather than the multiline signal [37, 38]. Although initially controversial, there is now general agreement that both the mukiline signal and the g-~ 4.1 signals arise from ground states (e.g., [39] and references therein). Representative multiline and g~4.1 spectra are shown in Fig. 4. Several forms of the OEC showing modified multiline signals have been prepared. Modifications include binding of NH 3 at a C1--independent site [40], replacement of Ca 2÷ with Sr 2÷ [41], and Ca 2÷ depletion coupled with binding of chelators [42, 43]. Finally, it has been found that samples prepared in sucrose with no added alcohol show not only the g ~ 2 multiline signal, but also excited state EPR signals, a different g ~ 4.1 signal [44] and a new g ~-6 signal [39], attributed to low-lying excited spin states of the cluster that gives rise to the mukiline signal.
Ca2+
Sr 2+
. . . . . . slgIlal multlllnC
.
g=4.1 signal
500
1600 1500 2600 2500 3600 3500 4600 4500
Magnetic Field (Gauss) Fig. 4. Examples of multiline and g = 4.1 EPR signals. Data shown are light minus dark
spectra for native and Sr2+-substituted samples. Redrawn from data in [87]
10
James E. Penner-Hahn
A large amount of information is encoded within the highly structured multiline signal. At least in principle, this should allow one to determine the magnetic and electronic properties of the cluster that give rise to the multiline signal. In practice, attempts to simulate the multiline signal have not produced a consensus regarding its origin. It was clear from the beginning that the presence of 18-20 lines, attributable to Mn hyperfine coupling, required that at least two, and possibly all four, Mn ions be magnetically coupled to the paramagnetic center that accounts for the multiline signal. 3.1.1
Dimer Interpretations of the Multiline Signal Initial simulations of the multiline signal focused on mixed-valence Mn dimers. It has been claimed [45, 46] that the observed spectral breadth can only be explained by invoking a tetranuclear cluster. More recently, it has been reported that the multiline EPR signals at S-, X- and Q-bands can all be modeled using a single set of parameters for an isolated Mn(III)/Mn(IV) dimer [47]. However, in order to account for the ca. 1900 G width of the multiline signal, the dimer simulations required large hyperfine anisotropy and extremely large nuclear quadrupolar couplings. These are much larger than values typically seen in model compounds, and more importantly, do not appear to be consistent with the Mn nuclear spin properties of the OEC, as probed directly by ESE-ENDOR [48]. A second problem with the dimer explanation is that the multiline signal arises from a cluster that has relatively low-lying excited spin states (either ca. 35 cm -1 in the presence of alcohols [49, 50] or only a few cm -1 in the absence of alcohols [39]). Within the dimer model, the excited spin states are explained by exchange coupling of Mn(III) (S = 2) and Mn(IV) (S -- 3/2), giving rise to a ladder of spin states ranging from Stota 1 = 1/2 to Stota 1 -~ 7/2. However, all known examples of di-p-oxo bridged Mn(III)/Mn(IV) dimers (see below for discussion of OEC structure) have ground and excited state splittings of ca. 200-300 cm -1. Is this a death knell for dimer models of the OEC multiline signal? It is perhaps possible that not all of the ENDOR intensity is seen or that there are other dimer simulations that would be compatible with the ENDOR results. However, the ENDOR results, together with the unusually small exchange coupling that is required to explain the temperature dependence make a strong case that an isolated Mn dimer cannot give rise to the multiline EPR signal. 3.1.2
Tetramer Interpretations of the Multiline Signal Recently, Zheng and Dismukes have described a detailed simulation of the multiline EPR signal [51]. These simulations were the first to make use of the experimental anisotropy in Mn hyperfine coupling. Simulations using a distorted tetrahedral magnetic coupling model (Fig. 5a) were, like the dimer simulations, able to account for both the X- and Q-band spectra, although in this case hyperfine parameters taken from structurally characterized models were used. To simplify the calculations, the interdimer couplings were assumed to be equal
11
StructuralCharacterizationof the Mn Sitein the PhotosyntheticOxygen-EvolvingComplex
334 J1/.t114/~ /j2 4.Mn4 Mnl Mn2
Mnl
,J Mn2 ,
J1z
JIZ
A
B
J34
3 Mn1
J12
Mn2
C
Fig. 5. Some of the possible magnetic coupling schemes for "dimer of dimers" models of the OEC [51].MnlMn2and Mn3Mn4are the two "dimer" pairs. A) Distorted tetrahedral coupling. B) Dimer of dimers couplingthat is equivalent to A) if J13= ]14= J23 -- J24- C) Magnetic coupling scheme with only a single interdimer interaction
(Ji3 = J14 = J23 = J24 = J). T h e H a m i l t o n i a n for this coupling s c h e m e is isomorphic with the simpler "dimer of dimers" Hamikonian (Fig. 5b). These models define a spin basis set [$12,$34; ST>, where $12 and $34 are the effective dimer spins, and ST is the total spin. The authors were able to obtain good fits to the native and the NH3-modified multiline signals signal using [7/2, 4; 1/2> and [7/2, 3; 1/2> basis sets, respectively. An important conclusion is that, if the simulations are limited to normal Mn hyperfine couplings, the ca. 1900 spectral width can only be explained by invoking large dimer spins (e.g., $12=7/2, $34=3,4). This appears to rule out pairwise coupling schemes such as that in Fig. 5c, where there is interdimer coupling between only one Mn of each dimer [51]. Like the dimer simulations, the distorted tetrahedral simulations appear to be inconsistent with other data about the OEC. They require a Mn(III)3Mn(IV) oxidation state while other data (see below) favor a Mn(III)Mn(IV)3 oxidation state for the $2 state. They require that the sign of the hyperfine anisotropy for Mn(III) be reversed from that observed in oxo-bridged Mn(III) model compounds, implying either trigonal bipyramidal or a very unusual tetragonally compressed Mn(III) environment; as well as ferromagnetic intradimer exchange coupling, to give 812 = 7/2 and $34= 3 or 4. There are no examples of ferromagnetic coupling in di-p-oxo bridged dimers. However, in a tetrahedral coupling scheme (Fig. 5 a) apparent ferromagnetic intradimer coupling can result, if the interdimer couplings are larger than the intradimer coupling: 1/4(J13+ J23+ J14+ J24) >[J34[. This condition also seems unlikely. The 1- 2 and 3- 4 dimers are generally considered to have di-poxo bridges (see below), which, in the model compounds made to date, are strongly antiferromagnetically coupled. The $12= 7/2 ground state would thus appear to require unexpectedly large interdimer coupling. Zheng and Dismukes suggest that a possible structural interpretation of their EPR model might be that the Mn ions in the dimers are bridged by hydroxide rather than oxide ligands. The EXAFS data (below) are probably inconsistent with this structure. The ability to simulate two different multiline signals with only small changes in coupling [51] is an important advance in understanding the OEC. These simulations were not able to account for the EPR spectrum observed in Ca2+-depleted, EGTA-treated samples. This could indicate that the Ca2+-deplet -
12
lames E. Penner-Hahn
ed samples have a unique structure, or could reflect the need to adopt a different coupling scheme. In addition, it is not clear that the coupling models in Fig. 5 can account for the existence [39] of excited spin states lying as close as a few cm -1 above the ground state. Although both the dimer simulation [47] and the dimer or dimers simulations [51] can account for the observed EPR spectra, neither appears consistent with other known properties of the OEC. The availability of several different multiline spectra, 55Mn ENDOR data [48], experimentally calibrated estimates of Mn hyperfine anisotropy [51, 52], and careful characterization of excited spin-state energies [39, 49] provide stringent criteria against which future multiline simulations will be judged. In addition, EXAFS data (see below) limit the range of possible OEC structures. 3.1.3 Interpretations of the g = 4.1 Signal
The g = 4.1 signal was originally attributed to an S = 3/2 mononuclear Mn(IV). However, the discovery [53] that the g ~ 4.1 signal for NH3-inhibited, oriented OEC multilayers shows resolved multiline structure demonstrated that at least this form of the g = 4.1 signal must also come from a multinuclear Mn cluster. Based on an analysis of the signal shape, g = 4.1 has been assigned to the middle Kramers transitions of a rhombic S = 5/2 state [54]. More recently, it has been suggested that an S = 3/2 description is required to account for the multiple-frequency EPR spectra of this signal [39]. Since the g = 4.1 signal arises from a ground state [50] and since an Mn(III)/Mn(IV) dimer (see below for oxidation state assignments) would have either an S = 1/2 or S = 7/2 ground state, the g = 4.1 signal must arise minimally from an Mn trimer. Since the g = 4.1 signal is produced at low temperature and converts to the multiline signal at higher temperatures, an early interpretation of the g--4.1 signal was that it represents a precursor to the multiline signal. This precursor state could be either a redox intermediate (Fig 6.I) or a conformational intermediate (Fig 6.II). Recently, Rutherford and co-workers [55] have been able to separate formation of the multiline from formation of the g = 4.1 signal. They found that near-IR (ca. 820 nm) radiation specifically converts the multiline form of $2 to the g -- 4.1 form. Since most earlier illumination studies used white light, samples were exposed to both visible and IR components. The near-IR result might be consistent with Figs. 6.I and II, if it were possible to drive the reaction backwards using near-IR light. However, Rutherford and co-workers also demonstrated that it is possible to prepare samples in the multiline $2 state using 130 K illumination, if no near-IR radiation is present. This demonstrates that the g = 4.1 form is not an intermediate between $1 and the multiline form of $2, but does not distinguish between a redox conversion (Fig. 6.III) and a conformational conversion (Fig. 6.IV) interpretation of the near-IR transition. The near-IR conversion shows a maximum at 150 K. Above 200 K there is no net conversion due to rapid relaxation of the g = 4.1 form back to the multiline form. Below 100 K, the conversion does not proceed, suggesting either that it is an excited state of the species giving rise to the multiline signal that is converted
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-EvolvingComplex
1 T
A
130 K illumination>
Aox
200 K annealing>
B
c
130 K illumination >
multiline
Cox
200 K annealing>
g=4.1
A
visible light
III B
820 nm, 150 K>
A
ZOO K annealing
Box
< B
g=4.1
multiline
IV
c
Co: multiline
Aox
>
A Box
g=4.1
II
13
visible light
820 nm, 150 K> >
Cox (
multiline
200 K annealing
Cox
g=4.1
Fig. 6. Possible interpretations of the production of multiline and g = 4.1 EPR signals for $2. I and II are the initial proposal based on the observation that formation of a g = 4.1 signal appeared to precede formation of a multiline signal; III and IV are modified models in the light of the recent finding that the multiline signal precedes the g = 4.1 signal [55]. Models I and III assume that the multiline and g = 4.1 signals originate from oxidation of different sites (A and B, respectively). Models II and IV assume that the mukiline and g = 4.1 signals originate from different conformations of a single cluster C (C and C', respectively)
by IR radiation, or that there is an activation barrier which m u s t be overcome in the conversion. If it is an excited state that is responsible for the IR conversion, this state m u s t lie significantly higher than the ca. 35 cm -1 excited spinstate found in relaxation studies [49]. The action s p e c t r u m of the multiline -+ g = 4.1 conversion has a m a x i m u m at about 820 n m [55]. The authors assigned this to an intervalence transition, converting for example an Mn(III)/Mn(IV) dimer to an Mn(IV)/Mn(III) dimer. The valence isomer might either be responsible itself for the g = 4.1 signal or might trigger a further redox reaction. Thus, if an Mn(III)/Mn(IV) dimer represented Aox in Fig. 6.III, the valence isomer Mn(IV)/Mn(III) might be able to oxidize B. Equivalent schemes can be drawn for possible valence isomerization within a single cluster (Fig. 6.IV), e.g. M n ( I I I ) / M n ( I V ) - M n ( I V ) / M n ( I V ) --> M n ( I V ) / M n ( I I I ) - M n ( I V ) / M n ( I V ) , possibly with further conversion to Mn(IV)/ Mn(IV)-Mn(III)/Mn(IV). Interestingly, low-temperature near-IR irradiation was able to convert the native multiline, the Sr2+-substituted multiline and the Ca2+-depleted, chelator-
14
James E. Penner-Hahn
treated multiline forms to g 4.1 forms, but was not able to convert the N H 3m o d i f i e d multiline to a g = 4.1 form [55]. This suggests that the species giving rise to the NH3-modified mukiline is in some way unique and different, perhaps in structure or ligation, from the other multiline forms. This is in contrast to the conclusion, based on multiline simulations [51], that it was the Ca2+-depleted form that appeared to have a unique structure. =
3.1.4
The ST State EPR Signal Although the $1 state is EPR-silent under normal conditions, it has been reported to have an EPR signal centered at g = 4.8 with a peak-to-peak width of ca. 600 G [56]. This signal, which is consistent with an integer spin, S = 1 state [57, 58], is seen only in parallel polarization EPR. The $1 EPR signal converts to the multiline EPR signal on illumination at 200 K. However, the most striking observation about the $1 EPR signal is that low-temperature illumination gave the g=4.1 S2 EPR signal (ca. 30% of maximum intensity, as judged by the extent of conversion to multiline on annealing at 200 K) without significant loss of the S1 EPR signal. If this observation is correct, it requires that the g = 4.1 $2 signal and the multiline $2 signal come from different clusters, not different conformations of a single cluster. The S1 EPR resuks were originally attributed to intermediacy of the g = 4.1 species between the oxidant (YZ.) and the multiline state [56]. With the discovery (above) that low-temperature production of the g = 4.1 signal is driven by a near-IR transition [55], this model requires modification. The scheme shown in Fig. 7 would be consistent both with the near-IR results and the observations made on the S~ EPR signal. In this scheme, A is the center that gives rise to the S~ EPR signal when reduced and the multiline signal when oxi-
30% >
o
>
A OX
St EPR )-
AOX
multiline
)
multiline
B
A S 1 EPR
B
g~.~. l
200 K annealing
visible
near-IR
light, 130 K
light, 130 K
200 K illumination
B
A 70%
S 1 EPR
B
A Sj EPR
B
Aox multiline
B
Fig. 7. A possible model for OEC redox conversions, incorporating results of both near-IR
conversion [55] and $1 EPR measurements [56] (see text for details)
StructuralCharacterizationof the Mn Site in the PhotosyntheticOxygen-EvolvingComplex
15
dized, while B is the center that gives the g = 4.1 signal when oxidized. The production of only 30% oxidized centers following 130 K illumination is due to competition from other, non-manganese electron donors at low temperature [591. More recently, it has been reported, based on measurements of spin-lattice relaxation rates for the YD" tyrosine radical, that the $1 state of the OEC is diamagnetic [60]. This would appear to be inconsistent with the existence of an EPR signal for $1. However, the relaxation measurements were made on longterm dark adapted samples, thus it is possible [4, 60] that a paramagnetic form of $1 (which gives rise to the $I EPR signal) converts, on long-term dark adaptation, to a diamagnetic resting form. 3.1 .S The "$3" State EPR Signals
A variety of treatments, most involving modification of the Ca :+ site, the C1site, or both, inhibit water oxidation by the OEC. When inhibited samples are illuminated they show a novel broad (100-230 G) radical EPR signal centered at g = 2.0. This effect was originally discovered for Ca2÷-depleted samples [42], but has subsequently been found [61-63] to result from a variety of inhibitory treatments, including C1- depletion, and fluoride, acetate, or ammonia treatment (which may operate by C1- displacement). A variety of lines of evidence, including flash yield and the ability to regenerate $2 on readdition of the disrupted cofactor (cf. [2, 64] for reviews) suggest that the new signal arises from a state that is one electron more oxidized than $2, thus making this formally an S3 EPR signal. The new signal has very limited stability at room temperature (hence the need to freeze under illumination in some of the preparations) and has limited stability even at low temperature [64]. The new signal lacks the breadth and hyperfine structure that would be expected for an Mn-centered paramagnetic center, but is much broader and much harder to saturate than would be expected for an isolated organic radical. It is generally agreed that the new signal can be explained as an organic radical that is dipole-coupled to a fast-relaxing paramagnetic center, the latter presumably being the OEC Mn cluster. This is supported by the observation that formation of S3 leads to loss of the S2 multiline signal as detected by cw EPR, but not by field-swept spin-echo EPR [65]. The new 100-230 G signal can thus be described as being due either to $3 or to S/X +, where X ÷ is the organic radical. Although the distinction is to some extent only semantic, the S3 designation implies that the state giving rise to the EPR signal is on the normal path to water oxidation, albeit altered in some way that makes it EPR-detectable. In contrast, the $2X÷ designation implies that the oxidation of X is an artefact of the inhibitory treatment, and is not necessarily relevant to the normal functioning of the OEC. It has been suggested, based on UV-visible difference spectra, that oxidized histidine is the radical responsible for the new EPR signal [66]. This could be consistent with either the $3 or the S2X+ designations, although a specific role for His- in water oxidation has been strongly favored by some authors. There have, however, been several suggestions that His. is not responsible for the "$3"
16
James E. Penner-Hahn
signal [67]. Recently, Britt and co-workers have reported ESEEM spectra for the "$3" signal that is produced in acetate-inhibited OEC samples [68]. By using a cyanobacterial (Synechocystis) source for the OEC, the authors were able to specifically deuterate the tyrosine residues, and to provide conclusive evidence that a tyrosyl radical is responsible for the "$3" EPR signal. Since Yz is the tyrosine that is close to the Mn cluster, at an estimated distance of 4.5 A [69], this is presumably the tyrosyl radical that is dipole-coupled to the Mn cluster. A more appropriate designation for the state that gives rise to the radical signal is thus S2Yz', a state on the normal path to water oxidation. The structural change that gives rise to the new signal appears to involve some modification of the Mn site so that it can no longer be oxidized byYz.. On removal of the inhibition (e.g., addition of Ca2+), the Yz" can oxidize the OEC to $3, thus continuing the Kok cycle. 3.2 Oxidation State Assignments It is only possible to have an S = 1/2 EPR signal for $2 if the cluster that gives rise to this signal contains an odd number of half-integer spin ions. For Mn, the only commonly available oxidation states with an odd number of electrons are Mn(II) and Mn(IV). The structural evidence (see below) that the OEC contains Mn(p-O)2Mn units, and the observation that di-p-oxo bridged structures are only found for Mn(III) and Mn(IV), has led to the suggestion that the $2 state must have an oxidation state of either Mn(III)3Mn(IV) or Mn(III)Mn(IV) 3. If S~ is one electron more reduced than $2 (see Sect. 3.4), this implies that $1 is either Mn(III) 4 or Mn(III)2Mn(IV)2. Several groups have used XANES to determine the oxidation state of S1, and the consensus is that the XANES data for S1 are most consistent with the latter oxidation state assignment [4, 26, 29, 70]. In model studies, it is relatively difficuk to distinguish between Mn(III) and Mn(IV) XANES spectra, such that the uncertainty in absolute (although not necessarily relative) oxidation state assignments is ca. +25% [70]. Although relatively poor, this precision is sufficiently good to distinguish unambiguously between average oxidation states of 3.0 and 3.5. It is perhaps worth recalling, however, that the ability to reduce the oxidation state question to one of distinguishing between average oxidation states of 3.0 and 3.5 is based on the assumption that only Mn spins contribute to the overall spin of the OEC. If another paramagnetic center was coupled to the cluster (as proposed e. g. in [39]), then it would be possible, from a spin-counting perspective, to have S~ as either Mn(III)3Mn(IV) or Mn(III)Mn(IV) 3. The best fit to the OEC XANES gives an average oxidation state of 3.5 + 0.25 for $1 [70], which could also be consistent with either of these possibilities. 3.3 Atomic Arrangements The Fourier transform of a recent EXAFS spectrum for the S1 state of the OEC is shown in Fig. 8. This spectrum is dominated by three peaks, correspond!ng to three principal shells of scatterers. The first two of these, at R + a = 1.6 A and
17
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex o
12
'
I
'
'
'
I
'
'
'
I
I
Mn-O nearestneighbors Mn-Mn at = 2.7 A
:18 ~ ~
4
•
2
~
0
I
0
2
Mn-Mn at = 3.3 A
|
I|
New interaction.*
4 6 Radius + a (/~_)
8
10
Fig. 8. Fourier transform of the EXAFS spectrum for the $1 state. Redrawn using data from [74]
2.3 A, are due to Mn-nearest neighbor interactions at an average distance of 1.9 A and Mn-Mn interactions at ca. 2.7 ~. These peaks were seen in the first EXAFS measurements of the OEC [71], and have dominated the structural models that have been proposed. The 2.7 A Mn-Mn distance is typical of the Mn-Mn distances found in di-p-oxo bridged Mn dimers, suggesting that this unit is a basic building block of the OEC Mn cluster. The distance for the Mnnearest neighbor peak is typical of Mn-O and Mn-N ligation (which cannot be distinguished by EXAFS), and is much shorter than Mn-S or Mn-C1 distances, thus suggesting that the Mn in the OEC must be ligated primarily by O- or Ncontaining ligands. Since only a single histidine nitrogen is seen in ESEEM measurements [72, 73], it is generally assumed that the Mn ions have predominantly oxygen ligation (oxo, hydroxo, aqua, carboxylato and possibly phenoxo or alkoxo ligands). 3.3.1
Mn-Neorest Neighbor Interactions
If the nearest neighbor interactions are modeled by a single Mn-O shell, the best fit is with ca. 3 oxygen atoms at an average distance of ca. 1.9 A [74]. This is clearly not a complete description of the nearest neighbor environment for at least two reasons. Firstly, the apparent coordination number is much too small. Secondly, if there are Mn(p-O)2Mn dimers in the OEC, there must be Mn-oxo bonds, which are exl~ected to be about 1.75/~ long, and there must be longer Mn-O bonds at ca. 2 A. This recognition has led many groups to analyze the first shell EXAFS in terms of two shells of scatterers [75- 79]. These fits typically give a shell of about two Mn-O at 1.75-1.8 • together with a second shell of variable coordination number at 1.90-2.25 A. The difficuky with these fits is that the number of variable parameters comes very close to the number of independent degrees of freedom in the data (see Sect. 2.3). If an attempt is made to correct
18
James E. Penner-Hahn
this when fitting the data [33], the significance of the second shell is marginal at best [79], and the two shell fits are probably not justified [74, 80]. We have argued [74] that the large (0.35 .~) range of distances reported for the second Mn-O shell reflects this fact. Much has been written regarding the question of whether there are one or two shells of nearest neighbors in the OEC. To some extent, this is an argument that is of interest only to EXAFS aficionados and is of little relevance to the OEC. After all, there is general agreement that the EXAFS data point to ligation by a disordered shell of primarily oxygen ligands. There are, however, two reasons why this issue is important. Firstly, if it is possible to resolve the Mn-oxo shell, then it is possible to use the apparent number of Mn-oxo interactions to define the structure of the OEC. Thus, it has been argued [4] that structural models with more than two Mn(p-O)2Mn units are unlikely, since these would give a higher Mn-oxo coordination number than is observed. If the two shell fits are not justified, than this argument cannot be used. Secondly, there is important information contained in the observation that two-shell fits are not significantly better than one-shell fits, and this information can be overlooked when two shells are forced to fit the data. Many Mn models and the Mn(p-O)zMn unit in Mn catalase do show two well-resolved shells of scatterers [31, 74, 81]. The models that do not show two well-resolved shells are those that have intermediate-range Mn-O interactions (ca. 1.9 A) in addition to both short Mn-O distances (N 1.8 A) and long Mn-(O/N) distances (N2.1 A) [82]. Thus, for example, the EXAFS data for the [Mn(IV)(salpn)(p-O)]2 dimer are best fitted by a single shell of 3 - 4 Mn-O scatterers at ~ 1.90 A, despite the presence of both short Mn-Ooxo distances and longer Mn-O/N distances [83, 84]. A similar situation may exist for the OEC [74], with Mn-O distances of 1.9 A coming from hydroxo, alkoxo, or phenoxo ligation. Unfortunately, the resolution of the OEC EXAFS data (ARN0.16 _~) is not sufficient to justify the use of three shells of nearest neighbors. 3.3.2
Mn-Mn Interactions
The apparent coordination number for the 2.7 ~ Mn-Mn peak is approximately 1.2 or 1.3 [4, 74, 75, 79]. This represents the number of Mn neighbors seen by each Mn, and thus corresponds to 2 or 3 Mn-Mn interactions for the Mn 4 cluster. This has generally been interpreted as evidence for two Mn(l.t-O)zMn dimers within the OEC [4, 77], a structure which would give an average coordination number of 1.0. EXAFS coordination numbers are not well defined. Nevertheless, it is intriguing that the apparent coordination number is consistently somewhat larger than expected for data measured by different groups, on different preparations, and analyzed using different parameters. Although disorder can easily lead to underestimation of EXAFS coordination numbers, it is rare to overestimate the coordination number. Thus, the apparent Mn-Mn number of 1.2-1.3 may indicate that there is an additional contribution to the 2.7 ]k peak. The most likely possibilities are either Mn-C from bidentate carboxylate residues or a third Mn-Mn interaction.
StructuralCharacterizationof the Mn Sitein the PhotosyntheticOxygen-EvolvingComplex
19
An important advance in characterizing the OEC structure was the development of polarized EXAFS measurements on oriented OEC multilayers. The initial measurements demonstrated that the 2.7 A peak is relatively isotropic, while the 3.3 A peak (see Sect. 3.3.3) has more pronounced orientation dependence [85]. A subsequent repetition of this work gave an average angle of 60 ° between the Mn-Mn vectors and the membrane normal [86]. Interestingly, the latter work found that the average Mn-Mn distance varied from 2.71 ,~ when the X-ray polarization was at 15° to the membrane normal to 2.74 A when it was at 75° [86]. Although this difference is small, the precision of EXAFS distance determinations is usually very good [74, 87], thus this difference is likely to be significant. The difference in Mn-Mn distance, which disappeared on oxidation to the S2 state (see Sect. 3.4), was interpreted [86] as evidence for two slightly different Mn-Mn distances in $1, which might be attributed to a Mn(III) 2 dimer and a Mn(IV)2 dimer. Subsequent EXAFS studies of inhibited samples (see Sect. 4) have confirmed the presence of two different Mn-Mn distances and have given refined Mn-Mn angles of 55° and 67° for the two different 2.7 ~ Mn-Mn vectors relative to the membrane normal [88]. The observed distance for the Mn-Mn shell is simply the weighted average of the different Mn-Mn distances that contribute to the shell. If the angles of 55° and 67° are attributed to dimers "A" and "B" respectively, R A (the Mn-Mn distance for dimer A) will contribute more when the X-ray polarization is parallel to the membrane normal, while RB will contribute more when the X-ray polarization is perpendicular to the membrane normal, thus accounting for the anisotropy in distance. However, angles of 55° and 67° give very nearly isotropic EXAFS and thus should not produce significant anisotropy. The anisotropy predicted by the measured angles [88] is so small that the true Mn-Mn distances would have to differ by > 0.2 .~ in order for two Mn-Mn vectors to give 0.03 A anisotropy in distance. An 0.2 A spread in distances is inconsistent with the isotropic EXAFS. It is not clear what this observation means for the OEC structure. It is possible that either the reported distance anisotropy or the reported angles are wrong. An alternative, however, is that there is a third scatterer (probably MnMn) which also contributes to the 2.7 A peak. If this interaction were oriented approximately parallel to the membrane normal (i.e., if it were very anisotropic) with a slightly shorter Mn-Mn distance, it would be possible to account for the angular anisotropy in distance [86] but still explain the lack of two resolvable 2.7 A Mn-Mn distances in the isotropic EXAFS. In this model, the apparent coordination number of 1.2 would be attributed to a slight underestimation of the true coordination number, due to slight disorder in the three Mn-Mn distances. 3.3.3 Outer Shell Interactions
From Fig. 8, it is clear that there is a third peak that is well above the noise level and there may even be a fourth peak. The latter is unconfirmed and will not be discussed further. The third shell peak reflects the long range structure in the
20
lames E. Penner-Hahn
OEC cluster. Interpretation of this peak has been controversial. The first suggestion that there might be scatterers at > 3 A was made in 1987 [89], but this suggestion was subsequently withdrawn and the peak was re-attributed to noisy data [28]. Later work using lower temperature measurements, showed reproducible features at R > 3 A, first in oriented [90] and subsequently in isotropic samples [30]. The observation that the third shell feature is seen only at low temperatures suggests that this feature arises from a fairly weak interaction. The third shell interaction shows greater dichroism than the 2.7 A Mn-Mn vectors, with an average angle of 43 ° relative to the membrane normal [86]. The relatively large intensity of the third shell peak, together with the absence of a significant number of histidine ligands (which give outer shell peaks) suggests that the third shell peak in the OEC is due to a Mn-metal interaction. The likely candidates for the metal are Mn and Ca. Although it has been reported that Mn-Ca interactions give better fits to this peak than do Mn-Mn interactions [78], it is now generally agreed that it is probably not possible to distinguish Mn-Mn from Mn-Ca scattering, at least not in these samples [79, 91]. Although the peak can be fitted with two shells (e. g. Mn-Mn+Mn-Ca), these fits were difficult to justify compared to single shell fits [79, 88]. This does not, of course, exclude the possibility that Mn, Ca (and potentially C) all contribute to the third shell [86, 92]. In addition to uncertainty over the identity of the distant scatterers, there is even uncertainty over the Mn-metal distance. We [30, 82, 87], together with George et al. [90] and the Klein/Sauer group [4, 77, 79, 86, 92] have consistently found a Mn-metal distance of approximately 3.3 A. In contrast, Nugent and coworkers have reported that the distance is 3.7 A [78, 91]. The origin of this discrepancy is not known. One possibility is that one of the reported distances is a result of the existence of multiple minima in EXAFS fits. Thus, for example, we have found that our data can be fitted using either distance, although we find the 3.3 A fits to be better [82]. In an effort to determine unambiguously whether Ca scattering contributes to the 3.3 A feature, several groups have removed Ca 2+ or replaced it with other metals and determined the effect of these perturbations on the EXAFS. One such study [93] found that replacement of Ca 2÷ by Sr 2÷ gave an increase in the amplitude of the 3.3 A interaction and that fits to the substituted data were slightly better with Sr + Mn scattering than with Ca+Mn scattering. Similarly, a decrease in amplitude was observed when Ca 2÷ was replaced by Na ÷ [91]. Both results suggest that at least some of the outer shell scattering is due to a Mn-Ca interaction. In contrast, we found no change in the third shell peak when Ca 2÷ was replaced by Sr2+, La3÷, or Dy3÷ [87]. In some cases, the fits were slightly better using Sr rather than Ca as the scatterer. However, this was true both for control samples that had never been exposed to Sr and for Sr-substituted samples, thus illustrating the difficulty in using fit quality to distinguish between fits for outer shell scattering [31]. One possible explanation for these apparently different results is that the samples were biochemically different. Our experiments were done in the absence of the 17 and 24 kDa extrinsic polypeptides, while these peptides were present in the other studies. If one or both peptides contributes ligands that
StructuralCharacterizationof the Mn Site in the PhotosyntheticOxygen-EvolvingComplex
21
/0 Mn~'!/NMn~=3. 3/~ ? '
H~
/M O
M=Ca2+,Sr2+,Dy3+,La3+
H
9. One possible explanation for the changein 2.7 the/~ Mn-Mn distancewhen the metal in the Ca site is changed Fig.
bridge the Mn-Ca interaction, the interaction would be expected to be stronger (and thus more readily detectable) in the presence of the extrinsic polypeptides. Akhough we did not find any changes in the 3.3 .~ feature, we did observe small but reproducible changes in the 2.7/~ peak when Ca was replaced [87]. The average Mn-Mn distance decreased slightly (0.013 _+0.004 .~) when Ca was replaced with Sr and increased slightly (0.015 + 0.004 .~) when Ca was replaced with a lanthanide. Similar results seem to have been observed in other studies [93], albeit with somewhat lower precision due to variability in E0 [87]. This is direct evidence of a role for Ca in controlling the M n core structure. The average Mn-Mn distance appears to increase with increasing Lewis acidity of the cation that is bound in the Ca site. Based on model studies, the observed changes in RMn.Mnare too small to be due to protonation [83, 84]. One possible explanation is that the Mn-Mn distance is affected by hydrogen bonding, with the Ca site modulating the acidity of a proton bound near an oxo bridge of one of the di-p-oxo bridged dimers (Fig. 9). A model such as Fig. 9 would explain the observation that Caa÷ is bound most tightly in $1 [3, 94, 95], since the basicity of the oxo-bridge, and thus the strength of the hydrogen bond, is expected to decrease when the Mn is oxidized [84]. The change in the multiline EPR spectrum following Sr substitution would be consistent with the fact that protonation alters intradimer exchange couplings [84]. Finally, the inability to form $2 when lanthanides are bound at the Ca site [96] and the stability of $2 when Ca is depleted [42] might be related, at least in part, to alterations in the redox potential of the Mn(p-O)/Mn dimers as a function of variations in hydrogen bonding. 3.4 S State Dependence of the OEC 3.4.1 The S2 State
As described in Sect. 3.1, the $2 state is EPR-active while S1 is EPR-silent under normal conditions. This suggested that Mn was oxidized in the $1~ $2 transition, and this was confirmed by early XANES measurements showing an i n -
22
James E. Penner-Hahn
crease in edge energy on formation of S 2 [97]. Very similar changes in energy are observed regardless of whether the multiline or the g ~ 4.1 forms of $2 are prepared [98, 99], although the details of the edge shape and the EXAFS are slightly different for the two forms of $2 [92]. This suggests that Mn is always oxidized in the S~-4 $2 transition, but that slightly different structures are formed, as expected given the different magnetic properties of the multiline and g-~4.1 forms. Several of the most recent reports of S state edge energies are summarized in Table 1. Although slightly different edge energies are obtained depending on the details of the data analysis (and possibly on sample preparation), similar results have been obtained in all of the studies of the $1-4 $2 transition, strongly supporting the conclusion that Mn is oxidized in this S-state transition. The 2.7 A Mn-Mn peak in the EXAFS for the g ~ 4.1 form of $2 is about 30% smaller than that seen for the multiline form of $2 [92]. This decrease in amplitude was modeled using two different Mn-Mn distances, at 2.72 and 2.85 A. This is probably not a unique fit since the spread in the apparent Mn-Mn distance is smaller than the resolution of the data. Nevertheless, it suggests a structural origin for the different spin-states of the two $2 forms. It is intriguing that the ratio of the coordination numbers for the short (2.72 ~k) and long (2.85 ~,) Mn-Mn shells is approximately 2:1. Similar behavior has been observed for inhibited $2 samples (see below). 3.4.2 The S3 state - XANES
In contrast to the general consensus on the S1--~ S2 transition, characterization of the $2-4 $3 transition has been controversial. There are three approaches to preparation of $3: The S1 state can be illuminated under conditions that allow two, and only two, turnovers; samples can be inhibited so that they cannot advance beyond $3; or illuminated with two short, saturating flashes and then frozen rapidly before they can be $3 can decay. The double-turnover approach was the first to be applied to the OEC [28]. Using chemical treatments to introduce a high-potential acceptor (oxidized non-heme Fe), it was possible to prepare samples with only 35 % of the normal $2 multiline EPR signal. The decrease in multiline intensity was attributed to formation of S3 in 65 % of the centers. The observed edge energy (see Table 1) was unchanged from that in $2, suggesting that there was no oxidation of Mn in the $2-4 $3 transition. Different results were obtained using inhibited samples. The advantage of inhibited samples is that they may permit preparation of more homogeneous samples. The disadvantage is that these samples really represent S*, i.e. a potentially modified $3. The observed edge energies increase by approximately 1 eV in both the S1 -~ $2 and the $2 -~ $3 transitions (see bottom of Table 1), suggesting that for the inhibited samples, Mn oxidation occurs in both transitions [100]. As described in Sect. 3.1.5, the inhibited $3 state is characterized by a broad radical signal in the EPR, now shown to be Yz'. This is consistent with the oxidation of Mn if the inhibited S* states contain a mixture of states, many of which have oxidized Mn and account for the shift in edge energy, and some
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex Table 1.
23
Edge energies for the OECa
State
Edge
S1 $2 $3 S~ $2 S~ $2 $3 $1 $2 $3 S~ S2 $3
6551.4 6552.3 6552.1 6551.6 6552.6 6551.7 6553.5 6553.8 6551.7 6552.5 6553.7 6551.8 6552.7 6553.2
Inhibited samples Acetate S* S* S* NaC1 S* S* S* Ammonia S* S* S*
6552.6 6553.2 6553.8 6551.0 6551.9 6553.2 6550.6 6551.5 6552.5
AE (eV)b
Ref. [28, 76]
0.9 0.0 [100] 1.0 [29] 1.8 0.3 [26] 0.8 1.2 [27] 0.9 0.5 [100] 0.6 0.6 [100] 0.9 1.3 [100] 0.9 1.0
a $3 state in [28] was trapped by chemical oxidation; other uninhibited S3 data were for flashed samples. b Change in energy between $2 and $1 or $3 and $2. of which have Yz" instead of oxidized Mn, and account for the broad radical in the EPR [100]. Recently, two groups have used flash illumination to prepare $3 and So [26, 29], reaching opposite conclusions. These experiments are inherently difficult because of the inevitable dephasing of S state composition as successive flashes are given. Dephasing can result from misses (centers that do not advance), double hits (centers that advance two S states on a single flash) or from centers that don't cycle (e. g. are inactive). In order to obtain "true" S state spectra, both groups were forced to estimate the S state composition after each flash and use these populations to deconvolute their measured spectra. The assumptions used in both of the deconvolutions are potentially problematic and may be responsible for the different conclusions. Neither group included a fraction of centers that did not turnover. This could potentially lead to an underestimation of the edge shifts. However, Ono et al., who have the smallest range of edge energies and thus would be most likely to have suffered from non-active centers, are the group that finds an energy
24 Table 2.
James E. Penner-Hahn Comparison of observed and fitted S state edge energies for flash experiments
Sample
Edge
Native
Observed 6551.7 Fit 6551.7 Observed 6551.7 Fit 6551.7 Observed 6551.8 Observed 6552.1
6552.4 6552.5 6552.4 6552.5 6552.7 6552.8
Observed 6551.7 Fit 6551.7
6553.5 6553.7 6553.5 6553.8
Native Native C1-depleted Native
0 flash/S, 1 flash/S2 2 flash/S3 3 flash/S0 4 flash 5 flash 6553.3 6553.7 6553.1 6553.6 6553.2 6552.9
6551.5 6550.7 6551.6 6550.9 6551.8 6552.7
6551.7
Ref. [26]
6551.7 6552.0 [101] 6551.6 6552.5
6550.5 6551.0 6552.9 6550.1
[27] [29]
See text for discussion of procedures used to calculate true S state spectra. change on $2--~ $3 (see Table 2 for a summary of reported edge energies). It thus seems unlikely that the different conclusions can be related to different fractions of inactive centers. Ono et al. illuminated their samples using Nd-YAG laser flashes (7 ns) and thus did not consider double hits [26, 101]. The rate of reduction of S states by YD was estimated from literature values and the percentage of misses (9 %) was refined to give the best fit to the observed edge energies. This approach has been criticized for lacking an independent measure of the S state composition [29]. However, the clear period-four oscillation in the measured edge energies (Table 2) suggests that there is not significant scrambling of the S state composition. In particular, the observed edge energy after 3 flashes is slightly lower than the starting energy ($1). This is the step at which virtually all of the models (see below) predict Mn should be reduced relative to $1. If the composition was too far off, the authors should not have seen reduction in edge energy at this step. Recently, Ono and co-workers have repeated these measurements on samples that were inhibited by either C1- [27] or Ca2÷ depletion [102]. The latter showed a resting (0-flash) edge energy that was decreased by ca. I eV relative to the control. This suggests the presence of Mn(II), and thus these data may be difficuk to interpret. For the C1--depleted data, however, the 0- and 1-flash energies are very similar to those in the control (see Table 2). In the C1--depleted samples, which do not advance beyond $2, the edge energy increases by only 0.1 eV on the second flash, while for the control, the second flash gives an increase of 0.5-0.9 eV. This difference is consistent with a model in which Mn oxidation is blocked in C1--depleted samples, and thus supports the proposal that Mn is oxidized in the $2---~$3 transition. However, these data are relatively noisy and must certainly be confirmed. Klein, Sauer, and co-workers [29] measured the relative concentration of $2 directly following each flash by using the intensity of the S2 multiline signal. With these data available for five flashes, they were able to treat both the miss fraction (12 %) and the double hit fraction (5 %) as adjustable parameters (the larger double hit fraction resuked from use of a flash lamp with a 14 l~s
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex
25
FWHM). This is clearly a more precise solution to the problem of S state deconvolution. It is not clear, however, that five flashes provide enough information to permit an accurate determination of the miss and double hit fractions. With data for only five flashes it is not possible to account for different miss fractions on different S state transitions (see Sect. 1.3). Perhaps more importanfly, the use of an average double hit parameter may underestimate the S state scrambling if, as has been found in some cases, the double hit fraction is large only on the first flash [8, 9]. A large double hit fraction would be one explanation for the 1.8 eV shift observed on the first flash, which is large for oxidation of only 25 % of the Mn. In model compounds, the change in edge energy is typically about 3 eV between Mn(III) and Mn(IV) [82]. If the $ 1 ~ Sa transition corresponds to oxidation of one out of four Mn atoms, it should show a shift of approximately 0.75 eV. If the $2 energy is overestimated, this would, in turn, lead to underestimation of the $2--~ $3 transition energy change. In addition to deconvolution differences, a second difference in these data concerns the method used to define the edge energy. Ono et al. used half-height, while Klein, Sauer, and their co-workers used the first inflection point. There is a rather pronounced change in the shape of the XANES spectrum between S~ and Sa (see Fig. 4 in [29]). The appearance of a new feature on the rising edge of the $2 spectrum may be, in part, responsible for the unusually large change in energy between S1 and S2. The change in edge shape (i. e. the new feature in the second derivative) causes the first inflection point, defined as the zero crossing in the second derivative, to shift to higher energy. In principle, this could lead to overestimation of the $2 energy. In summary, the XANES data for S3 are contradictory. Some of the data clearly suggest Mn oxidation in both $1---~$2 and $2---~$3. Other data show a very large shift in energy in $1---~S2 and a very small shift in Sz---~$3. This has been interpreted as evidence that Mn is not oxidized on the $2---~S3 transition [29], but might alternatively be interpreted as evidence that Mn is oxidized in both steps, but that the apparent S2 edge energy is too high. At the very least, it seems safe to conclude that the present XANES data do not provide compelling evidence for concluding that Mn is not oxidized in the $2---~$3 transition. 3.4.3 The $3 State - Other Measures of the Oxidation State
In addition to XANES, NMR, EPR and UV-visible spectroscopy have all been used to investigate the Mn oxidation state changes that accompany the S state transitions. The solvent 1H relaxation rate, which can be measured by NMR, should increase if Mn(II) or Mn(IV) ions are formed. The solvent relaxation rate increases in the $1 -~ $2 transition but not in the S2 --~ $3 transition [103]. This was interpreted as evidence for a mononuclear Mn(IV) in $2 and $3, but not in $1. However, strong relaxers other than mononuclear Mn(IV) might also be consistent with these data. Data for So are less well defined. It appears, however, that there is a further increase in relaxation rate in So. This was at-
26
James E. Penner-Hahn
tributed to formation of mononuclear Mn(II). The NMR data thus suggest that Mn is oxidized on So ~ $1 and $I ~ $2. The lack of change in relaxation rate on the $ 2 ~ $3 transition suggests either that Mn is not oxidized in this transition, or, if an Mn atom is oxidized that it does not contribute to the solvent water relaxation rate. The relaxation rate of the inactive tyrosyl radical, YD', has been measured by EPR relaxation [104, 105]. As with the NMR, the fastest relaxation rates were found for So, with $2 ~ $3 lower, and $1 lowest. If a single Mn dimer dominates the relaxation rate for YD', these data would suggest oxidation states of Mn(II) Mn(III) for S0, Mn(III) 2 for So, and Mn(III)Mn(IV) for Sa and $3. As with the NMR data, these data indicate that if an Mn atom is oxidized in the $2--~ $3 transition, it must not affect the spin-lattice relaxation rate for YD'. Both the NMR and EPR measurements are only indirectly sensitive to Mn oxidation state changes and may not necessarily be responsive to all of the Mn in the OEC. In contrast, UV-visible changes should be more directly sensitive to the Mn. These data have been controversial (see [106] for a review). There is now general consensus on spectral differences for S1--~ Sa and Sa---~$3. Both are characterized by a broad peak centered around 300 nm with AE-~ 4000-6000 M-~cm-~. Based on the similarity in peak wavelength, bandwidth, and amplitude, Dekker has interpreted both as M n ( I I I ) ~ Mn(IV) transitions [106]. The S0--~ S1 transition is the hardest to study, since the SOstate is the hardest to make pure and thus most susceptible to differences in data deconvolution. The S0---~$1 difference can also be interpreted as Mn(III)--~ Mn(IV), however this is by no means conclusive [106].
3.4.4 The Ss State - EXAFS
The first studies of S 3 found no change in the average Mn-Mn distance (2.72 A) on formation of $3 [28], although there was some increase in disorder of this shell. Fitting results suggested that if the disorder was due to the presence of 2 Mn-Mn distances, the spread in distance was approximately 0.15 .~. More recent measurements, using two saturating flashes to produce $3, suggest that there is a significant change in the Mn-Mn interactions when $3 is formed [4, 107], with both an increase in average distance and a decrease in the size of the Mn-Mn peak. The data can be fitted using either a single shell of two Mn-Mn interactions at 2.8 A, or with two Mn-Mn shells at 2.8 A and 3.0/~, giving an average distance of 2.9 ~. It is not clear why such different results have been obtained from $3. Although both experiments required deconvolution to obtain the true $3 spectrum, the%S3 that was formed in the two cases was very similar (65 % vs 70 %). Bond length differences of this magnitude are well outside the uncertainty of EXAFS, thus suggesting a difference of some sort between the samples. If the observation of a significant increase in both Mn-Mn distances is correct, this implies a significant structural change in the $2-~ $3 transition.
StructuralCharacterizationof the Mn Sitein the PhotosyntheticOxygen-EvolvingComplex
27
3.4.5
The So State The So state is the hardest to prepare by flash oxidation due to the dephasing which occurs. The XANES data for So all agree that this state is more reduced than S1 (see Table 2). Based on analysis of the shape of the XANES second derivative, it has been argued that Mn(II) is present in So [4, 29]. 4 Chemical Perturbations of the OEC
In addition to illumination, a variety of chemical perturbations have been used to probe the structure of the OEC. Inhibitors include potential mimics for water binding (NH3) or for H20 a (NH2OH), together with cofactor substitutions (e. g. F- for El-). 4.1 Inhibitors of Oxygen Evolution
Ammonia is a potential analog of substrate water. There are two independent NH 3 sites [108]. One, accessible to many amines, is competitive with CI-, while the other, accessible only to ammonia, is not competitive with C1-. In NH3-inhibited samples, NH 3 appears to be a ligand to Mn based on EPR [40] and ESEEM [109] measurements, and has been proposed to bind as a bridging amide (NHe) based on the observed asymmetry in the 14N quadrupolar interaction. Binding at the C1--independent site (which may be a water site) does not occur in SOor $1, or at low temperature, and gives rise to a modified multiline signal. The EXAFS data for such samples show a significant decrease in the amplitude of the Mn-Mn peak [88]. This decrease can be modeled either by doubling the Debye-Waller disorder term while keeping the distance the same (2.72 ~), or by including two different Mn-Mn shells at 2.71 and 2.86 A. This spread in distances is too small to allow reliable refinement given the limited resolution of the data. Consistent with this, the fit quality, when corrected for the number of variable parameters, is worse for two Mn-Mn shells than for a single shell. This does not mean that there are not two different distances but rather that they cannot be refined using isotropic data. However, the presence of two different Mn-Mn distances was confirmed by polarized measurements on oriented samples. The apparent distances were now 2.72 and 2.87 ~,, at angles of 55 ° and 67 ° respectively, relative to the membrane normal. The apparent coordination number was consistently larger (ca. 0.8) for the 2.72 A shell than it was for the 2.87 A shell. The larger contribution from the shorter Mn-Mn shell may account for the surprising result that the single shell fits gave a 2.71 ~ distance, not an average (ca. 2.8 A) distance. One interpretation of these data is that NH 3 has replaced an oxo bridge in one of the Mn(p-O)zMn dimers. This would explain both the ESEEM resuks (above) and the increase in the Mn-Mn distance. In addition, the ability to now
28
]ames E. Penner-Hahn
distinguish two different Mn-Mn distances in the polarized measurements supports the conclusion (above) that there are two different Mn dimers. 4.2 Modifications at the Chloride Site
Chloride is essential for full OEC activity [3]. Recent data suggest a single C1- is bound at the OEC, in one of two conformations [110]. In intact centers, C1- is bound in a tight-binding, slowly-exchanging site. In C1--free centers (i. e. once C1- has been removed), rebinding initially occurs at a low-affinity, fast-exchanging site which subsequently converts to a tight-binding site. Based on the correlation of 02 evolution rate with light intensity, it appears that the OEC is active in the absence of CI-, but at only 35 % of the rate seen for control samples [110]. In principle, it might be possible to use EXAFS to prove (or disprove) ligation of C1 to Mn. In reality, however, this is not possible using Mn EXAFS with the presently achievable data quality [30], although it may be possible to answer this question using CI (or Br) EXAFS. The best evidence that C1- is bound close to, if not necessarily on, an Mn atom, is the recent observation that replacement of C1- with F- causes a change in the Mn-Mn shell for the S2 state [111]. The change induced by F- is similar to, but much smaller than, the damping induced by NH3 (see Sect. 4.1). This can again be modeled using either a single shell of Mn with an increased Debye-Waller factor, or two shells of Mn separated by ca. 0.12 A. As with the NH3-inhibited samples, the two shell fits were generally worse than one shell fits (using a goodness-of-fit parameter corrected for the number of variable parameters), although marginally improved two shell fits could be obtained by constraining the Mn-Mn coordination numbers. A final similarity to the NH3-inhibited samples is that, in unconstrained fits, the apparent Mn-Mn coordination number for the longer shell is approximately half as large as that for the shorter shell. Polarized measurements to confirm the presence of two shells have not been reported for these samples. 4.3 Reduced Derivatives of the OEC
The OEC shows a two-flash delay in production of oxygen after treatment with potential substrate analogs such as NH2OH, N2H4, or H202 [112-114]. The delay is consistent either with a two-electron reduction of the Mn cluster to form an "S_I" state (see [ 115, 116] and references therein) or with formation of a state such as $1" NH2OH, where the bound NHzOH serves to reduce a higher S state (e.g., $2) following photooxidation [117, 118]. Initial XANES measurements suggested that NH2OH-treated samples were not reduced in the dark [76]. More recently, we have reported XANES data consistent with reduction of the Mn in the dark, and reoxidation to $I on exposure to light [70, 74]. The data in the two reports are, in fact, very similar. The different conclusions depend on whether the observed edge shifts are attributed to reduction of Mn to S_1 or to production of inactive Mn(II). In the latter study, the NHzOH effect was photoreversi-
Structural Characterization of the Mn Site in the PhotosyntheticOxygen-EvolvingComplex
29
ble, demonstrating that there was no significant decomposition, and suggesting that the observed edge shifts represent reduction of Mn by NH2OH in the dark. Regardless of the kinetic scheme (above) that is used to explain the two-flash NH2OH delay, the state produced by single turnover illumination should be So (or, more properly, S*, since this is not necessarily identical to the normal So). We found that NH2OH reduced the OEC to Mn(III) in S_~,as judged both by XANES [70] and by the lack of change in the 2.7 ~ Mn-Mn interactions on reduction [74]. Reduction to Mn(II) would be expected to cause loss of one of the MnMn interactions. An Mn(III) 4 oxidation state for S_1 implies an Mn(III)3Mn(IV) oxidation state for S0. In contrast, Guiles et al. found [76] evidence for Mn(II) when their NHaOH-treated samples were illuminated. This has been taken as evidence for Mn(II) in So, a conclusion that would be consistent with some [29] but not all [26] of the flash measurements of So. In addition to NH2OH, larger reductants, such as hydroquinone (H2Q), can also reduce the OEC. The latter gives a form of the OEC that contains ca. 50 % Mn(II), as judged by a six-line EPR signal, but that is nevertheless completely photoreversible, thus demonstrating that no Mn has been lost [119]. The Mn(II) produced by H/Q is not EDTA-extractable and does not cause enhancement of the solvent 1H-NMR relaxation, leading to the conclusion that this Mn(II) is sequestered near its active site [119]. Hydroquinone reduction gives a significant
Site A Low Potential Sl
Mrl~ !
MnIII O~'"~2QO~~ J"Mnll .... Mnll 1
n'¢ n'V
Site B
High Potential NH2OH(fast)/ N2H4 (siow)~ "So: ~0~, ~." •" " I Mnlll MnTM
S-1
I.o. ....
"~Of
~1/ Fast Mnll O~. S_ 1* MnIv MnlV ~,O d ~ NHpOH (fast)
Mnll
I ~04 I --O~
Mn~,
~ n Ill
O
HzOH
M nlP-----Mnll
"Mnl!
I
s_z I .o. Mn III
Mnlll
"~ 0 f'
NH20H
..........
Mnl1
) Mn release
~) ,Mnl~......~_~MnIU
S-s (unstable) Fig.10. Schemethat could account for the effectsof different reductants on the OEC structure within the "dimer of dimers" model. Alternate schemes based on different nuclearity would also be consistent. The connection between the two dimers is drawn for consistency with EPR data, but is not required by X-ray absorption measurements. Redrawn with permission from [74]
30
lames E. Penner-Hahn
decrease in XANES edge energy that is best modeled [70] by ca. 50% Mn(II), consistent with the EPR. As expected, the EXAFS spectra for H2Q-reduced samples show major structural changes [74], consistent with loss of the Mn-Mn interaction as a result of reduction of one Mn(p-O)zMn unit. The NHzOH and H2Q treatments show strong synergism in their ability to attack the OEC [119]. This, together with the EXAFS results, suggested [74] the two-site model shown in Fig. 10. Although the scheme is drawn using a dimer of dimers model, other arrangements of Mn would be consistent with observations. In this scheme, large reductants (e.g. H2Q) are only able to attack one site (A in Fig. 10). NH2OH preferentially attacks at a different site (B in Fig. 10) but will also slowly attack site A. When both A and B are reduced, the OEC is unstable and loses Mn to solution, thus accounting for the inactivation that accompanies long-term NH2OH incubation or simultaneous NH2OH + H2Q incubation. In this model, the structural change that accompanies H2Q reduction blocks the internal electron transfer (dashed line) that would normally prevent the production of different products when different reducing agents are used [74]. The significance of the two-site reduction model is that it suggests that, at least in the $1 and lower states, the OEC Mn is organized into two functionally distinct units. This would be consistent with the report (see Sect. 3.1.4) of two magnetically distinct Mn sites. If this model is correct and if it applies also to the higher oxidation states, it would suggest a two-step mechanism for water oxidation.
5 Structural Possibilities A wide variety of models for the OEC have been proposed at various times (see [2, 4] for discussion of possible structural models). Initial efforts centered on high-symmetry models, many of which have subsequently been shown [4, 79] to be inconsistent with the EXAFS data. A model that has dominated much of the recent discussion of the OEC has been the "dimer of dimers" model [77]. This contains two Mn (la-O)zMn units with 2.7 A Mn-Mn distances separated by a single 3.3 ]l oxo-bridged Mn-Mn, to give a C-shaped molecule. This is consistent with the EXAFS for S~, with the observation of two different Mn-Mn distances in inhibited samples and with suggestions (from reductant studies) that there are two separate sites of action. As discussed in Section 3.1.2, a dimer of dimers magnetic model can give an excellent fit to several different multiline EPR spectra [51]. The difficulty with the dimer of dimers model is that it is only possible to reproduce the EPR data by using magnetic couplings and Mn oxidation states that appear to be inconsistent with the EXAFS and XANES data. This incompatibility suggests that the dimer of dimers model requires modification in order to be consistent with both the EXAFS and the EPR. The goal of this section is to identify the key experimental observations that define the structure and to suggest ways in which existing models might be altered to be consistent with most, if not all, of the available data. At the outset, it is important to recall that EXAFS gives only average structural information. The consensus in the OEC field is that 4 Mn and 1 Ca are the
StructuralCharacterizationof the Mn Site in the PhotosyntheticOxygen-EvolvingComplex
31
metals present in the OEC [2] and this assumption is used in all of the interpretations of the EXAFS. It has been suggested [13], although not generally accepted (see [2] for a summary of the arguments) that there are more than 4 Mn. If there were more Mn, then the interpretation of the EXAFS coordination numbers would be different: An apparent coordination number of 1.2 - 1.3 with 6 Mn would represent 4 Mn-Mn interactions at 2.7/~. Similarly, contributions from metals other than Mn or Ca would not be distinguishable by EXAFS, but would alter significantly the interpretations of the EPR. One way to reconcile the EXAFS and EPR results would be to incorporate additional metals. Certainly, the history of bioinorganic chemistry contains many examples of enzymes whose true metal stoichiometry is larger than initially believed. However, it is premature to conclude that additional metals must be invoked in order to explain the observed data. The questions to be considered are: Does the OEC contain one or two Mn clusters? Is there an arrangement of Mn ions that would be consistent with both the EXAFS and the EPR? What are the Mn oxidation states? 5.1 One vs Two Clusters
At various times, the possibility of two different Mn clusters has been suggested. The presence of two dusters would be consistent with (although not proof of) a mechanism in which one cluster oxidizes water to H202, or a similar species, while the second cluster oxidizes H202 to Oz. It is thus important to establish whether two clusters are present. The key piece of evidence suggesting that there could be two independent clusters in the OEC is the observation that the $1 EPR signal converts to the multiline signal, but is uncorrelated to production of the g = 4.1 signal [56]. This observation requires that the g = 4.1 and multiline signals come from magnetically distinct species. The presence of two clusters would provide a ready explanation for the observed synergism between hydroquinone and NHaOH [119] and for the production of different reduced species [74]. However, these observations can also be explained within a single-cluster model. Similarly, the finding that it is possible to prepare both ground state and excited state g = 4.1 forms of Sa has been interpreted as evidence for two distinct clusters [39], but could also be consistent with multiple conformations of a single cluster. The principal weakness in the two-cluster model is that no additional reports of an $1 EPR signal have appeared. While this does not prove that this observation is incorrect, it raises the question of its reproducibility. The case for a single Mn cluster is based on two EPR observations: 1) The finding that, if the 55Mn ENDOR data are used to constrain the Mn parameters, the breadth of the multiline EPR signal can only be explained using at least 3 and probably 4 Mn ions [51]; and 2) The observation of multiline structure in the g = 4.1 signal, demonstrating that this signal must come from a cluster containing at least 3 Mn [53]. Thus, the g = 4.1 and multiline signals could only come from two different clusters if the OEC contained a minimum of 6 Mn ions. The discovery of a near-IR conversion from the multiline to the g = 4.1 forms of
32
James E. Penner-Hahn
Sa [55] removes any need to invoke two clusters (with the g ~ 4.1 signal coming from oxidation of an intermediate cluster). This does not, however, exclude the possibility of two clusters (see Fig. 6). The principal weakness in the singlecluster model is that mukiline character has only been observed for oriented, NH3-inhibited samples. This leaves open the possibility that the g ~ 4.1 signal in NH3-inhibited samples arises from a different source than the other g -- 4.1 signals (e. g., with NH3 bridging between the two putative clusters). Certainly, the structural changes observed on NH 3 inhibition [88] are larger than those seen for other inhibited samples. If there were two clusters, they would presumably be arranged as a monomer (giving the non-mukiline g ~ 4.1 signal) and a trimer (giving the multiline signals), since independent dimers appear inconsistent with the ENDOR data for the multiline form [48]. In summary, the data can probably be stretched to accommodate a twocluster model. The only reason to do so, however, is the S~ EPR signal. It is important to confirm or correct this report.
Note added in proof: A very recent report [120] confirms the presence of an $1 EPR signal that converts to the multiline $2 signal but not to the g ~ 4.1 $2 signal. The S~ signal arises from an excite state that is 2.5 K above the S = 0 ground state. 5.2 Arrangements of Mn Ions
It appears that the dimer of dimers scheme cannot account for both the EXAFS and EPR using a single structural model (i.e., oxo bridges for the EXAFS, hydroxo bridges for the EPR, as described above). One obvious way to modify this model would be to add a third Mn-Mn interaction at 2.7 A. This would give, for example, a strongly coupled triangle of Mn-Mn, perhaps with both p2-O and p3-O bridges. The spin-frustration that would be expected for such an arrangement might account for the small energy differences between the different OEC conformations that are seen in the EPR. The presence of three rather than two Mn-Mn interactions would explain the EXAFS coordination numbers. The presence of p3-0xo ligands, which typically have longer Mn-O bond lengths than pz-OXO, would explain both the disorder in the first shell EXAFS and the relatively long Mn-Mn distance (2.72 A for the OEC, vs 2.68 It for Mn catalase [81]). Finally, the presence of three Mn-Mn interactions, one of which is elongated in inhibited samples, would explain the apparent coordination numbers for inhibited samples and the anisotropy in Mn-Mn distances that was seen for S~. 5.3 Mn Oxidation States
If Mn is oxidized in each S state transition, then the mechanism of water oxidation can rely on conventional Mn redox chemistry. In contrast, if a non-manganese radical is produced in the higher S-state transitions, then this will have important mechanistic implications. Several models have been proposed in
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex
33
which a non-Mn radical plays a crucial role in the mechanism [4]. It is therefore important to evaluate the evidence for non-Mn radicals. The original suggestion that Mn is not oxidized in the $2~$3 transition came from XANES data [28]. This conclusion is inconsistent with many of the subsequent XANES measurements. Moreover, it is not clear that even the original data were necessarily proof that Mn was not oxidized, since, based on model studies, it seems possible that th e $1-~ Sa transition is anomalously large, thus making the apparent $2--~S3 energy difference appear small. The second line of evidence that Mn is not oxidized on $2--~$3 was the observation of a radical signal in the EPR. If an organic radical was produced on $2-4 S3, then Mn could not be oxidized in this transition. With the finding that the "$3 radical" is actually Yz" [68], this evidence against Mn oxidation disappears. It thus appears most likely that Mn is oxidized in all four S state transitions. A second redox question concerns the oxidation state of S0. It has been suggested [4] that one Mn dimer is redox-inert, with all of the water oxidation chemistry taking place on a second Mn dimer. The presence of Mn(II) in SO is consistent with this proposal [29]. However, if it is correct that the S_~ produced by NH2OH contains all Mn(III) [74], this would imply that at least 3, and probably all 4 of the Mn are redox-active (given $2 = Mn(III)Mn(IV)3; $3 = Mn(IV)4). As discussed above, these different conclusions may be related either to the difficulty of deconvoluting So in flash experiments or to the question of whether S_~truly represents a physiologically relevant state. Experiments of this sort should eventually clarify the role of Mn oxidation in water oxidation.
6 Conclusions and Future Prospects Despite a great deal of effort over many years, the detailed structure of the OEC remains elusive. Numerous models have been suggested and, as experimental methods have improved, most of these models have been shown to require modification. As the set of incorrect structures expands, the possibilities for the true structure become increasingly limited, and thus better defined. Recent EXAFS and EPR data appear to be inconsistent with the currently popular dimer-of-dimers model. Modifications based around triangular clusters may be one way to improve agreement with the available data. Acknowledgments. Supported in part by the NIH (GM-45205). I thank Profs. Charles Yocum, Vince Pecoraro, and Gerry Babcock for useful discussions.
References 1. Ghanotakis DF, Yocum CF (1990) Annu Rev Plant Physiol Plant Mol Biol 2. Debus RJ (1992) BBA 1102:269 3. Yocum CF (1992) In: V.L. Pecoraro (eds) Manganese Redox Enzymes VCH Publishers, New York, p 4. Yachandra VK, Sauer K, Klein MP (1996) Chem Rev (Washington, D C 96:2927 5. Deisenhofer J, Michel H (1991) Annu Rev Biophys Biophys Chem 20:247
34 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.
James E. Penner-Hahn Joliot P, Barbieri G, Chabaud R (1969) Photochem Photobiol 10:309 Kok B, Forbush B, McGloin M (1970) Photochem Photobiol 11:457 Shinkarev VP, Wraight CA (1993) Photosynth Res 38: 315 Shinkarev VP, Wraight CA (•993) Proc Nail Acad Sci U S A 90:1834 Berthold DA, Babcock GT, Yocum CF (1981) FEBS Letters 134:231 Kuwabara T, Murata N (1982) Plant Cell Physio123: 533 Ghanotakis DF, Demetriou DM, Yocum CF (1987) Biochimica et Biophysica Acta 891:15 Pauly S, Witt HT (1992) BBA 1099:211 Lee PA, Citrin PH, Eisenberger P, Kincaid BM (1981) Rev Mod Plays 53:769 Scott RA (1985) Methods Enzymol 117:414 Teo BK (1986) EXAFS: Basic Principles and Data Analysis, Springer-Verlag, New York Rehr JJ, Mustre dLJ, Zabinsky SI, Albers RC (1991) J Am Chem Soc 113:5135 Binsted N, Hasnain SS (1996) J Synchr Rad 3:185 Kirby JA, Goodin JA, Wydrynski T, Robertson AS, Klein MP (1981) J Am Chem Soc 103: 5537 Bart JCJ (1986) Adv Catal 34:203 Shulman RG, Yafet Y, Eisenberger P, Blumberg WE (1976) Proc Nail Acad Sci USA 73:1384 Wong J, Lyfle FW, Messmer RP, Maylotte DH (1984) Plays Rev B30: 5596 Rehr JJ, Albers RC, Zabinsky SI (1992) Phys Rev Lett 69:3397 Natoli CR (1983) In: A Bianconi, L Incoccia, and S Stipcich (eds) Springer Series in Chemical Physics Springer-Verlag, Berlin, p 43. Cramer SP, Eccles TK, Kutzler FW, Hodgson KO, Mortenson LE (1976) J Am Chem Soc 98:1287 Ono T-a, Noguchi T, Inoue Y, Kusunoki M, Matsushita T, Oyanagi H (1992) Science 258:1335 Ono T-a, Noguchi T, Inoue Y, Kusunoki M, Yamaguchi H, Oyanagi H (1995) J Am Chem Soc 117:6386 Guiles RD, Zimmermann JL, McDermott AE, Yachandra VK, Cole JL, Dexheimer SL, Britt RD, Wieghardt K, Bossek U, Sauer K, Klein MP (1990) Biochemistry 29:471 Roelofs TA, Liang W, Latimer MJ, Cinco RM, Rompel A, Andrews JC, Sauer K, Yachandra VK, Klein M (1996) Proc Nail Acad Sci USA 93:3335 Penner-Hahn JE, Fronko RM, Pecoraro VL, Yocum CF, Betts SD, Bowlby NR (1990) J Am Chem Soc 112: 2549 Riggs-Gelasco PJ, Stemmler TL, Penner-Hahn JE (1995) Coord Chem Rev 144:245 Lyfle FW, Sayers DE, Stern EA (1989) Physica B 158:701 Bunker G, Hasnain S, Sayers D (1991) In: S. Samar Hasnain (eds) X-ray Absorption Fine Structure Ellis Horwood, New York, p 751. Dismukes GC, Siderer Y (1980) FEBS Lett 121:78 Rutherford AW, Bouss ac A, Zimmermann JL ( 1991) New J Chem 15: 491 Miller A-F, Brudvig GW (1991) Biochim Biophys Acta 1056:1 Zimmermann JL, Rutherford AW (1984) Biochim Biophys Acta 767:160 Casey JL, Sauer K (1984) Biochim Biophys Acta 767:21 Smith PJ, Pace RJ (1996) Biochim Biophys Acta 1275:213 Beck WF, De Paula JC, Brudvig GW (1986) J Am Chem Soc 108:4018 Boussac A, Rutherford AW (1988) Biochemistry 27:3476 Boussac A, Zimmermann JL, Rutherford AW (1989) Biochemistry 28: 8984 Sivaraja M, Tso J, Dismukes GC (1989) Biochemistry 28:9459 Pace RJ, Smith P, Bramley R, Stehlik D (1991) Biochim Biophys Acta 1058:161 Zheng M, Dismukes GC (1992) Res Photosynth, Proc Int Congr Photosynth, 9th 2: Bonvoisin J, Bloudin G, Girerd JJ, Zimmermann JL (1992) Biophys J 61 : 1076 Ahrling KA, Pace RJ (1995) Biophys J 68:2081 Randall DW, Sturgeon BE, Ball JA, Lorigan GA, Chan MK, Klein MP, Armstrong WH, Britt RD (1995) J Am Chem Soc 117:11780 Lorigan GA, Britt RD (1994) Biochemistry 33:12072
Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex
50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89.
35
Hansson (),Aasa R,V~inng~trd T (1987) Biophys J 1987:825 Zheng M, Dismukes GC (1996) Inorg Chem 35:3307 Zheng M, Khangulov SV, Dismukes GC, Barynin VV (1994) Inorg Chem 33: Kim DH, Britt RD, Klein MP, Sauer K (1990) ] Am Chem Soc 112:9389 Haddy A, Dunham WR, Sands RH, Aasa R (1992) Biochim Biophys Acta 1099:25 Boussac A, Girerd J-J, Rutherford AW (1996) Biochemistry 35: 6984 Dexheimer SL, Klein MP (1992) ] Am Chem Soc 114:2821 Chan MK, Armstrong WH (1991) J Am Chem Soc 113:5055 Armstrong WH (1992) In: V.L. Fecoraro (eds) Manganese Redox Enzymes VCH Publishers, New York, p 261 de Paula ]C, Li PM, Miller A-F, Wu BW, Brudvig GW (1986) Biochemistry 25:6487 Koulougliotis D, Hirsh DJ, Brudvig GW (1992) J Am Chem Soc 114:8322 Andreasson LE, Lindberg K (1992) Biochim Biophys Acta 1100:177 MacLachlan DJ, Nugent JHA (1993) Biochemistry 32:9772 Baumgarten M, Philo IS, Dismukes GC (1990) Biochemistry 29:10814 SzalaiVA, Brudvig GW (1996) Biochemistry 35:1946 Zimmermann JL, Boussac A, Rutherford AW (1993) Biochemistry 32:4831 Boussac A, Zimmermann J-L, Rutherford AW, Lavergne J (1990) Nature 347:303 Hallahan BJ, Nugent JHA, Warden JT, Evans MCW (1992) Biochemistry 31:4562 Tang XS, Randall DW, Force DA, Diner BA, Britt RD (1996) ] Am Chem Soc 118:7638 Gilchrist ML, Ball JA, Randall DW, Britt RD (1995) Proc Natl Acad Sci, USA 92:9545 Riggs P, J., Yocum CF, Penner-Hahn JE, Mei R (1992) J Am Chem Soc 114:10650 Kirby ]A, Robertson JP, Smith AC, Cooper SR, Klein MP (1981) J Am Chem Soc 103: 5529 Britt RD, Tang XS, Gilchrist ML, Lorigan GA, Larsen BS, Diner BA (1994) Biochem Soc Trans 22: 343 Tang XS, Diner BA, Larsen BS, Gilchrist MLJ, Lorigan GA, Britt RD (1994) Proc Natl Acad Sci USA 91:704 Riggs-Gelasco PJ, Mei R, Yocum CF, Penner-Hahn ]E (1996) J Am Chem Soc 118:2387 Yachandra VK, Guiles RD, McDermott A, Britt RD, Dexheimer SL, Sauer K, Klein MP (1986) Biochim Biophys Acta 850:324 Guiles RD, Yachandra VK, McDermott AE, Cole JL, Dexheimer SL, Britt RD, Sauer K, Klein MP (1990) Biochemistry 29: 486 Yachandra VK, DeRose VJ, Latimer MJ, Mukerji I, Sauer K, Klein MP (1993) Science 260: 675 MacLachlan DJ, Hallahan BJ, Ruffle SV, Nugent JHA, Evans MCW, Strange RW, Hasnain SS (1992) Biochem J 285:569 DeRose VJ, Mukerji I, Latimer M], Yachandra VK, Sauer K, Klein MP (1994) J Am Chem Soc 116:5239 Penner-Hahn JE, Fronko RM, Waldo GS, Yocum CF, Bowlby NR, Betts SD (1990) Curt Res Photosynth, Proc Int Conf. Photosynth, 8th 1: Waldo GS, Yu S, Penner HJE (1992) ] Am Chem Soc 114:5869 Riggs-Gelasco PJ (1995) PhD thesis, University of Michigan Larson EJ, Riggs PJ, Penner-Hahn JE, Pecoraro VL (1992) J Chem Soc, Chem Commun 102 Baldwin MJ, Stemmler TL, Riggs GPJ, Kirk ML, Penner-Hahn JE, Pecoraro VL (1994) ] Am Chem Soc 116:11349 George GN, Prince RC, Frey TG, Cramer SP (1989) Physica B 158:81 Mukerji I, Andrews JC, DeRose VJ, Latimer MJ, Yachandra VK, Sauer K, Klein MP (1994) Biochemistry 33: 9712 Riggs-Gelasco P,Mei R, Ghanotakis D, Yocum CF, Penner-Hahn JE (1996) J Am Chem Soc 118: 2400 Dau H, Andrews JC, Roelofs TA, Latimer MJ, Liang W, Yachandra VK, Sauer K, Klein MP (1995) Biochemistry 34: 5274 Guiles RD, Yachandra VK, McDermott AE, Britt RD, Dexheimer SL, Sauer K, Klein MP (1987) Prog Photosynth Res, Proc Int Congr Photosynth, 7th 1:
36
James E. Penner-Hahn
90. George GN, Prince RC, Cramer SP (1989) Science 243:789 91. MacLachlan D], Nugent JHA, Bratt P], Evans MCW (1994) Biochim Biophys Acta 1186:186 92. Liang W, Latimer MJ, Dau H, Roelofs TA, Yachandra VK, Sauer K, Klein MP (1994) Biochemistry 33:4923 93. Latimer MJ, DeRose VJ, Mukerji I,Yachandra VK, Sauer K, Klein MP (1995) Biochemistry 34:10898 94. )~delroth P, Lindberg K, Andreasson LE (1995) Biochemistry 34:9021 95. Seidler A, Rutherford AW (1996) Biochemistry 35:12104 96. Bakou A, Ghanotakis DF (1993) Biochim Biophys Acta 1141:303 97. Goodin DB, Yachandra VK, Britt RD, Sauer K, Klein M (1984) Biochimica Biophysica Acta 767:209 98. Yachandra VK, Guiles RD, McDermott AE, Cole JL, Britt RD, Dexheimer SL, Sauer K, Klein MP (1987) Biochemistry 26: 5974 99. Cole ],Yachandra VK, Guiles RD, McDermott AE, Britt RD, Dexheimer SL, Sauer K, Klein MP (1987) Biochim Biophys Acta 890:395 100. MacLachlan DJ, Nugent JHA, Evans MCW (1994) Biochim Biophys Acta 1185:103 101. Kusunoki M, Ono T, Noguchi T, Inoue Y, Oyanagi H (1993) Photosynth Res 38:331 102. Ono Ta, Noguchi T, Inoue Y, Kusunoki M, Yamaguchi H, Oyanagi H (1993) FEBS Lett 330: 28 103. Sharp RR (1992) In: VL Pecoraro (eds) Manganese Redox Enzymes. VCH Publishers, New York, p 177 104. Bvelo RG, Styring S, Rutherford AW, HoffA] (1989) Biochim Biophys Acta 973:428 105. Styring SA, Rutherford AW (1988) Biochemistry 27:4915 106. Dekker JP (1992) In: V.L. Pecoraro (eds) Manganese Redox Enzymes VCH Publishers, New York, p 85 107. Liang W, Roelofs TA, Olsen GT, Latimer MJ, Cinco RM, Rompel A, Sauer K, Yachandra VK, Klein MP (1995) In: P. Mathis (eds) Photosynthesis: From Light to Biosphere Kluwer, Netherlands, p 413. 108. Sandusky PO, Yocum CF (1986) Biochim Biophys Acta 849:85 109. Britt RD, Zimmerman J-L, Sauer K, Klein MP (1989) J Am Chem Soc 111:3522 110. Lindberg K, Andreasson LE (1996) Biochemistry 35:14259 111. DeRose VJ, Latimer MJ, Zimmermann JL, Mukerji I, Yachandra VK, Sauer K, Klein MP (1995) Chem Phys 194:443 112. Bouges B (1971) Biochim Biophys Acta 234:103 113. Renger G, Bader KP, Schmid GH (1990) Biochimica et Biophysica Acta 1015:288 114. Mano J, Takahashi M-a, Asada K (1987) Biochemistry 26: 2495 115. Messinger J, Renger, G. (1993) Biochemistry 32:9379 116. Kretschmann H, Pauly S, Witt HT (1991) Biochim Biophys Acta 1059:208 117. Radmer R (1981) Biochimica et Biophysica Acta 637:80 118. Radmer R, Ollinger O (1982) FEBS Letters 144:162 119. Mei R, Yocum CF (1992) Biochemistry 31:8449 120. Yamauchi T, Mino H, Matsukawa T, Kawamori A, Ono T-a (1997) Biochemistry 35:7520
Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidases and Vanadium-ContainingEnzymes Albrecht Messerschmidt Max-Planck-Institut ffir Biochemie, Am Klopferspitz 18A, D-82152 Martinsried
E-mail:
[email protected]
The coordination geometries of metal sites in cupredoxins, mutants and metal derivatives of cupredoxins, multi-copper oxidases and a vanadium-containing chloroperoxidase as derived from X-ray crystallography are described. Correlations with their spectroscopic, electrochemical, electron transfer and catalytic properties are discussed. X-ray crystallography,EPR and Resonance Raman spectroscopy of copper sites in cupredoxins and mutants have led to a classification ranging from type 1 trigonal, type 1 distorted tetrahedral, type 1.5 to type 2. The mutation of copper ligands in azurin or amino acids close to the copper site changes the redox potential in a range of_+140 mV, only. The high redox potential of rusticyanin of 680 mV (azurin, 380 mV) should be mainly due to the special protein environment of the copper site (high proportion of hydrophobic residues). The type 1 and trinuclear copper centres of the multi-copper oxidases ascorbate oxidase, laccase and ceruloplasmin are presented. The metal sites of type 2 depleted, fully-reduced, peroxide and azide forms of ascorbate oxidase, as determined by X-ray crystallography, are discussed in terms of the mechanistic properties of these enzymes. The first X-ray structure of a vanadium-containing protein, namely of a chloroperoxidase from the fungus Curvularia inaequalis, is briefly discussed. The protein fold is mainly a-helical with two four-helix bundles. In the X-ray structure, which is an azide:enzyme complex, the vanadium exhibits a simple unexpected coordination geometry, namely, a trigonal bipyramidal coordination with three non-protein oxygen ligands (VO3 group), one nitrogen ligand from a histidine and one nitrogen from the exogenous azide ligand.
Keywords:protein crystallography, small blue copper proteins, multi-copper oxidases, vanadium-containing enzyme, electron transfer
List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Metal Sites in Small Blue Copper Proteins
38 ..............
39
1.3 1.4
Introduction ................................ Spatial S t r u c t u r e s of C u p r e d o x i n s a n d Mutants . . . . . . . . . . . The C u p r e d o x i n Fold . . . . . . . . . . . . . . . . . . . . . . . . . . . G e o m e t r i e s o f C u p r e d o x i n C o p p e r Sites . . . . . . . . . . . . . . . . S u b s t i t u t i o n o f C o p p e r b y Hg, Zn, C d a n d Ni . . . . . . . . . . . . . The CuA D i n u c l e a r C o p p e r Site . . . . . . . . . . . . . . . . . . . . . Spectroscopy of Cupredoxins ...................... Redox Potentials a n d Electron Transfer P r o p e r t i e s . . . . . . . . .
39 39 39 40 45 47 48 50
2
Metal Sites in Multi-Copper Oxidases . . . . . . . . . . . . . . . . .
52
2.1 2.1.1
The Trinuclear C o p p e r Active Site . . . . . . . . . . . . . . . . . . . Native O x i d i z e d E n z y m e . . . . . . . . . . . . . . . . . . . . . . . . .
52 52
1.1 1.2 1.2.1 1.2.2 1.2.3
1.2.4
Structure and Bonding, Vol. 90 © Springer Verlag Berlin Heidelberg 1998
38 2.1.2 2.1.3 2.1.4 2.1.5 2.2
Albrecht Messerschmidt Type-2 Depleted (T2D) Form of Ascorbate Oxidase . . . . . . . . . Fully-Reduced Form of Ascorbate Oxidase . . . . . . . . . . . . . . Peroxide Form of Ascorbate Oxidase . . . . . . . . . . . . . . . . . . Azide Form of Ascorbate Oxidase . . . . . . . . . . . . . . . . . . . Aspects of Electron Transfer and Mechanistic Properties . . . . . . . .
55 56 57 60 61
First X-ray Structure of a Vanadium-Containing Enzyme: Chloroperoxidase from the Fungus C u r v u l a r i a inaequalis
62
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
....
65
List of Abbreviations AzPwt AzADwt AzAXwt AzPH35Q AzPH35L AzPH35F AzADM121Q AzADM121H AzPwt: Zn AzADwt red AzPF 114A, AzPW48M: Ni AzPN47D: Zn AzPI7S AzPFll0S PcP PcE PcP: Hg ~p-AzAF AmPD AmTV CBP Scc AzAD: Cd NiR
AO TIHg
wild-type azurin from Pseudomonas aeruginosa (PA) [20, 21] wild-type azurin from Alcaligenes denitrificans (AD) [22] wild-type azurin from Alcaligenes xylosoxidans [23] azurin from PA, H35Q mutant [24] azurin from PA, H35L mutant [24] azurin from PA, H35F mutant [21] azurin from AD, M121Q mutant [25] azurin from AD, M121H mutant [26] wild-type azurin from PA containing Zn [27] wild-type azurin from AD, reduced form [28] azurin from PA, F114A mutant [29] azurin from PA, W48 M mutant containing Ni [30] azurin from PA, N47D mutant containing Zn [31] azurin from PA, I7S mutant [32] azurin from PA, F110S mutant [32] plastocyanin from Populus nigra [33, 34] plastocyanin from Enteromorpha prolifera [35] plastocyanin from Populus nigra, Hg substituted [36] pseudoazurin from Alcaligenesfaecalis [37, 38] amicyanin from Paracoccus denitrificans [39] amicyanin from Thiobacillus versutus [40] cucumber basic protein [41] cucumber stellacyanin [6] azurin from AD containing cadmium [42] copper-containing nitrite reductase from Achromobacter cycloclastes [15] ascorbate oxidase from Cucurbita pepo medullosa [43] mercury substituted type 1 site
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
39
1 Metal Sites in Small Blue Copper Proteins 1.1
Introduction
The unique properties of the small blue copper proteins have fascinated chemists for a long time. These proteins contain a mononuclear copper site with three characteristic properties: (1) an intense blue colour, ~ -600 nm, with absorption coefficients of 2000-6000 M-1 cm -1, due to an S(Cys)--~Cu(II) charge transfer; (2) an unusually narrow hyperfine coupling (All values of 0.0035-0.0063 cm -1) in the EPR spectrum of the Cu(II) protein; and (3) high reduction potentials (range 184-680 mV) as compared with the aqua Cu(II/I) couple (115 mV). This copper site has been classified as a type 1 copper site. The small blue copper proteins function as electron transfer proteins in such fundamental processes as photosynthesis and respiration. They have been termed cupredoxins to emphasise their common origin and their role as electron mediators in analogy to the iron-containing ferredoxins [1]. The research on cupredoxins has been summarised in several reviews in the past (see [2-4], for example). During the last few years cupredoxin research has benefited from the application of genetic engineering techniques in combination with modern spectroscopic techniques, X-ray crystallography, electrochemistry, kinetics and quantum theory. The main genetic engineering work has been done on azurins from Pseudomonas aeruginosa (PA) and Alcaligenes denitrificans (AD) in Leiden by Canters and co-workers (see [4], for example) and in G6teborg by V~nngard, Malmstr6m, Lundberg and colleagues (see [5], for example). Mutations were introduced in the direct copper ligand sphere and in the second coordination sphere to test their influence on the properties of the metal site. Mutations in different parts of the molecule have been introduced to probe their influence on other molecular properties. The first part of this article will focus only on new cupredoxin spatial structures, mutants of cupredoxins which influence the properties of the metal site, as well as the relation of the structure of the metal sites to their physicochemical properties. 1.2 Spatial Structures of Cupredoxins and Mutants 1.2.1 The CupredoxinFold
Cupredoxins fold into a single domain consisting mainly of a fl-sandwich or flbarrel. This fl-sandwich may comprise 6 to 13 fl-strands. The arrangement of the fl-strands is basically antiparallel showing the Greek-key motif. A typical cupredoxin fold represented by azurin is depicted in Fig. 1. In this case, the flsandwich consists of 8 fl-strands. The fl-sandwich of stellacyanin from cucumber, whose X-ray structure has been determined very recently, has only 6 flstrands [6]. This stellacyanin belongs to a subgroup of cupredoxins called
40
AlbrechtMesserschmidt
phytocyanins which are plant proteins. A characteristic feature of these phytocyanins is a disulphide bridge between a cysteine in the middle part of the amino acid sequence and a cysteine adjacent to the C-terminal histidine copper ligand as well as a 6- to 7-stranded fl-sandwich. The phytocyanin cucumber basic protein, whose X-ray structure is also known, has a 7-stranded fl-sandwich [7]. On the other hand, rusticyanin from Thiobacillusferrooxidans is characterised by a compact 13-stranded fl-sandwich, as determined by multidimensional NMR techniques [8] and X-ray crystallography [8a]. This compact fl-sandwich creates a hydrophobic core particularly rich in aromatic residues. Rusticyanin has a very high redox-potential (680 mV) and operates at extremely low pH (2.0). It is assumed that the unusual acid stability of the protein results from the high degree of secondary structure and the special hydrophobic core. Subunits II in cytochrome c oxidase [9, 10] and quinol oxidase [11] exhibit a cupredoxin fold in their membrane-exposed domain. These domains contain a binuclear copper site denoted as the CUA site. The cupredoxin domains of cytochrome c oxidase have a 10-stranded fl-sandwich. In quinol oxidase it is an 11-stranded fl-sandwich. The helical content in cupredoxins varies from 0 % in amicyanin to 14% in azurin. The blue, multi-copper oxidases ascorbate oxidase, laccase and ceruloplasmin consist of three (ascorbate oxidase, laccase) and six (ceruloplasmin) domains with the cupredoxin fold (see [13], for example). The mononuclear blue copper site is located in the C-terminal domain of ascorbate oxidase and laccase. Ceruloplasmin has three mononuclear coppers bound to domains 2, 4 and 6 [14]. A copper-containing nitrite reductase from Achromobacter cycloclastes consists of two domains with a cupredoxin fold which arrange to the enzymatically active trimer (six domains) [15]. The mononuclear copper site is found in the N-terminal domain. The cupredoxins and their multidomain relatives form a large evolutionary family. This family also comprises proteins that have lost their copper-binding capability during evolution (see [13], for example). 1.2.2
Geometriesof Cupredoxin CopperSites
The type 1 copper centre is located close to the surface of the protein (Fig. 1). In native type 1 proteins the copper has four protein side-chain ligands and in some cases (e.g. azurin) a weak main-chain carbonyl oxygen as the fifth ligand. The four canonical type 1 copper ligands are His, Cys, His, Met arranged in this sequence on the polypeptide chain. The Met ligand may be a glutamine in the subgroup of phytocyanins (e.g. stellacyanin from cucumber). The Cys, His, Met ligands come from a loop between the two C-terminal strands of the fl-sandwich. They are arranged in the sequence (Cys-Xn-His-Xm-Met(Gln)) with n and m varying from one protein to the next. The first histidine sits on a strand adjacent to these, but this strand is more than 30 residues away in the N-terminal direction due to the Greek key topology of the fl-sandwich. The methionine is always sandwiched between two hydrophobic residues, two residues apart, coming from a loop between two fl-strands at the N-terminus. One or more NH...S hydrogen bonds are formed between main-chain amide nitrogens and
Metal Sites in Small Blue Copper Proteins,Blue CopperOxidasesand Vanadium-Containing Enzymes
41
Fig. 1. Ribbon-type representation of an azurin from Pseudomonas aeruginosa, a typical small blue copper protein, type 1 copper binding site and disulphide bridge are included, MOLSCRIPT[12]
the cysteine ligand. The copper centres are usually not buried by more than 8 A. The C-terminal histidine, known as northern histidine, protrudes through a more or less extensive hydrophobic surface patch, which is probably one of the surface regions through which electron transfer takes place. The type 1 copper sites can be subdivided into T1 trigonal and T1 distorted tetrahedral based on the optical, EPR and RR spectroscopy [16-18], as well as X-ray crystallography [3]. The most abundant geometry of type 1 copper sites is between T1 trigonal and T1 distorted tetrahedral and found in plastocyanin, ascorbate oxidase, amicyanin, pseudoazurin and cucumber basic protein. This unusual copper coordination is a compromise between the preferred tetrahedral Cu(I) and tetragonal Cu(II) coordination. It reduces the reorganisation energy between both redox states which helps to speed up electron transfer [19]. As already mentioned, the genetic engineering work has been mainly focused on azurin. A relatively large number of high resolution X-ray structure analyses of different azurins, azurin mutants, azurin metal derivatives and apo-azurins has been carried out including structure determinations at different pH values and/or redox states. Their copper ligand bond distances are shown in Table 1. Most of these values are derived from high-resolution X-ray structures (resolution better
42
Albrecht Messerschmidt
than 2.0 ~) and the accuracy of the metal ligand bond distances has been estimated to be about + 0.1 ]k [22]. The copper coordination in azurin is slightly different to the T1 distorted tetrahedral coordination. It is T1 trigonal, a trigonal-bipyramidal coordination with the ND 1 atoms of the two histidines and the SG atom of the cysteine in the trigonal equatorial plane, the SD atom of the methionine and a main-chain carbonyl oxygen at the apices of the bipyramid (see Fig. 2). From attempts to correlate the spectroscopic properties with the coordination geometries of the type i sites it emerged that the cysteinate copper bond length and the distance of the copper ion from the N(His)-N(His)-S(Cys) plane may play a crucial role. These out-of-plane values are indicated in Table 1 for the cases where they were given in the corresponding references. Their meaning will be discussed in the context of the spectroscopy of the type 1 sites later. The Cu-S(Cys) bond of = 2.12 .~ in both T1 subtypes is extremely short as a consequence of having only three strong ligands. The T1 trigonal sites are further characterised by long Cu-S bonds to the axial methionine (>2.8 A) and out-of-plane values in the range of 0.02 to 0.13 _A.PA azurin, wild type and some mutant structures have Cu-S(Cys) bonds that are = 0.1 A longer than the normal value of 2.12 A in the other type 1 copper proteins. This is especially surprising among the azurin structures from different sources. It may be due to differences in the amino acid sequence
Fig.2. Type-1copper site in wild-type azurin from PA,MOLSCRIPT[12]
Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidasesand Vanadium-Containing Enzymes
43
Table 1. Bond distances of type 1 copper sites and metal derivatives
AzPwt, pH 5.5 AzPwt, pH 5.5, red AzPwt, pH 9.0 AzPwt, pH 9.0, red AzADwt AzADwt, red AzAXwt AzPH35Q AzPH35L AzPH35F AzPI7S AzPFll0S AzPFl14A AzADM121H, HP AzPwt: Zn AzPN47D: Zn AzAD" Cd AzPW48 M: Ni PcP PcP red, pH 7.8 PcE PcP: Hg AO ~o-AzAF ~o-AzAF red, pH 7.8 ~-AzAF red, pH 4.4 NiR AmPD AmTV CPB ScC AzADM121Q AzADM121Qred
CuN(H46)
CuS(Cl12)
CuN(Hll7)
CuS(M121)
Cu-O (G45)
out of NNSplane
2.11 2.14 2.09 2.14 2.08 2.13 2,02 2.03 2.09 1.94 2.10 2.09 2.19 2.06 2.01 2.09 2.25 2.15 1.91 2.12 1.89 2.34 2.11 2.16 2.16 2.19 2.06 1.95 2.04 1.93 1.96 1.94 1.97
2.25 2.29 2.26 2.27 2.14 2.26 2.12 2.05 2.20 2.08 2.24 2.10 2.23 2.15 2.30 2.27 2.39 2.49 2.07 2.11 2.12 2.38 2.08 2.16 2.17 2.16 2.17 2.11 2.13 2.16 2.18 2.12 2.09
2.03 2.10 2.04 2.15 2.00 2.05 2.02 2.05 2.03 2.12 1.94 1.96 2.46 2.06 2.07 2.04 2.20 2.07 2.06 2.25 2,17 2.36 2.08 2.13 2.29 3.09 2.00 2.03 2.13 1.95 2.04 2.05 2.67
3.15 3.25 3.12 3.17 3.14 3.23 3.26 3.04 3.01 3.06 3.10 3.18 3.04 2.22(N 8) 3.40 3.44 3.23 3.34 2.82 2.90 2.92 3,02 2.87 2.76 2.91 2.42 2.55 2.90 2.84 2.61 2.21(O e) 2.26(0 ~) 2.72(0 ~)
2.97 3.02 2.95 3.10 3.13 3.23 2.75 3.09 3,09 3.00 3.00 3.16 2.99 3.92 2.32 2.36 2.76 2.35 3.89 4.81 3.79 3.83 4.01 3.90 3.90 3.85 3.37 3.33
0.10 0.10 0.08 0.02 0.10 0.13 0.11 0.60 0.15 0.36 0.46 0.36 0.43 0.37
0.77 0.54 0.30 0.39 -
The resolution of the X-ray structure analyses is in most cases better than 2.0 ~..The omitting values were not given in the corresponding reference. The proteins are in the oxidized state if not indicated otherwise.
n e a r H i s 3 5 . H i s 3 5 is i n v a n d e r W a a l s c o n t a c t w i t h H i s 4 6 , a c o p p e r l i g a n d . I n PA a z u r i n a n H - b o n d is f o r m e d b e t w e e n t h e m a i n - c h a i n c a r b o n y l o f p r o l i n e 36 t o N D 1 o f H i s 3 5 . I n a z u r i n f r o m t h e Alcaligenes s p e c i e s denitrificans o r xylosoxidans the p r o l i n e is r e p l a c e d b y a v a l i n e w h i c h c a u s e s a c h a n g e i n t h e m a i n - c h a i n c o n f o r m a t i o n a t t h i s r e g i o n a n d t h e H - b o n d t o h i s t i d i n e is n o l o n g e r f o r m e d . The akered charge distribution at His35 could have an influence on the electron i c s t a t e o f t h e c o p p e r site g i v i n g r i s e t o t h e s e s m a l l a l t e r a t i o n s i n t h e c o p p e r site g e o m e t r y .
44
Albrecht Messerschmidt
The crystal structures of wild-type azurin at two pH values in both redox states have been determined [20, 21]. There are little changes in the COl~per site geometry. The bond lengths are slightly increased by about 0.05 to 0.1 A as also observed in reduced AD azurin [28], reduced poplar plastocyanin at pH 7.8 [34] and pseudoazurin from Alcaligenesfaecalis (AF) at pH 7.8 [38]. This is in line with a determination of the electronic structure of the reduced type 1 copper site [44]. In the reduced form of AF pseudoazurin at pH 4.4 the copper ion moves 0.69 A, mainly towards the SD atom of the axial methionine, and the imidazole of the northern histidine rotates by 26 ° around its CB-CG bond. The metal moves in the opposite direction with respect to its movement in the pH 7.8 reduced structure. The movements at the copper site resemble those in plastocyanin but the extent is quite different in these two cases [34]. The low-pH forms of PA azurin exhibit a pH-induced Pro36-Gly37 main-chain peptide bond flip. At the lower pH, the protonated imidazole ND1 of His35 forms a strong hydrogen bond with the carbonyl oxygen from Pro36, while at alkaline pH the deprotonated ND 1 acts as an acceptor of a weak hydrogen bond from the amide nitrogen of Gly37. In the crystal structures of the PA azurin mutants H35Q, H35L, H35F, I7S, Fll0S no remarkable influence on the copper site geometry can be noticed. This is not valid for the PA azurin mutant F114A [29]. The mutation was performed on residue Fl14, which exhibits a n-electron overlap with the copper ligand His117, to investigate its suggested role in the electron self-exchange reaction. Removal of steric constraints from the phenylalanine side chain created a somewhat different geometry around the copper site with an increased mobility of His117 resulting in an enlarged Cu-N bond length, which may be responsible for the slight differences obtained in the spectral properties of the mutant compared with the wild-type protein. Mutation of the axial methionine ligand in AD azurin to histidine generated a protein that has a green colour at higher pH values (6-7) and is blue below pH 3.8 [45]. The copper site geometry of the high-pH form (pH 7.0), as determined by X-ray crystallography [26], is shown in Fig. 3. The mutation generates a distorted tetrahedral copper coordination with an 0.6 ]~ out-of-plane value for the copper ion. The imidazole ND1 of His121 forms a strong bond with Cu, which leads to the distorted tetrahedral copper site. Copper sites that resuked from the replacement of the axial methionine by lysine, glutamic acid and histidine were found to have spectroscopic properties between those of type 1 and type 2 sites [46-48]. These sites were named type 1.5 sites [29, 42], but a clear understanding of their properties has still been lacking. The type 1 copper site of nitrite reductase from Achromobacter cycloclastes displays subtle geometric differences [15]. The bonds to the equatorial ligands of the type 1 site in oxidised nitrite reductase expand relative to those in plastocyanin with 2.17 ]k Cu-S(Cys) and 2.06 and 2.00 A Cu-N(His) bond lengths, while the axial Cu-S(Met) bond contracts to 2.55 •. Furthermore, the copper ion is raised further out of the equatorial NNS plane (0.54 A), and angular changes at the site, particularly with respect to the Cys and Met ligation, are evident. This perturbed geometry confers the protein a green colour.
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
45
Fig. 3. Copper site in M121Hmutant of azurin from AD, pH 7.0,MOLSCRIPT[12]
The copper sites of the M121Q mutant ofAD azurin both in the oxidised and reduced state, as derived from X-ray crystallography, are displayed in Fig. 4 [25]. The axial methionine ligand was mutated to glutamine because there was evidence from amino acid sequence alignments among the phytocyanins and molecular modelling based on the spatial structure of cucumber basic protein that glutamine is the axial ligand in stellacyanin [49] and the study of the mutant may shed new light on the structure of the copper site in the latter protein. In the oxidised form (Fig. 4a), the histidine and cysteine copper ligand distances and angles in the equatorial plane around the copper are very similar to the wild-type protein. Gln121 is coordinated in a monodentate fashion via its side-chain oxygen atom at a distance of 2.26 ~. The distance between the copper and the carbonyl group of Gly45 is increased to 3.37 A resulting in a distorted NNSO copper coordination. These values are equal, within experimental error, to the copper ligand bond distances in cucumber stellacyanin, whose Xray structure has been determined very recently [6]. In contrast to wild-type azurin, the copper site in M121Q azurin undergoes significant structural changes upon reduction (Fig. 4b). An increase of the Cu-OE Gln121 and ND1 His117 bond lengths to 2.73 and 2.68 A, respectively, with an SG Cys112-Cu(I)ND1 His46 angle of 156° produces an almost linear Cu(I) site with two strong bonds to the thiolate sull~hur atom of Cys112 (2.09/~) and the imidazole nitrogen atom of His46 (1.97 A), and two weak interactions with OE Gln121 and ND His117.
46
Albrecht Messerschmidt
Fig. 4. Copper site in M121Q mutant of azurin from AD. a) oxidised form, b) reduced form MOLSCRIPT [12]
Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidasesand Vanadium-Containing Enzymes
47
1.2.3 Substitution of Copper by Hg, Zn, Cd and Ni The apo-forms of the small blue copper proteins are able to bind other metals like Hg, Zn, Cd and Ni at their metal binding site. The Cu(II) has been substituted by Hg(II) in crystals of poplar plastocyanin and the crystal structure determined [36]. The quasi-tetrahedral coordination geometry found in Cu(II)plastocyanin remains almost unchanged in Hg(II)-plastocyanin, apart from the slight enlargement of the coordination polyhedron required to accommodate the mercury atom. When Hg(II) replaces Cu(II) in plastocyanin, a ring flip at Pro36 is observed. This ring flip is not unique. Pro36 undergoes a similar conformational change in the transition between the high and low-pH forms of reduced plastocyanin [34], and moves even further towards the CG-exo conformation in apoplastocyanin [50]. A zinc-containing azurin has been obtained as a by-product of heterologous expression of the gene encoding PA azurin in Escherichia coli and its X-ray structure determined [27]. The geometry at the metal binding site has been changed to some extent. The largest difference between Zn-azurin and Cu-azurin is at Gly45 O, which forms a weak bond (2.9 ~) with Cu but a strong bond (2.3 ~) with Zn. The movement of this atom and the slight adjustment of backbone atoms connected to it is probably induced by the preference of Zn for a tetrahedral coordination. A crystal structure of the PA azurin mutant N47D, which was prepared to test the influence of this mutation on the redox potential, showed that zinc was bound instead of copper. The zinc site of this mutant structure is almost identical to the wild-type zinc azurin structure [31]. A Cd-substituted AD azurin has been characterised by X-ray crystallography [42]. The Cd ion is not so close to the main-chain carbonyl of Gly45 (2.76 A) as the zinc ion in zinc azurin (2.32 ~). The distance between the Cd ion and the methionine ligand is 3.23 ~ compared to 3.40 .~ for the zinc-methionine bond distance in zinc azurin. This means that Cd behaves more like Cu than Zn in this protein. The structural data of this metal derivative are of interest for investigations on azurin and azurin mutants using i11mCd-perturbed angular correlation spectroscopy [51]. Copper can also be replaced by nickel in type 1 copper proteins. A crystal structure of the nickel-substituted W48M azurin mutant has been determined [30]. The nickel metal site of this mutant structure is in principle similar to the zinc metal site in wild-type zinc azurin, however, the Ni-S(Cys) bond length is increased to 2.49 A. 1.2.4
The CuAOinudear Copper Site The purple CuA copper site is one of four metal sites in cytochrome c oxidases and functions as a primary electron acceptor for cytochrome c (see [19, 52], for example). N20 reductase, another cupredoxin domain-containing enzyme, also contains a CuA site as one of its metal sites and has been characterised both in cytochrome c oxidase and N20 reductase as a mixed valence, Cu(1.5)-Cu(1.5) redox state, due to its seven-line EPR spectrum and other similar spectroscopic
48
Albrecht Messerschmidt
Fig.5. Schematic drawing of the dinuclear Cua site in CyoA(with permission from [11]) properties [53, 54]. Cua sites have been incorporated into amicyanin [55], azurin [56] and purple Cyo A (membrane-exposed domain from a quinol oxidase) [11]. The structure of the CUAsite in purple Cyo A, as derived from the crystal structure [11], is shown in Fig. 5. Two thiolate groups from different cysteines bridge the two coppers. Two histidines act as terminal ligands to each copper. The arrangement of these four ligands is symmetrical with respect to the copper pair. Asymmetry is introduced by the coordination of a methionine and main-chain carbonyl oxygen which each complete the distorted tetrahedral coordination of the individual coppers. All four Cu-S(Cys) bonds are ~- 2.2 A long. The copper-copper distance is 2.5 ~. The CuA sites in cytochrome c oxidases resemble that of Cyo A. The copper-copper distances in the X-ray structures of cytochrome c oxidase from Paracoccus denitrificans and bovine heart are 2.6 and 2.7 ~, respectively. It is remarkable that the distorted tetrahedral coordination of each copper is conserved and that the Cu2S2Im2 cluster allows complete electron delocalization, thereby ensuring that the small reorganisation energy is spread over both copper ions [19]. 1.3 Spectroscopy of Cupredoxins Cu(II) proteins have been classified as type 1, type 2, type 3 or CUA.The type 1, type 2 and CuAclassification is based on the nature and magnitude of their EPR hyperfine coupling and their VISible spectra (Fig. 6), whereas the type 3 copper is EPR-silent due to antiferromagnetic coupling of the two copper ions that constitute this copper species. The trigonal or bipyramidal type 1 site, as found in wild-type azurin, exhibits an axial EPR spectrum. This corresponds to the effective C3v symmetry of
49
Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidasesand Vanadium-Containing Enzymes
Type 1
Type 2
N~,
,,~N
"-cu .... S~f"
"~N
CuA
L,
""- .cu. . . ~mS/,,,
N/
S~"
/
,N
"""
Examples
Plastocyanin Azurin
EngineeredAzurin EngineeredSOD
Cytochrome c Oxidase N20 Reductase
Color Absorption max. in nm
Blue <->Green 460, 600
Yellow 400
Purple 480, 530, 800
EPR A H in 104 cm "1
< 90 (4 lines)
> 140 (4 lines)
< 30 (7 lines)
"L
Fig. 6. Copper coordination geometries in copper cysteinate proteins. Ligands denoted as S for cysteine thiolate,N for histidine imidazole and L for more weaklycoordinated methionine thioether or backbone carbonyl (with permission from [64])
the electronic distribution of the site and indicates that the detailed geometry and nature of the ligands in the NNS plane only weakly influence the shape of the electronic distribution. Distorted tetrahedral type 1 sites (the geometries of the different cupredoxin sites have been described in the previous paragraph) show a rhombic EPR spectrum consistent with a ligand field symmetry lowered to C2v. Stellacyanin, some type 1 copper site mutants, pseudoazurin and nitrite reductase belong to this subgroup. It is assumed that in these cases the rhombicity is due to a strong axial ligand field component [4]. W-band EPR, ENDOR and ESEEM studies have been carried out on frozen solutions and single crystals of wild-type azurin and the M121Q azurin mutant [57-60]. These studies more precisely define the character and position of the g-tensor and other spectroscopically relevant parameters. The data obtained can be used as a basis for an improved quantum-mechanical characterisation of the relevant copper sites. A strong band around 600 nm and a weaker one at 450 nm are found in the VISible region of the spectrum of type 1 sites. The strong absorption around 600 nm is mainly caused by a sulphur to Cu ligand-to-metal charge transfer (LMCT) transition. New theoretical studies identify the electron from a sulphur pn-orbital as being involved in the 600 nm LCMT transition [61]. The band around 450 nm has been assigned to a histidine to Cu LMCT transition but from recent resonance Raman (RR) studies on wild-type and mutant azurin as well as on superoxide dimutase mutants it is more probable that this band is caused by a second sulphur to Cu LMCT transition [62]. It is plausible that small changes in the copper coordination alter the electronic structure of the site and, as a consequence of this, the LMCT transfer properties. Associated with this is a change in the relative intensities of the 450 nm and 600 nm bands. Naturally occurring type 2 copper proteins do not contain sulphur ligands. Using genetic engineering, type 2 copper sites that contain a cysteinate as one of their ligands have been generated in azurin [63] and superoxide dismutase
50
Albrecht Messerschmidt
©
T1 Trigonal
T1 Distorted Tetrahedral
(430 - 405 cm-1)
(405 - 355 cm-1)
T1.5 Tetrahedral (360 - 340 cm-1)
/
T2 Distorted Tetragonal (365 - 340 cm-1)
T2 Tetragonal (320 - 300 cm-1)
Fig.7. Detailed coordination geometries in mononuclear copper-sulphur proteins (with per-
mission from [64]). Range of RR frequencies for the Cu-S(Cys) stretch in parentheses [16]. They exhibit a yellow colour (maximum absorbance at 400 nm). RR spectroscopy has been used to predict the coordination geometry in copper-sulphur proteins [64]. From these investigations it turns out that the mononuclear copper sites in copper-sulphur proteins can be subdivided in T1 trigonal, T1 distorted tetrahedral, T1.5 tetrahedral, T2 distorted tetragonal and T2 tetragonal (Fig. 7) [64]. This classification according to RR-spectroscopy is mainly based on the characteristic variation of the predominant Cu-S(Cys) stretching frequency and analyses of its correlation with ~460/E600absorption ratios and the geometry of the metal sites. 1.4 Redox Potentials and Electron Transfer Properties
The redox potential of type 1 copper sites in proteins varies from 185 mV for stellacyanin to 680 mV for rusticyanin, whereas the values for most are in the range from 260 mV to 380 mV. This is remarkably high as the redox potential
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
51
for the aqueous Cu(I)/Cu(II) pair (150 mV). It is thought that the stabilisation of the tetrahedral coordination in type 1 copper proteins is responsible for the higher redox potential because it is known from copper coordination chemistry that Cu(I) prefers a tetrahedral coordination and Cu(II) a tetragonal one. It was assumed that the nature of the axial ligand (usually methionine) would be responsible for tuning the redox potential [65]. This picture had to be extended as mutations of coordinating histidines or outside the first copper coordination sphere also changed the redox characteristics [66, 67]. An unexpected result was observed with the AD M121Q mutant. Met121 had been replaced to mimic a stellacyanin blue type 1 site. But the redox potential changed only by -23 mV with respect to the wild-type protein. The explanation came from the X-ray structure determination of both the oxidised and reduced state of the mutant. As already described in a previous paragraph, the oxidised form exhibits a strong axial coordination of the copper by OE of Gln121 stabilising the Cu(II) state. In the reduced form, a mainly two-fold coordination of the copper is found, which stabilises the Cu(I) state. Compared with the wild-type, both oxidation states are stabilised, resulting in an only moderate change in the redox potential. In PA azurin Met121 has been replaced by all other amino acids [46]. The redox potentials for this group of mutants varies from -105 mV to + 138 mV compared with the wild-type value at pH 7.0 [47]. Replacements of other copper ligands or in the second copper coordination sphere do not have drastic effects on the redox potential (changes are in the range for the Met121 mutations) (see [4], for example). Finally, titrating groups may have a considerable influence on the redox potential. The relatively low variation of the redox potential in the large number of azurin mutants designed to study the influence of the mutation on the redox potential supports the idea that the protein fold of the protein matrix around the redox site is also of importance. This is evident from the spatial structure of rusticyanin with its unusually high redox potential of 680 mV. The copper ligands in rusticyanin have been confirmed to be His, Cys, His, Met as in plastocyanin and many other blue type 1 copper proteins, for example [68]. Although the detailed geometry of the copper coordination site is not known yet it should be the significant abundance of hydrophobic residues in the loops closest to the copper site which causes this high redox potential by drastically changing the charge distribution in the environment of the redox centre. It is not the aim of this article to discuss the electron transfer properties of cupredoxins in detail. These have been described in several review articles (see [2, 4, 69], for example). Briefly, the type 1 copper sites in natural cupredoxins have been designed to make a rapid electron transfer possible. This has been accomplished by a copper site geometry that undergoes few changes upon alteration of the redox site (small reorganisation energy). The molecules have one or two surface regions for electron entrance and/or electron exit. One of these surface areas is a hydrophobic patch around the northern histidine as found in plastocyanin, azurin, and amicyanin. The structure of the electron transfer complex between methylamine dehydrogenase and amicyanin from Paracoccus denitrificans has been determined by X-ray crystallography [70]. The amicyanin docks at the light subunit of methylamine dehydrogenase with its hydro-
52
Albrecht Messerschmidt
phobic patch close to the northern histidine. The counterpart of the methylamine dehydrogenase is a hydrophobic surface area near the tryptophan tryptophylquinone (TTQ) cofactor. The copper atom of amicyanin and the redox factor of methylamine dehydrogenase are about 9.4 A apart, enabling a rapid electron transfer. The northern histidine is located between the two redox centres and may facilitate electron transfer between them. Plastocyanin has the hydrophobic patch and a remote acidic surface area. Cytochrome f, one of the natural redox partners, reacts at this site with plastocyanin [71]. The azurin system has been used to study long-range intramolecular electron transfer reactions. Farver, Pecht and other investigators started this research by following the electron transfer between the copper centre and a remote disulphide bridge [72-74]. Electron transfer rates were calculated using an electron pathway analysis based on the theory of Beratan and Onuchic [75]. The distant coupling problem has been studied in detail by Gray, Onuchic and associates [76]. In PA azurin positions 122, 124 and 126 of the C-terminal/3strand have been mutated to histidines. A [Ru(2,2"-bipyridine)2(imidazole)] 2÷ complex has been attached to the imidazole group of these residues and to His83 in the wild-type protein. The coupling between the copper ion and the Ru-complex coordinated to a histidine on the protein surface has been determined. In the case of the Ru-complex attached to histidines in positions 122, 124 and 126, the intervening medium was a single fl-strand, in the wild-type Rucomplex attached to His83 a section of a/3-sheet. On the basis of the results of these experiments a new theoretical approach for electron transfer beyond the single-pathway analysis has been developed. In the new theory, the protein matrix is reduced to only those relevant parts (tubes) which mediate the tunnelling matrix element. Such a tube is a tightly grouped family of pathways. These may be relevant for the coupling of a single tube only or for multiple tubes interfering with each other.
2 Metal Sites in Multi-Copper Oxidases 2.1 The Trinuclear Copper Active Site 2.1.1
Native Oxidized Enzyme The blue multi-copper oxidases ascorbate oxidase, laccase and ceruloplasmin catalyse the four-electron reduction of dioxygen to water with concomitant one-electron oxidation of the reducing substrate. They consist of domains with a cupredoxin fold. Ascorbate oxidase and laccase (their monomers have about 550 amino acids) are formed by three and ceruloplasmin (monomer about 1,000 amino acids) by six domains. The blue multi-copper oxidases have recently been reviewed [13] and a whole book has been devoted to the subject [77]. Apart from the blue type 1 copper sites, whose nature has been discussed in the previous section, the blue oxidases contain a trinuclear copper site which is located between the N- and C-terminal domains. The globular fold and the
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
53
Fig. 8. Schematic presentation of the monomer structure of ascorbate oxidase. The mono-
nuclear type 1 copper site is in the C-terminal domain whereas the trinuclear copper species is located between the N- and C-terminal domains
positions of the copper sites in the three-domain ascorbate oxidase are shown in Fig. 8. The atomic structure of the trinuclear copper site for ascorbate oxidase [43] is displayed in Fig. 9. The trinuclear cluster has eight histidine ligands symmetrically provided by the N- and C-terminal domains. It may be subdivided into a pair of copper atoms with histidine ligands whose coordinating N atoms (5 NE2 atoms and one ND1 atom) exhibit a trigonal prismatic arrangement. This pair is the putative type 3 copper. The remaining copper has two ligands and is the putative spectroscopic type 2 copper. Two oxygens are bound to the trinuclear species; as OH- or O2-, bridging the putative type 3 copper pair, and as OH- or H20 to the putative type 2 copper trans to the copper pair. An oxygen ligand in the centre of the three copper ions could not be detected. The bond lengths within the trinuclear copper site are similar to comparable binuclear model compounds [78, 79]. The average copper-copper distance in the trinuclear copper site of ascorbate oxidase is 3.74/~ and the individual distances do not deviate by more than 0.16 A from this mean value. Spectroscopic studies on laccase indicate tetragonal geometries for all three coppers in the cluster [80]. The tetragonal coordination geometries for all three coppers is not consistent with the structure. The coppers of the pair are both tetrahedrally coordinated, whereas the type 2 copper has three ligands. The existence of a central oxygen ligand would give rise to a pentacoordination of both copper pair atoms (but not a tetragonal-pyramidal coordination) and a square-planar coordination for the spectroscopic type 2-copper.
54
Albrecht Messerschmidt
t~
O
O
,.Q
8 0A
o~
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
55
There is experimental evidence in earlier studies that the type 2 copper is close to the type 3 copper and forms a trinuclear active copper site [81-86]. Solomon and associates described this metal binding site as a trinuclear active site, based on spectroscopic studies of azide binding to tree laccase [85, 86]. In ascorbate oxidase, the putative binding site for the reducing substrate is the type 1 copper [43]. Two channels which provide access from the solvent to the trinuclear copper site, the putative binding site of the dioxygen, could be identified. The structure of the trinuclear copper site in human ceruloplasmin, as determined by X-ray crystallography [14], is similar to that in ascorbate oxidase. However, in the current model of the 3.2 ~ resolution structure all histidine ligands are coordinated to the coppers by their RE2 atoms. 2.1.2
Type-2 Depleted (T2D) Form of Ascorbote Oxidose It is possible to selectively remove copper from the trinuclear species in the blue oxidases. This is documented by a loss of the type 2 EPR signal and of about 25% of the bound copper. The 2.5 .~ resolution X-ray structure of T2D ascorbate oxidase shows that about 1.3 copper ions per ascorbate monomer are removed [87]. The copper is lost from all three copper sites of the trinuclear copper species whereby the EPR-active type 2 copper is depleted somewhat preferentially (see Fig 10). Type-1 copper is not affected. The EPR spectra from polycrystalline samples of the native and T2D ascorbate oxidase were recorded. The native spectrum exhibits type 1 and type 2 EPR signals in a ratio of about 1:1 as expected from the crystal structure. The T2D spectrum reveals the characteristic resonances of the type 1 copper centre, as was also observed for T2D ascorbate oxidase in frozen solution, as well as the complete disappearance of the spectroscopic type 2 copper. The X-ray crystallography and the EPR spectroscopy seem to present a dilemma [88]. The X-ray structure shows all three sites depleted,
i2
°
!
Fig.lO. AveragedFOT2D-FCT2D-differenceelectrondensitymap plus atomicmodelaroundthe trinuclear copper site. Gontourlevels: - 18.0 solid line, 18.0 dashed line, magnitudesof holes less than -35.0
56
Albrecht Messerschmidt
whereas the EPR spectrum indicated selective removal of the type 2 copper. However, the EPR spectroscopy of the T2D form seems to be more complicated than usually assumed. A removal of copper from the trinuclear species means that we have a statistic mixture of copper pairs with the hole sitting on each copper, respectively. The relevant copper pairs are antiferromagnetically coupled and therefore EPR-silent. As a result of this, the T2D depleted duster does not produce an EPR signal. It is also conceivable that the remaining copper pairs are in the reduced state. In line with this are studies on T2D laccase which had been specificallylabelled with 63Cu and 65Cu.They show that there is either migration of Cu between the sites or a flip of ligands between the coppers [89].
2.1.3 Fully-ReducedForm of Ascorbate Oxidase The 2.2 ~ resolution X-ray structure analysis of fully-reduced ascorbate oxidase gave the following results [90]: The geometry at the type 1 copper remains much the same compared to the oxidised form as mentioned already in the previous section. The mean copper-ligand bond lengths of both subunits are increased by 0.04 A on average which is insignificant but may indicate a trend. A schematic drawing of the reduced form of ascorbate oxidase is shown in Fig 11.
~
,,P....~ ~, ,rto506
,
!
t
> HlO4
) Fig.11. Schematic drawing of the reduced form of ascorbate oxidase around the trinuclear copper site
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
57
The structural changes are considerable at the trinuclear copper site. Thus on reduction the bridging oxygen ligand OH1 is released and the two coppers CU2 and CU3 move towards their respective histidines and become tricoordinate, a preferred stereochemistry for Cu(I). CU2 and CU3 are each trigonally planar coordinated by their respective histidine ligands with equal bond lengths and bond angles within the accuracy of this X-ray structure determination. The copper-coopper distances increase from an average of 3.7 A to 5.1 A for CU2CU3, 4.4 A for CU2-CU4 and 4.1 A for CU3-CU4. The mean values of the copper-ligand distances of the trinuclear copper site are comparable to native oxidised ascorbate oxidase. CU4 remains virtually unchanged between reduced and oxidised forms. Coordinatively unsaturated copper(I) complexes are known from the literature. Linear bicoordinated [91] and T-shaped tricoordinated [92] copper(I) compounds have been reported. The copper nitrogen distances for both linearly arranged nitrogens are about 1.9 A, about 0.1 A shorter than copper nitrogen bond lengths in copper(II) complexes. In Tshaped copper(I) complexes, the bond length of the third ligand is increased. The copper ion CU4 has a T-shaped threefold coordination not unusual for copper(I) compounds. The structure of the fully reduced trinuclear copper site is quite different therefore from that of the fully oxidised resting form of the enzyme. 2.1.4
Peroxide Form of Ascorbote Oxidose
The 2.6 A resolution X-ray structure analysis of the peroxide form of ascorbate oxidase is illustrated in Fig. 12 [90]. The geometry at the type i copper site is not changed compared to the oxidised form. The copper-ligand average bond distances for both subunits show no significant deviations from those of the oxidised form. As in the reduced form, the structural changes are remarkable at the trinuclear copper site. The bridging oxygen ligand OH1 is absent, the peroxide binds terminally to the copper atom CU2 as hydroperoxide and the copper-copper distances increase from an average of 3.7 A to 4.8 A for CU2-CU3 and 4.5 ]l for CU2-CU4. The distance CU3-CU4 remains 3.7 A. The mean values of the copper-ligand distances of the trinuclear copper site are again comparable to native oxidised ascorbate oxidase and corresponding copper model compounds. The copper ion CU3 is tricoordinated, as in the reduced form, but the coordination by the ligating N atoms of the corresponding histidines is not exactly trigonal-planar and the CU3 atom is at the apex of a flat trigonal pyramid. The coordination sphere around CU4 is not affected and similar in all three forms. The copper atom CU2 is tetracoordinated to the NE2 atoms of the three histidines, as in the oxidised form, and by one oxygen atom of the terminally bound peroxide molecule in a distorted tetrahedral geometry. Its distance to CU3 increases from 4.8 A in the oxidised peroxide derivative to 5.1 A in the fully reduced enzyme. The bound peroxide molecule is directly accessible to solvent through a channel leading from the surface of the protein to the CU2-CU3 copper pair. This channel has already been described in [43] and its possible role as
58
Mbrecht Messerschmidt CU!
C507
J
f
H45o
H~II
~
/
\
x~104 Fig.
12. Schematicdrawingof the peroxide form of ascorbate oxidase around the trinuclear copper site
a dioxygen transfer channel has been discussed. An interesting feature is the close proximity of the imidazole ring of histidine 506 to the peroxide molecule. Histidine 506 is part of one possible electron transfer pathway from the type 1 copper to the trinuclear copper site and could indicate a direct electron pathway from CU1 to dioxygen. It may also help to stabilise important intermediate states in the reduction of dioxygen. The strong positive peaks at CU2 in both FONATrFOwoxand FOREDu-FOwox electron density maps could not be explained by a shift of CU2 alone. Occupancies of the copper atoms as well as of the oxygen atoms OH3 and the peroxide molecule were refined. Type-1 copper CU1 is almost unaffected. Copper atoms CU3 and CU4 are only partly removed, but copper atom CU2 is about 50% depleted. The oxygen ligands exhibit full occupancy. The treatment of crystals of ascorbate oxidase with hydrogen peroxide not only generates a well-defined peroxide binding but also a preferential depletion of the copper atom position CU2. In the copper-depleted molecules the coordinating histidine 106 adopts an alternative side-chain conformation as detected in the 2 FOFC-map calculated with the final peroxide derivative model coordinates.
Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidasesand Vanadium-Containing Enzymes
59
This map shows that Hisl06 moves away when the copper atom CU2 is removed and opens the trinuclear site even more. From the T2D crystal structure of ascorbate oxidase it is apparent that copper from all three metal binding sites of the trinuclear copper species are removed to different extents. The movement of the His 106 side-chain could explain how this process is accomplished. Copper depletion may also cause instability of the protein towards hydrogen peroxide. Reaction of hydrogen peroxide with ascorbate oxidase in solution in excess leads to a rapid degradation of the enzyme [93].This can be monitored in the UV/VIS PEOX-NATI difference spectrum by a negative band at 610 nm and a positive band at 305 nm. Adding four equivalents of hydrogen peroxide per monomer ascorbate oxidase does not lead to enzyme degradation and gives a positive peak at 305 nm indicative of peroxide binding. Unfortunately, it was not possible to monitor the UWVIS spectrum of dissolved crystals after X-ray data collection because of the dissociation of the bound peroxide in solution. The reaction of dioxygen with laccase or ascorbate oxidase has been investigated by several groups and is reviewed in [43] where the possible binding modes of dioxygen to binuclear and trinudear copper centres are also discussed. A novel mode of dioxygen binding to a binuclear copper complex was found in a compound synthesised by Kitajimia et al. [94] which should be mentioned in detail. The complex contains the peroxide in the p-e2:e2 mode i. e. side-on between the two copper(II) ions. Such a binding mode for dioxygen has been detected in the crystal structure of the oxidised form of Limulus Polyphemus subunit II hemocyanin [95]. However, the binding mode of dioxygen to the trinuclear copper site in the blue oxidases appears to be different, as can be seen from the Xray structure of the peroxide derivative of ascorbate oxidase. During its reaction with fully reduced laccase dioxygen binds to the trinuclear copper species and three electrons are very rapidly transferred to it resuking in the formation of an"oxygen intermediate" with a characteristic optical absorption near 360 nm [96, 97] and a broad low temperature EPR signal near g = 1.7 [98, 99]. The type 1 copper is concomitantly reoxidized when the low-temperature EPR signal is formed. The oxygen intermediate decays very slowly (tm ~- i to 15 s) correlated with the appearance of the type 2 EPR signal [100]. Solomon and co-workers [101-103] have identified and spectroscopically characterised an oxygen intermediate during the reaction of either fully reduced native tree laccase or T1Hg laccase with dioxygen. They concluded from their spectroscopic data that the intermediate binds as 1,1-/2 hydroperoxide between either CU2 and CU4 or CU3 and CU4. As it is unlikely that the dioxygen migrates or rearranges coordination during reduction, Messerschmidt et al. [90] proposed that the binding site and mode determined in the peroxide derivative of ascorbate oxidase is representative for all reaction intermediates of dioxygen and, using homology arguments, valid in all blue oxidases. Recently, a dicopper complex was synthesised that performs the reversible cleavage and formation of the dioxygen O-O bond within the complex [104]. The copper-copper distance is 3.56 A in the Cu2(/.I-~2:~2-O2) core where the O-O bond is not cleaved and 2.79 ~ in the Cu2(p-O)2 moiety with the cleaved O-O bond. This mechanism will be valid for oxygen-activating dinuclear copper enzymes, such as tyrosinase, but not for the trinuclear active
60
Albrecht Messerschmidt
copper site of the multi-copper oxidases. The copper-cop~per distance for the copper pair in fully reduced ascorbate oxidase is at 5.1 A m u c h too large to enable this binding mode. Furthermore, the end-on binding of the peroxide in ascorbate oxidase is indicative of an asymmetric action of the dioxygen or at least of its reaction intermediates.
2.1.5 Azide Form of Ascorbote Oxidose
The results of the 2.3 i resolution X-ray structure analysis are shown in Fig. 13 [90]. The geometry at the type 1 copper site is not changed compared to the native form. The copper-ligand bond distances averaged over both subunits show no significant deviations from those of the native form. Again, the structural changes are large at the trinuclear copper site. The bridging oxygen ligand OH 1 and water molecule 145 have been removed, CU2 moves towards the coordinating histidines and two azide molecules bind terminally to it. The copper-copper
//,=,,..~H 506
,,oo H448~_... ""
OH3---cu
~",.~z
"..
!i~/ 4
i l
• ~.~
/
I
\
104
Fig. 13. Schematicdrawing of the azideform of ascorbate oxidasearound the trinuclear copper site of ascorbate oxidase
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
61
distances increase from an average of 3.7 _~ to 5.1 ~ for CU2-CU3 and 4.6 ,~ for CU2-CU4. The distance CU3-CU4 is decreased to 3.6 A. The mean values for the copper-ligand distances of the trinuclear copper site are again comparable to native ascorbate oxidase and corresponding copper model compounds. The coordination of CU3 resembles that of the peroxide form. The threefold coordination by histidines is a very flat trigonal pyramid. The coordination sphere around CU4 is not affected. CU2 is pentacoordinated to the NE2 atoms of the three histidines, as in the reduced form, and to the two azide molecules. The two azide molecules are terminally bound at the apices of a trigonal bipyramid. Both azide molecules bind to the copper atom CU2, which is easily accessible from the broad channel leading from the surface of the protein to the CU2CU3 copper pair. It is not unexpected that the second azide molecule (az2 in Fig. 13) binds in a similar manner as the peroxide molecule, becausse azide is regarded as a dioxygen analogue. There is no azide molecule bridging either CU2 with CU4 or CU3 with CU4. The binding of azide in laccase as well as to ascorbate oxidase has been studied extensively by Solomon and co-workers [101,105-107] and by Marchesini and associates [108, 109] by spectroscopic techniques. The derived spectroscopic models involve the binding of two azide molecules for laccase and three azide molecules for ascorbate oxidase with different affinities. As the binding of the high-affinity azide molecules seemed to generate spectral features related to the type 2 and type 3 coppers, the spectroscopic data were interpreted as the binding of at least one azide molecule as a 1,3-/2 bridge between the type 3 copper ions and the type 2 copper ion. There are many structural studies of copper coordination compounds with azide ligands from mainly mononuclear and binuclear copper complexes but few with trinuclear copper complexes. A comprehensive review of copper coordination chemistry has been written by Hathaway (see [110], for example). Azide binds only terminally to mononuclear systems. Pentacoordination of nitrogen ligands, including azide, to Cu(II) is frequently found arranged as a trigonal bipyramid. In binuclear systems azide may bind terminally as 1,1-/2 or bridging as 1,3-/2. Similarly two azides may bind di-l,1-/2 or di-1,3-/2. The interaction with all three copper ions of a trinuclear complex may be either terminally as 1,1,1-/2 or bridging as 1,1,3-/2. In the X-ray crystal structure of ascorbate oxidase two azide molecules bind terminally to the type 3 CU2. Azide binding in ascorbate oxidase resembles therefore the binding of azide to an isolated copper ion. In fact there is little interaction of CU2 with CU3 and CU4 which are 5.1 ~ and 4.6/~ away, respectively. The coordination of the copper ion CU4 in the native oxidised structure is of some interest. It has only three ligating atoms at close distances forming a Tshaped coordination which is known for Cu(I) complexes (see the discussion of reduced the form). However, the ligand field is completed if we take into account the n-electron systems of the imidazole rings of histidines 62 and 450 (see Fig. 9). A ligand field with tetragonal-pyramidal symmetry around CU4 is then formed. The shortest distances of CU4 are 3.4/~ to CD2 450 and 3.6 .~ to CG 62. These distances are too long for strong copper n-electron interactions but the histidines will contribute to the CU4 ligand field.
62
Albrecht Messerschmidt
2.2 Aspectsof ElectronTransfer and MechanisticProperties The arrangement of the type 1 and the trinuclear copper redox centres in the muki-copper oxidases is crucial for their mechanistic properties. This arrangement as found in ascorbate oxidase [43] is illustrated in Fig. 14. The distances from the type 1 copper CU1 to the individual coppers of the trinuclear copper cluster, CU2, CU3 and CU4, are 12.2 A, 12.7 A and 14.9 A, respectively. The His506-Cys507-HisS08 amino acid sequence segment (ascorbate oxidase numbering) links the type 1 copper centre and the type 3 coppers as a bridging ligand. The same situation is found in the X-ray structure of ceruloplasmin [14]. The three type I copper sites of ceruloplasmin are arranged in a triangle whose corners are about 18 A aopart. The two type 1 copper sites located in domains 2 and 4 are more than 18 A away from the trinuclear cluster reducing their capability to directly interact with the trinuclear centre. Electron transfer properties and a proposal for the catalytic mechanism for ascorbate oxidase, which should be valid in principle for all blue oxidases, have been described in a recent review [111]. Briefly, the type 1 site in the C-terminal domain is where the electrons enter from the reducing substrate. In ceruloplasmin, the reducing substrate may react with the two other type 1 sites as well but the main entrance site will be the type 1 copper centre in the C-terminal domain. After reduction of the type 1 centre an intramolecular electron transfer to the trinuclear copper site takes place. The electron tunnelling may be completely through-space or follow individual electron transfer pathways. These may be through-bond, through-space or a combination of both. A through-bond pathway is available for both branches each with 11 bonds (Fig. 14). Two combined through-bond pathways can be chosen. The first one from CU1 to CU2 of the trinuclear centre involves a transfer from the SG atom of Cys507 to the main-chain carbonyl of Cys507 and through the hydrogen bond of this carbonyl to the ND1 atom of HisS06 (blue arrows in Fig. 14). The second one branches at the main-chain carbonyl of Cys507 and follows through bonds to CU3 (green arrows in Fig. 14). The catalytic mechanism contains the following stages: oxidised resting form; fully reduced form, OH-bridging ligand released, dioxygen binds probably to copper ion CU2; hydroperoxide intermediate, dioxygen has been reduced by two reduction equivalents, dioxygen double bond has been broken; oxygen radical intermediate O- coordinated to CU2, O-O bond cleaved, first water molecule released, CU1, CU3 and CU4 reoxidized; continuation of reduction of CU1 by a fifth reduction equivalent and electron transfer to CU3, release of second water molecule; continuation of reduction and attainment of the fully reduced state of the next catalytic cycle.
3 First X-ray Structure of a Vanadium-Containing Enzyme: Chloroperoxidase from the Fungus Curvularia inaequalis The first X-ray structure of a vanadium-containing protein, a chloroperoxidase from the fungus Curvularia inaequalis has been solved very recently [112]. The
Metal Sitesin Small BlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
63
Fig. 14. Region of the atomic model of ascorbate oxidase containing the type 1 and the trinuclear copper centres
chloroperoxidase from this fungus belongs to a class of vanadium enzymes that oxidise halides in the presence of hydrogen peroxide to the corresponding hypohalous acids. The mature enzyme consists of 609 amino acid residues with a calculated molecular mass of 67,488 Da. The 2.1 A crystal structure (R-value = 20%) of an azide chloroperoxidase complex reveals the geometry of the catalytic vanadium centre (Fig. 15). In the structure with a 2 mM azide mother liquor concentration, the vanadium has a trigonal bipyramidal coordination with three non-protein oxygen ligands (bond distances of about 1.65 A), one nitrogen ligand (NE2 atom) from His496 (bond distance of 2.25 ~) and an exogenous azide ligand (bond distance to coordinating nitrogen of 1.98 A). The negative charge of the VO3 group with vanadium in the oxidation state (V) is compensated by hydrogen bonds to several positively charged or hydrophilic protein side-chains and the main-chain amide nitrogen of Gly403. Oxygen OWl of the VO3 group forms hydrogen bonds to nitrogens NH1 of Arg360 (2.94 ~) and NH2 of Arg490 (2.93 A), oxygen OW2 to nitrogen NZ of Lys353 (2.72 A) and nitrogen N of Gly403 (2.99 A), and oxygen OW3 to oxygen OG of Ser402 (2.71 A)
64
Albrecht Messerschmidt
Fig. 15. Plot of the vanadium active site of chloroperoxidase from Curvularia inaequalis, MOLSCRIPT[12]
and nitrogen NE of Arg490 (3.04 ~). A water molecule from the solvent is hydrogen-bonded to the nitrogen atom N 1 of the bound azide molecule. The binding of vanadium as hydrogen vanadate (V) has now been confirmed by a difference Fourier map of the azide-free form. This difference map contains the VO4 group as its highest peak and one water molecule hydrogen-bonded to the apical oxygen atom. The difference electron density has the shape of a trigonal pyramid with the vanadium in the centre of the trigonal base plane, three oxygens at the vertices of this plane, and a fourth oxygen at the apex of the pyramid. The NE2 atom from His496 completes the trigonal bipyramidal coordination of the vanadium. The protein fold is mainly a-helical with two four-helix bundles as main structural motifs (Fig. 16) and an overall structure different from other structures. The vanadium centre is located on top of the second four-helix bundle. The molecule has an overall cylindrical shape with a length of 80 A and a diameter of 55 A. The secondary structure is mainly helical (about 44% of the atomic model), consisting of 20 helices, a small proportion of fl-structures and the rest extended strand and loop regions. Three residues near the azide or putative chloride binding site deserve special interest, the hydrophobic side chains of Trp350 and Phe397 as well as the imidazole ring of His404 (see Fig. 15). Both hydrophobic side-chains provide a hydrophobic environment, which seems to be necessary to stabilise chloride binding. His404 is on the other side of the putative chloride binding site and may play a crucial role as an acid-base group in catalysis. Steady-state kinetic data for vanadium chloroperoxidase show that the binding of peroxide is inhibited when a group with a pKa larger than 5 is protonated [113]. It may be that when His404 is protonated, binding of peroxide to the metal site is no longer possible as is also seen in heme iron peroxidases. This suggests some analogy to
Metal Sitesin SmallBlueCopperProteins,BlueCopperOxidasesand Vanadium-ContainingEnzymes
65
Fig. 16. Ribbon-type representation of the chloroperoxidase molecule including the vanadium binding site, MOLSCRIPT [12]
the reaction of heme-containing cytochrome c peroxidase with peroxide and the vanadium enzyme. A detailed catalytic reaction scheme has not been elaborated due to the lack of structural information about catalytic intermediates. The high thermostability of the enzyme is mainly due to the packing of the helices to a compact molecule. An amino acid sequence comparison with vanadium-containing bromoperoxidase from the seaweed Ascophyllum nodosum shows high similarities in the regions of the metal-binding site, with all hydrogen vanadate(V) interacting residues conserved except for Lys353, which is Asn. The crystal structure of the vanadium-containing bromoperoxidase from Ascophyllum nodosum has recently been solved by the Schomburg's group in Braunschweig, Germany, but has not been published yet. Acknowledgements. The author wishes to thank Prof. R. Huber for supporting the work on the reviewed projects and for his valuable suggestions. The Deutsche Forschungsgemeinschaft (Schwerpunktthema: Bioanorganische Chemie) is thanked for financial support.
References 1. Adman ET (1985) In: Harrison P (ed) Metalloproteins. Verlag Chemie, Weinheim, Germany, p 1 2. Sykes AG (1991) Adv Inorg Chem 36:377 3. Adman ET (1991) Adv Protein Chem 42:145 4. Canters GW, Gilardi G (1993) FEBS Lett 325:39
66
Albrecht Messerschmidt
5. Pascher T, Bergstr6m J, Malmstr6m BG, V/inngard T, Lundberg LG (1989) FEBS Lett 258: 266 6. Hart PJ, Nersissian AM, Herrmann RG, Nalbandyan RM,Valentine IS Eisenberg D (1996) Protein Science 5: 2175 7. Guss ]M, Merritt EA, Phizackerly RP, Hedman B, Murata M, Hodgson KO, Freeman HC (1988) Science 241:806 8. Hunt AH, Toy-Palmer A, Assa-Munt N, Cavanagh J, Blake II RC, Dyson HJ (1994) J Mol Biol 244: 370 8a. Walter RL, Ealick SE, Friedman AM, Blake II RC, Proctor C, Shoham M (1996) J Mol Biol 263: 730 9. Iwata S, Ostermeier C, Ludwig B, Michel H (1995). Nature 376:660 10. Tsukihara T, Aoyama H, Yamashita E, Tomizaki T, Yamaguchi H, Shinzawa-Itoh K, Nakashima R, Yaono R, Yoshikawa S (1996) Science 272:1136 11. Wilmans M, Lappalainen P, Kelly M, Sauer-Eriksson E, Saraste M (1995) Proc Natl Acad Sci USA 92:11955 12. Kraulis P (1991) J Appl Cryst 24:946 13. Messerschmidt A (1993) Adv Inorg Chem 40:121 14. Zaitseva I, Zaitsev V, Card G, Moshkov K, Bax B, Ralph A, Lindley P (1996) J Biol Inorg Chem 1: 15 15. Adman ET, Godden JW, Turley SJ (1995) ] Biol Chem 270:27458 16. Han J, Loehr, TM, Lu Y,Valentine JS, Averill BA, Sanders-Loehr (1993) J Amer Chem Soc 115:4256 17. Lu Y, LaCroix LB, Lowery MD, Solomon EI, Bender CJ, Peisach ], Poe JA, Gralla EB, Valentine JS (1993) J Amer Chem Soc 115:5907 18. Andrew CR, Yeom H, Valentine JS, Karlsson BG, Bonander N, van Pouderoyen G, Canters GW, Loehr TM, Sanders-Loehr J (1994) J Amer Chem Soc 116:11489 19. Ramirez BE, Malmstr6m BG, Winkler JR, Gray HB (1995) Proc Natl Acad Sci USA 92:11949 20. Nat H, Messerschmidt A, Huber R, van de Kamp M, Canters GW (1991) J Mol Biol 221 : 765 21. Nar H (1992) PhD thesis, Technical University Mfinchen 22. Baker EN (1988) J Mol Biol 203:1071 23. Dodd FE, Hasnain SS, Abraham ZHL, Eady RR, Smith BE (1995) Acta Cryst D51 : 1052 24. Nat H, Messerschmidt A, Huber R, van de Kamp M, Canters GW (1991) ] Mol Biol 218:427 25. Romero A, Hoitink CWG, Nar H, Huber R, Messerschmidt A, Canters GW (1993) J Mol Bio1229:1007 26. Messerschmidt A (1996) unpublished results 27. Nar H, Huber R, Messerschmidt A, Filippou AC, Barth M, Jaquinod M, van de Kamp M, Canters GW (1992) Eur J Biochem 205:1123 28. Shepard WEB,Anderson BF, Lewandowski DA, Norris GE, Baker EN (1990) J Amer Chem Soc 112:7817 29. Tsai L-C, Sj61inL, Langer V, Pascher T, Nar H (1995) Acta Cryst D51 : 168 30. Tsai L-C, Sj61inL, Langer V, Bonander N, Karlsson BG, V~inngard T, Hammann C, Nar H (1995) Acta Cryst DS1:711 31. Sj61inL, Tsai L-C, Langer V, Pascher T, Karlsson G, Nordling M, Nar H (1993) Acta Cryst D49:449 32. Hammann C, Messerschmidt A, Huber R, Nar H, Gilardi G, Canters GW (1996) J Mol Biol 255: 362 33. Guss JM, Bartunik HD, Freeman HC (1992) Acta Cryst B48:790 34. Guss JM, Harrowell PR, Murata M, Norris VA, Freeman HC (1986) J Mol Biol 192:61 35. Collyer CA, Guss JM, Sugimura Y,Yoshizaki F, Freeman HC (1990) J Mol Biol 211 :617 36. Church WB, Guss ]M, Potter ]J, Freeman HC (1986) J Biol Chem 261 : 34 37. Petratos K, Dauter D, Wilson KS (1988) Acta Cryst B44:628 38. Vakoufari E, Wilson KS, Petratos K (1994) FEBS Lett 347:203
Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidasesand Vanadium-Containing Enzymes
67
39. Cunane LM, Chen Z-W, Durley RCE, Mathews FS (1996) Acta Cryst D52:676 40. Romero A, Nar H, Huber R, Messerschmidt A, Kalverda AP, Canters GW, Durley R, Mathews FS (1994) J Mol Bio1236:1196 41. Guss JM, Merritt EA, Phizackerley RP, Freeman HC (1996) unpublished results 42. Baker EN, Anderson BF, Blackwell KA (1995) Protein Data Bank Reference 1AIZ 43. Messerschmidt A, Ladenstein R, Huber R, Bolognesi M, Avigliano L, Petruzzelli R, Rossi A, Finazzi-Agr6 A (1992) J Mol Biol 224:179 44. Guckert JA, Lowery MD, Solomon EI (1995) J Amer Chem Soc 117:2817 45. Kroes SJ, Hoitink, CWG,Andrew CR, Ai J, Sanders-Loehr, Messerschmidt A, Hagen WR, Canters GW (1996) Eur J Biochem 240:342 46. Karlsson BG, Nordling M, Pascher T, Tsai L-C, Sj61inL, Lundberg LG (1991) Protein Eng 4:343 47. Pascher T, Karlsson BG, Nordling M, Malmstr6m BG,V~inngard T (1993) Eur J Biochem 212:289 48. Kroes SJ, Andrew CR, Sanders-Loehr J, Canters GW (1995) J Inorg Biochem 59:661 49. Fields BA, Cuss JM, Freeman HC (1991) J Mol Biol 222:1053 50. Garrett TPJ, Clingeleffer DJ, Guss JM, Rogers SJ, Freeman HC (1984) J Biol Chem 259: 2822 51. Danielsen E, Bauer R, Hemmingsen L, Andersen M-L, Bjerrum MJ, Butz T, Troeger W, Canters GW, Hoitink CWG,Karlsson G, Hansson O, Messerschmidt A (1995) J Biol Chem 270:573 52. Gennis R, Ferguson-Miller S (1995) Science 269:1063 53. Kroneck PMH, Antholine WE, Riester J, Zumft WG (1988) FEBS Lett 242:70 54. Kroneck PMH, Antholine WE, Kastrau DHW, Buse G, Steffens GCM, Zumft WG (1990) FEBS Lett 268: 274 55. Dennison C, Vijgenboom E, de Vries S, van der Oost J, Canters GW (1995) FEBS Lett 365: 92 56. Hay M, Richards JH, Lu Y (1995) Proc Natl Acad Sci USA 93:461 57. Coremans, JWA, Poluetkov OG, Groenen EJJ, Canters GW, Nar H, Messerschmidt A (1994) J Amer Chem Soc 116:3097 58. Coremans JWA, van Gastel M, Poluetkov OG, Groenen EJJ, den Blaauwen T, van Pouderoyen G, Canters GW, Nar H, Hammann C, Messerschmidt A (1995) Chem Phys Lett 235: 202 59. Coremans JWA,Poluetkov OG, Groenen EJJ, Canters GW,Nar H, Messerschmidt A (1996) J Amer Chem Soc 118:12141 60. Coremans JWA, Poluetkov OG, Groenen EJJ, Warmerdam G, Canters GW, Nar H, Messerschmidt A (1996) J Phys Chem 100:19706 61. Solomon EI, Baldwin MJ, Lowery MD (1992) Chem Rev 92:521 62. Sanders-Loehr J (1993) In: Karlin KD, Tyeklar Z (eds) Bioinorganic Chemistry of Copper. Chapman & Hall, New York, NY, p 51 63. den Blaauwen T, Hoitink CWG, Canters GW, Han J, Loehr TM, Sanders-Loehr J (1993) Biochemistry 32:12455 64. Andrew CR, Sanders-Loehr J (1996) Acc Chem Res 29:365 65. Gray HB Malmstr6m BG (1983) Comments Inorg Chem 2:203 66. Hoitink CWG, Canters GW (1992) J Biol Chem 267:13836 67. Nishayama M, Suzuki J, Ohnuki T, Chang HC, Horinouchi S, Turley S, Adman ET, Beppu T (1992) Prot Eng 5:117. 68. Casimiro DR, Tyo-Palmer A, Blake RC, Dyson HJ (1995) Biochemistry 34:6640 69. Sykes AG (1990) Struct Bond 75:175 70. Chen L, Durley R, Poliks BI, Hamada K, Chen Z, Mathews FS, Davidson VL, Satow Y, Huizinga E, Vellieux FMD, Hol WGJ (1992) Biochemistry 31 : 4959 71. Beoku-Betts D, Chapman SK, Knox CV, Sykes AG (1985) Inorg Chem 24:1677 72. Farver O, Pecht I (1992) J Amer Chem Soc 114:5764 73. Farver O, Skov LK, Nar H, van de Kamp M, Canters GW, Pecht I (1992) Eur J Biochem 210:399
68
Albrecht Messerschmidt
74. Farver O, Pecht Iet al (1993) Biochemistry 32:7317 75. Beratan DN, Onuchic JN, Winkler JP, Gray HB (1992) Science 258:1740 76. Regan J], di Bilio A], Langen R, Skov LK, Winkler JR, Gray HB, Onuchic JN (1995) Chem & Biol 2:489 77. Messerschmidt A (1997) Multi-Copper Oxidases. World Scientific, Singapore 78. Karlin KD, Hayes JC, Gultneh Y, Cruse RW, McKnown JW, Hutchinson JP, Zubieta J (1984) J Amer Chem Soc 106:2121 79. Chaudhuri P, Ventor D, Wieghardt K, Peters E, Peters K, Simon A (1985) Angew Chem 97:55 80. Cole JL, Clark PA, Solomon EI (1990) J Amer Chem Soc 112:9534 81. Br/inden R, Deinum J (1977) FEBS Lett 73:144 82. Martin CT, Morse RH, Kanne RM, Gray HB, Malmstr6m BG, Chan SI (1981) Biochemistry 20: 5147 83. Morpurgo L, Desideri A, Rotilio G (1982) Biochem 1 207:625 84. Winkler ME, Spira DJ, LuBien CD, Thamann TJ, Solomon EI (1982) Biochem Biophys Res Com 107:727 85. Allendorf MD, Spira DJ, Solomon EI (1985) Proc Natl Acad Sci USA 82:3063 86. Spira-Solomon DJ, Allendorf MD, Solomon EI (1986) 108:5318 87. Messerschmidt A, Steigemann W, Huber R, Lang G, Kroneck PMH (1992) Eur J Biochem 209:597 88. Beinert H (1996) J Inorg Biochem 64:79 89. McMillin DR, Eggleston MK (1997) In: Messerschmidt A (ed) Multi-Copper Oxidases. World Scientific, Singapore, p 129 90. Messerschmidt A, Luecke H, Huber R (1993) J Mol Biol 230:997 91. Schilstra MJ, Birker PJMW,Verschoor GC, Reedijk J (1982) Inorg Chem 21:2637 92. Sorell TN, Malachowski MP (1983) Inorg Chem 22:1883 93. Marchesini A, Kroneck PMH (1979) Eur J Biochem 101:65 94. Kitajima N, Fujisawa K, Moro-oka Y (1989) J Amer Chem Soc 111: 8975 95. Magnus K, Ton-That H (1992) J Inorg Biochem 47:20 96. Andreasson L-E, Br~indenR, Malmstr6m BG, V~inngard T (1973) 32:187 97. Andreasson L-E, Br[inden R, Reinhammar B (1976) Biochim Biophys Acta 438:370 98. Aasa R, Br/inden R, Deinum J, Malmstr6m BG, Reinhammar B, V~inngard (1976) FEBS Lett 61:115 99. Aasa R, Br~inden R, Deinum J, Malmstr6m BG, Reinhammar B, V/inngard (1976) Biochem Biophys Res Com 70:1204 100. Br~inden R, Deinum J (1978) Biochim Biophys Acta 524:297 101. Cole JL, Clark PA, Solomon EI (1990) J Amer Chem Soc 112:9534 102. Cole JL, Ballou DP, Solomon EI (1991) J Amer Chem Soc 113:8544 103. Clark PA, Solomon EI (1992) J Amer Chem Soc 114:1108 104. Halfen ]A, Mahapatra S, Wilkinson EC, Kaderli S, Young Jr VG, Que Jr L, Zuberbuehler AD, Tolman WB (1996) Science 271:1397 105. Allendorf MD, Spira DJ, Solomon EI (1985) Proc Natl Acad Sci USA 82:3063 106. Spira-Solomon DJ, Allendorf MD, Solomon EI (1986) J Amer Chem Soc 108:5318 107. Cole ]L, Avigliano L, Morpurgo L Solomon EI (1991) J Amer Chem Soc 113:9080 108. Casella L, Gulotti M, Pallanza G, Pintar A, Marchesini A (1988) Biochem J 251:441 109. Casella L, Gulotti M, Pintar A, Pallanza G, Marchesini A (1989) J Inorg Biochem 37:105 110. Hathaway BJ (1987) Compr Coord Chem 5: 111. Messerschmidt A (1997) In: Messerschmidt A (ed) Multi-Copper Oxidases. World Scientific, Singapore, p 23 112. Messerschmidt A, Wever R (1996) Proc Natl Acad Sci USA 93: 392 113. van Schijndel JWPM, Barnett P, Roelse J, Vollenbroek EGM, Wever R (1994) Eur J Biochem 225:151
Structure and Function of the Xanthine-Oxidase Family of Molybdenum Enzymes M a r i a Joao Rom~to 1. a n d R o b e r t H u b e r 2 1 Instituto de Tecnologia Quimica e Biol6gica, Apt. 127, 2780 Oeiras and Instituto Superior T6cnico, Dep. Qulmica, 1096 Lisboa Codex, Portugal, E-maih
[email protected] 2 Max-Planck-Institut ffir Biochemie, am Klopferspitz 18a, D-82152 Martinsried, Germany
This work gives an account of the recent achievements which have contributed towards the understanding of the structure and function of the xanthine oxidase family of enzymes-the molybdenum hydroxylases. It is based essentially on the crystallographic data of the aldehyde oxido-reductase from Desulfovibrio (D.) gigas, a member of that family,whose structure is described in detail. Comparisons are made, whenever appropriate, with spectroscopic, kinetic and model compound studies. Mechanistic implications of the crystal structure of the D. gigas enzyme are considered and extended to the xanthine oxidase family. Keywords: Hydroxylase, molybdoenzymes, molybdopterin, protein crystallography, xanthine oxidase
List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
70
1
Introduction ..................................
70
2
Overview of the Xanthine Oxidase Family of Molybdenum Enzymes
71
3
Crystal Structure o f t h e A l d e h y d e O x i d o - R e d u c t a s e f r o m
Desulfovibrio gigas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
75
3.1 3.2 3.3 3.4 3.5 3.6
S t r u c t u r a l a n d D o m a i n A r r a n g e m e n t of the Protein . . . . . . . . . . The [2Fe-2S] Centers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The M o l y b d o p t e r i n Cofactor . . . . . . . . . . . . . . . . . . . . . . . . S t r u c t u r e a n d E n v i r o n m e n t o f the Metal Centers . . . . . . . . . . . . The M o l y b d e n u m Site a n d Its E n v i r o n m e n t . . . . . . . . . . . . . . . The D i m e r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
76 79 79 81 87 88
4
C o m p a r i s o n o f t h e A l d e h y d e O x i d o - R e d u c t a s e f r o m Desulfovibrio gigas w i t h the X a n t h i n e O x i d a s e F a m i l y o f E n z y m e s . . . . . . . . . .
89
5
Structure-Based Catalytic Mechanism
90
6
Conclusions
7
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
..................................
..................
92 93
* Corresponding Author. Structure and Bonding, Vol. 90 © Springer Verlag Berlin Heidelberg 1998
70
Maria Jofio Romgo • Robert H u b e r
List of Abbreviations AOR DMSOR Mop Mod MPT MCD MGD MAD MHD XO XDH
aldehyde ferredoxin oxido-reductase from Pyrococcusfuriosus dimethylsulfoxide reductase aldehyde oxido-reductase from Desulfovibrio gigas aldehyde oxido-reductase from Desulfovibrio desulfuricans ATCC27774 mononucleotide form of molybdopterin molybdopterin cytosine dinucleotide molybdopterin guanine dinucleotide molybdopterin adenine dinucleotide molybdopterin hypoxanthine dinucleotide xanthine oxidase xanthine dehydrogenase
1 Introduction Molybdenum-containing enzymes can be grouped into two classes: nitrogenase, which catalyzes the reduction of dinitrogen to ammonia and where molybdenum is part of a heterometal FeMo-cofactor [1]; and hydroxylases or oxotransferases, which promote a variety of two-electron oxidation-reduction reactions, whereby oxygen (oxo) atom transfer occurs. An essential role of molybdenum is the catalysis of the controlled oxo-transfer reaction coupled to an electron transfer between substrate and other cofactors, such as Fe/S centers, hemes or flavins. Coupling of both functions leads to a formal direct transfer of an oxygen atom from the metal center to the substrate (oxotransferase activity [2]). Among the molybdenum hydroxylases which are currently known (see [3] for a review), xanthine oxidases, sulfite oxidases, nitrate reductases, aldehyde oxidases and DMSO reductases have been characterized in greater detail. Molybdenum oxotransferase enzymes catalyze the following general reactions, where water has been shown to be the source of the incorporated oxygen atom [4], such that reducing equivalents are generated rather then consumed [60, 61] RH + H20 ~
ROH + 2e- + 2H +
(1.1)
RH = aldehyde or aromatic heterocycle
RR'E + H20 __k_>RR'E = O + 2e- + 2H +
(1.2)
E = N, S (simple oxo transfer without cleavage of a C-H bond)
Xanthine oxidase from bovine milk, due to its ready availability, is the prototype for molybdenum hydroxylases and has been intensively studied for the past forty years. Many other oxomolybdenum enzymes have been investigated within the last decade. In a recent review on molybdenum oxotransferases [3], an extensive list of 68 enzymes is given, according to the information available at the end of 1995. These enzymes have been classified into three families on the basis of the reactions they catalyze, as well as on the basis of the characteristics of their molybdenum centers: (1) The xanthine oxidase family, (2) the sulfite
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
71
oxidase and assimilatory nitrate reductase family and (3) the DMSO reductase family. This classification is supported by amino acid homologies within these protein families, which are found in a wide range of organisms. The large xanthine oxidase family of enzymes may be considered one of the true hydroxylases, with the substructure dithiolene of the molybdopterin cofactor (fac) coordinated to a MoOS (H20) unit. The MoOS coordination was suggested by EXAFS experiments [8, 9, 57], but the presence of an additional water ligand was established by crystallography [6, 56]. Members of this family have been found broadly distributed within eukaryotes, prokaryotes and archaea. They catalyze the oxidative hydroxylation of aldehydes and aromatic heterocycles in reactions involving C-H bond cleavage (reaction 1.1). Sulfite oxidase and assimilatory nitrate reductases possess afacMo02 unit. There is uncertainty about additional coordination positions of the molybdenum. They have been found in eukaryotes and catalyze a simple oxo transfer to the lone pair of sulrite in sulfite oxidase (E=S, reaction 1.2-a) or the reverse reaction in nitrate reductase (E=N, reaction 1.2-b). In the DMSO reductase family, other enzymes such as the biotin-S-oxide reductase and bacterial dissimilatory nitrate reductases have been included, which also follow the overall stoichiometry of reaction 1.2-b. This family of enzymes possesses the Mo-cofactor with a bisdithiolene coordination of the molybdenum, established by recent X-ray structural data [59, 65]. These studies showed that the Mo atom is also coordinated by the side-chain oxygen of a serine (Ser 147). This class of enzymes seems to be structurally more diverse when compared to the other two, and its members have only been found in bacteria and archaea. Enzymes have been included in this class on the basis of significant sequence homology or spectral similarity. It is the purpose of this review to focus on the xanthine oxidase family of molybdenum enzymes with particular emphasis on the crystal structure of the aldehyde oxido-reductase from Desulfovibrio gigas (Mop), as the first representative of this group of enzymes for which a three-dimensional structure is available [6]. In the following, structural data on Mop will be provided and, whenever possible, comparisons will be made with literature data derived from spectroscopic - EPR, EXAFS, resonance Raman, MCD, ENDOR - and kinetic techniques, as well as from electron transfer studies. The mechanistic implications of the Mop structure for a general catalytic mechanism for the xanthine oxidase family of molybdenum enzymes are presented.
2 Overview of the Xanthine Oxidase Family of Molybdenum Enzymes Molybdenum oxotransferases in general possess a common Mo=O group in the metal coordination sphere, which is the reason why they have been called oxomolybdenum enzymes [7]. The molybdenum hydroxylases, which constitute the xanthine oxidase family (containing an MoOS unit), have been assigned on the basis of their irreversible inhibition by cyanide which reacts with Mo=S, releasing thiocyanate. Until X-ray structural data were available for this class of enzymes, the main techniques used to study the molybdenum coordination sphere were EXAFS [8, 9] ([58] for Mop) and EPR from the Mov species [10]
72
Maria Jofio Romio • Robert Huber
([11, 52] for Mop). EXAFS m e a s u r e m e n t s have shown that the metal center in its oxidized form has at least one oxo group with an Mo=O distance of about 1.7 ~, as well as (at least) two thiolate ligands at distances of about 2.4 A (see Fig. 1). The latter feature is c o m m o n to all o x o m o l y b d e n u m enzymes, and is due to the coordination of m o l y b d e n u m to an organic cofactor c o m m o n l y designated as " m o l y b d o p t e r i n ' . In spite of the lability of this cofactor when isolated from the protein matrix, its basic pterin ring structure and dithiolene side-chain were proposed on the basis of chemical and spectroscopic analysis of the cofactor iso2.33/~ (2.35)
2.15 A (2.18)
~
1.67A (1.74)
f
1.66/k (1.68)
___=.
2.00/k (2.14)
2.44 A (2.47) (1.98A )
b) reduced enzyme (Mo TM) [81([57])
a) oxidized enzyme (MoVO [81([57]) 1.67/~. (1.70)
SH-~
1.66 A
(~\ .H
>o
fs,,,
2.38/k (2.36) 2.27/k (2.34) c) reduced enzyme-alloxanthine complex (MOW)
[91([571)
SH-')
2.40 A
~_..-N~_LNH
r
Y--
2.08 A
d) reduced enzyme-violapterin complex (MOW) [91 2.11/~
O
,----s
~
C 1.66/~ kgH~,
1.68 A
! #o -
2.47/~
'~ ~-1.90a 1.68/~
e) oxidized Mop (desulfo) (MovI)
[581
1.
~..S_.~.~M0...~ ? 2.4oA f) reduced Mop (desulfo) (Mo w) [58]
Fig. The molybdenum coordination, as seen in Mop, with metal ligand distances obtained by EXAFS for various derivatives of XO and Mop: a xanthine oxidase, oxidized [8, 57]; b xanthine oxidase, reduced [8, 57]; c complex of xanthine oxidase with alloxanthine, reduced [9, 57]; d complex of xanthine oxidase with violapterin, reduced [9]; e desulfo form of Mop, oxidized [58]; f desulfo form of Mop, reduced [58]
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
73
lated from the enzymes in different modified or inactivated forms [12-15]. However, the presence of a pyran ring in the structure was established by crystallography for the tungsten-containing aldehyde oxido-reductase from Pyrococcus furiosus [16], for the aldehyde oxido-reductase from Vesulfovibrio gigas (Mop) [6] and for the DMSO reductases from two Rhodobacter species [59, 65]. The pyran ring closure may occur by attack on the side-chain 9"OH to C7 of a dihydropterin, possibly after incorporation and binding to the enzyme. The molybdopterin cofactor is coordinated to the metal via its dithiolene function and may be present either in dinucleotide forms or in the simpler monophosphate form. In Mop, the cofactor is the dinucleotide of cytosine molybdopterin cytosine dinucleotide (MCD) -, but in other enzymes from prokaryotic sources is found as the guanine (MGD) [17], adenine (MAD) [18] or hypoxanthine (MHD) [18] dinucleotide. The simpler monophosphate form (MPT), also found in some bacterial enzymes [3], is the only form present in all known enzymes from eukaryotic sources, and the diversity of the pterin cofactor within known molybdenum-containing enzymes seems to be related to the species of origin rather than to the enzymatic function, as shown within the xanthine oxidase family, where MPT has been found in eukaryotic enzymes and MCD in bacterial enzymes reported so far [3]. Members of the xanthine oxidase family show about 25 % sequence identity and 60-70 % sequence similarity, with higher conservation in segments involved in the binding of the metal centers and redox-active sites. Enzymes belonging to this class are either: homodimers, a2, or dimers of heterotrimers, a2/J2y2. Mop, aldehyde oxidases and eukaryotic xanthine oxidases and xanthine dehydrogenases are organized as a2 homodimers, with all redox-active cofactors confined within a single polypeptide chain. The common folding pattern for this group of enzymes, starts with the two [2Fe-2S] domains, followed by a flavin domain (which is absent in Mop and replaced by an extended connecting segment, cf. Sect. 3.2) and finishes with two large domains responsible for binding the molybdopterin cofactor [5, 19]. Other groups of hydroxylases have been included in the xanthine oxidase family of enzymes, due to analogies in their molybdenum centers and the reactions catalyzed: the CO dehydrogenases [20-22], the isoquinoline oxidoreductases [23-29] and nicotine dehydrogenases [30-33], all from bacterial and archaeal sources. They are organized as a2fl2y2 structures, where the a subunit harbors the two [2Fe-2S] centers, the fl subunit the flavin and the y subunit the molybdenum cofactor. Xanthine oxidase is the prototype of the molybdenum hydroxylases and has been intensively studied using spectroscopic and kinetic techniques, which have contributed to characterizing it both structurally and functionally. Relations to a number of inorganic model systems could be deduced from such studies. These results have been described and summarized in a number of recent reviews [3, 34, 35, 101]. We will now focus on some general features of xanthine oxidase, highlighting relevant points which can be illustrated with details from the crystal structure of Mop. Xanthine oxidase catalyses the oxidation of xanthine to uric acid, using dioxygen as the physiological electron acceptor (or NAD + in the case of xanthine dehydrogenases).
74
Maria Jo~o Romgo • Robert Huber O
o¢-
O
2 H2°*+"
NH
" O~NH
N
> NH
O*H
+ H202
In addition to xanthine, it can also oxidize other aromatic heterocycles and aldehydes, although with little specificity. Xanthine is oxidized at the molybdenum center, whereby MovI is reduced to MoTM and reducing equivalents are transferred to 02 at the flavin site. Electron transfer between the molybdenum center and the FAD is mediated by the Fe/S centers. Intramolecular electron transfer within the xanthine oxidase family of molybdenum enzymes is an essential aspect of catalysis and has been studied by different techniques: flash photolysis [36, 37], pulse radiolysis [38, 39] and pH-jump perturbation [40-42]. Rate constants for electron transfer between redox-active centers have been determined and show that the equilibration of reducing equivalents within xanthine oxidase is rapid in comparison to the overall catalysis and is not rate-determining [40]. EPR was one of the first tools employed for the study of the molybdenum center of xanthine oxidase and a variety of Mov species were detected, either in the course of equilibrium reductive titrations, or transiently within reaction with substrate [10]. Other characteristic EPR signals have been observed from complexes of the enzyme with inhibitors such as arsenite [43-45], methanol [46], ethylene glycol [47] and alloxanthine [48]. The alloxanthine complex, characterized also by EXAFS (Fig. 1), gives, upon a one-electron reduction, an EPR signal resembling the so-called "very rapid" signal, observed very early and transiently in the course of the reaction of xanthine oxidase with excess alloxanthine. While this "very rapid" signal is quite anisotropic and exhibits no proton hyperfine coupling, the so-called"rapid" Mov EPR signal, which also arises in the course of reaction with substrate, belongs to two different types, "rapid type 1" and "rapid type 2", depending on the nature of the proton hyperfine coupling observed [7, 10]. The Mov EPR signals of ,type 2", exhibited by xanthine oxidase upon reaction with xanthine [49] (gl.2,3= 1.9951, 1.9712, 1.9616), are quantitatively very similar to the corresponding signals reported for Mop upon reaction with salicylaldehyde [11] (gl,a,3= 1.9882(3), 1.9702(3), 1.9643(3)), suggesting similarity in their active sites. Also the two [2Fe, 2S] centers - Fe/S I and Fe/S II - which are clearly distinguished on the basis of their characteristic EPR signals observed at low temperature, display similar features in Mop [52] and xanthine oxidases. The so-called Fe/S type I shows characteristic g-values similar within eukaryotic xanthine oxidases (milk XO [46] gl,2,3=2.022, 1.935; 1.899) and Mop [52] (gl,2,3 = 2.021, 1.938; 1.919) typical of spinach ferredoxin [3] (g1,2,3=2.02(1), 1.93(1), 1.90(1)). The center Fe/S II, on the other hand, exhibits broader lines than Fe/S I, with g-values comparable in Mop [52] (2.057, 1.970, 1.900) and milk xanthine oxidase [53] (2.12, 2.01, 1.91), but displaying larger variations within members of the xanthine oxidase family than the Fe/S I center. M6ssbauer spectroscopy also allows one to distinguish between both Fe/S centers in the reduced enzymes: one of the centers (probably Fe/S I) exhibits a
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
75
rather normal quadrupole splitting AEQ of 2.4 mm/s and 2.69(2) mm/s for the ferrous site in xanthine oxidase (at 175 K) [54] and Mop (at 180 K) [55] respectively, while the other center exhibits an unusually large quadrupole splitting of 3.2 mm/s and 3.14(2) mm/s for xanthine oxidase and Mop respectively.
3 Crystal Structure of the Aldehyde Oxido-Reductase from Desulfovibrio gigas (Mop) The crystal structure of the aldehyde oxido-reductase from Desulfovibrio gigas (Mop) represents the first structure of a molybdenum oxotransferase and is a valid model for the interpretation of the weakh of experimental data for xanthine oxidase and related enzymes. It was solved at high resolution in its native desulfo form [6], as well as in its sulfo, oxidized, reduced and alcohol-bound forms [56], allowing a detailed look at the several structural aspects of the molybdenum hydroxylases relevant for catalysis: domain architecture; structure of the cofactors and binding mode within the polypeptide chain; molybdenum center environment; metal coordination and its role in catalysis. In analogy to eukaryotic xanthine oxidases, xanthine dehydrogenases and aldehyde oxidases, Mop is an ot2 homodimer [63] consisting of two 100 kDa subunits (2 × 907 amino acids) [62]. The redox-active cofactors are found in discrete domains within a single polypeptide chain. Mop possesses, per subunit,
o!s) o 4~
9 o
o 'o' 2""
l[
3-
SH
HSx~[
O
NH~/J~NH
(Dihydropterin) Fig.2. The MCD molybdopterin cofactor as established by crystallography in Mop [6]. The open-chain form as suggested by chemical analysis is also shown and an equilibrium between these two forms suggested in solution
76
M a r i a Joao R o m e o • Robert H u b e r
two different kinds of [2Fe-2S] centers [52], as well as a molybdopterin cofactor (molybdopterin cytosine dinucleotide-MCD) (Fig. 2), but lacks the flavin and its domain which is present in most molybdenum hydroxylases. 3.1 Structural and Domain Arrangement of the Protein
From the electron density map interpretation the two [2Fe-2S] centers were recognized in the earlier stages of the analysis, while the molybdopterin cofactor was identified only at a later stage of the structure solution as molybdopterin cytosine dinucleotide (MCD), later confirmed by chemical means [64]. For some of the higher resolution data sets which were analyzed [56], isopropanol molecules as well as magnesium ions, both from the crystallization solution, were identified in density and accordingly refined. The presence of alcohol molecules, particularly the one close to the molybdenum site, was important for modelling the Michaelis complex of an aldehyde substrate molecule (cf. below). The protein molecule is roughly globular with an approximate diameter of 75 .~ (Fig. 3). Its secondary structure shows 28% a-helical and 21%//-sheet conformations, with a total of 33 helical and 27 fl-strand segments longer than 3 amino acid residues. The molecule folds into four distinct domains (Fe/Sa, Fe/Sb, Mol and Mo2) (Fig. 3) of which the first two (Fe/Sa and Fe/Sb) bind the two iron-sulfur clusters, while the larger domains Mol and Mo2 bind the MCD cofactor in extended conformation by a network of hydrogen bonding interactions: Mol contributes with two single molybdopterin binding segments and Mo2 binds the other side of the pterin system and provides all of the dinucleotide binding segments (Fig. 4). These two large domains also surround the molybdenum catalytic site and define, at their interface, a 15 k-deep tunnel, wide open at the surface and constricted in the middle, which leads substrate molecules into the buried molybdenum catalytic site (Fig. 5).
Fig. 3. Stereo plot o f the m o l e c u l a r s t r u c t u r e of Mop w i t h the four i n d e p e n d e n t d o m a i n s rep r e s e n t e d in different colors a n d cofactors s h o w n as colored spheres. Fe/Sb - blue, residues 1 76; Fe/Sa - red, residues 84 156; c o n n e c t i n g peptide - white, residues 158 195; M o l - green, residues 196 581; Mo2 - purple, residues 582 907
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
77 c~ 0
2=
8 =
2-0
0
~
,-~
0
0
=
,'~ ~.~ 0
.
0
=8
"~ ~ ~ 0 ~:~0 0 ~
=
~
E~
0
78
Maria Jo~o Romao • Robert Huber
Fig. 5. Stereo Cc' representation of the Mop molecule with cofactors highlighted as colored spheres. Two isopropanol molecules are also represented as well as the three buried waters close to the molybdenum site. The entrance of the tunnel is marked by one of the isopropanol molecules (IPP1) and by Pro258
The first Fe/Sb (residues 1 to 76) domain shows the CXXGXCXXC motif common to the plant-type ferredoxin class of iron-sulfur proteins [ 19]. Also the chain fold is similar to that of plant (Spirulinaplatensis) [66], alga (Aphanothece sacrum) [80] and cyanobacterium Anabaena [81] [2Fe-2S] ferredoxins. It shows the topology of a five-stranded fl-half barrel with an a-helix running almost perpendicular to the strand direction. Apart from the absence of a 19-residue loop toward the C-terminus of this domain and a / J - t u r n , shorter by six-residues, between the first two strands, the overall fold can be superimposed to Spirulina platensis ferredoxin [66] with a deviation of less than 0.5 A in the C° atom positions of the iron-sulfur cluster-binding turns. The second iron-sulfur domain Fe/S a (residues 84 to 156) reveals a new ferredoxin-type fold: a two-fold symmetric four-helical bundle with two longer central helices flanked by two oblique shorter a-helices. The iron-sulfur cluster lies at the N-termini of the two central helices. Domain Fe/S a is connected to the molybdenum-binding domain Mol via a long, extended segment with irregular secondary structure which juts about 50 A across a rather concave region of the surface of the molecule. This concave region extends from one side of the protein to the other and appears to be the most likely site for the insertion of the flavin domain present in xanthine oxidases. This region of the molecule also appears to be the most probable site of interaction with flavodoxin, which has been shown to be able to accept electrons from Mop/aldehydes in vitro [67]. The two larger Moco binding domains, Mol (residues 196 to 581) and Mo2 (residues 582 to 907) are in close contact to each other, accommodating the molybdenum catalytic site at their interface. Domain Mol is rather elongated (ca 75 ~ long and 28 .~ wide) and organized in two subdomains: a larger N-ter-
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
79
minal part, which consists of a seven-stranded incomplete fl-barrel with one ahelix filling its central cavity, and two additional helices flanking the barrel and exposed to solvent. The smaller C-terminal subdomain consists of a fivestranded mixed parallel-antiparallel fl-sheet flanked by two helices which run approximately parallel to the strand direction. Domain Mo2 is also organized in two subdomains, each with a similar basic fold (a four-stranded fl-sheet, which bends around a pair of helices), and in part dyad related. Both subdomains resemble two large wings spanning over 80 A and with the cofactor MCD lying at the intersection of the two wings. 3.2 The [2Fe-2S] Centers
Both 2Fe-2S clusters appear planar with the iron atoms and cysteine sulfur atoms defining a plane orthogonal to the plane of the [2Fe-2S] group. All iron atoms are tetrahedrally coordinated by the sulfur atoms of the cysteines, with Fe-S bond lengths of 2.2 A and 2.3 A for the iron-sulfide and for the iron-cysteine SVrespectively. The N-terminal plant-type ferredoxin iron cluster (Fe/Sb) has one of the iron atoms linked to C40 and C45, while the other iron atom is coordinated by C48 and C60. The second iron cluster (Fe/Sa) has one of the iron atoms bound to C100 and C139 and the other to C103 and C137.While the Fe/Sb center is close to the protein surface with its Gys S ~ 0 exposed to the solvent, the FeS a center is deeply buried and in contact with the molybdopterin. The assignment of both Fe/S centers to the two spectroscopically distinguishable [47, 52, 53] iron-sulfur centers, type I and type II, is still unclear and under debate [3]. Data from magnetic coupling with molybdenum [52, 53, 78, 79] and electron transfer rate studies [3, 40], suggest that Fe/S I corresponds to Fe/Sa (helix bundle) and Fe/S II to Fe/Sb (plant-type ferredoxin). This assignment is also in agreement with the distribution of reducing equivalents among the redox centers of xanthine oxidase, which is kinetically fast in comparison to the catalytic turnover and follows their relative redox potentials [70]: for XO at pH 7.7 [42] MoVI/MoV-373mV, MoV/MorV-377mV, Fe/S 1-310mV, Fe/S II-255mV, FAD/FADH'-332mV, FADH"/FADH2-234 mV, for Mop at pH 7.6 [67, 69] MoVr/MoV-450mV, MoV/MoIV-530mV, Fe/S 1-280mV, Fe/S II-285mV. However, center type I is the one with g values quite similar to spinach ferredoxin. A possible explanation for this apparent contradiction is that details of coordination protein environment and solvent accessibility, rather than the general polypeptide fold, determine the EPR features of the iron centers [3]. For these reasons, we maintain the designation Fe/S a and Fe/Sb throughout this work. 3.3 The Molybdopterin Cofactor
'
The structure of the molybdopterin cofactor was established for Mop by crystallography as a cytosine dinucleotide (MCD) (Fig. 2). The bicyclic pterin structure and dithiolene side-chain had been proposed by chemical and spectroscopic analysis in general [12-15], but the fused pyran ring resulting from a
80
Maria Joao Romao • Robert Huber
tricyclic structure had remained undetected. It is now well established by crystallography and has been found in three different enzymes: in the aldehyde ferredoxin oxido-reductase (AOR) from Pyrococcusfuriosus [16], in Mop [6] and, more recently, in the DMSO reductase (DMSOR) from Rhodobacter sphaeroides [59] and capsulatus [65]. In both AOR and DMSOR, the metal atom (W and Mo respectively) coordinates two molybdopterins through their dithiolene groups but in distinct ways. The AOR from Pyrococcusfuriosus has two molybdopterins coordinating the tungsten atom and additionally linked by a magnesium ion bound to their phosphate groups [16]. In DMSOR (Fig. 6) we find two molybdopterin guanine dinucleotides (MGD) at the molybdenum atom resulting in a very elongated structure (N 35 ~) [59, 65]. Neither the tungstoprotein nor the DMSO reductases share homology with Mop or the xanthine oxidase family; they are members of three different protein families, of which the DMSO reductase family of oxomolybdenum enzymes is likely to be structurally more diverse than the xanthine oxidase family [3]. Nevertheless, the basic structure of the tricyclic system of the cofactor is similar in all three enzyme families. When isolated, the cofactor is bicyclic, suggesting that the pyran ring is closed by a reversible intramolecular nucleophilic addition of the hydroxyl 9'-OH to the C7 carbon atom of the double bond of a dihydropterin system [77]. The pyran ring closure may occur in situ subsequent to binding of the open-chain cofactor to the enzyme favored by the numerous interactions between the cofactor and the surrounding polypeptide chain as observed in Mop [6] (Fig. 4), DMSORs [59, 65] and AOR [16]. In all structures the pyran is approximately 40° tilted relative to the two ring pterin system (<30 ° in one of the pterins of DMSOR). The fused pyran ring has three chiral centers C6, C7 and C9', whose con-
N
NH " ~
o
7 O(S)
~a-I'y~s,,,,,...~o~O 6 ~ sf "~orh
~
S
........Mo"' '"OyS147
/
NH2
Fig. 6. Molybdenum cofactor structures crystallographically determined for a Mop [6, 56] and for b DMSO reductase from Rhodobacter capsulatus [65]. The dithiolene sulfurs of the second molybdopterin are located at the apical position but are not ligands of the metal, at a distance of 3.5/~ and 3.9 ~k,respectively. The role of this spectator cofactor in catalysis is unclear
Structureand Functionof TheXanthine-OxidaseFamilyof MolybdenumEnzymes
81
figurations are R,R,R in Mop, as well as in the other structures. At variance with suggestions [16, 68] based on functional and structural inorganic models for DMSOR [68], neither the pterin nucleus [68] nor the dinucleotide [16] coordinate directly to the metal. In DMSOR the protein contributes to the metal coordination with the side-chain oxygen of a serine residue, which may be a cysteine or selenocysteine in other members of the DMSOR diverse family. Diversity is also reflected in the observation that even closely related DMSORs from Rhodobacter sphaeroides and R. capsulatus show significant differences in the metal coordination. In R. capsulatus only one dithiolene is covalently bound, and the second is too far for bonding (spectator cofactor) (Fig. 6), while in R. sphaeroides DMSOR, one of the sulfurs from the second cofactor is bound to the metal. In Mop the pterin cofactor is in the dinucleotide MCD form and it is well documented [3] that the variants molybdopterin (MPT) and molybdopterin dinucleotides (MGD, MCD, MAD, MHD) are widely distributed among eukaryotes and prokaryotes. In enzymes from eukaryotes the pterin has always been found in the MPT form. The role of the ribonucleotide portion of the molybdopterin dinucleotides in some bacterial enzymes remains obscure. The MCD present in Mop is extended, with a widest distance of 17 _~,between the cytidine (N4" atom) and the pterin (04 atom). The functional groups of the cytidine are hydrogen-bonded to main chain atoms of domain Mo2 (Fig. 4), and the cytosine dinucleotide has the pyrimidine base in anti conformation, while the D-ribofuranose is twisted with C2"endo and C3"exo. The structure of this cofactor is likely to represent the general form of dinucleotide cofactors of other molybdenum hydroxylases with differences in stabilization of the base. Differences are to be expected in those segments (of domain Mo2) which bind the dinucleotide part, absent in the eukaryotic hydroxylases. This is reflected by the lower degree of homology in this part of the primary sequence, when Mop is compared with eukaryotic xanthine dehydrogenases [19] (Fig. 7). 3.4 Structure and Environment of the Metal Centers
The redox-active cofactors of Mop are inserted in the protein matrix in close proximity to each other, and define the electron transfer pathway (Fig. 8). While the first iron-sulfur center h is rather exposed to solvent via its cysteine 60, the iron-sulfur center a is approximately 15 ~ below the molecular surface and has no direct contact with the solvent. The closest distance between the iron atoms of the two Fe/S centers is ca. 12 ~ while the molybdenum atom lies 15 ~ away from the nearest iron atom of center a (Fig. 4). The molybdenum site is also buried but accessible to the protein surface through a 15 ~ deep tunnel as described above. The pterin system of Mo-co is in direct contact with the nearest iron-sulfur center through a hydrogen bond between the amino group at C2 and the ysulfur atom of Cys139, which coordinates one iron atom. The two iron-sulfur centers are in contact via a chain of seven covalent bonds and a single hydrogen bond between the amide of residue Ala136 and the carbonyl of Cys45. The [2Fe-2S] iron centers are bound by covalent bonds between the irons and
82
Maria Jo~o Romeo • Robert Huber
Structureand Functionof TheXanthine-OxidaseFamilyof MolybdenumEnzymes
83
84
Maria Joao Romeo • Robert Huber
•"
~"~
~ o
~
~
o~
~
~.~
~ 0 ~
m . ~
0 ~ - ~
,
0
~
o'~
~,..~ ~_~
~
~
~
~ ~_~m " . ~ - ~
~
~
0
~
~
~
~
~'~
~-~
,~
'~ ~
-~',o O~
~m
. ~
-~
~ ~0
;~
o~ ~.~
o o
~m.~'~
<,~
~
~-,
~
~
~ ~
~,~'~ ~
~.~ ~
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
85
Fig. 8. Stereomolecular representation of the molybdopterin cytosine dinucleotide of Mop (resulfurated) and both close contacting [2Fe-2S] centers. For the sake of clarity, residues contacting with the dinucleotide part are omitted, and those included show the same color code as Fig. 3. The isopropanol molecules show the direction of the tunnel
cysteine residues and by a number of NH-S hydrogen bonds between their sulfur atoms and the amide groups of the surrounding protein main chain as depicted in Fig. 5. The pterin tricycle is bound by two hydrogen bonds of the carbonyl oxygen atom 04 to amides of the segment Thr420-Phe421 from domain Mol, while the phenyl ring of Phe421 is coplanar with the pterin, favoring a n- n interaction at a short distance of 3.4 .~. The side-chain of Gln807 from domain Mo2, anchors the other edge of the pterin establishing hydrogen bonds to nitrogens N1 and N8 (Fig. 8). The channel which leads from the surface towards the molybdenum site is coated with hydrophobic residues exclusively from domains Mol and Mo2. It exhibits a constriction at half-way by a cluster of hydrophobic residues Phe425, Phe494, Leu497 and Leu626. These side-chains must move away in order to let bulkier aromatic substrate molecules pass and reach the catalytic site. After passing the hydrophobic channel, reducing substrates, aldehydes in the case of Mop, react and are oxidized at the molybdenum center. Mow is reduced to Mow and reducing equivalents are transferred through the partially conjugated system of the pterin and the hydrogen bond, pterin-NH2-Sy-C139, to the Fe/S center _a.Electron transfer proceeds via seven covalent bonds and one hydrogen bond (NH Ala136-O=C C45) towards the exposed Fe/S center b. From here, electrons will flow to an unknown physiological electron acceptor in the case of Mop or, in the case of xanthine oxidase, will be intramolecularly transferred to the flavin center and from there to molecular oxygen, generating superoxide. Intramolecular electron transfer within the xanthine oxidase family of enzymes has been recognized as an integral aspect of catalysis and has been extensively studied using a number of techniques [37-43]. The proposed pathway assigns an important role to the pterin ring for the electron transfer from molybdenum to the Fe/S center. Since the molybdenum atom is not coordinated by any protein side-chain (cf. Sect. 3.5), the pterin cofactor is essential in
86
Maria Joao Romao • Robert Huber
anchoring the molybdenum within the protein. A third important role may be attributed to the pterin cofactor by modulating the redox properties of the molybdenum. Crystallographic data of Mop [56] obtained under reducing conditions show conformational changes in the cofactors: In the reduced crystals the dithiolene sulfurs are wider apart (larger S-Mo-S angle) with a sulfur-sulfur distance of 3.5 ]k instead of 3.0 A as found in the oxidized crystals. Additionally, in the reduced derivatives, the dithiolene molybdenum cycle is less planar with the molybdenum atom deviating ca 0.4-0.7 A from the mean plane of the ring towards the apical ligand. This distortion also suggests that the dithiolene double bond participates in the redox reaction. In the oxidized state of the enzyme, the short distance between the dithiolene sulfur atoms (3.0/~) is closer than their van der Waals contacts (2 x 1.85 ~), which implies a partial disulfide bond formation lost upon reduction of the molybdenum. As a resuk, the dithiolene ligand donates electron density to the metal and consequently the reduction potential of the molybdenum center would be expected to decrease. Additionally, the electron-donating capacity of the ligand to the metal may favor a five-fold coordinated MovI center (see Sect. 3.5). Such a role for the pterin cofactor in modulating the redox state of the metal had been proposed [74] on the basis of model compound studies for a Mo vI cysteamine-derived complex L2MoO2 (L=SC(CH3)2CH2NH(CH3)), which also exhibits an S-S bond of 2.76 A, indicative of a disulfide bond character. These authors suggest that the dithiolene group may participate in the MoW~ M o TM r e dox process by involving a disulfide-thiolate reduction, associated with formal changes of the oxidation state of the metal. More recent studies with other model compounds [73] have proven that tetrahydropterin is able to reduce Mow with disulfur coordination ([MoO2(diethyldithiocarbamate)2]) to Mo TM in [MoO(diethyldithiocarbamate)2]. These results suggested that oxidation/reduction states of the pterin were involved in the Mo oxidation state during substrate turnover. However, more recently [82], the same authors reinterpreted those results on the basis of isolated complexes where molybdenum is found to bind directly to the pterin. Nevertheless, extrapolations from these model systems to molybdenum-containing enzymes were not satisfactory. Although the synthesis of such complexes of molybdenum coordinated to pterin species has deserved much attention [35] and has often been successful [68, 82, 89], the relationship to the reaction mechanism of molybdenum hydroxylases is not clear. There have been proposals in the literature concerning the role of the pterin cofactor in the enzymatic mechanism of the molybdenum oxotransferases. Studies of the state of reduction of MPT in sulfite oxidase [71, 72] and in xanthine oxidase [71] have led to the suggestion of direct involvement of the pterin in catalysis, either via electron transfer or modulation of the oxidation/reduction properties of the metal. The Mop crystal structure analyzed in oxidized and reduced states [56] supports these suggestions. Unfortunately, there are very few mutational studies of xanthine dehydrogenases and fewer mutations have been characterized. The non-conservative mutation G1011E (Glu623 in Mop) in the DrosophilaMelanogasterxanthine dehy2 drogenase leads to an inactive enzyme [75, 76]. This residue is close to the molybdenum and involved in formation of the pocket filled with the internal buried
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
87
Fig. 9. Stereo representation of the molybdenum catalytic site (sulfo form) and surrounding residues within hydrogen bonding distance. IPP1 points in the direction of the channel
water molecules (Fig. 9). The mutation may thus aker the structure considerably, precluding detailed interpretations. Mutants of Aspergillus nidulans pufine hydroxylase are available but require further analysis [93, 94]. 3.5 The Molybdenum Site and Its Environment
The molybdenum coordination sphere was defined on the basis of the 2.25 crystallographic data [6] as square pyramidal geometry, with no protein ligand binding to the metal, although a glutamate (Glu869, conserved within the XO family of enzymes [19](Fig. 7)) is only 3.5 A away from the metal. With higher resolution data to 1.8 A [56], the molybdenum site was defined in greater detail and the metal ligands unambiguously identified. In the approximate square pyramidal arrangement, the dithiolene ring system defines the equatorial plane and contributes with two sulfur ligands. The three remaining positions were assigned as three oxygen ligands identified as two oxo (Mo=O) and one water ligand (Mo-OH2), with the latter occupying the position trans to sulfur $7' of the dithiolene (Fig. 6). The two oxo groups were identified on the basis of shorter distances to the metal (range of 1.6-1.9 A for the different crystals analyzed [56]), as well as lack of hydrogen bonds to surrounding residues. The water ligand shows a longer bond to the metal (range of 2.1-2.5 A) and is within hydrogen-bonding contact to the hydroxyl group of isopropanol IPP3, to amide of Gly697 and to carboxylate of Glu869 (Fig. 9). The apical oxo ligand is replaced by a sulfur atom in Mop crystals resulfurated by incubation with sulfide under turnover conditions [56]. This result allowed one to clearly identify the catalytically essential Mo--S group. The structure (Fig. 6) is in agreement with EXAFS data for xanthine oxidases and for Mop: these data show that an oxo group is present in both oxidized and reduced forms of xanthine oxidases (Fig. 1 a, b) and of Mop (Fig. 1 e, f), while three sulfur ligands are assigned for XO with one longer Mo-S bond for the reduced state of the enzyme (Fig. 1b-d). However, EXAFS data failed to provide evidence for the second oxygen ligand, which has now been identified as a coordinated water molecule.
88
Maria Joao Romeo • Robert Huber
Relevant residues and well-ordered water molecules within the molybdenum catalytic site are represented in Fig. 9 for the sulfo form of the enzyme. In close vicinity to the molybdenum is the conserved Glu869 which is approximately trans to the apical position and may bind directly to the metal by a small rotation of its carboxylate. The apical Mo=S group has the imidazole of His653 as nearest neighbor, within hydrogen-bonding distance (-3.2 .~). The Mo=O ligand is in a quite constrained environment, close to the amide of Arg533 (3.3 A), to the carbonyl oxygen of Phe421 (3.8 -~) and to C~ of Gly422. The metal-bound water is facing the mouth of the extended channel and is the most accessible ligand. The neighboring isopropanol molecule (IPP3), which is in contact with this water ligand, is from the crystallization solution. Alcohols are known to inhibit Mop [67] (as well as xanthine oxidase [47]) and this IPP3 site provides a model for the Michaelis complex of the reaction of Mop with aldehydes (cf. Sect. 5). The isopropanol molecule takes part in a network of hydrogen bonds to a chain of three buried water molecules (137 W, 138 W, 105 W). Two of these (137 W, 138 W) are stabilized by hydrogen bonds to the surrounding polypeptide, but the innermost water, 105 W, is located in a particularly apolar environment (PheS05, Phe763 and Tyr622). Apart from IPP3, other isopropanol molecules were located in the crystal structure of Mop, one of which was identified close to the mouth of the channel (Fig. 5), which may explain the known inhibitory effect of alcohols in Mop (as well as in xanthine oxidase). To summarize, one can distinguish as most relevant features of this catalytic site: (1) Its accessibility via an extended hydrophobic channel, which has a constriction separating a wide-open outer compartment from a very restricted inner binding site at the molybdenum atom; (2) The presence of a water ligand which is the most accessible one, and probably the source of the transferred hydroxyl group and binding site of reaction intermediates; (3) The chain of wellordered buried water molecules, capable of replenishing the vacant coordination site after product dissociation; (4) The conserved residue Glu869, close to the molybdenum, which may act as a proton acceptor of the water molecule and a transient ligand of the metal (cf. Sect. 5). 3.5 The Dimer
As with other molybdenum hydroxylases, Mop is a homodimer. The subunits are related by a crystallographic dyad but are, however, functionally independent. Redox cofactors are relatively separate in both subunits with a shortest distance of 39.3 X between the nearest iron atoms (from center FeSb and its symmetry partner). The molecular dyad runs along the contact of both Fe/Sb domains through Gly32, Leu31, such that residues from FeSb and Mol domains contact symmetry partners from domains Fe Sb* and Mo 1". These symmetrical contacts are mainly of a hydrophobic nature and involve residues Metl, Ile11, Phe16 (from domain FeSb), Leu199, Met203, Pro204, Phe374, Asp379, Tyr376 (from domain Mo 1) which are conserved in the xanthine dehydrogenase family (see below).
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
89
4 Comparison of the Aldehyde Oxido-Reductase from Desulfovibrio gigos with the Xanthine Oxidase Family of Enzymes The amino acid sequence of Mop had first demonstrated its relationship to the xanthine oxidase family of proteins [19, 6]. The multiple alignment with xanthine oxidases from mammalian, insect and fungal sources, showed that the Mop amino acid sequence is highly conserved with ca. 52 % homology and 25 % identity, suggesting a close structural relationship (Fig. 7). Homology extends to different chains of bacterial carbon monoxide dehydrogenase, isoquinoline-1oxidoreductase and nicotine dehydrogenase complexes, which harbor the iron sulfur clusters and the molybdopterin cofactor, respectively. The homology is particularly high in those segments associated with binding of the redox cofactors (but not of the cytosine dinucleotide!) as well as within residues of the substrate-binding pocket, and for residues of the tunnel. As shown in Fig. 7, the binding segments of the cofactors are well conserved with a high proportion of invariant residues and secondary structures, fl-strands and helices are contiguous. Interruptions by deletions and insertions occur only in loop regions and at the N- and C-termini. The long contiguous deletion of about 400 residues in the Mop structure, between residues 176 and 177, corresponds to the FAD domain in the xanthine dehydrogenases and is absent in Mop. This additional domain must be placed, in the XO family of enzymes, somewhere along the extended connecting segment (white part in Fig. 3). The FAD binding domain has been tentatively assigned on the basis of mutant studies. A mutation which has received considerable attention is the Y395F Xanthine dehydrogenase from Drosophila melanogaster [85] which was shown to be enzymatically inactive. This residue had been previously shown to be important in the NAD÷ binding by the enzyme, by performing chemical modification of this tyrosine in chicken liver XDH [86] and in rabbit liver aldehyde oxidase [87]. In chicken liver XDH that residue has been identified as Tyr419 [84]. In all cases, these chemical modifications identify the NAD binding site within the insertion found in the Mop sequence [19, 84]. There have also been proposals for the location of the molybdenum domain within the primary sequence of XDH on the basis of chemically modified lysine residues with fluorodinitrobenzene [83, 84, 88] (identified as Lys742 and Lys759 in the rat XDH [88b], modifications which affect the rate of the enzymatic reaction). This indirect evidence is also now validated with the Mop structure: In the primary sequence alignment shown on Fig. 7, these two lysine residues are located on weakly conserved regions of the domain Mol and align with residues Gly378 and Leu394, respectively. Interestingly, Gly378 is close to the surface and takes part in the crystallographic dimeric contact, close to Met2 of the other subunit. Leu394 is ca 10 A from the molybdenum atom, and is part of the residues from the tunnel. This may explain the observed interference in the enzymatic reaction of XDH with chemically modified lysines [88], probably due to blocking of the tunnel. The close relationship between Mop and the xanthine oxidase family of proteins implies a common mechanism of action of these enzymes. Essential struc-
90
Maria Jo~o Romao • Robert Huber
tural aspects of Mop do indeed agree with a multitude of experimental data for the XO enzymes, as discussed in the next session.
5 Structure-Based Catalytic Mechanism Molybdenum hydroxylases and in particular xanthine oxidases have been investigated by a variety of spectroscopic techniques to gain insight into the mechanism of this class of enzymes (cf. Sect. 2). Several aspects have been recently reviewed [3, 7, 61]. In parallel, much effort has been dedicated to the search for synthetic models of Moco and to the analysis of their structures as well as their spectroscopic and functional properties. These model systems and their contribution to the understanding of enzymatic mechanisms have been reviewed by Holm [2, 34] and Enemark and Young [35]. In previous sections we have described relevant features of the Mop structure with important mechanistic implications: (1) The intramolecular electron transfer among the redox centers, mediated by the pterin cofactor, to which severn functions are now clearly associated; (2) The domain organization of the protein and cofactor environment which defines a buried catalytic site with unique features; (3) The clear definition of the metal coordination sphere, consistent with EXAFS data, as well as the unambiguous location of the catalytically important sulfido molybdenum ligand. The water ligand is revealed for the first time and proposed as the catalytically labile oxygen. The possibility of the water ligand being the source of the labile oxygen is corroborated by recent 170 -EPR studies [90] of model compounds, on which basis a metal-bound hydroxide was proposed as the labile site; (4) The isopropanol site, present in the second coordination sphere of molybdenum is used as a valid model for the Michaelis complex of an aldehyde substrate (Fig. 10, upper), whereas the carbonyl oxygen replaces the hydroxyl group and the aldehyde hydrogen occupies the methyl group closer to the metal. In this orientation, the carbonyl oxygen establishes two hydrogen bonds with water 137 W and the water ligand, thereby polarizing it. The nucleophilicity of the water ligand is promoted by Glu869. The validity of using the isopropanol site as a model for the substrate-binding site is indeed supported by several arguments from literature data: Alcohols are substrate analog inhibitors of Mop [67] as well as of xanthine oxidase [46], while ethylene glycol is a slow substrate of xanthine oxidase [91]. A number of aldehydes are found to inactivate bovine milk xanthine oxidase [92], while reactivation occurs immediately after removing excess aldehyde. EPR data from xanthine oxidase inhibited with methanol and formaldehyde produced identical "inhibited" signals [95], which suggests similar binding modes for alcohols and aldehydes. The binding of substrate molecules in the second coordination sphere of molybdenum in the Michaelis complex is also supported by spectroscopic data: In the reaction of xanthine oxidase with 2-hydroxy-6-methylpurine (a slowly reacting substrate), no proton coupling due to the C8-H proton is detected in the "rapid type 1" EPR signal [96], although coupling is detected after substrate oxidation and hydride transfer [50, 96]. Also the strong inhibitor 8bromoxanthine [97] has been shown to interact with xanthine oxidase in a way
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
91
reductive half-cycle Fig. 10. The reductive half-cycleof the hydroxylation reaction of Mop and xanthine oxidase.
Lower: Stereo representation of the molybdenum cofactor and bound isopropanol, buried waters and Glu869. Upper: Hypothetical structures off a, the Michaeliscomplexwith aldehyde
substrate close to MoVI;b, the enzyme carboxylic acid product complexwith MoW;and c, intermediate after-product dissociation, in which Glu869 is bound to the metal typical of purine substrates and the Mo-Br distance was found to be larger than 4 A [8]. The structure of the Michaelis complex as represented in Fig. 10 shows a simple pathway for the enzymatic reaction: the carbonyl carbon atom suffers nucleophilic attack by the activated water ligand (transferred as OH-), concertedly with hydride transfer to the sulfido group (a), whereby the carboxylic acid product is generated but remains bound to the molybdenum via the transferred oxygen atom (b). Bond formation requires a further approach of the atoms involved, possibly via anchoring of the carbonyl oxygen to water 137 W, and the aldehyde hydrogen must come closer to the sulfido group. The direct transfer of a hydrogen from substrate (a) to xanthine oxidase has been detected as a proton from the enzyme strongly coupled to Mo v in the rapid EPR signal [10, 48- 50, 99]. The close proximity of the carbonyl carbon atom to the metal as in (b) was proposed in the "inhibited" species of xanthine oxidase by ENDOR studies [98], while EXAFS has indicated product analogs bound to Mo TM [9, 57]. In the last step (c), the carboxylic acid product is released, which is probably
92
Maria Joao Romeo • Robert Huber
facilitated by a transient binding of Glu869 to the metal to maintain a 5-fold coordination. The Mo water site may then be regenerated from the chain of internal buried water molecules. This reaction scheme agrees with proposals for the reaction of xanthine oxidase [7, 102] but gives details on the exact stereochemical environment. However, it differs by suggesting the water ligand, instead of the oxo group (probably present just as the so-called "spectator oxygen", known from molybdenum coordination chemistry [100]), as the labile oxygen to be transferred to the substrate as supported by recent 170-EPR studies [90], 170- and 13C-ENDOR experiments and kinetic studies [98] on xanthine oxidase.
6 Conclusions Xanthine oxidase is a central enzyme in purine catabolism. It converts xanthine, the product of all major purine nucleotide or dinucleotide catabolic pathways to uric acid, which may be excreted by some organisms, or, in others, further degraded via intermediates to urea. Xanthine oxidase harbors four redox-active centers, a molybdenum molybdopterin, two iron-sulfur centers and a flavin, in a large protein representing an electron transport chain within a single polypeptide. All cofactors have characteristic spectroscopic features which allow one to monitor their redox states using physical techniques. For a long time, xanthine oxidase has attracted the attention of not only physiologists, biochemists and spectroscopists but also that of pharmacologists because elevated levels of uric acid in body fluids and its deposition in joints lead to gout, a disease manifested by painful joint inflammation. Gout is treated by the administration of a potent inhibitor of XO, allopurinol, which reduces the uric acid levels. The xanthine oxidase family of molybdenum enzymes is exemplified by the aldehyde oxido-reductase from D. gigas (Mop), which shares common structural features: homodimers with the molybdenum bound to a single pterin cofactor, two [2Fe-2S] centers with similar spectroscopic features in Mop and in XOs, and large homology of other sequences. Recent 3D structurers of members of other classes of Mo oxo-transferases (DMSO reductases [59, 65] and formate dehydrogenase (FDHase) from E. coli [109, 110], show no homology to the XO group of enzymes: in both cases, the Mo atom is coordinated by two molybdopterin guanine dinucleotide cofactors and, in the case of the FDHase, to an essential selenocysteine residue. While DMSOR possesses no additional prosthetic cofactors, this FDHase contains a [4Fe-4S] cluster. These enzymes are unrelated to the XO group and the oxido-reductase from D. gigas remains the prototype of the xanthine oxidase family of molybdenum enzymes. Its crystal structure represents a milestone towards our understanding at the molecular level of the fascinating XO family of enzymes.
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
93
7 References 1. Rees DC, Kim J, Georgiadis MM, Komiya H, Chirino AJ, Woo D, Schlessman J, Chan MK, Joshua-Tor L, Santillan G, Chakrabarti, Hsu BT (1993) ACS Symposia 535:170 2. Holm, RH (1987) Chem Rev 87:1401 3. Hille R (1996) Chem Rev (in press) 4. Murray, KN, Watson JG, Chaykin S (1966) J Biol Chem 241 : 4798 5. Wooton JC, Nicolson RE, Cock JM, Walters DE, Burke JF, Doyle WA, Bray RC (1991) Biochim Biophys Acta 1057:1575 6. Romfio MJ, Archer M, Moura I, Moura JJG, LeGall J, Engh R, Schneider M, Hof P, Huber R (1995) Science 167:1167 7. Hille R (1994) Biochim Biophys Acta 1994:143 8 Cramer SP, Hille R (1985) J Am Chem Soc 107:8164 9. Hille R, George GN, Eidsness MK, Cramer SP (1989) Inorg Chem 28:4018 10. Bray RC (1988) Quaterly Rev Biophysics 21:299 11. Turner N, Barata BAS,Bray RC, Deistung J, LeGall J, Moura JJG (1987) Biochem J 243: 755 12. Johnson JL, Hainline BE, Rajagopalan KV (1980) J Biol Chem 255:1783 13. Johnson JL, Rajagopalan KV (1982) Proc Nail Acad Sci USA 79:6856 14. Johnson JL, Hainline BE, Rajagopalan KV, Arison BH (1984) J Biol Chem 259:5414 15. a) Cramer SP, Johnson JL, Ribeiro AA, Millington DS, Rajagopalan KV (1987) J Biol Chem 262:16357; b) Johnson JL, Wuebbens MM, Rajagopalan KV (1989) J Biol Chem 264: 13440; c) Rajagopalan KV, Johnson JL (1992) J Biol Chem 267:10199 16. Chan MK, Mukund S, Kletzin A, Adams MWW, Rees DC (1995) Science 267:1463 17. Johnson JL, Bastian NR, Rajagopalan KV (1990) Proc Natl Acad Aci USA 87:3190 18. B6rner G, Karrasch K, Thauer RK (1991) FEBS Lett 290:31 19. Thoenes U, Flores OL, Neves A, Devreese B, Beeumen JJV, Huber R, Romeo MJ, LeGall J, Moura JJG, Rodrigues-Pousada C (1994) Eur J Biochem 220:901 20. Meyer O (1982) J Biol Chem 257:1333 21. Krfiger B, Meyer O (1986) Eur J Biochem 157:121 22. Schfibel U, Kraut M, M6rsdorf G, Meyer O (1995) J Bacteriol 177:2197 23. Tchisuaka B, Kappl R, Htittermann J, Lingens F(1993) Biochem 32:12928 24. Peschke B, Lingens F (1990) Biol Chem Hoppe-Seyler 372:1081 25. Schach S, Tchisuaka B, Fetzner S, Lingens F (1995) Eur J Biochem 232:536 26. Lehmann R, Tchisuaka B, Fetzner S, R6ger P, Lingens F (1994) J Biol Chem 269:11254 27. Bauer G, Lingens F (1992) Biol Chem Hoppe-Seyler 373:699 28. Fetzner S, Lingens F (1993) Biol Chem Hoppe-Seyler 374:363 29. Sauter M, Tchisuaka B, Fetzner S, Lingens F (1993) Biol Chem Hoppe-Seyler 374:1037 30. Dilworth GL (1982) Arch Biochem Biophys 219:30 31. Nagel M, Andreesen JR (1990) Arch Microbiol 154:605 32. Freudenberg W, K6nig K, Andreesen JR (1988) FEMS Microbiol Lett 52:13 33. Grether-Beck S, Iglio GL, Pust S, Schilz E, Decker K, Brandsch R (1994) Mol Microbio113: 929 34. Holm RH (1990) Coord Chem Rev 100:183 35. Enemark JH, Young CG (1993) Adv Inorg Chem 40:1 36. Bhattacharyya A, Tollin G, Davis M, Edmonson DE (1983) Biochem 22:5270 37. Walker MC, Hazzard JT, Tollin G, Edmonson DE (1991) Biochem 30:5912 38. Anderson RF, Hille R, MasseyV (1986) J Biol Chem 261:15870 39. Hille R, Anderson RF (1991) J Biol Chem 266:5608 40. Porras AG, Palmer G (1982) J Biol Chem 257:11617 41. Barber, MJ, Siegel LM (1982) Biochem 21 : 1638 42. Hille R, MasseyV (1986) J Biol Chem 261 : 1241 43. George GN, Bray RC (1983) Biochem 22:1013 44. Hille R, Stewart RC, Fee JA, Massey (1983) J Biol Chem 258:4849 45. Barber, MJ, Siegel LM (1983) Biochem 22:618
94 46 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77.
78. 79. 80. 81. 82. 83. 84. 85.
Maria Jofio Romeo • Robert Huber Barber MJ, Coughlan MP, Rajagopalan KV, Siegel LM (1982) Biochemistry 21:3561 Lowe DJ, Barber MJ, Pawlik RT, Bray RC (1976) Biochem J 155:81 Hawkes TR, George GN, Bray RC (1984) Biochem J 218:961 Gutteridge S, Tanner SJ, Bray RC (1978) Biochem J 175:869 Gutteridge S, Tanner SJ, Bray RC (1978) Biochem J 175:887 Gibson JF, Bray RC (1968) Biochim Biophys Acta 153:721 Bray RC, Turner NA, LeGall J, Barata BAS, Moura JJG (1991) Biochem J 280:817 Lowe DJ, Lynden-Bell RM, Bray RC (1972) Biochem J 130:239 Morpeth FF, George GN, Bray RC (1984) Biochem J 220:235 Barata BAS, Liang J, Moura I, LeGall J, Moura JJG, Hanh Huynh B (1992) Eur J Biochem 204, 773 Huber R, Duarte R, Moura I, Moura JJG, LeGall J, Liu M, Hille R, Romao MJ (1996) Proc Natl Acad Sci USA, 93:8846 Turner NA, Bray RC, Diakun GP (1989) Biochem J 260:563 Cramer SP, Moura JJG, Xavier AV,LeGall J (1984) J Inorg Biochem 20:275 Schindelin H, Kisker C, Hilton J, Rajagopalan KV, Rees DC (1996) Science 272:1615 Stiefel EI (1993) ACS Symposia 535:1 Hille R (1993) ACS Symposia 535:22 Romeo MJ, Barata BAS, Lobeck K, Moura I, Carrondo MA, LeGall J, Lottspeich F, Huber R, Moura JJG (1993) Eur J Biochem 215:729 Moura JJG, XavierAV, BruschiM, LeGall J, Hall DO, CammackR (1976) Biochem Biophys Res Commun 72: 782 Caldeira J, Moura I, Romeo MJ, Huber R, LeGall J, Moura JJG (1995) J Inorg Biochem 59:739 Schneider F, L6we J, Huber R, Schindelin H, Kisker C, Kn~iblein J (1996) J Mol Biol (in press) (a) Fukuyama K, Hase T, Matsttmoto S, Tsukihara T, Katsube Y, Tanaka N, Kakudo M, Wada K, Matsubara H (1980) Nature 286:522; (b) Tsukihara T, Fukuyama K, Nakamura M, Katsube Y, Tanaka N, Kakudo M, Wada K, Hase T, Matsubara H ( 1981) J Biochem 90:1763 Barata BAS, LeGall J, Moura JJG (1993) Biochemistry 32:11559 Fischer B, Schmalle H, Dubler E, Sch~iferA, Viscontini M (1995) Inorg Chem 34:5726 Moura JJG, Xavier AV, Cammack R, Hall DO, Bruschi M, LeGall J (1978) Biochem J 173: 419 Olson IS, Ballou DP, Palmer G, MasseyV (1974) J Biol Chem 249:4363 Gardlik S, Rajagopalan KV (1990) J Biol Chem 265:13047 Gardlik S, Rajagopalan KV (1991) J Biol Chem 266:4889 Burgmayer SJN, Baruch A, Kerr K, Yoon K (1989) J Am Chem Soc 111 : 4982 Stiefel EI, Miller KF, Bruce AE, Corbin JL, Berg JM, Hodgson KO (1980) J Am Chem Soc 102:3624 Doyle W, Chovnick A, Whittle JRS, Bray RC (1996) Biochem Soc Trans 24:14S Bray RC, Bennett B, Burke JF, Chovnick A, Doyle WA, Howes BD, Lowe DJ, Richards RL, Turner NA, Ventom A, Whittle JRS (1996) Biochem Soc Trans 24:99 Hilton JC, Rajagopalan KV (1996) Archives Biochem Biophys 325:139 Lowe DJ, Bray RC (1978) Biochem J 169:471 Barber MJ, Salerno JC, Siegel LM (1982) Biochemistry 21 : 1648 Tsukihara T, Fukuyama K, Mizushima M, Harioka T, Kusunoki M, Katsube Y, Hase T, Matsubara H (1990) J Mol Biol 216:399 Rypniewski R, Breiter DR, Benning MM, Wesenberg G, Oh BH, Markley JL, Rayment I, Holden HM (1991) Biochemistry 30: 4126 Burgmayer SJN, Arkin MR, Bostick L, Dempster S, Everett KM, Layton HL, Paul KE, Rogge C, Rheingold AL (1995) J Am Chem Soc 117:5812 Hughes RK, Doyle WA, Chovnick A, Whittle JRS, Burke JF, Bray RC (1992) Biochem J 285: 507 Sato A, Nishino T, Noda K, Amaya Y, Nishino T (1995) J Biol Chem 270:2818 Doyle WA, Burke JF, Chovnick A, Dutton FL, Russell C, Whittle JRS, Bray RC (1996) Biochem Soc Trans 24:31S
Structureand Functionof The Xanthine-OxidaseFamilyof MolybdenumEnzymes
95
86. Nishino T, Nishino T (1989) J Biol Chem 264:5468 87. Turner NA, Doyle WA, Ventom AM, Bray RC (1995) Eur J Biochem 232:646 88. a) Nishino T, Tsushima K, Hille R, Massey V (1982) J Biol Chem 257:7348; b) Nishino T, Amaya Y, Nishino T (1990) 10th International Symposium on Flavins and Flavoproteins, Como, Italy 89. Burgmayer SJN, Everett K, Bostick L (1993) ACS Symposia 535:114 90. Greenwood R], Wilson GL, Pilbrow J R, Wedd A G (1993) ] Am Chem Soc 115:5385 91. Tanner SJ, Bray RC (1978) Biochem Soc Trans 6:1331 92. Morpeth FF, Bray RC (1984) Biochemistry 23:1332 93. Scazzocchio C, Sealy-Lewis HM (1978) Eur J Biochem 91:99 94. Glatigny A, Scazzocchio C (1995) J. Biol Chem 270:3534 95. Pick FM, McGartoll MA, Bray RC (1971) Eur J Biochem 18:65 96. Hille R, Kim JH, Hemann C (1993) Biochemistry 32:3973 97. Hille R, Stewart RC (1984) J Biol Chem 259:1570 98. a) Howes BD, Bray RC, Richards RL, Turner NA, Bennett B, Lowe DJ (1996) Biochemistry 35:1432; b) Howes BD, Bennett B, Bray RC, Richards RL, Lowe DJ (1994) J Am Chem Soc 116:11624 99. Bray RC, George GN (1985) Biochem Soc Trans 13:560 100. Rapp6 AK, Goddard WA III (1982) J Am Chem Soc 104:3287 101. Collison D, Garner CD, Joule JA (1996) Chem Soc Rev : 25 102. Kim JH, Ryan MG, Knaut H, Hille R (1996) J Biol Chem 271:6771 103. Terao M, Cazzaniga G, Ghezz P, Bianchi M, Falciani F, Perani P, Garattini E (1992) Biochem J 283:863 104. Amaya Y, Yamazaki K, Sato M, Noda K, Nishino T, Nishino T (1990) J Biol Chem 265: 14170 105. Houde M, Tiveron MC, Bregegere F (1988) EMBL data library 106. Lehmann R, Tchisuaka B, Fetzner S, Lingens F(1995) J Biol Chem 270:14420 107. Lee CS, Curtis D, McCarron M, Love C, Gray H, Bander W, Chovnick A (1987) Genetics 116:55 108. Yasokochi Y, Kanda T, Tamura T, unpublished
Nickel-Iron Hydrogenases: Structural and Functional Properties M. Frey Laboratoire de Cristallographie et de Cristallogen~se des Prot~ines, Institut de Biologie Structurale Jean-Pierre Ebel CEA-CNRS, 41 Avenue des Martyrs, 38027 Grenoble CDX 1, France, E-mail:
[email protected]
Hydrogenases are proteins which catalyze the reversible two-electron oxidation of the most simple of chemical compounds, molecular hydrogen, following the reaction: H2 ~ 2H++2e -. These enzymes are the central feature of hydrogen metabolism which is essential to many microorganisms of great biotechnological interest, such as methanogenic, acetogenic, nitrogen-fixing, photosynthetic and sulfate-reducing bacteria. For the last twenty years hydrogenases have enjoyed renewed interest, mostly in view of their capacity to produce molecular hydrogen, a clean source of energy. The determination of the first three-dimensional atomic structure of a hydrogenase by single crystal X-ray diffraction, together with infrared (IR) and electron paramagnetic resonance (EPR) spectroscopy have revealed that the active site consists of a bimetallic center containing for [NiFe] hydrogenases one nickel and one iron atom and three non-protein diatomic ligands (one CO and two CN-) to the iron. This evidence with stoichiometric redox-titrations has led to new insights into the catalytic mechanism. Moreover, genetic studies have cast light on the biosynthesis of the protein, including the incorporation of metal atoms found at the active site. A brief overall view of the structural and functional properties of hydrogenase is presented. The crystal structure is described with emphasis on the determination of the atomic content and arrangement of the active site from combined X-ray and spectroscopic data. Some aspects of the catalytic mechanism are discussed in the light of these new results.
Keywords:Hydrogenases, hydrogen metabolism, nickel enzyme, biological electron and proton transfer, iron-sulfur proteins
Abbreviations ....................................
98
1
Introduction .................................
99
2
[Ni-Fe] H y d r o g e n a s e s
2.1 2.2 2.2.1 2.2.2
Structural Properties ............................ Functional Properties ............................ Isotopic Exchanges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Redox States o f Nickel as seen b y EPR a n d R e d o x - T i t r a t i o n s . . . . . . .
102 103 103 103
3
X-Ray C r y s t a l l o g r a p h y . . . . . . . . . . . . . . . . . . . . . . . . . . .
105
3.1 3.2 3.3 3.4
Introduction ................................. Purification a n d Crystallization o f the E n z y m e . . . . . . . . . . . . X-Ray Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The X-Ray A n o m a l o u s Dispersion: a Tool for I d e n t i f y i n g Metal Ions . . . . . . . . . . . . . . . . . . . . .
105 106 106
...........................
102
107
Structure and Bonding, Vol. 90 © Springer Verlag Berlin Heidelberg 1998
98
M. Frey
3.5 3.6 3.6.1 3.6.2 3.6.3
Solving the Crystal Structure: an Outline . . . . . . . . . . . . . . . . Modelling the Active Site . . . . . . . . . . . . . . . . . . . . . . . . . . Identification of the Metal Ions at the Active Site . . . . . . . . . . . The Three Iron Non-Protein Diatomic Ligands . . . . . . . . . . . . . A Small Ni-Fe Bridging Non-Protein Ligand . . . . . . . . . . . . . .
107 107 108 110 110
4
Architecture of the Molecule
110
4.1 4.2 4.3 4.3.1 4.3.2 4.3.3
Reliability of the X-Ray Model . . . . . . . . . . . . . . . . . . . . . . The Iron-Sulfur Clusters . . . . . . . . . . . . . . . . . . . . . . . . . . The Active Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Assignment of the Three Diatomic Iron Ligands . . . . . . . . . . . . Assignment of the Non-Protein Ni-Fe Bridging Ligand . . . . . . . . Ni, Fe and Ligands Coordinations . . . . . . . . . . . . . . . . . . . .
111 113 113 113 114 114
5
Functional Significance of the Structure
115
5.1 5.1.1 5.1.2 5.1.3 5.2 5.2.1 5.2.2 5.2.3 5.2.4
Access to the Active Site . . . . . . . . . . . . . . . . . . . . . . . . . . Electron Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proton Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular H y d r o g e n Transfer . . . . . . . . . . . . . . . . . . . . . . . The Active Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Form C .................................... SI State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F o r m AB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spectroscopy of the Iron . . . . . . . . . . . . . . . . . . . . . . . . . .
115 115 115 115 116 116 118 118 119
6
The D. gigas H2ase Model is a Prototype of [NiFe] Hydrogenases
119
7
Biosynthesis of the Enzyme . . . . . . . . . . . . . . . . . . . . . . . .
121
8
Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . .
122
9
References
123
.......................
................
..................................
Abbreviations
Chromatium vinosum photosynthetic bacterium C. vinosum Desulfovibrio gigas sulfate-reducing bacterium D. gigas T. roseopersicina Thiocapsa roseopersicina photosynthetic bacterium Ni-EPR
Ni-(A, B, C) Ni-A EPR signal Ni-B EPR signal Ni-C EPR signal Form A, B, C
Electron paramagnetic resonance of nickel Ni-center states yielding an EPR signal. (Values correspond to D. gigas hydrogenase) g = 2.31, 2.23, 2.02 g = 2.33, 2.16, 2.02 g = 2.19, 2.16, 2.01 Forms of the e n z y m e containing Ni-A, Ni-B and Ni-C, respectively.
Nickel-Iron Hydrogenases:Structural and Functional Properties
State (SI, SU) State (R) XAS
[FeS]
99
Ni-EPR silent and infrared (IR)-identified states more reduced than Forms AB and less than Form C Ni-EPR silent and IR-identified states more reduced than Form C X-ray absorption spectroscopy Iron sulfur cluster
1 Introduction Hydrogenases (hydrogen:acceptor oxido-reductases) constitute a class of proteins, discovered by Stephenson and Stickland in 1931 [1]. These enzymes have the unique ability to catalyze the reversible two-electron oxidation of the most simple of chemical compounds, molecular hydrogen, following the reaction: H 2 ¢=>2H++2e. Hydrogenases are the central feature of hydrogen metabolism [2, 3] which is essential to many microorganisms of great biotechnological interest such as methanogenic, acetogenic, nitrogen-fixing, photosynthetic and sulfate-reducing bacteria [4, 5, 6]. Hydrogenases either oxidize H2 to provide the organism with a source of strong reductants or generate hydrogen as sinks of excess electrons. Thus, for Table 1. Physico-chemical and catalytic properties of the t). gigas H2ase
Class Organism
[NiFe] Sulfate-reducing bacterium
Desulfovibrio gigas Cellular localization Molecular weight (x 10 3) pI Chemical sequence (from nudeotide sequence) Nb. of iron atoms Nb. acid-labile sulfide Nb. [4Fe-4S] 2÷/1+clusters Nb. [3Fe-4S] 1+/° dusters Nb. Nickel atom Diatomic non-protein ligands H2 production 3 ([NiFe], [Fe] 4) Haase H 2 consumption3 ([NiFe], [Fe] 4) H2as e Inhibitors
Periplasm 89.5 59.459 + 6; 28321 _+41 6.0 61480 (large subunit) 2 28436 (small subunit) 12 12 2 1 1 2 CN-; 1CO 440 (10400)
[s51 [211 [i0] [291 [471
450 (50000)
[4, 29, 861
CO NO
[87 and in ref. therein] [881
[10,21, 44] [10,211 [10,22] [10,22] [10,19, 20, 44] [12,44,55] [4,861
1 Mass spectrometric analysis of the mature large and small subunit polypeptide chain. 2 Including the 15 residues C-terminal peptide which is cleaved upon maturation of the enzyme. 3 The specific activity is in micromoles of hydrogen produced (or consumed) per minute per mg of protein. 4 D. vulgaris (Hildenborough)
100
M. Frey
example, the strictly anaerobic sulfate-reducing bacteria can grow on H 2 as an electron donor with sulfate and thiosulfate as terminal electron acceptors. Conversely, these same bacteria produce hydrogen upon fermentative growth on pyruvate in the absence of sulfate [3, 4]. For the last twenty years hydrogenases have been the subject of a wealth of physiological, biochemical, physicochemical and genetic studies, mostly with a view to a potential use of H2 as a clean fuel and alternative energy source [for reviews, 4 and 5]. Hydrogenases, so far discovered, are, but with one exception [7], metalloenzymes which contain iron and also in many cases nickel [8, 9]. They have been 2088 / 2076
. .[0.01 O.D.
1950 ' 1937 /~ / 1921
2060 .~ <
)2044
I
B
I
1
Io.oiO.D.
I
1898
!
2059
g .< "~
\ 2076 2088
2100
~] 1950 I
I
I
I
2050
2OO0
1950
1900
1850
Wavenumber (c/n "t ) Fig. 1. Infrared spectra (IR) of Chromatium vinosum bydrogenase reduced and activated by 30 rain incubation with I bar H z at 50 °C, cooled on ice and then equilibrated with I °/6 Hz. Absolute IR absorbances of the enzyme (A) in the dark at 23 K; (B) after illumination with white light for 10 rain at 23 K; (C) after photolysis (trace B),warming at 200 K for 15 rain and cooling in the dark at 23 K. (D) IR difference spectrum (B-A). Spectral resolution was I cm -I. The reference spectrum for the absolute IR spectra shown in A, B and C was an empty infrared cell at room temperature. The absolute infrared difference spectra were baseline-corrected using the same parabolically shaped curves for traces A, B and C. The difference spectrum shown in trace D was calculated using the raw data for traces A and B. From [12] with permission
Nickel-Iron Hydrogenases:Structural and Functional Properties
101
g=2.32
FormA
A
.....
/I g=2.23
/I
FormB
g=2.01
g=2.33
SI
=
~ g=2.01
g=2.19
FormC
!, g=2.14
........ I
2800 3000 32'00 34'00 36'00 Magnetic Field (Gauss)
1.2o¢.- 1.0G.) 0
0.8-
0
~- 0.6-O
.N -~ 0 . 4 o
Z
b
S
--Form A ---- FormB ........SI .... FormC ........R
0.20.0I
t300
I
8320
I
I
I
8340 8360 8380 X-rayEnergy(eV)
I
8400
Fig. 2. X-ray absorption spectroscopic studies of the [NiFe] hydrogenase from Thiocapsa (T.) roseopersicina hydrogenase, a Nickel EPR (Ni-EPR) spectra of the [NiFe] hydrogenase from T. roseopersicina hydrogenase samples prior to exposure to synchrotron radiation. The spectra were recorded at 77 K. Conditions of measurements: Microwave power, 20 mW, frequency, 9.62 GHz, Modulation amplitude, 4G. b Ni-K edge XANES spectra at five Ni-EPR identified redox states of T. roseopersicina hydrogenase. From [11] with permission
102
M. Frey
classified accordingly in two major classes: Iron-only ([Fe]) hydrogenases [for review, 5] and Nickel-Iron ([NiFe]) hydrogenases [for review, e.g. 8 and 9]. Catalytically, [NiFe] hydrogenases are generally thought to be both less active and more specific for molecular hydrogen than the [Fe] enzymes and more resistant to oxidation (e.g. Table 1). The [NiFe] enzymes have been the most extensively studied at the molecular level through a multidisciplinary approach, as recently exemplified in an exhaustive and clear review concerning their physiological, structural and functional properties [9]. More recently, the determination of the crystal structure of the [NiFe] hydrogenase from the sulfate-reducing bacterium Desulfovibrio gigas (the first hydrogenase structure and the first structure of a nickel containing enzyme to have been determined) [10] as well as physico-chemical [e.g. 11-13] (Figs. 1 and 2) and genetics studies [e.g. 14] have given a completely new insight into the molecular and catalytic properties of the molecule and its biosynthesis. The aim of the present paper is to briefly summarize the overall structural and functional properties of [NiFe] hydrogenases and present and discuss the recent results in the light of earlier models.
2 [Ni-Fe] Hydrogenases In the [NiFe] family, the hydrogenases from the photosynthetic bacteria Chromatium vinosum or Thiocapsa roseopersicina and from the sulfate-reducing bacteria belonging to the Desulfovibrio genus have been subject to many multidisciplinary studies which are referred to below. To avoid repetitions they are abbreviated as: D. gigas (or vulgaris etc.) HRase, C. vinosum H2ase and Z roseopersicina Hgase.
2.1 Structural properties [NiFe] hydrogenases are mostly heterodimers containing a large subunit and a small subunit with molecular weights (m.w.) of about 60,000 and 30,000 respectively [8, 9]. A large number of amino-acid sequences of [NiFe] hydrogenases has been determined from the structural genes encoding the two subunits [for a review, e.g. 15]. Analyses of their homologies revealed that the sequences of the large subunits are fairly conserved whereas those of the small subunits are comparatively more variable [e.g. 9 and references therein]. Some [NiFe] hydrogenases contain also selenium in the form of selenocysteine [14, 16]. The presence of one nickel atom, which characterizes this class of hydrogenases, has been detected by physico-chemical analysis, EPR and isotope substitution (61Ni) and never been questioned since [ 17 - 20]. By contrast, the number of iron atoms present in the enzyme was less certain. For example, the D. gigas H2ase was first thought to contain 12 (+ 1) Fe atoms distributed among three iron-sulfur [4Fe-4S] clusters [21] and later on 11, after EPR and M6ssbauer spectroscopy had shown that one of the three iron-sulfur ([FeS]) cluster is a [3Fe-4S] cluster [22- 25].
Nickel-IronHydrogenases:Structuraland FunctionalProperties
103
2.2 Functional Properties 2.2.1
Isotopic exchanges
Besides catalyzing the reversible two-electron oxidation of molecular hydrogen, hydrogenases are able to catalyse proton-deuterium or proton-tritium exchanges in the absence of electron donors or acceptors following the reaction H 2 + D20 ¢=~HD + HDO and convert ortho-hydrogen to para-hydrogen. Moreover, several lines of evidence have shown that for hydrogenases, including D. gigas H2ase, the HD/H2 ratio in the exchange reaction is pH-dependent [16]. From isotope exchange experiments it has been inferred that molecular hydrogen is heterolytically cleaved to a proton and a hydride ion with the subsequent formation of an enzyme-hydride intermediate [26, 27]. In this context, a base near the catalytic site might help to bind the proton during the catalytic process. Furthermore, since hydrogenases split molecular hydrogen heterolytically, their active site is most likely to be involved in the transfer of two electrons at a time although they reduce one-electron substrates [16, 28]. The isotope exchange reaction does not involve an electron or donor. Therefore it is often used to probe the functional state of the active site [29, and references therein]. 2.2.2
Redox states of Nickel as seen by EPR and Redox-Titrotions
Electron Paramagnetic Resonance (EPR) spectroscopy (at 77 K) combined with isotope substitution (61Ni) and X-ray absorption spectroscopy (XAS) have shown that hydrogen activation by hydrogenases takes place in the vicinity of the nickel atom [e.g. 19, 22]. Several EPR signals corresponding to the nickel atom (Ni-EPR) have been identified according to states of the enzyme which are related to changes in its apparent activity [20, 29, 30-33] (Figs. 2a and 3). For example, samples of D. gigas H2ase, as purified under aerobic conditions, are oxidized and catalytically inactive. They are characterized by two distinct Ni-EPR signals which are the signatures of two different paramagnetic states of a nickel atom (called Ni-A and Ni-B respectively). The relative intensities of the two Ni-A and Ni-B EPR signals may vary depending on the preparation of the protein [for review, 29]. Hydrogenase samples yielding mostly the Ni-A EPR signal can be activated slowly by strong reductants or after incubation under hydrogen for hours, whereas those yielding mostly the Ni-B EPR signal can be activated by reductants within few seconds or minutes [32]. Accordingly, the two redox states Ni-A and Ni-B have been termed"unready" (Niu) and"ready" (Ni~) respectively [9]. In the present paper, we term Form A and Form B the protein samples yielding the Ni-A and Ni-B EPR signals respectively. A typical aerobically purified sample of D. gigas H2ase contains about 40% Form A, 10% Form B and a 50% fraction which does not yield any Ni- EPR signal. Upon hydrogen activation or reductive titration of these oxidized inactive samples of hydrogenase, the Ni-A and Ni-B EPR signals disappear. The resulting
104
M. Frey g-factor 2.4
I
2.2
I
2
I
I
I
i
-•••••"•
Ni-A
Ni-B
Ni-C
~
Ni-C split
Ni-L2
I
I
270
290
I
1
I
310
330
350
Magnetic field, mT
Fig. 3. Representative EPR spectra of the nickel atom from [NiFe] hydrogenases; Ni-A, (mea-
sured at a temperature of 92 K); Ni-B (92 K); Ni-C (60 K); Ni-C split by interaction with the reduced iron-sulfur clusters (4.2 K); Ni-L2, induced by illumination of Ni-C (30 K). Ni-A, Ni-B are from the Desulfovibrio (D.)fructosovorans hydrogenase; Ni-C, Ni-C split and Ni-L2 (previously known as Ni-C*) from D. gigas hydrogenase. For clarity, spectral substraction has been used to remove small amounts of the Ni-A and Ni-B signal respectively.Conditions of measurement: Microwave power, 20 roW, frequency, 9.2 GHz, Modulation amplitude, 1 mT. Adapted by Richard Cammack, from [29] with permission
protein samples become Ni-EPR silent. Consequently, this state of the enzyme has been termed SI (for silent intermediate). Further reduction results in the appearance of a third EPR signal which is the signature of a new paramagnetic state of the nickel (called Ni-C) [32, 33]. Form C, yielding the Ni-C EPR signal corresponds to a redox state of the protein which is involved in the catalytic cycle [e. g. 33, 34]. Finally, full reduction of the samples leads again to an Ni-EPR silent state termed R (for reduced) [19, 35].
Nickel-Iron Hydrogenases:Structural and Functional Properties
105
Form C, which exists in all [NiFe] hydrogenases and corresponds to a partially reduced form of the enzyme where Ni-C could be bound to a hydrogen species, has been proposed to be either H2, H + or H- [e.g. 36, 37]. Interestingly, the compounds of Ni-C with these hydrogen species or the competitive inhibitor CO are dissociated upon illumination below 77 K [30, 34]. The rate of this photochemical conversion is slower in D20 than in H20 which supports the role of Ni-C in catalysis. Moreover, since the redox potentials of Ni-C are pH-dependent it was inferred that the nickel ion is involved in hydrogen binding [34 and ref. therein]. The various forms and states of hydrogenase as probed by EPR spectrocopy of the nickel site upon reductive activation of the as-purified enzyme, can also be observed by subsequent controlled oxidation. Analysis of stoichiometric oxidative titrations of D. gigas H2ase have demonstrated that Form B, state SI, Form C and state R are separated by sequential one-electron reductions (or oxidations) of the nickel atom in the following order: Form B ~ SI Form C 6-~ R. A cyclic catalytic mechanism involving Ni(SI) present in the SI state (as the active H 2 oxidant), Ni-C, and Ni(R) present in the R state (as the active H ÷ reductant) has been proposed [35]. Together with the active site, D. gigas Hzase has two [4Fe-4S] clusters with mid-point redox potentials (- -290 mV and -340 mV at pH 7, [13, 25, 34]), similar to those of the Ni(SI)/Ni-C and Ni-C/Ni(R) couples, respectively. Consequently, a full description of any overall state of this hydrogenase is a combination of the states of three redox-active centers. This results in a complex combination of twelve possible microstates, forming three cycles, for which relative populations have been determined at various potentials [57]. It is also of interest to note that the mid-point potentials of the nickel and the [4Fe-4S] clusters are pH-dependent (-60 mV/pH unit, [34]). By contrast, the mid-point redox potential of the [3Fe-4S] cluster is far more positive ( - - 30 mV to - 80 mV; [20, 22, 13]) which has led to the questioning of its involvement in the electron transfer process associated with the catalytic activity of the enzyme. This point will be discussed below. The most characteristic structural and functional properties of the [NiFe] D. gigas H2ase are summarized in Table 1.
3 X-Ray Crystallography 3.1 Introduction Single crystal X-ray diffraction is, so far, the only experimental method available for obtaining reliable three-dimensional atomic models of biological macromolecules with a molecular weight above 20,000. In short, a productive X-ray crystallographic study results in a three-dimensional map of the electron density of the molecule. This map should be as detailed and precise as possible to allow the building (or fitting like hand in glove) of a reliable three-dimensional atomic model of the protein using all the available biochemical and structural data (cf. Table 1). An interpretable map typi-
106
M. Frey
cally requires diffraction data at resolutions of 3.2 ~ or better and accurate phases. This in turn is highly dependent on the size and quality of the crystals [e. g. 38-40]. 3.2 Purification and Crystallization of the Enzyme
Sulfate-reducing bacteria produce large amounts (typically 40 mg for 260 g of cells, wet weight) of [NiFe] hydrogenase which is relatively stable in air [29]). This could explain why X-ray-suitable crystals have only been reported for the [NiFe] hydrogenases from two sulfate-reducing bacteria: D. vulgaris Miyazaki and D. gigas [41, 42, 10]. Both D. vulgaris and D. gigas H2ases are purified, under aerobic conditions, by chromatographic methods, including preparative high-performance liquid chromatography (HPLC) as a final step of the process [29, 41]. To limit further oxidation damage the purified enzyme is kept in liquid nitrogen. Single crystals are obtained from these preparations by microdialysis (D. vulgaris Haase) or sitting and hanging drop vapor diffusion methods (D. vulgaris and D. gigas H2ases). Remarkably large (> 1 mm) and stable crystals of D. vulgaris H2ase are grown by introducing microcrystals in the final crystallization droplets [41, 43]. They are kept at 4 °C in an X-ray capillary tube filled with nitrogen or argon. By contrast, the crystals ofD. gigas Haase are smaller (ca. 0.2 mm) and relatively unstable. They are stabilized with glycerol and stored either under strict anaerobic conditions (5 ppm oxygen) or in liquid nitrogen after flash-cooling [44]. EPR analysis has shown that D. gigas crystals obtained with aerobically purified samples contain, as expected, a mixture of the two oxidized paramagnetic states Ni-A (-40%) and Ni-B (-10%) and an Ni-EPR-silent unknown fraction (N50%) ([10], P. Bertrand and B. Guigliarelli, personal communication). We will discuss the consequences of these observations when we analyse the crystal structure. Moreover, the enzyme obtained by dissolving the i). vulgaris and D. gigas H2ase crystals can be activated after prolonged incubation under hydrogen in the presence of methyl viologen ([42], E.C. Hatchikian, Y. Higuchi, personal communications). To determine the phases of the diffracted data, isomorphous crystals of various heavy-atom derivatives are prepared in the usual way by soaking the native protein crystals (which comprise about 42 % by volume of solvent) in solutions containing heavy-atom reagents at typically several mM concentrations [10, 43]. 3.3 X-Ray Diffraction
In the early stages, the intensities of X-rays diffracted by crystals of the native D. vulgaris and D. gigas H2ase and isomorphous crystals of heavy-atom derivatives were measured, at 10 °C and room temperature respectively, with conventional rotating-anode X-ray sources and area detectors [10, 41]. Presently, diffracted intensities of D. gigas [44] and D. vulgaris Haases (Y. Higuchi, personal
Nickel-IronHydrogenases:Structuraland Functional'Properties
107
communication) crystals are routinely measured with synchrotron radiation and image-plates or CCD detectors. To limit the damage induced by the powerful photon flux of synchrotrons, crystals are flash-cooled and kept at 100 K during the experiments. This has allowed the acquisition of accurate X-ray diffracted intensities at 1.8 A resolution for the native D. vulgaris H2ase [Y. Higuchi, personal communication] and at 2.54 .& for the native D. gigas Hzase[44]. 3.4 The X-ray Anomalous Dispersion: A Tool for Identifying Metal Ions
A small fraction of the X-ray radiation which is absorbed by an atom is remitted by inner shell electrons. This ,,anomalous" scattering differs in phases and amplitudes from the normal scattering; it depends on both the nature of the atom and the X-ray wavelength. At wavelengths which are characteristic of each element, the anomalous intensity° changes abruptly (absorption edge; e.g. at A = 1.743 A for iron and A= 1.487 A for nickel; Fig. 4). Due to the high brilliance and tunability of synchrotron radiation it is now possible to obtain accurate anomalous intensities at chosen wavelengths [45]. This has been crucial for locating and identifying the redox centers of D. vulgaris H2ase by mukiwavelength anomalous diffraction (MAD) [43] and for determining unambiguously the nature of the metal ions present in the active site of D. gigas H2ase[44]. 3.5 Solving the Crystal Structure: An Outline
The initial phases of the diffraction data of D. vulgaris H2ase and D. gigas H2ase crystals have been derived by combining multiple isomorphous (MIR) replacement (with, in each case, six isomorphous heaW atom derivatives) and anomalous diffraction methods. The resulting electron density maps, at 4 (D. vulgaris H2ase, [43]) and 5 A (D. gigas Haase, [46]) resolution, revealed the globular shape of the molecules and four strong features which were tentatively assigned to the two [4Fe-4S] and one [3Fe-4S] clusters, respectively, and the nickel site. To obtain the phases at higher resolution, the molecular envelope was taken into account and, in the case of D. gigas Haase, the electron densities of the two independent molecules present in the crystal cell were averaged. This led to electron density maps at 3.0 A (D. vulgaris H2ase) and 3.3 A (D. gigas H2ase) resolution to which the two [4Fe-4S] and the [3Fe-4S] clusters and most of the chemical sequence [47] were fitted manually by interactive computer graphics [Y Higuchi personal communication, 10]. 3.6 Modelling the Active Site
A strong feature of the electron density map ofD. gigas H2ase was found close to the center of the molecule and partially sandwiched by cysteines 65, 68, 530 and 533 of the large subunit. These latter residues are found in all hydrogenases and thought to be involved in nickel binding [9 and ref. therein]. Moreover, amino-
108
M. Frey
Fig. 4. X-ray anomalous dispersion of nickel (red) and iron metals (dark blue),f' indicates the real and f" the imaginary component of the respective anomalous dispersion factors. Both f' and f" components have been used to calculate the phases at 4 ~ resolution for the D. vulgaris H2ase [43] by the multiple anomalous dispersion,method (MAD, [45]) with five wavelengths (1.040 ~, 1.487 _A,1.730 A, 1.743 ~., 1.750 ~). f have been used to locate and identify the nickel and iron atoms (including the iron at the active site) of D. gigas H2asewith four wavelengths (0.905 .~, 1.5418 ~, 1.733 h (A1),1.750 .~ (.,12)) [44] Courtesy of Elsa Garcin acid sequence alignment has shown that Cys530 in D. gigas H2ase is homologous to a selenocysteine in the [NiFe] D. baculatus H2ase which was demonstrated to be a ligand of the nickel, by EXAFS studies [48] and EPR measurements on the 77Se-enriched enzyme [49]. All this evidence pointed to the assignment of this region of the electron density of D. gigas H2ase to the nickel atom. However, it appeared that the bulk of this active-site-assigned electron density could in fact accommodate one additional metal ion (distant by about 2.9 ~) and three non suppress di-atomic protein ligands [10] (Fig. 5) in addition to the nickel atom Therefore, in view of these unexpected results, great care has been taken to identify the additional metal ion and its ligands. 3.6.1 Identification of the Metal Ions at the Active Site
Atomic absorption spectrometry, EPR spectroscopy and inductively coupled plasma (ICP) analysis show that the D. gigas H2ase contains one nickel and 12
Nickel-Iron Hydrogenases:Structural and Functional Properties
109
Fig.5. Stereo view of the active site of the D. gigas hydrogenase fitted to an electron density map at 2.54/~ resolution. The two major peaks corresponding to the nickel (blue) and iron (red) sites are contoured at 7or.L1, L2 and L3 identify the three diatornic non-protein ligands respectively(violet sticks). The question mark (?) is at the site of the putative monoatomic oxo species bridging the two metals. Adapted from [44]
(+ 1) iron atoms [21, 10]. Since the determination of the crystal structure had unambiguously confirmed that 11 iron atoms were distributed among the three [FeS] clusters of the molecule it has been concluded that the "remaining" twelfth iron atom could be the second mysterious metal ion modelled in the electron density of the active site in the vicinity of the nickel. To verify this hypothesis, two anomalous difference density maps were calculated using single crystal X-ray diffraction data collected on both sides of the Fe absorption edge (A1 =1.733 A and ./12=1.750 A; Fig. 4). The ".,11"map shows significant density (with heights about 11 times the background level, a) at the sites of the 11 Fe atoms of the [3Fe-4S] and the two [4Fe-4S] clusters, respectively, and at the site of the unidentified metal ion close to the nickel. The "Ae" map shows only noise at all metal positions. After scaling the two maps and subtracting one from the other, the resulting "A1-Ae" difference map shows 12 peaks (11 a) corresponding, again, to the 11 [FeS] cluster iron atoms and the second metal at the active site. Since no other peaks are found besides the ironsulfur clusters and the active site, the presence of 12 Fe ions per hydrogenase molecule is confirmed. This experiment has established, beyond any reasonable doubt, that the second active site metal ion is iron [44]. Moreover, an anomalous map was calculated with the imaginary components (f") of the anomalous X-ray data obtained at 2.54 A resolution with A = 0.905 A. This map also shows 11 peaks (10.2 a) corresponding to the 11 iron ions of the clusters and two peaks (13.5 a, 10.5 a) at the putative location of the nickel and iron atom at the active site, respectively. Since the relative heights of Ni-to-Fe assigned peaks closely match the f" (Ni) to f" (Fe) ratio at A = 0.905 ~, this confirms the presence in the D. gigas H2ase of a nickel atom liganded to the four cysteines 65, 68,530 and 533 of the large subunit (Table 2).
M. Frey
110
Table 2. Selected interatomic distances and angles in the active site of D. gigas H2ase. Adapted from [44] Atoms
d (A)
Atoms
angles (o)
Ni-Sy 65 Ni-Sy68 Ni-Sy 530 Ni-Sy 533 Ni-O(?) Ni-Fe Fe-S~r 68 Fe-Sy 533 Fe-O(?) Fe-CL1 Fe-CL2 Fe-CL3
2.2 2.6 2.3 2.6 1.7 2.9 2.2 2.2 2.1 1.9 1.7 1.9
Ni-Sy 68-Fe Ni-Sy 533-Fe Sy 68-Fe-Cm Sy533-Fe-C L2 Ni-O(?)-Fe C51-Fe-CL2 CL1-Fe-CL3 CL2-Fe-CL3
74 74 168 171 97 94 83 91
The Fe ligands L1 L2, putatively hydrogen-bonded to the protein (Fig. 7 a), could correspond to the two CN- and L3 to the single CO, identified in the homologous [NiFe] C. vinosum H2ase [55]. The average errors on all the atomic coordinates are 0.2 A [10, 44].
3.6.2
The ThreeIron Non-Protein Diatomk Ligands
Three small and nearly spherical features close to the, then putative, iron atom were first detected in the 2.85 A resolution electron density map. They were modelled as water molecules, although it was clear that their nature remained to be determined [10]. At 2.54 A resolution, these features appear protuding from the bulk of the active site electron density as three ellipsoids which can be modelled only as diatomic non-protein molecules (named L1, L2 and L3; Fig. 5). Indeed, neither smaller or larger ligands could be fitted correctly to the electron density [44]. No other structural non-protein features have been found close to the active site. However, at that stage, the chemical nature of the active site iron diatomic ligands remained uncertain. This point will be discussed below. 3.6.3
A small Ni-Fe Bridging Non-Protein Ligand
The 2.54 A resolution maps also revealed the presence of a small feature of the electron density bridging the two metal Ni and Fe sites (Fig. 5). Due to the protein environment, this site has only room to be modelled as a monoatomic species [44].
4 Architecture of the Molecule The D. Gigas H2ase is a roughly globular heterodimer (radius 28 + 5 A) with two subunits interacting extensively (Figs. 6a, b). The large subunit contains the
Nickel-IronHydrogenases:Structuraland FunctionalProperties
111
Fig. 6a, b. Schematic drawings of the three-dimensional structure of the oxidized [NiFe] hy-
drogenase from the sulfate-reducing bacterium Desulfovibrio gigas as obtained by single crystal X-ray diffraction, b is a side view (from the right) of a. The secondary structure elements: helices (ribbon) and fl-sheets (arrows) are highlighted. The small and large subunits are colored in blue and pink respectively but for the 15 residues of the C-terminal part of the large subunit (colored in grey). These residues are probably involved in the incorporation of the nickel during the biosynthesis of the enzyme [ 10,14, 83]. The nickel, iron and sulfur atoms of the four metal centers are coloured in blue, red and yellow respectively. The N and C terminus ends of the protein chain of the small(s) and the large subunits (1) are indicated but, for clarity, the C-terminus end of the large subunit which is deeply buried in the large subunit in the vicinity of the active site. The heterobimetallic active site inside the large subunit includes one nickel (blue) and one iron atom (red) and three ligands (small grey and green sticks). It is deeply buried and very close to the center of gravity of the molecule. The three [FeS] clusters are located in the small subunit. They are nearly aligned and about 12/~ apart; the [3Fe-4S] cluster being between the two [4Fe-4S] clusters. The[4Fe-4S]p cluster proximal to the active site (about 14 ,~ apart) is bound to the N-terminal domain (Ns) of the small subunit (in dark blue). This domain closely resembles the electron transfer flavodoxins but for the substitution of the flavin cofactor by a [4Fe-4S] cluster. The most distal cluster to the active site, [4Fe-4S]d, is close to the molecular surface. Adapted from [10, 44]
hydrogen-activating site (Figs. 7) which is deeply buried inside the protein whereas the small subunit holds the three [FeS] clusters. 4.1 Reliability of the X-Ray Model The current atomic three-dimensional m o d e l of the [NiFe] D. gigas, H2ase includes residues 7 to 536 o f the large subunit, residues 4 to 264 residues of the small subunit, the 3 [FeS] clusters, the nickel and iron atom and the four nonprotein ligands at the active site. This model was refined against native X-ray data at 2.54 ,~ in the usual way by interleaving cycles of simulated annealing, least-squares energy minimization and manual refitting in periodically updated electron density maps. On the whole, the correlation between the observed electron density and the density calculated from the final atomic m o d e l is very g o o d (e. g. Fig. 9 a). Moreover, the stereochemical parameters of the a m i n o acid chains of the large and small subunits and iron-sulfur clusters fit within ex-
112
M. Frey
Fig.
7. a The [NiFe] hydrogenase Desulfovibrio gigas: A stereoview of the active site and its immediate environment in the oxidized inactive form. The putative hydrogen bonds between the non-protein diatomic ligands L1 and L2 and the protein are represented by magenta dashed lines. The putative oxo or hydroxo metal bridging ligand is represented by a small light-blue sphere. (red*) indicates the sixth vacant coordination site of the nickel which is about 4 A away from the extremity to a network of hydrophobic channels connecting the active site and the external medium, b A dose-up view of the the bimetallic [NiFe] center coordination. Atoms are colour-coded: green for carbon, yellow for sulfur, red for iron and blue for nickel. The three ligands could be cyanides (L1 and L2) and carbon monoxide (L3) [44, 55]; in the oxidized inactive form, the metal bridging grey coloured atom (?) is most likely an oxo or hydroxo ligand. Adapted from [44]
pected error with those f o u n d in the literature [50, 51]. No electron density has been seen for several residues of the N-terminus end of the large (6) and the small subunit (4). As these small stretches of the protein are exposed to the solvent they are probably disordered within the crystals. As noted above, the oxidized D. gigas Haase crystals used to determine the atomic structure contain a mixture of Ni-EPR detectable Form A ( - 4 0 % ) and F o r m B (N10%) and an Ni-EPR-silent u n k n o w n fraction ( - 5 0 % ) . However, structural differences beween these different redox-states of the active site cannot be distinguished by crystallographic analysis (at least at 2.54 ~ resolution).
Nickel-IronHydrogenases:Structuraland FunctionalProperties
113
The only detectable differences concern the nickel ion, the St" atom of Cys530 and the Ni-Fe bridging ligand as their temperature factors are significantly higher than those of the neighbouring atoms at the active site. Since X-ray data have been collected at 100 K, these relatively high values indicate some static disorder and hence subtle structural differences between the different oxidized states of the protein present in the crystal [44]. Interestingly, Ni-EPR studies of the homologous D. vulgaris H2ase single crystals suggest that the observed changes of g-values and tensor axes orientations between Ni-A and Ni-B might reflect modifications in a cysteine sulfur ligand of the nickel [52]. 4.2 The Iron-Sulfur Clusters
The three [FeS] clusters are disposed along an almost straight line, at about 12 from each other, with the [3Fe-4S] cluster located half-way between the two [4Fe-4S] clusters (Figs. 6 a, 6b, 8). One of the two [4Fe-4S] clusters is about 13 from the active site (proximal (p) cluster), whereas the other [4Fe-4S] cluster (distal (d) cluster) is close to the surface of the molecule with the imidazole ring of one histidine ligand partially exposed to solvent. It should be pointed out, again, that the stereochemical parameters of the three [FeS] clusters, as derived from the 2.54 A resolution X-ray study, do not significantly differ from those observed for model compounds [51] or iron-sulfur proteins [53 and references therein]. 4.3 The Active Site 4.3.1
Assignment of the Three Diotomk Iron Ligonds
The active site of the oxidized protein contains a binuclear metallic center with one nickel and one iron atom, one small Ni-Fe bridging non-protein ligand and three Fe non-protein diatomic ligands. The presence of such non-protein ligands in a protein was quite unexpected. Therefore, much thought and experimental effort has been made to verify whether the ligands are an intrisic part of the protein rather than an artefact and determine their chemical nature. Infrared studies on the inhibition of the [NiFe] C. vinosum hydrogenase by carbon monoxide had detected, besides a band at 2060 cm -1 assigned to a metal-CO complex, three additional and unexplained high-frequency bands at 2082, 2069 and 1929 cm -1 [54]. Further IR studies on the C. vinosum (and more recently on D. gigas, [44]) Haase detected these three bands for all the stable redox states of the enzyme and showed that their frequencies shift in a concerted fashion as a function of the redox state of the nickel [12]. Therefore, the authors concluded [12, Note added in proof] that the three unusual IR bands observed for C. vinosum Haase are the signatures of intrinsic non-protein groups which could be the three non-protein ligands of the, then putative, iron atom [10] subsequently identified by X-ray anomalous dispersion [44]. They
114
M. Frey
also suggested that the high frequencies of the bands could be assigned to groups containing either triply bonded species or adjacent double bonds. Given these lines of evidence, it has been proposed that the three non-protein diatomic ligands of the crystallographic model at 2.54 ~ resolution could be either carbon monoxide (CO) or cyanide [44]. Moreover, the higher intensity and lower frequency of one of the IR bands relative to the other two suggested that a mixture of these different ligands is bound to the active site iron. This has indeed been demonstrated by a recent IR analysis of the I3C- and iSN-enriched [NiFe] hydrogenase from Chromatium vinosum which has shown that the three absorption bands are the signatures of two cyanide groups and one carbon monoxide molecule in accordance with the release of 0.5 mol carbon monoxide and 1.6 mol cyanide per mol enzyme during careful denaturation of the enzyme
[55]. 4.3.2
Assignment of the Non-Protein Ni-Fe Bridging Ligand When the reduced active [NiFe] C. vinosum H2ase is exposed to 1702, broadened EPR spectra are observed for the resulting Ni-A and Ni-B EPR signals indicating that an exogenous oxygen species is bound close to the nickel in these states [56]. By analogy, the electron density of the monoatomic species bridging the nickel and iron atoms in the D. gigas. H2ase active site has been modelled as fully occupied by an p-oxo species derived from the reduction of 02. 4.3.3
Ni, Fe and Ligands Coordination The nickel atom has three close and two distant ligands in a highly distorted square pyramidal conformation, with a vacant axial sixth ligand site, whereas the iron atom has six ligands in a distorted octahedral conformation (Figs. 7a and 7b, Table 2). It is rather striking that in total only four protein atoms are involved in the liganding of the metal ions. The protein environment of each of the three Fe diatomic ligands is different: L1 could accept two hydrogen bonds from the peptide NH and the OH group of Ser486 and L2 two hydrogen bonds from the peptide NH and guanidinium group of Arg463, respectively; L3 makes contacts only with hydrophobic residues [44] (Fig. 7a). The active site is quite remarkably buried in the center of the protein. This suggests that the present structure of the oxidized forms of the enzyme, is probably not prone to undergo major structural changes upon hydrogen activation.
Nickel-IronHydrogenases:Structuraland FunctionalProperties
115
S Functional significance of the Structure 5.1 Access to the Active Site 5.1.1 Electron Transfer
Redox-titrations, EPR and M6ssbauer spectroscopy, show that the three [FeS] clusters are involved in the redox process [e. g. 22, 24]. Moreover, stoichiometric oxidative and reductive titrations of D. gigas Hzase have established that the mid-point potentials of the two [4Fe-4S] clusters are compatible with those of the active redox states of the nickel and that the [4Fe-4S] cluster interacting with the nickel has the more negative potential [13, 35, 57]. These lines of evidence led to the proposal of an electron-transfer route where the clusters serve as relays between the active site and the surface (Figs. 6a, 6b, 8) whether through-space [58] or through-bond orbitals with occasional through-space jumps [59] mentionedabove.The halfway [3Fe-4Fe] cluster has a mid-point potential about 300 mV more positive [20, 22, 13] than those of the two [4Fe-4S] dusters [13, 25, 34]. Therefore, it has been suggested that the [3Fe-4S] cluster, not present in many [Ni-Fe] hydrogenases, could be used in the D. gigas H2ase to control the rate or the direction of the electron flow, or both [60]. Residues surrounding the partially exposed histidine ligand of the distal [4Fe-4S] cluster could be involved in contacts with the redox partner of hydrogenase which could be a soluble [e. g. 61] or membrane-bound [62] polyhemic cytochrome c3. 5.1.2 Proton Transfer
It is widely assumed that protons move in proteins through very short displacements (smaller than 1~) by rotational movements of donor and acceptor groups [e.g. 63, 64]. In D. gigas Hzase such proton path(s) between the active site and the molecular surface have not yet been identified with certainty. The proton path which has been proposed based on the 2.85 A resolution X-ray model [10] has to be questioned since, at 2.54 ~ resolution, it is not possible to fit residue 482 of the large subunit [44], predicted to be a histidine by the nucleotide sequence [47], to the more detailed electron density map. Alternative possible pathways including histidine, glutamate (which have suitable pK values for transferring protons) and internal water molecules are currently being discussed. 5.1.3 Molecular Hydrogen Transfer
Early proposals suggested that several pathways exist for molecular hydrogen that depend on fluctuations in the protein conformation, as in the case of CO access to the haem group of myoglobin [10, for review, e.g. 65]. Analysis of the
116
M. Frey
X-ray model suggests that molecular hydrogen exchanges could actually be facilitated by hydrophobic channels linking the active site and the surface of the molecule. This hypothesis is currently being probed. 5.2 The Active Site
Most of the early interpretations concerning the activity of hydrogenase assumed that the nickel atom was the mononuclear active site of the enzyme and that the [FeS] clusters were responsible for transfering electrons between the active site and the redox partners of the protein [for recent reviews, 9 and 66]. In this context several mechanisms of hydrogen activation have been proposed. In particular, in view of the appearance and disappearance of different Ni EPR signals upon activation of the enzyme, it has been suggested that the formal Ni oxidation state changes during the catalytic cycle (from Ni° or Ni+lto Ni+3) [13, 57, 67, 68]. This proposal was strongly questioned when it was observed that the X-ray absorption edge of the Ni ion in T. roseopersicina H2ase does not change significantly when the enzyme is poised in the different redox states including the Ni-EPR silent states SI and R and the Ni-EPR detectable photolyzed product Ni-C* [11] (Figs. 2a and 2b). This observation has been confirmed by recent XAS analyses of several [NiFe] hydrogenases (including D. gigas Haase) poised in the different redox states which suggest only a slight increase in the electron density on the nickel upon reduction from Form A and Form B to the SI state and no significant increase upon further reduction to Form C and R state [69]. As seen above, X-ray crystallography combined with infrared spectroscopy yields a picture of the active site which is rather complex as it contains a binuclear metal center, with one nickel and one iron atom, and three diatomic non-protein ligands to the iron. In view of these latter results we will highlight below some of the interpretations (or reinterpretations) of the catalytic mechanisms which have been proposed recently [44, 70, 71]. 5.2.1 Form C
As mentioned above, EPR studies of 61Ni-enriched D. gigas H2ase show that the Ni-AB and Ni-C signals arise from paramagnetic states of the nickel in the active site which, consequently, may correspond to either NP + or Ni 1÷ [e.g. 19]. Ni-C seems to be the only paramagnetic state which is part of the catalytic activity. Moreover, the presence of Ni-C is associated with a photolabile hydrogenic species (H ÷, H- or H2) and can reversibly bind the competitive inhibitor CO [36, 37]. Studies of model compounds have essentially shown that: (a) Ni3+species can neither bind hydride nor CO whereas Ni 1+ adducts can; (b) bridging thiolates should help to stabilize low oxidation states of nickel and (c) hydride adducts arise from a base-assisted heterolytic cleavage of H2 [66, 72 and references therein].
Nickel-Iron Hydrogenases:Structural and Functional Properties
117
The X-ray structure indicates that the Ni ion is liganded to four cysteine sulfur atoms. As in model compounds, this should help in stabilizing a Ni +1 state and consequently allow the formation of the H- complex with Ni-C. EPR and ENDOR studies on D. gigas H2ase rule out the assignment of a hydride bound axially to Ni-C and suggest the presence of an in-plane hydride [73]. Therefore, the hydride ligand could bind at the metal bridging site which is occupied by the putative oxo species in the crystal structure of the oxidized forms. Its negative charge would be stabilized by the two active site metals and possibly the guanidinium group of Arg 463 of the large subunit. Moreover, such MHM bonds are quite common in compounds containing two or more transition metal atoms [e. g. 66], including Fe and Ni [74]. Hydrogenase takes up the two-electron reductant dihydrogen to reduce oneelectron substrates (this reaction can be viewed as a,,redox-switch" [75 and for review, 76]). During catalysis the two electrons are likely to be transferred, one at a time, to the proximal [4Fe-4S] cluster and then to the putative hydrogenase redox partner cytochrome c3 through the electron transfer route described above (Fig. 8). Therefore the "second electron" must, most probably, remain at
Fig. 8. A close-up view showing a possible electron transfer route between the [NiFe] active site, buried in the large subunit, and His185 exposed at the surface of the small subunit. Atoms are colour-coded:green for carbon,yellow for sulfur, red for iron and blue for nickel.The bimetallic [NiFe] center is represented. The (#) mark indicates residues of the large subunit. A through-bond pathway [58] could include hydrogen bonds (green lines), covalentbonds of Cys65#, Gly66#, Cys148, Cys249, Cys246 and His185 and ligands interactions between clusters of iron and cysteine sulfur atoms. Alternativelythe"organic glass" [59] model would involve the redox centers, including the cysteine Sy ligands. The nearest edge-to-edge distances between these centers, are represented by black lines and their values indicated in ~. Adapted from [10]
118
M. Frey
the active site until the proximal cluster is reoxidized. Moreover, the concerted displacements of the three high-frequency IR bands according to the redox states of the nickel (Fig. 1, [12, 44]) indicate that changes in charge density are occurring in the immediate environment of the three atomic ligands to the iron. These changes might sense the "second electron" in the vicinity of the iron which could then have a redox role in the catalytic process [44, 70, 71]. Illumination of C. vinosum H2ase induces large shifts of the three special IR bands to lower frequencies (Fig. 1, [12]) which might be, again, the signature of changes in charge density in the vicinity of the Fe center. This could also mean that the putative hydride bridging Ni and Fe changes into a terminal Fe-bound hydride. Stereochemical analysis of the active site supports such a possibility. In conclusion, all this evidence points to a mechanism whereby the two-electron reduction of the active site would be associated with a base-assisted heterolytic cleavage of H 2 by the Ni ion with concomitant H- binding at the metal bridging site in the Ni-C state without any change in the formal charges of both active site Ni and Fe ions. Subsequent one-electron oxidations would involve primarily the Fe ion [70, 71, 44]. These hypotheses are currently being discussed. Thus, a recent EPR study of the light-sensitive Ni-C suggested that the hydrogen species associated with Ni-C is not in the first coordination sphere of the nickel atom but bound to a sulfur atom of its cysteine ligands Gys 65 and Cys 530 (Fig. 7b) [77]. 5.2.2 $1 State
Analysis of oxidative titrations of D. gigas H2ase assumed that SI is the only state capable of spontaneously oxidizing H 2 t o protons [57]. A formal Ni+1 assignment to SI may be considered because, when H 2 is passed through solutions of reduced Ni+1, model compound hydride adducts are formed in much higher yields [66]. However, since SI is an Ni-EPR silent state, such Ni +1 assignment requires a strong magnetic coupling with the nearby Fe in the active site. 5.2.3 Form AB
Modelling of stoichiometric reductive titrations of D. gigas H2ase have indicated that Ni-AB is most likely to be two-electron more oxidized that Ni-G [13]. If Ni-C is a hydride (or H2) species the two electrons would reside in the hydride (or H2) ligand, implying that the formal valencies of both active site metal ions do not change upon reduction from the oxidized Form A or Form B to Form C. In addition, the fact that the three high-frequency IR bands of Form AB and Form C have similar positions in C. vinosum H2ase [12] and in D. gigas Hzase [44] (Table 3) indicates that the iron bound to the non-protein diatomic ligands has most probably the same formal charge in both forms. However, it has been argued that Ni-A, Ni-B and Ni-C cannot be in the same oxidation state since the
Nickel-IronHydrogenases:Structuraland FunctionalProperties
119
Table 3. Wavenumbers (in cm-1) of infrared absorbance bands corresponding to the different
Ni EPR assigned redox states of 1). gigas and C. vinosum H2ase.Adapted from [44] and [12] D. gigas H2ase
Form A Form B (SU) (SI-1) (SI-2) Form C (R)
1947 1946 1950 1914 1934 1952 1940
C. vinosum H2ase
2083 2079 2089 2055 2075 2073 2060
2093 2090 2099 2069 2086 2086 2073
Form AB
1944 2081 20931
(SI)
1910 2051 20671
Form C (R)
1950 2076 2088 1936 2060 2075
1 Only one set of bands has been observed. N.B. See [12] and [44] for the experimental conditions:
oxidized Forms AB do not bind CO (which favors a Ni +3 state) whereas the reduced Form C does (which favors Ni ÷1) [57]. A simple explanation could be that Ni-A and Ni-B in the Ni +I state do not bind CO because the CO binding site to the nickel is occupied, for example, by some oxo or hydroxo species. Further experimental evidence is required to clarify that point. 5.2.4
Spectroscopy of the Iron It is rather intriguing that EPR studies do not provide any indication of the presence of a nearby paramagnet close to the Ni. The only possible explanation is that the Fe ion is diamagnetic in all EPR detectable states of the nickel, Ni-AB and Ni-C. This requires a low-spin Fe2÷ formal oxidation state, which is normally only observed in a strong octahedral ligand field [44]. This is in accordance with the proposed assignment of three diatomic ligands as one CO and two CN- to the Fe coordination sphere [55]. EPR spectroscopy on 57Fe and 61Ni-enriched C. vinosum H2ase showed broadening or splitting only in the 61Ni-enriched preparations. This suggested that unpaired spins are localized on the nickel rather than on the iron. However, as seen above, IR spectroscopy indicates that changes in the charge density are occurring most probably in the vicinity of the iron. Therefore, it has been suggested that electronic changes on the Ni ion could be transferred to the Fe ion [78]. M6ssbauer spectroscopy of the same C. vinosum H2ase revealed magnetic interactions between the [3Fe-4S] cluster, an unidentified (X) paramagnetic moiety (with a mid-point potential of + 150 mV) and the nickel atom [79]. X was tentatively assigned to a low-spin Fe ion located between the nickel ion and the [3Fe-4S] cluster. The crystal structure (Figs. 68, 6b, 8) shows that X cannot be assigned to the iron atom found at the active site which is opposite the [3Fe-4S] cluster with respect to the nickel. The identification of X is undoubtedly one of the most interesting challenges concerning hydrogenase in the near future.
120
M. Frey
0.0 g
1.0
I b
I -8
,
I
,
-4 Ve locit:v
I 0
,
I 4
,
I 8
(mm/sJ
Fig. 9. Crystallographic and spectroscopic evidence of the presence of a [3Fe-4S] cluster. a The [3Fe-4S] ° cluster of the D. gigas hydrogenase and its immediate environment as modelled in an electron density map at 2.54 A resolution contoured at a 2o level. The iron and inorganic sulfur atoms of the cluster are represented as red and green spheres respectively. Adapted from [10]. b M6ssbauer spectra of the Desulfovibrio gigas hydrogenase poised at a redox potential of -207 mV for which the [3Fe-4S] cluster was reduced but the two [4Fe-4S] clusters were still oxidized. The experimental data (C. vertical bars) were recorded at 4.2 °K with a parallel field of IT. (A) theoretical spectrum of the [3Fe-4S] ° cluster. (B) theoretical spectrum of the two [4Fe-4S] 2+ clusters. (C) Experimental spectrum (C vertical bars) superimposed with the theoretical spectrum (solid line) including both the [3Fe-4S] ° cluster and the two [4Fe-4S] 2+ clusters. From [25] with permission
121
Nickel-Iron Hydrogenases:Structural and Functional Properties
6 The D. gigas H2ase model is a Prototype of [NiFe] Hydrogenases Functional and sequential analogies led to the description of hydrogenases as composed of three parts (we term as domains; Fig. 10) [9]. Domain 1 contains the hydrogen-activating site, domain 2 is the electron-transfer part and domain 3 the electron acceptor or donor. The crystal structure of the [NiFe] D. gigas H2ase fits this scheme remarkably well as the large subunit contains exclusively the active site and the small subunit holds the probable electron transfer route. Domain 3 is most probably a multihemic cytochrome associated with the enzyme [e.g. 60, 61]. In the crystal structure domain 2 is composed of two subdomains: (A) the Nterminus domain, binding the [4Fe-4S] cluster proximal to the active site, which is remarkably similar to that of the redox protein flavodoxins, and (B) the C-terminus domain binding the [3Fe-4S] and [4Fe-4S] distal clusters (Fig. 6a). Since the N-terminus (A) domain, including the proximal [4Fe-4S] cluster, is found in the small subunit of many [NiFe] hydrogenases [15, 80], it is probably essential to H2 activation in [Ni-Fe] hydrogenases [9, 10]. Several residues of the large subunit interacting with this same [4Fe-4S] cluster are also very conserved pointing further to the role of the [4Fe-4S] proximal cluster as a direct partner of the catalytic site [10]. This scheme can be extended to more complex [Ni-Fe] hydrogenases such as the NAD+-dependent hydrogenase from Alcaligenes eutrophus, which contains one hydrogenase moiety consisting of two subunits named fl and 6. The fl subunit contains a nickel ion, most probably involved in the catalytic activity, and a [4Fe-4S] cluster. The 6 subunit contains a single [3Fe-4S] cluster which might transfer electrons to the NAD+-reducting moieties a and y [81]. The three unusual infrared bands which have been assigned to one CO and two CN- liganded to the iron atom [55] are also observed for many [NiFe] and [Fe] hydrogenases [44, 78]. This and sequence homologies suggest that the active site as seen in the crystal structure of the D. gigas H2ase is similar in all hydrogenases with a binuclear centre (Ni/Fe or Fe/Fe), a low-spin iron and three non-protein ligands [78].
H2
part2
Envlronm
2e-
~____~yAH 2
'
,
rganlsm +
2H
+
A +2H l I
~ative
II II
Variable
I I
Fig. 10. Schematic representation of [NiFe] hydrogenases as an oxidoreductase. Part I contains the hydrogen-activating site. Part 2 serves as an electronic interface. Part 3 is an electrochemical interface which accepts electrons from Part 2 and reduces a substrate in the organism. From [9] with permission
122
M. Frey
Fig. 11. Stereoscopicview of the C-terminal end of the large subunit of the D. gigas H2asefitted to an electron density map at 2.54/~ resolution contoured at a 2a level. The putative hydrogen bonds are represented by dashed lines coloured in magenta. Adapted from[10].
7
Biosynthesis of the Enzyme The biosynthesis of hydrogenases is a complex process which involves particularly incorporation of the active site nickel and iron atoms and the CO and two CN- ligands [61, 81- 84]. Interestingly, the determination of the crystal structure casts light on some steps in the maturation of the enzyme which involve the nickel atom [10]. Interpretation of the electron density maps at 2.85 A and 2.54 -~ resolution showed that the large subunit could not be modelled beyond His 536 (Fig. 11), that is 15 amino acids short of the 551 residues predicted by the nucleotide sequence [47]. This was confirmed by a mass spectrometric analysis of the samples which shows that the relative molecular mass of the large subunit (59,459 + 6) is close to the theoretical value of 59,704 calculated after omitting the 15-residue cleaved peptide [10]. These results are likely to be biologically relevant since it has been reported that the proteolytic cleavage of these last 15 residues (including the cysteines 530 and 533 liganding the nickel and iron atoms) by a specific protease is an obligatory step for the maturation of the enzyme which is associated with the formation of the active site [83 and for review, 84].
8 Concluding Remarks The first structure of hydrogenase, the enzyme which catalyzes the activation of molecular hydrogen, has been elucidated at the atomic level in its oxidized form. X-ray crystallography combined with spectroscopic studies (EPR and IR) have shown that the catalytic site of the enzyme contains a bimetallic center with one nickel and one iron atom (about 2.9 A apart from each other) coordi-
Nickel-Iron Hydmgenases:Structural and Functional Properties
123
nated to the protein by only four cysteine sulfur ligands. The iron atom is itself bound to three non-protein ligands (i.e. one CO and two CN-). During catalysis, the nickel atom is most likely responsible for a base-assisted heterolytic cleavage of the hydrogen molecule, whereas the iron atom could be redox-active. Specific transfer routes are probably required for the transfer of the chemical reaction partners (H2, H + and e-) between the active site, deeply buried inside the protein, and the molecular surface. In the near future genetic studies combined with X-ray and spectroscopic analysis should help in a better understanding of the hydrogenase function through, for example, appropriate mutations of some amino acids which seem to play a role in the catalytic mechanism (e.g. those involved in the active site or in the putative electron transfer route). X-ray crystallographic analyses of fully reduced (R) and CO-complexed D. gigas H2ase, combined with spectroscopic analyses and redox titrations, should provide a more precise picture of the active site during activation and catalysis (e.g. the formal oxidation states of the nickel and iron atoms during the catalytic cycles) and an indirect view of the substrate binding site. Genetic studies are also expected to unravel the mechanism of protein maturation including the incorporation of the complex active site. New developments are also expected in the chemistry of model compounds mimicking the hydrogenase activity in view of the recent synthesis of a binuclear Ru-Fe complex which transforms a two-electron reductant (such as H2) into two single-electron reducing equivalents (redox switch reaction, 75, 76). Figs. 5, 7a, 9a and 11 were made with the O package [89]. Figs. 6a, 6b, 7b and 8 were made with MOLSCRIPT [90] and the Raster3D molecular graphics package [91, 92]. The atomic X-ray model of the oxidized hydrogenase from D. gigas at 2.85 (lfrv) is deposited with the Protein Data Bank (http://pdb.pdb.bnl.gov). Acknowledgements,I am indebted to Dr. Claude Hatchikian for his outstanding and generous
collaboration and Dr. Juan-Carlos Fontecilla-Campsfor his enthusiastic and stimulating involvement in the hydrogenase crystallographic project. I gratefully acknowledge the highquality work of Dr. Anne Volbeda particularly during some crucial steps of the determination of the crystal structure and its interpretation. Protein crystallography is nothing without crystals. For that my colleagues and myself are greatly indebted to Dr Marie-H~l~neCharon and Claudine Piras for managing to obtain X-ray-suitablecrystals reproducibly.I also want to say many thanks to our young colleagues Elsa Garcin, YaWlMontet and Xavier Vern~de for their motivation and high quality contributions. Many thanks also to Drs. Simon Mbracht, Richard Cammack and Victor Fernandez for interesting and illuminating discussions, Drs. Kim Bagleyand Mike Maroney for their kind permission to reproduce figures of their results and Dr. Yoshiki Higuchi for communicating experimental results concerning the determination of the Desulfovibrio vulgaris hydrogenase X-ray structure prior to publication. This work was financially supported by the Commissariat ~ 1' Energie Atomique (CEA), the Centre National de la Recherche Scientifique (CNRS) and the European Union BiotechnologyProgram.
124
M. Frey
9 References 1. Stephenson M, Sticldand LH (1931) Biochem J (London) 25:205 2. Odom JM, Peck HD Jr (1984) Annu Rev Microbiol 38:551 3. Hatchikian EC, Fernandez V, Cammack R (1990) The hydrogenases of sulfate-reducing bacteria: physiological, biochemical and catalytic aspects. In: Belaich JP, Bruschi M, Garcia JL (eds) Microbiology and Biochemistry of Strict Anaerobes Involved in Interspecies Hydrogen Transfer, Plenum Press, New York and London, p53 4. Adams MWW, Mortenson LE, Chen J-S (1981) Biochim Biophys Acta 594:105 5. Adams MWW (1990) Biochim Biophys Acta 1020:115 6. Postgate JR (1984) The sulphate-reducing bacteria, 2nd edn. Cambridge University Press, Cambridge, UK 7. Zirngibl C, Van Dongen W, Schw6rer B, yon Bfinau R, Richter M, Klein A, Thauer RK (1992) Eur J Biochem 208:511 8. Przybyla AE, Robbins J, Menon N, Peck HD (1992) FEMS Microbiology Reviews 88:109 9. Albracht SPJ (1994) BiochimBiophys Acta 1188:167 10. Volbeda A, Piras C, Charon MH, Hatchikian EC, Frey M, Fontecilla-Camps JC (1995) Nature 373:580 11. Bagyinka C, Whitehead, JP, Maroney, MJJ (1993) J Am Chem Soc 115:3576 12. Bagley KA, Duin EC, Roseboom W, Albracht SPJ and Woodruff WH (1995) Biochemistry 34:5527 13. Roberts LM, Lindahl PA (1995) J Am Chem Soc 117:2565 14. Sorgenfrei O, Linder D, Karas M, Klein A (1993) Eur J Biochem 213:1355 15. Voordouw G (1992) Evolution of hydrogenase genes. In: Cammack R, Sykes AG (eds) Adv Inorg Chem, Academic Press, San Diego, London 38:397 16. Teixeira M, Fauque GD, Moura I, Lespinat PA, Berlier Y, Prickril BC, Peck Hd Jr, Xavier AV,LeGall J, Moura JJG (1987) Eur J Biochem 167:47 17. Lancaster JR (1982) Science 216:1324 18. Graf E-G, Thauer RK (1981) FEBS Lett 136:165 19. LeGall J, Ljungdahl P, Moura I, Peck HD Jr, Xavier AV,Moura JJG, Teixeira M, Huynh B-H, DerVartanian DV (1982) Biochem Biophys Res Comm 106:610 20. Cammack R, Patil D, Aguirre R, Hatchikian EC (1982) FEBS Lett 142:289 21. Hatchikian EC, Bruschi M, LeGall J (1978) Biochem Biophys Res Comm 82:451 22. Teixeira M, Moura I, Xavier AV,DerVartanian DV, LeGall J, Peck HD Jr, Huynh B-H, Moura JJG (1983) Eur J Biochem 130:481 23. Krfiger H-J, Huynh BH, Ljungdahl PO, Xavier AV,DerVartanian DV, Moura I, Peck HD Jr, Teixeira M, Moura JJG, LeGall J (1982) J Biol Chem 257:14620 24. Huynh B-H, Patil DS, Moura I, Teixeira M, Moura JJG, DerVartanian DV, Czechowski MH, Prickril BC, Peck HD Jr, LeGall J (1987) J Biol Chem 262:795 25. Teixeira M, Moura I, Xavier AV,Moura JJG, LeGall J, DerVartanian DV, Peck HD Jr, Huynh B-H (1989) J Biol Chem 264:16435 26. Farkas A, Farkas L, Yudkin J (1934) Proc Roy Soc 115:373 27. Krasna AI (1978) Methods in Enzymology 53:296 28. Lespinat PA, Berlier Y, Fauque G, Czechowski M, Dimon B, LeGall J (1986) Biochimie 68: 55 29. Cammack R, Fernandez VM, Hatchikian EC (1994) Nickel-iron hydrogenases. In: Peck HD Jr, LeGall J (eds) Methods in Enzymology. Academic Press, San Diego, London 243:43 30. Van der Zwaan JW, Albracht SPJ, Fontijn RD, Slater EC (1985) FEBS Lett 179:271 31. Fernandez VM, Aguirre R, Hatchikian EC (1984) Biochim Biophys Acta 790:1 32. Fernandez VM, Hatchikian EC, Cammack R (1985) Biochim Biophys Acta 832: 69 33. Fernandez VM, Hatchikian EC, Patil DS, Cammack R (1986) Biochim Biophys Acta 883: 145 34. Cammack R, Patil D, Hatchikian EC, Fernandez VM (1987) Biochim Biophys Acta 912: 98 35. Barondeau DP, Roberts LM, Lindahl P (1994) J Am Chem Soc 116:3442
Nickel-Iron Hydrogenases:Structural and Functional Properties
125
36. Van der Zwaan JW, Albracht SPJ, Fontijn RD and Mul P (1987) Eur J Biochem 169:377 37. Whitehead JP, Gurbiel RJ, Bagyinka C, Hoffman BM, Maroney MJ (1993) J Am Chem Soc 115:5629 38. Blundell TL, Johnson LN (1975) Protein Crystallography.Academic Press, London 39. Drenth J (1994) Principles of protein X-ray Crystallography. Springer, Berlin Heidelberg New York 40. Ducruix A, Gi~g~ R (1992) Crystallization of nucleic acids and proteins. A practical approach, IRL, Oxford 41. Higuchi Y, Yasnoka N, Kakudo M, Katsube Y, Yagi T, Inokuchi H (1987) J Biol Chem 262: 2823 42. Nivi~re V, Hatchikian EC, Cambillau C, Frey M (1987) J Mol Biol 195:969 43. Higuchi Y, Okamoto T, Fujimoto K, Misaki S, Morimoto Y, Yasuoka N (1994) Acta Cryst D50:781 44. Volbeda A, Garcin E, Piras C, de Lacey AL, Fernandez VM, Hatchikian EC, Frey M, Fontecflla-Camps JC (1996) J Am Chem Soc 118:12989 45. Fourme R, Hendrickson WH (1990) In: Hasnain (ed) Synchrotron and biophysics. Ellis Horwood, Chichester, p 156 46. Volbeda A, Piras C, Charon MH, Hatchikian EC, Frey M, Fontecilla-Camps JC (1993 ) Joint CCP4 ESF-EACBM Newsletter on Protein Crystallography Daresbury Laboratory Warrington, UK 28:30 47. Voordouw G, Menon NK, LeGall J, Choi E-S, Peck HD Jr, Przybyla AE (1989) J Bacteriol 171:2894 48. Eidsness MK, Scott RA, Prickrfl BC, DerVartanian DV, LeGall J, Moura I, Moura JIG, Peck HD Jr (1989) Proc Natl Acad Sci USA 86:147 49. He S-H, Teixeira M, LeGall J, Patil DS, Moura I, Moura JJG, DerVartanian DV, Huynh B-H, Peck HD Jr (1989) J Biol Chem 264:2678 50. Engh RA, Huber R (1991) Acta Cryst A47:392 51. Holm RH (1992) Trinuclear cuboidal and heterometallic cubane-type iron-sufur clusters: new structural and reactivity themes in Chemistry and Biology. In: Cammack R and Sykes AG (eds) Advances in Inorganic Chemistry Vol 32, Iron-Sulfur Proteins. Academic Press, San-Diego London, p 1 52. Gessner Ch, Trofanchuk O, Kawagoe K, Higuchi Y,Yasuoka N, Lubitz W (1996) Chem Phys Lett 256:518 53. Cammack R (1992) Iron-Sulfur Clusters in Enzymes: Themes and variations. In: Cammack R and Sykes AG (eds) Advances in Inorganic Chemistry Vol 38, Iron-sulfur proteins. Academic Press, San-Diego, p 281 54. Bagley KA,Van Garderen CJ, Chen M, Duin EC, Roseboom W,Albracht SPJ,WoodruffWH (1994) Biochemistry 33:9229 55. Happe RP, Roseboom W, Pierik AJ, Albracht SPJ, Bagley KA (1997) Nature 385, 126 56. Van der Zwaan JW, Coremans JMCC, Bouwens ECM, Albracht SPJ (1990) Biochim Biophys Acta 1041:101 57. Roberts LM, Lindahl PA (1994) Biochemistry 33:14339 58. Onuchic JN, Beratan DN, Winkler JR, Gray HB (1992) Annu Rev Biophys Biomol Struc 21: 349 59. Moser CC, Keske IM, Warncke K, Farid RS, Dutton PL (1992) Nature 355:796 60. Cammack R (1995) Nature 373:556 61. Nivi~re V, Hatchikian EC, Bianco P, Haladjian J (1988) Biochim Biophys Acta 935:34 62. Voordouw G (1996) Function and assembly of periplasmic hydrogenases in Desulfovibrio. In: Regulation of Hydrogenase Biosynthesis. European Union COST Action 818 Workshop. Berlin 12th-14th Sept 1996 63. Okamura MY, Feher G (1992) Annu Rev Biochem 61:861 64. Williams RJP (1995) Nature 376:643 65. Boxer SG, Aufinrud PA (1994) Nature Struc Biol 11:749 66. Marganian Goldman C, Mascharak PK (1995) Comments Inorg Chem 18:1
126
M. Frey
67. Moura JJG, Teixeira M, Moura I, LeGall J (1988) [Ni-Fe] hydrogenases from sulfatereducing bacteria: nickel catalytic and regulatory roles. In: Lancaster JR (ed) Nickel in biochemistry. VCH, New-York, p 191 68. Cammack R (1988) Nickel in metalloproteins. In: Sykes AG (ed) Advances in Inorganic Chemistry Vo132, Iron-Sulfur Proteins. Academic Press, San-Diego London, p 297 69. Gu Z, Dong J, Allan CB, Choudhury SB, Franco R, Moura JJG, Moura I, LeGall J, Przybyla AE, Roseboom W, Albracht SPJ, Axley MJ, Scott RA, Maroney M (1996) J Am Chem Soc 118:11155 70. Fontecilla-Camps JC (1996) J Bio Inorg Chem 1:91 71. FonteciUa-Camps JC, Volbeda A, Frey M (1996) Trends in Biotechnology 14:417 72. Marganian CA, Vazir H, Baidya N, Olmstead MM, Mascharak PK (1995) J Am Chem Soc 117: 1584. 73. Fan C, Teixeira M, Moura J, Moura I, Huynh B-H, LeGall J, Peck HD Jr, Hoffman BM ( 1991) J Am Chem Soc 113:20 74. Ceriotti A, Chini P, Fumagalli A, Koetzle TF, Longoni G, Takusagawa F (1984) Inorg Chem 23:1363 75. Hembre RT, McQueen JS and Day VW (1996) J Am Chem Soc 118:798 76. Collman JP (1996) Nature Structural Biology 3:213 77. Dole F, Medina M, More C, Cammack R, Bertrand P, Guigliarelli B (1996) Biochemistry 35:16399 78. Van Der Spek TM, Arendsen AF, Happe RP, Yun S, Bagley KA, Stufkens DJ, Hagen WR, Albracht SPJ (1996) Eur J Biochem 237:629 79. Surerus KK, Chen M, van der Zwaan JW, Rusnak F, Kolk M, Duin EC, Albracht SPJ, Miinck E (1994) Biochemistry 33:4980 80. Wu L-F, Mandrand MA (1993) FEMS Microbiology Reviews 104:243 81. Friedrich B, Schwartz E (1993) Ann Rev Microbiol (1993) 47:351 82. Vignais PM, Toussaint B (1994) Arch Microbiol 161:1 83. Menon NK, Robbins J, Der Vartanian M, Patil D, Peck HD Jr, Menon AL, Robson RL, Przybyla AE (1993) FEBS Lett 331:91 84. Maier T, B6ck A (1996) In: Hausinger RP, Eichorn GL, Marzilli LG (eds). Advances in Inorganic Biochemistry: Mechanisms of Metallocenter Assembly.VCH, New York, p 173 85. Nivi~re V, Bernadac A, Forget N, Fernandez V, Hatchikian EC (1991) Arch Microbiol 155: 579 86. Lissolo T, Pulvin S, Thomas D (1984) I Biol Chem 259:11725 87. Berlier YM, Fauque GD, LeGall J, Lespinat PA, Peck HD Jr (1987) FEBS Lett 221:241 88. Berlier YM, Fauque GD, LeGall J, Choi ES, Peck HD Jr, Lespinat PA (1987) Biochem Biophys Res Comm 146:147 89. Jones TA, Zou T-Y, Cowan SW, Kjelgaard M (1990) Acta Cryst A47:110 90. Kraulis PJ (1991) I Appl Cryst 24:946 91. Bacon DJ, Anderson WF (1988) J Mol Graphics 6:219 92. Merritt EA, Murphy MEP (1994) Acta Cryst DS0:869
Coordination Sphere Versus Protein Environment as Determinants of Electronic and Functional Properties of Iron-Sulfur Proteins F r a n c e s c o Capozzi 1, Stefano Ciurli 1, Claudio Luchinat 2, 1 Institute of Agricultural Chemistry, University of Bologna, Viale Berti Pichat 10, 1-40127 Bologna, Italy 2 Department of Soil Science and Plant Nutrition, University of Florence, P.le delle Cascine 28,1-50144 Florence, Italy The aim of this article is to critically discuss the information available on the influence of the protein environment on the electron transfer properties of Fe-S proteins. First, the intrinsic redox and structural features of synthetic Fe-S clusters are illustrated, in order to provide a background on which the effects of the protein matrix can be subsequently highlighted. The parameters found to influence the reduction potential in metalloproteins are then discussed, and a detailed analysis of structure-function relationships among the different classes of Fe-S proteins is carried out. This analysis reveals that, within each class, one of the parameters is often more important than the others in determining the reduction potential of the Fe-S center; the parameter changes upon passing from one class to the other. Within the class of high potential Fe-S proteins the major determinant for the variation of reduction potential is the difference in charged residues, whereas in rubredoxins and [2Fe-2S], [3Fe-4S], and [4Fe-4S] ferredoxins, this effect is attenuated and overwhelmed by solvent accessibilityto the cluster core and/or inner-fractional charges in the protein itself. A deeper understanding of the influence of the structural properties of the protein is gained by analysing the microscopic reduction potentials of the individual Fe ions in Fe-S centers. The latter are obtained from the electronic distribution within each type of cluster core, which, in turn, is determined from a theoretical analysis of the electron-nucleus coupling derived from NMR and other spectroscopic analysis.
Keywords:FeS proteins, redox properties, HiPIP, ferredoxins, iron-sulfur clusters.
1 Introduction ...................................
128
2 Classes o f Iron-Sulfur Clusters
131
........................
3 Protein-Induced Variability o f Reduction Potentials Within Each Type o f Cluster . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
135
4 Electronic Structures and Microscopic Reduction Potentials
148
5 References
....................................
......
157
* Correspondence: Prof. Claudio Luchinat, Department of Soil Science and Plant Nutrition, University of Florence, P.le delle Cascine 28,1-50144 Florence, Italy. Structure and Bonding,Vol.90 © SpringerVerlag BerlinHeidelberg1998
128
E Capozzi • S. Ciurli • C. Luchinat
1 Introduction Electron transfer reactions occur in living organisms over a wide range of reduction potentials. Metal ions have proven to be more versatile than organic redox molecules for this task. Indeed, different metal ions with different redox properties are available to living organisms and, for a given metal ion, the reduction potential is strongly dependent on the coordination sphere (nature, number and stereochemistry of the donor atoms). Furthermore, for a given metal ion and coordination sphere, the reduction potential can still be modified over a relatively wide range. This last effect is due to the properties of the surrounding protein matrix. It is perhaps less understood but not less important than the others. Copper and iron are among the most important metal ions used in redox metalloproteins [1]. Copper is found in many metalloproteins with different functions, but in electron transfer processes it is found almost exclusively in blue proteins. Blue proteins share the same coordination sphere [2, 3], and their redox potentials span a range of slightly over 150 mV [4]. Iron is found in two main classes of electron transfer proteins, cytochromes and iron-sulfur proteins. Cytochromes all contain iron bound to a porphyrin (heme iron), but are further subdivided according to the type of heme-protein linkage and to the presence and type of axial ligands (Fig. 1) [5, 6]. It appears that axial ligands are very important in determining the value of the reduction potential [7-12]. However, within each subclass, a large modulation induced by the protein matrix can still be appreciated. In the case of cytochromes, a further means of modulating the reduction potential is provided by the possibility of embedding more than one heine in the same protein matrix [13, 14]. Here, besides differential protein effects on each heme, electrostatic effects among the hemes themselves could be important [15]. Perhaps the most striking effects of protein-induced modulations, coupled with metal-metal interactions, are found in iron-sulfur proteins [16]. They can also be grouped into different classes according to i) the number of iron ions contained in the cluster and ii) the number and type of clusters contained in the same protein. Striking variations in reduction potentials are observed on passing from one type of cluster to another, and also within the same type of cluster (Fig. 2) [17]. A close-up analysis of the factors that may play a role in differentiating the E°' values in Fig. 2 is the aim of this review. The picture is still under development and many issues remain to be solved. The unsolved issues regard not only the adequacy of the present theoretical models (see for instance, the contrasting views expressed in references [18-21]), but als0 the accuracy of the structural data on which E°" predictions are made and the accuracy and homogeneity of the literature E°" values to be reproduced. Nevertheless, this analysis may already prove helpful at this stage for a better understanding of the functional properties of this class of proteins.
Coordination Sphere Versus Protein Environments
|29 Q 0 tc)
A
A
4-
m m tJ
I
¢J
° o
~%
o °
~ ' ~
°~
m~ m,,-i m..im~ m~ m y m o
o
o o
~
~.~
~
o
p
~J
~.~ ~o
I. I
I
o c~
~
I
0
(3 O
m o
~
~.~ .~ ~
° Lr)
O
~,~ .
m
130
E C a p o z z i • S. Ciurli • C. L u c h i n a t o o t~
0 0
0 0
C~
'~
o
N 0 "0
N
0 '0
k
k k
',0
I
o o
oo
-
o
C)
0
° o
o
~d 0
o
°
Coordination Sphere Versus Protein Environments
131
2 Classes of Iron-Sulfur Clusters All iron-sulfur centers known to exist in electron transfer proteins have their counterpart in terms of synthetic coordination compounds [22, 23]. In order to make a comparison between the properties of iron-sulfur centers in proteins and model compounds, we briefly review in this paragraph the main structural and electronic properties of the latter systems, in which no or little constraints are imposed on the metal centers by the ligands to the Fe ions. The relevant structures are shown in Fig. 3. The [Fe(SCFs)4] 1-/2- center present in rubredoxins [24-27] (Fig. 3A) can be compared with the model analogue [Fe(Sa-o-xylyl)~] 1-/a- [28] (Table 1. In oxidized rubredoxins, the Fe(III)-SCys distance is 2.29 A, while in the isoelectronic [Fe(Sa-o-xylyl)a] 1- model compound, the Fe-SR distance is 2.267 .~, in agreement with the structure of the biological system. Upon reduction of the model compound, the Fe-SR distance increases significantly to 2.356 A, while nonsignificant changes can be observed in reduced rubredoxins, maybe because of the lower resolution attainable in the crystal structure of the higher molecular weight protein. The increase in the metal-to-ligand distance upon reduction reflects the increase of the Fe ionic radius, and is common, to a variable extent, to all other Fe-S dusters which have been structurally characterized. The reduction potential of the couple [Fe(S2-o-xylyl)2]l-/2-in DMF (- 1030 mV vs NHE) [28] is much more negative than potentials for the Rdox/Rdrea couple, as shown
Fig. 3. Coordination geometries of [Fe(SR)4] (A), [Fe2S2(SR)4] (B), [Fe3S4(SR)3] (C), and
[Fe4S4(SR)4] (D) systems in proteins (carbon atoms in black, sulfur atoms in light gray, and iron atoms in dark gray)
132
F. Capozzi • S. Ciurli • C. Luchinat
Comparison of average structural and redox properties of rubredoxins and a representative model compound
Table 1.
Redox State
Core Unit Formal Iron Oxidation State Fe-SR Distance (4) E°' (mV) vs. NHE Protein Data Bank Structure Codes
1-Fe Rubredoxins
Model Compound [28]
Oxidized
Oxidized
Reduced
Reduced
[Fe(SCys)4] 1- [Fe(SCys)4] 2Fe(III) Fe(II)
Fe(III)
Fe(II)
2.29
2.267
2.35
2.30
- 60 + +43 IlRO 1CAD 1CAA 8RXN 6RXN 1RDG 7RXN
[Fe(S2-o-xylyl)2] I- [Fe(S2-o-xylyl)2] 2-
- 1030 mV in DMF -
-
in Table 1. Substantially more negative potentials in non-aqueous media compared to protein potentials in aqueous solution are properties common to all 1-, 2-, 3-, and 4-Fe synthetic analogues. This effect can be ascribed to the higher dielectric constant of water (80.2) as compared to organic polar solvents such as DMF (36.7) or MeCN (37.5), and to the presence of hydrogen bonding which causes a stabilization of the doubly charged [Fe(SR)4] 2- center. This point will be discussed later in more detail. The known structures of the [Fe2S2(SCys)4]2- centers in [2Fe-2S] ferredoxins [29-32] (Fig. 3 B), formally containing two Fe(III) ions, show structural parameters (Table 2) typical for other [Fe2S2] units found in polynuclear iron-sulfur clusters. In order to make a structural comparison with a synthetic analogue, the compound [Fe2S2(S2-o-xylyl)2] 2- is taken as a representative compound [33]. As shown in Table 2, the metric features again indicate a substantial invariance of the cluster core structure. A striking difference is again observed however, when one compares the reduction potential with the [Fe2S2(SR)4]2-/3- couple: the model compound shows a much more cathodic process, indicating a stabilization of the oxidized core. The effect of the core environment can be again considered to drastically control the redox properties of the iron-sulfur cluster. A comparison with the reduced form of the cluster either in the protein or in synthetic analogues cannot be carried out in this case because of the absence of available structures. The [Fe3S4(SCys)3]3- cluster core found in proteins [34-36] (Fig. 3C) has been recently synthesized using the semi-rigid cavitand ligand LS3 [23]. Both in proteins and in the synthetic analogue, the [Fe3S4] unit consists of a cuboidal core, in which one vertex, usually occupied by an Fe ion in [4Fe-4S] ferredoxins, is missing. The structural features show a striking similarity between the biological and the synthetic unit (Table 3), indicating that the role of the protein matrix in stabilizing such a core is readily also performed by an organic ligand,
Coordination SphereVersus Protein Environments
133
Table 2. Comparison of average structural and redox properties of [2Fe-2S] ferredoxins and
a representative model compound Redox State
Core Unit Formal Iron Oxidation State Fe-SR Distance (A) Fe-S* Distance (~)
[2Fe-2S] Ferredoxins
Model Compound [33]
Oxidized
Reduced
Oxidized
Reduced
[Fe2S2 (SCys)4]2Fe(III) Fe(III) 2.25
[Fe2S2 (Says)4] sFe(II) Fe(III)
[Fe2S2 (S2-o-xylyl)2]2Fe(III) Fe(III) 2.304
[Fe2S2 (S2-o-xylyl)2]3Fe(III) Fe(III)
2.21
E°' (mV) vs. NHE - 430 + - 240 Protein Data Bank 1FRR Structure Codes 1FRD 1DOI 4FXC 1FXI
2.208 - 1500 in DMF
which, although it has a preferential conformation, is not completely rigid This allows one to conclude again that the structure of Fe-S cores is an intrinsic property not necessarily enforced by the protein matrix. As in the previous examples the reduction potential in MeCN is much more negative than that observed in proteins (Table 3). The first [Fe4S4] 2+ cluster in biological systems (Fig. 3D) was structurally characterized and reported in 1972 [37], and shortly afterwards the analogue compound [Fe4S4(SCH2Ph)4] 2- was synthesized [38]. The structures of the [Fe4S4] cluster in the other two oxidation states, [Fe4S4] 3+ [39-42] and [Fe4S411+[36], both known to be accessible in biological environments, also have their counterpart in synthetic model chemistry [43-45]. Table 4 shows that a slight dependence of the Fe-SR and Fe-S* distances from the core oxidation state exists in proteins, which suggests that upon reduction by one electron an increase in the cluster core dimension is induced. The low resolution of the protein structures does not allow a more detailed discussion. If one considers that in mononuclear Fe(SR)4 centers the reduction of the metal ion oxidation state by one unit induces an increase in the Fe-SR distance of the order of 0.1 .&, and that in [Fe4S4] cores the increase in negative charge is delocalized over at least two Fe ions, one cannot expect to detect such a small increase in protein structures, due to the low resolution attainable. A more defined increase is instead appreciable in model compounds, because the resolution of the crystal structures is higher (Table 4). The Fe-SR distance increases by 0.04 and 0.05 ~, upon reduction of the [Fe4S4]3+ core by one and two electrons, respectively, while the Fe-S* distance increases by 0.03 and 0.03 ~. The Fe-Fe distances do not vary appreciably. Furthermore, crystallographic studies indi-
134
E Capozzi • S. Ciudi • C. Luchinat
Comparison of average structural and redox properties of [Fe3S4] dusters in ironsulfur proteins and a representative model compound Table 3.
Redox State
Core Unit Iron Oxidation State Formal
Fe-SR Distance (•) Fe-S* Distance Fe-S3Distance (~)
[3Fe-4S] Ferredoxins
Model Compound [23]
Oxidized
Reduced
Oxidized
Reduced
[Fe3S 4 (SCys)3] 2-
[Fe3S4
-
[Fe3S4(LS3)] 3-
Fe(III) Fe(III) Fe(III) 2.28
Fe(III) Fe(III) Fe(II) 2.29
Fe(III) Fe(III) Fe(III) -
Fe(III)
2.29 2.29
2.28 2.26
-
2.26 2.31
[(SCys)3] 3-
E°' (mV) vs. NHE -450 + - 130 Protein Data Bank 1FXD 1FDB Structure Codes 1FDA 1FDC 5FD1 5ACN
Fe(III) Fe(II) 2.32
- 1060 mV in MeCN
cate that the structure of the [Fe4S4] 3+ cluster and 12 out of 14 structures of the [Fe4S4] 2+ clusters possess compressed D2a core geometry (4 short and 8 long Fe-S* bonds), while 3 out of 8 X-ray structures of [Fe4S4] 1+ clusters show elongated D2d core geometry (4 long and 8 short Fe-S* bonds). These effects could be due to solid state effects, because resonance Raman solution studies suggest a T d structure for [Fe4S4]2+ clusters in solution [46]. A comparison of the structural features of [Fe4S4] clusters in biological and non-biological environments then shows, once again, a substantial invariance, confirming that the intrinsic structural properties of the cluster cores impose the geometry of the iron-sulfur moiety on the protein matrix. However, a large difference is observed when one compares the reduction potentials of the redox couples functioning in low-potential ([Fe4S4]2+11+),or high-potential ([Fe4S4]3+/2+) ferredoxins with the corresponding potentials observed in model compounds, as shown in Table 4. The much lower values characterizing the synthetic models [38, 45] can be ascribed to the lower dielectric constant or the lack of H-bonding capability of the organic solvents in which the model compounds are usually studied (MeCN or DMF), which stabilize the oxidized form because of the lower charge associated to it, as compared to water. The reduction potential of a synthetic [Fe4S4(SR4)]2- cluster in water (R=CH2CH2OH) was indeed found to be -480 mV [47], i.e. more than 700 mV higher than in MeCN. This comparison would suggest that the iron-sulfur cluster in a protein, including the cysteine residues utilized to anchor it to the protein backbone, is only partially shielded from the water molecule dipoles by the protein matrix itself. Clearly, however, the solvent effect is still very strong. Indeed, the reduction potentials of most [4Fe-4S] ferredoxins are similar to that of the model complex in water.
Coordination Sphere Versus Protein Environments
135
Table 4. Comparison of average structural and redox properties of [4Fe-4S] ferredoxins and
representative model compounds Redox State
Core Unit
[4Fe-4S]Ferredoxins
Model Compounds
Oxidized HiPIP
Reduced HiPIP Oxidized Fd
Reduced Fd
[45]
[Fe4S 4
[Fe4S 4
[Fe4S4 (SCyshP-
[Fe4S4 [Fe4S4 (S-2,4,6- (SCH 2 (i-Pr) 3 Ph)4] 2C6H4)4]l-
Ph)4]3-
(SCys)4]'- (SCyshl2-
[44, 451
[43, 44]
[We4S4 (SCH 2
Formal Iron Fe(III) Oxidation State Fe(III) Fe(III) Fe(II) Fe-SR Distance 2.2
Fe(III) Fe(IlI) Fe(II) Fe(II) 2.3
Fe(III) Fe(II) Fe(II) Fe(II) 2.31
Fe(III) Fe(III) Fe(III) Fe(II) 2.21
Fe(III) Fe(III) Fe(II) Fe(II) 2.25
Fe(III) Fe(II) Fe(II) Fe(II) 2.30
Fe-S* Distance
2.28
2.28
2.26
2.29
2.32
(~)
2.25
(h) E°' (mV) -650 + -280 vs. NHE + 90 + + 460 Protein Data 1HIP 1FCX 1FDC Bank Structure 1HPI 1FCA Codes IISU 2FXB 2HIP 5FD1
- 1200 in MeCN -150 in MeCN
A systematic study of the correlation between cluster solvent accessibility, calculated from the protein structure, and the reduction potential of the protein would be desirable. Such a calculation performed on different classes of Fe-S cores using the WHATIF program [48] indicates that within each class the solvent accessibility is largely constant, but it differs noticeably among the different classes: in particular, we find ca. 10 ,~2/Cys in [4Fe-4S] low potential ferredoxins, ca. 8 AZ/Cys in rubredoxins, ca. 6.5 ~.2/Cys in [2Fe-2S] ferredoxins, and a much lower accessibility (ca. 1 ,~2/Cys) in HiPIPs (see below). This trend might justify the different reduction potentials found in the different classes, but also indicates that other factors are responsible for the modulation of the potential within each protein class.
3 Protein-Induced Variability of Reduction Potentials Within Each Type of Cluster Before we begin the discussion of the experimental data available on the various classes of iron-sulfur proteins, let us examine the problem of protein-induced effects from a theoretical point of view [49, 15, 50,18-21]. To do so, we make the important assumption that a given type of cluster possesses an "intrinsic" reduc-
136
E Capozzi • S. Ciurli • C. Luchinat
Protein-Induced Effects on Cluster Reduction Potential 1. Alteration of Cluster Geometry and Symmetry 2. Electrostatic Effects Due to: i. unit charges ii. fractional charges iii. atomic clouds polarizability 3. Structural Changes Between Oxidized and Reduced States (Reorganization Energy) 4. Solvent-Cluster Interactions 5. Solvent Shielding of Protein-Cluster Electrostatic Interactions Scheme 1
tion potential (see Sect. 2), whose value could also be guessed using quantum mechanical calculations [51]. Embedding this cluster as such in a solvent or in a protein only adds positive or negative contributions to this reference value. The possible protein-induced effects are summarized in five points (Scheme 1). First, the proteins can alter the cluster geometry and symmetry, and can do this in different ways due to strains derived by the different folding of the polypeptide chains. Second, the protein matrices provide different sets of electrostatic interactions with the cluster. These interactions can be further, somewhat artificially, defined as arising i) from interactions of the cluster with protein unit charges (provided by the negatively charged Asp and Glu residues and the positively charged Arg, Lys and His residues at acidic pH), ii) from fractional charges on all individual atoms of the protein, and iii) from the polarizability of atomic clouds. Third, the proteins and the embedded clusters can differ in geometry between the oxidized and reduced forms. Slight geometric changes can result in non-negligible energy differences between the two forms, which obviously contribute to the overall E°" of the system. This energy difference, known as "reorganization energy" [52], in addition to its thermodynamic relevance, is also important in the kinetics of the electron-transfer processes. Fourth, protein-induced solvent-cluster interactions should be considered. This is a complicated matter. An isolated cluster embedded in a solvent (e.g. water) obviously displays strong solvent-cluster interactions. When the cluster is surrounded by a protein, part of these interactions is actually lost. The remaining interactions, however, can still be important and they strongly depend on the details of the protein folding around the cluster. Fifth, the solvent can play an important role in modulating the protein-cluster electrostatic interactions discussed in point 2 above. It should be pointed out that the interactions involving the solvent are the most difficult to treat from a theoretical point of view [15, 50] (see also ref. [21]). While the electrostatic interactions discussed in points i and 2, as well as the energy differences discussed in points 1 and 3 of Scheme 1, are essentially enthalpic effects, solvent effects are free energy effects, i.e. they may contain strong entropic contributions. Reorientation of water molecules is for instance the origin of the shielding of all electrostatic interactions in the medium, including protein-cluster interactions. It should be apparent from the above discussion that the problem of proteininduced effects is a difficult one, both from a theoretical and experimental point of view. Experimentally, all effects sum up, and it is particularly challenging to
Coordination Sphere Versus Protein Environments
137
factorize out one from the other. A meaningful example is provided by the [Fe4S4] clusters. As shown in Fig. 2, and discussed in Sect. 2, this cluster type is present in two main classes of proteins, and the redox properties in the two classes are strikingly different. Solid state and solution structures of several proteins from both classes are available, so that a critical examination of the origin of protein-induced variability is in principle possible. Two representative structures of members of each class are shown in Fig. 4 and Fig. 5. It appears that, although the geometry of the cluster itself is the same in the two classes, within experimental error, differences are clearly present already in the Cfl-S-Fe-S dihedral angles (Table 5). In particular, three out of four such angles in HiPIPs, but only two out of four in ferredoxins, are close to 60°. These differences may influence the Fe-S interactions and therefore affect point 1. From the available sequences and the overall folding of the proteins, points 2 and 4 - 5 of Scheme 1 are also likely to play a role. Differences in point 3 are difficult to assess. In any case, a direct comparison of the two classes promises to prove a difficult task. This will be tackled later. It should be an easier task to attempt an analysis of the variability within a single class of proteins. In this case, HiPIPs are the class of choice, as the E°" range covered within the class is relatively large, and structural information is available for several members of the class. The range is from 90 to 460 mV (Table 6). Fig. 6 shows a superposition of protein backbones in the vicinity of the cluster for the four known HiPIP X-ray structures [92, 40-42]. The superposition is striking, and shows that the effect of differences in point 1 within the series is likely to be small. For the same reasons, differences in reorganization
Fig. 4. Structure of the HiPIP from Rc. tenuis [41] drawn using the MOLMOL program [138]
138
E Capozzi • S. Ciurli • C. Luchinat
Fig. 5. Structure of the [8Fe-8S] ferredoxin from C. acidiurici [53] drawn using the MOLMOL program [138]
Table 5. Cfl-S-Fe-S Dihedral angles a in [Fe4S4] clusters of a representative HiPIP[41] and fer-
redoxin [53] Cfl-S Gys(XX)
Fe-S(A)
HiPIP (R. tenuis, model A) [IISU] 22 [E~] 25 - 60.2 40 ~ 55 - 66.8 HiPIP (R. tenuis, model B) 22 ~ 25 40 55
Fe-S(B)
92.3 70.5
Fe-S(C)
144.8 177.0 169.4 178.9
94.2 60.4 67.2 67.9
142.0 179.7 175.1 177.8
Ferredoxin (C. acidi-urici, cluster 1) [ 1FCA] 8 - 64.7 [5~ 11 ~ 65.2 14 - 96.1 47 ~ 92.3
178.4 138.1 150.0
Ferredoxin (C. acidi-urici, cluster 2) 18 [Z~5~ 37 - 64.9 40 - 73.9 43 - 87.4
148.8 176.8 171.4 148.5
89.9
175.7
a The dihedral angle that is smallest in absolute value is highlighted.
CoordinationSphereVersusProtein Environments
139 .n
0
0 ~J
o.
0
~.
c~
LJ 0 0 ~
0
~
0 0 0
0
0
O 0
~n3
~
~0 ~J
I
•
I
I
~
I
I
0 0 0
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
o
I
0
¢)
o 0
oh
o
~0
o
m
~J 0 v)
~
~
~
=~
o
~ . ~
u~LJ
140
E C a p o z z i • S. C i u r l i • C. L u c h i n a t
I
I
I
I I I I I I I I I I I I I I I I l l l ~ l ~ l l ~ l ~
u~uuuuuuwuwuuvuuuuuuuuuuu
Z
".~
~
Z
Coordination Sphere Versus Protein Environments
141
Fig. 6. Best superposition of the a-carbon backbone of HiPIPs from C. vinosum (yellow) [39], E. vacuolata iso-II (cyan) [42],E. halophila iso-I (green) [40], and Rc. tenuis (red) [41] determined using the WHATIFprogram [48].The [Fe4S4]cluster belonging to C. vinosum HiPIP is
also shown
energies within the series are likely to be minor. Solvent effects may be somewhat different, but in general the solvent exposure of the cluster is negligible and that of its ligands relatively small (Table 6), so solvent effects are expected to be modest on an absolute scale (see also later). We are thus left with point 2 and its various sub-cases, and point 5, as the more likely candidates. Electrostatic effects of type 2 are enthalpic effects and easy to calculate, whereas those of type 5 are not. Several theoretical approaches are available, differing mainly in the way the behavior of the solvent is treated. We have performed these calculations using two different methods [94, 95], and the results are strikingly similar: they clearly point to the effect of unit charges as the major determinant for the differences. This is illustrated in Fig. 7, where the calculated contributions to E°' are plotted versus the experimental ones for a series of HiPIPs. Although the correlation is far from being perfect, all other contributions except that of unit charges, show only minor variations within the series. The solvent contribution may not be negligible in all cases although it may be difficult to account for satisfactorily (further discussion on this point is found below). We conclude that the major determinant for the variation of E°' within the HiPIP series is the difference in charged residues within the series. To simplify things further, as the Coulombic interaction follows a l/r dependence (in the
142
F. Capozzi • S. Ciurli • C. Luchinat i
40
i
1100
Vac I1 lhd I I
Vac I
llal ii
Ten Gel C.Vin
•
20
850
600
[] []
350 ~
[]
[]
-20 [] D
-40
+ -60 0
+
+
+
100
+ +
I
I
i
I
I
I
I
I
50
100
150
200
250
300
350
400
-150
450
Eexp (mV) Fig. 7. Calculated vs. experimental reduction potentials for a series of HiPIPs ([]) and contributions to the electrostatic energy from unit charges (O), CONH dipoles ( i ) and solvation (+) [95] absence of polarization effects on the intervening medium), and since within the proteins the distances of the various charged residues (mostly on the surface) from the center of the cluster are within a factor of two, we expect E°' to correlate with the total charge of the protein. This is indeed the case (Fig. 8) [94], and the correlation is not appreciably worse than the one obtained with a quantitative estimate of the whole electrostatic field [94, 95]. Furthermore, E°" shows the expected dependence on ionic strength, the dependence being larger for the more charged and smaller for the less charged proteins (unpublished resuits from our laboratory). Finally, variations in one unit charge, either through pH titrations of histidines or through point mutations, always yield the predicted sign and magnitude of variation [94, 95]. We should point out that such variation is indeed small (ca 10-30 mV/unit charge), and the conclusions drawn on the present class, akhough valid, may be overridden in other classes by other effects when present. This situation will be encountered in most of the following examples. A rough correlation of E°" with protein unit charges could also be noted (Table 6) for the classes of rubredoxins (Fig. 9) and [2Fe-2S] ferredoxins (Fig. 10). For rubredoxins, some surface mutations which introduce or remove charges close to the metal site also seem to confirm the trend [54]. However, other mutations at the same sites show opposite behavior [54]. Clearly, the relationship of E°' with unit charges in this series is more labile. In rubredoxins the metal ion is also rather shielded from the solvent, but is more peripheral than the [Fe4S4] cluster in HiPIPs and two of the four cysteines are rather exposed
143
Coordination Sphere Versus Protein Environments
10
[]
AO
Q;
0
ro
[]
0
.r,-i
I:I
0 []
-5 0J o
[] 0
I~ - 1 0 O ....
-15 0
I .... 1 O0
I .... 200
I .... 300
I .... 400
500
E°(mV) Fig. 8. Correlation o f the experimental reduction potential with the unit charges o f the protein for a series o f HiPIPs at p H 5 ([3), 7 (A), and 8 (O) [94]
Fig. 9. Structure o f the rubredoxin from P. furiosus [27] d r a w n using the MOLMOL p r o g r a m
[138]
144
F. Capozzi • S. Ciufli • C. Luchinat
Fig. 10. Structure of the [2Fe-2S] ferredoxin from Spirulina platensis [139] drawn using the MOLMOLprogram [138]
(Tables 6 and 7). Therefore, solvent effects may be expected to be relatively more important. On the other hand, both rubredoxins and [2Fe-2S] ferredoxins have a more elongated shape with respect to HiPIPs and the metal sits at one end (cfr. Fig. 4 and 5), so that the effects of unit charges at the other end of the protein are likely to be smaller. Interestingly, however, binding of a tripositive chromium(III) complex to plant-type ferredoxins increases their reduction potential by almost 200 mV, and even makes the Fe(II)-Fe(II) state accessible [96]. Most of the substitutions that seemingly negate the importance of charges in rubredoxin are on the bulky side-chain of Val 8 [54]. It could be that less bulky side-chains increase the solvent accessibility of the metal. Since, in these systems, reduction implies a change of the metal site charge from - 1 to -2, i. e. an increase in absolute value, an increased solvent exposure should stabilize more the reduced state, and thus increase the reduction potential. This working hypothesis is currently being tested using various structural and spectroscopic techniques. Proteins containing the [Fe3S4] and [Fe4S4] clusters are usually smaller than HiPIPs and ferredoxins, and the clusters are strongly exposed to the solvent (Figs 5, 11, 12 and 13). Although the variations in unit charges within the various classes are sizable (Table 6), the variations in reduction potentials are often smaller and do not correlate with charges. Apparently, increased solvent accessibility has a two-fold effect of i) decreasing the sensitivity of the E°' va-
Coordination SphereVersus Protein Environments
145
Fig. 11. Structure of the [3Fe-4S] ferredoxin from Desulfovibrio gigas [35] drawn using the MOLMOL program [ 138]
Fig. 12. Structure of the [4Fe-4S] ferredoxin from Bacillus thermoproteolyticus [80] drawn using the MOLMOL program [138]
146
E Capozzi • S. Ciurli • C. Luchinat
Fig. 13. Structure of the [7Fe-8S] ferredoxin from Azotobacter vinelandii [77] drawn using the MOLMOLprogram [138]
lues to both unit and fractional protein charges, because of the charge-shielding effect of water, and ii) making the E°' values more sensitive to solvent exposure itself, and therefore to the even minor variations in solvent exposure from one protein or another within the same class. HiPIPs and ferredoxins, taken together, are the most striking example of protein-induced effects. The effects are such that proteins belonging to the two classes have different redox states accessible, the HiPIPs using the [Fe4S413+! [Fe4S4 ] 2+ pair and the ferredoxins using the [Fe4S4] 2+/[Fe4S4] + pair. As anticipated in Sect. 2, the E°' values are in the + 90/+ 460 mV range for the former and in the -450/-650 mV range for the latter (Table 6). To some extent, however, the large difference between these two sets of values may give a misleading picture. One should actually compare the E°' values for the same reduction step, either the first or the second, for both classes. Until recently, this has not been possible. Attempts to oxidize ferredoxins to the [Fe4S4] 3+ state produce loss of one or of all four irons in the cluster [97], while reduction of HiPIPs to the [Fe4S4] + state could only be achieved under denaturing conditions [98]. In the absence of experimental data, computational attempts have been made to identify the major determinants for the difference in behavior. Using the same arguments as for rubredoxins and [2Fe-2S] ferredoxins, a larger solvent exposure is expected to increase the reduction potential of the [Fe4S4] ferredoxin clusters with respect to HiPIPs (Table 6). This effect is in the right direction, since the higher reduc-
CoordinationSphereVersusProteinEnvironments
147
tion potentials of HiPlPs is referred to the preceding reduction step. An extensive study was performed using the Protein Dipoles Langevin Dipoles method [99], which takes into account points 2, 4, and 5 in Scheme 1, and has recently been improved by considering molecular dynamics-averaged structures [20]. The calculations yielded reduction potentials for proteins of the two classes (P. aerogenes and A. vinelandii ferredoxins on one hand and C. vinosum HiPIP on the other) differing by about 1.2 V, C. vinosum being lower. The difference was mainly attributed not to solvent effects but to the much larger number and proximity of CONH peptide dipoles oriented with the positive end toward the cluster in ferredoxins [99, 20]. Indeed, this difference is also apparent from column C. C. S. of Table 6 which simply shows the sum of the electrostatic effects of the fractional charges of the above atoms. Recently, Hagen et al. succeeded in measuring the reduction potential of the second reduction step of a HiPIP under non-denaturing conditions [100]. The HiPIP was chosen as th e one displaying the highest reduction potential for the first step (R. globiformis + 460 mV) [90, 89], with the reasonable hope that the second step would also be the highest and thus the most accessible. A value of -600 to -900 mV was measured, depending on the method. Such a value is only 300-400 mV lower than the E°' of normal ferredoxins, instead of being 1.2 V lower as suggested by calculations [99, 20]. This assigns to the protein matrix an easier task to achieve differentiation of the two redox pairs. According to these data, the two reduction steps in [Fe4S4] clusters are separated by not much more than one volt rather than by two volts. Such a difference is also more in line with the lessons from model clusters, where some success has been obtained in providing ligand environments which stabilize one or the other oxidation state. Whether the 300-400 mV difference between ferredoxins and HiPIPs should be mainly ascribed either to CONH dipoles as suggested by calculations [99], or to solvent effects or other factors, remains to be seen. Unit charges seem to play a minor role here, at variance with HiPIPs [94]. The smaller number of hydrogen bonds between main-chain amide NH groups and terminal or bridging sulfur atoms of the cluster found in HiPIPs (5) relative to ferredoxins (8) can also contribute to the difference in reduction potential and stabilization of the higher electron density in the cluster core associated with the reduced state [140]. This hypothesis was indeed the first attempt to rationalize the difference in reduction potentials between HiPIP and Fd, this approach being subsequently generalized, including the effect of the more distant CONH dipoles. The question then arises as to why a ferredoxin with E°' of, for example, -650 mV [79] could not be superoxidized, as its further reduction potential is expected to be around or only slightly above + 0.5 V. This leads to an extention of our considerations of the chemical stability of the clusters in the various oxidation states. It is well known that [Fe4S4]3+ model clusters are most stable in non-polar solvents and when the ligands are bulky enough to shield the cluster itself from the solvent; otherwise, solvolysis occurs [45]. The loss of one iron atom from ferredoxins upon addition of ferricyanide [97, 101] can be ascribed to the same phenomenon, aided by the driving force of precipitation of polymeric iron cyanide species. Conversely, it has been shown that HiPIPs can stand extensive unfolding in guanidinium hydrochloride solutions when
148
E Capozzi • S. Ciurli • C. Luchinat
in the reduced state, but not when in the oxidized state [102]. The overall picture that results from these observations is that HiPIPs and ferredoxins differ less than often thought in intrinsic redox properties, and more in the stability of the 3 + state, which dramatically depends on the shielding of the cluster from the solvent. At the end of this section a comment is due on the role of the various possible determinants of the redox properties of iron-sulfur proteins [21]. We have seen that, depending on the system, one or another effect may be the leading one. This may depend on the absence or constancy of the other effects, and also on the different sensitivity of each effect to other parameters. One additional parameter to be considered is the size of the protein [103]. The larger the protein, and the more buried the redox center, the more likely are partial protein charges to play a leading role, while the role of solvent and of unit charges of acidic and basic residues on the surface are less important. The smaller the protein, the more important are the unit surface charges and the solvent. The closer the redox center to the surface, the less important are surface charges with respect to direct solvent effects on the redox center. Coordination sphere effects are likely to be size-independent.
4 Electronic Structures and Microscopic Reduction Potentials From the above discussion it appears that, although the general principles underlying the overall effect of the protein on the reduction potential of the iron-sulfur center are well understood, the relative importance of these effects in the various cases is seldom clearcut. Additional experimental information would be desirable. One piece of such information could be the knowledge of the microscopic reduction potentials of the individual iron ions in di-, tri- and tetrametallic systems. If available, differential protein and solvent effects could be considered in the same system, instead of limiting the comparison to series of homologous systems. For coordinatively symmetric clusters, this information tells directly us how much the effect of the protein and solvent can differentiate between two or more otherwise equal reduction potentials. Determination of microscopic reduction potentials is not an easy task. First, it should be established whether microscopic reduction potentials can be defined at all. This depends on the fate of the additional electron entering the system upon reduction. It can either localize over one iron ion of the cluster, delocalize over all iron centers or, in intermediate cases, delocalize over only some of the iron ions. Even in the case of localized behavior, the actual measurement of individual E°' values may not be easy. Sometimes, spectroscopic rather than electrochemical methods are more suitable for this purpose. The presence of partial or total delocalization can also be assessed by spectroscopic methods, notably taking advantage of the relationship between individual oxidation states and individual spin states of the iron ions. This latter theme has been the subject of intensive research in the past years [104]. [2Fe-2S] ferredoxins cycle between the [Fe2S2]2+ and the [Fe2S2]÷ states. The [Fe2S2] ° s t a t e is hardly accessible. Asymmetry in the electron distribution may
149
Coordination Sphere Versus Protein Environments
s \_,,~ .............. ,,,/ l~eA
/
Fe,B
\s /
E~
\ ~x~E~
\ F "'''''' S ........ iw/ /Fe#k / e% s
Kiso ~
\ _ m,,,...S ........ iV' l-eA Fe~ / \S / \
14. Redox equilibria between the oxidized form of a [2Fe-2S] ferredoxin and the two electronic isomers of its reduced form Fig.
be expected only in the [Fe2S2]+ state. This asymmetry could be revealed by measuring the two reduction steps shown in Fig. 14, but this is obviously difficult because of the rapid equilibration that is likely to occur between the two reduced species. Estimates of the equilibrium constant between these two species, that have been recently termed electronic isomers [105], Kiso, can be obtained using NMR spectroscopy, taking advantage of the fact that signals originating from the two cysteines bound to one of the iron ions show paramagnetic effects on chemical shifts and temperature dependencies typical of an S=2 spin, and signals from the two cysteines bound to the other iron ion reflect their interaction with an S=5/2 ion [106, 107]. Since the two spins are antiferromagnetically coupled, the magnetic moment of the ion with smaller spin is aligned opposite to the external magnetic field in the ground state [108, 109]. If the ground state were the only populated state at the temperature of the NMR experiment, opposite signs of the hyperfine shifts should be observed, namely downfield shifts for the nuclei coupled to the ferric S=5/2 spin and upfleld shifts for those coupled to the ferrous S--2 spin. The observed effect is less dramatic (Fig. 15) [110], because higher spin levels of the coupled system are appreciably populated around room temperature, so that both sets of signals are actually shifted
a-d HI3 Cys Ill,IV
f-i HlO Cys I,II
a c,d
r
ppm ....................
30
~'"- -.,
L a
1
~
j.
e
..e
,40
130 ............."
IV
fg h~ k)
..~"""2;~:2".:::~... f
20 .............~"-- h b
,20
. --'"""
........................ k t0
cd "~"~" ppm
40
"*'~"~ 30
~'-..,~ 20
10
0
110
-" " " " ' " ' ' " " 3.2
0 3.4
3.0
3.2
3.4
1000~(1/K) a
b
Fig. 15. XHNMR spectrum of the reduced form of the [2Fe-2S] ferredoxin from spinach and temperature dependence of the hyperfine shifted signals [106]
150
E Capozzi • S. Ciurli • C. Luchinat
downfield. However, those associated with the ferric ion (signals a-d) are far downfield, while those associated with the ferrous ion (signals f-i) are shifted much less downfield and, more importantly, tend to move upfield with decreasing temperature, the opposite of what one would expect for nuclei interacting with a normal paramagnet. The normal temperature dependence is termed of Curie type, and the anomalous one displayed by the smaller spin is said to be of anti-Curie type [ 108]. NMR spectroscopy thus unravels which set of signals (in a sequence-specific sense) belongs to which iron ion (in an oxidation state sense) [106, 111], and at the same time demonstrates that i) the valencies are localized and one of the two electronic isomers prevails, otherwise both sets of signals would show the same intermediate behavior [104], and ii) the iron which becomes reduced is the one bound to Cys I and Cys II (FeA), while the one bound to Cys III and Cys IV (FEB) remains essentially ferric in the reduced state too [111]. Fe A is the ion more exposed to solvent. The above pattern seems to be a general feature of [2Fe-2S] ferredoxins [104]. From the estimates of the equilibrium constant, Kiso, [112, 104] a difference between the two reduction potentials of the order of 50-100 mV can be predicted. Attempts have been made to account for E °' differences of this order of magnitude using the same computational approaches as those used to analyze reduction potential trends within series of proteins. Computations have been successful in predicting which iron ion is more reducible, and also the order of magnitude of the E°' difference with the other ion (Table 8) [95]. According to these calculations, the major determinant seems to be the differential exposure to solvent (mostly through Cys I, see Table 7) of FeA which apparently stabilizes the larger negative charge of the ferrous FeAS4 center better. Differences in unit or fractional protein charges seem to play a role in the same direction (Table 6 and Fig. 16). A modest contribution may also arise from the slight differences in the coordination sphere, notably differences in S-Fe-S Solvent accessibility to cluster bound cysteines for some representative iron-sulfur proteins a Table 7,
Protein
Cys Acc. I (~2)
Gys Acc. II (~2)
Gys Acc. III (~2)
Gys Acc. IV (~2)
P. furiosus Rdx H. marismortui 2Fe-Fdx D. gigas 3Fe-Fdx C. acidi-urici 8Fe-Fdx (cl. 1) C. acidi-urici 8Fe-Fdx (cL 2) E. halophila HiPIP I (rood. A) E. halophila HiPIP I (rood. B) C. vinosum HiPIP E. vacuolata HiPIP II R. tenuis HiPIP (rood. A) R. tenuis HiPIP (mod. B)
6 63 8 8 18 31 31 43 34 22 22
9 68 14 11 37 34 34 46 37 25 25
39 71 50 14 40 48 48 63 51 40 40
42 102
14.5 1.8
47 43 64 64 77 65 55 55
7.2 3.0 0.0 0.0 0.0 0.0 0.0 3.0
0.2 13.2 7.0 6.6 6.7 0.0 0.0 0.0 0.0 0.0 0.0
a Obtained with the program WHATIF [48].
17.1 0.3 3.3 21.1 7.5 1.3 1.5 0.2 0.0 1.0 0.6
0.4 0.8 5.4 3.4 20.4 2.9 2.8 1.9 1.8 3.0 2.9
151
Coordination Sphere Versus Protein Environments . . . . . . .
o
.
.
.
.
.
.
.
.
. o.
.
.
.
.
,
g
~'~ ~ ~,~
~-- o k'3
M-I
1',3 I
0,1 I
~I
0" ~ ~ I
~I
I'0 '
..
0
~
G3
I
I
',~-
1"3
~
~0-o ~
.
I "'~ 04
~
0
r o
I ~
I
I~'0 0
'M0
1"3 0
0 04
..
-o ~0
00
~0
~1 0
r'O 0
0
C'~l 1"3
~ "-G ~ ~ 0 -~ ~,.~ 4,~
O2 ~
~.~
•~ ;~ &.~N ~ ~
~.~.~
.~
.......
....
4"t
o2
, ^ [.~ • I
~'~
~
~
o2
o2
~.~
~
"~
o2
~
N N ~N~
angles and distances and S-Fe-S-Cfl dihedral angles [113]. Even in a symmetric model cluster, a tendency to unsymmetric electron localization has been observed, although not as marked as in the protein system [114, 115]. This type of detailed analysis is not available for [2Fe-2S] proteins of the Rieske type, essentially due to the lack of estimates of the equilibrium constant between the two electronic isomers. However, other spectroscopic evidence points to the iron bound to the two nitrogens as the more reducible one. In this case the difference is probably mostly due to coordination sphere effects, and' protein-induced effects, although undoubtedly present, will be more difficult to
152
F. Capozzi • S. Ciurli • C. Luchinat
Table8. Calculated reduction potential and variation in free energy between the reduced and the oxidized form for a [Fe2S2]2+/1÷moiety in S. platensis in the case of reduction of FeAor FeB. The values are calculated with Eprotei n = 4
AAGsolv (kJ/mol) AGch (kJ/mol) AGdip (kJ/mol) AGred (kJ/mol) AE (mV)
FeA
FeB
AAB
75.5 21.6 - 25.6 71.6 - 400
82.8 22.5 - 24.4 80.9 - 497.1
- 7.3 - 0.9 - 1.2 - 9.3 97.1
assess. Recent crystallographic data [116] have shown that the two histidines bound to the more reducible iron are exposed to solvent, and their NH protons are titratable and constitute a further source of modulation of redox properties. As for [2Fe-2S] ferredoxins, [Fe3S4] clusters can also show asymmetry only in the reduced state. The latter formally contains one ferrous ion, while the oxidized state contains all ferric ions. Mossbauer evidence shows that the extra electron in the reduced state is delocalized over one pair, which thus contains two iron ions with an average oxidation state of +2.5, while the third iron ion is essentially ferric [ 117, 118]. According to simple spin-coupling theories in these types of systems, the pair bearing the extra electron undergoes the phenomenon of double exchange, which results in a reduction of the effective antiferromagnetic coupling constant, J, over that pair [ 119]. It may be speculated that the mixed valence pair may actually form preferentially over the two iron ions having already the smallest J coupling in the oxidized state. The corresponding cysteines have been recently identified as Cys III and Cys IV from NMR experiments [120, 121]. In the absence of direct sequence-specific information on which cysteines are bound to the mixed valence pair in the reduced systems, one could provisionally suggest that the mixed valence pair forms on the iron ions bound to Cys III and IV. No particular preference for Cys III and IV as the preferred reduction site appears from inspection of Fig. 16 or Table 7. Direct NMR evidence is available on the identity of the most abundant electronic isomers in reduced ferredoxins containing the [Fe4S4] + cluster, as well as in oxidized HiPIPs containing the [Fe4S4] 9+ cluster. In both cases, experimental data and theoretical considerations point to a predominant electronic isomer containing one mixed valence pair and one ferrous (in ferredoxins) or one ferric (in HiPIPs) pair. The [Fe4S4] 2+ cluster, present in both oxidized ferredoxins and reduced HiPIPs, always shows a highly symmetrical distribution of the excess electrons, yielding four essentially equivalent Fe 2-5÷ions [122, 104]. The assignment of the proton NMR signals of the cysteine protons coordinating the two clusters in the reduced ferredoxin from C. pasteurianum could be achieved by EXSY experiments [123], which allowed a 1:1 relationship to be established with the assigned signals in the oxidized form [124]. The experiments also yielded a reasonable estimate of the small difference in reduction potential between the two clusters (14 mV) [123]. This small difference is expected because the clostridial-type ferredoxins originate from gene duplica-
Coordination Sphere Versus Protein Environments
153
tion, and the two clusters are in very similar secondary structure environments. With the assignment of the ligand proton signals of the reduced form in hand, advantage can be taken of the differential paramagnetic effects of the mixed valence pair with respect to the ferrous pair. The former has a larger subspin than the latter (S=9/2 vs. S=4) and, similarly to the behavior described above for [2Fe-2S] systems, the antiferromagnetic coupling between the two pairs forces the magnetic moment of the pair with smaller S to align opposite to the field at low temperature [104]. The temperature dependence of the proton signals arising from the ferrous pair is therefore of anti-Curie type, while that of the proton signals arising from the mixed valence pair is of Curie type [123]. From this analysis, it appears that for both clusters the extra electron upon reduction goes preferentially on the iron ions coordinated to Cys II and IV [124]. Cys II is by far the most solvent-exposed cysteine residue in all clostridial ferredoxins (Table 7). Even more detailed information is available for HiPIPs, which have recently been shown to be involved in the photo-induced electron transfer in bacterial photosynthesis [125, 126]. Oxidized HiPIPs again show evidence of the extra electron (referred to an all-ferric iron cluster) being delocalized over one pair [127]. Mossbauer data indicate that the mixed valence pair has a larger subspin than the ferric pair [128]. S values of 9/2 for the mixed valence and 4 for the ferric pairs have been proposed [129], together with values of 7/2 and 3, respectively [130]. A recent extension of the theoretical model that describes the cluster in terms of resonance between less symmetric spin-pairing situations suggests that the ground state is actually a mixture of 9/2,4 and 7/2,4 wavefunctions [131,132], the former having a larger weight. Independent of the detailed ground state wavefunction, the interpretation of the proton NMR data for oxidized HiPIPs is straightforward. By taking the particular HiPIP II from E. halophila as an example (Fig. 17), the set of four downfield signals (A-D) is clearly attributable to protons interacting with the mixed valence pair (larger spin) and the set of four upfield signals (W-Z) to protons interacting with the ferric pair (smaller spin) [133, 134]. At variance with the two-iron and four-iron ferredoxin cases, the signals of the protons interacting with the smaller spin are not downfield with an anti-Curie temperature dependence but upfield with a typically strong Curie-type temperature dependence which has been termed pseudoCurie [104]. The observation of upfield signals is apparently due to a larger population of the ground spin state of the coupled system at room temperature, in turn due to a larger J value, when compared with, for instance, plant-type [2Fe2S] ferredoxins. The sequence-specific assignment of the above signals allows us to identify the mixed valence pair as the one bound to Cys II and Cys III, and the ferric pair as the one bound to Cys I and Cys IV (Fig. 17). Table 7 shows that Cys II and Cys III are the most exposed to solvent, while Fig. 16 shows that Cys III is also more stabilized by favorable orientation of CONH groups. This is confirmed by calculations performed using the same approach as that described for [2Fe-2S] systems. The extra electron characterizing the mixed-valence pair has in turn been placed on each of the six possible pairs in the HiPIP cubane. The calculations correctly yield the pair of iron ions bound to Cys II and Cys III as the most reducible [95]. This result is noteworthy,
154
F. Capozzi • S. Ciurli • C. Luchinat A H{32 Cys Ill B H{32 Cys II C H{31 Cys III D H{B1 Cys II
W X Y Z
H}~2 Cys IV HL31Cys IV H{32Cys I H!31 Cys I
9O
80
B
60! C
A
A
1oo
8 (ppm)
8'o
6'o
4o 2b 8 (ppm)
d
-2'0
15 IC
-~0
-10
B
140[
,
,
,
3,3 3.4 3.5 1000/T (l/K)
,
3.6
Fig. 17. 1H NMR spectrum of the oxidized HiPIP iso-II from Ectothiorhodospira halophila and temperature dependence of up- and downfield-shiftedsignals [133, 134]
although the calculated difference in microscopic reduction potential is too small by far, with respect to the lower limit that can be estimated from the NMR results. NMR data on the other HiPIPs are less straightforward to interpret. Only two signals are observed upfield, while six are downfield. However, two of them have an anti-Curie temperature dependence. By analogy with the [2Fe-2S] ferredoxins, the latter are associated with the smaller spin of the ferric pair like the two upfield signals. Among all the HiPIPs investigated, Cys I is always bound to a ferric iron, Cys III to a mixed-valence iron, while Cys II and IV seem to share different amounts of ferric and mixed-valence iron from one protein to another. Qualitative considerations from Table 7 and Fig. 16 are in agreement with this finding. This behavior has been rationalized in terms of the electronic isomer equilibrium shown in Fig. 18. The shifts and temperature dependences of the protons belonging to Cys II and Cys IV reflect the position of the equilibrium for the various HiPIPs (Fig. 19). Such equilibrium is about 45/55 % in the HiPIP from C. vinosum (i. e. Cys IV is about 45 % ferric). Again, calculations show that two out of the six pairs that show mixed valence character give the lowest energy when bearing the extra electron [95]. As noted above the iron ion which is always ferric is the farthest from the protein surface and the one in the most hydrophobic environment, while the iron ion which is always mixed valence is the most solvent-exposed and stabilized by CONH groups (Table 7 and Fig. 16). All the above considerations regarding the microscopic reduction potentials of iron-sulfur clusters rely on the hypothesis that spins are related to charges. Indeed, spectroscopy yields information on the spin quantum numbers associated to individual iron ions or to iron pairs, and we translate this information
155
Coordination Sphere Versus Protein Environments
S
I
I V Fe+~Fe~iZis+i( I
I
S
Fe3+
rLe S I
Kiso
E~+eB
AII E~.~+e
Fe3+
II
I
S
Fe25+
i'%L = II Fig. 18. Redox equilibria between two of the six electronic isomers of the oxidized form of a HiPIP and its reduced form
into (average) oxidation numbers. This has been successful for both [2Fe-2S] and [4Fe-4S] systems, while [3Fe-4S] systems should await the availability of NMR data for the reduced form. Of course, it would be desirable to test the predictive power of our approach, for instance by site-directed mutagenesis. This is not straightforward, since one should design a mutant that selectively affects the charge on one iron ion without introducing appreciable geometric distortions or other effects that may also influence the charges of the other iron ions. Then, the consequences of this change of charge should become measurable as changes in the spin distribution or, equivalently, in the mixture of electronic isomers. The HiPIP from C. vinosurn which is characterized by a 45/55 % mixture of electronic isomers in the oxidized form (Fig. 18 and Fig. 19) was chosen. A Cys-~ Set mutation was planned at one of the two iron sites (Cys II or Cys IV) whose valence depends on the position of the electronic isomer equilibrium. The C77S mutant (Cys IV=Cys77) was found to be stable [135], and its structure, particularly in the neighborhood of the cluster, was essentially the same as that of the wild-type protein [136]. However, the 1H NMR spectra of the mutant clearly showed a reversal of the temperature dependence of the cysteine proton signals of Cys II and Cys IV, the former indicating a larger share of mixed valence character and the latter a larger share of ferric character of the associated iron ion. This corresponds approximately to a transition from spectrum C to spectrum B in Fig. 19. A change in the electronic isomer mixture from 45/55 to about 60/40% could be estimated [135]. The change may not seem large but it is significant, as it involves a reversal of the order of the microscopic reduction potentials of the iron ions associated to Cys II and Cys IV. In our view, there is now sufficient evidence to at least qualitatively validate the postulated relationship between spins and charges in iron-sulfur systems. It would be desirable to assess whether a relationship could be established between spins and cluster distortions, especially in the case of [Fe4S4] clusters.
156
E Capozzi • S. Ciurli • C. Luchinat I S
I
Fe3+
/ iS
~7
Fe3+
II
A
II
III ]I
IV I
AAAA B
Ill
II
AA C
IV
I
AA AA, AA 0
i ~
III
AA III
,AAAA
I
ppm
AA AA, AA IV
AAAA
I
, AAAA
0
ppm
Fig. 19. Schematic representation of the 1H NMR hyperfine-shifted signals of the four cysteine fl-CH2 pairs of oxidized HiPIPs observed as a function of the position of the electronic isomer equilibrium. A: 100% to the left (E. halophila iso-II HiPIP); B: 60% to the left; C: 60% to the right; D: 100% to the right
From the high-resolution X-ray structures of the model compounds discussed in Section 2 we learn that the only [Fe4S4]3÷ model available, and most of the [Fe4S4]2+ models show compression along a binary axis, while most [Fe4S4]+ clusters do not show appreciable distortions. In terms of Heisenberg exchange coupling, a compressed cubane would lead to two smaller and four larger J values, the smaller ones associated with the mixed valence and ferric pairs. According to one theoretical model, this could be the case for [Fe4S4]3+ clusters [132]. [Fe4S4]2+ clusters could also be described satisfactorily within this frame [137]. At present, this is only a working hypothesis which needs more experimental data to be tested. In particular, high resolution X-ray data on protein systems would be desirable to test whether the electronic isomer distributions found from spectroscopic data may also arise from small geometric distortions imposed by the protein, in addition to electrostatic contributions. Acknowledgements. We thank Ivano Bertini for discussion and critical reading of the manuscript, and Robert A. Scott and Jean-Marc Moulis for sharing with us unpublished data.
Coordination Sphere Versus Protein Environments
157
5 References 1. Bertini I, Gray HB, Lippard SJ, Valentine JS, Eds., Bioinorganic Chemistry (University Science Books, Mill Valley, CA, 1994) 2. Adman ET (1985) Structure and function of blue copper proteins, Macmillan, New York 3. Adman ET (1991) Adv Protein Chem 42:144 4. Solomon El (1981) Copper proteins, John Wiley & Sons, New York 5. Moore GR, Pettigrew GW (1990) Cytochrome c: evolutionary, structural and physicochemical aspects, Springer-Verlag, Berlin 6. Scott RA, Mauk AG (1996) Cytochrome c. A multidisciplinary approach, University Science Books, Sausalito, CA 7. Banci L, Rosato A, Turano P (1996) JBIC 1:364 8. Goodin DB (1996) JBIC 1:360 9. Gross Z (1996) JBIC 1:368 10. Poulos TL (1996) JBIC 1:356 11. Rietjens IMCM, Osman AM, Veeger C, Zakharieva O, Antony J, Grodzicki M, Trautwein AX (1996) JBIC 1:372 12. Weiss R, Mandon D, Wolter T, Trautwein AX, Muther M, Bill E, Gold A, Jayaraj K, Terner J (1996) JBIC 1:377 13. Mathews FS (1985) Prog Biophys Mol Bio145:1 14. Louro RO, Catarino T, Salgueiro CA, LeGall J, Xavier AV (1996) JBIC 1:34 15. Sharp K, Honig B (1990)Ann Rev Biophys Biophys Chem 19:301 16. Cammack R, Sykes AG, Eds., Iron sulfur proteins (Academic Press, San Diego, 1992) 17. Cammack R (1992) Adv Inorg Chem 38:281 18. Gunner MR, Honig B, in Cytochrome c Sourcebook Scott RA, Mauk AG, Eds. (University Science Books, Mill Valley, CA, 1996) pp. 347 19. Gunner MR, Nicholls A, Honig B (1996) J Phys Chem 100:4277 20. Stephens PJ, Jollie DR, Warshel A (1996) Chem Rev 96: 2491 21. Commentary Articles (1997) JBIC 2:108 22. Berg JM, Holm RH (1982) Iron sulfur proteins, Wiley Interscience, New York 23. Zhou J, Hu Z, Munck E, Holm RH (1996) J Am Chem Soc 118:1966 24. Watenpaugh KD, Sieker LC, Jensen LH (1980) J Mol Biol 138:615 25. Frey M, Sieker LC, Payan F, Haser R, Bruschi M, Pepe G, LeGall J (1987) J Mol Biol 197: 525 26. Adman ET, Sieker LC, Jensen LH (1991) J Mol Biol 217:337 27. Day MW, Hsu BT, Joshua-Tor L, Park J-B, Zhou ZH, Adams MWW, Rees DC (1992) Protein Sci 1 : 1494 28. Lane RW, Ibers JA, Frankel RB, Papaefthymiou GC, Holm RH (1977) J Am Chem Soc 99:84 29. Tsukihara T, Fukuyama K, Nakamura M, Katsube Y, Tanaka N, Kakudo M, Wada K, Hase T, Matsubara H (1981) J Biochem 90:1763 30. Tsukihara T, Fukuyama K, Mizushima M, Harioka T, Kusunoki M, Katsube Y, Hase T, Matsubara H (1990) J Mol Biol 216:399 31. Jacobson BL, Chae YK, Markley JL, Rayment I, Holden HM (1993) Biochemistry 32: 6788 32. Frolow F, Harel M, Sussman JL, Mevarech M, Shoham M (1996) Nat Struct Biol 3:452 33. Mayerle JJ, Frankel RB, Holm RH, Ibers JA, Phillips WD, Weiher JF (1973) Proc Nail Acad Sci, USA 70:2429 34. Robbins AH, Stout CD (1989) Proteins: Structure, Function, and Genetics 5:289 35. Kissinger CR, Sieker LC, Adman ET, Jensen LH (1991) J Mol Biol 219:693 36. Stout CD (1993) J Biol Chem 268:25920 37. Sieker LC, Adman ET, Jensen LH (1972) Nature 235:40 38. Herskovitz T, Averill BA, Holm RH, Ibers JA, Phillips WD, Weiher JF (1972) Proc Natl Acad Sci USA 69: 2437
158
E Capozzi • S. Cinrli • C. Luchinat
39. Carter CWJ, Kraut J, Freer ST, Xuong N-H, Alden RA, Bartsch RG (1974) J Biol Chem 249:4212 40. Breiter DR, Meyer TE, Rayment I, Holden HM (1991) J Biol Chem 266:18660 41. Rayment I, Wesemberg G, Meyer TE, Cusanovich MA, Holden HM (1992) J Mol Biol 228: 672 42. Benning MM, Meyer TE, Rayment I, Holden HM (1994) Biochemistry 33:2476 43. Laskowski EJ, Frankel RB, Gillum WO, Papaefthymiou GC, Renaud J, Ibers JA, Holm RH (1978) J Am Chem Soc 100:5322 44. Reynolds JG, Laskowski EJ, Holm RH (1978) J Am Chem Soc 100:5315 45. O'Sullivan T, Millar MM (1985) J Am Chem Soc 107:4096 46. Czernuszewicz RS, Macor KA, Johnson MK, Gewirth A, Spiro TG (1987) J Am Chem Soc 109:7178 47. Hill CL, Renaud J, Holm RH, Mortenson LE (1977) J Am Chem Soc 99:2549 48. Vriend G (1990) J Molec Graph 8:52 49. Warshel A, Russell ST (1984) Quart Rev Biophys 17:283 50. Lee FS, Chu ZT, Warshel A (1993) J Comp Chem 14:161 51. Noodleman L, Case DA (1992) Adv Inorg Chem 38:424 52. Marcus RA, Sutin N (1985) Biochim Biophys Acta 811:265 53. Tranqui D, Jesior JC (1995) Acta Crystallogr 51 : 155 54. Zeng Q, Smith ET, Kurtz DM, Jr, Scott RA (1996) Inorg Chim Acta 242:245 55. SeryA, Housset D, Serre L, Bonicel J, Hatchikian C, Frey M, Roth M (1994) Biochemistry 33:15408 56. Shimizu F, Ogata M, Yagi T, Wakabayashi S, Matsubara H (1989) Biochimie 71 : 1171 57. Adams MWW (1992)Adv Inorg Chem 38:341 58. Hormel S, Walsh KA, Prickril BC, Titani K, Le Gall J, Sieker LC (1986) FEBS Lett 201:147 59. Stenkamp RE, Sieker LC, Jensen LH (1990) Proteins: Structure, Function, and Genetics 8:352 60. Kok M, Oldenhuis R, van der Linden MP, Meulenberg CH, Kingma J, Witholt B (1989) J Biol Chem 264:5442 61. Saeki K, Jain MK, Shen GJ, Prince RC, Zeikus JG (1989) J Bacteriol 171:4736 62. Lee WY, Brune DC, LoBrutto R, Blankenship RE (1995) Arch Biochem Biophys 318:80 63. Meyer TE, Przysiecki CT, Watkins JA, Bhattacharyya A, Simondsen RP, Cusanovich MA, Tollin G (1983) Proc Nail Acad Sci USA 80:6740 64. O'Dell SE, Kurtz DM, Jr, Scott RA, Smith ET (1993) J Inorg Biochem 43:18 65. Meyer J, Galliard J, Lutz M (1995) Biochem Biophys Res Commun 212:827 66. Woolley KJ, Meyer TE (1987) Eur J Biochem 163:161 67. Bohme H, Schrautemeier B (1987) Biochim Biophys Acta 891:1 68. Rypniewski WR, Breiter DR, Benning MM, Wesenberg G, Oh B-H, Markley JL, Rayment I, Holden HM (1991) Biochemistry 30:4126 69. Hurley JK, Salamon Z, Meyer TE, Fitch JC, Cusanovich MA, Marldey JL, Cheng H, Xia B, Chae YK, Medina M, Gomez-Moreno C, Tollin G (1993) Biochemistry 32:9346 70. Mason JR, Cammack R (1992) Ann Rev Microbiol 46:277 71. Pochapsky TC, Mei Ye X, Ratnaswamy G, Lyons TA (1994) Biochemistry 33:6424 72. Wada K, Masui R, Matsubara H, Rogers LJ (1988) Biochem J 252:571 73. Bottin H, Lagoutte B (1992) Biochim Biophys Acta 1101:48 74. Cammack R, Rao KK, Bargeron CP, Hutson KG, Andrew PW, Rogers LJ (1977) Biochem J 168:205 75. Stonbaugh NA, Sundquist JEI Burris RH, Orme-Johnson WH (1976) Biochemistry 15:2633 76. Armstrong FA, George SJ, Cammack R, Hatchikian EC, Thompson AJ (1989) Biochem J 264: 265 77. Stout CD (1989) J Mol Biol 205:545 78. Martin AE, Burgess BK, Stout CD, Cash VL, Dean DR, Jensen GM, Stephens PJ (1990) Proc Natl Acad Sci USA 87:598 79. Shen B, Jollie DR, Stout CD, Diller RC, Armstrong FA, Gorst CM, La Mar GN, Stephens PJ, Burgess K (1994) J Biol Chem 269(11):8564
CoordinationSphereVersusProteinEnvironments
80. 81. 82. 83. 84.
159
Fukuyama K, Matsubara H, Tsukihara T, Katsube Y (1989) J Mol Bio1210:383 Langen R, Jensen GM, Jacob U, Stephen PJ, Warshel A (1992) J Biol Chem 267:25625 Smith ET, Feinberg BA (1990) J Biol Chem 265 : 14371 Guerlesquin F, Moura JIG, Cammack R (1982) Biochim Biophys Acta 679:422 Sticht H, Wildegger G, Bentrop D, Darimont B, Sterner R, Roesch P (1996) Eur J Biochem 237: 726 85. Okawara N, Ogata M, Yagi T, Wakabayashi S, Matsubara H (1988) Biochimie 70: 86. Asso M, Mbarki O, Guigliarelli B, Yagi T, Bertrand P (1995) Biochem Biophys Res Commun 211 : 198 87. Adman ET, Sieker LC, Jensen LH (1976) J Biol Chem 251:3801 88. Lode ET, Murray CL, Rabinowitz JC (1976) J Biol Chem 251 : 1683 89. Luchinat C, Capozzi F, Borsari M, Battistuzzi G, Sola M (1994) Biochem Biophys Res Commun 203:436 90. Przysiecki CT, Meyer TE, Cusanovich MA (1985) Biochemistry 24: 2542 91. Van Driessche G, Ciurli S, Hochkoeppler A, Van Beeumen I (1997) Eur J Biochem 244:371 92. Freer ST,Alden RA, Carter CW, Jr, Kraut J (1975) J Biol Chem 250:46 93. Noodleman L, Norman JG, Osborne JH, Aizman A, Case DA (1985) J Am Chem Soc 107:3418 94. Banci L, Bertini I, Ciurli S, Luchinat C, Pierattelli R (1995) Inorg Chim Acta 240:251 95. Banci L, Bertini I, Gori Savellini G, Luchinat C (1996) Inorg Chem 35:4248 96. Im SC, Kohzuma T, McFarlane W, Gaillard J, Sykes AG (1997) Inorg Chem 36:1388 97. Thomson AJ, Robinson AE, Johnson MK, Cammack R, Rao KK, Hall DO (1981) Biochim Biophys Acta 637: 423 98. Cammack R (1973) Biochem Biophys Res Commun 54:548 99. Jensen GM, Warshel A, Stephen PJ (1994) Biochemistry 33:10911 100. Heering H& Y.B. B, Hagen WR, Meyer TE (1995) Eur J Biochem 232:811 101. Benelli B, Bertini I, Capozzi F, Luchinat C (1994) Gazz Chim Ital 124:469 102. Bertini I, Cowan JA, Luchinat C, Natarajan K, Piccioli M (1997) Biochemistry 36:9332 103. Zhou H-X (1997) JBIC 2:109 104. Bertini I, Ciurli S, Luchinat C (1995) Struct Bonding 83:1 105. Bertini I, Luchinat C (1996) ACS series 653:57 106. Banci L, Bertini I, Luchinat C (1990) Struct Bonding 72:113 107. Luchinat C, Ciurli S (1993) Biological Magnetic Resonance 12:357 108. Dunham WR, Palmer G, Sands RH, Bearden AJ (1971) Biochim Biophys Acta 253:373 109. Bertini I, Luchinat C (1996) NMR of paramagnetic substances, 150 Elsevier, Amsterdam 110. Bertini I, Lanini G, Luchinat C (1984) Inorg Chem 23:2729 111. Dugad LB, La Mar GN, Banci L, Bertini I (1990) Biochemistry 29:2263 112. Benini S, Ciurli S, Luchinat C (1995) Inorg Chem 34:417 113. Carloni P, Corongiu G (unpublished) 114. Ding X-Q, Bill E, Trautwein AX, WinNer H, Kostikas A, Papaefthymiou V, Simopoulos A, Beardwood P, Gibson JF (1993) J Chem Phys 99:6421 115. KrOckelM, Grodzicki M, Papaefthymiou V, Trautwein AX, Kostikas A (1996) JBIC 1: 173 116. Iwata S, Saynovits M, Link TA, Michel H (1996) Structure 4:in press 117. Emptage MH, Kent TA, Huynh BH, Rawlings J, Orme-Johnson WH, Munck E (1980) J Biol Chem 255:1793 118. Huynh BH, Moura JIG, Moura I, Kent TA, LeGall J, Xavier AV, Munck E (1980) J Biol Chem 255: 3242 119. Papaefthymiou V, Girerd J-J,Moura I, Moura JIG,Munck E (1987) J Am Chem Soc 109:4703 120. Aono S, Bertini I, Cowan JA, Luchinat C, Rosato A, Viezzoli MS (1996) JBIC 1:523 121. Bertini I, Dikiy A, Luchinat C, Macinai R, Viezzoli MS, Vincenzini M. Biochemistry 36:3570 122. Bertini I, Capozzi F, Luchinat C, Piccioli M (1993) Eur J Biochem 212:69 123. Bertini I, Briganti F, Luchinat C, Messori L, Monnanni R, Scozzafava A, Vallini G (1992) Eur J Biochem 204:831
160
E Capozzi • S. Ciurli • C. Luchinat
124. Bertini I, Capozzi F, Luchinat C, Piccioli M, Vila A (1994) J Am Chem Soc 116:651 125. Hochkoeppler A, Kofod P, Ferro G, Ciurli S (1995) Arch Biochem Biophys 322:313 126. Hochkoeppler A, Ciurli S, Zannoni D, Meyer TE, Cusanovich MA, Tollin G (1996) Proc Natl Acad Sci USA 93: 6998 127. Dickson DPE, Johnson CE, Cammack R, Evans MCW, Hall DO, Rao KK (1974) Biochem J 139:105 128. Middleton P, Dickson DPE, Johnson CE, Rush JD (1980) Eur J Biochem 104:289 129. Noodleman L (1988) Inorg Chem 27:3677 130. Gloux J, Gloux P, Lamotte B, Mouesca JM, Rius GJ (1994) J Am Chem Soc 116:1953 131. Belinskii MI, Bertini I, Galas O, Luchinat C (1995) Z Naturforsch 50a:75 132. Belinskii MI, Bertini I, Galas O, Luchinat C (1996) Inorg Chim Acta 243:91 133. Banci L, Bertini I, Briganti F, Luchinat C, Scozzafava A, Vicens Oliver M (1991) Inorg Chim Acta 180:171 134. Banci L, Bertini I, Capozzi F, Carloni P, Ciurli S, Luchinat C, Piccioli M (1993) J Am Chem Soc 115:3431 135. Babini E, Bertini I, BorsariM, Capozzi F, Dikiy A, Eltis LD, Luchinat C (1996) J Am Chem Soc 118:75 136; Bentrop D, Bertini I, Capozzi F, Dikiy A, Eltis LD, Luchinat C (1996) Biochemistry 35: 5928 137. Bertini I, Briganti F, Luchinat C, Scozzafava A, Sola M (1991) J Am Chem Soc 113:1237 138. Koradi R, Billeter M, Wuthrich K (1996) J Mol Graphics 14:51 139. Fnkuyama K, Hase T, Matsumoto S, Tsukihara T, Katsube Y (1980) Nature 286:522 140. Backes G, Mino Y, Loehr TM, Meyer TE, Cusanovich MA, Sweeney WV, Adman ET, Sanders-Loehr J (1991) J Am Chem Soc 113:2055
The Bio-lnorganic Chemistry of Tungsten W.R. Hagen 1'2,* and A . E Arendsen 1 ] Department of Biochemistry, Wageningen Agricultural University, 2 Department of Molecular Spectroscopy, University of Nijmegen, The Netherlands
E-mail:
[email protected]
A steady flow of reports over the last seven years on the molecular characterization of tungsten-containingproteins from a wide range of microorganisms has drastically changed our appreciation of tungsten in redox enzymes. Biological tungsten is not an odd remnant of evolution but a widespread, versatile catalytic entity for the activation of the carbonyl group both in carbon dioxide and in a broad spectrum of aldehydes and carboxylic acids. This review starts off with an outline of the boundary conditions for the biological use of tungsten dictated by inorganic chemistry. Subsequently, the possibilities for spectroscopy on tungsten proteins are reviewed. The molecular properties of tungsten enzymes are described and are related to their physiology and their mechanism of action. Gaps in our current knowledge of the bio-inorganic chemistry of tungsten are identified, and possible directions of future research are indicated.
Keywords:Tungsten, molybdenum,pterin, tungstoenzymes, aldehyde oxidoreductase. List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
162
1
162
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1 Tungsten is a C o m m o n Bioelement in Microorganisms
........
163
2
Chemistry of Tungsten and its Biological Implications
........
164
2.1 2.2 2.3 2.4
Tungsten Atomic Structure . . . . . . . . . . . . . . . . . . . . . . . . . Aqueous Chemistry of Tungsten . . . . . . . . . . . . . . . . . . . . . . Sequestering and Transport of Tungstate . . . . . . . . . . . . . . . . . Bio-Availability o f Tungsten . . . . . . . . . . . . . . . . . . . . . . . . .
165 166 168 169
3
Spectroscopy of Biological Tungsten
170
3.1 3.2 3.3 3.4 3.5 3.6
Elemental Analysis of Tungsten . . . . . . . . .............. Tungsten E P R S p e c t r o s c o p y . . . . . . . . . . . . . . . . . . . . . . . . . T u n g s t e n N M R Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . Tungsten M6ssbauer Spectroscopy . . . . . . . . . . . . . . . . . . . . . Tungsten X-ray Absorption Spectroscopy . . . . . . . . . . . . . . . . . Tungsten Optical Spectroscopies . . . . . . . . . . . . . . . . . . . . . .
...................
170 171 173 174 175 176
* Correspondence to: Department of Biochemistry, Dreijenlaan 3, NL-6703 HA Wageningen, The Netherlands. Structure and Bonding,Vol.90 © SpringerVerlag BerlinHeidelberg1998
162 4
W.R. Hagen. A.E Arendsen Microbiology of Tungsten . . . . . . . . . . . . . . . . . . . . . . . . . .
177
4.1 Reactions Catalyzed by Tungsten Enzymes . . . . . . . . . . . . . . . .
178
5
Biochemistry of Tungsten . . . . . . . . . . . . . . . . . . . . . . . . . .
179
5.1 5.2 5.3 5.4
Generic Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pterin Cofactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Crystal Structural Analysis . . . . . . . . . . . . . . . . . . . . . . . . . Redox Chemistry at the Active Center . . . . . . . . . . . . . . . . . . .
181 184 184 186
6
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
188
7
References . . . . . . .
189
............................
List of Abbreviations AOR CAR CEPT DMSOrd ENDOR EPR ESEEM ESR EXAFS FAD FDH FMDH FMF FOR GAP GAPOR INEPT MCD MFR NADH NMR OAT RR RTP XANES
aldehyde oxidoreductase carboxylic acid reductase coupled electron-proton transfer dimethyl sulfoxide reductase electron-nuclear double resonance electron paramagnetic resonance electron spin echo envelope modulation electron spin resonance extended X-ray absorption fine structure flavin adenine dinucleotide formate dehydrogenase formylmethanofuran dehydrogenase formylmethanofuran formaldehyde oxidoreductase glyceraldehyde-3-phosphate glyceraldehyde-3-phosphate oxidoreductase insensitive nucleus enhanced by polarization transfer magnetic circular dichroism methanofuran reduced nicotinamide-adenine dinucleotide nuclear magnetic resonance oxo-atom transfer resonance R e n a n red tungsten protein X-ray absorption near edge structure
1 Introduction The single overriding criterion for living cells to adapt to the use of an element is its bio-availability. This is a function of terrestrial abundance, geological dis-
The Bin-Inorganic Chemistry of Tungsten
163
tribution, solubility in water, and speciation. A first, crude impression of bioavailability is obtained by comparing abundance in seawater with abundance in rock [1]. For tungsten this ratio, < 10 -4, is unfavourable by at least two orders of magnitude compared to the value of 10-2 for its congener, molybdenum. This is in line with our knowledge that molybdenum, in apparent contrast to tungsten, is widely distributed in nature (including humans) as an essential trace element. There are, however, certain niches in nature for which this ratio of abundances is reversed in favor of tungsten. These are, for example, the submarine hydrothermal vents. Here we find the hyperthermophilic microorganisms [2] whose very rich tungsten biochemistry is only now being unveiled. Furthermore, the absolute amount of Mo and W in the earth's crust is comparable (N 1 ppm), and tungsten is constantly being redistributed by human activity. The most obvious route is of course via the disposal of innumerable light bulbs produced worldwide during this century. Another bulk application involves the tungsten carbide of tool tips. Also, WS2 is a catalyst in the oil industry. Redistribution in the form of tungstate as a minor component of fertilizers (cf. [3]) is perhaps less expected, but possibly quite efficient. Research laboratories increasingly make their own small contribution with numerous widespread applications that involve tungsten, e.g. color reactions (protein determination, phosphate determination, tissue staining), solution deproteination, catalysis of oxidation by H202, particles for gene transfer by bombardment, tungsten microelectrodes. This redistribution can be expected to result in a more even, and finer mesh of the element in nature and thus may well increase our chances of finding new forms of tungsten biochemistry. 1.1 Tungsten is a Common Bioelement in Microorganisms
The positive influence of tungstate on microbial enzyme activity was first described in the early seventies by Andreesen et al. [4, 5]. Shortly afterwards the first tungsten enzyme was purified from Clostridium thermoaceticum and characterized in part [6- 8]. A pterin cofactor as ligand to the tungsten was identified in this [8] and (later) in other tungsten enzymes [9, 10] akhough the exact structure of the tungsten-containing prosthetic group remained obscure for two decades to come [ 11]. Until relatively recently, tungsten in biology was considered to be a rare oddity or at best a remnant from earlier times in evolution on its way out to extinction [1]. Over the last seven years some dozen tungsten enzymes have been described from a broad range of microorganisms including hyperthermophilic and mesophilic archaea, Gram-positive bacteria, namely, thermophilic and mesophilic Clostridia, and Gram-negative bacteria including an acetylene fermenter, a methylotroph, and a sulfate reducer. It thus appears that tungsten is a common bio-element in archaea and in bacteria. No biological function has been identified for tungsten in any eukaryotic cell. Tungsten enzymes were reviewed in 1994 by Adams [12]. Since then several developments have taken place to justify a second review. Perhaps the most sig-
164
w.R. Hagen. A.F.Arendsen
nificant of these is the first structure determination by X-ray crystallography of a tungsten enzyme [11] (shortly followed by the first 3D structure of two molybdenum enzymes with similar pterin cofactors [12, 14, 15]). Other events of significance are the growing awareness that Mo and W enzymes are not simply variations on a single theme, the discovery of tungsten enzymes in mesophilic bacteria, the availability of several structural gene sequences, and the first characterization of enzymic tungsten redox chemistry. Very recently, Kletzin and Adams published an extensive review on the microbiology of tungsten [16]. In the present review we briefly outline microbiological aspects, focusing instead on biochemical, structural, and spectroscopic aspects.
2 Chemistry of Tungsten and Its Biological Implications The position of tungsten in the periodic table of elements is unique. It is the only 5d series element with a known biological function. Figure 1 shows "the cell's view" of the block of transition ions. The elements given are those that are predominantly soluble in aqueous solution under "typical" conditions of near neutral pH [17]. The elements that actually have a function in cells are highlighted. For the other elements in Fig. 1 until now no bona fide biological functions have been established. The obvious explanation for Ag and Cd would be their tendency to bind to cysteine residues in polypeptides and proteins. The trivalent hard acids Y, La-Lu appear not to fulfil a similar role to the ubiquitous calcium ion, to which they are otherwise quite similar in their coordination chemistry. No biological function has been established for technetium or for rhenium, although these elements have significant similarity to tungsten. Their terrestrial abundance is very low. The bio-inorganic exploration of the 4d/5d island of Mo, W, Tc, Re has been particularly promoted by the field of nitrogen fixation. The biological activation of the inert N2 molecule is by means of a metal-sulfur cofactor in the enzyme nitrogenase. Three different nitrogenase systems have been identified and subsequently purified to homogeneity. The "standard" system contains an MoFeTSs cofactor [18]; the "vanadium" system contains a similar cofactor with V instead of Mo; the "third" or "alternative" or "iron-only" system presumably
6
7
Y
Mo
Tc
Ln
W
Re
3
4
5
8
9
10
11
12
Ag
Cd
L M°IFelColN'IOulZn
Fig. 1. Tungsten in the periodic table of biological elements. The figure gives transition elements that are reasonablysoluble in aqueous solution at near-neutral pH and a potential of 0 V.The boxed elements are those for which a biological function has been established
The Bio-lnorganicChemistryof Tungsten
165
also contains a similar cofactor possibly with Fe replacing the Mo [19]. With reference to the periodic table of Fig. 1, combining group relations (isoelectronics), diagonal relations (isoradial ions), and bioavailability (aqueous solubility) has led to the suggestion that W, Tc, Re are other possible candidates for yet other alternative nitrogenases. This idea has not materialized. It has been suggested that tungsten is built into a W-Fe-S cofactor; however, no relation with activity has been established [20]. Recently, a mutant strain of Azotobacter vinelandii has been reported, which, it is claimed, fixes nitrogen in the presence of W but not in the presence of Mo [21]. As no nitrogenase has yet been characterized the matter of W-related activity remains unsettled. The presence of up to 1 mM technetium (99Tc; tl/2 = 2.14 x 105 years) in the growth medium ofA. vinelandii does not affect nitrogenase activity (H. Haaker, personal communication). With the perrhenate ion in bacterial growth media no positive correlation has been found with nitrogenase activity [22]. For the time being, our vantage point is narrowed down to a smaller peninsula formed by Mo and W. The elements are not only quite similar chemically, but also (cf. below) biologically, as they figure as pterin-metal cofactors in similar enzymes. An important question to ask is: how similar or how different are tungsten and molybdenum? As a starting point we note from a chemical perspective that: 1) W and Mo ions of a given oxidation state have almost identical ionic radii due to the lanthanide contraction effect. 2) in aqueous chemistry W and Mo are stabilized in high oxidation states. 3) The reduction potentials for the (V/VI) and the (IV/V) couples differ by a small amount and frequently even cross over (i.e. electron-pair reduction). Also: 4) from their position in the periodic system W/Mo in coordination can be seen to be acids of intermediate hardness, like most of the bio-elements of the first (3d) transition series. Therefore, a rich coordination chemistry can be anticipated enabling complexation with all the common biological ligand atoms C, N, O, S, Se. 2.1 Tungsten Atomic Structure
Elemental tungsten is [Xe]4f145d46s 2, the ionic electron configurations are given in Table 1. W 2+ and W 3+ may not appear to be particularly likely candidates for bio-inorganic chemistry; however, low oxidation states of Mo have been discussed in the frame of model reactions for nitrogenase reactivity [23]. Also, molybdenum-carbon bond formation has very recently been proposed in xanthine oxidase-catalyzed reactions [24]. The standard Frost diagram in Fig. 2 provides a guide to the redox chemistry of tungsten [25, 26]. The most stable oxidation state is W(VI), although the Table 1. Electronic configurations of tungsten ions
configuration
5d°
5d1
5d2
5d3
5d4
oxidation state
W 6÷
W 5+
W 4÷
W 3+
W 2+
electron spin low spin
0 -
1/2
1 0
3/2 1/2
1 0
166
W.R. Hagen. A.E Arendsen 0 -1 -2
-3 e"
-4
-5
-6
-7 r
r
r
I
i
r
0
1
2
3
4
5
6
Oxidation number Fig. 2.
Frost diagram for tungsten in aqueous solution at acid pH (A) and alkaline pH (0)
effect is not pronounced at acid pH. Certain ligands may result in a quite different diagram: the octacyano coordination stabilizes the 5+ form of W and Mo [251. The inorganic chemistry of tungsten is rich in double, triple, quadruple metal-metal bonds [27]. The occurrence of metal-metal bonds in biological systems is controversial. They do not occur in tungsten enzymes as the metal is mononuclear, cofactor-bound. However, it is not known at present if/how tungsten is stored. Multiple bonds of tungsten with p-block elements are also common, e. g., W=O, W=N-H, W--N, W--=C-H [28], and this is certainly relevant to biology. The 74th element consists of five stable isotopes: ~8°W (0.14%); 182W (26.41%); 183W (14.4%); 184W (30.64%); 186W (28.41%). All isotopes have a nuclear spin I=0 except ~83Wwhich has I= 1/2. There are no stable quadrupole nuclei. The nuclear spin of 183W can be used in spectroscopy, e.g., isomer shift in W-NMR or hyperfine splitting in W-EPR. Because of its considerable natural abundance of 14.4% the ~83Wisotope is relatively cheap and obtainable in relatively pure form. Mass differences, e.g., between ~82Wand 186W,are of potential use in resonance Raman spectroscopy. A suitable radioactive isotope is the fl-emitter ~85Wwith a half life of 76 days. This isotope has been used in several studies on the incorporation of tungsten in archaea and bacteria (cf. [29]). 2.2 Aqueous Chemistry of Tungsten The hydrated W 6+ ion does of course not exist freely in water, as it would be a very strong acid. The aqueous chemistry of W (and Mo) is predominantly that of oxoanions. W(VI) is tungstate, WO42-, a feebly basic anion (pKbx= 9.4) as with
167
The Bio-lnorganic Chemistry of Tungsten
WO 3 >
wo -
o
-1
7
14
pH
Fig. 3. Pourbaix diagram of predominance of tungsten species in water. Broken diagonal lines indicate reduction potentials for the H20/O2 couple and the H2/H+ couple. The thick broken vertical line is for a tungstate concentration of 10 pM
most M Qz- anions because the basicity is a function of the number of oxogroups and of the charge of the anion [28]. This and other basic properties of W in water, and hence in natural waters and biological fluids, can be readily appreciated from a look at Fig. 3: the Pourbaix diagram [17] of species predominance as a function of pH and solution redox potential. A practical problem is immediately obvious: at slightly acidic pH the oxide WO 3 predominates, and this species is insoluble in water. From a biological point of view this may well signify an important difference between W and Mo, where the latter has a predominance area of reasonably soluble HMoO~ in between molybdate and the trioxide. It should be noted that the predominance diagram of Fig. 3 is one of standard conditions, i.e. molar activities at ambient temperature and pressure. In more dilute solutions, soluble species have larger predominance areas [17]. Therefore, in natural waters (including W-rich niches) and in laboratory growth media the tungstate species will predominate at lower pH values. In Fig. 3 this effect is indicated by the vertical, broken line for a tungstate concentration of 10 pM (a typical value used in batch cultures of microorganisms; see below). Note also that about half of the microbial species from which tungsten enzymes have been isolated are hyperthermophiles, i.e., they proliferate in nature and in the laboratory at temperatures close to the boiling point of water and at higher than atmospheric pressure. Another key point, which is not obvious from the standard Pourbaix diagrams, is the tendency of oxoanions of Mo and W to polymerize in the form of multiple oxo-octahedra [30, 31]. For example, in the biologically particularly relevant pH range 5 to 7.8 this process for tungsten at least involves the species WO 2-, W6020(OH)26-, W7026~, HW7025~, H2W12 O10- [32]. Under standard condi-
168
W.R. Hagen. A.E Arendsen
tions the rate of some of these conversions is quite slow; equilibrium is typically reached over periods of hours to weeks. The size of oxoanions is not significantly determined by the size of the central atom but predominantly by the number of oxo groups. Therefore, all MO~anions have similar size. Since, as already mentioned, the basicity of the anion is also determined by the charge, the W042- ion is quite similar to the common oxoanions MoO2-, S042-, SeO2-, HPO2-, HAs042-, but it is not so similar to the common oxoanions of Tc and Re (MO~) and of V (H2VO4, H3V207); it is similar to the species HVO~- which predominates at mildly alkaline pH. 2.3 Sequestering and Transport of Tungstate
Nutritional elements like tungsten have to be internalized by organisms. To this goal single-cell microorganisms typically possess complex systems, e.g., consisting of sequestering agents, receptors, membrane transporters, intracellular carriers [33]. Internalizing systems for tungsten have thus far not been described in the literature; however, one can make some straightforward predictions on the basis of the aqueous chemistry of tungsten. The most common form in aqueous solution is the tungstate oxoanion, W042-. This is also the form in which tungsten is added to microbial growth media in the laboratory, i. e. Na2WO4. Tungstate is quite soluble in water at pH _>7. Thus, it is perhaps less likely that microorganisms in these media will synthesize specific sequestering agents (see below). This is of course in contrast with the extremely strong complexing agents, the siderophores, for the extremely poorly soluble ferric ion [34, 35]. Internalizing systems of Gram-negative bacteria have been described in considerable detail for sulfate and phosphate, and to a lesser extent for molybdate, and they have been shown to have many properties in common (cf. [33]). In view of the strong similarity of the W042- ion to these other oxoanions, it is reasonable to assume the existence of a similar system for tungstate. A minimal picture for such a transport system is outlined in Fig. 4. It consists of an aspecific porin dimer, a soluble periplasmic anion-binding protein, and an ATPdependent pump made up of a transmembrane anion transporter and an ATPhydrolyzing protein. For example a high-affinity molybdate transport system is chromosomally encoded by the mod-operon in E. coli [36], A. vinelandii [37], or R. capsulatus [38]. This operon encodes the proteins A (more specifically: its precursor carrying an export sequence), B, and C, but not the aspecific porin. Molybdenum internalization in these bacteria encompasses much more than just the modoperon, e.g. regulatory genes and genes for cofactor synthesis. However, no genes encoding specific extracellular complexing agents for molybdenum have been identified. Catecholato-type ligands have been proposed as possible natural complexing agents for bacterial sequestering of molybdenum [34, 39]. A recent report suggests that tungstate internalization proceeds via a system similar to the modABC one. The product of a modA overexpressing E. coli strain has been purified. The protein binds both molybdate and tungstate with high
The Bio-lnorganicChemistryof Tungsten
169
INNER Membrane
OUTER
Membrane
Q B Porin
Fig.4. General scheme for an ion-transporter system in Gram-negative bacteria. A is a soluble, periplasmic, anion-binding protein; BC is an ATP-dependent pump consisting of a transmembrane transporter (B) and an ATP-hydrolyzing protein (C)
affinity (K d = 5 pM) but has no affinity for sulfate, chromate, selenate, phosphate, or chlorate [36]. While this inability to bind similar oxoanions remains to be explained on a protein molecular level, it is noted in passing that a negative correlation betwee n phosphate fertilizing and molybdenum uptake by plants is a well known fact in agriculture (cf. [40]). The above suggestion of a high-affinity tungstate transport system modeled after the molybdate modABC system should be placed in proper perspective with the following considerations: 1) no genuine tungsten transport system has yet been identified; 2) bacteria frequently carry the genetic information for more than one transport system for the same anion, a high-affinity and a lowaffinity one [33]; 3) the system of Fig. 4 applies to Gram-negative bacteria while most tungsten enzymes have been purified (see below) from Gram-positive bacteria and from archaea. For completeness, we note that in 185W labeling experiments of C. formicoaceticurn a 5.5 kDa protein has been detected, whose function has not been established [41]. Also, we recall an early report describing in some detail a Mostorage protein in A. vinelandii and suggesting the synthesis of a similar system for W storage as a response to growth on W [42]. 2.4 Bioavailability of Tungsten
It is clear (2.2, above) that the speciation of tungsten in near-neutral water, and therefore its typical bioavailability, is strongly dependent on the pH. When tungstate is converted at low pH into the trioxide WO3, it becomes insoluble and thus essentially not available to the cell. Thus, in nature not all habitats will allow for W-dependent forms of life. Acidophiles that thrive in media with an unusually low pH, are not the first organisms to be scrutinized in search of
170
w.R. Hagen. A.E Arendsen
tungsten biochemistry. The complex polymerization pattern of tungstate at pH _<7 will produce species that, though soluble, would be expected to be internalized by cells only with great difficulty, if at all. In the laboratory many microorganisms, particularly those concerning us here, e. g. extremophiles [43], are typically grown in chemostats on a medium of near neutral pH, or in batch cultures of a medium that initially is at near neutral pH, and that will usually acidify when growth proceeds. For example, Pyrococcus furiosus, from which three different W enzymes have been purified, can grow around neutral pH on a medium without added tungstate [44, 45]; growth essentially stops at pH _<5 [46]. However, the medium commonly contains a rich source of nutrients, yeast extract and tryptone (3 nmol W per g dry weight [47]). The stimulatory effect of tungsten has been tested by the addition of 10 pM of tungstate (e.g. [48-51]). This is an unusually high concentration; the optimal concentration established in studies on Clostridia [41, 52] and D. gigas [53] is 0.1 pM. One cannot help but wonder whether the complex pH dependence and, specifically, time-dependence of the polymerization of tungstate should not be a consideration in the composition of growth media for tungsten enzyme-synthesizing microorganisms. Another relevant factor is the very low solubility of calcium tungstate. However, the pH of many natural waters and soils will not exclude the sustenance of W-dependent life. This is also true for cultivated soils that are frequently kept at alkaline pH by liming.
3 Spectroscopy of Biological Tungsten The purification of tungsten proteins is a relatively recent event, and, therefore, most of the spectroscopy is still to be developed. No W-NMR nor W-M6ssbauer data are yet available. Some EXAFS has been reported; however, the redox state of tungsten in these studies is not clear (see below). MCD and RR studies have just been initiated. The only data set of significant size concerns the X-band EPR of W s+ (g- and A-values). ENDOR or ESEEM studies have not been carried out yet. Consequently, our discussion of some spectroscopies below is limited to summarizing the boundary conditions and predicting applicability to tungsten proteins.
3.1 Elemental Analysisof Tungsten Quantitative determination of tungsten in purified proteins is straightforward. The colorimetric dithiol method for the determination of Mo in proteins [54] can be modified for the detection of W (or Mo and W) [20, 55]. Inductively coupled plasma mass spectrometry, ICP-MS [56], and instrumental neutron activation analysis, INAA [57], are alternative methods suitable for determining W in purified proteins (cf. [58]). Internalization of W by cells and biosynthesis into proteins can be monitored analytically with the radioactive tracer 18sW [29, 41, 52].
The Bio-lnorganicChemistryof Tungsten
171
With the two extremes in W concentration covered (i. e. whole cells versus purified proteins) the question arises as to whether one could monitor tungsten proteins during purification on a preparative scale on the basis of W content. The answer is probably affirmative. A sensitive polarographic detection of W from soil samples has been described based on the principle of adsorption of a W-complexing agent, mercaptobenzothiazole, onto the hanging mercury drop electrode. The quoted detection range is approximately 0.004-1.4 ppm [59]. An equally sensitive alternative (0.002-0.02 ppm) for the simultaneous determination of Mo+W may be solid-phase spectrofluorometry of the carminic acid complex [60]. A sensitive (0.05- 30 ppm) determination of Mo, without interference of W, has been reported based on the catalytic action of Mo on the reduction of nile red with hydrazine dihydrochloride [61]. 3.2 Tungsten EPR Spectroscopy The 5d 1 system of W(V) has an electron spin S = 1/2 whose EPR spectroscopy should be straightforward. Spin-lattice relaxation is relatively slow and the W(V) signal is readily detectable up to nitrogen temperatures without significant lifetime broadening. The sign of the spin-orbit coupling constant for less than half-filled d-shell systems is negative. Therefore, all g-values are expected to be less than the free-electron value of g = 2.002. The 183Wisotope (14.4 %) has a nuclear spin I = 1/2 and central hyperfine interaction should be detectable in non-enriched samples. W(V) EPR data are available from synthetic tungsten coordination complexes, e.g., Refs. [62-64]. For these octahedral [WOX5] or square pyramidal [WOX4] complexes one of the g-values is, contrary to expectation, sometimes slightly greater than 2.00. For the core structure [WO(SeR)4] the deviation is considerable: gz= 2.09 [63]. The magnitude of 183Whyperfine splitting for thiolato complexes is typically 5 mTesla and anisotropy is not particularly pronounced. EPR results for W enzymes and for W-substituted Mo enzymes are summarized in Table 2. Remarkably, more than half of the reported spectra have one or two g-values greater than the free-electron value. In some cases this deviation is very pronounced. A ligand-field model has been proposed for synthetic complexes involving mixing in of a low-energy charge-transfer excited state [62]. This model has been cited subsequently to explain the unusual g-values in tungsten enzymes [65, 67]. The situation is unsatisfactory because the original model predicts a stronger effect for heavier ligands (e. g. Se which is not a ligand in the enzymes) and a stronger effect for Mo over W (not observed in the enzymes). Also, this interpretation cannot be generally valid where half of the proteins have "regular" g-values and all complexes are presumably pterin derivatives. It seems to us that the problem of the g-tensor in W complexes is worthy of reconsideration. Only limited data are available on 183Whyperfine splittings in proteins. The A-values given in Table 2 are from one W enzyme (see the spectrum in Fig. 5) and two W-substituted Mo enzymes. All three sets have been obtained on non-
172 Table 2.
W.R. Hagen. A. E Arendsen W(V) EPR of tungsten enzymes
Enzyme
M. wolfei FMDH2 (W/Mo)3 P.furiosus AOR P.furiosus GAPOR C. thermoaceticum FDH C.formicoaceticum CAR
g-values
2.049, 2.012, 1.964 1.982, 1.953, 1.885 1.946, 1.885, 1.829 2.10, 1.98, 1.95 2.035, 1.959, 1.899 2.03, 2.01, 2.002 D. gigas AOR 1.943, 1.902, 1.850 1.974, 1.932, 1.869 Rat liver sulfite oxidase (W/Mo) 1.982,1.982, 1.936
A-values1
Reference
3.1, 4.6, 5.0 4.2, 4.5, 7.7 nr 4 nr nr nr nr nr 4, 4, 9
[65] [66] 5 [67] [68]
[69] [70]
1 183W (1=1/2) hyperfine splitting values in mTesla. 2 FMDH,formylmethanofuran dehydrogenase; AOR, aldehyde oxidoreductase; GAPOR,glyceraldehyde-3-phosphate oxidoreductase; FDH, formate dehydrogenase; CAR, carboxylic acid reductase. 3 W-substituted Mo enzyme. 4 Not reported. 5 Our unpublished data.
enriched samples. The values are "typical", i. e. comparable to values for thiolato model complexes [63]. The W-substituted rat liver sulfite oxidase exhibits a resolved superhyperfine doublet from an I -- 1/2 ligand [70]. This proton splitting is quite characteristic for the EPR of Mo enzymes, however, it appears to be not generally present in the EPR of W enzymes or, for that matter, in the EPR of metalloproteins in general. The nature of the proton and why it is observed in Mo spectra and not in W spectra are questions that remain to be addressed. A possible approach to this question and to the study of the fine details of the electronic structure around the tungsten ion in general would be ENDOR and/or ESEEM spectroscopy. No data have been reported yet. The T 1 relaxation rate of W(V) in proteins is comparable to Cu(II) and somewhat faster than Mo(V) ([70] and our unpublished observations), and this suggests that application of these spectroscopies should not pose any particular problem. An important, and problematic (see below), observation is that P. furiosus AOR exhibits a W(V) signal upon oxidation, which does not disappear at high redox potentials. The apparent implication of this would be that the tungsten is never found as W(VI). Similarly, W(V) signals have been observed in isolated enzyme, which disappeared upon reduction for D. gigas AOR [69] and for P.furiosus GAPOR (our unpublished observations). Finally, we note that W(V) is not the only paramagnetic oxidation state of tungsten. W(IV) and Mo(IV) can be high-spin, S = 1. Because the zero-field splitting may well be significantly larger than the energy of standard EPR equipment, these d 2 systems, like the d 8 systems Ni(II) and Co(I), are expected to be interesting objects for the high-frequency (typically _>90 GHz) EPR configurations that are presently being realized in several laboratories worldwide.
173
The Bio-lnorganicChemistryof Tungsten
1.982
1.953
1.886
$
I
320
340
360
B/mT
Fig. 5. EPR spectrum at 40 K and simulation of W(V) in P. furiosus AOR oxidized to E > + 180 mV. The simulation is based on 14.4% natural abundance for the I = 1/2 nucleus ~83W; see Table 2 for parameters [66] 3.3 Tungsten NMR Spectroscopy
NMR spectroscopy on tungsten in proteins has yet to be reported. NMR of 183W (I=1/2) is difficult because the transition has a low intrinsic sensitivity. However, the S7Fe(I = 1/2) nucleus has an even lower NMR sensitivity, yet a few cases of S7Fe -NMR in iron proteins have been reported [71-73]. In Table 3 relevant NMR parameters of 57Fe and ls3W are compared to those of the familiar I = 1/2 nuclei ZH and 13C. Here sensitivity is defined as the third power of the ratio of the nucleus gyromagnetic ratio to that of the proton. This n u m b e r gives only an approximate indication of relative sensitivity. It is not corrected for isotope abundance. Also, it will be further dependent on both T 1and T2 relaxation times. Both for iron and tungsten proteins isotope enrichment is indicated. A final enrichment of 57Fe to 70-80 % in bacterial batch cultures can be routinely obtained (cf. [74]). A similar percentage enrichment in ls3W (i. e. approximately six Table 3. Comparison of proton, carbon, iron, and tungsten NMR nuclei
Nucleus (% abundance)
Frequency (MHz)
Sensitivity (versus 'H)
Typical ~ range (ppm)
1H (99.98) 13C (1.11) 57Fe (2.19) ls3W (14.4)
100.00 25.14 3.24 4.19
1 0.016 0.000034 0.000073
10 200 12,000 4,000
174
w.R. Hagen •A.E Arendsen
times) should be easily obtainable because I83W is an order of magnitude cheaper than 57Fe and because standard chemicals for growth media contain much less W than Fe. Thus the data collection time for ~83W-enriched sampes should be down by one or two orders of magnitude. Indirect NMR detection of W by polarization transfer to a sensitive nucleus through the INEPT technique has been described for a variety of W(VI) coordination compounds [75-79]. Subsequently, direct W-NMR spectroscopy has been reported [80, 81] recently also on W-carbohydrate complexes [82, 83]. Typically, much higher concentrations are used than attainable in protein NMR. An estimation of the required amount of 183W-enriched protein for P. furiosus AOR has been made by extrapolation of the signal-to-noise ratio from W(VI)-toluene-3,2-dithiol in CDC13 (6=2702 ppm versus tungstate) at 300 MHz (F. Merkx and J. Vervoort, personal communication). Half-a-man-year of continuous purification effort was indicated for one NMR sample of AOR to give a signal-to-noise ratio of 10 after 10 hours of data collection. It is clear that the W-NMR on proteins will only be practical under specific conditions, e.g., availability of overproducing clones and/or samples of relatively small molecular mass, e.g., small proteins or isolated cofactors. As a final, possibly crucial consideration we note that only W(VI) is purely diamagnetic (cf. Table 1) while redox titration data (see below) indicate that W(V) is the highest oxidation state occurring in tungsten proteins. 3.4 Tungsten Mi~ssbauerSpectroscopy M6ssbauer spectroscopy on biological systems has thus far been limited to 57Fe. This is perhaps not surprising as 57Fe is the ideal M6ssbauer nucleus and iron is a ubiquitous bio-element. There are, however, several other nuclei of biological interest for which the M6ssbauer effect has been detected. Perhaps the most obvious one is 67Zn,which unfortunately is experimentally an extremely difficult system because of the extremely narrow natural linewidth of the M6ssbauer transition. Attempts in the seventies to develop M6ssbauer spectroscopy on zinc enzymes failed (R.H. Sands, personal communication). Another potentially interesting case is 61Ni: conventional M6ssbauer spectroscopy is not practical because the parent nucleus has a half-life of only 99 minutes. The use of synchrotron radiation is presently being explored for M6ssbauer spectroscopy on 61Ni systems, one of the aims being to include nickel enzymes in the future (A.X. Trautwein, personal communication). The only other naturally occurring stable M6ssbauer nuclei are I27Iand the four isotopes of tungsten. In Table 4 a few properties of the M6ssbauer transitions of Fe, Ni and W are compared to estimate the feasibility of biological tungsten M6ssbauer spectroscopy. In general, the following considerations hold: the isotope natural abundance should be high; the energy, E, of the M6ssbauer transition should be low; the natural linewidth, F, should be small; the resonant absorption cross-section, a, should be large; the lifetime of the parent nucleus, tl/2, should be long. Our reference point is the popular 14.4 keV transition of 57Fe which has been very
The Bio-lnorganicChemistryof Tungsten
175
Table 4. Comparison of iron, nickel and tungsten M6ssbauer transitions. Nucleus (% abundance)
E (keV)
F (ram/s)
a (10-18cm2)
tl/2 (days)
57Fe (2.17) 57Fe (2.17) 61Ni (1.25) 182W (26.4) x83W (14.4) 183W (14.4) 184W (30.6) 186W (28.4)
14.4 136.3 67.4 100.1 46.5 99.1 111.2 122.6
0.19 0.23 0.78 2.00 31 3.9 1.92 2.21
2.57 0.35 0.72 0.29 0.23 0.14 0.27 0.31
270 270 0.07 115 5.1 5.1 38 3.75
(Co) (Co) (Co) (Ta) (Ta) (Ta) (Re) (Re)
E is transition energy; F is linewidth; a is resonant absorption cross-section; t m is lifetime of radioactive parent nucleus. Data taken from [84].
extensively used in biology. At the other extreme is the 136.3 keV transition of 57Fe, which is difficult to employ and in fact has never been used in biological studies. In between is the nickel transition, which at this time at least has the benefit of the doubt. From Table 4 it can be seen that tungsten is somewhat less suitable than nickel because the M6ssbauer effect is somewhat weaker and the natural linewidth is larger. On the other hand, the lifetime of the parent nucleus for the 182W isotope readily allows for conventional M6ssbauer spectroscopy, and there is hardly any need for isotope enrichment. In principle the M6ssbauer effect of e.g. 182W in tungsten proteins should be detectable by conventional methods. The conventional tungsten spectroscopy will be more expensive than the iron counterpart because of differences in the source and detection system. However, these items do not make up the bulk of the price tag of a M6ssbauer facility. Also, it would be worthwhile to consider tungsten as the next target for synchrotron M6ssbauer spectroscopy once the benchmarks for nickel have been established. The spectroscopy itself will also be somewhat different because the isomer shifts of W (and also of Ni) compounds are small compared to the natural linewidth. It may prove to be quite difficuk to extract information on the oxidation state of the metal. However, useful information on structure and bonding should be extractable from quadrupole patterns. A difference with the well-known iron quadrupole doublets is that all even-numbered isotopes of W have a nuclear spin I = 2 in their excited M6ssbauer state resulting in more complex patterns.
3.5 Tungsten X-Ray Absorption Spectroscopy X-ray absorption data on metalloproteins using synchrotron radiation are commonly obtained at the K-edge of the metal. For tungsten, the excitation energy of a 1s electron is impractically high, therefore, data are taken at the L3-edge. Thus far only two studies'on tungsten enzymes have been reported.
176
W.R. Hagen. A.E Arendsen
An early paper reports on a single broad asymmetric peak in the room temperature FT-EXAFS spectrum of C. thermoaceticum formate dehydrogenase in the presence of dithionite [85]. The data appear inconsistent with W = O and are interpreted in terms of _>2 SR ligands at 2.39 ~ and >__2 (N,O) ligands at 2.13 with the caveat of a large uncertainty in the coordination number. The paper should perhaps be taken as a first exercise as it cites several experimental problems: a severe "glitch" problem with the monochromator and low enzyme activity ascribed to the presence of apoprotein. A tungsten L-edge study at 4 K on the Rfuriosus aldehyde oxidoreductase resulted in clearly resolved peaks in the FT-EXAFS spectrum [86]. The derived coordination was two W=O bonds with a bond length of 1.74 A, three S ligands at 2.41 ~, and possibly an (O,N) ligand at 2.1 X. One important conclusion from this study is that no W-Fe interactions are observed, therefore the tungsten is not part of an iron-sulfur heterometal cluster (as originally proposed in [12, 87]). This also implies that the extensive model chemistry on W-Fe-S clusters (cf. [88, 89]) has no relevance for tungsten enzymes (it may still be relevant for putative biological tungsten storage). The proposed structure is not fully consistent with the subsequently resolved X-ray crystallographic structure (4 sulfur and 0-2 oxygens; see below). It has been suggested [86] that the oxidation state of tungsten in the first study was a mixture of IV, V and VI. In the second study the oxidation state was assumed to be VI [86], however, the preparation was enzyme as isolated anaerobically in the presence of dithionite. Results from an EPR redox titration study [66] indicate that the tungsten may have been IV. In both studies the processed EXAFS data are presented and not the original L-edge spectra; no comment is made on the structure of the L3-edge (XANES spectrum) and possible pre-edge features. It seems to us that that there is still much to be explored in the X-ray absorption spectroscopy of tungsten proteins.
3.6 Tungsten Optical Spectroscopies All tungsten enzymes purified thus far appear to contain iron-sulfur clusters (at least one Fe/S cluster per W, see below). The optical absorption spectra, in sofar as they have been reported [9, 69, 87, 90]), show the broad, featureless peak around 400 nm characteristic of Fe/S-containing enzymes. Some of these spectra are reported for air-oxidized, i.e. inactivated enzyme. The redox state of the tungsten in these preparations is not defined. No attempt has been reported yet to deconvolute any contribution of the tungsto-pterin cofactor to the UVvisible spectrum from the Fe/S contribution. A possible approach to this problem would be to make use of the differential magnetism and charge-transfer properties of the chromophores, i.e. to study the magnetic circular dichroism spectroscopy and the resonance Raman spectroscopy of tungsten enzymes as a function of redox state. The group of MK Johnson has recently announced preliminary MCD and RR results on tungsten enzymes from three archaea [91, 92].
The Bio-lnorganicChemistryof Tungsten
177
4 Microbiology of Tungsten The microbiology of tungsten has been extensively reviewed very recently [16]. Here, we limit ourselves to a few general observations. Table 5 lists some microorganisms for which an effect (stimulatory or inhibitory) has been reported on specific activities as a response to tungsten addition. The listing is alphabetic in the species name, however, several other arrangements may be practical. For example, singling out the archaeal species Table S. Effect on enzyme activities of addition of tungstate to the growth medium of microorganisms Microorganism
Enzyme activity
Effect Reference
Alcaligenes eutrophus Arthrobacter nicotinovorans Azotobacter vinelandii Azospirillum brasilense Sp7
formate dehydrogenase nicotine dehydrogenase nitrogenase nitrate rednctase nitrite reductase carboxylic acid reductase carboxylic acid reductase formate dehydrogenase formate dehydrogenase carboxylic acid reductase formate dehydrogenase formaldehyde oxidoreductase carboxylic acid reductase formate dehydrogenase formate dehydrogenase aldehyde oxidoreductase dimethylsulfoxide reductase formate dehydrogenase formate hydrogen lyase FMF dehydrogenase FMF dehydrogenase formate dehydrogenase FMF dehydrogenase formate dehydrogenase formate dehydrogenase nitrate reductase acetylene hydratase aldehyde oxidoreductase GAP dehydrogenase formaldehyde oxidoreductase sulfhydrogenase dimethylsulfoxide reductase quinaldic acid 4-oxidoreductase nitrate reductase formaldehyde oxidoreductase aldehyde oxidoreductase
+ + + + + + + + + + + + + + + + + + + + + + + +
Butyribacterium methylotrophicum Clostridium aceticum Clostridium acidiurici Clostridium cylindrosporum Clostridium formicoaceticum Clostridium thermoaceticum Clostridium thermoautotrophicum Desulfovibrio gigas Escherichia coli Eubacterium limosum Methanobacterium formicicum Methanobacterium thermoautotr. Methanobacterium woIfei Methanococcus vannielli Methanosarcina barkeri Methylobacterium sp RXM Mycobacterium vaccae 10 Paracoccus denitrificans Pelobacter acetylenicus Pyrococcus furiosus
Rhodospirillum rubrum Serratia marcescens 2GC-1 Synechococcus R2 Thermococcus litoralis Thermococcus, strain ES- 1
FMF = formylmethanofuran; GAP = glyceraldehyde-3-phosphate.
[93] [94] [20] [95] [95] [47] [47] [52] [52] [41] [5] [8] [96] [4] [97] [53] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [51 ] [109] [110] [ 111] [ 112] [113] [ 114] [110] [ 115]
178
W.R. Hagen • A.E Arendsen
(Methanobacterium, Methanosarcina, Pyrococcus, Thermococcus) will show that within this Urkingdom stimulation is the rule. It appears that tungsten plays a prominent role in the biochemistry of archaea. On the other hand, in the bacterial world we frequently encounter inhibitory effects of tungsten. When the entries of Table 5 are grouped according to enzyme activity, it can be seen that inhibition applies to activities that have been well characterized as carried by molybdenum enzymes (e.g., nitrate reductase) or to activities that are dependent on molybdenum enzymes (e. g., nitrite reductase). Also, stimulated activity appears to be restricted to strict anaerobic or micro-oxygen tolerant (D. gigas, cf. [116]) species. Finally, from Table 5 it can be seen that tungsten stimulation occurs in archaea and bacteria, in mesophilic, thermophilic, and hypertermophilic species, in Gram-negative and in Gram-positive bacteria. The effect of tungsten on gene expression of species stimulated by tungsten has not been developed at the molecular level. It has been suggested that tungstate plays a role in transcriptional regulation of Mo enzymes, e.g. nitrate reductase (cf. [16]) and nitrogenase [117]. 4.1 Reactions Catalyzed by Tungsten Enzymes Roughly seven distinct tungsten-enzyme activities have been reported; these are listed in Table 6. The microbiological significance of some activities is evident; other putative activities are under discussion. Formylmethanofuran dehydrogenase, FMDH, catalyzes the first step in CO 2 fixation by archaeal methanogens with the use of methanofuran (MFR) as cofactor [102]. The CO 2 ends up as the formyl group covalently bound to the cofactor: N-formylmethanofuran. Similarly, tungsten-containing formate dehydrogenase, FDH, catalyzes COg fixation in some bacteria using the cofactor NADPH as the direct source of reducing equivalents [7]. The product of the reaction is the free formate anion. Formaldehyde oxidoreductase, FOR [110], aldehyde oxidoreductase, AOR [119], and glyceraldehyde-3-phosphate oxidoreductase, GAPOR [109], from hyperthermophilic archaea are related enzymes which catalyze the reduction of aldehydes at the expense of reducing equivalents presumably derived from fer-
Table6. Reactions catalyzed by tungsten enzymes Enzyme
Reaction
Reference
FMDH FDH FOR AOR GAPOR CAR C2H2hydratase
CO2 + MFR + 2[H] --> CHO-MFR + H20 CO2 + NADPH + H+ --->HCOOH + NADP+ HCOH + H20 + Fd(ox) --->HCOOH + 2[H] + Fd(red) RCHO + H20 + Fd(ox) --> RCOOH + 2[H] + Fd(red) R'COH + H20 + Fd(ox) --~ R'COOH + 2[H] + Fd(red) RCHO + 2MV++ + OH- --~ RCOO- +2H+ + 2MV+ C2H2+ H20 --->CH3CHO
[102] [118] [110] [119] [109] [118] [108]
MFR = methanofuran; Fd = ferredoxin; R' = -CHOHCH2OPO23-;MV = methyl viologen.
TheBio-lnorganicChemistryofTungsten
179
redoxin. As their respective names indicate, these enzymes differ in substrate specificity, and therefore in their role in archaeal metabolism. Their precise placement on the metabolic chart is not always clear as the chart itself has been, and appears to continue to be, a matter of debate [119 - 122]. This discussion has been reviewed very recently [16] and will not be repeated here. The bacterial counterpart of the ferredoxin-dependent archaeal aldehyde oxidoreductase family is the AOR from the sulfate reducer D. gigas [69] and the Clostridial carboxylic acid reductase, CAR [96]. Recently, the latter have also been referred to as AORs [9]. The natural redox partner of these enzymes has not been established. The physiological role(s) of bacterial AOR and CAR is not known. Acetylene hydratase catalyses the first step in the fermentation of acetylene to acetate and ethanol by P. acetylenicus [108]. The product of this non-redox hydration reaction is acetaldehyde. 5 Biochemistry of Tungsten Analytical data on tungsten enzymes that have been purified to near homogeneity have been collected in Table 7. At first sight the composition of these complex enzymes may appear to vary considerably between species, however, we feel that most of this variation stems from two factors: the presence of apoprotein and the adaptation to different entry/exit routes for reducing equivalents. The metal data in Table 7 suggest the presence of 0.4-1.1 W atoms per minimal catalytic unit. Most or all of these enzymes are unstable on contact with air (although those from hyperthermophilic sources are thermostable) and non-rigorous anaerobic conditions might be related to substoichiometric metal content in purified protein. The active center of tungsten enzymes is thought to consist of a single tungsten ion coordinated by a pterin cofactor (see below). The additional prosthetic group and cofactor content reflects the variation in electron transfer routes that is quite characteristic of redox metalloenzymes (cf. nickel hydrogenases [124]) and may involve combinations of Fe/S dimers, Fe/S cubanes, flavin, and possibly other groups. Dimerization or polymerization, apparently not related to function, further contributes to complicating the molecular picture. 5.1 Generic Reactions Reactions catalyzed by Mo enzymes (except nitrogenase) have been discussed in terms of two formal equations (reviewed, e. g. in [ 125, 126]), namely, the"oxotransfer formulation" reaction or "oxo-atom transfer" (OAT) reaction: X + [O] --9 XO
(1)
and the "coupled electron-proton transfer" (CEPT) reaction: X + H20 --~ XO + 2H++ 2e-
(2)
180
W.R, Hagen, A.R Arendsen
%
0&
~
a-?
~
'..o
,.-,
r~
•
~ h
~
~
,-2
.,~
,.,,..,
"u~
.,..u
- o .,--, ~. o ~ o . ~
~
"
'~
,..,
~ ~ . ~
o
r...p
o
~
~
o
r,..p
~
r..p
d
e 0
o
<
0
~
~
~
~
u<
<
~
'~ '.U r~
o
The Bio-lnorganicChemistryof Tungsten
181
The formulations differ in the working mechanism that each suggests for the enzyme. In the OAT formulation an oxygen atom is simply transferred from the metal (molybdenum) to the substrate or vice versa, and this formally changes the oxidation state of Mo from VI to IV. The CEPT formulation suggests that protonation of a ligand plays a role (i.e. via modification of ligand pKa through an oxidation-state change at the metal) and implies that water is the ukimate oxygen donor. A complete substrate reaction for molybdoenzymes may be written as: RH + H20 + (A) --~ ROH + 2H++ (D)
(3)
where (A) is a generalized electron acceptor, i. e. the/an oxidized natural redox partner of the enzyme (the electron stoichiometry is not specified). The mechanism of action of tungsten enzymes has not yet been discussed in the literature in any detail. Since the generic reactions catalyzed by tungsten enzymes (except acetylene hydratase) are a subset of Eq.3, namely: RCHO + HzO + (A) --) RCOOH + 2H++ (D)
(4)
the OAT vs. CEPT discussion may also apply. Note, however, that until recently the oxidation state change has generally been assumed to be localized on the metal with the cofactor in the role of a non-redox ligand. The formate dehydrogenase reaction is catalyzed in the reverse direction by tungsten FDHs: HCOOH + (A) (-- CO2 + 2H++ (D)
(5)
In this CEPT formulation the water molecule is absent. For an OAT formulation of this process one has to assume that the redox reaction is preceded by a hydration of the CO2 to carbonic acid. Finally, we note that the acetylene-hydratate reaction, that is apparently catalyzed by a tungsten enzyme in P. acetylenicus [108]: Call2+ H20 --q CH3CHO
(6)
does not fit in any way in the above scheme because it is not a redox reaction. Remarkably, this enzyme apparently produces a product that is normally a substrate for tungsten enzymes. Also, contrary to the situation for all other tungsten enzymes, this activity is not inactivated by oxygen [108]. The quoted paper ends with the sentence: "Thus, a possible physiological function of acetylene hydratase beyond acetylene hydration cannot be defined at present". 5.2 Pterin Cofactors
The cofactor for W/Mo in proteins has the trivial name tungstopterin or molybdopterin. A pterin is dicyclic, however, the cofactor is a tricyclic system. Pterin is a derivative of the four-nitrogen ring system pteridine (1); see Fig. 6. The fivenitrogen system 2-amino-4-hydroxy pteridine is pterine (2). The cofactor is thought to be derived from tetrahydro pterin-6-carboxylic acid (3).
182
W.R. H a g e n . A.F. Arendsen
2
3 H
N
H2N
N
N
N
y.y.h
N, ~ . , . , y
OH
O
OH
H.aN
H N
N
CHOI-ICH2OPO3=
O
SH
SH
4 H
O
SH
5 NH
,yy 0
SH OH
Fig.6
6
OH
The Bin-InorganicChemistryof Tungsten
183 o
\
/ \
o
/
O P O ~ - / Mg----'OsPO \
7
o
0/
\0
s\
7
I
'w' / s /
I-I2
H
I
°
\\ s
H
N-~
Fig. 6. Structure of the tungsten cofactor and related compounds. (1) pteridine; (2) pterine;
(3) tetrahydro-pterin-6-carboxylic acid; (4) pro-metallopterin; (5) metallopterin; (6) metallopterin-cytosine-dinucleotide; (7) tungsten cofactor in JR.furiosus AOR. Two of the dangling oxygen ligands to the Mg in (7) are from carbonyl backbone groups; the other two are presumably from water molecules [11]
The original proposal for the structure of the molybdopterin cofactor, based on the work of Johnson and Rajagopalan (cf review [127]), was structure (4). From the recent X-ray crystallographic studies on W/Mo enzymes [11, 13, 14] we now know that the system is modified by an intramolecular cyclization to a pyran ring thus forming a three-ringed structure (5). To avoid the confusion that comes from the fact that both (4) and (5) are called molybdopterin, we suggest the trivial names pro-metallopterin (4) and metallopterin (5). A numbering system has recently been proposed for metallopterin (5) [14]. The cofactor coordinates the metal through its 1'-2' cis-dithiolene group. In bacterial and archaeal molybdenum enzymes the phosphate group at the 4' end of (5) is usually 5'-bonded to a nucleoside by a pyrophosphate ester bond to form molybdopterin-X-dinucleotide. X stands for G (Guanine), C (Cytosine), A (Adenine) or I (Inosine) [127]. For example (6) is the metallopterin cytosine dinucleotide that provides the dithiolene ligand for Mo in D. gigas Mo-AOR [13]. Dinucleotide forms of the cofacor have thus far not been identified in tungsten enzymes. Four tungsten enzymes from hyperthermophiles were all shown to contain nucleoside-free metallopterin [10]. The cofactor ligand for tungsten in P.furiosus AOR is a dimer of two metallopterins whose phosphate groups are linked through a magnesium ion (7). The remainder of the octahedral Mg(II) coordination is formed by two water molecules and two carbonyl oxygens from the protein backbone. Both dithiolene groups coordinate to the tungsten [11].
184
W.R. Hagen. A. F. Arendsen
5.3 Crystal StructuralAnalysis One single crystal structure of a tungsten enzyme has been determined: AOR of P. furiosus. A 2.3 ~ resolution was obtained by multiple isomorphous replacement and multiple crystal form averaging [11]. The oxidation state of tungsten in the crystals has not been established. The analysis has not been unambiguous with respect to the coordination of tungsten. Specifically, the coordination number is uncertain. P. furiosus AOR has a single subunit of 605 residues [128]. The protein is a homodimer in the crystal [11] and probably also in solution [66]. Three different metal centers were identified in the crystal analysis, a tungstopterin and a [4Fe-4S] cubane within each subunit, and a single tetrahedral metal at the dimer interface. The topology of these centers is indicated in the cartoon of Fig. 7. Several important structural conclusions have been drawn with confidence from the X-ray analysis. The organic part of the cofactor is a three-ringed system, namely the metallopterin structure (5) in Fig. 6, and the complete cofactor is the dimeric pterin (Mg,W) structure (7). The cofactor is directly bonded by the protein through two backbone carbonyl (Asn-93 and Ala-183) oxygen bonds to the coordinatively saturated magnesium. These two residues are conserved in the FOR of T. litoralis [128]. In Fig. 8 the metal coordination is schematically outlined in the form of idealized octahedra and compared to the two Mo enzymes for which crystal data are available [13-15]. The non-sulfur ligands to the tungsten are problematic. According to the original paper "the electron density around the tungsten suggests that two additional coordination sites may be occupied by glycerol or oxo ligands or both" [ 11]. The EXAFS spectroscopy on this enzyme suggested two oxo ligands (in addition to three sulfurs) [86]. However, both for the EXAFS study and the crystal study the oxidation state of W has not been established. In both cases the protein was purified under anaerobic reducing conditions, and this apparently results in W(IV) acf
O.
N~ (
W ~
,
5nml
( S . W ~ S~
nm
Cys
CyS~Fe~Fe 0.6nm
/
CFS ~ C y s / / J
Fig. 7. Topologyand approximate distances of metal centers in the homodimeric AOR from P.furiosus (data from [11])
185
The Bio-lnorganic Chemistry of Tungsten
D.gi#as
/7.sphaero/des
P. furiosus
AOR
DMSOrd
AOR
s
I
0
O?
I
ISer-147] -~
or,?; ° o.2
s
;
',,
°
glycerol? .07
s1
o [Glu-86gl
[Asn-g3]
O~
4>
[Ala-1831
°
Fig. 8. Idealized octahedral representation of metal coordination in W- and Mo-enzymes as
deduced from X-ray crystallography (cf [11, 13-15]). The encircled P symbolizesthe protein cording to the EPR redox titration [66]. It thus appears premature to draw conclusions about the stereochemical consequences of the tungsten coordination for the mechanism of action [ 11 ]. This stereochemistry has been reviewed recently for Mo enzymes based on the crystal-structural information [13-15] on the assumption that oxidized crystals have Mo(VI) and reduced crystals have Mo(IV) [129]. Unfortunately, the Mo enzymes are not readily comparable in detail to P. furiosus W-AOR, because D. gigas Mo-AOR has a five-coordinated Mo(VI?) with a single pterin, and in R. sphaeroides DMSO reductase one of the ligands to the Mo(VI?) is oxygen from a protein residue (cf. Fig. 8). The W-AOR of P.furiosus contains a single iron-sulfur cluster per subunit coordinated by four Gys residues, in a pattern of three, C-X-X-C-X-X-X-C, with a fourth Cys further downstream. Although this pattern, which is conserved in T. litoralis FOR [128], is not exactly the canonical ferredoxin-type sequence, the cluster appears to be a regular [4Fe-4S]/2+;z+/, single-electron transferring cubane. The reduction potential is Era,8= - 410 mV at 20 °C [87]; the ground state of the oxidized cluster is diamagnetic; the reduced, mixed-valence cluster is a physical mixture of S = 312 and S = 1/2 [66, 87]. The cubane is the proposed electronic link between the enzyme"s natural redox partner(s), presumably ferredoxin, and the tungstopterin active center. The distances to the protein surface and to the tungstopterin (Fig. 7) are consistent with this assignment.
186
W.R.Hagen. A.F.Arendsen
The third, subunit-bridging metal center is enigmatic. In the original paper: "the metal site is tentatively identified as iron on the basis of anomalous scattering of the atom and a metal analysis of the AOR protein although direct confirmation of this assignment by M6ssbauer spectroscopy is needed" [11]. Unfortunately, details of the anomalous scattering experiment are not given. In the present reviewers view anomalous scattering would allow for a more straightforward and specific assignment than M6ssbauer spectroscopy (one out of nine tetrahedral iron sites per protein dimer). Also, details of metal analysis are not given, however, again, to differentiate eight versus nine Fe in a protein of 132 kDa is far from a trivial matter. We have briefly reported on EPR studies of the anaerobically oxidized enzyme that indicated the bridging metal to be highspin (S=5/2) ferric iron in an environment of intermediate rhombicity [130]. However, our subsequent studies indicate that this signal is not always found in AOR preparations and does not appear to be related in a simple way to activity (unpublished resuks). For the metal in reduced enzyme Adams and collaborators have recently claimed on one occasion high-spin ferrous, S=2 [92] and on another occasion low-spin ferrous, S =0 [16]. The two ligands per subunit to the metal are Glu-332 and His-383, and they occur in two separate E-X-X-H patterns per subunit. In their recent review Kletzin and Adams note that in the protein sequence a total of three of these patterns are found, and they recall that a double E-X-X-H pattern is a putative binding motif for a dinuclear Fe-O-Fe cluster. They also call two of the patterns "conserved". However, this is not borne out by their comparison to the sequence of T. litoralis FOR [16, 128]. Finally, we know of no evidence, spectroscopic or otherwise, for the presence of dinuclear iron-oxo clusters in tungsten proteins. At this time we tentatively conclude that ,,the subunit-bridging iron" in P. furiosus AOR appears to be a nonfunctional accident of purification. 5.4 Redox Chemistry at the Active Site
The fact that the reduction-potential values E(V/VI) and E(IV/V) in Mo/W coordination complexes are frequently found to be close, or even crossed over, combined with the relative stability of the Mo/W-oxo group in the hexavalent state, naturally suggests a central role for the metal of Mo/W enzymes in biological OAT or CEPT reactions: group transfer and localization of valence. Thus, it is a long-standing paradigm of molybdenum biochemistry that the metal shuttles between Mo(VI) and Mo(IV) (e.g. [125, 126]) and that the pterin cofactor does not enter in the redox chemistry [131]. It would appear to be only natural to extend these ideas also to tungsten enzymes. However, several independent observations have been recently made that force us to reconsider this central paradigm of Mo/W biochemistry. The AOR of P.furiosus was initially purified in an inactive form on the basis of its reddish color in the presence of dithionite: RTP or red tungsten protein [87]. This protein was studied in EPR-monitored pH- and temperature-dependent redox titrations. No W(V) signals were observed in RTP. The oxidized protein was EPR-silent; reduction gave the S--3/2 and S = 1/2 mixture of the [4Fe-
The Bin-InorganicChemistryof Tungsten
187
4S] 1+ cubane; upon further reduction to very low potentials, <-0.4 Volt, a very complex spectrum appeared with effective g-values approximately ranging from 10 to 1.3. It was concluded that E(W5+/W 6+) ~ - 0.4 V [87]. Subsequently a rapid purification procedure was described to obtain the protein in an active form: AOR, and it was claimed that the EPR of AOR is indistinguishable from RTP [119]. However, it was recently found that this is not correct; the complex interaction signal at low potential is found only with inactive RTP not with active AOR [66]. Active AOR exhibits a W(V) signal in the oxidized state with Em,7.s(W4+/W5+)----+180 mV at ambient temperature. This observation suggests that the tungsten is never W(IV) and probably not even W(V) under physiological conditions. Similarly, the enzymes AOR from D. gigas and GAPOR from P.furiosus both exhibit W(V) signals in their "as isolated" state (Table 2); these signals disappear upon reduction. The geometrical arrangement of the pterin ligand in P. furiosus AOR in between the cubane and the tungsten has been taken to suggest that the ligand "does not merely play a passive structural role, but may be an active participant in the redox chemistry" [ 11]. Also, observations on molybdenum systems erode the foundations of the central paradigm. Cyclic voltammetry studies on molybdopterin model compounds have revealed multiple redox transitions [132]. In the enzymes studied by crystallography the pterin appears to undergo significant structural changes upon reduction [ 14, 15]. Furthermore, Mo enzymes can be"stabilized" in a nonoxidizable Mo(V) form with substrate analogues [133]. Also, redox chemistry in some molybdenum model compounds has been described as a coupled process of "induced redox reaction": two-electron reduction of a ligand is coupled to oxidation of Mo(V) to Mo(VI) with the net result that single-electron reduction of the complex corresponds to oxidation of the metal [134, 135]. Finally, pterinbased stable radicals have recently been reported for two Mo enzymes [136]. Our summarizing interpretation of the previous is that the redox chemistry at the active site of tungsten enzymes (and at at least some of the molybdenum enzymes) is far from clear. In our view this is the central problem that should be solved as a first step towards an understanding of the mechanism of action of these enzymes. To this goal we formulate five minimum hypotheses for the oxidation-state changes at the metal site: Model-l: the change in oxidation state is fully localized on the metal and E(WS+/W 6+) ~ E(W4+/WS+); this corresponds to the central paradigm of molybdenum biochemistry. Model-2: the catalytic redox chemistry is an interplay of the tungstopterin and a second redox cofactor, e.g., [4Fe-4S] (2+;~+).Model-3: the tungsten does not change oxidation state, it remains, e.g., W(IV); all redox chemistry is pterin-based. Model-4: induced redox chemistry or the multipleelectron reduction of the pterin cofactor is coupled to one-electron oxidation of the metal, e. g., W(V) to W(VI). Model-5: The redox process is localized on the metal, however, it shuttles between W(V) and W(III). The actual redox chemistry of a particular tungsten enzyme may be described by one of the above models or perhaps by a combination of more than one model. We conclude with a few considerations that we expect to be relevant to future studies of this subject. All oxidation states, except VI, are potentially
188
W.R. Hagen. A.F.Arendsen
paramagnetic and can be identified by appropriate spectroscopies. Two pterins will form a softer base ligand system than one pterin, and can be expected to relatively stabilize lower oxidation states of the metal. The majority of purified tungsten enzymes is from thermophilic species. The significance of temperature for the redox chemistry has not been established.
6 Conclusions Tungsten is not a particularly rare element on this planet. Human activity helps in making an ever finer grid of global tungsten distribution. Hexavalent tungsten in the form of the oxoacid tungstate is quite soluble in aqueous solutions at near-neutral to alkaline pH. Tungstate is not a highly reactive chemical; it is structurally similar to oxoacids widely used in biology, sulfate, phosphate, molybdate. Tungsten is a widespread bio-element in the microbial world. In enzymes tungsten is always coordinated by the ligand metallopterin; there is no convincing evidence for a tungsten variant of the heterometal-iron-sulfurhomocitrate cofactor of nitrogenase. Replacement of Mo by W in Mo enzymes usually, but not always, leads to abolishment of activity. W enzymes catalyzed reactions are similar, but not identical to Mo enzymes catalyzed reactions. Some species synthesize a Mo version and a genetically different W version of a catalyst for apparently the same reaction. W enzymes catalyze oxidation of a broad spectrum of aldehydes. The formate dehydrogenase reaction catalyzed by Mo enzymes is reversed under the influence of W enzymes to function in CO2 assimilation. The general concepts from Mo biochemistry of oxo-atom transfer and of coupled electron-proton transfer may be useful for W enzymes too. The redox chemistry at the active site of W enzymes is a key problem. The potentially rich spectroscopy of biological tungsten is still to a large extent to be developed. Tungsten uptake, transport, storage, and W enzyme biosynthesis are essentially virginal areas of research. A function of tungsten in eukaryotic cells has not been established. This does not mean that such a function is not expected to be found. It most certainly does not imply that tungsten is never present in eukaryotes. Some of the microorganisms from which tungsten enzymes have been purified (cf. Table 7) are quite common colonizers of the intestinal flora of animals including humans. The level of tungsten in the feed affects the rumen microbiological population [137]. Tungsten complexes are the subject of many studies in medicine. Antimony polytungstate complexes are active against virus-induced tumors [138]. A tungsten complex of gluthatione activates cAMP-dependent phosphodiesterases [139]. Heteropolytungstates are antiviral, including anti-HIV, agents [140]. Tungstate is an insulin mimic and so are peroxotungstate compounds [141]. The latter prevent deactivation of phosphatases [142]. We do not know the amount of tungsten in a normal human diet, but perhaps the following observation is not a frivolous one: humans require significant amounts of molybdenum; a bottle of wine accounts for some 1/10th of our daily requirement of Mo; contrary to the situation in seawater the amount of W in wine is approximately equal to the amount of Mo [143].
The Bio-lnorganicChemistryof Tungsten
189
Note added in proof: A m a p of h y d r o g e n b o n d s b e t w e e n p r o t e i n a n d cofactor for P. furiosus AOR is p r o v i d e d in [144]. MCD a n d EPR f r o m m u l t i p l e f o r m s of W ( V ) in P.furiosus AOR are r e p o r t e d in [145]. A s y n t h e t i c m o d e l for the active site of AOR f r o m P. furiosus has AOR a c t i v i t y [146]. A gene in Methanopyrus kandleri e n c o d e s a W - F M D H w i t h an active site s e l e n o c y s t e i n e [147]. The crystal s t r u c t u r e for E. coli M o - F D H w i t h selenocysteine is d e s c r i b e d in [148]. 7 References 1. Frafisto da Silva JJR, Williams RJP (1991) "The biological chemistry of the elements; The inorganic chemistry of life". Clarendon, Oxford 2. Stetter KO, Fiala G, Huber G, Huber R, Segerer A (1990) FEMS Microbiol Rev 75:117 3. Charter RA, Tabatabai MA, Schafer JW (1995) Commun Soil Sci Plant Anal 26:3051 4. Andreesen, JR, Ljungdahl LG (1973) J Bacteriol 116:867 5. Andreesen JR, E1 Ghazzawi E, Gottschalk G (1974) Arch Microbiol 96:103 6. Andreesen JR, Ljungdahl LG (1974) J Bacteriol 120:6 7. Ljungdahl LG, Andreesen JR (1975) FEBS Lett 54:279 8. Yamamoto, I, Saiki T, Liu S-M,Ljungdahl LG (1983) J Biol Chem 258:1826 9. Strobl G, Feicht R, White H, Lottspeich F, Simon H (1992) Biol Chem Hoppe-Seyler 373: 123 10. Johnson JL, Rajagopalan KV, Mukund S, Adams MWW (1993) J Biol Chem 268:4848 11. Chan MK, Mukund S, Kletzin A, Adams MWW, Rees DC (1995) Science 267:1463 12. Adams MWW (1994) In: King RB (ed) Encyclopedia of inorganic chemistry. John Wiley, Chichester, p 4284 13. Romao MJ, Archer M, Moura I, Moura JJG, LeGall J, Engh R, Schneider M, Hof P, Huber R (1995) Science 270:1170 14. Schindelin H, Kisker C, Hilton J, Rajagopalan KV, Rees DC (1996) Science 272:1615 15. Huber R, Hof P, Duarte RO, Moura JJG, Moura I, Liu M-Y, LeGall J, Hille R, Archer M, Romao MJ (1996) Proc Natl Acad Sci 93:8846 16. Kletzin A, Adams MWW (1996) FEMS Microbiol Rev 18:5 17. Pourbaix M (ed) (1963) Atlas d'equilibres electrochimiques h 25°C. Gauthier-Villars,Paris 18. Kim J, Rees DC (1992) Science 257:1677 19. Smith BE, Eady RR (1992) Eur J Biochem 205:1 20. Hales BJ, Case EE (1987) J Biol Chem 262:16205 21. Kajii Y, Kobayashi M, Takahashi T, Onodera K (1994) Biosci Biotechnol Biochem 58:1179 22. Golan U, Schneider K, Mfiller A, Schfiddekopf K, Klipp W (1993) Eur J Biochem 215:25 23. Coucouvanis D, Mosier PE, Demadis KD, Patton S, Malinak SM, Kim CG, Tyson MA (1993) J Am Chem Soc 115: 12193; Pickett CJ (1996) J Biol Inorg Chem 1:601 24. Howes BD, Bray RC, Richards RL, Turner NA, Bennett B, Lowe DJ (1996) Biochemistry 35:1432 25. Latimer WM (1952) The oxidation states of the elements and their potentials in aqueous solution, 2nd ed. Prentice-Hall, New York 26. Bard AJ, Parsons R, Jordan J (1985) Standard potentials in aqueous solutions. Dekker, New York 27. McCleverty JA (1994) In: King RB (ed) Encyclopedia of inorganic chemistry. John Wiley, Chichester, p 4240 28. Wulfsberg G (1991) Principles of Descriptive Inorganic Chemistry. University Science Books, Mill Valley, CA 29. Zellner G, Winter J (1987) FEMS Microbiol Lett 40:81 30. Greenwood NN, Earnshaw A (1984) Chemistry of the elements. Pergamon Press, Oxford 31. Cotton FA, Wilkinson G (1988) Advanced inorganic chemistry, 5th ed. John Wiley, New York
190
W.R. Hagen. A. F. Arendsen
32. 33. 34. 35.
Cruywagen JJ, van der Merwe IFJ (1987) J Chem Soc (Dalton Trans) 1701 Silver S, Walderhaug M (1992) Microbiol Rev 56:195 Hider RC (1984) Struct Bonding 58:25 Crichton RR (1991) The inorganic chemistry of iron metabolism. Ellis Horwood, Chichester Rech S, Wolin C, Gunsalus RP (1996) J Biol Chem 271:2557 Luque F, Mitchenall LA, Chapman M, Christine R, Pau RN (1993) Mol Microbiol 7:447 Wang G, Angermiiller S, Klipp W (1993) J Bacteriol 175:3031 Cornish AS, Page W] (1995) Biometals 8:332 Heuwinkel H, Kirkby EA, Le Bot J, Marschner H (1992) J Plant Nutr 15:549 Leonhardt U, Andreesen JR (1977) Arch Microbiol 115:277 Pienkos PT, Brill WJ (1981) J Bacteriol 145:743 Lowe SE, Jain MK, Zeikus JG (1993) Microbiol Rev 57:451 Brown SH, Kelly RM (1989) Appl Environ Microbiol 55:2086 Malik B, Su W-W,Wald HL, Blumentals II, Kelly,RM (1989) Biotechnol Bioeng 34:1050 Fiala G, Stetter KO (1986) Arch Microbiol 145:56 White H, Simon H (1992) Arch Microbiol 158:81 Bryant FO, Adams MWW (1989) J Biol Chem 264:5070 Raven N, Ladwa N, Cossar D, Sharp R (1992) Appl Micorbiol Biotechnol 38:263 Schicho RN, Snowden L], Mukund S, Park J-B, Adams MWW, Kelly RM (1993) Arch Microbiol 159:380 Mukund S, Adams MWW (1996) J Bacteriol 178:163 Wagner R, Andreesen JR (1987) Arch Microbiol 115:277 Hensgens CMH, Nienhuis-Kuiper ME, Hansen TA (1994) Arch Microbiol 162:143 Hart LI, McGartoll MA, Chapman HR, Bray RC (1970) Biochem J 116:851 Cardenas J, Mortenson LE (1974) Anal Biochem 60:372 Olivares JA (1988) Methods Enzymol 158:205 Versieck J (1988)Methods Enzymol 158:267 Pierik, AJ, Hagen WR, Redeker JS, Wolbert RBG, Boersma M, Verhagen MFJM, Grande HJ, Veeger C, Mutsaers PHA, Sands RH, Dunham WR (1992) Eur J Biochem 209:63 Dan D, Re J (1992) Talanta 39:119 Capitan F, Sanchez-Palencia G, Navalon A, Capitan-Vallvey LF, Vilchez JL (1992) Anal Chim Acta 259: 345 Ensafi AA (1992) Anal Lett 25:2339 Kon H, Sharpless, NE (1966) J Phys Chem 70:105 Hanson GR, Brunette AA, McDonell AC, Murray KS, Wedd AG (1981) J Am Chem Soc 103:1953 Hanson GR, Wilson GL, Bailey TD, Pilbrow JR,Wedd AG (1987) J Am Chem Soc 109:2609 Schmitz RA, Albracht SPJ, Thauer RK (1992) FEBS Lett 309:78 Arendsen, AF, de Vocht, M, Bulsink, YBM, Hagen WR (1996) J Biol Inorg Chem 1: 292 Deaton JC, Solomon EI, Watt GD, Wetherbee PJ, Durfor CN (1987) Biochem Biophys Res Commun 149:424 Huber C, Caldeira J, ]ongejan JA, Simon H (1994) Arch Microbiol 162:303 Hensgens CMH, Hagen WR, Hansen TA (1995) J. Bacteriol 177:6195 Johnson JL, Rajagopalan KV (1976) J Biol Chem 251:5505 Lee HC, Gard JK, Brown TL, Oldfield E (1985) J Am Chem Soc 107:4087 Baltzer L (1987) J Am Chem Soc 109:3479 Chung J, Lee HC, Oldfield E (1990) J Magn Reson 90:148 Pierik AJ, Hagen, WR, Dunham WR, Sands RH (1992) Eur J Biochem 206:705 Christina H, McFarlane E, McFarlane W, Rycroft DS (1976) J Chem Soc (Dalton Trans) 1616 Brevard C, Schimpf R (1982) J Magn Reson 47:528 Benn R, Brennecke H, Heck J, Rufinska A (1987) Inorg Chem 26:2826 Benn R, Rufinska A (1988) Magn Reson Chem 26: 895 Benn R, Rufinska A (1989) J Organomet Chem 376:359
36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67.
68. 69. 70. 71. 72. 73. 74. 75. 76. 77.
78. 79.
The Bio-lnorganicChemistryof Tungsten 80. 81. 82. 83.
191
Acerete R, Hammer CF, Baker, LCW (1979) J Am Chem Soc 101:267 Kraatz H-B, Aramini JM, Gao X, Boorman PM, Vogel HJ (1992) Inorg Chem 32:3976 Chapelle S, Verchere JF (1995) Carbohydr Res 266:161 Hlaibi M, Benaissa M, Busatto C, Verchere JF, Chapelle S (1995) Carbohydr Res 278: 227 84. Greenwood NN, Gibb TC (1971) M6ssbauer spectroscopy. Chapman and Hall, London 85. Cramer SP,Liu C-L, Mortenson LE, Spence JT, Liu S-M,Yamamoto I, Ljungdahl LG (1985) J Inorg Biochem 23:119 86. George GN, Prince RC, Mukund S, Adams MWW (1992) J Am Chem Soc 114:3521 87. Mukund S, Adams MWW (1990) J Biol Chem 265:11508 88. Holm RH, Simhon ED (1985) In: Spiro TG (ed) Metal ions in biology Vol 7 Molybdenum enzymes, Ch 1. John Wiley, New York 89. Holm RH (1992) In: Cammack R, Sykes AG (eds) Advances in inorganic chemistry Vol 38 Iron sulfur proteins, Ch 1. Acad Press, San Diego 90. Durfor CN, Wetherbee PJ, Deaton JC, Solomon EI (1983) Biochem Biophys Res Commun 115:61 91. Mukund S, Koehler BP,Roy R, Johnson MK,Adams MWW (1995) J. Inorg Biochem 59:540 92. Koehler BP,Mukund S, Conover RC, Crouse BR, Fu W,Adams MWW,Johnson MK (1995) J Inorg Biochem 59:560 93. Friedebold J, Bowien B (1993) J Bacteriol 175:4719 94. Grether-Beck S, Igloi GL, Pust S, Schilz E, Decker, K, Brandsch R (1994) Mol Microbiol 13:929 95. Chauret C, Knowles R (1991) Can J Microbiol 37:744 96. White H, Strobl G, Feicht R, Simon H (1989) Eur J Biochem 184:89 97. Ljungdahl LG (1986) Ann Rev Microbio140:415 98. Rothery RA, Grant JLS,Johnson JL, Rajagopalan KV,Weiner JH (1995) J Bacterio1177: 2057 99. Le Bloas P, Guilbert N, Loubiere P, Lindley ND (1993) J Gen Microbiol 139:1861 100. Wu WM, Hickey RE Jain MK, Zeikus JG (1993) Arch Microbiol 159:57 101. Bertram PA, Schmitz RA, Linder D, Thauer RK (1994) Arch Microbiol 161:220; Hochheimer A, Schmitz RA, Thauer RK, Hedderich R (1995) Eur J Biochem 234:910 102. Schmitz RA, Richter M, Linder D, Thauer RK (1992) Eur J Biochem 207: 559; Schmitz RA, Albracht SPJ, Thauer RK (1992) Eur J Biochem 209: 1013; Schmitz RA, Bertram PA, Thauer RK (1994) Arch Microhiol 161 : 528 103. Jones JB, Stadtman TC (1981) J Biol Chem 256:656 104. Bertram PA, Karrasch M, Schmitz RA, B6cher R, Albracht SPJ, Thauer RK (1994) Eur J Biochem 220: 477 105. Girio FM, Marcos JC, Amaral-Collaco MT (1992) FEMS Micriobiol Lett 97:161 106. Karzanov VV, Correa CM, Bogatsky YG, Netrusov AI ( 1991) FEMS Microbiol Lett 81 : 95 107. Chauret C, Barraquio WL, Knowles R (1992) Can J Microbiol 38:1042 108. Rosner BM, Schink B (1995) J Bacteriol 177:5767 109. Mukund S, Adams MWW (1995) J Biol Chem 270:8389 110. Mukund S, Adams MWW (1993) J Biol Chem 268:13592 111. Juszczak A, Aono S, Adams MWW (1991) J Biol Chem 266:13834 112. Sajitz P, Klemme JH, Koch HG, Molitor M (1993) Zeitschr Naturforch C Biosci 48:812 113. Fetzner S, Lingens F (1993) Biol Chem Hoppe-Seyler 374:363 114. Shearer G, Schneider JD, Kohl DH (1991) J Gen Microbiol 137:1179 115. Heider J, Ma K, Adams MWW (1995) J Bacteriol 177:4757 116. Timkovich R, Burkhalter RS, Xavier AV,Chen L, LeGall J (1994) Bioorg Chem 22:284 117. Raina R, Bageshwar UK, Das HK (1992) FEMS Microbiol Lett 98:169 118. Huber C, Skopan H, Feicht R, White H, Simon H (1995) Arch Microbiol 164:110 119. Mukund S, Adams MWW (1991) J Biol Chem 266:14208 120. Kengen SWM, Stams AJM (1994) Arch Microbiol 161 : 168 121. Kengen SWM, de Bok FAM,van Loo N-D, Dijkema C, Stams AJM, de Vos WM (1994) J Biol Chem 269:17537 122. Schafer T, Xavier KB, Santos H, Sch6nheit P (1994) FEMS Microbiol Lett 121 : 107
192
W.R. Hagen. A. E Arendsen
123. White H, Feicht R, Huber C, Lottspeich F, Simon H (1991) Biol Chem Hoppe-Seyler 372: 999; White H, Huber C, Feicht R, Simon H (1993) Arch Microbiol 159:244 124. Albracht SPJ (1994) Biochim Biophys Acta 1188:167 125. Burgmayer SJN, Stiefel EI (1985) J Chem Educ 62:943 126. Young CG, Wedd AG (1994) In: King RB (ed) Encyclopedia of inorganic chemistry. John Wiley, Chichester, p 2330 127. Rajagopalan KV, Johnson JL (1992) J Biol Chem 267:10199 128. Kletzin A, Mukund S, Kelley-Crouse TL, Chan MK, Rees DC, Adams MWW (1995) J Bacteriol 177:4817 129. Hille R (1996) J Biol Inorg Chem 1:397 130. Arendsen, AF, de Vocht ML, Bulsink YBM, Hagen WR (1995) J Inorg Biochem 59:542 131. Hille R, MasseyV (1982) J Biol Chem 257:8898 132. Garner CD, Armstrong EM, Ashcroft MJ, Austerberry MS, Birks JH, Collison D, Goodwin AJ, Larsen L, Rowe DJ, Russell JR (1993) In: Stiefel EI, Coucouvanis D, Newton WE (eds) Molybdenum Enzymes, Cofactors, and Model Systems. ACS Symposium Series 535, Ch 7, Washington DC 133. Morpeth FF, Bray RC (1984) Biochemistry 23:1332 134. Coyle CL, Harmer MA, George GN, Daage M, Stiefel EI (1990) Inorg Chem 29:14 135. Murray HH, Wei L, Sherman SE, Greaney MA, Eriksen KA, Cartensen B, Halbert TR, Stiefel EI (1995) Inorg Chem 34:841 136. de Vries S (1996) In: Feiters MC, Hagen WR, Veeger C (eds) Proceedings 3rd European conference on bio-inorganic chemistry. NSR Center, Nijmegen, Abstract G1 137. Takahashi J, Ando S, Watanabe M, Yamamoto N, Fujita H (1993) Animal Sci Technol 64: 637 138. Haiduc I, Silvestru C (1990) Coor Chem Rev 99:253 139. Thompson WJ, Tan BH, Strada S] (1991) J Biol Chem 266:17011 140. Cholewa M, Legge GJF,Weigold H, Holan G, Birch C] (1994) Life Sci 54:1607 141. Bevan AP, Drake PG,¥ale JF, Shaver A, Posner BI (1995) Mol Cell Biochem 153:49 142. Haque SJ, FlatiV, Deb A, Williams BRG (1995) J Biol Chem 270:25709 143. Eschnauer HR, Stoeppler M (1992) In: Stoeppler M (ed) Hazardous metals in the environment. Elsevier, Amsterdam, p 49 144. Kisker C, Schindelin H, Rees DC (1997) Annu Rev Biochem 66:233 145. Koehler BP, Mukund S, Conover RC, Dhawan IK, Roy R, Adams MWW, Johnson MK (1996) J Am Chem Soc 118:12391 146. Das SK, Biswas D, Maiti R, Sarkar S (1996) J Am Chem Soc 118:1387 147. Vorholt ]A, Vaupel M, Thauer RK (1997) Mol Microbio123:1033 148. Boyington JC, Gladyshev VN, Khangulov SV, Stadtman TC, Sun PD (1997) Science 275:1305
Author Index Volumes 1-90
Aegerter MA (1996) Sol-Gel Chromogenic Materials and Devices. 85:149-194 Ahrland S (1966) Factors Contributing to (b)-behavior in Acceptors. 1:207-220 Ahrland S (1968) Thermodynamics of Complex Formation between Hard and Soft Acceptors and Donors. 5:118 - 149 Ahrland S (1973) Thermodynamics of the Stepwise Formation of Metal-Ion Complexes in Aqueous Solution. 15:167-188 Aisen P, see Doi K (1980) 7 0 : 1 - 2 6 Alcock NW, see Leciejewicz ] (1995) 82:43- 84 Allen GC, Warren KD (1971) The Electronic Spectra of the Hexaflnoro Complexes of the First Transition Series. 9:49-138 Allen GC, Warren KD (1974) The Electronic Spectra of the Hexafluoro Complexes of the Second and Third Transition Series. 19:105-165 Alonso ]A, Balbas LC (1993) Hardness of Metallic Clusters. 80:229-258 Alonso ]A, Balb~s LC (1987) Simple Density Functional Theory of the Electronegativity and Other Related Properties of Atoms and Ions. 66:41-78 Andersson LA, Dawson ]H (1991) EXAFS Spectroscopy of Heme-ContainingOxygenases and Peroxidases. 74:1 - 40 Antanaitis BC, see Doi K (1988) 7 0 : 1 - 2 6 Ardon M, Bino A (1987) A New Aspect of Hydrolysis of Metal Ions: The Hydrogen-Oxide Bridging Ligand (H30~2). 65:1 - 28 Arendsen AF, see Hagen WR (1998) 90:161 - 192 Armstrong FA (1990) Probing Metalloproteins by Voltammetry. 7 2 : 1 3 7 - 221 Augustynski ] (1988) Aspects of Photo-Electrochemical and Surface Behavior of Titanium(IV) Oxide. 69:1-61 Auld DS (1997) Zinc Catalysis in Metalloproteases. 89: 29-50 Averill BA (1983) Fe-S and Mo-Fe-S Clusters as Models for the Active Site of Nitrogenase. 53:57-101
Babel D (1967) Structural Chemistry of Octahedral Fluorocomplexes of the Transition Elements. 3:1 - 87 Bacci M (1984) The Role of Vibronic Coupling in the Interpretation of Spectroscopic and Structural Properties of Biomolecules. 55: 6 7 - 99 Baekelandt BG, Mortier W], Schoonheydt RA (1993) The EEM Approach to Chemical Hardness in Molecules and Solids: Fundamentals and Applications. 8 0 : 1 8 7 - 2 2 8 Baker EC, Halstead GW, Raymond KN (1976) The Structure and Bonding of 4f and 5f Series Organometallic Compounds. 25: 21 - 66 Balbfis LC, see Alonso ]A (1987) 66:41-78 Balb~isLC, see Alonso ]A (1993) 80:229-258 Baldwin AH, see Butler A (1997) 89:109-132 Balsenc LR (1980) Sulfur Interaction with Surfaces and Interfaces Studied by Auger Electron Spectrometry. 39:83 - 114
194
Author Index Volumes 1 - 90
Banci L, Bencini A, Benelli C, Gatteschi D, Zanchini C (1982) Spectral-Structural Correlations in High-Spin Cobalt(II) Complexes. 52: 37- 86 Banci L, Bertini I, Luchinat C (1990) The 1H NMR Parameters of Magnetically Coupled Dimers - The Fe2S2 Proteins as an Example. 72:113-136 Baran EJ, see Mfiller A (1976) 26:81 - 139 Bartolotti LJ (1987) Absolute Electronegativities as Determined from Kohn-Sham Theory. 66:27-40 Ban RG, see Teller R (1981) 4 4 : 1 - 8 2 Baughan EC (1973) Structural Radii, Electron-cloud Radii, Ionic Radii and Solvation. 15:53-71 Bayer E, Schretzmann P (1967) Reversible Oxygeniernng von Metallkomplexen. 2:181-250 Bearden A], Dunham WR (1970) Iron Electronic Configuration in Proteins: Studies by M6ssbauer Spectroscopy. 8:1 - 52 Bencini A, see Banci L (1982) 52:37-86 Benedict U, see Manes L (1985) 59/60:75 - 125 Benelli C, see Banci L (1982) 52:37-86 Benfield RE, see Thiel RC (1993) 81 : 1-40 Bergmann D, Hinze J (1987) Electronegativity and Charge Distribution. 66:145 - 190 Berners-Price SJ, Sadler PJ (1988) Phosphines and Metal Phosphine Complexes: Relationship of Chemistry to Anticancer and Other Biological Activity. 70: 27-102 Bertini I, see Banci L (1990) 72:113-136 Bertini I, Ciurli S, Luchinat C (1995) The Electronic Structure of FeS Centers in Proteins and Models. A Contribution to the Understanding of Their Electron Transfer Properties. 83:1-54 Bertini I, Luchinat C, Scozzafava A (1982) Carbonic Anhydrase: An Insight into the Zinc Binding Site and into the Active Cavity Through Metal Substitution. 48:45-91 Bertrand P (1991) Application of Electron Transfer Theories to Biological Systems. 7 5 : 1 - 4 8 Bill E, see Trautwein AX (1991) 7 8 : 1 - 9 6 Bino A, see Ardon M (1987) 6 5 : 1 - 2 8 Blanchard M, see Linar~s C (1977) 33:179-207 Blasse G, see Powe11RC (1980) 42:43-96 Blasse G (1991) Optical Electron Transfer Between Metal Ions and its Consequences. 76:153-188 Blasse G (1976) The Influence of Charge-Transfer and Rydberg States on the Luminescence Properties of Lanthanides and Actinides. 26:43 -79 Blasse G (1980) The Luminescence of Closed-Shell Transition Metal-Complexes. New Developments. 42:1 -41 Blauer G (1974) Optical Activity of Conjugated Proteins. 18: 69 - 129 Bleijenberg KC (1980) Luminescence Properties of Uranate Centres in Solids. 42: 97-128 Boca R, Breza M, Pelik~n P (1989) Vibronic Interactions in the Stereochemistry of Metal Complexes. 71 : 57- 97 Boeyens JCA (1985) Molecular Mechanics and the Structure Hypothesis. 63:65-101 B6hm MC, see Sen KD (1987) 66:99-123 Bohra R, see Jain VK (1982) 52:147-196 Bollinger DM, see Orchin M (1975) 23:167-193 Bominaar EL, see Trautwein AX (1991) 7 8 : 1 - 9 6 Bonnelle C (1976) Band and Localized States in Metallic Thorium, Uranium and Plutonium, and in Some Compounds, Studied by X-ray Spectroscopy. 31 : 23 - 48 Bose SN, see Nag K (1985) 63:153-197 Bowler BE, see Therien MJ (1991) 75:109-130 Bradshaw AM, Cederbaum LS, Domcke W (1975) Ultraviolet Photoelectron Spectroscopy of Gases Adsorbed on Metal Surfaces. 24:133-170 Braterman PS (1972) Spectra and Bonding in Metal Carbonyls. Part A: Bonding. 10:57-86 Braterman PS (1976) Spectra and Bonding in Metal Carbonyls. Part B: Spectra and Their Interpretation. 26:1 - 4 2
AuthorIndexVolumes1-90
195
Bray RC, Swann JC (1972) Molybdenum-ContainingEnzymes. 11 : 107-144 Brec R, see Evain M (1992) 79:277-306 Brese NE, O'Keeffe M (1992) Crystal Chemistry of Inorganic Nitrides. 79: 307 - 378 Breza M, see Boca R (1989) 71:57-97 Briggs LR, see Kustin K (1983) 53:137-158 Brooks MSS (1985) The Theory of 5f Bonding in Actinide Solids. 59/60:263 -293 Brown DG, see Wood JM (1972) 11:47-105 Buchanan BB (1966) The Chemistry and Function of Ferredoxin. 1: 109-148 Bucher E, see Campagna M (1976) 30: 99-140 Buchler JW, Dreher C, Kiinzel FM (1995) Synthesis and Coordination Chemistry of Noble Metal Porphyrins. 84:1 - 7 0 Buchler JW, Koldsch W, Smith PD (1978) Cis, Trans, and Metal Effects in Transition Metal Porphyrins. 34: 79-134 Bulman RA (1978) Chemistry of Plutonium and the Transuranics in the Biosphere. 34: 39- 77 Bulman RA (1987) The Chemistry of Chelating Agents in Medical Sciences. 67:91 - 141 Burdett JK (1987) Some Structural Problems Examined Using the Method of Moments. 65:29-90 Burdett JK (1976) The Shapes of Main-Group Molecules: A Simple Semi-Quantitative Molecular Orbital Approach. 31:67-105 Butler A, Baldwin AH (1997) Vanadium Bromperoxidase and Functional Mimics. 89:109-132
Campagna M, Wertheim GK, Bucher E (1976) Spectroscopy of Homogeneous Mixed Valence Rare Earth Compounds. 30: 99-140 Capozzi F, Ciurli S, Luchinat C (1998) Coordination Sphere Versus Protein Environment as Determinants of Electronic and Functional Properties of Iron-Sulfur Proteins 90:127-160 Carter RO, see Mtiller A (1976) 26:81 - 139 Cauletti C, see Furlani C (1978) 35:119 - 169 Cederbaum LS, see Bradshaw AM (1975) 24:133-170 Cederbaum LS, see Schmelcher PS (1996) 86: 27-62 Ceulemans A, Vanquickenborne LG (1989) The Epikernel Principle. 71 : 125 - 159 Chandrasekhar V, Thomas KR, Justin KR (1993) Recent Aspects of the Structure and Reactivity of Cyclophosphazenes. 81 :41 - 114 Chandrashekar TK, see Ravikanth M (1995) 82:105 - 188 Chang J, see Therien MJ (1991) 75:109-130 Chapman SK, Daft S, Munro AW (1997) Heme: The Most Versatile Redox Centre in Biology? 88:39-70 Chasteen ND (1983) The Biochemistry of Vanadium. 53:103 - 136 Chattaraj PK, Parr RG (1993) Density Functional Theory of Chemical Hardness. 80:11 -26 Cheh AM, Neilands JP (1976) The y-AminoevulinateDehydratases: Molecular and Environmental Properties. 29:123 - 169 Chimiak A, Neilands JB (1984) Lysine Analogues of Siderophores. 58: 89- 96 Christensen JJ, see Izatt RM (1973) 16:161 - 189 Ciampolini M (1969) Spectra of 3d Five-Coordinate Complexes. 6:52-93 Ciurli S, see Bertini I (1995) 83:1-54 Ciurli S, see Capozzi F (1998) 90:127-160 Clack DW,Warren KD (1980) Metal-Ligand Bonding in 3d Sandwich Complexes. 39:1 - 141 Clarke MJ, Fackler PH (1982) The Chemistry of Technetium: Toward Improved Diagnostic Agents. 50:57- 58 Clarke MJ, Gaul JB (1993) Chemistry Relevant to the Biological Effects of Nitric Oxide and Metallonitrosyls. 81 : 147-181 Clarke RJH, Stewart B (1979) The Resonance Raman Effect. Review of the Theory and of Applications in Inorganic Chemistry. 36:1 -80 Codling K, Frasinski LJ (1996) Molecules in Intense Laser Fields: an ExperimentalViewpoint. 86:1-26
196
Author Index Volumes 1- 90
Cohen IA (1980) Metal-Metal Interactions in Metalloporphyrins, Metalloproteins and Metalloenzymes. 40:1 - 37 Connett PH, Wetterhahn KE (1983) Metabolism of the Carcinogen Chromate by Cellular Constituents. 54: 93 - 124 Cook DB (1978) The Approximate Calculation of Molecular Electronic Structures as a Theory of Valence. 35: 37- 86 Cooper SR, Rawle SC (1990) Crown Thioether Chemistry. 72:1 -72 Corbett JD (1997) Diverse Naked Clusters of the Heavy Main-Group Elements. Electronic Regularities and Analogies. 87:157-194 Cotton FA,Walton RA (1985) Metal-Metal Mukiple Bonds in Dinudear Clusters. 6 2 : 1 - 4 9 Cox PA (1975) Fractional Parentage Methods for Ionisation of Open Shells of d and f Electrons. 24:59-81 Cox MC, see Sun H (1997) 88: 71 - 102 Cras JA, see Willemse J (1976) 28: 83 - 126 Cremer D, see Frenking G (1990) 73:17-96 Crichton RR (1973) Ferritin. 17:67-134
DaffS, see Chapmann SK (1997) 88:39-70 Dance J-M, see Tressaud A (1982) 52: 87-146 Darriet ], see Drillon M (1992) 79: 55-100 Daul C, Schl~ipfer CW, yon Zelewsky A (1979) The Electronic Structure of Cobalt(II) Complexes with Schiff Bases and Related Ligands. 36:129-171 Dawson JH, see Andersson LA (1991) 7 4 : 1 - 4 0 Deeth R] (1995) Computational Modelling of Transition Metal Centres. 82:1 - 42 Degen J, see Schmidtke H-H (1989) 71:99-124 Dehnicke K, Shihada A-F (1976) Structural and Bonding Aspects in Phosphorus ChemistryInorganic Derivates of Oxohalogeno Phosphoric Acids. 28:51 - 82 Denning RG (1992) Electronic Structure and Bonding in Actinyl Ions. 79:215- 276 DhubhghaiU OMN, Sadler PJ (1991) The Structure and Reactivity of Arsenic Compounds. Biological Activity and Drug Design. 78:129-190 Diehn B, see Doughty MJ (1980) 41:45 -70 Diemann E, see Miiller A (1973) •4:23 -47 Dirken MW, see Thiel RC (1993) 81 : 1-40 Dobifi~ B (1984) Surfactant Adsorption on Minerals Related to Flotation. 56:91 - 147 Doi K, Antanaitis BC, Aisen P (1988) The Binuclear Iron Centers of Uteroferrin and the Purple Acid Phosphatases. 70:1 -26 Domcke W, see Bradshaw AM (1975) 24:133 - 170 Dophin D, see Morgan B (1987) 64:115 -204 Doughty MJ, Diehn B (1980) Flavins as Photoreceptor Pigments for Behavioral Responses. 4 1 : 4 5 - 70 Drago RS (1973) Quantitative Evaluation and Prediction of Donor-Acceptor Interactions. 15: 73-139 Dreher C, see Buchler JW (1995) 8 4 : 1 - 7 0 Drillon M, Darriet J (1992) Progress in polymetallic Exchange-Coupled Systems, some Examples in Inorganic Chemistry. 79: 55-100 Duffy JA (1977) Optical Electronegativity and Nephelauxetic Effect in Oxide Systems. 32:147-166
Dunham WR, see Bearden AJ (1970) 8:1 - 52 Dunn MF (1975) Mechanisms of Zinc Ion Catalysis in Small Molecules and Enzymes. 23: 61 - 122 Eatough DJ, see Izatt RM (1973) 16:161 - 189 Eller PG, see Ryan RR (1981) 46:47-100 Emmerling A, see Fricke J (1991) 77: 3 7 - 88
AuthorIndexVolumes1-90
] 97
Emsley E (1984) The Composition, Structure and Hydrogen Bonding of the fl-Diketones. 57:147-191
Englman R (1981) Vibrations in Interaction with Impurities. 43:113-158 Epstein IR, Kustin K (1984) Design of Inorganic Chemical Oscillators. 56:1 - 33 Ermer 0 (1976) Calculations of Molecular Properties Using Force Fields. Applications in Organic Chemistry. 27:161 - 211 Ernst RD (1984) Structure and Bonding in Metal-Pentadienyl and Related Compounds. 57:1-53 Erskine RW, Field BO (1976) Reversible Oxygenation. 2 8 : 1 - 5 0 Evain M, Brec R (1992) A New Approach to Structural Description of Complex Polyhedra Containing Polychalcogenide Anions. 79: 277- 306
Fackler PH, see Clarke MJ (1982) 5 0 : 5 7 - 5 8 Fajans K (1967) Degrees of Polarity and Mutual Polarization of I0ns in the Molecules of Alkali Fluorides, SrO and BaO. 3:88-105 Fan M-E see Lin Z (1997) 87:35-80 Fee JA (1975) Copper Proteins - Systems Containing the "Blue" Copper Center. 23:1 - 60 Feeney RE, Komatsu SK (1966) The Transferrins. 1 : 149- 206 Fehlner TP (1997) Metalloboranes. 87:111 - 136 Felsche J (1973) The Crystal Chemistry of the Rare-Earth Silicates. 13: 99 - 197 Ferreira R (1976) Paradoxical Violations of Koopmans' Theorem, with Special Reference to the 3d Transition Elements and the Lanthanides. 31 : 1 -21 Fichtinger-SchepmanAMJ, see Reedijk J (1987) 67: 53-89 Fidelis IK, Mioduski T (1981) Double-Double Effect in the Inner Transition Elements. 47:27-51
Field BO, see Erskine RW (1976) 2 8 : 1 - 5 0 Fischer J, see Mathey F (1984) 55:153 -201 Follmann H, see Lammers M (1983) 54:27-91 Fournier JM, Manes L (1985) Actinide Solids. 5f Dependence of Physical Properties. 59/60:1 - 56 Fournier JM (1985) Magnetic Properties of Actinide Solids. 59/60:127-196 Fraga S,Valdemoro C (1968) Quantum Chemical Studies on the Submolecular Structure of the Nucleic Acids. 4:1 - 62 Frasinski LJ, see Codling K (1996) 8 5 : 1 - 2 6 Frafisto da Silva JJR, Williams RJP (1976) The Uptake of Elements by Biological Systems. 29:67-121
Frenking G, see Jorgensen CK (1990) 73:1 - 16 Frenking G, Cremer D (1990) The Chemistry of the Noble Gas Elements Helium, Neon, and Argon - Experimental Facts and Theoretical Predictions. 73:17- 96 Frey M (1998) Nickel-Iron Hydrogenases: Structural and Functional Properties. 90:97-126 Fricke B (1975) Superheavy Elements. 21:89-144 Fricke J, Emmerling A (1991) Aerogels-Preparation, Properties, Applications. 77: 3 7 - 88 Friebel C, see Reinen D (1979) 37:1-60 Friedrich H (1996) Field Induced Chaos and Chaotic Scattering. 86: 9 7 - 1 2 4 Friesen C, see Keppler BK (1991) 78: 97-128 Fuhrhop J-H (1974) The Oxidation States and Reversible Redox Reactions of Metalloporphyrins. 18:1 - 67 Furlani C, Cauletti C (1978) He(I) Photoelectron Spectra of d-metal Compounds. 35:119-169 Gani D, Wilkie J (1997) Metal Ions in the Mechanism of Enzyme Catalysed Phosphate Monoester Hydrolyses. 89:133-176 Gallagher TF (1996) Microwave Mukiphoton Exitation and Ionization. 86:125 - 148 Galland P, see Russo VEA (1980) 41:71 - 110 Galvfin M, see Gfizquez JL (1987) 6 6 : 7 9 - 9 8
198
Author Index Volumes 1 - 90
Gatteschi D, see Banci L (1982) 5 2 : 3 7 - 8 6 Gaul JB, see Clarke MJ (1993) 81 : 147-181 Gavezzotti A, see Simonetta M (1976) 27:1 -43 Gfizquez JL, Vela A, Galv~inM (1987) Fukui Function, Electronegativity and Hardness in the Kohn-Sham Theory. 66: 7 9 - 98 Gazqu~z JL (1993) Hardness and Softness in Densitiy Functional Theory. 8 0 : 2 7 - 4 4 Gerloch M, Harding JH, Woolley RG (1981) The Context and Application of Ligand Field Theory. 4 6 : 1 - 4 6 Ghijsen J, see Naegele JR (1985) 5 9 / 6 0 : 1 9 7 - 2 6 2 Gilbert TR, see Kustin K (1983) 5 3 : 1 3 7 - 1 5 8 Gillard RD, Mitchell PR (1970) The Absolute Configuration of Transition Metal Complexes. 7: 46- 86 Gleitzer C, Goodenough JB (1985) Mixed-Valence Iron Oxides. 61:1 - 7 6 Gliemann G, Yersin H (1985) Spectroscopic Properties of the Quasi One-Dimensional Tetracyanoplatinate(II) Compounds. 62: 8 7 - 1 5 3 Golovina AP, Zorov NB, Runov VK (1981) Chemical Luminescence Analysis of Inorganic Substances. 47:53 - 119 Goodenough JB, see Gleitzer C (1985) 61 : 1 - 7 6 Gr~itzelM, see Kiwi J (1982) 49: 37-125 Gray HB, see Therien M] (1991) 7 5 : 1 0 9 - 1 3 0 Green ]C (1981) Gas Phase Photoelectron Spectra of d- and f-Block Organometallic Compounds. 43: 37-112 Grenier JC, Pouchard M, Hagenmuller P (1981) Vacancy Ordering in Oxygen-Deficient Perovskite-Related Ferrites. 47:1 -25 Grice ME, see Politzer P (1993) 80:101 - 114 Griffith JS (1972) On the General Theory of Magnetic Susceptibilities of Polynuclear Transitionmetal Compounds. 10: 8 7 - 1 2 6 Grisham CM, see Mildvan AS (1974) 2 0 : 1 - 2 1 Gubelmann MH, Williams AF (1984) The Structure and Reactivity of Dioxygen Complexes of the Transition Metals. 55:1 - 65 Gtidel HU, see Ludi A (1973) 1 4 : 1 - 2 1 Guilard R, Lecomte C, Kadish KM (1987) Synthesis, Electrochemistry, and Structural Properties of Porphyrins with Metal-Carbon Single Bonds and Metal-Metal Bonds. 64: 205 - 268 Guillaumont R, see Hubert S (1978) 34:1 - 18 Gtitlich P (1981) Spin Crossover in Iron(II)-Complexes. 44: 83-195 Gutmann V, see Mayer U (1972) 12:113 - 140 GutmannV,Mayer H (1976) Application of the FunctionalApproach to Bond Variations Under Pressure. 31 :49- 66 Gutmann V, Mayer U (1973) Redox Properties: Changes Effected by Coordination. 15:141 - 166 Gutmann V, Mayer U (1972) Thermochemistry of the Chemical Bond. 1 0 : 1 2 7 - 1 5 1
H~ider D-P, see Nuksch W (1980) 41 : 111 - 139 Hagen WR, Arendsen AF (1998) The Bio-Inorganic Chemistry of Tungsten. 90:161 - 192 Hagenmuller P, see Grenier ]C (1981) 47:1 -25 Hale JD, see Williams RJP (1966) 1:249-281 Hale ]D, see Williams RJP (1973) 15:1 and 2 Halet J-F,Saillard ]-Y (1997) Electron Count Versus StructuralArrangement in Clusters Based on a Cubic Transition Metal Core with Bridging Main Group Elements. 87:81 - 110 Hall DI, Ling ]H, Nyholm RS (1973) Metal Complexes of Chelating Olefin-Group V Ligands. 15:3-51 Halstead GW, see Baker EC (1976) 25:21-66 Hamstra BJ, see Slebodnick C (1997) 89:51 - 108
Author Index Volumes 1 - 90
199
Hanack M, see Schultz H (1991) 74:41 - 146 Harding ]H, see Gerloch M (1981) 46:1-46 Harnung SE, Sch~iffer CE (1972) Phase-fixed 3-G Symbols and Coupling Coefficients for the Point Groups. 12:201 - 255 Harnung SE, Sch~ffer CE (1972) Real Irreducible Tensorial Sets and their Application to the Ligand-Field Theory. 12: 257- 295 Hathaway B] (1984) A New Look at the Stereochemistry and Electronic Properties of Complexes of the Coppes (II) Ion. 57:55 - 118 Hathaway B] (1973) The Evidence for "Out-of-the Plane" Bonding in Axial Complexes of the Copper(II) Ion. 14: 49- 67 Hawes ]C, see Mingos DMP (1985) 63:1-63 Hellner EE (1979) The Frameworks (Bauverb~nde) of the Cubic Structure Types. 37: 61-140 Hemmerich P, Michel H, Schug C, Massey V (1982) Scope and Limitation of Single Electron Transfer in Biology.48:93-124 Henry M, ]olivet JP, Livage J (1991) Aqueous Chemistry of Metal Cations: Hydrolysis, Condensation and Complexation. 77:153- 206 Hider RC (1984) Siderophores Mediated Absorption of Iron. 57: 25- 88 Hill HAO, R6der A, Williams RIP (1970) The Chemical Nature and Reactivity of Cytochrome P-450.8:123-151 Hilpert K (1990) Chemistry of Inorganic Vapors. 73: 97-198 Hinze ], see Bergmann D (1987) 66:145 - 190 Hoffman BM, Natan M], Nocek JM, Wallin SA (1991) Long-Range Electron Transfer Within Metal-Substituted Protein Complexes. 75: 85-108 Hoffmann BM, see Ibers ]A (1982) 50:1-55 Hoffmann DK, Ruedenberg K, Verkade ]G (1977) Molecular Orbital Bonding Concepts in Polyatomic Molecules - A Novel Pictorial Approach. 33: 57- 96 Hogenkamp HPC, Sando GN (1974) The Enzymatic Reduction of Ribonucleotides. 20:23-58 Housecroft CE (1997) Clusters with Interstitial Atoms from the p-Block: How Do Wade's Rules Handle Them? 87:137-156 Huber R, see Romao M] (1998) 90:69-96 Hubert S, Hussonois M, Guillaumont R (1978) Measurement of Complexing Constants by Radiochemical Methods. 34:1 - 18 Hudson RF (1966) Displacement Reactions and the Concept of Soft and Hard Acids and Bases. 1:221-223 Hulliger F (1968) Crystal Chemistry of Chalcogenides and Pnictides of the Transition Elements. 4:83 - 229 Hussonois M, see Hubert S (1978) 34:1 - 18 Hyde BG, see Makovicky E (1981) 46:101 - 170 Hyde BG, see O'Keeffe M (1985) 61:77-144
Ibers JA, Pace LJ, Martinsen J, Hoffmann BM (1982) Stacked Metal Complexes: Structures and Properties. 50:1 -55 Ingraham LL, see Maggiora GM (1967) 2:126-159 Iqbal Z (1972) Intra- and Inter-Molecular Bonding and Structure of Inorganic Pseudohalides with Triatomic Groupings. I0: 25 - 55 Izatt RM, Eatough DJ, Christensen J] (1973) Thermodynamics of Cation-Macrocyclic Compound Interaction. 16:161 - 189
Jain VK, Bohra R, Mehrotra RC (1982) Structure and Bonding in Organic Derivatives of Antimony(V). 52:147 -196 ]erome-Lerutte S (1972) Vibrational Spectra and Structural Properties of Complex Tetracyanides of Platinum, Palladium and Nickel. 10:153 - 166
200
Author Index Volumes 1 - 90
Johnston RL (1997) Mathematical Cluster Chemistry. 87:1 -34 Johnston RL, see Mingos DMP (1987) 68:29-87 Jolivet JP, see Henry M (1991) 77:153 -206 Jorgensen CK, see Miiller A (1973) 14:23-47 Jorgensen CK, see Reisfeld R (1982) 49:1-36 Jorgensen CK, see Reisfeld R (1988) 69:63-96 Jorgensen CK, see Reisfeld R (1991) 77:207-256 Jorgensen CK, Frenking G (1990) Historical, Spectroscopic and Chemical Comparison of Noble Gases. 73:1 - 16 Jorgensen CK, Kauffmann GB (1990) Crookes and Marignac - A Centennial of an Intuitive and Pragmatic Appraisal of "Chemical Elements" and the Present Astrophysical Status of Nucleosynthesis and "Dark Matter". 73: 227- 254 Jorgensen CK, Reisfeld R (1982) Uranyl Photophysics. 50:121 - 171 Jorgensen CK (1976) Deep-LyingValnce Orbitals and Problems of Degeneracy and Intensitites in Photo-Electron Spectra. 30:141 - 192 Jorgensen CK (1966) Electric Polarizability, Innocent Ligands and Spectroscopic Oxidation States. 1 : 234- 248 Jorgensen CK (1990) Heavy Elements Synthesized in Supernovae and Detected in Peculiar A-type Stars. 73:199- 226 Jorgensen CK (1996) Luminescence of Cerium(III) Inter-Shell Transitions and Scintillator Action. 85:195-214 Jorgensen CK (1976) Narrow Band Thermoluminescence (Candoluminescence) of Rare Earths in Auer Mantles. 25:1 -20 Jorgensen CK (1975) Partly Filled Shells Constituting Anti-bonding Orbitals with Higher Ionization Energy than Their Bonding Counterparts. 22: 49- 81 Jorgensen CK (1975) Photo-Electron Spectra of Non-Metallic Solids and Consequences for Quantum Chemistry. 24:1 - 58 Jorgensen CK (1978) Predictable Quarkonium Chemistry. 34:19-38 Jorgensen CK (1966) Recent Progress in Ligand Field Theory. 1:3-31 Jorgensen CK (1967) Relationship Between Softness, Covalent Bonding, Ionicity and Electric Polarizability.3:106-115 Jorgensen CK (1981 ) The Conditions for Total Symmetry Stabilizing Molecules, Atoms, Nuclei and Hadrons. 43:1-36 Jorgensen CK (1973) The Inner Mechanism of Rare Earths Elucidated by Photo-Electron Spectra. 13:199-253 Jorgensen CK (1969) Valence-ShellExpansion Studied by Ultra-violet Spectroscopy. 6: 94-115 Justin KR, see Chandrasekhar V (1993) 81:41 - 114 Kadish KM, see Guilard R (1987) 64:205-268 Kahn O (1987) Magnetism of the Heteropolymetallic Systems. 68: 89 - 167 Kalyanasundaram K, see Kiwi J (1982) 49:37-125 Kauffmann GB, see Jorgensen CK (1990) 73:227-254 Keijzers CP, see Willemse J (1976) 28:83-126 Kemp TJ, see Leciejewicz J (1995) 82:43- 84 Keppler BK, Friesen C, Moritz HG, Vongerichten H, Vogel E (1991) Tumor-Inhibiting Bis (fl-Diketonato) Metal Complexes. Budotitane, cis-Diethoxybis (1-phenylbutane-1,3-dionato ) titanium(IV). 78: 97-128 Kimura E, Koike T, Shionoya M (1997) Advances in Zinc Enzyme Models by Small, Mononuclear Zinc(II) Complexes. 89:1 -28 Kimura T (1968) Biochemical Aspects of Iron Sulfur Linkage in None-Heine Iron Protein, with Special Reference to "Adrenodoxin". 5:1-40 Kitagawa T, Ozaki Y (1987) Infrared and Raman Spectra of Metalloporphyrins. 64: 71 - 114 Kiwi J, Kalyanasundaram K, Gratzel M (1982) Visible Light Induced Cleavage of Water into Hydrogen and Oxygen in Colloidal and Microheterogeneous Systems. 49: 37-125
Author Index Volumes 1-90
201
Kjekshus A, Rakke T (1974) Considerations on the Valence Concept. 19: 45 - 83 Kjekshus A, Rakke T (1974) Geometrical Considerations on the Marcasite Type Structure. 19:85-104 Klabunde T, Krebs B (1997) The Dimetal Center in Purple Acid Phosphatases. 89:177-198 Koike T, see Kimura E (1997) 8 9 : 1 - 2 8 Kokisch W, see Buchler JW (1978) 34:79-134 Komatsu SK, see Feeney RE (1966) 1 : 149 -206 Komorowski L (1993) Hardness Indices for Free and Bonded Atoms. 80:45-70 K6nig E (1991) Nature and Dynamics of the Spin-State Interconversions in Metal Complexes. 76:51-152 K6nig E (1971) The Nephelauxelic Effect. Calculation and Accuracy of the Interelectronic Repulsion Parameters I. Cubic High-Spin d2, d3, d7 and d8 Systems. 9:175-212 K6pf H, see K6pf-Maier P (1988) 7 0 : 1 0 3 - 1 8 5 K6pf-Maier P, K6pf H (1988) Transition and Main-Group Metal Cyclopentadienyl Complexes: Preclinical Studies on a Series of Antitumor Agents of Different Structural Type. 70:103-185 Koppikar DK, Sivapullaiah PV, Ramakrishnan L, Soundararajan S (1978) Complexes of the Lanthanides with Neutral Oxygen Donor Ligands. 34:135 - 213 K6ren B, see Valach F (1984) 55:101 - 151 Krause R (1987) Synthesis of Ruthenium(II) Complexes of Aromatic Chelating Heterocycles: Towards the Design of Luminescent Compounds. 67:1 -52 Krebs B, see Klabunde T (1997) 8 9 : 1 7 7 - 1 9 8 Krumholz P (1971) Iron(II) Diimine and Related Complexes. 9:139 - 174 Kubas GJ, see Ryan RR (1981) 46:47-100 Kuki A (1991) Electronic TunnelingPaths in Proteins. 75:49-84 Kulander KC, Schafer KJ (1996) Time-Dependent Calculations of Electron and Photon Emission from an Atom in an Intense Laser Field. 86:149 - 172 Kiinzel FM, see Buchler JW (1995) 8 4 : 1 - 7 0 Kustin K, see Epstein IR (1984) 56:1 - 33 Kustin K, McLeod GC, Gilbert TR, Briggs LR (1983) Vanadium and Other Metal Ions in the Physiological Ecology of Marine Organisms. 5 3 : 1 3 7 - 1 5 8
Labarre IF (1978) Conformational Analysis in Inorganic Chemistry: Semi-Empirical Quantum Calculation vs. Experiment. 35:1 - 35 Lammers M, Follmann H (1983) The Ribonucleotide Reductases: A Unique Group of Metalloenzymes Essential for Cell Proliferation. 54:27-91 Le Brun NE, Thomson AJ, Moore GR (1997) Metal Centres of Bacterioferritins or Non-HeamIron-ContainingCytochromes b557. 8 8 : 1 0 3 - 1 3 8 Leciejewicz J, Alcock NW, Kemp TJ (1995) Carboxylato Complexes of the Uranyl Ion: Effects of Ligand Size and Coordinat. Geometry Upon Molecular and Crystal Structure. 82: 43- 84 Lecomte C, see Guilard R (1987) 64:205-268 Lee YJ, see Scheidt WR (1987) 64:1 - 70 Lehmann H, see Schultz H (1991) 74:41 - 146 Lehn ]-M (1973) Design of Organic Complexing Agents. Strategies Towards Properties. 16:1-69
Li H, see Sun H (1997) 88:71 - 102 Licoccia S, Paolesse R (1995) Metal Complexes of Corroles and Other Corrinoids. 84:71-134 Lin Z, Fan M-F (1997) Metal-Metal Interactions in Transition Metal Clusters with n-Donor Ligands. 87: 35- 80 Linar~s C, Louat A, Blanchard M (1977) Rare-Earth Oxygen Bonding in the LnMO4 Xenotime Structure. 33:179 -207 Lindskog S (1970) Cobalt(II) in Metalloeuzymes. A Reporter of Structure-FunctionRelations. 8:153-196
202
Author Index Volumes 1- 90
Ling JH, see Hall DI (1973) 15:3-51 Liu A, Neilands JB (1984) Mutational Analysis of Rhodotorulic Acid Synthesis in Rhodotorula philimanae. 58: 97-106 Livage J, see Henry M (1991) 77:153-206 Livorness ], Smith T (1982) The Role of Manganese in Photosynthesis. 48:1 -44 Llin~isM (1973) Metal-Polypeptide Interactions: The Conformational State of Iron Proteins. 17:135 -220 Louat A, see Linar~s C (1977) 33:179-207 Luchinat C, see Banci L (1990) 72:113-136 Luchinat C, see Bertini I (1982) 48:45 -91 Luchinat C, see Bertini I (1995) 83:1- 54 Luchinat C, see Capozzi F (1998) 90:127-160 Lucken EAC (1969) Valence-Shell Expansion Studied by Radio-Frequency Spectroscopy. 6:1-29 Ludi A, Giidel HU (1973) Structural Chemistry of Polynuclear Transition Metal Cyanides. 14:1-21 Lutz HD (1988) Bonding and Structure of Water Molecules in Solid Hydrates. Correlation of Spectroscopic and Structural Data. 69:125 Lutz HD (1995) Hydroxide Ions in Condensed Materials - Correlation of Spectroscopy and Structural Data. 82:85-104
Maaskant WJA (1995) On Helices Resulting from a Cooperative Iahn-Teller Effect in Hexagoal Perovskites. 83: 55 - 88 Maggiora GM, Ingraham LL (1967) Chlorophyll Triplet States. 2:126-159 Magyar B (1973) Salzebullioskopie III. 14:111 - 140 Makovicky E, Hyde BG (1981) Non-Commensurate (Misfit) Layer Structures. 46:101 - 170 Manes L, see Fournier JM (1985) 59/60:1 -56 Manes L, Benedict U (1985) Structural and Thermodynamic Properties of Actinide Solids and Their Relation to Bonding. 59/60:75-125 Mann S (1983) Mineralization in Biological Systems. 54:125 - 174 March NH (1993) The Ground-State Energy of Atomic and Molecular Ions and Its Variation with the Number of Electrons. 80:71 -86 March NH (1996) Semiclassical Theory of Atoms and Ions in Intense External Fields. 86:63-96 Martinsen J, see Ibers ]A (1982) 50:1-55 Mason SF (1980) The Ligand Polarization Model for the Spectra of Metal Complexes: The Dynamic Coupling Transition Probabilities. 39:43- 81 Massey V, see Hemmerich P (1982) 48:93 - 124 Mathey F, Fischer J, Nelson JH (1984) Complexing Modes of the Phosphole Moiety. 55:153 -201 Mauk AG (1991) Electron Transfer in Genetically Engineered Proteins. The Cytochrome c Paradigm. 75:131-158 Mayer U, see Gutman V (1972) 10:127-151 Mayer U, see Gutman V (1973) 15:141 - 166 Mayer H, see Gutman V (1976) 31:49-66 Mayer U, Gutmann V (1972) Phenomenological Approach to Cation-Solvent Interactions. 12:113-140 Mazumdar S~ Mitra S (1993) Biomimetic Chemistry of Heroes Inside Aqueous Micelles. 81:115-145 McGrady JE, see Mingos DMP (1992) 79:1-54 McLendon G (1991) Control of Biological Electron Transport via Molecular Recognition and Binding: The "Velcro" Model. 75:159-174 McLeod GC, see Kustin K (1983) 53:137-158 Mehrotra RC, see Jain VK (1982) 52:147-196
Author Index Volumes 1-90
203
Mehrotra RC (1991) Present Status and Future Potential of the Sol-Gel Process. 77:1-36 Meier PC, see Simon W (1973) 16:113-160 Melnik M, see Valach F (1984) 55:101 - 151 Messerschmidt A (1998) Metal Sites in Small Blue Copper Proteins, Blue Copper Oxidases and Vanadium-ContainingEnzymes. 90:37- 68 Michel H, see Hemmerich P (1982) 48:93-124 Mildvan AS, Grisham CM (1974) The Role of Divalent Cations in the Mechanism of Enzyme Catalyzed Phosphoryl and Nucleotidyl. 20:1-21 Mingos DMP, Hawes JC (1985) Complementary Spherical Electron Densitiy Model. 63:1 - 63 Mingos DMP, Johnston RL (1987) Theoretical Models of Cluster Bonding. 68:29-87 Mingos DMP, McGrady JE, Rohl AL (1992) Moments of Inertia in Cluster and Coordination Compounds. 79:1 - 54 Mingos DMP,Zhenyang L (1990) Hybridization Schemes for Coordination and Organometallic Compounds. 72:73-112 Mingos DMP, Zhenyang L (1989) Non-Bonding Orbitals in Coordination Hydrocarbon and Cluster Compounds. 71 : 1- 56 Mioduski T, see Fidelis IK (1981) 47:27-51 Mitchell PR, see Gillard RD (1970) 7:46 - 86 Mitra S, see Mazumdar S (1993) 81:115-145 Moody DC, see Ryan RR (1981) 46:47-100 Moore GR, see Le Brun NE (1997) 88:103-138 Moreau-Colin ML (1972) Electronic Spectra and Structural Properties of Complex Tetracyanides of Platinum, Palladium and Nickel. 10:167-190 Morf WE, see Simon W (1973) 16:113-160 Morgan B, Dophin D (1987) Synthesis and Structure of Biometric Porphyrins. 64:115 -204 Moritz HG, see Keppler BK (1991) 78:97-128 Morris DFC (1968/1969) An Appendix to Structure and Bonding. 4; 6:157-159 Morris DFC (1968) Ionic Radii and Enthalpies of Hydration of Ions. 4:63 - 82 Mortensen OS (1987) A Noncommuting-GeneratorApproach to Molecular Symmetry. 68:1 - 28 Mortier JW (1987) ElectronegativityEqualization and its Applications. 66:125 - 143 Mortier WJ, see Baekelandt BG (1993) 80:187-228 Moura I, see Xavier AV (1981) 43:187-213 Moura JJG, see Xavier AV (1981) 43:187-213 Mullay JJ (1987) Estimation of Atomic and Group Electronegativities. 66:1 - 25 Mfiller A, Baran EJ, Carter RO (1976) Vibrational Spectra of Oxo-, Thio-, and Selenometallates of Transition Elements in the Solid State. 26: 81-139 Mfiller A, Diemann E, Jorgensen CK (1973) Electronic Spectra of Tetrahedral Oxo, Thio and Seleno Complexes. Formed by Elements of the Beginning of the Transition Groups. •4:23-47 Mfiller U (1973) Strukturchemie der Azide. 14:141 - 172 Mfiller W, Spirlet J-C (1985) The Preparation of High Purity Actinide Metals and Compounds. 59/60: 57 - 73 Munro AW,see Chapman SK (1997) 88:39-70 Murray JS, see Politzer P (1993) 80:101 - 114 Murrell JM (1977) The Potential Energy Surfaces of Polyatomic Molecules. 32:93-146
Naegele JR, Ghijsen J (1985) Localization and Hybridization of 5f States in the Metallic and Ionic Bond as Investigated by Photoelectron Spectroscopy. 59/60:197 -262 Nag K, Bose SN (1985) Chemistry of Tetra- and Pentavalent Chromium. 63:153 - 197 Nalewajski RF (1993) The Hardness Based Molecular Charge Sensitivities and Their Use in the Theory of Chemical Reactivity. 80:115-186 Natan MJ, see Hoffman BM (1991) 75:85 - 108 Neilands JB, see Liu A (1984) 58: 97-106 Neilands JB, see Chimiak A (1984) 58: 89- 96
204
Author Index Volumes 1 - 90
Neilands JB (1972) Evolution of Biological Iron Binding Centers. 11:145-170 Neilands JB (1984) Methodology of Siderophores. 58:1 -24 Neilands JB (1966) Naturally Occurring Non-Porphyrin Iron Compounds. 1:59 - 108 Neilands JP, see Cheh AM (1976) 29:123 - 169 Nelson JH, see Mathey F (1984) 55:153-201 Nickerson DP, see Wong L-L (1997) 88:175 -208 Nieboer E (1975) The Lanthanide Ions as Structural Probes in Biological and Model Systems. 22:1-47 Nocek JM, see Hoffman BM (1991) 75:85-108 Nomoto K, see Sugiura Y (1984) 5 8 : 1 0 7 - 1 3 5 Nova& A (1974) Hydrogen Bonding in Solids. Correlation of Spectroscopic and Crystallographic Data. 1 8 : 1 7 7 - 2 1 6 Nultsch W, H~ider D-P (1980) Light Perception and Sensory Transduction in Photosynthetic Prokaryotes. 41 : 111 - 139 Nyholm RS, see Hall DI (1973) 15:3-51
O'Keeffe M, see Brese NE (1992) 7 9 : 3 0 7 - 3 7 8 O'Keeffe M, Hyde BG (1985) An Alternative Approach to Non-Molecular Crystal Structures with Emphasis on the Arrangements of Cations. 61 : 77-144 O'Keeffe M (1989) The Prediction and Interpretation of Bond Lengths in Crystals. 71 : 161 - 190 O dora JD (1983) Selenium Biochemistry. Chemical and Physical Studies. 54:1 -26 Oehme I, see Wolfbeis OS (1996) 85:51-98 Oelkrug D (1971) Absorption Spectra and Ligand Field Parameters of Tetragonal 3d-Transition Metal Fluorides. 9:1 26 Oosterhuis WT (1974) The Electronic State of Iron in Some Natural Iron Compounds: Determination by M6ssbauer and ESR Spectroscopy. 20: 59- 99 Orchin M, Bollinger DM (1975) Hydrogen-Deuterium Exchange in Aromatic Compounds. 23:167-193 Ozaki Y, see Kitagawa T (1987) 64:71 - 114 -
Pace LJ, see Ibers JA (1982) 50:1-55 Padhye SB, see West DC (1991) 7 6 : 1 - 5 0 Paolesse R, see Licoccia S (1995) 84:71 - 134 Parr RG, see Chattaraj PK (1993) 80:11-26 Patil SK, see Ramakrishna VV (1984) 56:35 - 90 Peacock RD (1975) The Intensities of Lanthanide f ~ f Transitions. 22:83 - 122 Pearson RG (1993) Chemical Hardness - An Historial Introduction. 80:1 - 10 Pecoraro VL, see Slebodnick C (1997) 89:51 - 108 Pelikfin P, see Boca R (1989) 7•:57-97 Penfield KW, see Solomon EI (1983) 53:1-56 Penneman RA, Ryan RR, Rosenzweig A (1973) Structural Systematics in Actinide Fluoride Complexes. 13:1 - 52 Penner-Hahn JE (1998) Structural Characterization of the Mn Site in the Photosynthetic Oxygen-Evolving Complex. 90:1 -36 Perlman ML, see Watson RE (1975) 24: 83 - 132 Politzer P, Murray JS, Grice ME (1993) Charge Capacities and Shell Structures of Atoms. 80:101 - 114 Pouchard M, see Grenier JC (1981) 4 7 : 1 - 2 5 Powell AK (1997) Polyiron Oxides, Oxyhydroxides and Hydroxides as Models for Biomineralisation Processes. 88:1 - 38 Powell RC, Blasse G (1980) Energy Transfer in Concentrated Systems. 42: 43 - 96 Que Jr. L (1980) Non-Heme Iron Dioxygenases. Structure and Mechanism. 40:39-72
AuthorIndexVolumes1-90
205
Rakke T, see Kjekshus A (1974) 19:45-83 Rakke T, see Kjekshus A (1974) 19: 85-104 Ramakrishna VV, Patil SK (1984) Synergic Extraction of Actinides. 56:35-90 Ramakrishnan L, see Koppikar DK (1978) 3 4 : 1 3 5 - 2 1 3 Rao VUS, see Wallace WE (1977) 33:1 -55 Raphael AL, see Therien MJ (1991) 75:109-130 Ravikanth M, Chandrashekar TK (1995) Nonplanar Porphyrins and Their Biological Relevance: Ground and Excited State Dynamics. 8 2 : 1 0 5 - 1 8 8 Rawle SC, see Cooper SR (1990) 7 2 : 1 - 7 2 Raymond KN, see Baker EC (1976) 25:21-66 Raymond KN, Smith WL (1981) Actinide-Specific Sequestering Agents and Decontamination Applications. 43:159 - 186 Reedijk J, Fichtinger-Schepman AMJ, Oosterom AT van, Putte P van de (1987) Platinum Amine Coordination Compounds as Anti-Tumour Drugs. Molecular Aspects of the Mechanism of Action. 67:53 - 89 Rein M, see Schultz H (1991) 74:41 - 146 Reinen D, Friebel C (1979) Local and Cooperative Jahn-Teller Interactions in Model Structures. Spectroscopic and Structural Evidence. 37:1 -60 Reinen D (1970) Kationenverteilung zweiwertiger 3dn-Ionen in oxidischen Spinell-, Granat und anderen Strukturen. 7:114-154 Reinen D (1969) Ligand-Field Spectroscopy and Chemical Bonding in Cr3+-Containing Oxidic Solids. 6: 30- 51 Reisfeld R, see Jorgensen CK (1982) 50:121 - 171 Reisfeld R, Jorgensen CK (1988) Excided States of Chromium(III) in Translucent GlassCeramics as Prospective Laser Materials. 69:63- 96 Reisfeld R (1996) Laser Based on Sol-Gel Technology. 85:215- 234 Reisfeld R, Jorgensen CK (1982) Luminescent Solar Concentrators for Energy Conversion. 49:1-36 Reisfeld R (1996) New Materials for Nonlinear Optics. 85:99-148 Reisfeld R, Jorgensen ChK (1991) Optical Properties of Colorants or Luminescent Species in Sol-Gel Glasses. 77: 207- 256 Reisfeld R (1976) Excited States and Energy Transfer from Donor Cations to Rare Earths in the Condensed Phase. 30: 65- 97 Reisfeld R (1975) Radiative and Non-Radiative Transitions of Rare Earth Ions in Glasses. 22:123-175
Reisfeld R (1973) Spectra and Energy Transfer of Rare Earths in Inorganic Glasses. 13:53-98 Reisfeld R, see Wolfbeis OS (1996) 85:51- 98 Reslova S, see Thomson AJ (1972) 11 : 1-46 R6der A, see Hill HAO (1970) 8:123 - 151 Rohl AL, see Mingos DMP (1992) 79:1 -54 Romao MJ, Huber R (1998) Structure and Function of the Xanthine-Oxidase Family of Molybdenum Enzymes. 90: 69- 96 Rosenzweig A, see Penneman RA (1973) 13:1 -52 Rfidiger W (1980) Phytochrome, a Light Receptor of Plant Photomorphogenesis. 40:101 - 140 Ruedenberg K, see Hoffmann DK (1977) 33:57-96 Runov VK, see Golovina AP (1981) 47:53-119 Russo VEA, Galland P (1980) Sensory Physiology of Phycomyces Blakesleeanus. 41 : 71 - 110 Ryan RR, see Penneman RA (1973) 13:1-52 Ryan RR, Kubas GJ, Moody DC, Eller PG (1981) Structure and Bonding of Transition MetalSulfur Dioxide Complexes. 46: 47 - 1O0
Sadler PJ, see Berners-Price SJ (1988) 70:27-102 Sadler PJ, see Dhubhghaill OMN (1991) 7 8 : 1 2 9 - 1 9 0
206
Author Index Volumes 1- 90
Sadler PJ, see Sun H (1997) 88:71 - 102 Sadler PJ (1976) The Biological Chemistry of Gold: A Metallo-Drug and Heavy-Atom Label with Variable Valency.29:171 - 214 Saillard J-Y,see Haler J-F (1997) 87:81 - 110 Sakka S, Yoko T (I 991) Sol-Gel-Derived Coating Films and Applications. 77: 89-118 Sakka S (1996) Sol-Gel Coating Films for Optical and Electronic Application. 8 5 : 1 - 50 Saltman P, see Spiro G (1969) 6:116-156 Sando GN, see Hogenkamp HPC (1974) 20: 23 - 58 Sankar SG, see Wallace WE (1977) 33:1 - 55 Sch~iffer CE, see Harnung SE (1972) 12:201-255 Sch~ffer CE, see Harnung SE (1972) 12: 257- 295 Sch~ffer CE (1968) A Perturbation Representation of Weak Covalent Bonding. 5:68-95 Sch~iffer CE (1973) Two Symmetry Parameterizations of the Angular-Overlap Model of the Ligand-Field. Relation to the Crystal-Field Model. 14: 69-110 Scheidt WR, Lee YJ (1987) Recent Advances in the Stereochemistry of Metallotetrapyrroles. 64:1-70 Schl~pfer CW, see Daul C (1979) 36:129-171 Schmelcher PS, Cederbaum LS (1996) Two Interacting Charged Particles in Strong Static Fields: A Variety of Two-Body Phenomena. 86: 27- 62 Schmid G (1985) Developments in Transition Metal Cluster Chemistry. The Way to Large Clusters. 62: 51 - 85 Schmidt H (1991) Thin Films, the Chemical Processing up to Gelation. 77:115-152 Schmidt PC, see Sen KD (1987) 66:99-123 Schmidt PC (1987) Electronic Structure of Intermetallic B 32 Type Zinfl Phases. 65:91-133 Schmidt W (I 980) Physiological Bluelight Reception. 41 : 1-44 Schmidtke H-H, Degen J (1989) A Dynamic Ligand Field Theory for Vibronic Structures RationalizingElectronic Spectra of Transition Metal Complex Compounds. 71:99-124 Schneider W (1975) Kinetics and Mechanism of Metalloporphyrin Formation. 23:123 - 166 Schoonheydt RA, see Baekelandt BG (1993) 80:187-228 Schretzmann P, see Bayer E (1967) 2:181-250 Schubert K (1977) The Two-Correlations Model, a Valence Model for Metallic Phases. 33:139-177
Schug C, see Hemmerich P (1982) 4 8 : 9 3 - 1 2 Schultz H, Lebmann H, Rein M, Hanack M (1991) Phthalocyaninatometal and Related Complexes with Special Electrical and Optical Properties. 74:41-146 Schutte CJH (1971) The Ab-Initio Calculation of Molecular Vibrational Frequencies a n d Force Constants. 9:213 - 263 Schweiger A (1982) Electron Nuclear Double Resonance of Transition Metal Complexes with Organic Ligands. 51 : 1-122 Scozzafava A, see Bertini I (1982) 48: 45- 91 Sen KD, B6hm MC, Schmidt PC (1987) Electronegativity of Atoms and Molecular Fragments. 66: 99-123 Sen KD (1993) Isoelectronic Changes in Energy, Electronegativity, and Hardness in Atoms via the Calculations of
. 80:87-100 Shamir J (1979) Polyhalogen Cations. 37:141 -210 Shannon RD, Vincent H (1974) Relationship Between Covalency, Interatomic Distances, and Magnetic Properties in Halides and Chalcogenides. 19:1 -43 Shihada A-F,see Dehnicke K (1976) 28:51- 82 Shionoya M, see Kimura E (1997) 8 9 : 1 - 2 8 Shriver DF (1966) The Ambident Nature of Cyanide. 1:32-58 Siegel FL (1973) Calcium-BindingProteins. 17:221 - 268 Sima J (1995) Photochemistry of Tetrapyrrole Complexes. 84:135 - 194 Simon A (1979) Structure and Bonding with Alkali Metal Suboxides. 36:81 - 127 Simon W, Morf WE, Meier PCh (1973) Specificity of Alkali and Alkaline Earth Cations of Synthetic and Natural Organic Complexing Agents in Membranes. 16:113-160
AuthorIndexVolumes1-90
207
Simonetta M, Gavezzotti A (1976) Extended Hfickel Investigation of Reaction Mechanisms. 27:1-43
Sinha SP (1976) A Systematic Correlation of the Properties of the f-Transition Metal Ions. 30:1-64 Sinha SP (1976) Structure and Bonding in Highly Coordinated Lanthanide Complexes. 25:67-147
Sivapullaiah PV, see Koppikar DK (1978) 34:135 - 213 Sivy P, see Valach F (1984) 55:101-151 Sj6berg B-M (1997) Ribonucleotide Reductases - A Group of Enzymes with Different Metallosites and a Similar Reaction Mechanism. 8 8 : 1 3 9 - 1 7 4 Slebodnick C, Hamstra BJ, Pecoraro VL (1997) Modeling the Biological Chemistry of Vanadium: Structural and Reactivity Studies Elucidating Biological Function. 89:51 - 108 Smit HHA, see Thiel RC (1993) 81 : 1-40 Smith DW, Williams RJP (1970) The Spectra of Ferric Haems and Haemoproteins. 7:1-45 Smith DW (1978) Applications of the Angular Overlap Model. 35:87 - 118 Smith DW (1972) Ligand Field Splittings in Copper(II) Compounds. 12:49-112 Smith PD, see Buchler JW (1978) 34:79-134 Smith T, see Livorness J (1982) 4 8 : 1 - 4 4 Smith WL, see Raymond KN (1981) 43:159 - 186 Solomon EI, Penfield KW, Wilcox DE (1983) Active Sites in Copper Proteins. An Electric Structure Overview. 53:1 - 56 Somorjai GA, Van Hove MA (1979) Adsorbed Monolayers on Solid Surfaces. 38:1 - 140 Sonawane PB, see West DC (1991) 7 6 : 1 - 5 0 Soundararajan S, see Koppikar DK (1978) 34:135- 213 Speakman JC (1972) Acid Salts of Carboxylic Acids, Crystals with some "Very Short" Hydrogen Bonds. 1 2 : 1 4 1 - 1 9 9 Spirlet J-C, see Mfiller W (1985) 59/60: 5 7 - 73 Spiro G, Saltman P (1969) Polynuclear Complexes of Iron and Their Biological Implications. 6:116-156 Steggerda JJ, see Willemse J (1976) 28: 83-126 Stewart B, see Clarke MJ (1979) 36:1-80 Strohmeier W (1968) Problem und ModeU der homogenen Katalyse. 5: 96 - 117 Sugiura Y, Nomoto K (1984) Phytosiderophores - Structures and Propterties of Mugineic Acids and Their Metal Complexes. 5 8 : 1 0 7 - 1 3 5 Sun H, Cox MC, Li H, Sadler PJ (1997) Rationalisation of Binding to Transferrin: Prediction of Metal-Protein Stability Constants. 88:71 - 102 Swann JC, see Bray RC (1972) 11:107-144 Sykes AG (1991) Plastocyanin and the Blue Copper Proteins. 75:175 - 224
Takita T, see Umezawa H (1980) 40:73-99 Tam S-C, Williams RJP (1985) Electrostatics and Biological Systems. 63:103-151 Teller R, Bau RG (1981) Crystallographic Studies of Transition Metal Hydride Complexes. 44:1 - 82 Therien MJ, Chang J, Raphael AL, Bowler BE, Gray HB (1991) Long-Range Electron Transfer in Metalloproteins. 75:109-130 Thiel RC, Benfield RE, Zanoni R, Smit HHA, Dirken MW (1993) The Physical Properties of the Metal Cluster Compound Au55(PPh3) 12C16.81 : 1-40 Thomas KR, see Chandrasekhar V (1993) 81:41 - 114 Thompson DW (1971) Structure and Bonding in Inorganic Derivatives of fl-Diketones. 9: 27- 47 Thomson AJ, Reslova S, Williams RJP, (1972) The Chemistry of Complexes Related to cis-Pt(NH3)2C12. An Anti-Tumor Drug. 11 : 1 - 46 Thomson AJ, see Le Brun NE (1997) 88:103 - 138 Tofield BC (1975) The Study of Covalency by Magnetic Neutron Scattering. 21 : 1 - 87 Trautwein AX, Bill E, Bominaar EL, Winlder H (1991) Iron-Containing Proteins and Related Analogs-Complementary MBssbauer, EPR and Magnetic Susceptibility Studies. 7 8 : 1 - 96
208
Author Index Volumes 1 - 90
Trautwein AX (1974) M6ssbauer-Spectroscopy on Heme Proteins. 20:101-167 Tressaud A, Dance ]-M (1982) Relationships Between Structure and Low-Dimensional Magnetism in Fluorides. 52: 87-146 Tributsch H (1982) Photoelectrochemical Energy Conversion Involving Transition Metal d-States and Intercalation of Layer Compounds. 49:127-175 Truter MR (1973) Structures of Organic Complexes with Alkali Metal Ions. 16:71 - 111
Umezawa H, Takita T (1980) The Bleomycins: Antitumor Copper-Binding Antibiotics. 40:73-99
Vahrenkamp H (1977) Recent Results in the Chemistry of Transition Metal Clusters with Organic Ligands. 32:1- 56 Valach F, Kdren B, Siv~ P, Meln~ M (1984) Crystal Structure Non-Rigidity of Central Atoms for Mn(II), Fe(II), Fe(III), Co(II), Co(III), Ni(II), Cu(II) and Zn(II) Complexes. 55:101-151 Valdemoro C, see Fraga S (1968) 4:1-62 van Bronswyk W (1970) The Application of Nuclear Quadrupole Resonance Spectroscopy to the Study of Transition Metal Compounds. 7: 87-113 van de Putte P, see Reedijk I (1987) 67:53-89 Van Hove MA, see Somorjai GA (1979) 38:1 - 140 van O osterom AT, see Reedijk I (1987) 67:53 - 89 Vanquickenborne LG, see Ceulemans A (1989) 71:125 - 159 Vela A, see G~zquez JL (1987) 66:79-98 Verkade JG, see Hoffmann DK (1977) 3 3 : 5 7 - 9 6 Vincent H, see Shannon RD (1974) 19:1-43 Vogel E, see Keppler BK (1991) 78:97-128 von Herigonte P (1972) Electron Correlation in the Seventies. 12:1-47 yon Zelewsky A, see Daul C (1979) 36:129-171 Vongerichten H, see Keppler BK ( 1991) 78: 97 - 128
Wallace WE, Sankar SG, Rao VUS (1977) Field Effects in Rare-Earth Intermetallic Compounds. 33:1-55 Wallin SA, see Hoffman BM (1991) 75:85-108 Walton RA, see Cotton FA (1985) 6 2 : 1 - 4 9 Warren KD, see Allen GC (1974) 19:105 - 165 Warren KD, see Allen GC (1971) 9:49 - 138 Warren KD, see Clack DW (1980) 39:1-141 Warren KD (1984) Calculations of the ]ahn-Teller Coupling Constants for dx Systems in Octahedral Symmetry via the Angular Overlap Model. 57:119 - 145 Warren KD (1977) Ligand Field Theory of f-Orbital Sandwich Complexes. 33:97-137 Warren KD (1976) Ligand Field Theory of Metal Sandwich Complexes. 27:45-159 Watson RE, Perlman ML (1975) X-Ray Photoelectron Spectroscopy. Application to Metals and Alloys. 24:83 - 132 Weakley TJR (1974) Some Aspects of the Heteropolymolybdates and Heteropolytungstates. 18:131-176 Weissbluth M (1967) The Physics of Hemoglobin. 2:1-125 Wendin G (1981) Breakdown of the One-Electron Pictures in Photoelectron Spectra. 45:1 - 130 Wertheim GK, see Campagna M (1976) 30 : 99-140 Weser U (1967) Chemistry and Structure of some Borate Polyol Compounds. 2:160-180 Weser U (1968) Reaction of some Transition Metals with Nucleic Acids and Their Constituents. 5: 41 - 67 Weser U (1985) Redox Reactions of Sulphur-Containing Amino-Acid Residues in Proteins and Metalloproteins, an XPS Study. 61 : 145 - 160
Author Index Volumes 1-90
209
Weser U (1973) Structural Aspects and Biochemical Function of Erythrocuprein. 17:1-65 West DC, Padhye SB, Sonawane PB (1991) Structural and Physical Correlations in the Biological Properties of Transitions Metal Heterocyclic Thiosemicarbazone and S-alkyldithiocarbazate Complexes. 76:1- 50 Westlake ACG,see Wong L-L (1997) 88:175-208 Wetterhahn KE, see Connett PH (1983) 54: 93 - 124 Wilcox DE, see Solomon EI (1983) 53:1-56 Wilkie J, see Gani D (1997) 89:133-176 Willemse J, Cras ]A, Steggerda JJ, Keijzers CP (1976) Dithiocarbamates of Transition Group Elements in "Unusual" Oxidation State. 28: 83-126 Williams AF, see Gubelmann MH (1984) 55:1-65 Williams RIP, see Frafisto da Silva JJR (1976) 29:67-121 Williams RIP, see Hill HAO (1970) 8:123 - 151 Williams RIP,see Smith DW (1970) 7:1-45 Williams RIP, see Tam S-C (1985) 63:103-151 Williams RIP,see Thomson AJ (1972) 11 : 1-46 Williams RJP, Hale JD (1973) Professor Sir Ronald Nyholm. 15:1 and 2 Williams RIP, Hale JD (1966) The Classification of Acceptors and Donors in Inorganic Reactions. 1:249-281 Williams RJP (1982) The Chemistry of Lanthanide Ions in Solution and in Biological Systems. 50: 79-119 Wilson ]A (1977) A Generalized Configuration - Dependent Band Model for Lanthanide Compounds and Conditions for Interconfiguration Fluctuations. 32:57-91 Winkler H, see Trautwein AX (1991) 78:1-96 Winkler R (1972) Kinetics and Mechanism of Alkali Ion Complex Formation in Solution. 10:1-24
Wolfbeis OS, Reisfeld R, Oehme I (1996) Sol-Gels and Chemical Sensors. 85:51-98 Wong L-L,Westlake ACG,Nickerson DP (1997) Protein Engineering of Cytochrome P450ca m. 88:175 - 208 Wood JM, Brown DG (1972) The Chemistry of Vitamin B:2-Enzymes. 11:47-105 Woolley RG, see Gerloch M (1981) 46:1-46 Woolley RG (1982) Natural Optical Acitivity and the Molecular Hypothesis. 52:1 -35 Wfithrich K (1970) Structural Studies of Hemes and Hemoproteins by Nuclear Magnetic Resonance Spectroscopy. 8: 53 - 121
Xavier AV,Moura JJG, Moura I (1981) Novel Structures in Iron-Sulfur Proteins. 43:187-213
Yersin H, see Gliemann G (1985) 62:87-153 Yoko T, see Sakka S (1991) 77:89-118
Zanchini C, see Banci L (1982) 52:37-86 Zanello P (1992) Stereochemical Aspects Associated with the Redox Behaviour of Heterometal Carbonyl Clusters. 79:101-214 Zanoni R, see Thiel RC (1993) 81:1-40 Zhenyang L, see Mingos DMP (1989) 71 : 1-56 Zhenyang L, see Mingos DMP (1990) 72: 73-112 Zorov NB, see Golovina AP (1981) 47:53-119 Zumft WG (1976) The Molecular Basis of Biological Dinitrogen Fixation. 29:1 - 65