Current TOPES in Membranes and Transport Vohnne 9
Advisory Board
Robert W. Berliner 1. S . Edelman 1. M . Glynn FranCois Morel Shmuel Razin Aser Rothstein H . J . Schatzmann Stanley G . Schultz Philip Siekevitx Daniel C.Tosteson
Contributors
Johannes Boonstra S . Roy Caplan Mortimer M. Civan Alvin Essig M . Marlene Hosey F . IsmailBeigi Wil N . Konings Stuart Mchughlin Leena Mela Mordechai Shporer Mariano Tao
Current Topics in Membranes and Transport VOLUME 9
Edited by Felix Bronner Department of Oral Biology University of Connecticut Health Center Fannington, Connecticut and
Amort Kleinzeller Department of Physiology University of Pennsylvania School of Medicine Philadelphia, Pennsylvania
1977
Academic Press
New York San Fmncirco London
A Subsidiary of Harcourt Brace Jovanouich, Publkhers
COPYRIGHT 0 1977, BY ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED OR TRANSMlTTED IN A N Y FORM OR BY A N Y MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITINQ FROM THE PUBLISHER.
ACADEMIC PRESS, INC.
111 Fifth Avenue, New Yak, New York 10003
United Kingdom Edition publkhed by ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road. London NWI
LIBRARY OF CONGRESS CATALOG CARD NUMBER: 70-1 17091 ISBN 0- 12- 15 3309-3 PRINTED IN THE UNITED STATES OF AMERICA
List of Contributors, ix Contents of P n v i w s Volumes, xi The State of Water and Alkali Cations within the Intmcellular Fluidr: The Contribution of NMR Spectmscopy MORDECHAI SHPORER AND MORTIMER M. CIVAN I. Introduction, 1 11. Principles of NMR Spectroscopy, 6 111. Techniques of NMR Spectroscopy, 17 IV. NMR Studies of Water, 19 V. NMR Studies of Alkali Cations, 41 VI. Conclusions, 59 Symbols and Abbreviations, 61 References, 62
Elaetrostatic Potentials at Membrane-Solution Interfaces STUART MCLAUGHLIN 1. Introduction, 71 11. Fixed Charges at Membrane-Solution Interfaces, 73 111. Adsorption of Charged Molecules to Membranes, 93
IV. Molecular Dipoles at Membrane-Solution Interfaces, 108 V. Electrostatic “Boundary” Potentials, 113 VI. Biological Implications, 118 Appendix I, 130 Appendix 11, 132 Apendix 111, 133 References, 135
A Thermodynamic Treatment of Active Sodium Transport S. ROY CAPLAN AND ALVIN ESSIG
I. Introduction, 145 11. Theory of the Nonequilibrium Thermodynamic (NET) Approach, 147 111. Experimental Evaluation of the NET Approach, 149 V
vi
CONTENTS
IV. V. VI. VII. VIII.
Theory of the Equivalent Circuit Model, 162 Experimental Evaluation of the Equivalent Circuit Model, 165 Utility of the Thermodynamic Affinity A, 165 Experimental Comparison of ENaand A, 169 Some General Comments, 170 IX. Conclusions, 173 References, 173
Anaerobic Electron Transfer and Active Transport in Bacteria WIL N. KONINGS AND JOHANNES BOONSTRA I. Introduction, 177 11. Anaerobic Electron Transfer Systems, 180 111, Phosphorylation Coupled to Electron Transfer, 195
IV. Anaerobic Active Transport, 199 References, 219
Protein Kinases and Membrane Phosphorylation
M. MARLENE HOSEY AND MARIAN0 TAO I. Introduction, 233 11. Protein Kinases, 237
111. Membrane Phosphorylation, 258 IV. Membrane-Bound Phosphoprotein Phosphatases, 299 V. Concluding Remarks, 303 References, 304
Mechanism and Physiological Significance of Calcium Transport across Mammalian Mitochondrial Membranes LEENA MELA
I. Introduction, 322 11. Early Experiments Leading to the Discovery of Mitochondrial Ability to Accumulate Ca*+Ions, 322 111. Three-Step Mechanism of Mitochondrial Ca2+Accumulation, 326 IV. Role of Mitochondria in the Physiological Control of Cellular Ca2' Concentration, 337
V. Physiological Significance of Mitochondrial Ca2+Accumulation in Different Tissues, 344 VI. Some Aspects of the Pathophysiology of Mitochondrial Caz+Accumulation, 350 VII. Summary, 354 References, 354
CONTENTS
vii
Thyroidal Regulation of Active Sodium Transport
F. ISMAIL-BEIGI I. Introduction, 367 11. Thyroid Status and Sodium Transport-Dependent Respiration (Qoz(t)),369
111. Possible Pathways of Thyroid Hormone-Induced Increase in Q,,,(t), 372 IV. Thyroid Status and Transmembrane Electrochemical Potential Differences of Nat and K+, 375 V. Thyroid Status and Membrane NaK-ATPase Activity, 379 VI. Thyroid Status and Tissue Adenine Niicleotide Content, 383 VII. Summary and Conclusions, 384 References, 385
Subiect Index, 389
This Page Intentionally Left Blank
Numbers in parentheses indicate the pages on which the authors’ contributions begin. Johannes Boonstm, Department of Microbiology, Biological Center, University Cron-
ingen, Haren, The Netherlands (177) 5. Roy Caplan, Department of Membrane Research, Weizmann Institute of Science, Rehovot, Israel (145)
of Physiology and Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania (1)
MoHimer M. Civan, Departments
of Physiology, Boston University School of Medicine, Boston, Massachusetts (145)
Alvin Euig, Department
M. Marlene Mosey, Department of Biological Chemistry, University of Illinois at the
Medical Center, Chicago, Illinois (233) F. Ismail-Beigi, Department of Internal Medicine and Pahlavi Medical Research Unit,
Pahlavi University, School of Medicine, Shiraz, Iran (367) Wil N. Konings, Department of Microbiology, Biological Center, University Groningen,
Haren, The Netherlands (177)
of Physiology and Biophysics, Health Sciences Center, State University of New York, Stony Brook, New York (71)
Stuae Mclaughlin, Department
of Surgery and Biochemistry and Biophysics, University of Pennsylvania, Philadelphia, Pennsylvania (321)
Leena Mela, Departments
Mordechai Shpomr,* Departments of Physiology and Medicine, University of Pennsyl-
vania School of Medicine, Philadelphia, Pennsylvania (1) Mariano Tao, Department of Biological Chemistry, University of Illinois at the Medical
Center, Chicago, Illinois (233)
* On leave from the Department of Isotope Research, The Weizmann Institute of Science, Rehovot, Israel. ix
This Page Intentionally Left Blank
Contents of Previous Volumes Volume 1
Some Considerations about the Structure of Cellular Membranes MAYNARD M. DEWEYAND LLOYDBARR The Transport of Sugars across Isolated Bacterial Membranes H. R. KABACK Calactoside Pennease of Escherichiu coli ADAMKEPES Sulfhydryl Croups in Membrane Structure and Function ASER ROTHSTEIN Molecular Architecture of the Mitochondrion DAVIDH. MACLENNAN Author Indexsubject Index Volume 2
The Molecular Basis of Simple Diffusion within Biological Membranes W. R. h E B AND w. D. STEIN The Transport of Water in Erythrocytes ROBERTE . FORSTER Ion-Translocation in Energy-Conserving Membrane Systems B. CHANCEAND M. MONTAL Structure and Biosynthesis of the Membrane Adenosine Triphosphatase of Mitochondria
TZAGOLOFF ALEXANDER Mitochondria1 Compartments: A Comparison of Two Models HENRYTEDESCHI Author Index-Subject Index
Volume 3
The Na+, K+-ATPase Membrane Transport System: Importance in Cellular Function ARNOLDSCHWARTZ, AND GEORGEE. LINDENMAYER, JULIUSC. ALLEN Biochemical and Clinical Aspects of Sarcoplasmic Reticulum Function ANTHONYMARTONOSI The Role of Periaxonal and Perineuronal Spaces in Modifying Ionic Flow across Neural Membranes w. J . ADELMAN,JR. AND Y. PALTI Properties of the Isolated Nerve Endings GEORGINAR O D ~ G U EDE Z LORES ARNAIZAND EDUARDODE ROBERTIS Transport and Discharge of Exportable Proteins in Pancreatic Exocrine Cells: In Vitro Studies J . D. JAMIESON xi
xii The Movement of Water across Vasopressin-Sensitive Epithelia RICHARDM. HAYS Active Transport of Potassium and Other Alkali Metals by the Isolated Midgut of the Silkworm WILLIAMR. HARVEYAND KARL ZERAHN Author Index-Subject Index Volume 4 The Genetic Control of Membrane Transport CAROLYN W. SLAYMAN Enzymic Hydrolysis of Various Components in Biomembranes and Related Systems KUMAR JAIN MAHENDRA Regulation of Sugar Transport in Eukaryotic Cells HOWARDE. MORGANAND CAROLF. WHITFIELD Secretory Events in Gastric Mucosa RICHARD P. DURBIN Author Index-Subject Index Volume 5 Cation Transport in Bacteria: K+, Na+, and H+ FRANKLIN M. HAROLDAND KARLHEINZ ALTENDORF Pro and Contra Carrier Proteins; Sugar Transport via the Periplasmic GalactoseBinding Protein WINFRIEDBoos Coupling and Energy Transfer in Active Amino Acid Transport ERICH HEINZ The Means of Distinguishing between Hydrogen Secretion and Bicarbonate Reabsorption: Theory and Applications to the Reptilian Bladder and Mammalian Kidney AND WILLIAMA. BRODSKY THEODORE P. SCHILB Sodium and Chloride Transport across Isolated Rabbit Ileum STANLEY G . SCHULTZ AND PETERF. CURRAN
CONTENTS OF PREVIOUS VOLUMES
A Macromolecular Approach to Nerve
Excitation AND ICHIJI TASAKI EMILIO CARBONE Shbject Index Volume 6 Role of Cholesterol in Biomembranes and Related Systems KUMAR JAIN MAHENDRA Ionic Activities in Cells A. A. LEV AND W. McD. ARMSTRONG Active Calcium Transport and Ca2+Activated ATPase in Human Red Cells H. J. SCHATZMANN The Effect of Insulin on Glucose Transport in Muscle Cells TORBEN CLAUSEN Recognition Sites for Material Transport and Information Transfer HALVORN. CHRISTENSEN Subject Index Volume 7 Ion Transport in Plant Cells E. A. C. MACROBBIE H+ Ion Transport and Energy Transduction in Chloroplasts RICHARD A. DILLEYAND ROBERTT. GIAQUINTA The Present State of the Carrier Hypothesis PAULG . LEFEVRE Ion Transport and Short-circuit Technique WARRENS. REHM Subject Index Volume 8 Chemical and Physical Properties of Myelin Proteins M. A. MOSCARELLO The Distinction between Sequential and Simultaneous Models for Sodium and Potassium Transport P. J. GARRAHAN AND R. P. GARAY
CONTENTS OF PREVIOUS VOLUMES
Soluble and Membrane ATPases of Mitochondria, Chloroplasts, and Bacteria: Molecular Structure, Enzymatic Properties, and Functions W K A PANET AND D. RAO SANADI Competition, Saturation, and Inhibition-Ionic Interactions Shown by Meni-
xiii brane Ionic Currents in Nerve, Muscle, and Bilayer Systems ROBERTJ. FRENCH AND WILLIAMJ. ADELMAN,JR. Properties of the Glucose Transport System in the Renal Brush Border Membrane R. KI”E Subject Index
This Page Intentionally Left Blank
Current Topics in Membranes and Transport Volume 9
This Page Intentionally Left Blank
The State of Water and Alkali Cations within the Intracellular Fluids: The Contribution of NMR Spectroscopy MORDECHAZ SHPORER* AND MORTZMER M . CZVAN Departments of Physiology and Medicine University of Pennsylvania School of Medicine Philadelphia, Pennsylvania
Introduction .......... ................................ Principles of NMR Spec ...................... Techniques of NMR Speckoscopy ...... .......................... NMR Studies of Water . . . . . . . .......................... A. Ordered Model Systems .......................................... B. Protein Solutions . . . . . . . . . ........................... C. Intracellular Fluids ........................ .................. V. NMR Studies of Alkali Cations . . . . . . .......................... A. Model Systems . . . . . . . . . . . . . . . . . .......................... B. Intracellular Fluids ............. .......................... VI. Conclusions ........................ .......................... Symbols and Abbreviations ........................................... References ........................................................... I. 11. 111. IV.
1.
1 17 19 21 27 34
41 44
48 59 61 62
INTRODUCTION
Over the past decade, a large number of papers have been published dealing with applications of nuclear magnetic resonance (NMR) spectroscopy to physiological problems. Of particular interest have been the efforts to define the nature and state of the intracellular fluids. The present review in no way constitutes an exhaustive compilation of the published literature; rather, we have specifically tried to communicate the nature and limitations of present NMR techniques
* On leave from the Department of Isotope Research, The Weizmann Institute of Science, Rehovot, Israel. 1
2
MORDECHAI SHPORER AND MORTIMER M. CIVAN
and the basic concepts that have been proposed to account for the data and to assess the relative significance and promise of the various NMR approaches in current use. In this introductory section we would like to place the biological applications of NMR spectroscopy in perspective, albeit briefly. Our main point here is the fact that NMR techniques constitute one of several powerful approaches currently being used to define the state of the intracellular fluids and that these different approaches provide different complementary information. It has long been recognized that the nature and state of intracellular fluids is one of the most crucial problems confroilting cell physiologists. The state and intracellular distribution of ions clearly play a pivotal role in regulating certain cellular functions; as a familiar example, the intracellular distribution of Ca2+ between the cytoplasm and sarcoplasmic reticulum is a primary determinant of the state of contraction of striated muscle. To what extent subcellular redistribution of water and ions may play a role in regulating other physiological and pathophysiological processes, such as cell volume regulation, transepithelial transport, gene activity, blast transformation, and malignant transformation, is conjectural. However, it is clear that definition of the transport processes responsible for movement of water and ions across the plasma membrane is altogether hopeless until the driving forces acting on these species can be measured. Since the composition of the extracellular fluids can usually be determined, the central problem is reduced to a definition of the activities of ions and water within the cytoplasm an’d within the subcellular organelles. As an initial step toward obtaining this information, the total intracellular contents of water and ions have been measured in a large number of tissues. Implicit in such measurements are the assumptions that: (i) the molecule marking the extracellular space, be it inulin, polyethylene glycol, sucrose, mannitol, or S042-, does not penetrate the intracellular space; and (ii) both the ions under study and the much larger extracellular marker are distributed identically throughout the interstitial space. Quite apart from the validity of these assumptions, such measurements can provide estimations only of the average ionic concentrations (cJ within the cell. Because of potential subcellular compartmentalization, structuring of intracellular water and binding of intracellular ions, such data are insufficient for calculating cytoplasmic ionic activities. By “subcellular compartmentalization,” we refer to the establishment of concentration gradients of free ions between the cytoplasm and subcellular organelles, such as the nucleus and mitochondria. By
NMR STUDIES OF WATER AND IONS WITHIN CELLS “
3
structuring of intracellular water,” we refer to the average degree of geometric correlation; some forms of structured water may indeed have different solvation properties from those of bulk water. The concept of binding of ions by macromolecular sites may seem intuitively obvious. The intent is to refer to a decrease in the electrochemical activity of the ions caused b y the presence of the macromolecules. However, leaving the concept in such terms has provoked spirited semantic debate. For example, is an ion bound if it is confined to, but free to travel within, an equipotential surface delimiting a macromolecule? For the purposes of this review, the “binding of ions” will be defined specifically in terms of measurable NMR parameters, following presentation of the basic principles of NMR spectroscopy. A number of techniques have been utilized in order to detect the presence, and quantify the degrees, of subcellular compartmentalization and immobilization of ions and water. Kinetic analysis of radioactive ionic fluxes across the plasma membrane has been the approach perhaps most widely applied. First-order kinetics have been considered to reflect uncomplicated membrane transport. However, such kinetics are rarely observed, and multiexponential kinetics may reflect processes other than intracellular compartmentalization and binding. For example, cellular heterogeneity within the tissue, effects of the changing intracellular ionic concentration either directly upon the membrane permeability or indirectly by altering the transmembrane electrical potential, as well as technical artifacts can all lead to nonexponential behavior. Radioautography could, in principle, provide far more directly interpretable information. However, the high energy of the radiation emitted by the available nuclides, and ionic diffusion during the prolonged periods of exposure required even at low temperatures, seriously reduce the spatial resolution, so that this approach has been primarily applicable to very large cells (Dick et al., 1970; Horowitz and Fenichel, 1970). The three most promising approaches to the problem have been: (i) direct measurement of the ionic contents of the different subcellular organelles, (ii) introduction of ion-selective microelectrodes into the cell, and (iii) NMR analysis of whole cells and tissues. As discussed below, the data provided by these three techniques are different from, and complementary to, each other. Samples of subcellular organelles have been obtained for analysis in several ways. The most direct general method has been to fractionate broken cell preparations either in hydrophobic solvents or in
4
MORDECHAI SHPORER AND MORTIMER M. CIVAN
aqueous media containing high concentrations of sucrose (Siebert, 1967); however, during disruption of the cell, even at low temperatures, ions and water may in principal be redistributed among the intracellular organelles and cytoplasm. Samples of nucleoplasm and cytoplasm have also been extracted from whole cells with fine tungsten needles; it is'uncertain how representative these samples may be of the intracellular fluids. A less problematic technique has been the microdissection and chemical microanalysis of nuclei from frozen specimens; this approach is practicable only for very large cells (Century et
al., 1970). In recent years, the development of electron-probe X-ray microanalysis has permitted estimation of the subcellular distribution of the ionic contents of certain cells, without resorting to cell fractionation procedures (Tousimis, 1969; Andersen, 1973, Hall et al., 1974; Echlin and Galle, 1975; Goldstein and Yakowitz, 1975).After preparing a section of frozen hydrated or dehydrated tissue, a beam of electrons, often 1 pm in diameter, is focused on the specimen. Electrons are ejected from the inner shells of some of the atoms by the incident electrons. Each ejected electron is replaced by an electron from a higher energy state; this transition is, therefore, accompanied by a loss of energy, irradiated as a characteristic X-ray of the nuclide under study. Mcnitoring the X-rays emitted with either a crystal- or energy-dispersive spectrometer provides a measure of the intracellular content of the element. By focusing the electron beam on different sites within the cell, the intracellular distribution of the element can be determined. This approach promises to be extremely powerful but is currently very much in a developmental phase. Optimal preparative procedures, most appropriate calibrations, and theoretical corrections for the many technical problems are all yet to be fully defined. The technique solely measures the content, not the electrochemical activity of the element under study. The second of the three major approaches to define the nature and composition of the intracellular fluids has been intracellular recording with ion-selective microelectrodes. Suitable electrodes can be fabricated directly from ion-sensitive glasses (Lev and Armstrong, 1975) or prepared indirectly from open-tipped micropipettes by introducing either liquid ion-exchanging resins (Walker, 1971) or, in the case of C1-, a chlorided silver wire into the tip (Neild and Thomas, 1973). Such electrodes directly measure the electrochemical activity of the ion under study, subject to the technical problems of selectivity, sensitivity, and contamination of the electrode tips by cellular proteins (Neild and Thomas, 1974). In order to derive the chemical activity
NMR STUDIES OF WATER AND IONS WITHIN CELLS
5
from measurements of the electrochemical activity, it is also necessary to measure the absolute value of the transmembrane electrical potential. The latter measurement is subject to an intrinsic ambiguity arising from the liquid junction potential, although a reasonable estimate can be obtained. It will be appreciated that microelectrodes sample the electrochemical activities of ions only within highly restricted volumes surrounding their tips. Under these circumstances, the observation of a low value (as is commonly the case) for the activity coefficient of Na+, in comparison to that in the extracellular fluid, could reflect either subcellular compartmentalization or binding. Actually, compartmentalization, at least within the nucleus, seems to be of less quantitative significance than might be thought. Recent evidence obtained with K+-selective (Palmer and Civan, 1975) and C1--selective (Palmer and Civan, 1976) microelectrodes indicates that there is no gradient in chemical activity for K+ or C1- across the nuclear membranes of the giant cells of Chironomus salivary glands. The third major technique, nuclear magnetic resonance (NMR), constitutes the subject of the current review. The second section of this manuscript will present the formal basis for understanding the discussions in Sections IV and V. Section I11 is devoted to a brief presentation of the NMR techniques applied to the study of the intracellular fluids. NMR is different from the electron probe and from microelectrodes in at least two ways. First, it senses the state not only of Na+ and K+, but also of water molecules. Second, it senses the atomic nuclei distributed throughout the entire sample. Each spectral line constitutes a superposition of signals from all the nuclei that are in the same state. Thus, NMR analysis does not provide direct information concerning the spatial distribution of the nuclide within the tissue. For this purpose, indirect techniques involving, e.g., ionic substitutions or addition of paramagnetic substances are required for reasonable inferences to be drawn. As a spectroscopic tool, NMR is similar to the electron probe in being far more specific than ion-selective microelectrodes for the nuclide under study. On the other hand, NMR is like ion-selective microelectrodes and, unlike the electron probe in providing information about not only the amount, but also the state of the nuclide under study. Specifically, the technique quantifies the freedom of molecular tumbling, which need not correlate precisely with the equilibrium activity coefficent obtained with microelectrodes. As will be emphasized later, even small
6
MORDECHAI SHPORER AND MORTIMER M. CIVAN
fractions of bound nuclides may be detected by NMR, depending upon the conditions of exchange among the heterogeneous populations of nuclei. No single one of the above techniques can both define the electrochemical activities within the cell and quantify the degrees of structuring of water and ionic compartmentalization and immobilization. Such information may be provided, under a variety of physiologically important conditions, when all of the above techniques are applied to the same tissues. The technique of electron probe X-ray microanalysis has been presented in a number of reviews and monographs (Tousimis, 1969; Andersen, 1973; Hall et al., 1974; Echlin and Galle, 1975; Goldstein and Yakowitz, 1975). Ion-selective microelectrodes have been discussed in a review that recently appeared in this review series (Lev and Armstrong, 1975). The remainder of the current review will b e concerned with the principles and application of NMR spectroscopy.
II. PRINCIPLES OF NMR SPECTROSCOPY
In this section, we intend to introduce the reader to the concepts and equations we shall be applying to the NMR data in Sections IV and V. The equations are presented in the same explicit form that will be used later in the text. This section in no way constitutes a general introduction to NMR spectroscopy. Atomic nuclei with odd mass numbers, such as 'H, I7O, 23Na,and 39K, possess the properties of nonzero nuclear spin quantum number ( I ) (a dimensionless integer or half-integer) and magnetic moment ( p ) (erg * gauss-'); nuclei with even mass numbers but with odd atomic numbers, such as 2D,also possess these properties. The basis for NMR spectroscopy is the Zeeman interaction of such nuclear magnetic moments with a large imposed magnetic field (Ho) (gauss). A nucleus interacting solely with a magnetic field of intensity H , is characterized by 21 1distinct energy levels, separated by differences in energy of yH, (rad * sec-'), where y (rad gauss-' * sec-l) is the gyromagnetic ratio characteristic of each nuclide; each energy level is characterized by another quantum number m, which can assume the values (I), (I - l ) , ( I - 2) and so on, through ( - I ) . Energy transitions between these levels can be observed as spectroscopic absorption lines at the Larmor resonance frequency (w,) (rad * sec-') given by yH,. Nuclei with spin numbers greater than 1/2 possess electrical quadrupole moments, which interact with the gradient in the local elec-
+
NMR STUDIES
OF WATER AND IONS WITHIN CELLS
7
tric field generated by the molecular environment. This nuclear quadrupolar interaction modifies the magnitudes of the energy differences between the different nuclear energy states. The magnitude of the nuclear quadrupolar interaction is highly sensitive to the symmetry and electron density of the immediate environment of the nucleus under study; the interaction vanishes entirely in the case of spherical or cubic symmetry. There are two modes of application of NMR spectroscopy: continuous wave (cw) NMR and pulsed NMR. Continuous wave techniques are entirely analogous to other spectroscopic methods. As in other spectroscopic techniques, cw NMR is concerned with the measurements of frequency displacements of wo (chemical shifts) and with the shapes and widths of the spectroscopic lines. Continuous wave NMR has been most widely appliedas high resolution spectroscopy of nonviscous liquids where sharp resonance lines occur, providing fingerprints of the molecules studied, and permitting characterization of the electronic structure of molecules containing the nuclide examined. These techniques have also been applied to solids and other phases with restricted molecular motion, which are characterized by broad spectral lines, permitting characterization of the spatial structure of the molecule studied and of its surroundings. Nuclei solely subject to the Zeeman interaction are characterized b y 21 1 equally spaced energy levels. Transition is possible only between two neighboring energy states. Since all such transitions will be associated with the same increment in energy, the spectrum will be characterized by a single absorption line. In the event of additional magnetic dipolar or electrical quadrupolar interactions in the solid state, these energy levels will be shifted, resulting in a spectrum composed of two or more absorption lines; in this discussion, we are limiting ourselves to a consideration only of first order effects on the Zeeman energy levels. In the case of dipole-dipole interactions between pairs of protons, as within water molecules, the spectral line is split into two distinct lines. The magnitude of the line splitting (Am), is given by:
+
(Aw)H
=
(3/4)(yH2h/rH3)(1 - 3 cos28H)
(1)
where yH is the gyromagnetic ratio for 'H, h is Planck's constant (h)divided by 2 r , r(cm) is the interproton distance, and 8 is the angle between the axis of interaction and the steady magnetic field @Io). Here, 8H is the angle between the proton-proton interaction axis and Ho. For nuclei with spin numbers greater than 1/2, the dominant nu-
8
MORDECHAI SHPORER AND MORTIMER M. CIVAN
clear interaction, in addition to the Zeeman interaction, is quadrupolar. As a specific example, Z = 1 for 2D. The magnitude of the nuclear quadrupolar interaction (e2qQ)(rad* sec-') is given by the product of the electric field gradient, and the nuclear quadrupole constant characterizing 2D. Here, the Iine splitting is given by:
(AoJ), = (3/8)(e2qQ),(1- 3 COSz8e,)
(2)
when there is axial symmetry of the electric field gradient; 8, is the angle between the major axis of quadrupolar interaction for 2Dand the steady magnetic field. In the absence of axial symmetry, Eq. (2) must be modified by introducing a dimensionless asymmetry factor r) characterizing the electric field gradient. Specifically, r) is defined as (Vxz - V,,)/V,,, where V,,, V,,, and V,, (= eq) are second derivatives of the electrical potential and constitute the principal components of the electric field gradient. According to the usually used convention, [V,,l B IV,,J 1 /Vzxl (Barnes, 1974). Since V,, V,, + V,, = 0, it is possible to characterize all three components by the two parameters r) and V,? For the most highly asymmetric field gradient, r ) = 1, and for perfect axial symmetry, = 0. For axial asymmetry:
+
(Ao), = (3/8)(e2qQ)D[1- 3 cos28D +
r)
sin20D(sinz@ - cos2@)D1 (3)
where eQ is the nuclear quadrupole moment, and @ is the angle between V,, and the vector perpendicular to the plane containing H o and V z p For nuclei such as 23Naand 39K with a spin number of 3/2, a first order nuclear quadrupolar effect in the solid state produces a central unshifted spectral line (reflecting the transition between the energy levels characterized by rn = ? 1/2); in addition, a pair of satellite lines (reflecting the transitions between the energy levels characterized by rn = 3/2 to 1/2 and by rn = -1/2 to -3/2) is displaced equally about the center. The separation between the satellite lines, in the case of axial symmetry, is given by: = (1/4)(e2qQ),=3d1-
3 cos28) (4) The magnitude of the central line is 40% of the total integrated intensity of the entire spectrum, with the remainder distributed equally between the two satellite lines (Abragam, 1961). / ~given by an expression In the absence of axial symmetry, ( A o J ) , = ~is analogous to Eq. (3): (Ao)l=3/2
( A U ) , = ~ /=~ (1/4)(e2qQ),=,/2[1- 3 cos20 +
r)
sin20(sin2@- cos2@)] ( 5 )
When the nuclear spin number equals 5/2, the nuclear quadrupolar
NMR STUDIES
OF WATER AND IONS WITHIN CELLS
9
interaction produces an additional pair of satellite lines. The line splitting is given by an equation entirely analogous to Eqs. (4)and (5), in the presence and absence of axial symmetry, respectively. Once again, the magnitude of the effect is proportional to a constant characteristic of the nuclide and proportional to the same geometric factor. In this case, the central spectral line constitutes 27%of the total integrated spectral intensity (Abragam, 1961). In the event that the sample studied consists not of a single crystal, but of a large number of randomly oriented crystals, a superposition of the contributions from all possible values of 8 and @ will be obtained. The spectrum of Fig. 1 characterizes such polycrystalline samples for nuclides with a spin number of 3/2 in the presence of axial symmetry. It should be noted that the stronger the quadrupolar interaction, the greater the frequency range of the shifted spectral contributions. The relative integrated intensity of the contributions from the satellite signals is fixed at 60% of the total integrated intensity. Therefore, the greater the quadrupolar interactions, the smaller must be the peak intensities of the satellite lines, and the less favorable will be their signal-to-noise ratios. In a rotating molecule, the average value of (1 - 3 cos28)may b e expressed in terms of the cosines of two other related angles. For any molecule, it is always possible to define a principal axis of molecular rotation. We define r#~to be the angle between the principal axis of rotation and the steady magnetic field H,;p is defined as the angle
w
FIG. 1 . CW spectrum from nuclide with I = 3/2 subjected to a first-order quadrupolar interaction with axial symmetry, in a polycrystalline sample. The abscissa is the frequency of the applied rf field. The ordinate is the intensity of the energy absorbed.
10
MORDECHAI SHPORER AND MORTIMER M. CIVAN
between the principal axis of rotation and V,? In the case of cylindrical symmetry, both for the electric field gradient and for the molecular movement, the geometric factor can be related to @ and p by Eq.
(6): (1 - 3 cos2e) = ( 3 co~2p- i)(i- 3
~ 0 ~ 2 4 ) ~
(6)
where the bar indicates that an average has been taken over the range of molecular movement. In the absence of axial symmetry for the field gradient, but with axial symmetry of molecular motion (Resing, 1976). 1 - 3 cos2e - r] sin28(sin24- cos2+) = [3 cos2p - 1 - r] sin2p(sin2p- cos2p)l(1 - 3 c0s24)/2 (7)
where /3 is the angle between V,, and the vector perpendicular to the plane containing V,, and the principal axis of rotation. The angles p and p are characteristic, then, of the nucleus within the molecule, while 4 is characteristic of the molecule with respect to the external
field. In certain sample systems, as in liquid crystals, there is a preferred axis of molecular alignment. Under these conditions, Eq. (7) must be modified by the introduction of a dimensionless molecular ordering factor cf,), expressing the degree of orientation of the molecule with time. A value offo = 1 implies that the molecule is constantly ordered along one axis; a value of 0 implies that the molecule is randomly oriented with time. The final expression for the line splitting becomes:
(Aw)= (strength of interaction)(fo)[(3 cos2p - 1) - r] sin2p(sin2p- c o s 2 ~ )1 l (- 3 c0s24)/2
(8)
Comparing Eq. (8)with Eqs. ( l ) , (2), and (4),it will be appreciated that the final form of the line splitting is similar to that seen in the solid state, but with the effective interaction scaled down b y the molecular ordering factorf,, and by the geometric factor, which depends on the angles p and p. Pulsed NMR techniques are different from other spectroscopic approaches, Although the two NMR techniques are not completely independent, equivalent results are occasionally obtained from cw techniques and from Fourier transforms of pulsed NMR data, as described in detail by Farrar and Becker (1971).The parameters of relaxation obtained with pulsed methods provide measures of the dynamic state of the molecules examined, and have been particularly promising in the study of the intracellular fluids. The remainder of this section will be
11
NMR STUDIES OF WATER AND IONS WITHIN CELLS
concerned with a brief presentation of the relaxation parameters measured by pulsed NMR. In general, a magnetic absorption line can be characterized by two relaxation times, TI and T,, both expressed in seconds. At equilibrium, the imposed steady magnetic field H o causes preferential alignment of the nuclei along the z or longitudinal axis, producing a x component ( M , ) of the bulk magnetization. In the xy plane transverse to Ho,there is no preferred direction of alignment for the nuclear dipoles; therefore, at equilibrium, the x and y components (M,and M,, respectively) of the bulk magnetization are both zero. TI, called the longitudinal or spin-lattice relaxation time, is a measure of the time required for M , to achieve its equilibrium value. T,, termed the transverse or spin-spin relaxation time, characterizes the time required for M , and M , to return to zero. The rate of transverse relaxation determines the spectral line width of the signal observed with cw NMR: at halfpeak intensity, this width = 2 / T , (rad sec-I). Longitudinal relaxation proceeds by an exchange of energy between the Zeeman energy of the aligned dipoles and the several modes of energy in the surroundings of the lattice. Transverse relaxation may reflect an exchange between different modes of energy or an exchange of magnetic energy exclusively among the nuclear spins. Therefore, T , can never exceed T,. Both T I and T , are influenced by time-dependent magnetic and electrical interactions with the nuclei studied. These interactions are either dipolar (for the proton-proton interaction of water) or quadruI7O, 23Na,and 39K). Nuclear interactions are particularly polar (for 2D, effective in enhancing longitudinal relaxation when the fluctuation frequencies are high, approaching the Larmor frequency. On the other hand, transverse relaxation is enhanced even more by slowly or nonfluctuating interactions. Although the nuclear times can be influenced b y many sources of interaction, in this review w e shall be solely concerned with systems where TI and T , are determined specifically by fluctuating magnetic dipolar and electrical quadrupolar interactions. Under these conditions, T , and T , may be related to three factors: the experimental Larmor frequency (wo) determined by Ho,the magnitude of the fluctuating interactions, and the correlation time (T,)(sec), which is a measure of the rate of the fluctuations. The type of application of pulsed NMR dealt with in this review is largely based upon our ability to estimate and monitor T~ from experimental measurements of the relaxation times. We now proceed to equations that describe transverse and longitu-
-
12
MORDECHAI SHPORER AND MORTIMER M. CIVAN
dinal relaxation behavior under the assumption that the relaxation processes are primarily determined by rotational tumbling. This model will be appropriate for our later discussion. However, diffusional translational motion modulates the absolute magnitude of the intermolecular dipolar interaction and, in a given experimental condition, translational motion may also modulate the absolute magnitude of the quadrupolar interaction of ions (Sutter and Harmon, 1975). In any case, the functional interrelationships among the various NMR parameters will be similar. The aim of the remainder of this section is to provide the reader with an appreciation of how the parameter of primary physiological interest, 7,, depends upon the directly measurable parameters, T,, T 2 ,and ow We shall be discussing only the three nuclides of water (‘H, 2D,and 1 7 0 ) and the alkali cations 23Naand 39K. Among these nuclei, only the relaxation times of water protons are determined by magnetic dipolar interactions. Each proton of water is subject to both an intramolecular interaction with the other water proton, and an intermolecular interaction with the protons of neighboring water molecules. The rate of longitudinal (l/TJ and of transverse (l/TZ) relaxation may each be considered to consist of intramolecular and intermolecular contributions.
(1/TJ = (l/Ti)intra + (J-/TJinter
(9)
(1/7’2) = (1/T2)intra + (l/Tz)inter
(10)
In normal liquid water the intramolecular interaction contributes motions modulating the intramolecular dipolar interactions, the relaxation rates may be defined in closed forms:
where r is the internuclear distance and ii = h/(27r). For the primarily translational motions involved in the intermolecular interactions, precise formulations are not available in closed form. However, a similar functional dependence on T, and oo holds, and 7, is of the same approximate order of magnitude for translational and for rotational motion in water (Abragam, 1961). Apart fiom ‘H, the nuclides to be discussed are characterized by spin numbers I > 1/2, whose nuclear relaxations are, therefore, primarily determined by electrical quadrupolar interactions. The spe-
NMR STUDIES OF WATER AND IONS WITHIN CELLS
13
cific dependence of the relaxation times upon wo and T, is determined by the value of 1. In each case, the treatment is based upon fluctuations in the direction, but not the magnitude, of the gradient in electric field applied locally to the nucleus; this is the most probable basis for fluctuations in the quadrupolar interactions to be examined. For those nuclides such as 2Dwith I = 1, the relationship can be expressed in the following closed form (Abragam, 1961):
For nuclides such as 23Naand 39K with I = 3/2, both T , and T , consist of two components (Hubbard, 1970; Rubinstein et al., 1971).
Only 20% of the signal intensity decays with the longitudinal relaxation rate (l/Tl)I,while the remaining 80% decays at the rate (l/TJlI, Since the two rates differ at most by a factor of 4, and since ( 1/TJI reflects only 20% of the total signal, (l/Tl), is not readily detectable. The two rates of transverse relaxation for nuclides with Z = 3/2 are:
( 1/T2)1characterizes the decay of 40% of the signal and can be attributed to the transition between the + (1/2) and - (1/2) Zeeman energy levels. The remaining 60%of the signal decays at the rate ( 1/T2)11, and can be attributed to the transitions between the +(3/2) and + (1/2), and between the - (1/2) and - (3/2) energy levels. It is important to recognize that the same fractionation into two components comprising 40% and 60% of the total signal must be observed in cw experiments, as well. The widths of the two superimposed lines will b e 2( 1/T2)1and 2(1/T2)11.
14
MORDECHAI SHPORER AND MORTIMER M. CIVAN
For nuclides with I 2 5/2, the relationships among the relaxation times, wo and T,, are considerably more complex. For nuclei with a spin number of 5/2 such as 1 7 0 , both T , and T , consist of three components, which cannot be expressed in closed form. To a first approximation, these components exhibit dependences upon wo and T~ similar to those of Eqs. (11)-(18). However, according to Rubinstein et al. (1971), only one of the three components of T , would be dominant for all values of T~ and a,,. The rate of longitudinal relaxation can b e well approximated (Rubinstein et al., 1971) by:
On the other hand, for certain values of T~ and wo, several components of the transverse relaxation could be observed for a single homogeneous population of nuclei. In addition to the effects arising from rotational fluctuations, fluctuations in the magnitude of the electric field gradient generated by diffusion may also be of significance for ions in aqueous media (Sutter and Harmon, 1975). Even in the latter case, the dependence of the relaxation times upon wo and T~ will assume a similar form. However, for nuclides with 1 ? 3/2, the components of the relaxation rates may converge to a single term. All of the above Eqs. (11-19) express a functionally similar dependence of the relaxation rates upon T, and wo, both for dipolar and quadrupolar interactions. If T~ is very short (w07, << l),the condition of “motional narrowing” observed for small molecules in nonviscous solutions, all of the relationships converge to precisely the same form:
(l/T,)
=
( 1/T2)
(strength of interaction).r,
(20)
This condition is commonly met, since in aqueous solution at room temperature, T~ for free ions and water molecules is of the order of magnitude of 10-llsec. Under these conditions, the relaxation times are independent of the Larmor frequency. For nuclei with 1 > 1, each relaxation rate also consists of only one component. If the correlation time is very prolonged, the longitudinal relaxation process becomes inversely dependent on 7, and is, thereby, ineffective. However, under these circumstances, one or more components of the transverse relaxation process becomes directly proportional to T,, and therefore, maximally effective. When wo becomes sufficiently large so that ( W ~ T , ) >> 1, ( l / T J becomes strikingly dependent upon the Larmor frequency [Eqs. ( 1 1,13,
NMR STUDIES OF WATER AND IONS WITHIN CELLS
15
15, 16, and 19)l. This effect is less obvious for most of the components of (1/T2) because of the first term within brackets in Eqs. (12, 14, and 18). The phenomenon of the frequency dispersion of ( l/Tl) appears to be of considerable importance in biological applications; its significance as a biological probe will be considered later. The bulk of the data to be discussed have been obtained at values of wo ranging from 10' to 109 rad * sec-'. Therefore, the condition that W ~ > T>~ 1 necessarily implies much more prolonged correlation times than those noted above, characteristically observed for ions and water in aqueous solution. It should be pointed out that the above equations have been developed under the condition that (strength of interaction) * T~ < 1. Baram et d.(1973) have, however, extended the treatment to slower molecular motions. Until now, we have considered only homogeneous samples of nuclei characterized by single values of TC and nuclear dipolar or quadrupolar interactions. However, biological systems generally contain heterogeneous populations of nuclei, each population characterized by a different set of NMR parameters. Under these circumstances, the behavior of the relaxation processes will depend upon an additional set of parameters, the mean lifetimes ( 7 ) within each population. Although the measured values of both T I and T , are influenced by ) the components of the sample, T , is the rates of exchange ( 1 / ~among also influenced by the magnitudes of the chemical shifts characterizing the different populations. When the chemical shifts are small (i.e., less than the width of the spectroscopic line, using cw NMR), as in the systems we shall consider here, T , and TI are influenced similarly by the phenomenon of exchange between populations of nuclei. Specifically, three possible cases of exchange can be distinguished (Woessner, 1961b; Woessner and Zimmerman, 1963). In the event of ~< l/Tl, l/T,), the NMR signal (cw NMR) and slow exchange ( 1 / < decays of magnetization (pulsed NMR) consist of superimposed components independent of T reflecting the relative magnitudes of the separate populations. In the event of an intermediate rate of exchange ( 1 / ~ 1 /TI, 1/T2), the multiple components are still distinguishable, but the observed magnitudes of T1,TZ,and of the relative intensity of the signal component for each population will depend both upon the intrinsic values of TI, T,, and P i (the relative mole fraction) and upon 1 / ~It. is only in this case that measurements of the relaxation rates ~> l/Tl, permit estimation of 7. In the event of fast exchange ( 1 / > 1/T2), even a heterogeneous system will behave as an apparently homogeneous sample, characterized by a single set of NMR parame-
-
16
MORDECHAI SHPORER AND MORTIMER M. CIVAN
ters; each of these parameters constitutes a weighted average, independent of 7, of the contributions from the different populations.
2Pt=1
(23)
i
It is for this reason that differentiation between a simple homogeneous system and a heterogeneous one with rapid exchange is a common, important and difficult problem in NMR analysis. Analysis becomes even more complex when dealing with heterogeneous populations of a nuclide with I 2 3/2. As discussed above, the relaxation times of even homogeneous samples of such nuclei consist of multiple components. Fortunately, considerable simplicity is provided by the fact that each of the two components of f,for nuclei with I = 3/2 behaves as if exchanging independently with the corresponding component from other fractions (Bull, 1972).In the event of rapid exchange among different populations of a nuclide, two components will be observed if the relaxation process of one or more of the populations consists of two components (i.e., if wort 2 1 for one or more of the populations). It should be emphasized that the relative magnitudes observed for the two components of TI and T, must necessarily remain constant, irrespective of the fractional distribution among the several populations. The two components will be expressed similarly in cw experiments. For those nuclei with I 2 5/2, analysis is more complex, and it is not possible to express in closed form the combined effects of multiple components and population heterogeneity on the transverse and longitudinal relaxation behaviors. On the basis of the principles of NMR spectroscopy presented earlier in this section, we can now provide a working definition of “binding” in terms of the correlation time rC.At this point, it should be clear that NMR provides a measure of the rate of molecular tumbling. In this context, we take binding to refer to a marked decrease in this rate of tumbling, reflected by a much greater value of the correlation time rc than that in bulk aqueous solution. Because of the many connotations associated with “binding.” we shall use the more restrictive term “immobilization” in the remainder of the review. The above definition of immobilization has been found to be opera-
NMR STUDIES OF WATER AND IONS WITHIN CELLS
17
tionally useful. Ionophores that form stable defined complexes with Na+ have also been found to lengthen the 7c of 23Na(Haynes et al., 1971; Shchori et al., 1971; Shchori et al., 1973); here, the measured value for T , approaches that for the larger ionophore molecule. In general, 7, will be directly dependent upon the size of the complex. However, if the mean lifetime ( T ~of) the complex is shorter than the correlation time of the complex, the measured value of T~ will reflect T M . In contrast to the interaction between alkali cations and ionophores, a simple coulombic interaction between these ions and negatively , charged sites on macromolecules would not markedly prolong T ~ and would therefore not be considered an example of immobilization. 111.
TECHNIQUES OF NMR SPECTROSCOPY
It is clearly beyond the scope of this review to present a detailed discussion of current NMR techniques. There is merit, however, in providing a brief summary of these methods, at least in order to acquaint the reader with the nomenclature of this field. In both cw and pulsed NMR experiments, a large steady magnetic field of intensity H , is established along the z axis. Typically, H , is of the order of 14-20 kG, but may be three to four times larger with superconducting magnets. Following the establishment of a bulk magnetization vector within the sample along the z axis, a much smaller rf signal is established along the y axis, inducing a nonzero component of the magnetization in the xy plane; the magnitude of the rf field can range from a fraction of a gauss in typical cw experiments to tens of gauss in pulsed NMR studies. The receiver coil oriented in the xy plane detects the appearance of an induced component in this plane. The magnitude of the rf energy absorbed is negligible until the frequency of the rf field approaches that of the Larmor frequency of the nuclide under study and is maximal at this resonance frequency (w,). In a cw experiment, either the frequency ( w ) of the applied rf field is vaned, keeping Ha (and therefore 0,) constant, o r w is held constant and w, varied ( by causing Ha to scan over a very small excursion), depending upon the specific instrument used. In pulsed NMR, the direction of the equilibrium magnetization vector is temporarily displaced by an angle J, from the positive z axis, and measurements are made of the rate at which equilibrium is reestablished. With suitable adjustment of the magnitude ( H I )and duration ( A t ) of a brief rf pulse along the x axis, any desired displacement can be achieved. y H , A t = J, (24)
18
MORDECHAI SHPORER AND MORTIMER M. CIVAN
Usually, At ranges from several to several hundred microseconds. It is particularly convenient to apply a pulse (90" pulse) displacing M by go", or a pulse (180" pulse) displacing M by 180". After a 90" pulse, the initial magnitude of the induced component in the xy plane (M,) is that of the equilibrium value ( M , ) for M , just before the pulse. M , then decays back to its equilibrium value (zero) with a time constant Th. This decay of M , following a 90" pulse is called the free induction decay (FID). Simultaneous with the FID, M , returns to M o with a time constant TI. In a perfectly homogeneous field, Ti will be identical with the true T , characterizing transverse relaxation within the sample. However, in the presence of a nonhomogeneous magnetic field, the course of a FID can be considerably shorter. This phenomenon can be most easily appreciated by considering the linewidth of a cw signal. The width of the line is, as noted above, (2/T,) at half-peak signal intensity. Field inhomogeneity causes slightly different local field strengths within the sample, inducing a spectrum of Larmor frequencies, rather than a single value. The resonance signal will therefore be broadened, equivalent to decreasing the observed value of T,. It is technically possible to minimize the effect of field inhomogeneity by applying a particular sequence of rf pulses, termed the Carr-Purcell sequence, or one of its modifications. This technique is particularly effective in the study of slow rates of transverse relaxation. The approach constitutes an improvement of the original spin-echo method devised by Hahn (1950), a technique which is still used occasionally. For further details concerning these experimental techniques, the reader is encouraged to consult the particularly lucid monograph of Farrar and Becker (1971). Although T , may be measured directly from the free induction decay following a single 90" pulse, measurement of TI is necessarily slightly more indirect. A 180" pulse is applied to the sample, inverting the equilibrium magetization M,. In principle, calculation of TI is based upon measurement of the time decay of M , back to M,, once again aligned in the direction of H,,. However, insofar as the receiving coil is oriented in the xy plane, M , cannot be measured directly. Instead, a second pulse-90" in this case-is applied at a given time after the initial 180" pulse. The value of the magnetization vector just after the 90" rotation is taken to be the value of M , just before the second pulse. M , can thus be measured as a function of the time interval between the applied pairs of pulses. The measurement ofT, at very low Larmor frequencies is of particular theoretical interest. As may be appreciated from Eqs. ( l l ) , (13),
NMR STUDIES OF WATER AND IONS WITHIN CELLS
19
(15),and (19), T , is a function of both T~ and w,. In order to calculate T~ directly from measurements of T , , it is necessary to measure T , over a range of w, where the frequency dependence is clearly measurable, i.e., where W , T ~ -- 1. We are most interested in measuring T~ for immobilized complexed ions, where T~ is markedly prolonged. Therefore, it is of great interest to measure T , in the frequency range where a, is several orders of magnitude smaller than that usually achieved by conventional values of H , . It is not practicable to examine T , over this low-frequency domain simply by lowering H,; the result would be to reduce unacceptably the instrumental sensitivity, which is directly dependent on H , . Instead, two imaginative experimental approaches have been used to determine T , at very low field strengths. Koenig and his colleagues have measured TI at low fields by means of a rapid switching technique (Koenig and Schillinger, 1969a). After permitting the nuclei under study to come into equilibrium with a large external field, the sample is subjected to a much lower field. After varying time intervals, the sample is returned to the high field and the magnetization vector measured, permitting measurement of the decay of the magnetization at the low field. In principle, T , may be examined at any value of H , , permitting determination of the entire functional dependence of TI on wo. This approach is limited by the delay time (of some 10 msec for the equipment used by Koenig) required for switching the field, so that nuclei with very rapid longitudinal relaxation rates cannot be studied with this instrument. A second less direct approach has been the measurement of ‘‘TI,” or ‘‘T, of the rotating frame,” equivalent to measuring T , at low magnetic fields of the magnitude of H , (Farrar and Becker, 1971). After permitting equilibration with a large external field, a 90” pulse is applied, followed by another long rf pulse of intensity H , , 90” out of phase with the first. At the conclusion of the second pulse, the magnitude of the magnetization is determined. Thus, the decay of magnetization can be studied as a function of the duration of the second pulse. The net effect of this approach is to apply a low magnetic field of intensity equal to that of the rf field H , . The frequency range practicable with this approach is limited. rf fields greater than tens of gauss cannot be achieved because of technical problems. IV.
NMR STUDIES OF WATER
In this section we examine the NMR properties of water in certain model systems of biological interest, as well as within the intracellular
20
MORDECHAI SHPORER A N D MORTIMER M. CIVAN
fluids. The N M R properties of intracellular water were reviewed by Walter and Hope (1971); considerably more information is now available. Certain broad conclusions can b e . drawn from the published data. However, despite the considerable attention devoted to this problem, our detailed interpretation of the N M R data, both in model and biological systems, remains incomplete. First, we consider data obtained with ordered model systems and with aqueous protein solutions of particular relevance to biological tissue, and then we shall discuss some of the published studies of water within the intracellular fluids. It should be emphasized once again that we are not attempting to present an exhaustive compilation of the literature. Rather, the aim of this review is to provide a coherent view, based on NMR studies, of our current knowledge of the intracellular state of water and alkali cations. Three nuclides of water are available for N M R analysis: 'H, 2D,and "0. In bulk aqueous solutions at room temperature, the molecular tumbling of water is isotropic, and is customarily characterized by a single correlation time (7J. At room temperature, 7c = 3 x sec (Hertz, 1973; Deverell, 1969), so that under standard experimental conditions, w,gC << 1. Under these circumstances, each of the nuclides of water is characterized by a single spectral line (by cw analysis), and the spin-lattice (TI)and spin-spin (T,) relaxation times consist of single components that are equal. The rates of longitudinal relaxation (l/Tl) and of transverse relaxation ( 1/T2)for water protons are determined by both intramolecular and intermolecular dipolar interactions. The rates of relaxation for both 2Dand 1 7 0 arise from intramolecular quadrupolar interactions. In bulk aqueous solution at 23"C, the total rates of nuclear relaxation for 'H are 0.294 sec-I, while the intramolecular contributions have been determined to be 0.163 sec-' (Krynicki, 1966). The relaxation rates for 2Dand ''0 are 2.34 sec-' (Woessner, 1964) and 145 sec-' (Hindman et al., 1970), respectively. The longitudinal relaxation rate of deuterons has been measured in a solution containing more than 99.5% 2D20;therefore, for purposes of comparison, the value given above must b e divided by 1.24, the relative viscosity of 2D,0 to that of H 2 0 at 23°C (Hardy and Cottington, 1949). The relaxation rate of interest for 2D becomes then 1.87 sec-'. We shall not discuss further the physicochemical properties of water and simple solutions, as reflected in the published NMR data. For this purpose, the interested reader is referred to the comprehensive review of Hertz (1973).
NMR STUDIES OF WATER AND IONS WITHIN CELLS
21
A. Ordered Model Systems
Doublet spectra have been reported for water 'H or 2D in zeolites (Ducros, 1960), montmorillonoid clays (Hecht et al., 1966; Woessner and Snowden, 1969a,b; Hecht and Geissler, 1970), collagen (Berendsen, 1962; Migchelsen and Berendsen, 1967; Dehl and Hoeve, 1969; Fung and Trautmann, 1971; Fung and Siegel, 1972; Fung and Wei, 1973), concentrated soap pastes (Charvolin and Rigny, 1969; Blinc e t al., 1970), Li-DNA (Migchelsen et al., 1968),keratin (Lynch and Haly, 1970), rayon (Dehl, 1968), and a polysaccharide polymer (Woessner and Snowden, 1973). Quadrupole splitting has been observed for l'0 from H2170 in unoriented liquid crystals of sodium linoleate in water (Civan and Shporer, 1972) and in oriented crystals of sodium decyl sulfate-decyl alcohol-NazS04 in water (Fujiwara et al., 1974). Although some degree of ordering of the water molecules is clearly present in these systems, further analysis is limited if only a single water nuclide is studied in each experiment. However, from an analysis of two or more of the nuclides in the same system, we may hope to arrive at a more detailed understanding of the orientation of the water molecules. The basis for this is the expectation that the geometric factor cited above, which depends upon the angles p and p within the molecular frame, should be different for all three nuclides (Verhoeven et al., 1969). A number of comparative studies of the line splittings of 2D and 'H have been carried out with preparations of oriented material. In these systems, it was possible to change the orientation of the system with respect to the external magnetic field H,. The magnitudes of the line splittings for 2D and 'H were strongly dependent upon 0, as expected from Eqs. (1) and (2). However, the ratio of the line splittings (AwH/AwD) remained constant. In general, in the solid state, the line splitting for each of the three nuclides would be expected to exhibit a different angular dependence. The fact that 2Dand 'H were characterized b y the same angular dependence indicates that the molecular motion of water results in movement of these two nuclides about a common axis. The quadrupolar and dipolar coupling coefficients are likely to be similar in bulk solutions and in the model systems cited. T h e dipolar coupling constant of 'H does not depend on phase, since it is solely a function of the intramolecular proton-proton separation. In addition, the quadrupolar coupling constant for 2D has been calculated to b e
22
MORDECHAI SHPORER AND MORTIMER M. CIVAN
-
approximately 1.45 x lo6 rad sec-' in water (Powles et al., 1966) on the basis of measurements of T , . The value in ice has been found to be 1.35 x lo6 rad sec-' (Waldstein et al., 1964) on the basis of quadrupolar splitting in NMR experiments at high magnetic Aelds, and 1.34 x los rad * sec-' from more direct measurements of pure quadrupole resonance (Edmonds and Mackay, 1975). The quadrupolar coupling constant for ''0 in water is approximately 49.6 x lo6 rad sec-' calculated from measurements of T , (Hindman et al., 1970); in ice, it is 42.1 x lo6 rad sec-' from NMR experiments at high magnetic field (Spiess et al., 1969) and 40.8 x lo6rad * sec-' obtained directly from pure quadrupole resonance (Edmonds and Zussman, 1972).Therefore, even on the basis of so extreme a comparison as that between liquid and solid phases of water, the ratio of the quadrupolar constant for 2D to that of 1 7 0 is unlikely to be very different in the model systems. The ratio of the quadrupolar constant for 2Dto the intramolecular dipolar interaction constant for 'H is likely to differ even less in the model systems. These considerations suggest that the coupling constants for l'0, 2D,and 'H are likely to be approximately the same in bulk water and in these ordered systems. We may express, then, the quadrupolar and dipolar coupling constants in terms of the rates of longitudinal relaxation for the water nuclides in bulk aqueous solution, and substitute these expressions into Eq. (8).Here, for simplicity, cylindrical symmetry is assumed for the molecular motion of the water; in the absence of such symmetry, a different equation incorporating an additional set of parameters must be used (Saupe, 1964). From Eq. ( l l ) ,under conditions of motional nar, < 11, the intramolecular contribution to the rate of rowing [ O J ~ T< longitudinal relaxation (l/T1)" for 'H is given by:
-
(I/TI)H = (3/2)(yH4h2/r6)(Tc)
(25)
We assume spherical symmetry for the molecular movement of water, so that the correlation times for the three nuclides become identical. Therefore, from Eqs. (13) and (19),the rates of longitudinal relaxation for 2D and 1 7 0 are given by:
(~/TI)D = (3/8)(1 -k V D ' / ~ ) ( ~ ' ~ Q ) D Y ~ C )
(26)
Replacing the dipolar or quadrupolar coupling constants of Eq. (8)by the appropriate expressions from Eqs. (25-27), we may obtain the ratios of the geometric factors for the three nuclides:
NMR STUDIES OF WATER AND IONS WITHIN CELLS
23
(3 Cos2pH - 1) [3 COS’PD - 1 - q D sinz~(sin2pD/cos2pD)1 = (AwH/AoD)[l/( 1 + r)D2/3)]1’2[( l/T])D/( 1/T1)HI1” (28) [3 cos’p, - 1 - qo sin2po(sin2j30- cos2po)] [3 C O S ‘ ~ ~ , 1sin2pD(sin2PD - CoS’pD)] = 2*53(Ao,/b)[(l
+ 7)0’/3)/(1+ qD2/3)I1”[(
1/~~)D/(1/~1)01”’
(29)
In those experimental systems where the line splittings of two or more nuclides have been measured at the same time, application of Eqs. (28) and (29) should provide further information concerning the orientation of the water molecules. The line splittings for 2Dand ‘H of water have been measured in several model systems; only the absolute values, and not the signs, of these splittings can be determined. The most precise data have been obtained with collagen (Chapman and McLauchlan, 1969; Migchelsen and Berendsen, 1973) and with montmorillonoid clays (Woessner and Snowden, 1969a,b). Chapman and McLauchlan found that the ratio of the line splitting of ‘H (A& to that of 2D (AoD) was 0.312. The data of Migchelsen and Berendsen are consistent with this value; from their results, the ratio can be calculated to be between 0.297 and 0.314, depending on the water content of the preparation. Woessner and Snowden found ( A w H / A w D ) to be 0.260. Similar values have been found by other investigators for collagen (Migchelsen and Berendsen, 1967; Dehl and Hoeve, 1969), Li-DNA (Migchelsenet al., 1968), and rayon (Dehl, 1968).As pointed out by Woessner and Snowden (1973), the similarity of the ratio of the line splittings in these chemically very different systems suggests that the interface, rather than the substrate, is the primary determinant of the line splittings. Unfortunately, the line splittings have been determined simultaneously only once for 170(Awo) and ‘D (AwD). From the results of Fujiwara et al. (1974), (Awo)/(AwD) may be calculated to be 10.13. By applying Eqs. (28) and (29) to these data, and using measured values of qo and qD,it is possible to demonstrate that the two simplest and most often suggested interpretations of the NMR measurements of water nuclides in these model systems cannot be correct. The value of T - / ~has been measured to be 0.135 (Verhoeven et al., 1969) for ‘H’DO in the gas phase. Edmonds and Mackay (1975) have found qDto b e 0.121 for ‘H’DO and 0.112 for 2 D z 0in the ice phase. The value for r), is close to one and is highly sensitive to the molecular environment of the 0 nuclide. In gas, rl0 is 0.75 (Verhoeven et uZ., 1969), and in ice, qo = 0.925 (Edmonds and Zussman, 1972). In
24
MORDECHAI SHPORER AND MORTIMER M. CIVAN
crystals of BaC10, * H2170,vo has been found to be 0.91 from pure quadrupole measurements (Shporer and Achlama, 1976). Introducing the experimental values into the right-hand sides of Eq. (2%
( 3 COSP ' H - 1) = 0.9 to 1.1 (30) [3 cos'p, - 1 - (qDsin2pD)(sin2pD - cos'pD)1 while for the single simultaneous study of
and 'D,
170
[3 cos2po - 1 - (v0 sin2po)(sin2Po- C O S ~ / = ~ ~3.3 )] [3 C O S ' ~ D - 1 - ( q D sin2pD)(sin2PD - COs'pD)]
(31)
On the basis of Eqs. (30)and (31), we are now in a position to examine the possible physical bases for the NMR data obtained with water nuclides in ordered model systems. The simplest interpretation of the data obtained with these model systems has been that of a rapid exchange between a small fraction of immobilized water on the surface of the ordered nonaqueous phase with a much larger fraction of molecules having the properties of bulk water. However, from such a model, we would expect different angular dependences for the line splittings of the three nuclides of water, contrary to the experimental results obtained comparing protons and deuterons. This experimental observation requires us to conclude that at least the protons and deuterons of the water molecule must share a common axis of rotation within the molecular frame. In order to pursue this analysis, we must now consider the geometry of the whole water molecule and of the principal axes of interaction characterizing each of the nuclides. The H-0-H angle of water in the gas phase has been found to be 104.4' by infrared spectroscopy (Herzberg, 1945). In the solid phase, this angle is considerably widened. From X-ray analysis, Peterson and Levy (1957) obtained a value of 109.2". From NMR studies of 'D in Chiba (1963) has measured a D-0-D angle of 110". BaClO, * 2Dz0, From fine structure analysis of pure nuclear quadrupole resonance of I7O, Shporer and Achlama (1976) have estimated the H-0-H angle to be 110". It seems likely that the H-0-H angle in the model systems is closer to values observed for water in the solid phqse, in ice, and in hydrated salt. We will therefore presume the angle to be 109.2' in the subsequent discussion. We can also identify the orientation of the principal axis of quadrupolar interaction for 2D with considerable certainty. Chiba (1963) found from his NMR studies of hydrated BaC10, that this axis must
NMR STUDIES OF WATER AND IONS WITHIN CELLS
25
have been directed along the O-D bond, or close to it. Similarly, in the gas phase, Verhoeven et uZ. (1969) found that the principal axis of interaction for the 'D of ' H 2 D 0 lay within 1.4"of the O-D bond; this slight deviation may have arisen from the H-D asymmetry of the water molecule studied. The second greatest component of interaction (V,,) for 2Dhas been found to be perpendicular to the H-0-H plane for water in the gas phase (Verhoeven et al., 1969). Since qD within water is very little different in the gas and solid phases, it seems highly likely that V,, is also perpendicular to the plane of the water molecule within the model systems investigated. Both the values of the quadrupolar coupling constant and qo characterizing the I7O of water are considerably less well defined, and more sensitive to the absolute value of the H-0-H angle than in the case angle is close to a perfect tetrahedral angle, so of *D. The H-0-H that the two major components of the electric field gradient approximately equal each other; the third component is much smaller. Therefore, in this case, qo is close to one. In the gas phase, the H-0-H angle is slightly less than the perfect tetrahedral angle. Here, qo = 0.75, the smallest component is oriented along the bisector of the H-0-H angle, the largest component is perpendicular to the molecular plane, and the intermediate component lies within the molecular plane perpendicular to the bisector. On the other hand, in ice, the H-0-H angle is not significantly different from the perfect tetrahedral angle (Peterson and Levy, 1957), and q,, = 0.935. In this case, the smallest component of the electric field gradient remains oriented along the bisector. However, the largest component (Vzz)and the intermediate component (VaJ may either retain the same orientations found in water vapor, or they may actually angle beswitch their orientations with one another if the H-0-H comes greater than the perfect tetrahedral angle. The orientation of the components of the electric field gradient exerted on the I7O nucleus within water in the solid state has also been examined with pure quadrupole resonance studies of hydrated BaClO, (Shporer and Achlama, 1976). In this case, the data obtained for the 1H-'70 hyperfine interactions can best be fitted with V,, oriented within the molecular plane, perpendicular to the bisector, contrary to the case for water vapor. Here, the quadrupole coupling constant was found to be 7.7 MHz (48.4 x lo6 rad . sec-I), a value close to that found from T, measurements of liquid water, and intermediate between the 10.2 MHz (64.1 x lo6rad * sec-l) obtained from uhf measurements of water vapor and the 6.7 MHz (42.1 x lo6
26
MORDECHAI SHPORER AND MORTIMER M. CIVAN
rad - sec-I) obtained from pure quadrupole resonance measurements of ice; T~ was found to be 0.91 for the 1 7 0 of Ba(C10,), * H2170. On the basis of these data and for the purposes of the following discussion, we take qo = 0.935, and we consider both possible orientations of V,,, perpendicular to the molecular plane and within the molecular plane, perpendicular to the bisector of the H-0-H angle. As noted above, we take the H-0-H angle to be 109.2". With these values, we can now consider the simplest possible model of oriented water. A fraction of the water molecules are considered characterized by a preferred axis of rotation colinear with the bisector of the H-0-H angle and perpendicular to the substrate surface. Within the framework of this model, pH = go", p,, = 54.6", and PI,= 0"; po = 90" and Po = 0" for both orientations of the V,, applied to the 1 7 0 nucleus. Inserting all of these values into Eqs. (30)and (31), we may calculate the ratio of the geometric factor for 'H to that for 2D to be 11.5, far larger than the measured value of 0.9 to 1.1. The corresponding ratio for I7Oto 2Dmay be calculated to be 0.7, as opposed to the single measured value of 3.3. It should be emphasized, however, that the significance is vastly different for these two comparisons. In the case of 'H and *D,the experimental data have been obtained in several different experimental systems. Similarly, the theoretical calculation is based on reasonably exact estimations of the asymmetry factor for 2Dand for the angles involved. Therefore, the discrepancy between the measured and expected ratios is clearly significant, excluding the simple model considered. On the other hand, the significance of the comparison of the experimental and theoretical ratios for I7O and 2Dis highly doubtful. (i) The experimental results were obtained as a secondary aspect of a single study (Fujiwaraet al., 1974).(ii) The theoretical ratio is strongly dependent upon the precise value of the H-0-H angle; small but highly important changes in that angle could markedly change the results. As pointed out by Woessner and Snowden (1969b), the discrepancy between the measured and theoretical estimates of the geometric factor for 'H to that for 2Ddisappears if we assume that the molecules of water are preferentially oriented with their planes parallel to the substrate surface. However, the physical basis and significance for this orientation is unclear. Other, more complicated models may also be developed on an ad hoc basis. We should emphasize that the above discussion is based on certain simplifying assumptions, which currently appear to be reasonable. (i) We presume, as is done generally, that a single correlation time char-
27
NMR STUDIES OF WATER AND IONS WITHIN CELLS
acterizes all three nuclides of water in bulk aqueous solution. (ii) We have assumed axial symmetry for the molecular movement in the ordered systems. (iii) We have presumed that the quadrupolar and dipolar interaction constants are similar within bulk water and within these model systems. In summary, given the paucity of the data and the simplifying assumptions involved, the problem cannot yet be resolved. Some preferred orientation of the water molecules in these ordered systems must be present. The precise nature of this orientation remains unclear, although the underlying mechanism appears common to several very different model systems. It should be emphasized that, within these ordered systems, line splitting is observed only when the water content is low; usually the water concentration is ~ 1 0 0 %(w/w). Furthermore, the molecular ordering factors may be calculated to be very small, indeed. Both of these facts indicate that the interaction between the water and substrate molecules is weak. This concept is further strengthened by the studies of Hoeve and Kakivaya (1976) of the heat capacity of water-collagen preparations, who have concluded that the water molecules must be in the liquid state, even when the water concentration is as low as 1% (w/w). B. Protein Solutions
The rates of nuclear magnetic relaxation of water protons are enhanced by the presence of macromolecules. For protein concentrations of less than 100 mg/ml, the rates of both longitudinal and transverse relaxation are both linearly dependent on the concentration of macromolecule (Daszkiewicz et al., 1963). This observation is based on an interaction of water molecules with the protein surface and suggests the presence of a rapid exchange between surface and bulk fractions of water. It is clear, however, that the water associated with proteins in aqueous solutions is not rigidly immobilized on the macromolecular surface. Koide and Carstensen (1976) have studied the ultrahighfrequency absorption spectrum of water hydrated to Sephadex. They have pointed out that, although the immobilized water tumbles at frequencies lower than that of bulk water, the dispersion occurs at a frequency in excess of logHz (6.3 x lo9rad sec-l). Thus, even water of hydration on a macromolecular surface possesses a high degree of freedom of molecular movement. Kuntz and his colleagues have also presented evidence that the e
28
MORDECHAI SHPORER AND MORTIMER M. CIVAN
water of hydration of polypeptides and proteins in solution is far from “icelike.” Specifically, the cw spectral line of some of the water protons in such solutions remains relatively narrow, even when the temperature is reduced to a value as low as -60°C (Kuntz et al., 1969, Kuntz, 1971a,b). Defining the water of hydration as that water which remains unfrozen at low temperatures, proteins contain 0.3-0.5gm water/gm protein; nucleic acids contain three to five times as much water, using the same criterion (Kuntz et al., 1969). These estimates are consistent with the concept of a single layer of water of hydration covering the surface of the protein. For lysozyme, at least, X-ray data directly indicate a monolayer of water in well-defined positions (A. Yonath, private communication, cited by Koenig et al., 1975; Blake et al., 1967). These studies have suggested that the layer of water molecules constituting the interface between the bulk water molecules, on the one hand, and the protein substrate on the other, has special properties. The increased fluidity of this layer need not, however, reflect special properties of the macromolecular substrate, but may rather characterize the interfacial water layer bounding ice crystals, in general. This concept has been supported by NMR studies of finely divided ice above - 10°C.A narrow spectral line of water was found to be superimposed on the broad spectral line of ice (Kvlividze et al., 1974); this suggests that a fluid layer bounds ice crystals, in general. Additional confirmation of this concept has been provided by application of other physical techniques to the study of ice (Mazzega et al., 1976). One of the most striking characteristics of water nuclides in protein solutions is the phenomenon of relaxation dispersion. From Eqs. (11)-(19), it is clear that the rates of nuclear relaxation for ‘H, 2D,and 1 7 0 should all depend upon the Larmor frequency (o,,). Under the usual conditions of motional narrowing in water (o,,~,<< l), this dependence is too small to be detected. However, a frequency dependence of the magnetic relaxation for water protons has been clearly demonstrated in diamagnetic protein solutions by Koenig and his colleagues (Koenig and Schillinger, 1969a).A similar nuclear magnetic relaxation dispersion phenomenon has been noted for the deuterium of water in such protein solutions (Koenigzt al., 1975); from measurements at 8.1 MHz (50.9 X lo6 rad sec-’) and 4.3 MHz (27.0 x lo6 rad * sec-‘), a similar dispersive effect has been inferred for the 1 7 0 of water (Koenig et al., 1975).An example of this relaxation dispersion is presented in Fig. 2. From a measurement of the Larmor frequency at the inflection point of such curves, it is possible to estimate the correlation time (7,)
-
29
NMR STUDIES OF WATER AND IONS WITHIN CELLS
I00
10 I
12
I
IK
IOK
I
I
25°C OD280'171 15wt% pH 7.7
10-
8-
-Y 6 -
-
1 -
'4-
-
2-
0
1--Protein free ' I
1 1 - 7
0.1
buffer
'
I
' I + '
1.0
'
.* 1-
I
lo
'
. I
26 L
0
FREQUENCY (MHz)
FIG.2. Nuclear magnetic relaxation dispersion data for apotransferrin (Koenig and Schillinger, 1969a).The solid curve has been entered from an empirical equation. It is clear from the data that T1-l is strongly dependent upon the frequency applied over the range from 0.1 to 6 MHz. (Reprinted with the permission ofthe authors and Thelournu1 of Biological Chemistry.)
for the water nuclides in each protein solution. The value of T, calculated in this way is directly dependent upon the size of the protein used. Studies of lysozyme (Koenig et al., 1975) of molecular weight (MW) 14,700, apotransferrin (Koenig and Schillinger, 1969a) (MW = 84,000), hemoglobin (Lindstrom and Koenig, 1974) (MW = 64,000), carbonic anhydrase (Fabry et aZ., 1970) (MW = 30,000), concanavalin A (Koenig et al., 1973) (MW = 54,000), bovine serum albumin (Kimmich and Noack, 1970, 1971) (MW = 69,000), ceruloplasmin (Koenig and Brown, 1973) (MW = 150,000), and hemocyanin (Koenig et d., 1975) (MW = 9 x lo6) have been published in extended form. A more exhaustive quantitative study of a wide range of proteins has been recently carried out by Hallenga and Koenig (1976). Even more strikingly, the value of the T, for the water nuclides, calculated from the relaxation dispersion, is approximately equal to the orientational relaxation time T~ (Koenig and Schillinger, 1969a), calculated for the protein solute using Stokes' law with a correction for anisotropy (Woessner, 1961a). Thus, the rate of the molecular tumbling of the water molecules appears dependent upon the thermal rotational motion of the solute protein molecules. In addition to this major frequency dispersion, another dispersion of much smaller mag-
30
MQRDECHAI SHPORER AND MORTIMER M. CIVAN
nitude is noted at high frequencies applied to water protons; e.g., such an effect accounts for the fact that the data points for frequencies above 10 MHz fall below the solid curve of Fig. 2. )( is a measure It should be pointed out that the product ( 1 / ~ ~1/TJw=,, of the strength of interaction of the relaxation mechanism. Recently, Hallenga and Koenig (1976) have pointed out that this product (normalized to concentration) is directly proportional to the molecular weight of a variety of proteins. This suggests that the relaxation mechanism reflects a physical interaction, rather than a specific chemical interaction. The simplest interpretation of the data would be that a fraction of water molecules immobilized to the protein rapidly exchanges with a larger fraction of free water molecules (Koenig and Schillinger, 1969a). This model appears unlikely for at least two reasons. First, the number of water molecules calculated to be firmly attached to the protein molecules is surprisingly small. The calculation is carried out by first estimating T~ from the inflection point of the relaxation dispersion, and obtaining a theoretical estimate of the dipolar interaction constant from the known value of the intramolecular proton-proton distance. Introducing these values into Eq. (11) permits calculation of the intramolecular contribution to the longitudinal relaxation of the protons ordered on the protein surface. The longitudinal relaxation in water at room temperature is 0.294 sec-'. Inserting these values into Eqs. (21)and (23), the mole fractions ofthe ordered and free water can be calculated. Of the water molecules constituting the first hydration layer, only a few percent would turn out to be actually immobilized. For example, in the case of lysozyme, only three water molecules are calculated to be immobilized on each protein molecule within the framework of this simple model (Koenig et al., 1975). Second, comparison of the relaxation times for 'H, 2D, and 1 7 0 for the water in different protein solutions imposes severe restrictions on the possible mean lifetime ( T M ) of a water molecule on the surface of the protein. For the immobilized water molecules to sense the molecular tumbling of the protein, TM must be both 2 T~ and 5 T l b , the relaxation time of the nuclide when on the protein. For solutions of lysozyme at 22°C. T~ = 4.2 x sec, while T l b = 7.8 x sec for the 1 7 0 of H2170in such solutions. Therefore, in this case (Koenig et al., 1975),
7.8 x
low72 T M
L
4.2 x
(32)
From similar considerations for the 2D of water in solutions of hemocyanin (Koenig et a1., 1975),
31
NMR STUDIES OF WATER AND IONS WITHIN CELLS
1.5 x
1.6 x
(33) It is presumed that the interactions of water with lysozyme and hemocyanin are similar. However, from Eqs. (32) and (33),the permitted ranges for the lifetimes of the immobilized water are highly restricted, and are also different for the two proteins. The likelihood that the mean lifetimes satisfy these requirements seems remote, casting doubt on the validity of the basic model. An alternative interpretation of the data is that of anisotropic tumbling of the solvent molecules, resulting in ordering of the water molecules with respect to the protein surface along some preferred principal axis of rotation. From data obtained with all three water nuclides in the same protein solution, and from considerations similar to those used in the preceding section of this chapter, certain models based on ordering can be excluded. In aqueous solution with isotropic tumbling, the rates of longitudinal relaxation for ‘H, 2D, and 1 7 0 are given by Eqs. ( l l ) , (13), and (19),respectively. Ordering of the water molecules with respect to the protein surface scales down the dipolar or quadrupolar interaction by the molecular ordering and geometrical factors; this results in an effective interaction which is modulated by the rate of tumbling of the protein. Making the same reasonable assumption discussed earlier of the similarity of the nuclear interaction constants in water and in the protein solutions, and assuming once again cylindrical symmetry for The subscripts the molecular movement of water, we obtain Eq. (34). “W” and “P” refer to pure water and to protein solutions, respectively. 2 T~ 2
+
[ 1+
7p ( 0 0 7 ~ ) ~1
[ 1+
7p (W07p)l
[ 1+
7p (007~)~
4Tp
]
-k 4(007p)~
(34)
+ 1 i- 47p ] 4(W07p)~
(35)
+ 1 -k 4Tp ] d(W07p)~
(36)
where the molecular and geometric factors were defined in the pre-
32
MORDECHAI SHPORER AND MORTIMER M. CIVAN
ceding section. It will be appreciated from Eqs. (34)-(36) that the rates of longitudinal relaxation for each nuclide are greater in protein solution than in pure water because of three factors. Two of these , identical for all factors, (fJ2 and the argument dependent upon T ~ are three nuclides. The third factor, the square of the geometric factor, is different for each nuclide. If the water molecules were polarized with respect to the protein surface, the most reasonable principal axis of rotation for the ordered water molecules would be along the bisector of the H-0-H angle perpendicular to that surface. Therefore, the angles between the principal axis of molecular orientation and the axes of interaction would be pH = 90" and pD = 54.6'. We take, once again, j 3 D = 0" and qD= 0.121. From these values and Eqs. (34-36), the ratio [( l/Tl)H/( 1/Tl)DlP/[(l/Tl)H/( 1/Tl)DlW
may be calculated to be very large (> 100). As discussed earlier, the ratio of the geometric factor for 1 7 0 to that for 2D is subject to a very large uncertainty, so that the ratio [( l/Tl)O/(
l/Tl)D]P/[( l/Tl)O/( 1/Tl)DlW
cannot be specified from the data at hand. Actually, simultaneous determinations of the relaxation rates of the three nuclides in aqueous solutions of lysozyme at room temperature have demonstrated (Koenig et al., 1975) that the first ratio is close to 2 and the second ratio is close to 1, excluding this model from further consideration. It should be appreciated that, even if the value of qDis varied over a reasonable range, the calculated [( l / T 1 ) H / ( l / T 1 ) D I P / [ ( l / T ~ ) H / ( l / T 1 ) D I W
ratio based on such a polarization model remains much larger than 2. As discussed in the previous section dealing with ordered systems, other less obvious models based on the concept of oriented water are possible. Specifically, the water molecules could be preferentially oriented with their molecular planes parallel to the protein surface. However, the physical basis for this and more complex orientations is unclear. In addition to the as yet unresolved specific orientation of the water molecules on the protein substrate, another datum has yet to be satisfactorily explained. Partial replacement of water protons with deuterons does not alter the longitudinal relaxation of the 'H to the extent expected from studies of water protons in pure water (Anderson and Arnold, 1956; Smith and Powles, 1965). As previously noted, the
NMR STUDIES OF WATER AND IONS WITHIN CELLS
33
spin-lattice relaxation of protons in pure water arises both from intramolecular dipolar interactions and from intermolecular dipolar interactions of 'H with other molecules of water. The magnetic dipole moment of 'D is considerably smaller than that of 'H. Therefore, partial substitution of deuterons for protons produces an approximately proportional reduction in the measured rate of longitudinal relaxation of 'H in pure water. The effect noted has been much less than expected in certain protein solutions (Kimmich and Noack, 1970, 1971; Hilton and Bryant, 1976). This might suggest either that a paramagnetic contaminant had been present in the protein solutions, or that an additional significant determinant of the proton relaxation was a specific interaction of the first layer of hydration with the macromolecular substrate. In both cases, substitution of 2D for 'H in the water molecules would not necessarily induce a measurable change in the TI of 'H. The studies of Hilton and Bryant (1976)using EDTA suggest that paramagnetic impurities are not likely to play an important role; these authors suggest, on the other hand, that the exchange of water protons with protein protons may provide an important contribution to the observed effect. We see that the fractional enhancement of the longitudinal rate in protein solutions is the same for 2D and 1 7 0 , but twice as great for 'H. From the studies of deuteron-proton substitutions, it appears that an additional intermolecular process contributes to the rate of proton relaxation. It seems likely then that the relaxation mechanisms operating in pure water are enhanced in protein solution to the same extent for all three nuclides. However, an additional relaxation mechanism now contributes to the measured rate of proton relaxation. To summarize this section thus far, no simple model appears to accommodate satisfactorily all of the data obtained with water nuclides in protein solution. It is difficult to reconcile all of the published results with a model solely of oriented water. On the other hand, a rapid exchange solely between fractions of immobilized and free water also appears inadequate. Possibly, elements of both mechanisms may play a role. If one to two monolayers of surface water are characterized by anisotropic tumbling, the molecular ordering parameter (fo) must be very small; as in the case of the ordered model systems, the strength of interaction with the substrate must be very small. Although most of the studies of water in protein solutions have been concerned with the nature of the association between the solvent and substrate molecules, the coefficient of diffusion of the water molecules on the protein surface has also been measured (James and
34
MORDECHAI SHPORER AND MORTIMER M. CIVAN
Gillen, 1972).These measurements have been carried out in the presence of magnetic field gradients, using a modification of the Carr-Purcell sequence of pulses mentioned in Section 111. Obtained in this way, the coefficient for self-diffusion in 10% albumin solution has been found to be 1.83 x cm*/sec, very little different from that in distilled water of 2.07 x lop5cm2/sec. These NMR data have been interpreted within the model of a rapid exchange between free and immobilized water, each characterized by a different self-diffusion coefficient and mole fraction. Within this framework, the results correspond to an immobilized fraction of 2.6 monolayers of water surrounding the protein (James, 1975). C. lntracellular Fluids
The simplest mammalian cell is the mature erythrocyte, containing neither nucleus nor mitochondria. As expected from the relative histological simplicity, the NMR behavior of the water nuclides within the red cell is similar to that in the protein solutions discussed. Lindstrom and Koenig (1974) have measured the rate of longitudinal relaxation for water protons as a function of Larmor frequency. A relaxation dispersion induced b y hemoglobin was noted for the water protons in packed erythrocyte pellets, in whole blood, and in aqueous solutions of hemoglobin. In all three cases, the magnitudes of the enhancement and inflection frequency were determined solely by the hemoglobin concentration. These data indicate that encapsulation of hemoglobin within the cell membrane has little effect on its molecular motion. The TI of intracellular I7O from H,"O has also been found, at an applied frequency of 8.1 MHz (50.9 X lo6 rad sec-I), to be four to five times shorter than in the supernatant, similar to the enhancement of proton relaxation by hemoglobin in erythrocytes and hemoglobin solutions (Shporer and Civan, 1975).All ofthese data are consistent with the view that hemoglobin modifies the molecular dynamics of water inside cells and in aqueous solutions in the same way. NMR techniques have also been applied to human red cells in order to determine the permeability of erythrocyte membranes to water. Based on measurements of tritiated water, the most rigorous estimate ~ ) adult red cells of the reciprocal mean lifetime for water ( 1 / ~ within has been 91 & S.D. 22 sec-' (Barton and Brown, 1964).This rate is far greater than that for the enhanced longitudinal relaxation of water protons within the cells of 1.8 sec-' measured at 157 x lo6 rad * sec-' (Fabry and Eisenstadt, 1975). Therefore, the measured relaxation rate e
NMR STUDIES OF WATER AND IONS WITHIN CELLS
35
of water protons in erythrocytes is a weighted average of the proton relaxation rates within the intra- and extracellular phases, and it is insensitive to the precise value of the rate constant for diffusion. When, however, Mn2+is added to the plasma medium (Conlon and Outhred, 1972; Fabry and Eisenstadt, 1975), the water protons of the two phases can be distinguished, and the influence of diffusion detected. The rate of diffusion of water across erythrocyte membranes can also be determined by NMR techniques without adding any foreign material, b y studying the 1 7 0 from H2170.The distinct advantage of using H2I7Ois that the rate of longitudinal relaxation of 1 7 0 in bulk water at room temperature is 145 sec-', three orders of magnitude greater than that for 'H. Thus, ( l / T J for 1 7 0 is comparable to ( 1 / ~ ~ ) . The longitudinal relaxation time (T,)of the 1 7 0 was determined separately in samples of red cell suspensions, packed cells, and supernatant (Shporer and Civan, 1975).At room temperature, the rate of longitudinal relaxation of the intracellular 1 7 0 , measured with the cell pellets, was simply exponential and proceeded at 600-700 sec-', four to five times faster than that of the supernatant. The longitudinal relaxation of 1 7 0 in the erythrocyte suspensions was, however, nonexponential, reflecting water exchange across the cell membranes, as well as relaxation processes inside and outside the cell. At room temperature, calculation of the rate constant for the water exchange was of limited precision, because 1/T, within the cell was an order of magni~ the . other hand, raising the temperature of tude larger than 1 / ~On the samples decreased the rates of longitudinal relaxation within the intra- and extracellular phases, while increasing the rate of water exchange across the erythrocyte membranes. For these reasons, the data provided far more satisfactory estimates of rWat 37"C, which was, in any event, the temperature of physiological interest. In addition to these studies of erythrocytes, considerable effort has been devoted to more complex tissues as well. Striated muscle has been particularly well studied. Since striated muscle cells exhibit a high degree of order, we might expect to observe line splittings of the water nuclides, similar to those described for ordered model systems. However, the fractional water content of biological cells is far greater than that of the oriented model systems discussed earlier. Presumably for this reason, line splittings of the cw spectra of the water nuclides have not generally been observed. Any line splittings characterizing a small fraction of surface water would be obscured by rapid exchange with the greater fraction of nonordered water. Two reports of line splittings in biological preparations have, however, appeared. First, Chapman and McLauchlan (1967) observed a splitting of the
36
MORDECHAI SHPORER AND MORTIMER M. CIVAN
'H spectra1 line from water protons of rabbit sciatic nerve; the splitting depended upon the angle between the steady magnetic field and sample axis. This splitting was not observed, however, in the NMR spectrum of another nerve preparation (Fritz and Swift, 1967). If the line splitting reported by Chapman and McLauchlan truly reflected ordering of the sample water, we would expect the line splitting to depend on the strength of the dipolar or quadrupolar interaction of the water nuclide, the molecular ordering factor, and the geometric factor characterizing the nuclide, and not to depend on the strength of the external magnetic field. On the other hand, the line splitting could arise from a bulk susceptibility effect (Shporer et al., 1974a); this phenomenon has actually formed the basis for measuring the magnetic susceptibility of certain materials (Douglas and Evetiells, 1963). In the latter case, the splitting should depend upon the magnetic susceptibility and geometry of the sample, the external magnetic field, and the gyromagnetic ratio of the nuclide studied. Of particular interest here is the fact that the ratio of the line splittings of 'H and 2D ( A w ) ~ / ( A w )has ~ been found to be about 0.3 for a wide variety of ordered model systems, while that ratio would be expected to be 6.5 for a bulk susceptibility effect, on the basis of the ratio of gyromagnetic ratios. In order to examine this problem further, Klein and Phelps (1969) studied the cw spectrum of the 2Dof water in rat phrenic nerve. They were unable to detect any splitting whatsoever in their preparation and suggested that the observation of Chapman and McLauchlan arose as an artifact of their sample geometry. They were, in fact, successful in simulating the bulk susceptibility effect by moistening sample strips of twine. More recently, Fung (1975) has reported evidence for line splitting, both of water protons and deuterons, within frog striated muscle. The magnitude of the splitting for 2D was considerably smaller than that for 'H, consistent with a susceptibility effect. Fung, however, chose to ascribe this finding to the different geometric factors that might As noted above, the ratio of the geometric characterize 'H and 2D. factors for these two nuclides has been found to be close to unity for a wide variety of ordered model systems. Therefore, it is far more likely that the splittings noted for water protons, both in nerve and in muscle, reflect bulk susceptibility effects. This matter could be fully resolved by examining the magnitude of the line splittings as a function of the intensity of the external magnetic field. NMR techniques have also been applied to tissues in an effort to measure the self-diffusion constant of water within the intracellular
NMR STUDIES OF WATER AND IONS WITHIN CELLS
37
fluids. These investigations have been based upon the same Carr-Purcell sequence discussed earlier for measuring the coefficient of self-diffusion for water in protein solutions. In striated muscle (Abecedarskaya et al., 1968; Hansen, 1971; Finch et al., 1971; Chang et al., 1972), liver (Abecedarskaya et al., 1968), and brain (Hansen, 1971), the coefficient is reduced by a factor of only 1.3-2.4, in comparison to that for pure water. These data can be accounted for by (i) restriction of diffusion by macromolecules and membrane surfaces, and (ii) interaction with macromolecular substrate surfaces. In general, the NMR characteristics of water within biological tissue are different from those characterizing pure bulk water. First, there appears to be a fraction of water that does not freeze, even at temperatures as low as - 60°C. From measurements of the heights of the free induction decay signals of water protons within frog striated muscle as a function of temperature, Belton et al. (1972) have suggested that this fraction may be as high as 20% of the total tissue water. This finding is similar to that reported by Kuntz and his colleagues (Kuntz et al., 1969; Kuntz, 1971a,b) for water in protein solutions. As discussed above, this phenomenon may reflect the properties of the surface layer of water bounding the ice crystals formed within the cell, rather than reflect any special properties of the macromolecular substrate surface (Kvlividze et al., 1974). In addition, there are several other striking characteristics of intracellular water: the rate of longitudinal relaxation (l/Tl) is enhanced, the rates of longitudinal and transverse (l/T2)relaxation are not equal, and (l/Tl) is a function of the Larmor frequency applied. Not only is the longitudinal relaxation enhanced for all three water nuclides, but also the relaxation behavior, at least for 1 7 0 , is considerably more complex in fresh striated muscle than in bulk water. Specifically, the spin-lattice relaxation process for I7O from intracellular H,I7O has been found to b e nonexponential, both at 20-22°C and at 2-3°C (Civan and Shporer, 1974). The data could be fit by two exponentials; this suggests the presence of two functionally different populations of approximately equal size. With deterioration of the muscle over periods of observation of 1 to 2 days, the longitudinal relaxation became simply exponential; this final rate of relaxation was the same as the initial rate of relaxation of the fresh samples. It will be appreciated that the initial slope of an ensemble of several superimposed exponential decays represent, on a semilog scale, a weighted average of the individual slopes of the several components. Under conditions of rapid exchange among the several populations, a single relaxation rate constituting the same weighted average is ex-
38
MORDECHAI SHPORER AND MORTIMER M. CIVAN
pected. Thus, the loss of nonexponential behavior and the constancy of the initial slope during the course of deterioration of the muscle sample suggest that necrosis leads to mixing of the different compartments of water within the tissue, yielding a single final population of water, in terms of NMR parameters. The fact that it was possible to detect this process of exchange between the different populations of water with 1 7 0 arises from the far more rapid rate of relaxation of 1 7 0 than of the other two water nuclides, markedly improving the time resolution. In all subsequent discussions of the rate of longitudinal relaxation of intracellular 170, we shall be referring to the initial (average) rate of relaxation. A similar nonexponential process of longitudinal relaxation has been noted for normal and malignant rat lymphocytes (Shporer et al., 1976). This preparation is of particular interest because it provides an opportunity indirectly to study the NMR properties of the nucleoplasm. There are considerable technical difficulties in isolating nuclei in exactly the same state as within the normal cell. However, unlike most biological cells, the volume of the small lymphocyte is largely occupied by the nucleus. Therefore, in this cell, the contribution of intranuclear water would be expected to be considerably more prominent than, e.g., in the muscle cell. As in the case of muscle, the longitudinal relaxation of the fresh viable cells was nonexponential, but became simply exponential with eventual necrosis of the cells. Likewise, the rate of spin-lattice relaxation could be well fitted by a sum of two exponentials. The average mole fraction of the molecules subject to the slower rate of relaxation was approximately two-thirds of the total water. The value of T, characterizing the single exponential decay of the nonviable cells was not appreciably different from that characterizing the initial average rate of relaxation of the fresh cells. In the case of frog striated muscle, it was not possible to determine the anatomic site of the functionally different populations of water. In the case of these lymphocytes, however, the simplest interpretation of the data is that some two-thirds of the cell water is located within the nucleus and is characterized by a slower rate of relaxation than the one-third of the cell water in the cytoplasm because of the different macromolecular compositions of the two subcellular compartments. One of the most striking characteristics of intracellular water nuclides is the dependence of their rates of longitudinal relaxation upon the Larmor frequency (oJ. In striated muscle, TI appears to be monotonically dependent upon wo over the range 2.4-60 MHz (15 X lo6to 377 x lo6rad * sec-l) (Belton et al., 1972; Outhred and George, 1973;
39
NMR STUDIES OF WATER AND IONS WITHIN CELLS
Held et al., 1973; Fung and McGaughy, 1974). In order to examine this phenomenon in greater detail, the longitudinal behavior of the three water nuclides was studied in the same samples of frog striated muscle by Civan and Shporer (1975) at two different frequencies, 8.1 and 4.3 MHz (50.9 and 27.0 x 106 rad sec-', respectively). Once again, given the different axis of interaction for each nuclide within the molecular frame, ordering might be expected to affect each of the three relaxation behaviors differently because of the different geometric factors. The results indicated that: (i) the longitudinal relaxation rates of all three have an identical frequency dependence over the range studied, ~ 7the 0 1 same in muscle water and pure water, (ii) the ratio [ ( T l ) ~ ~ ( T l )is is 2.1 times greater in pure water than it is while the ratio [(T,)1H/(T,)2D] in muscle water, and (iii) 30-49% substitution of 2D for 'H has very little effect on the spin-lattice relaxation of tissue water protons. These data are identical to those obtained with protein solutions (Koenig and Schillinger, 1969a,b; Koenig et al., 1975); the same considerations are relevant here. Specifically, the results indicate that the molecular dynamics of intracellular water are determined only b y the macromolecular composition and not directly by the physiological state of the cell. However, changes in the physiological state of the cell can certainly alter the distribution of macromolecules within the intracellular fluids, indirectly altering the observed NMR properties of the intracellular water. The rate of transverse relaxation of water protons has also been reported to be nonexponential in rat striated muscle (Hazlewood et al., 1974). The decay was thought to consist of contributions from three populations of water: (i) a bulk phase, consisting of 82% of the total tissue water; (ii) an extracellular phase, consisting of 10%of the total water; and (iii) an intracellular phase of immobilized water, consisting of 8% of the total tissue water. The rate of transverse relaxation ( 1/T2) of the protons within the bulk intracellular water was enhanced to an extent an order of magnitude greater than that for (l/Tl).The enhancement of ( 1/T2) characterizing the apparently immobilized fraction was even greater. It must be kept in mind that not all of the proton spectrum arises from water protons. Protons from the tissue proteins can also contribute to the observed signal, so that the measured (l/Tz) becomes greater. In addition, dipolar interactions between the 'H of water and proteins can result in nonexponential behavior of the transverse relaxation process. For these reasons, the precise significance of the most rapidly relaxing fraction, constituting 8% of the signal intensity of the tissue protons, must remain open to question.
-
40
MORDECHAI SHPORER AND MORTIMER M. CIVAN
The large magnitude of the enhancement of (l/TJ characterizing the water protons within the bulk phase of the intracellular water has been puzzling. From measurements of the relaxation dispersion of (l/Tl) for the water protons, it is possible to calculate the apparent value of T,, with the aid of Eq. (1 1).From this value, and from Eqs. (11) and (12), the anticipated value of (1/T2) can be calculated for any given Larmor frequency. These calculated values of (l/Tz) prove to be considerably lower than those measured directly. This indicates that an additional relaxation process characterized by a correlation time longer than that of the protein must be playing a role. The additional process contributing to the measured rate of transverse relaxation could reflect local magnetic field gradients, arising from heterogeneity of the magnetic susceptibility within the sample. The local field gradient generated by inhomogeneities in magnetic susceptibility should be proportional to the strength of the steady magnetic field, In this case, and if diffusion across the local field graActudient plays a role, T2should be inversely proportional to ally, Hazlewood et al. (1974) have found that T , of water protons within rat striated muscle is essentially the same at 50 MHz (314 x los rad sec-l) and at 25 MHz (157 x lo6rad * sec-l), so that heterogeneity of sample susceptibility does not appear to play a dominant role. In addition to the studies of (l/T J and (1/T2)of water nuclides in biological cells, several recent reports have been concerned with measurements of ( l/TIP),a parameter equivalent to ( l/Tl) at low Larmor frequencies. From the data obtained with mouse muscle and spleen (Thompson et ul., 1973; Knispel et al., 1974), mouse adenocarcinoma (Knispel et al., 1974), and frog muscle (Finch and Homer, 1974), it is clear that intracellular water protons undergo relaxation dispersions at two distinct critical frequencies. In addition to the inflection point noted at a Larmor frequency of several megahertz discussed above, a second inflection point is observed at some tens of kilohertz. Although the possible basis for this second relaxation dispersion has been discussed in the literature (Raaphorst et al., 1975; Diegel and Pintar, 1975), the mechanism involved is not yet clear. From the studies cited dealing with water nuclides in cells and in protein solutions, it seems likely that the relaxation dispersion appearing at a high Larmor frequency arises from interaction of water with macromolecules within the cell. If so, it may be that the second relaxation dispersion, appearing at a low Larmor frequency, arises &om the fact that the intracellular proteins are grouped into ordered domains. Diffusion of water between these domains could introduce an additional longer correlation time, reflecting the mean lifetime of
41
NMR STUDIES OF WATER AND IONS WITHIN CELLS
the water within each domain, and modifying the relaxation behavior of the water nuclides. The net effect would then be an enhancement, both of (1/T2) at all Larmor frequencies and of (l/Tl) at low frequencies. It should be pointed out that the NMR properties of the water nuclides within a variety of cancer cells have been intensively studied. Over the past 4 or 5 years, it has become increasingly clear that both (l/Tl) and (l/T2)for the water protons in these cells are significantly reduced (Damadian, 1971; Weisman et al., 1972; Floyd et al., 1975). The relaxation rates for the I7O from H,I7O in malignant rat thymocytes have also been found to be reduced (Shporer et al., 1976). This prolongation of the relaxation rates of water nuclides in tumors could reflect at least two phenomena. The effect could arise solely from the increased water content/dry weight that is present in the more rapidly growing cancer cells. In addition, the effect could reflect, in part, changes in the size, shape, distribution, and state of aggregation of macromolecules within the neoplastic tissue. Recent studies of the two relaxation dispersions characterizing intracellular water protons suggest that the latter possibility is probably correct. Knispel et al. (1974) have examined water protons within mouse adenocarcinoma cells. At Larmor frequencies of 17-45 MHz (107 to 283 x 106 rad . sec-'), (l/Tl) was larger for normal muscle samples than for tumor-infiltrated samples; from measurements of T,, over the range of 103-105Hz (6 x lo3to 6 x lo5rad sec-l), (l/TJ was smaller for normal than for tumor-infiltrated muscle. In this case, a simple increase in the cell water content could not account for the change produced in the relaxation dispersions. If confirmed, this datum suggests that the altered rates of nuclear relaxation reflect changes in the macromolecular composition or state of aggregation within the intracellular fluids. For example, if the average molecular weight of the intracellular macromolecules were to increase, either by addition of new large species or by aggregation of smaller units, the inflection point characterizing the relaxation dispersion at high magnetic field would be shifted to a lower frequency. Under these circumstances, the rate of longitudinal relaxation would be reduced at frequencies greater than the inflection point and enhanced at frequencies less than the inflection point.
-
V.
NMR STUDIES OF ALKALI CATIONS
In this section we examine the NMR properties of alkali cations in systems of biological interest, primarily within the intracellular fluids
42
MORDECHAI SHPORER AND MORTIMER M. ClVAN
of biological cells, but also in certain model systems. Most of the published data deal with 23Na,the common isotope of sodium in nature. Very few papers have presented studies of 39K, the common isotope of potassium. Unfortunately, the relative sensitivity of 39Kat constant field is 2 orders of magnitude smaller than that for 23Na.Thus, despite its relatively high concentration, intracellular K+ presents a signal which is at least an order of magnitude smaller than that due to intracellular Na+. The formidable technical problem of achieving a satisfactory signal-to-noise ratio can be largely resolved by the use of superconducting magnets; this provides fields three to four times greater than those that have been routinely used thus far. Both Na+ and K+ have a spin number of 3/2, and the relaxation rates of both nuclides are determined primarily by nuclear quadrupolar interactions. Thus, although we shall be primarily discussing Na+, the same considerations hold for K+ as well. As noted in Section 11, each of the relaxation times for a homogeneous population of Na+ consist of one or two components [Eqs. (15)-( IS)], depending on the values of a,,and T ~ In . addition, a given sample may consist of more than one population of Na+ nuclei, each population comprising a different mole fraction ( P i ) of Na+. If the rates of nuclear relaxation are observed to depend on wo, at least one of the populations must be immobilized to a considerable extent. In this case, it is possible to estimate the T~ of the immobilized fraction (Shporer and Civan, 1974). If the immobilized fraction is in rapid exchange with a free fraction of Na+, we can also calculate the values of Pi, at least indirectly. Using Eqs. (21)-(23) in addition to Eqs. (15)-(18), it is possible to estimate the product (Pi)(e2qQ)2. Thus, in order also to calculate P,,we must obtain an independent estimate of the quadrupole coupling coefficient. For the purposes of simplicity and convenience, we denote the quadrupole coupling constant by the expression e2qQ here and in the remainder of the review. However, the constant also should be understood to include the factor 11 + (qz/3)l1’*,which, as noted previously, may vary from 1.0 to 1.3, depending upon the asymmetry of the electric field gradient applied to the nucleus. In principle, it should be possible to measure (e2qQ)in the solid state, using the technique of nuclear quadrupole resonance (NQR). The characteristic NQR absorption lines of Naf are found at low frequencies, where sensitivity to Na+ is poor, rendering conventional application of NQR impractical. However, the recent development of double resonance techniques has greatly increased the sensitivity of NQR, raising the possibility of its future use in measuring the quadrupole coupling constant (Edmonds et al., 1974).
43
NMR STUDIES OF WATER AND IONS WITHIN CELLS
Without direct measurements of the value for (e24Q)characterizing intracellular Na+, it is necessary to estimate the quadrupole coupling coefficient from theoretical considerations. The quadrupole coupling constant is directly dependent upon the symmetry of the electric field applied to the Na+ nucleus. For example, within a crystal of NaCl with perfect cubic symmetry, e24Q = 0. In aqueous solution, the most symmetrical local electric field possible should be provided by the sphere of water molecules surrounding the Na+ nucleus. In that case, the nuclear quadrupole constant characterizing Na+ arises from randomly fluctuating distortions in the sphere of hydration caused by the finite rate of movement of the water molecules. We wish to estimate an effective average quadrupole coupling coefficient for the hydrated sodium ion in order to obtain a lower bound for that characterizing intracellular Na+. The correlation time for the reorientation of water molecules in bulk water is 3 x sec (Deverell, 1969; Hertz, 1973).Generally, rs characterizing the mean lifetime of water within the hydration spheres of the alkali cations is not very different from this value (Deverell, 1969; Hertz, 1973). This lifetime characterizes the fluctuations of the field gradient applied to the alkali ion. The translational motion of the water molecules near Li+ has been characterized b y a rs of sec (O'Reilly and Peterson, 1969); the correlation time for 7x water in the vicinity of the other alkali cations falls monotonically with increasing size of the nonhydrated nucleus (Samoilov, 1972). Therefore, the correlation time for the movement of water molecules surrounding Na+ lies within the range of 3-7 x lO-"sec. From Eq. (16) in the limit of motional narrowing, and from the value of (l/Tl) = 18 sec-' for Na+ in dilute aqueous solution (Eisenstadt and Friedman, 1967), the value of quadrupole coupling coefficient characterizing hydrated Na+ should be 4.8-7.6 x lo6 rad sec-'. This estimate agrees with the value of 4.8 x lo6 rad * sec-' used by Chen and Reeves (1972). In an immobilized state, a less symmetric arrangement of ligands would be expected than in dilute aqueous solution; therefore a larger value for the coefficient should prevail. The above model for estimating the quadrupolar interaction may not b e entirely appropriate. The rapid entry and exit of water molecules from the ionic sphere of hydration might, in fact, produce fluctuations in the magnitude, as well as direction, of the electric field gradient. The possibility of this phenomenon has been dealt with in a model developed by Sutter and Harmon (1975) for the quadrupolar relaxation of Lit in aqueous solution. From their equations [which are of a general form similar to Eqs. (11)-(19)]and the known values for the coefficient of self-diffusion of water and the T , of Na+ in dilute
-
44
MORDECHAI SHPORER AND MORTIMER M. CIVAN
aqueous solution, we arrive at an average value of the quadrupole coupling coefficient which is similar to the one above. We conclude, therefore, that 4.8 x lo6 rad * sec-' is a conservative lower limit for (e%Q). A. Model Systems
Jardetsky and Wertz (1956, 1960; Wertz and Jardetsky, 1956) were the first to study the effect of complexation on the cw signal of 23Na. When they used small complexing molecules, the line width was broadened (equivalent to an increase in the value of 1/T, for Na),; because of the rapid exchange between complexed and free Na+, the observed value of 1/T, reflected an average of the two Na+ populations. However, in the presence of an ion-exchanging resin, a reduction in the size, rather than broadening, of the observed signal was noted. Under these latter conditions, exchange between the free fraction of Na+ and that fraction affected b y the resin was slow; this results in a simple superposition of the signals contributed by the two populations. Because of the enhanced transverse relaxation, the linewidth of the signal reflecting the Na+ interacting with resin was so broadened as to be undetectable under their experimental conditions. In this specific circumstance, the relative integrated intensity of the visible signal was directly proportional to the concentration of free Na+ in solution. Unfortunately, as discussed below, uncritical extension of these results to other biological systems has led to certain simplistic, and probably incorrect interpretations. These early investigations were extended to studies of other artificial systems. Haynes et a2. (1971) reported line broadening of the cw signal in the presence of each of a group of low molecular weight ionophores: monactin, enniatin B, valinomycin, monensin, cyclohexyl ether, and nigericin. More detailed study of the interactions between Na+ and a cyclic polyether (Shchori et al., 1971, 1973), cryptate (Ceraso and Dye, 1973), and valinomycin (Shporer et al., 1974b), and between K+ and a cyclic polyether (Shporer and Luz, 1975) have permitted characterization of the relaxation times and mean lifetime of the complexed and free populations of the alkali cations. The study of cryptate differed from the others in being based on measurements of the chemical shift of the complexed Na+; the magnitude of this shift is large with respect to the linewidth of the cw signal. In the other studies of ionophores, the relaxation times of the complexed species were faster, and determined by longer correlation times, than was the
NMR STUDIES OF WATER AND IONS WITHIN CELLS
45
case for free Na+. It should be pointed out, however, that because of the relatively small molecular size of the ionophores, the correlation times were still small enough for the condition of motional narrowing to hold ( W ~ T ,<< 1). Investigation of these model systems has been particularly instructive in facilitating analysis of each of the three ranges of exchange rates between free and complexed populations of cations. At high temperatures, exchange was sufficiently fast to result in a single average observed value for 1/T, and for l / T z . At very low temperatures, exchange was sufficiently slowed to result in the superposition of a narrow on a wide signal in the cw mode, or the appearance of two distinct components of the relaxation rates reflecting each of the populations in pulsed experiments. In practice, only the narrow line in cw experiments, or the slower component in pulsed experiments, is detected. In the intermediate range of temperatures, the specific value of the mean lifetime for exchange influences the experimental results, permitting calculation of the rate of exchange between the two populations. These studies have served then not only to provide detailed information on the kinetics of complexation of Na+ and K+ in these model systems, but also to provide a general approach for examining the more complex properties of intracellular Na+ in biological tissue. James and Noggle (1969a,b, 1972a,b) have studied the binding constants characterizing the interaction between 23Na and a series of small molecules by the use of NMR techniques. I n the event of rapid exchange between the complexed and free fractions of Na+, Eqs. (21)-(23) hold. In their systems, fast exchange prevails, so that the observed relaxation times represent weighted averages of the values characterizing the two populations. The binding or dissociation constants for the complexation, as well as the relaxation rates of the complex, have been obtained by measuring the relaxation times of the Na+ while varying the total Na+ and ligand concentrations. This approach can be extended to measure the values of the same parameters for Na+ complexed to polyelectrolyte macromolecules, a model system more closely analogous to the intracellular fluids. Thus, the longitudinal relaxation rates for Na+ apparently bound with soluble RNA, DNA, and NA+,K+-activatedadenosinetriphosphatase have been found to be 222 (James and Noggle, 1969a), 170 (Reuben et al., 1975), and 200 (Ostroy et al., 1974) sec-', respectively. The similarity of these values for 1/T, is remarkable. The enhancement of the longitudinal relaxation should depend upon the degree of immobilization determined either by T, or 7, at the binding site, and on the field gradient determined by the ligand organization. Despite the great dif-
46
MORDECHAI SHPORER AND MORTIMER M. ClVAN
ferences among these macromolecules, longitudinal relaxation was enhanced by almost precisely the same amount. The possible importance of this observation is further strengthened by the data of van der Klink et al. (1974) obtained with solutions of sodium polyacrylate. These investigators analyzed their results in a different way. However, their Fig. 4 gives ( l/Tl) for Naf in the presence of polyacrylic acid, as a function of Na+ added at a fixed degree of neutralization. From this figure, l/Tl for the apparently complexed Na+ can be calculated by the foregoing method to be approximately 200 sec-’. In an earlier study from the same laboratory, Kielman and Leyte (1973) also examined the longitudinal relaxation of 23Nain polyphosphate solutions. l/Tl was directly and strongly dependent on the degree of polymerization of the sodium polyphosphate (NaPO,), until n was approximately 60. Increase in n from 60 to 338 produced no further change in longitudinal relaxation. It is of considerable interest that in dilute solution, (NaPOJ,, was characterized by a longitudinal relaxation rate of some 160 sec-’. Comparing their Figs. 1 and 2, it is clear that polyphosphate macromolecules (n > 60) would also be characterized by l/Tl = 200 sec-’ in dilute solution. In more concentrated solution, the longitudinal relaxation is further enhanced, possibly because of an effect on the viscosity of the water. From their calculations of the longitudinal relaxation times of the Na+ soluble RNA “complex,” and from an estimate of the correlation time from the Debye-Stokes relationship, James and Noggle (1969a) calculated a value for the quadrupole coupling constant, equivalent to e2qQ = 3.1 X lo5 rad sec-’. This estimate, which is more than an order of magnitude smaller than the lower bound for e29Q calculated above, raises the question of the physical significance of the apparent dissociation constant measured by those authors. Using the same approach, Reuben et al. (1975) obtained remarkably similar results for Na+-DNA. As they pointed out, if their data were calculated by the technique of James and Noggle, e2qQ would be estimated to be even lower than the above value because of the greater size, and, therefore, longer correlation time, of the complex. From so prolonged a T, for the “bound” Na+, we would expect a strong dependence of 1/T,on ow A small but significant effect of frequency was found. On the assumption that all of the Na+ associated with the DNA is immobilized and that the entire frequency dependence arises from this immobilized Na+, T, was calculated to be 5.5 nsec; this value is three orders of magnitude greater than that of free Nat in water, but some five orders of magnitude shorter than that of DNA alone. The longitudinal relaxation rate (l/Tlb) for the complexed Na+ was 170
NMR STUDIES OF WATER AND IONS WITHIN CELLS
47
sec-I, ten times greater than that for free Na+. From these experimental values of 7, and (l/Tlb), e'4Q may be calculated from Eq. (15) to be 6.4 x lo5 rad * sec-', an order of magnitude smaller than our conservative estimate calculated earlier. This large discrepancy, as well as the above considerations, strongly indicate that the basic formulation cannot be correct, and that presumably only some fraction of the total Na+ affected by the presence of the DNA can be physically immobilized. An alternative formulation that accommodates the data far more satisfactorily is as follows. Polyelectrolyte solutions may be considered to consist of uniformly charged particles at the center of unit elements (Katchalsky, 1971). Between the macromolecules, Na+ may be considered free and characterized by the same properties as in water free of the polyelectrolytes. However, for macromolecules with sufficiently high charge densities, some of the Na+ will condense on the polyelectrolyte surface (Manning, 1969; MacGillivray, 1972). Although the sodium ions in the condensed phase will be constrained to travel along some equipotential surface surrounding the macromolecule, the great bulk of these ions cannot be considered immobilized. Under certain circumstances, the ionic mobility of the counterion can be greater in the condensed phase than in the bulk water phase (Katchalsky, 1971). Van der Klink et al. (1974) have calculated the expected rate of longitudinal relaxation for Na+ under such circumstances. Since (l/Tl) for free Na+ in bulk water is insensitive to frequency, we may ascribe all of the change in the observed value of (l/TJ for Na+-DNA to Na+ in the condensed phase. The precise basis for the small frequency dependence noted by Reuben et al. (1975) is unclear. One possibility is that a fraction of the condensed Na+ was immobilized and in rapid exchange with the remaining condensed Na+. If so, an upper limit for the fraction immobilized (Pb) can be calculated in the following way. As discussed above, our conservative estimate ofe2qQ for the immobilized Na+ is 4.8 X lo6 rad sec-'. Taking the correlation time of 5.5 x sec, obtained from the slight frequency dependence measured b y Reuben et d . as characterizing the immobilized species of Na+, 1/T,b may be calculated from Eq. (15) to be at least 14,900 sec-'. On the assumption that the remainder of the Na+ behaves as does free Na+ in bulk water, and that a rapid exchange occurs between the two subpopulations within the condensed phase, Pb can be calculated from Eqs. (21) and (23) to be no more than 1%of the total condensed Na+.
48
MORDECHAI SHPORER A N D MORTIMER M. CIVAN
It should be emphasized that far less than 1% of the condensed Na+ could possibly be immobilized. Otherwise, the immobilization would constitute the dominant factor responsible for the enhanced relaxation of Na+ in DNA solution, and the close similarity for the values of l/T,b found for Na+ apparently bound to very different macromolecules would be entirely fortuitous. Rather, it seems likely that one single factor is determining the enhanced longitudinal relaxation of Na+ in each of these polyelectrolyte solutions. We may speculate that the electrostatic interaction with the polyelectrolyte modifies the field gradient on the nucleus and that the rate of diffusion of the hydrated Na+ on the surface of each macromolecule determines 7, for Na+ within the condensed phase; these two factors are not greatly different among these polyelectrolytes. The molecular basis for the factors determining 7, in the bulk and condensed phases may not be immediately apparent. It will be recalled that in bulk water, 7, for free Na+ is determined by the rate of movement of the water molecules within the sphere of hydration; here, it is the entry and exit of the water molecules that is primarily responsible for the fluctuating field gradient. Within the condensed phase, fluctuations in the positions of the neighboring water molecules are less important in fixing the direction and size of the field gradient. Because of the proximity of the charged sites of the polyelectrolyte, the direction of the field will be relatively constant until the Na+ nucleus itself begins to diffuse away from the site. In summary, NMR studies of Na+-ionophore systems have provided a working model for quantifying the extent and effect of immobilization in biological systems. These agents, which have been thought to complex Na+, cause an enhancement of the longitudinal relaxation time, which is attributed to an increase in the correlation time, and presumably to an increase in the field gradient, as well. These effects are quantitatively different for each ionophore. On the other hand, different polyelectrolytes, which have not been considered specific complexing agents for Na+, seem to enhance the longitudinal relaxation of Na+ to very similar extents. The mechanism of action appears not to be by immobilizing the ion, but by producing a condensed phase of ions free to move along the surface of the macromolecule. As will be seen from the following discussion, a similar mechanism of action appears to prevail in the intracellular fluids. B. lntracellular Fluids
Cope (1965, 1967)was the first to study the cw NMR spectrum of intracellular 23Na,measuring the relative integrated intensity of the visi-
49
NMR STUDIES OF WATER AND IONS WITHIN CELLS
ble spectral line obtained from samples of muscle, kidney, and brain. Subsequently, the samples were ashed, solvent added to restore the volumes to their initial values, and the relative integrated intensities redetermined. Cope reported that the relative integrated intensity of each sample was only some 30-40% of that following ashing. Drawing on the data of Jardetsky and Wertz obtained with artificial systems under very different conditions, Cope concluded that two populations of Na+ must be present in these tissues. He reasoned that 60-70% of the intracellular Na+ was immobilized, and believed that the cw line of this fraction was so broadened as to be indistinguishable from background noise. The remaining 30-40% of the total Na+ was considered free, and, therefore, characterized by a far sharper and more detectable signal. Cope’s hypothesis initially seemed to be supported by two lines of evidence. (i) Although all of the 23Nawithin human erythrocytes could be detected (Yeh et al., 1973), a number of investigators obtained results similar to Cope’s using a wide variety of more complex tissues (Table I). (ii) It was subsequently possible to identify two components of the transverse relaxation process, using two different approaches. TABLE I PERCENTAGEOF *%”
Tissue Frog muscle
Rat muscle Homogenized frog muscle Rabbit muscle actomyosin Frog skin Rat testicle Rabbit kidney Rat kidney Rabbit brain Rat brain Rabbit myelinated nerve Frog liver Rat liver Pseudomonas aeruginosa
NMR SIGNALUNDETECTED m SAMPLESOF BIOLOGICAL.TISSUES Percentage undetected
Reference
63-72 58-65 37 55-63 65-72 62 65-74 48 57-64 44 24 58-70 67 60-70 67 56 66 61 60
Cope (1967) Ling and Cope (1969) Martinez et nl. (1969) Czeisler et 01. (1970) Cope (1967) Cope (1970a) Cope (1967) Cope (1967) Rotunno et al. (1967) Reisin et al. (1970) Reisin et al. (1970) Cope (1967) Cope (1970a) Cope (1967) Cope (1970a) Cope (1970b) Martinez et nl. (1969) Monoi (1974) Magnuson and Magnuson (1973)
50
MORDECHAI SHPORER AND MORTIMER M. CIVAN
Czeisler et al. (1970) observed the effect on the cw line of increasing the field strength of the rf‘field.From the anomalous saturation behavior of the visible spectral line for 23Na,they concluded that a second broader line must have also been present. In addition, the presence of at least two components of the transverse relaxation process was more directly established by pulsed NMR measurements of the free induction decay (Cope, 1970a; Berendsen and Edzes, 1973; Shporer and Civan, 1974). On the basis of these observations, there could be no doubt that the cw spectrum of 23Na consisted of at least two distinct components. However, as first pointed out by Shporer and Civan (1972), the data did not necessarily reflect the existence of two distinct populations of intracellular Na+. In fact, partly on the basis of the 23Naspectrum of smectic liquid crystals of sodium linoleate in water, they suggested an alternative, more likely interpretation. These authors examined the sharp central cw signal of 23Nain a control solution of NaOH, and in an experimental sample containing an equimolar amount of Na+ in the same volume of linoleate. Glycerol was added to the control test tube in order to equalize the width of the spectral lines. The relative integrated intensity of the experimental sample was only 34-39% of that observed for the control (Fig. 3),a result similar to what had been obtained with biological tissues. However, Shporer and Civan examined the cw spectrum with the aid of low fi-equency audio modulation and phase-sensitive detection, a technique commonly applied to detect wide spectral lines. In this
n
200 cps c t
L
W
FIG.3. Absorption spectrum of ‘“a obtained from sodium linoleate in water (left) and from a reference sample of NaOH in water and glycerol (right). The integrated intensity of the visible experimental signal is 39% of that for the reference signal, although both samples contained the same concentration and quantity of 23Nanuclei (Shporer and Civan, 1972). (Reprinted with the permission of The Rockefeller University Press.)
51
NMR STUDIES OF WATER AND IONS WITHIN CELLS
-
I w
FIG. 4. Derivative of the 23Na signal obtained from sodium linoleate in water (Shporer and Civan, 1972). In addition to the central signal, satellite lines are easily distinguished. (Reprinted with the permission of The Rockefeller University Press.)
more sensitive mode of signal detection, instead of the absorbed rfenergy, the derivative of the absorbed energy with respect to the rf frequency ( w ) is plotted as a function of w . A satellite spectral line was found on either side of the central line (Fig. 4). It will be noted that the trace of Fig. 4 is in the form of a first derivative (with respect to w) of the spectrum of Fig. 1. Using the same reasoning applied by Cope and others to biological samples, one might have concluded that some 34-39% of the Na+ nuclei were free in the sodium linoleate sample, whereas immobilization of the remaining nuclei broadened the 23Nasignal beyond detection. However, the appearance of satellite lines was inexplicable within the framework of separate populations of free and immobilized nuclei. Rather, the data clearly reflected the presence of a first order nuclear quadrupolar interaction for 23Na, as outlined for randomly oriented crystals in Section 11. The separation between the two satellite lines was 7.0 x lo4 rad * sec-I. From this value, and with the aid of Eqs. (4) and ( 5 ) , the quadrupolar coupling constant ( e z q Q )may be calculated to be 14 x 1oQ rad * sec-'. Similar data have since been obtained with unoriented liquid crystals of lecithin-Na+-cholate in water (10.5 x lo4 rad . sec-'; Lindblom, 1971), oriented liquid crystals of sodium decyl sul-
52
MORDECHAI SHPORER AND MORTIMER M. CIVAN
-
fate-decyl alcohol-Na,S04 in water (7.9 x lo4rad sec-'; Chen and Reeves, 1972),and oriented DNA samples (1.3-31.4 x lo4rad * sec-'; Edzes et al., 1972). In each of these cases, the values estimated for (e29Q)have been similar. Since all of these values are more than an order of magnitude smaller than that characterizing Na+ in aqueous solution (see Section 11), the quadrupolar interaction cannot arise from immobilization of all the Na+ in these model systems. It is more likely that the quadrupolar effect underlying the data arises from an ordering parameter characterizing the liquid crystals (Lindblom et al., 1971; Buckingham and McLauchlan, 1967). This ordering effect may reflect either or both of two possible phenomena. First, rapid exchange may exist between a pool of free Na nuclei and a fraction of bound nuclei of indeterminate size, which may be very small; the central spectral line is then a weighted average of the 1/2 to - 1/2 transitions of all the free and bound Na ions. Similarly, the satellite signals are weighted averages of the remaining two permitted transitions of all the bound plus free 23Nanuclei. Second, an additional possible basis for the ordering effect would be that the Na nuclei are in anisotropic domains, producing ordering of the electric field gradients. Ordering in these model systems might arise from polarization of the sphere of Na+ on the surface of the liquid crystals. Some such general electrostatic mechanism is suggested by the similarity of the values estimated for e29Q from data obtained with a variety of model systems. It should be pointed out that the rates of nuclear relaxation of =Na may consist of two components even in the absence of ordering. From Eqs. (15) and (16), it may be appreciated that, on theoretical grounds, two distinct Lorentzian 23Na signals may arise in isotropic liquids ~ )1, and need not arise solely from quasicharacterized by ( 0 ~ 7 2 solids. Experimental evidence supporting this theoretical possibility has been provided by studies of 23Nain agarose (Andrasko, 1974). On the basis of the discussion in Section 11, a similar situation would prevail even if a fraction of the Na nuclei, characterized by w07, 2 1, were in rapid exchange with the bulk of the Na ions characterized by fast tumbling. From these considerations, it seemed clear that the two components of the spectrum obtained for intracellular 23Na within biological tissues (Table I) might well reflect different energy transitions of all the Na+, rather than distinct populations of Na+. In fact, the alternative interpretation appeared the more likely (Shporer and Civan, 1972). The visible 23Nasignal obtained from almost all of the tissues studied
NMR STUDIES OF WATER AND IONS WITHIN CELLS
53
had been close to 40% of that anticipated on the basis of the total Na+ content. It seemed remarkable that so wide a variety of tissues as muscle, kidney, frog skin, liver, brain, and nerve should all apparently immobilize the same fraction of Na (Table I), although the protein and phospholipid contents of the tissues are very different. If the relative integrated intensity of the central visible 23Naline were a direct measure of the concentration of free Na+, the constancy of the estimates of the apparent fractional binding would be inexplicable other than as a fortuitous coincidence. However, this constancy is only to be expected from a first-order quadrupolar interaction with the Na+ nucleus. This alternative hypothesis has received considerable support. Of particular importance was the study of intracellular 23Na in striated muscle by Berendsen and Edzes (1973).Although at least two components of the spin-spin relaxation time were observed, only a single spin-lattice relaxation time could be detected. This finding, which has been confirmed (Shporer and Civan, 1974), strongly suggests that two distinct populations of immobilized and free Na+ are not present in muscle. Otherwise, two populations with such very different values of T , would be expected to be characterized by two distinctly different values of TI as well. From these considerations, it appears unlikely that the central signal from intracellular 23Na observed with cw NMR arises solely from a minor fraction of free Na+. Rather, the observed signal reflects all of the intracellular Na+, which however, is subjected to a nuclear quadrupolar interaction. The precise basis for this electrical interaction has been unclear, but could reflect either: (i) rapid exchange between free and immobilized fractions of Na+, or (ii) some sort of ordering of macromolecules and/or water within the cell. If the quadrupolar interaction does reflect the presence of a fraction of immobilized Na+, it would be of considerable interest to quantify this fraction. As Berendsen and Edzes (1973) have pointed out, the relatively modest broadening of the central visible signal strongly suggests that relatively little intracellular Na+ is immobilized. However, if the field gradient is modulated by rotational tumbling, the line-width is not a single-valued function of T,, reaching a maximum value when w07, = 1 [Eq. (17)]. Therefore, a narrow central signal may b e observed when T, is very small or very large. If, however, T, is large, we would expect both 1/T, [Eqs. (15) and (16)l and 1/T2 [Eq. (17)] to be strongly dependent upon the Larmor frequency oo. This reasoning has provided a basis for experimentally estimating the fractional immobilization of intracellular Na+.
54
MORDECHAI SHPORER AND MORTIMER M. CIVAN
Shporer and Civan (1974)have examined the effect of w,, on both nuclear relaxation rates of 23Nawithin frog striated muscle. Changing the Larmor frequency from 7.85 MHz, equivalent to 49.3 x lo6 rad sec-l, (0,) to 15.7 MHz, equivalent to 98.6 x lo6 rad sec-I, (2wa)produced only a small change in ( 1/T2),of 21 sec-' and an even smaller change in ( l/Tl) of 4 sec-'. The absolute values of (l/T2)Iand (1/TJIIat 15.7 MHz were 73 and 350 sec-I, respectively. This observation supported the presumption of Berendsen and Edzes (1973) that the intracellular fraction of immobilized Na+ must be very small. The experimental results may be subjected, however, to a semiquantitative analysis in order to obtain an explicit upper bound for the fractional immobilization. The simplest possible model would be to assume the presence of a single homogeneous population of intracellular Na+, all of which is immobilized. With this assumption, the data of the preceding paragraph may be used with Eq. (17) in order to calculate a lower limit for T ~ Specifically, . Eq. (17) permits us to express the ratio of ( 1/T2)1at the two Larmor frequencies as a function of w,T,:
-
From the values of ( 1/T2)1at the two frequencies, and Eq. (37),T , may be estimated to be 3.33 x lo+ sec. With this estimate of T , and the known values of (1/T2)1,2ws and 2wa, Eq. (17) may be applied once again in order to obtain an estimated upper bound for e2qQ.Using this approach, it is found to be no more than 9.6 x lo5 rad * sec-', far smaller than our conservative estimate calculated for free Na+ in aqueous solution. Clearly, this model cannot accommodate the data; all of the intracellular Na+ cannot be immobilized. The second simplest model consists of a rapid exchange between immobilized and free species of Na+ within an isotropic medium, resulting in a single population of 23Nawith respect to NMR parameters. Specifically, one species (f) is considered free in solution within the cell, and is characterized by a very short correlation time Tf, by relaxation times (Tlf)and (TB),and by the mole fraction Pf; ooTf << 1. The second species (b), considered immobilized to macromolecules within the cell, is characterized by a much longer correlation time q,, the relaxation times ( T l b ) , (T2b)I, and (T2b)II, and by P b , the mole fraction of Na+ immobilized; WoTb b 1. Species b is considered entirely responsible for the observed frequency dependence of the relaxation times. Within the framework of this model, Pb may be calculated from the measured dependence of ( 1/T2)1on frequency, the measured absoand our conservative estimate of e2qQ. lute value of ( 1/T2)11,2wa,
NMR STUDIES OF WATER AND IONS WITHIN CELLS
55
Using this approach, P b may be calculated to be 0.36 X lo-' (Shporer and Civan, 1974). Even in the unlikely and extreme case that (e'qQ) is as low as 3.2 X lo6 rad sec-I, P b can be calculated to be 0.91 X Therefore, within the framework of the simplest possible interpretation of the data, and subject to the realistic estimate of ( e 2 4 Q ) ,less than 1% of the total intracellular Na+ appears to be immobilized. An even smaller fractional immobilization would be estimated, basing the calculation on the data obtained for (l/Tl)instead of (l/T*). Although we have attempted to quantify the maximal possible fractional immobilization of the intracellular Na+ by a detailed analysis of one specific model, a similar conclusion is reached by analyzing other more complicated models as well. The central concept involved in such analyses is the fact that immobilization of an ion is equivalent to a prolongation of the correlation time characterizing that ion. When the correlation time is large, ( l / T J should strongly depend upon the Larmor frequency. The fact that little or no frequency dependence is actually observed demands that the fractional binding of intracellular Na+ must be very small. Analysis may be carried out within a more sophisticated framework than the model chosen. For example, the correlation time characterizing immobilized Na+ may be modulated both by translational and rotational processes. In addition, the bound species may be characterized by a spectrum of correlation times, rather than b y a single value. However, choice of the specific model for analysis cannot substantially alter the basic conclusion. Therefore, if the nuclear quadrupolar interaction of intracellular 23Naarises from fractional immobilization of Na+,the fraction immobilized must b e very small, indeed. For example, a fraction of a few percent niight be immobilized, not contributing to the signal at all. On the other hand, the nuclear quadrupolar effect might arise from ordering, rather than from any immobilization of the Na+ (Shporer and Civan, 1972; Berendsen and Edzes, 1973). Berendsen and Edzes (1973) have examined this possibility in detail. Specifically, the electric field gradient imposed on the Na+ is likely to be aligned along one preferred axis. As already discussed, a condensed phase of counterions on the surface of charged macromolecules would modify the electric field gradient imposed on the Na+ nucleus. To the extent that such macromolecules are incorporated into membrane structures or other intracellular organelles, long-range ordering would result. In such an event, the electric field gradient imposed on the Na+ nuclei would not average out to zero with time, but would lead to an effective nuclear quadrupolar interaction. The Na+ within each ordered environment or domain will be characterized by
56
MORDECHAI SHPORER AND MORTIMER M. CIVAN
a residence time rMand a value of eZqQ(e2qQ)dom,which is smaller than that for Na+ in free solution. On the basis of the observation that two components of T , for intracellular Na+ may be distinguished, Berendsen and Edzes (1973) have calculated a lower limit for the extent of the long-range order associated with each domain. Exchange of Na+ between adjoining domains is considered to take place by diffusion. If diffusion were to occur very rapidly (o0rM << l), a single narrow cw line would result, reflecting the entire 23Na signal, contrary to experimental results. On the other hand, if diffusion were to be very slow [ o ~ T> ~ > (ezqQ)domrM >> 13, a powder pattern similar to that of Fig. 1 would be expected, giving rise to satellite signals with cw NMR; such satellite signals have not been observed, either because of signal-tonoise limitations or because they do not, in fact, exist. The conditions for the presence of two detectable components of T , in the true absence of satellite signals is given by (Baram et al., 1973). WO~M
1
( e b Q)domTM
(38)
Under these boundary conditions, rM2 lo-* sec. The rms displacement (d) of an ion in this time will be approximately given by
d = (60~M)~” (39) where D is the coefficient of diffusion for Na+. If the value of D is equal to that for Na+ in free solution (^I 10-5cm2* sec-l), d may be calculated to be no smaller than about 80 A. Thus, the lower limit for the extent of each domain can be roughly estimated to be of the order of 100 A. In summary, the NMR properties of 23Nawithin the intracellular fluids are different from those in aqueous solution. Although the results of initial studies were thought to reflect the presence of two distinct bound and free populations of Na+, the larger fraction being bound, subsequent studies have strongly suggested that this is not the case. Rather, the quadrupolar interaction with intracellular 23Naappears to cause two of the three permitted energy transitions to be characterized by shorter transverse relaxation times. This quadrupolar effect could arise either from rapid exchange between a small fraction (<1%) of immobilized Na+ in rapid exchange with the bulk free Na+, or from diffusion of Na+ between domains of ordered polyelectrolytes containing entirely free Na+. The interaction between Na+ and the polyelectrolytes is electrostatic and does not significantly limit the freedom of motion of the ion. Far fewer NMR studies have been devoted to an examination of in-
NMR STUDIES OF WATER AND IONS WITHIN CELLS
57
tracellular 3sK because of its less favorable relative sensitivity. Magnuson et al. (1973)have reported that the NMR relaxation parameters of both 23Naand 30Kaccumulated by etiolated pea stem are similar to those of the free ions within dilute solution. However, as they pointed out, most of the intracellular fluid of their preparation was contained within vacuoles, which may not be representative of the intracellular fluids within animal cells. More recently, Damadian and Cope (1973)and Cope and Damadian (1974) have found that the rate of longitudinal relaxation time was increased by a factor of 5 to 8 for 3sK within the intracellular fluids of rat muscle and brain. On the other hand, these authors claimed that the rate of transverse relaxation time for intracellular 3sK is some 200 times faster than that for 3sK in dilute aqueous solution. Their data were interpreted to indicate substantial fractional immobilization of intracellular K+. However, the apparently marked enhancement of ( 1/T2)could also have primarily reflected inhomogeneity of their applied magnetic field. Although Cope and Damadian (1974) considered the latter possibility, they rejected it in the belief that the free induction decay used to measure (1/T2) necessarily assumes a Gaussian shape when limited by field inhomogeneity. (Actually, field inhomogeneity may cause the FID to assume a variety of different shapes; even when field inhomogeneity plays a dominant role, the F I D can frequently be well fitted to a single exponential decay.) Since the shape of their free induction decays was gaussian for 3sK in aqueous solution, and exponential for 3sK in muscle, they concluded that field inhomogeneity played a role in their studies of solution, but not of muscle. In order to examine this problem in greater detail, Civan et al. (1976) have studied samples of 1 M KCl solution and intact frog striated muscle at a frequency of 10 MHz, and at 4-7°C and/or at 21-22°C. Field inhomogeneity was minimized by using small sample volumes and b y using a superconducting magnet designed specifically to provide highly homogeneous fields. In these experiments, magnetic field inhomogeneity was measured to contribute < 15%to the free induction decay observed for intracellular 30K. The signalto-noise ratio of the measurements was enhanced by means of extensive time averaging. The rates of nuclear relaxation for 30K in aqueous solution were 22 3 (mean f 95% confidence limits) sec-' at 4-7"C, and 15 f 2 sec-' at 21-22°C. For intracellular 39K, (1/T2) was measured to be 327 2 22 sec-' and 229 f 10 sec-* at the lower and higher temperatures, respectively. The corresponding values for ( l / T J in the same
*
58
MORDECHAI SHPORER AND MORTIMER M. CIVAN
*
muscle samples were 198 ? 31 sec-' and 79 15 sec-' at 4-7°C and at 21 to 22"C, respectively. Thus, in aqueous solution, the rates of nuclear relaxation of 39K and 23Naare similar. Within frog striated muscle, the relaxation rates for 39K are increased only two to three times more than for 23Na.The slightly greater enhancement for 39K is likely to arise from the greater number of electrons and therefore greater value of its Sternheimer antishielding factor (Deverell, 1969). Such a mechanism also appears responsible for the greater enhancements of the relaxation rates of 87Rbover those of 23Naby aqueous solutions of DNA (Reuben et al., 1975); 87Rbis characterized by a larger antishielding factor than either 23Naor 3gK (Deverell, 1969). The larger the alkali cation, the more effectively will its electronic cloud be polarized by a given electrostatic field, resulting in a larger electric field gradient. This observation that the relaxation rates for 39K are increased only two to three times more than for 23Na,and the observation that the ratio [(1/T1)/(1/T2)]is similar for the two nuclides suggest that a common mechanism determines the NMR properties of both intracellular ions.
VI.
CONCLUSIONS
It will be appreciated from the foregoing material that, as in all techniques, there are technical limitations and conceptual uncertainties. However, application of the techniques of NMR spectroscopy permit certain clear generalizations to be drawn concerning the nature and composition of the intracellular fluids. Studies of the water nuclides have demonstrated that, over the range of Larmor frequencies commonly applied, the basic mechanism determining the rates of longitudinal relaxation is exactly the same within biological cells and within model systems. Encapsulation within the plasma membrane confers no new properties on the water molecules. In each case, the longitudinal relaxation times of the water nuclides sense the tumbling time and concentration of the macromolecule in solution. In cells that are more complicated than erythrocytes, the rates of transverse relaxation are determined by the same mechanism that affects (l/TJ. In addition, however, ( 1/T2)appears to be affected by the diffusion of water molecules between intracellular domains of ordered macromolecules. The detailed dynamics of the interaction of water molecules with the surfaces of proteins in solution are not clear, either within simple aqueous solutions of the macromolecules or within the intracellular fluids. However, it should be emphasized that the large measurable
NMR STUDIES OF WATER AND IONS WITHIN CELLS
59
changes in the NMR parameters characterizing the water nuclides in macromolecular solutions seem to reflect very small perturbations in the physical state of the water molecules involved. As discussed earlier, if we try to accommodate the NMR data within the framework of rapidly exchanging fractions of immobilized and free water molecules, we calculate that only some 1%of the surface water coating the proteins need be immobilized. Similarly, if we describe the data within the framework of an ordered homogeneous phase of water, the ordering parameter calculated is extremely small. From these considerations, it seems unlikely that the large NMR changes noted for water in solutions of macromolecules have any direct physiological significance. On the other hand, NMR measurements of the water nuclides may prove to be of considerable indirect practical value. The water molecules can serve as a sensor of the physical state of the macromolecules within the intracellular fluids. For example, NMR studies of the relaxation processes of the water nuclides may permit characterization of the nature and concentrations of macromolecules within the cell under pathological conditions. This approach has already proved useful in studying sickle cells at high and low p 0 2 tensions (Lindstrom et al., 1976). It may also help define the malignant state of biological cells (Damadian, 1971; Weisman et al., 1972). In this respect, an NMR application of potentially great value is the technique of zeugmatography devised by Lauterbur (1973). This approach permits definition of the amounts of water and the NMR parameters characterizing the water in spatially different areas within biological tissue. The technique has undergone further development by Ernst and his colleagues (Kumar et al., 1975), but has not yet been widely applied to biological preparations. Of the major intracellular ions, Na+ has been most extensively studied. In both model systems and in the intracellular fluids, the rates of nuclear magnetic relaxation are enhanced. This enhancement appears to be nonspecific, currently best explained by the existence of a condensed phase of cations associated with the surface of charged polyelectrolyte macromolecules. This condensation does not immobilize Na+, whose freedom of motion remains very large. It should b e emphasized that the differences in the NMR properties of Na+ within the intracellular fluids and within water are not great, either in absolute terms or in terms of the frequency dependence of the rates of nuclear relaxation. Therefore, the fractional immobilization of intracellular Na+ appears to be very small. As in the case of the water nuclides, in cells more complicated than erythrocytes, the NMR properties of intracellular Na+ are affected by the presence of functional domains of organized charged macromole-
60
MORDECHAI SHPORER AND MORTIMER M. CIVAN
cules. The correlation time characterizing the mean lifetime within each domain appears short enough to permit the appearance of two exponential components of the transverse relaxation process. It should be possible to estimate the size of these domains from measurements of the T,, of intracellular Na+. If the concept of diffusion between functional domains is valid, these estimates should conform to estimates based on measurements of the T , , of the intracellular water nuclides, taking into account differences in diffusion coefficients. Relatively little information is available concerning intracellular 39K. However, from recent data obtained in our laboratory, it seems likely that the NMR properties of %K are very similar to those obtained for 2SNa. It is of interest to relate these findings to the measurements of the intracellular activities of Na+ and K+ obtained with ion-selective microelectrodes. Data obtained for a variety of tissues indicate that the electrochemical activity coefficient for K+ within the intracellular fluids is very close to that characterizing K+ in the external bathing medium (Lev and Armstrong, 1975). The activity coefficient for Na+ within the intracellular fluids is roughly half that found for Na+ in the external bath. A certain amount of caution must be exercised in interpreting this observation because of the problem of correcting for the extracellular Na+ trapped in the tissues subjected to chemical analysis; this problem is negligible for K + and is of considerable importance for Na+ because of the usually great differences in external concentration of the two ions. Indeed, some evidence has been presented suggesting that part of the difference noted between the apparent activity coefficients for Na+ and K+ may be related to this technical problem for some (Hinke, 1961) but not for all cells (Dick and McLaughlin, 1969). If, however, the observation is basically correct that the two activity coefficients are substantially different, a significant fraction of the intracellular Na+ could be either immobilized or compartmentalized. The NMR data discussed in this review strongly suggest that fractional immobilization of intracellular Na+ is very slight (< 1%). By exclusion, it seems likely that subcellular compartmentalization is playing a role. Recent studies of the salivary gland cells of Chironomus larvae using ion-selective microelectrodes indicate that K+ (Palmer and Civan, 1975), C1- (Palmer and Civan, 1976), and Na+ (Palmer and Civan, 1977) are distributed homogeneously throughout the necleus and cytoplasm. Presumably, subcellular compartmentalization takes place within organelles other than the nucleus. In summary, further advances in our detailed understanding of the
NMR STUDIES OF WATER AND IONS WITHIN CELLS
61
small perturbations in the molecular dynamics affecting the NMR measurements of water and ions within the intracellular fluids will depend upon continued studies of model systems. Nevertheless, it is already possible to use the NMR properties of intracellular nuclides to sense the nature and state of aggregation of the macromolecules within the intracellular fluids.
SYMBOLS AND ABBREVIATIONS b cw
D d ebQ
f
subscript referring to immobilized species continuous wave mode of NMR spectroscopy coefficient of d i h s i o n rms displacement of a diffuring species quadrupole coupling constant subscript referring to free species molecular ordering factor free induction decay magnitude of steady magnetic field magnitude of rf magnetic field Planck’s constant
h/2n
7W
Q
spin quantum number magnetic quantum number value of macroscopic magnetization vector at equilibrium components of the macroscopic magnetization vector molecular weight degree of polymerization relative mole fraction interproton distance longitudinal or spin-lattice relaxation time transverse or spin-spin relaxation time TI is the rotating frame principal components of the electric field gradient angle between V , , and the vector perpendicular both to V,, and to the principal axis of rotation gyromagnetic ratio asymmetry factor angle between the axis of interaction and H o magnitude of the nuclear magnetic moment angle between the principal axis of rotation and V,, correlation time mean lifetime of molecular complex correlation time for a protein in solution correlation time characterizing molecular tumbling of water within the hydration spheres of ions mean lifetime for water within cells angle between the principal axis of rotation and H o
62
MORDECHAI SHPORER AND MORTIMER M. ClVAN
aJ
angle between V,, and the vector perpendicular both to H o and to V,, angular displacement of the magnetization vector following an NMR pulse frequency of applied alternating magnetic field Larmor resonance frequency
JI 0 “0
ACKNOWLEDGMENTS Supported in part by a research grant from the National Institutes of Health (5 R01 AM 16586-03). Dr. Shporer received support from the University of Pennsylvania-Israel Exchange Program, and Dr. Civan was an Established Investigator of the American Heart Association during the preparation of this review. The authors wish to acknowledge a special debt of gratitude to Dr. Rene‘ Bloch, who died tragically on February 10,1976. Dr. Bloch first brought the authors together, stim‘dating a collaboration that has been maintained over the past six years. REFERENCES Abecedarskaya, L. A,, Miftahutdinova, F. G., and Fedotov, V. D. (1968).State of water in live tissues (results ofinvestigation by the N.M.R. spin echo method). Biophysics (USSR) 13,750-758. Abragam, A. (1961). “The Principles of Nuclear Magnetism.” Oxford Univ. Press (Clarendon), London and New York. Andersen, C. A., ed. (1973).“Microprobe Analysis.” Wiley, New York. Anderson, W. A., and Arnold, J. T. (1956).Proton relaxation times in H,O-D,O mixtures. Phys. Reu. [2] 101,511-512. Andrasko, J. (1974). Nonexponential relaxation of 23Na+in agarose gels. J . Magn. Reson. 16,502-504. Baram, A., Luz, Z., and Alexander, S. (1973). Resonance line shapes for semi-integer spins in liquids. J. Chem. Phys. 58,4558-4564. Barnes, R. G. (1974). Deuteron quadrupole coupling tensors in solids. Advan. Nucl. Quadrupole Reson. 1,335-355. Barton, T. C., and Brown, D. A. J. (1964).Water permeability ofthe fetal erythrocyte. J. Gen. Physiol. 47, 839-849. Belton, P. S., Jackson, R. R., and Packer, K. J. (1972). Pulsed NMR studies of water i n striated muscle: I. Transverse nuclear spin relaxation times and freezing effects. Biochim. Biophys. Acta 286, 16-25. Berendsen, H. J. C. (1962). Nuclear magnetic resonance study of collagen hydration.J . Chem. Phys. 36,3297-3305. Berendsen, H. J. C., and Edzes, H. T. (1973). The observation and general interpretation of sodium magnetic resonance in biological material. Ann. N.Y. Acad. Sci. 204, 459-480. Blake, C. C. F., Mair, G . A,, North, A. C. T., Philips, D. C., and Sarnia, V. R. (1967).On the conformation ofthe hen egg-white lysozyme molecule. Proc. Roy. Soc. London, Ser. B. 167,365-377. Blinc, R., Easwaran, K., PirS, J., Volfan, M., and Z u p a n G , I. (1970).Self-diffusion and molecular order in lyotropic liquid crystals. Phys. Rev. Lett. 25, 1327-1330. Buckingham, A. D., and McLauchlan, K. A. (1967). High resolution nuclear magnetic resonance in partially oriented molecules. Progr. N.M.R. Spectrosc. 2,63-109. Bull, T. E. (1972). Nuclear magnetic relaxation of spin 3/2 nuclei involved in chemical exchange. J . Magn. Reson. 8,344-353. Century, T. J., Fenichel, I. R., and Horowitz, S. B. (1970). The concentrations of water,
NMR STUDIES
OF WATER
AND IONS WITHIN CELLS
63
sodium and potassium in the nucleus and cytoplasm of amphibian oocytes. J . Cell Sci. 7, 5-13. Ceraso, J. M., and Dye, J. L. (1973). nuclear magnetic resonance study of exchange rates. Sodium cryptate in ethylenediarnine. J. Am. Chem. SOC. 95, 4432-4434. Chang, D. C., Hazlewood, C. F., Nichols, B. L., and Rorschach, H. E. (1972).Spin echo studies on cellular water. Nature (London) 235, 170-171. Chapman, G . E., and McLauchlan, K. A. (1967).Oriented water in the sciatic nerve of rabbit. Nature (London)215,391-392. Chapman, C . E., and McLauchlan, K. A. (1969). The hydration structure of collagen. Proc. Roy. SOC. London, Ser. B 173,223-234. Charvolin, J., and Rigny, P. (1969).Mesure du temps de relaxation nucle‘aire transversal des deuterons de I’eau lourde adsorbee sur une surface. C . R . Acad. Sci., Ser. B 269, 224-227. Chen, D. M., and Reeves, L. W. (1972). Sodium magnetic resonance in lyotropic nematic phases and the implications for observation in living systems.J. Am. Chem. Soc. 94,4384-4386. Chiba, T. (1963).Deuteron magnetic resonance study of barium chlorate monohydrate. J . Chem. Phys. 39,947-953. Civan, M. M., and Shporer, M. (1972). 1 7 0 NMR spectrum of H,”O in frog striated muscle. Biophys. J. 12,404-4 13. Civan, M. M., and Shporer, M. (1974). Pulsed NMR studies of I7O from H,I7O in frog striated muscle. Biochim. Biophys. Acta 343,399-408. Civan, M. M., and Shporer, M. (1975).Pulsed NMR study of I7O,,D, and ‘H of water in frog striated muscle. Biophys. J. 15, 299-306. Civan, M. M., McDonald, G . C . ,Pring, M., and Shporer, M. (1976).Pulsed nuclear magnetic resonance study of ”K in frog striated muscle. Biophys. J. 16, 1385-1398. Conlon, T., and Outhred, R. (1972). Water diffusion permeability of erythrocytes using an NMR technique. Riochitn. Biophys. Acta 288, 354-361. Cope, F. W. (1965). Nuclear magnetic resonance evidence for complexing of sodium ions in muscle. Proc. Nutl. Acud. Sci. U.S.A. 54,225-227. Cope, F. W. (1967).NMR evidence for complexing of Na+ in muscle, kidney and brain, and by actinomysin. The relation of cellular complexing of Na+ to water structure and to transport kinetics.J. Cen. Physiol. 50, 1353-1375. Cope, F. W. (1970a). Spin-echo nuclear magnetic resonance evidence for complexing of sodium ions in muscle, brain and kidney. Biophys. J. 10, 843-858. Cope, F. W. (1970b).Complexing of sodium ions in myelinated nerve by nuclear magnetic resonance. Physiol. Chem. Phys. 2,545-550. Cope, F. W., and Damadian, R. (1974). Biological ion exchanger resins: IV. Evidence for potassium association with fixed charges in muscle and brain by pulsed nuclear magnetic resonance of 58K. Physiol. Chem. Phys. 6,763-771. Czeisler, J. L., Fritz, 0. G . , Jr., and Swift, T. J. (1970). Direct evidence from nuclear magnetic resonance studies for bound sodium in frog skeletal muscle. Biophys. J . 10,260-268. Damadian, R. (1971).Tumor detection by NMR. Science 171, 1151-1153. Daskiewicz, 0. K., Hennel, J. W., Lubas, B., and Szczepkpwski, T. W. (1963). Proton magnetic relaxation and protein hydration. Nature (London)200, 1006-1007. Dehl, R. E. (1968). Broad line NMR study of H,O and D,O in oriented rayon fibers.J. Chem. Phys. 48,831-835. Dehl, R. E., and Hoeve, C. A. J. (1969).Broad line NMR study of H,O and D,O in collagen fibers.]. Chem. Phys. 50,3245-3251. Deverell, C. (1969).Nuclear magnetic resonance studies ofelectrolyte solutions. Progr. N.M.R. Spectrosc. 4, 235-334.
64
MORDECHAI SHPORER AND MORTIMER M. CIVAN
Dick, D. A. T., and McLaughlin, S. G. A. (1969.) The activities and concentrations of sodium and potassium in toad oocytes. J. Physiol. (London) 205,61-78. Dick, D. A. T., Fry, D. J., John, P. N., and Rogers, A. W. (1970). Autoradiographic demonstration of inhomogeneous distribution of sodium in single oocytes by Bufo bufo. J. Physiol. (London) 231, 19-29. Diegel, J. G., and Pintar, M. M. (1975). Origin ofthe nonexponentiality of the water proton spin relaxations in tissues. Biophys. J. 15,855-860. Douglas, D. C., and Evetiells, A. (1963).Volume magnetic susceptibility measurement by NMR.J. Chem. Phys. 39,3161-3162. Ducros, P. (1960). Study by dielectric relaxation and nuclear magnetic resonance of the mobility of water and of cations in some zeolites. Bull. SOC.Fr. Miw’ral. Cristallogr. 83,85-112. Echlin, P., and Galle, P. (1975). Conference 0rganizers.J. Microsc. Biol. Cell. 22(2-3), 1-520. Edmonds, D. T., and Mackay, A. L. (1975). The pure quadrupole resonance ofthe deuteron in ice. J. Magn. Reson. 20,515-519. Edmonds, D. T., and Zussman, A. (1972). Pure quadrupole resonance of 1 7 0 in ice. Phys. Lett. A41, 167-169. Edmonds, D. T., Hunt, M. J., Mackay, A. L., and Summers, C. P. (1974).The high sensitivity detection of pure quadrupole resonance. Aduan. Nucl. Quadrupole Reson. 1, 145-157. Edzes, H. T., Rupprecht, A., and Berendsen, H. J. C. (1972). Observation of quadrupolar NMR signals of ?Li and 23Nain hydrated oriented DNA. Biochem. Biophys. Res. Commun. 46,790-794. Eisenstadt, M., and Friedman, H. L. (1967). Nuclear magnetic relaxation in ionic solution. 11. Relaxation of 23Na in aqueous solutions of various diamagnetic salts. J. Chem. Phys. 46,2182-2193. Fabry, M. E., and Eisenstadt, M. (1975). Water exchanges between red cells and plasma: A measurement by nuclear magnetic relaxation. Biophys. J. 15, 1101-1110. Fabry, M. E., Koenig, S. H., and Schillinger, W. E. (1970).Nuclear magnetic relaxation dispersion in protein solutions. IV. Proton relaxation at the active site of carbonic anhydrase. J , Biol. Chem. 245,4256-4262. Farrar, T. C., and Becker, E. D. (1971). “Pulse and Fourier Transform NMR.” Academic Press, New York. Finch, E. D., and Homer, L. H. (1974). Proton nuclear magnetic resonance relaxation measurements in frog muscle. Biophys. J. 14,907-921. Finch, E. D., Harmon, J. F., and Miller, B. H. (1971).Pulsed NMR measurements ofthe diffusion constant of water in muscle. Arch. Biochem. Biophys. 147,299-310. Floyd, R. A., Yoshida, T., and Leigh, J. S., Jr. (1975). Changes of tissue water proton relaxation rates during early phases of chemical carcinogenesis. Proc. Natl. Acud. Sci. U.S.A. 72,56-58. Fritz, 0.G., and Swift, T. J. (1967).The state of water in polarized and depolarized frog nerves: A proton magnetic resonance study. Biophys. J. 7,675-687. Fujiwara, F., Reeves, L. W., Tracey, A. S., and Wilson, L. A. (1974). Studies of membrane processes. 111. Deuterium magnetic resonance as a tool in studies of the lipophilic region of membrane systems. J. Am. Chem. SOC.96,5249-5250. Fung, B. M. (1975).Orientation of water in striated frog muscle. Science 190,800-802. Fung, B. M., and McGaughy, T. W. (1974). The state of water in muscle as studied by pulsed NMR. Biochirn. Biophys. Acta 343,663-673. Fung, B. M., and Siegel, M. M. (1972). The ordering and relaxation times of water adsorbed on collagen fibers. Biochim. Biophys. Acta 278, 185-187.
NMR STUDIES OF WATER AND IONS WITHIN CELLS
65
Fung, B. M., and Trautmann, P. (1971). Deuterium NMR and EPR of hydrated collagen fibers in the presence of salts. Biopolymers 10, 391-397. Fung, B. M., and Wei, S. C. (1973).The effect of alkali and alkaline earth salts on the structure of hydrated collagen fibers as studied by deuterium NMR. Biopolymers
12, 1052-1062. Goldstein, J. I., and Yakowitz, H., eds. (1975).“Practical Scanning Electron Microscopy: Electron and Ion Microprobe Analysis.” Plenum, New York. Hahn, E. L. (1950).Spin echoes. Phys. Rev. 80,580-594. Hall, T., Echlin, P., and Kaufmann, R., eds. (1974).“Microprobe Analysis as Applied to Cells and Tissues.” Academic Press, New York. Hallenga, K., and Koenig, S. H. (1976).Protein rotational relaxation as studied by solvent ‘H and ‘H magnetic relaxation. Biochemistry 15,4255-4263. Hansen, J. R. (1971).Pulsed NMR study of water mobility in muscle and brain tissue.
Biochim. Biophys. Acta 230, 482-486. Hardy, R. C., and Cottington, R. L. (1949). Viscosity of deuterium oxide and water from 5” to 125”.J. Chern. Phys. 17,509-510. Haynes, D. H., Pressman, B. C., and Kowalsky, A. (1971).A nuclear magnetic resonance study of 23Na+complexing by ionophores. Biochemistry 10,852-860. Hazlewood, C. F., Chang, D. C., Nichols, B. L., and Woessner, D. E. (1974).Nuclear magnetic resonance transverse relaxation times of water protons in skeletal muscle.
Biophys. J. 14,583-606. Hecht, A. M., and Geissler, E. (1970). Nuclear magnetic resonance and relaxation ofadsorbed water in synthetic fluor montmorillonite. J . Colloid Interface Sci. 34,32-42. Hecht, A. M., Dupont, M., and Ducros, P. (1966).Nuclear magnetic resonance of the transport of adsorbed water in certain clay minerals. Bull. Soc. Fr. Mine‘ral. Cristallogr. 89,6-13. Held, G., Noack, F., Pollak, V., and Melton, B. (1973). Proton spin relaxation and mobility of water in muscle tissue. 2. Naturforsch., Teil C 28,59-62. Hertz, H. G. (1973).Nuclear magnetic relaxation spectroscopy. In: “Water: A Comprehensive Treatise” (F. Franks, ed.), Vol. 3,Chapter 7,pp. 301-399. Plenum, New York. Herzberg, G. (1945). “Infrared and Raman Spectra of Polyatomic Molecules.” Van Nosband-Reinhold, Princeton, New Jersey. Hilton, B. D., and Bryant, R. G . (1976).Proton magnetic resonance relaxation in lysozyme solutions: An isotope dilution experiment. J. Magn. Reson. 21, 105-108. Hindman, J. C., Svirmickas, A., and Wood, M. (1970).Spin-lattice relaxation of oxygen-17 in water. J. Phys. Chem. 74,1266-1269. Hinke, J. A. M. (1961).The measurement of sodium and potassium activities in the squid axon by means of cation-selective glass micro-electrodes. J . Physiol. (London)
156,314-335. Hoeve, C. A. J., and Kakivaya, S. R. (1976).On the structure of water absorbed in collagen. J . Phys. Chem. 80,745-749. Horowitz, S. B., and Fenichel, I. R. (1970). Analysis of sodium transport in the amphibian oocyte by extractive and radioautographic techniques. J. Cell Biol. 47,120-131. Hubbard, P. S. (1970).Nonexponential nuclear magnetic relaxation by quadrupole interactions. J . Chem. Phys. 53,985-987. James, T. L.(1975).“Nuclear Magnetic Resonance in Biochemistry: Principles and Applications.” Academic Press, New York. James, T. L., and Gillen, K. T. (1972).Nuclear magnetic resonance relaxation time and self-diffusion constant of water in hen egg white and yolk. Biochirn. Biophys. Acta
286, 10-15. James, T. L., and Noggle, J. H. (1969a).‘“a
nuclear magnetic resonance relaxation
66
MORDECHAI SHPORER AND MORTIMER M. CIVAN
studies of sodium ion interaction with soluble RNA. Proc. Natl. Acad. Sci. U.S.A. 62,644-649. James, T. L., and Noggle, J. H. (196913).Sodium-23 nuclear magnetic resonance studies of sodium aminocarboxylic acid complexes.J. Am. Chem. Soc. 91,3424-3428. James, T. L., and Noggle, J. H. (1972a). '"a NUR studies pf sodium ion interactions: Cysteine, aspartic acid, and citric acid. Bioinorg. Chem. 2,69-74. James, T. L., and Noggle, J. H. (1972b). 23NaN M R studies of sodium ion interactions. Phosphatidyl serine, phosphoserine, and phosphoethanolamine. Anal. Biochem. 49,208-217. Jardetsky, O., and Wertz, J. E. (1956). The complexing of sodium ion with some common metabolites. Arch. Biochem. Biophys. 65,569-572. Jardetsky, O., and Wertz, J. E. (1960). Weak complexes of sodium ion in aqueous solutions studied by nuclear spin resonance. J . Am. Chem. SCC.82,318-323. Katchalsky, A. (1971). Polyelectrolytes. Pure A&. Chem. 26,327-373. Kielman, H. S., and Leyte, J. C. (1973). Nuclear magnetic relaxation of sodium-23 in polyphosphate solutions. J . Phys. Chem. 77, 1593-1594. Kimmich, R., and Noack, F. (1970). Nuclear magnetic relaxation in protein solutions. 2. Naturforsch. Teil A 25, 1680-1684. Kimmich, R., and Noack, F. (1971). Nuclear magnetic relaxation in solutions of proteins and polypeptides. Ber. Bunsenges. Phys. Chem. 75,269-272. Klein, M. P., and Phelps, D. E. (1969). Evidence against orientation of water in rat phrenic nerve. Nature (London)224, 70-71. Knispel, R. R., Thompson, R. T., and Pintar, M . M. (1974). Dispersion of proton spinlattice relaxation in tissues. ]. Magn. Reson. 14, 44-51. Koenig, S. H., and Brown, R. D. (1973). Anomalous relaxation of water protons in solutions of copper-containing proteins. Ann. N.Y. Acad. Sci. 222, 752-763. Koenig, S. H., and Schillinger, W. S. (1969a). Nuclear magnetic relaxation dispersion in protein solutions: I. Apotransferrin. J . Biol. Chem. 244,3283-3289. Koenig, S . H., and Schillinger, W. E. (1969b).Nuclear magnetic relaxation dispersion in protein solutions. 11. Transferrin. J . B i d . Chem. 244,6520-6526. Koenig, S. H., Brown, R. D., and Brewer, C. F. (1973). Solvent proton magnetic relaxation dispersion in solutions of concanavalin A. Proc. Natl. Acad. Sci. U.S.A. 70, 475-479. Koenig, S. H., Hallenga, K., and Shporer, M. (1975). Protein-water interaction studied by solvent 'H, 'H, and "0 magnetic relaxation. Proc. Natl. Acad. Sci. U.S.A. 72,2667-267 1. Koide, G . T., and Carstensen, E. L. (1976). Dielectric properties of Sephadex and its water of hydration. J . Phys. Chem. 80, 55-59. Krynicki, K. (1966). Proton spin-lattice relaxation in pure water between 0°C and 100°C. Physicu (Utrecht) 32, 167-178. Kuniar, A., Welti, D., and Ernst, R. R. (1975). NMR Fourier zeugmatography.]. Magn. Reson. 18,69-83. Kuntz, I. D., Jr. (1971a). Hydration of macromolecules. 111. Hydration of polypeptides. J . Am. Chern. SOC. 93,514-516. Kuntz, I. D., Jr. (1971b). Hydration ofmacromolecules. IV. Polypeptide conformation in frozen solutions. J . Am. Chem. Soc. 93,516-518. Kuntz, I. D., Jr., Brassfield, T. S., Law, G. D., and Purcell, G. V. (1969). Hydration of macromolecules. Science 163, 1329-1331. Kvlividze, V. I., Kiselev,V. F., Kurzaev, A. B., and Ushakova, L. A. (1974). Mobile water phase on ice surfaces. Surf. Sci. 44,60-68. Lauterbur, P. C. (1973). Image formation by induced local interactions: Examples employing nuclear magnetic resonance. Nature (London) 242, 190-191.
NMR STUDIES OF WATER AND IONS WITHIN CELLS
67
Lev, A. A., and Armstrong, W. McD. (1975).Ionic activities in cells. Cum. Top. Membr. Transp. 6, 59-123. Lindblom, G. (1971).Ion binding in liquid crystals studied by NMR. 111. 23Naquadrupolar effects in a model membrane system. Acta Chem. Scand. 25,2767-2768. Lindstrom, T. R., and Koenig, S. (1974). Magnetic-field-dependent water proton spinlattice relaxation rates of hemoglobin solutions and whole blood. J. Magn. Reson. 15,344-353. Lindstrom, T. R., Koenig, S. H., Boussios, T., and Bertles, J. F. (1976). Intermolecular interaction of oxygenated sickle hemoglobin molecules in cells and cell-free solutions. Biophys. J. 16,679-689. Ling, G. N., and Cope, F. W. (1969). Potassium ion: Is the bulk of intracellular K+ adsorbed? Science 163,1335-1336. Lynch, L. J., and Haly, A. R. (1970). NMR study of the anisotropy of water absorbed by keratin. Kolloid 2. Z. Polym. 239,581-586. MacGillivray, A. D. (1972).Analytic description of the condensation phenomenon near the limit of infinite dilution based on the Poisson-Boltzmann Equation. J. Chem. Phys. 56, 83-85. Magnuson, N. S., and Magnuson, J. A. (1973).=Na interaction with bacterial surfaces: A comment on nuclear magnetic resonance invisible signals. Biophys. J. 13, 1117-1 119. Magnuson, J. A., Magnuson, N. S., Hendrix, D. L., and Higinbotham, N. (1973). Nuclear magnetic resonance studies of sodium and potassium in etiolated pea stem. Biophys. J. 13, 763-771. Manning, G . S. (1969). Limiting lows and counterion condensation in polyelectrolyte solutions. I. Colligative properties. J. Chem. Phys. 51, 924-933. Martinez, D., Silvidi, A. A., and Stokes, R. M. (1969). Nuclear magnetic resonance studies of sodium ions in isolated frog muscle and liver. Biophys. J. 9, 1256-1260. Mazzega, E., del Pennino, U., Loria, A,, and Mantovani, S. (1976). Volta effect and liquidlike layer at the ice surface.J. Chem. Phys. 64, 1028-1031. Migchelsen, C., and Berendsen, H. J. C. (1967). Deuteron magnetic resonance on hydrated collagen. In “Magnetic Resonance and Relaxation” (R. Blinc, ed.), pp. 761-766. North-Holland Publ., Amsterdam. Migchelsen, C., and Berendsen, H. J. C. (1973).Proton exchange and molecular orientation of water in hydrated collagen fibers. An NMR study of H 2 0and D 2 0 .J . Chem. Phys. 59,296-305. Migchelsen, C., Berendsen, H. J. C., and Rupprecht, A. (1968). Hydration of DNA. Comparison of nuclear magnetic resonance results of oriented DNA in A, B and C form. J. Mol. Biol. 37,235-237. Monoi, H. (1974). Nuclear magnetic resonance of tissue 23Na: I. 23Nasignal and Na+ activity in homogenate. Biophys. J. 14, 645-651. Neild, T. O., and Thomas, R. C. (1973). New design for a chloride-sensitive microelectrode. J . Physiol. (London) 231, 7-8P. Neild, T. O., and Thomas, R. C. (1974). Intracellular chloride activity and the effects of acetylcholine in snail neur0ns.J. Physiol. (London) 242,453-470. O’Reilly, D. E., and Peterson, E. M. (1969). Relaxation rates of alkali metal and chloride nuclei in aqueous solution. Quadrupole moment of ‘”Cs. J. Chem. Phys. 51, 4906-4908. Ostroy, F., James, T. L., Noggle, J. H., Sarrif, A., and Hokin, L. E. (1974). Studies on the characterization of the sodium-potassium transport adenosinetriphosphatase: Nuclear magnetic resonance studies of 23Nabinding to the purified and partially punfied enzyme. Arch. Biochem. Biophys. 162,421-425.
68
MORDECHAI SHPORER AND MORTIMER M. CIVAN
Outhred, R. D., and George, E. P. (1973).Water and ions in muscles and model systems. Biophys. J . 13,97-103. Palmer, L. G., and Civan, M. M. (1975). Intracellular distribution of free K+ in Chironomus salivary glands. Science 188, 1321-1322. Palmer, L. G., and Civan, M. M. (1976). Distribution of free C1- across the nuclear membrane of salivary gland cells of Chironomus. Clin. Res. 24,469A. Palmer, L. G., and Civan, M. M. (1977). Distribution of Na+, K+ and C1- between nucleus and cytoplasm in Chironomus salivary gland cells.]. Membr. B i d . (In press.) Peterson, S. W., and Levy, H. A. (1957). A single-crystal neutron-diffraction study of heavy ice. Acta Crystallogr. 10, 70-76. Powles, J. G., Rhodes, M., and Strange, J. H. (1966). Deuteron spin lattice relaxation in benzene, bromobenzene, water and ammonia. Mol. Phys. 11,515-523. Raaphorst, G. P., Kruuv, J., and Pintar, M. M. (1975). Nuclear magnetic resonance study of mammalian cell water: Influence of water content and ionic environment. Biophys. /. 15,391-402. Reisin, I. L., Rotunno, C. A., Corchs, L., Kowalewski, V., and Cereijido, M. (1970). The state of sodium in epithelial tissues as studied by nuclear magnetic resonance. Physiol. Chem. Phys. 2, 171-179. Resing, H. A. (1976). Quadrupole coupling constants in adsorbed water. Effects of chemical exchange. J . Phys. Chem. 80, 186-193. Reuben, J., Shporer, M., and Gabbay, E. J. (1975).The alkali ion-DNA interaction as reflected in the nuclear relaxation rates of 23Naand *%b. Proc. Natl. Acud. Sci. U.S.A. 72,245-247. Rotunno, C. A., Kowalewski, V., and Cereijido, M. (1967).Nuclear spin resonance evidence for complexing of sodium in frog skin. Biochim. Biophys. Actu 135,170-173. Rubinstein, M., Baram, A., and Luz, Z. (1971). Electronic and nuclear relaxation in solutions of transition metal ions with spin S = 312 and 512. Mol. Phys. 20,67-80. Samoilov, 0. Ya. (1972). Residence times of ionic hydration. In “Water and Aqueous Solutions: Structure, Thermodynamics, and Transport Processes” (R. A. Horne, ed.), pp. 597-612. Wiley (Interscience), New York. Saupe, A. (1964). Nuclear (magnetic) resonances in liquid crystals and in liquid-crystal solutions. Z. Nuturforsch., Teil A 19, 161-171. Shchori, E., Jagur-Grodzinski, J., Luz, Z., and Shporer, M. (1971). Kinetics ofcomplexation of macrocyclic polyethers with alkali metal ions. I. 23NaNMR of Na+-dibenzo 18-crown-6 in N,N-dimethyl formamides. J . Am. Chem. SOC. 93,7133-7138. Shchori, E., Jagur-Grodzinski, J., and Shporer, M. (1973). Kinetics of complexation of macrocyclic polyethers with sodium ions by NMR spectroscopy. 11. Solvent effects. J. Am. Chem. SOC.95,3842-3846. Shporer, M., and Achlama, A. (1976). Fine structure in the pure quadrupole resonance of‘ oxygen-17 in Ba(C103)2. ~ ~ by 1nuclear ~ double 0 resonance. J . Chem. Phys. 65,3657-3664. Shporer, M., and Civan, M. M. (1972). NMR ofSNa linoleate-water: Basis for an alternative interpretation of 23Naspectra within cells. Biophys. J . 12, 114-122. Shporer, M., and Civan, M. M. (1974). Effects of temperature and field strength on the NMR relaxation times of 23Nain frog striated muscle. Biochim. Biophys. Acta 354, 291-304. Shporer, M., and Civan, M. M. (1975). NMR study of I7Ofrom H,I7O in human erythrocytes. Biochim. Biophys. Acta 385,81-87. Shporer, M., and Luz, Z. (1975). Kinetics of complexation of potassium ions with dibenzo-18crown-6 in methanol by 3BK NMR.1. Am. Chem. SOC. 97,665-666. Shporer, M., Vega, A. J., and Frommer, M. A. (1974a). Diamagnetic susceptibility effects in NMR measurements of the properties of water in polymeric membranes.J. Polym. Sci., Polym. Phys. Ed. 12,645-652.
NMR STUDIES OF WATER AND IONS WITHIN CELLS
69
Shporer, M., Zemel, H., and Luz, Z. (1974b). Kinetics of complexation of sodium ions with valinomycin in methanol by 23NaNMR spectroscopy. FEBS Lett. 40,357-360. Shporer, M., Haas, M., and Civan, M. M. (1976). Pulsed NMR study of l'o from H2I7Oin rat lymphocytes. Biophys. J . 16, 601-611. Siebert, G. (1967). Isolation of nuclei. Methods Cancer Res. 2,287-301. Smith, D. W. G., and Powles, J. G. (1965). Proton spin-lattice relaxation in liquid water and liquid ammonia. Mol. Phys. 10,451-463. Speiss, H. W., Garrett, B. B., Sheline, R. K., and Rabideau, S. W. (1969). Oxygen-17 quadrupole coupling parameters for water in its various phases. J . Chem. Phys. 51,1201-1205. Sutter, E. J., and Harmon, J. F. (1975). Molecular motion in super-cooled liquids. I. Pulsed nuclear magnetic resonance of lithium 7 in 11M aqueous lithium chloride. J . Phys. Chem. 79,1958-1961. Thompson, R. T., Knispel, R. R., and Pintar, M. M. (1973). A study of the proton exchange in tissue water by spin relaxation in the rotating frame. Chem. Phys. Lett. 22,335-337. Tousimis, A. J. (1969). Electron probe microanalysis of biological structures. In "X-ray and Electron Probe Analysis in Biomedical Research" (K. M. Earle, and A. J. Tousimis, eds.), pp. 87-103. Plenum, New York. van der Klink, J. J., Zuiderweg, L. H., and Leyte, J. C. (1974). Nuclear magnetic relaxation of 23Nain polyelectrolyte solutions. J . Chem. Phys. 60,2391-2399. Verhoeven, J . , Dymanus, A., and Bluyssen, H. (1969). Hyperfine structure of HD"O by beam-maser spectroscopy. J . Chem. Phys. SO, 3330-3338. Waldstein, P., Rabideau, S. W., and Jackson, J . A. (1964). Nuclear magnetic resonance of single crystals of D20 ice.]. Chem. Phys. 41,3407-3411. Walker, J. L., Jr. (1971). Ion specific liquid ion exchanger microelectrodes. Anal. Chem. 43,89A-93A. Walter, J. A., and Hope, A. B. (1971). Nuclear magnetic resonance and the state of water in cells. Progr. Biophys. Mol. Biol. 23,3-20. Weisman, I. D., Bennett, L. H., Maxwell, L. R., Sr., Woods, M. W., and B u k , D. (1972). Recognition of cancer in vivo by NMR. Science 178,1288-1290. Wertz, J. E., and Jardetsky, 0. (1956). Nuclear spin resonance of aqueous sodium ion. J. Chem. Phys. 25,357-358. Woessner, D. E. (1961a). Nuclear spin relaxation in ellipsoids undergoing rotational Brownian motion. J . Chem. Phys. 37,647-654. Woessner, D. E. (1961b). Nuclear transfer effects in nuclear magnetic resonance pulse experiments. J . Chem. Phys. 35,41-48. Woessner, D. E. (1964). Molecular reorientation in liquids. Deuterium quadrupole relaxation in liquid deuterium oxide and perdeuterobenzene. J . Chem. Phys. 40, 2341-2348. Woessner, D. E., and Snowden, B. S., Jr. (1969a). A study ofthe orientation of adsorbed water molecules on montmorillonite clays by pulsed NMR. J . Colloid Znterface Sci. 30, 54-68. Woessner, D. E., and Snowden, B. S., Jr. (1969b). NMR doublet splitting in aqueous montmorillonite gels. J. Chem. Phys. SO, 1516-1523. Woessner, D. E., and Snowden, B. S., Jr. (1973). A pulsed study of dynamics and ordering of water molecules in interfacial systems. Ann. N.Y. Acad. Sci. 204,113-124. Woessner, D. E., and Zimmerman, J. R. (1963). Nuclear transfer and anisotropic motional spin phenomena: Relaxation-time temperature dependence studies of water adsorbed on silica gel. IV. J . Phys. Chem. 67,1590-1600. Yeh, H. J. C., Brinley, F. J., Jr., and Becker, E. D. (1973). Nuclear magnetic resonance studies on intracellular sodium in human erythrocytes and frog muscle. Biophys. J. 13,56-71.
This Page Intentionally Left Blank
Electrostatic Potentials at Membrane-Solution Interfaces STUART MCLAUGHLlN Depurtment of Physiology and Biophysics Health Sciences Center Stute Unicersity of N e w York Stony Brook, New York
I. Introduction . . . .
...................................................
71
......................
B.
Experimental Tests of the Gouy Equation
A.
The Conductance-Voltage Curves of Excitable Membranes ..... Distribution of Charged Lipids in Biological Membranes . . . . . . . . . . . 122
B.
E . Photochemical Reactions ....................... F. Osmolarity of Solutions in Small Vesicles . . . G. Other Effects . .
............. .............
..............
Appendix I1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix 111 ...... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
I.
126 127
132 133 135
INTRODUCllON
There is now strong evidence that the lipids in the membranes of all cells and subcellular organelles are arranged in the form of a bilayer, 71
72
STUART MCUUGHLIN
with the hydrocarbon tails sequestered away from the water and the polar head groups exposed to the aqueous environment (Stoeckenius and Engelman, 1969; Singer, 1971; Singer and Nicolson, 1972; Branton and Deamer, 1972; Tfauble and Overath, 1973; Bretscher, 1973; McLaughlin et al., 1975a). About 10-20% of the lipids in the membranes of many cells (e.g., nerves, muscles) and organelles (e.g., mitochondria, synaptic vesicles) bear a net negative charge, whereas positively charged lipids are extremely rare (White, 1973). As a phospholipid in a bilayer occupies an area of about 60 (Fettiplace et d., 1971; Levine and Wilkins, 1971; Haydon and Hladky, 1972), the average charge density on the bilayer portion of a membrane comprised of 20% negative lipids is = 1 electronic charge/300 &. These charges produce a negative electrostatic potential in the aqueous phase immediately adjacent to the membrane, the potential in the bulk aqueous phase being defined as zero. When the concentration of monovalent ions in the bulk aqueous solution is 10-’ M and the temperature is 25”C, the Gouy-Chapman theory of the difhse double layer predicts that the surface potential will be - 60 mV. This is a substantial potential, when compared to the value of kT/e = RTIF, which is 25 mV for a monovalent ion in solution at room temperature, and will directly influence a variety of membrane-related phenomena. The concentration of monovalent cations at the surface of the bilayer will, for example, be an order of magnitude higher than the concentration of these ions in the bulk aqueous phase. The local pH will, therefore, be one unit lower than in the bulk pH, a phenomenon that will affect many enzymatic processes. As the permeability of the membrane to ions is related to the interfacial rather than the bulk aqueous concentrations of the ions, these membrane permeabilities will be affected by the surface potential. The surface potential produced by charged lipids, is, moreover, dependent on the salt concentration in the bulk aqueous phase, and a seminal paper by Chandleret aZ. (1965) illustrated the importance of this effect in understanding the electrical properties of nerves. These and other biological examples will be discussed in Section VI. In Section II,A the predictions of the diffuse double layer theory of Gouy (1910)and Chapman (1913)will be discussed briefly. There are many serious theoretical objections to this theory, so Section II,B will consider the various experimental tests of the theory that have been conducted on artificial bilayer membranes. It must be admitted that the theory has been tested at the mercury-water interface for many years by electrochemists, and found to be only fair at describing the measured changes in capacitance (e.g., Bockris and Reddy, 1973).
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
73
Lipid bilayers are, however, the system of choice for those of us who are ultimately interested in the application of this physical concept to biology and medicine. The pioneering work of Mueller et al. (1963)in developing planar black lipid membranes and of Bangham in developing spherical liposomes (e.g., Bangham e t al., 1974)has provided us with model systems that are identical to the lipid bilayer portion of the biological membranes we wish to study. The absence ofproteins, polysaccharides, and other macromolecules normally present in biological membranes can be considered an advantage if the objective is to test how well the theory of the diffuse double layer describes the electrostatic potential produced by charges on lipids. The statement is not meant to imply that charges on macromolecules in membranes are unimportant, but this topic is difficult to approach, both experimentally and theoretically, and is beyond the scope of this review. Section I11 will deal with the hydrophobic adsorption of charged molecules to bilayer membranes. It will be argued that the simplest possible theoretical description of the adsorption is, in fact, consonant with the available experimental evidence. Section IV will deal with the electrostatic potential produced by molecular dipoles at membrane-solution interfaces. There is little question that the dipole potential is large in magnitude, but its origin is obscure and its biological relevance uncertain at the present time. Section V will deal with the electrostatic boundary potential produced by charges located in the interior of the membrane a few angstroms from the interface. Finally, Section IV will deal with a few examples of the possible biological significance of these electrostatic surface potentials. II. FIXED CHAROES AT MEMBRANE-SOLUTION INTERFACES A. Theoretical Description of the Diffuse Double layer
Many excellent reviews have been written about the theory of the diffuse double layer (e.g., Grahame, 1947; Venvey and Overbeek, 1948; Haydon, 1964; Delahay, 1965; Mohilner, 1966; Barlow, 1970). The history of this concept since its inception b y Gouy (1910) and Chapman (1913) and the various modifications which have been added to the simplest theory in the past 60 years are discussed by Bockris and Reddy (1970, pp. 623-843) in their textbook on Modern EEectrochemistry. There is, therefore, little need for either another detailed or historical treatment of double layer theory. The interested biologist who wishes an introductory mathematical treatment is re-
74
STUART MClAUGHLlN
ferred to the relevant section of a recent book by Aveyard and Haydon (1973, pp. 31-57) and to Appendix I of this paper. In this Section, I will attempt, in a relatively nonmathematical, heuristic manner, to give the reader an intutitive grasp of the theory. Some of the salient predictions of the theory will then be illustrated graphically. Figure 1A illustrates, in a highly schematized manner, the distribution of ions at a given instant in time near a charged surfwe immersed in an aqueous solution. We may consider the surface to be a bilayer membrane consisting of a mixture of zwitterionic and negatively
0
A
10
20
DISTANCE, X
(8)
x)
B
-1Ix-
I
I
I
0
10
20
DISTANCE, X
(A)
D
FIG.1. (A) Schematic diagram of the distribution of ions near a negatively charged membrane. (B) The potential profile predicted by the Couy-Chapman theory of the diffuse double layer when 20% of the lipids in the membrane bear a net negative charge. (C)The concentrations of anions and cations adjacent to the membrane, as predicted by the Gouy-Chapman theory. (D) A parallel plate capacitor model of the diffuse double , Debye length from layer. We assume that the counterions are located a distance 1 / ~the the membrane. The average distance between the charges on the surface of the membrane is d = 18 A. See text for details. The temperature was 25°C.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
75
charged phospholipids. The negative charges on the lipids produce an electric field that attracts counterions, ions of the opposite sign to the charge on the membrane, and repels coions, ions of the same sign to the charge on the membrane. In the absence of any specific short range or “chemical” interactions, the counterions d o not remain at the surface for the same reason that the earth’s atmosphere does not collapse to the ground. Both the gas molecules in the earth’s atmosphere and the ions in the “atmosphere” at the surface of the membrane have thermal energy, which manifests itself as a statistical tendency for the molecules or counterions to diffuse from regions of high to low concentration. A balance is struck, in the case of both the earth’s and the membrane’s atmospheres, between the attractive forces generated by the gravitational or electrostatic fields and the statistical tendency of the gas molecules or counterions to diffuse away from the surface at which they are concentrated.’ In terins of the Gouy-Chapman theory of the diffuse double layer, the magnitude of the electrostatic potential decreases with distance from the membrane, as illustrated in Fig. 1B. If we assume that 20%of the lipids bear a single net negative charge (e.g., phosphatidyl serine, PSL; phosphatidyl glycerol, PG-; phosphatidyl inositol, PI-) and that the remaining 80% of the lipids are either zwitterions (e.g., phosphatidy1 choline, PC’; phosphatidyl ethanolamine, PE’) or neutral (e.g., cholesterol, C’), the average charge density is u -- 1 electronic charge/300 di2 because a phospholipid occupies an area of = 60 A2. If the concentration of salt in the bulk aqueous phase is C = lo-’ M , the theory predicts that the potential at the aqueous side of the membrane-solution interface is Jlo = -60 mV, as illustrated in Fig. 1B. When the surface potential is not too high, the magnitude of the potential decreases in an exponential manner with distance frcm the membrane. As indicated in Fig. lB, the distance at which the potential falls to l / e its value at the surface of the membrane is called the Debye length is about 10 A when the bulk Debye length, 1 / ~ The . concentration of inonovalent ions is lo-’ M and about 100 A when the M. The concentration of ions concentration of monovalent ions is at any distance away from the membrane may be calculated from the T h e analogy is not perfect, and there are important differences between the two “atmospheres.” T h e gas molecules in the earth’s atmosphere d o not significantly modify the gravitational attraction and the number of particles at a given height, N ( h ) ,falls of!; for an isothermal atmosphere, merely as predicted by the Boltzmann relation or barometer forinula: N ( h ) = N ( o ) . e(-mah’kT1, where tn is the mass, h is the height, g is the gravitational constant, k is Bolkmann’s constant, and T is the temperature. I n the ionic atmosphere near the membrane, the ions do modify the electric field that attracts them to the surface.
76
STUART MCLAUGHLIN
potential. illustrated in Fig, 1B via the Boltzmann relation. These concentrations are illustrated in Fig. 1C. If C = lo-' M and $o = - 60 mV, for example, the Boltzmann relation predicts that the concentration of monovalent cations at the surface of the membrane is 1M and that the concentration of monovalent anions is 1C2M . A simple analogy may be of help in illustrating two important features of the diffuse double layer. We first note that the membrane plus any volume of fluid which extends for more than a few Debye lengths must be electroneutral. The excess number of counterions in the diffuse double layer must, therefore, be exactly equal to the number of charges on the membrane. As a crude approximation, we can consider all these counterions placed at an average distance from the membrane (Fig. 1D); this average distance, as shown below, is the Debye length. We are thus considering the diffuse double layer (Fig. lB, 1C) to be analogous to a parallel plate capacitor (Fig. 1D). The analogy has some historical significance (Bockris and Reddy, 1970) and, as discussed below, is most valid for low values of the surface potential. For a capacitor, the field is constant or, equivalently, the potential falls in a linear manner with distance between the plates, as illustrated in Fig. 1D. The electric field or gradient of the electrostatic potential predicted by the capacitor analogy is only identical to the field predicted by the theory of the diffuse double layer at the membrane-solution interface (compare Figs. 1B and lD), but the analogy does illustrate how the surface potential depends on the charge density and the salt concentration. In terms of the model illustrated in Fig. l D , increasing the charge on the membrane will increase the potential at the surface of the membrane. (The voltage V across a capacitor is related to the charge density (+ via V = a/C'where C' is the capacitance per unit area. For a parallel-plate capacitor, C' = E ~ E ~ where K , E~ is the dielectric constant of the medium separating the plates, c0 is the permittivity of free space, and 1 / is~ the spacing between the plates.) If the capacitance remains constant, V will increase linearly with (+.The dependence of surface potential on charge density predicted by the Gouy-Chapman theory is illustrated in Fig, 2, Note that for low charge densities the magnitude of the surface potential does increase linearly with the charge density. When the concentration of ions in the bulk phase is reduced, the Debye length or average distance of the counterions from the membrane increases. In terms of the model presented in Fig. l D , this is equivalent to moving the capacitor plates farther apart. From the capacitor equation presented in the previous paragraph, it is apparent
77
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
0
-20 -40
3
-60
x
.3;
-80
-100
-120
-140
0
10
x)
30
DISTANCE C i )
FIG.2. The potential profiles predicted by the Gouy-Chapman theory for different values of the surface charge density. For low values of the charge density, the magnitude of the surface potential increases in an approximately linear manner with the charge density, and the magnitude of the potential falls in an approximately exponential manner with distance from the membrane-solution interface.
that an increase in 1 / will ~ produce an increase in V for a given charge density u.This is in fact the behavior predicted by the theory of the diffuse double layer. As illustrated in Fig. 3, a decrease in the salt concentration increases the value of the Debye length and therefore increases the magnitude of the surface potential. The dependence of the potential at the surface of the membrane on the charge density u and salt concentration C is quantitatively predicted by the Gouy equation from the theory of the diffuse double layer. This equation is derived in Appendix I, but for those who lack the time or the inclination to follow derivations of equations, a brief outline of the approach is given here. The electrostatic attraction of the counterions and repulsion of the coions from the membrane is described by Poisson’s equation [Eq. (lA), Appendix I], one of the four Maxwell equations. (Poisson’s equation is the differentia1 form of Gauss’ Law, and this reduces, for a single point charge, to Coulomb’s Law.) The statistical tendency of the counterions to diffuse away from, and of the coions to diffuse toward, the membrane is expressed by the Boltzmann relation from statistics [Eqs. (2A) and (3A)l. The combination of these two equations results in the Poisson-Boltzmann relation
+,,
78
STUART MCLAUGHLIN 0 -20 -40
-60
>
-F
-80
X
g -100 -120
-140
-160
-180
0
20
40
60
80
100
DISTANCE (8)
FIG.3. values of Note that when the
The potential profiles predicted by the Couy-Chapman t..eory for different the concentration of monovalent electrolyte in the bulk aqueous solution. both the Debye length and the magnitude of the surface potential increase salt concentration decreases.
[Eq. (4A)] which can be solved, utilizing the appropriate boundary conditions, to yield an expression [Eqs. (6A)-(8A)] for the potential at any distance from the membrane. These profiles are illustrated in Figs. 2 and 3. Finally, by invoking the conditjon of bulk electroneutrality, one obtains the following relation between Jlo, the electrostatic potential in the aqueous phase at the surface of the membrane located at x = 0, and u the charge density:
Au/*
= sinh(~e$~/2kT)
(1)
where k is the Boltzmann constant, T is temperature, e is electronic charge, z is the valence of the symmetrical electrolyte solution, and C is the bulk aqueous electrolyte concentration. A = ~ / ( ~ N E , E ~ ~ T ) ” ~ where N is Avogadro’s number, E , the dielectric constant, and e0 the permittivity of free space. Values of A are given in Table I for different temperatures. If T = 25”C, for example, and we express u in electronic charges/square angstrom and C in moles/liter, then 1 3 6 . 6 u / f i = sinh(zJIo/51.38), where Jlo is in millivolts.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
79
INTERFACES
TABLE I VALUES OF THE CONSTANTS A = 1 / ( 8 c , ~NkT)1’2AND kT/e = RT/F IN EQ. (1) AT DIFFERENT VALUES OF THE TEMPERATURE T
VC)
5 10 15 20 22
kT/e (mV)
A ( 6 A2)
T
kT/e
A
23.96 24.40 24.82 25.26 25.43
135.1 135.4 135.8 136.2 136.4
25 30 35 40
25.69 26.12 26.55 26.98
136.6 137.0 137.5 138.0
” To use these values o f A , express the concentration C in moles dm-3 (M) and cr in electronic charges/%r’.If cr in Eq. (1) is expressed in SI units (C m-*),A must be divided by 16.0.Values of all constants were taken from the Hnndbook ofClteniistry nndl‘hysics.
Note that for any parameter x, sinh x = (ex - e-S)/2 and that for x << 1, sinh x + x. Thus, for small potentials, Eq. (1) reduces to: (T
=
EpEoKqo
(2)
where 1 / ~ is the Debye length: K
= (2e2z2NC/~r~OkT)1’2
(3)
Equajion (2) is identical in form to the equation for a parallel-plate capacitor, provided we interpret 1 / ~ as the distance between the plates. Equations (2) and (3) indicate that the Debye length 1 / and ~ surface potential I!,I~ vary inversely with the square root of the salt concentration C. The rationale for considering the diffuse double layer as analogous to a parallel-plate capacitor is valid only at low surface potentials, when we can approximate Eq. (1) by Eq. (2). At high negative potentials, sinh (zet,b0/2kT)= - 1/2 exp( - zeq0/2kT) and Eq. (1)reduces to: 4A2u2/C = exp(- zet,bo/kT)
(41
The surface potential, qotincreases in proportion to the charge density for low values of Jlo [Eq. (2)]but in proportion to the log of the charge density [Eq. (4)l for high values of t,bo, as illustrated in Fig. 2. Equation ( l ) , the Gouy equation, predicts the dependence of the surface potential on salt concentration in a solution of symmetrical electrolytes. Divalent counterions should have a substantially larger effect than monovalent ions on the surface potential, a prediction with biological significance. It is apparent from Fig. 3 or from consideration of Eq. (1)that if u = - 1/300 A2, C = lo-’ M , T = 25”C, z = 1,
80
STUART MCLAUGHLIN
then q0 = -60 mV. To calculate the concentration of divalent ions that will produce the same surface potential, we insert x = 2, J !,I~ = - 60 mV, T = 25"C,A = 136.6, u = - 1/300 Az into Eq. (1) and solve for C?+,the concentration of divalent electrolyte in the bulk aqueous phase. We obtain a value of C2+ = 8 mM. Thus, at a value of u and q0that we expect to find on many biological membranes, divalent ions are predicted to be an order of magnitude more effective than monovalent ions in changing the surface potential via a nonspecific double layer effect. An apparent paradox now arises. The concept of ionic strength (e.g., Tanforg, 1961, p. 466; Moore, 1972, p. 443) predicts that divalent ions should be only a factor of four more potent in exerting electrostatic effects than are monovalent ions. The concept of ionic strength, however, arises from the Debye-Huckel theory of weak electrolytes, and it is not applicable to highly charged membranes." The Gouy relqtion, Eq. (l),is only valid for a symmetrical (z-z; e.g., MgS04,NaC1) electrolyte solution, although the valence of the coion is of little consequence. The relevant equation from diffuse double layer theory for a solution of mixed electrolytes was derived by Grahame (1947) and is presented as Eq. (11A) in Appendix I. Some theoretical curves illustrating the dependence of the surface potentials of membranes on the concentration of divalent ions in the presence of a fixed concentration of monovalent ions are given by McLaughlin et al. (1970, 1971) and Muller (1971), while Abraham-Schrauner (1975) discusses a method of calculating the dependence of the potential on distance under these conditions. It should be stressed that a great many implicit assumptions have entered into the derivation of the Gouy relation, Eq. (l),or the more general Eq. (11A). These assumptions are discussed briefly by Aveyard and Haydon (1973, pp. 43-46), and in more detail in the references cited in the first paragraph of this section. From the point of In the Debye-Huckel theory of weak electrolytes, the Poisson-Boltzmann equation [Eq. (4A) in Appendix I] i s linearized. That is, it is assumed that the potential $, << kT/e = 25 mV, the right-hand side of Eq. (4A)is expanded as a power series, and only the first term is retained. The linearization is obviously incorrect when potentials of60 mV or higher are encountered, as they will be on many biological membranes. As an aside, we note that, for symmetrical electrolytes (e.g., B-z salts), the odd terms in the power series expansion of the right-hand side of Eq. (4A)cancel, and the expansion is thus equivalent to retaining the first two terms. This is not so, however, for mixed e l e c trolytes. Biological solutions usually contain cations of different valences, so linearization is a particularly poor assumption for these solutions. To calculate the potential adjacent to a charged membrane exposed to a solution of mixed electrolytes, one must use Eq. ( 1 IA).
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
81
view of applying the theory to a bilayer or biological membrane, it would seem particularly inappropriate to assume that (i) ions are point charges; (ii) the dielectric constant i s equal to its bulk value up to the surface of the membrane; (iii) image charge effects can be ignored: (iv) the surface charge is smeared uniformly over the membrane. We discuss briefly here the fourth assumption because Haynes (1974) has suggested that discrete charge effects may be responsible for the apparent discrepancies between some of his observations on bilayers and the predictions of the Gouy-Chapman theory. The discrete charge effect arises because the charges on the lipids are not literally smeared over the interface: but are arrayed as shown in Fig. 4. An ion in a medium ofhigh dielectric constant, such as water, will b e attracted back towards the bulk aqueous phase as it approaches a medium of low dielectric constant, such as a membrane. The attraction is due to ion-dipole forces, and these forces can be calculated by the mathematical method of images, hence the name “image” forces. For a calculation ofthese forces adjacent to a membrane, the interested reader is referred to Neumcke and Lauger (1969),Haydon and Hladky (1971),Andersen and Fuchs (1975), and Bradshaw and Robertson (1975). Lipids in both artificial bilayers and biological membranes are capable of rapid translational motion, their diffusion constants beingD = cm2/sec (for a review, see Edidin, 1974). Two-dimensional diffusion problems must be approached with caution (Saffman and Delbriick, 1975), but the value of D has been estimated by three independent experimental techniques and is surely qualitativelycorrect. For motion in a plane, a relation derived by Einstein (1956; p. 17) states that r 2 = 4 D t , where rTis the mean square displacement, and t is the time. In 1psec, a lipid in a fluid membrane will diffuse about 20 A, the distance between charged lipids in a membrane containing 20% negative lipids. It follows that we may regard the charges on the lipids as beingsmeared uniformly over the interface if we are considering certain nonequilibrium processes that last longer than 1 psec. Lipid-soluble ions such as tetraphenylborate and the noncmz/sec actin-K+ complex, for example, should have diffusion coefficients ofD = 5. in a bilayer membrane. In the absence ofany long-range or image forces, they will thus require a time t = x2/W = sec to diffuse across a membrane 30 A thick. ConducL sec after the application of a voltage tance measurements made at times longer than clamp will, as discussed in more detail by Andersen and Fuchs (1975), depend on a “pseudostationary” rather than an equilibrium distribution of ions. The equilibrium concentration of the membrane-permeant ions, reflects, on the other hand, the timeaverage electrostatic potential. This potential, provided it could be estimated by rapid conductance measurements via Eqs. (5) or (6),is not necessarily the “smeared charge” potential. It is also distinct from the electrostatic potential that is inferred from actual conductance measurements. The difference between these two electrostatic potentials near the center of the membrane, the region that determines the conductance (Andersen and Fuchs, 1975, see Fig. 14), is subtle and probably of little consequence. In brief, it can be argued quite strongly, on both temporal and spatial grounds, that discrete charge effects should not affect such conductance measurements. They could, however, affect the adsorption of charges to interfaces, which depends on yet a different time-average value of the electrostatic potential within the membrane. (See footnote 10.)
82
STUART MCLAUGHLIN
FIG. 4. Schematic description of the field lines emanating from a fixed array of charges located a few angstroms from the hydrocarbon region of the membrane. Field lines show the direction ofthe electric vector (e.g.,Feynmanet ul., 1964, pp. 4-1 1). The density of the lines illustrates the strength of the electric field, which falls off rapidly with distance from the membrane.
If the charge density were uniform, the field lines would be everywhere perpendicular to the surface of the membrane. It is apparent from the diagram that the lines of force are approximately perpendicular to the surface a short distance away from the membrane. The discrete charge effect should be most important when the spacing between the charges is high (i:e., at low charge densities) and when the Debye length is short (i.e., at high salt concentrations). I n view of all the simplifying assumptions that enter into the theory, one has every right to be suspicious of the validity of the Gouy equation. As discussed in the next section, however, five independent lines of experimental evidence indicate that the Gouy-Chapman theory of the diffuse double layer provides a remarkably good description of the electrostatic potential due to charges at the surface of artificial bilayers. 6. Experimental Tests of the Gouy Equation
1. CARRIERSAS “PROBES”OF THE SURFACEPOTENTIAL As the potential we wish to investigate falls essentially to zero in a few tens of angstroms in a physiological solution (Fig. lB), it would obviously be futile to attempt to measure it with a device like an open-tip microelectrode. A “molecular voltmeter” is required, and the molecule that has been most widely used for this purpose is the antibiotic nonactin. This nautral carrier (for recent reviews see McLaughlin and Eisenberg, 1975; Hladky, 1977) functions by binding an alkali metal cation such as K+, solubilizing the ion in the low di-
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
83
electric interior of the membrane, and thereby increasing the conductance of the bilayer. Nonactin (Fig. 5A) solubilizes the cation by two means; it places a hydrophobic coat on the K+ ion, and it increases the size of the ion from about 2 to 5 A. This latter factor decreases the Born charging energy required to move the ion from a water into a hydrocarbon phase from about 40 to 20 kcal/mole (Parsegian, 1969).The hydrophobic “coat” reduces the energy further by about 25 kcal. The complex thus partitions favorably into the membrane (Haydon and Hladky, 1972). It is still not clear whether the movement of the carrier-ion complex should be described in terms of a NernstPlanck diffusion or absolute rate theory processes, (Zwolinski et al., 1949; Ciani, 1965), but there is a consensus that the carrier binds an ion at one interface, transports it across the membrane, releases it on the other side, then returns to the first side to complete the cycle (Fig. 5B). The conductance G+ produced by this process depends, as one might intuitively expect, on the concentration of carrier-ion complexes inside the membrane. At equilibrium, the number of these complexes in the membrane is proportional to exp (- ei,bo-/kT), where i,bo- is defined as the electrostatic potential within the membrane, more specifically within any dipole layer located at the interface, measured with respect to the potential in the bulk aqueous
aqueous phase
d A
B
FIG. 5. (A) Diagram of the molecular structure of the nonactin-K+ complex. The carbons are represented by filled circles, the oxygens by open circles, and the potassium by a heavy circle. The hydrogen and methyl side groups have been omitted for clarity. (B) Diagram of carrier-mediated ion translocation across a membrane, defining the various rate constants (adapted from McLaughlin and Eisenberg, 1975).
a4
STUART MCLAUGHLIN
phase, which we define to be zero. This potential can arise from both surface charges and surface dipoles. It is, in principle, unmeasurable (Guggenheim, 1929; 1930), but if we assuine that the thickness of the membrane and both the mobility and the standard chemical potential of the charged complex in the membrane are unaffected by changes in the potential Jl0-, these changes can be estimated by means of conductance measurements. If G' and G" designate the conductances of two membranes with different surface potentials:
G 'I /G
> = exp (- eAt,bo-/kT)
(5) where A$,- = - $&-isthe difference between the electrostatic potentials in the interior of the two membranes. It is also necessary to assume that the interfacial reactions occur rapidly and that the application of a small potential does not, therefore, significantly perturb the equilibrium (Neumcke, 1970). These assumptions can all be tested experimentally with control experiments. The first test of the double layer theory on bilayer membranes was made by Lesslauer et al. (1967). They observed that iodide ions greatly enhanced the conductance of black lipid membranes and that the addition of indifferent electrolyte (KCI) increased the iodidemediated conductance when the membranes were formed from the negative lipid phosphatidyl inositol. They proposed that the addition of KC1 reduced the magnitude of the diffuse double layer potential, as predicted by Eq. (1) and illustrated in Fig. 3. This, they reasoned, should increase the concentration of iodide anions within the membrane and, therefore, the conductance. The ratio of the anion conductances, measured before and after the addition of indifferent electrolyte, is given by GII/GL = exp (+ eA$,-/kT)
(6)
The change in the potential within the membrane, as calculated from Eq. (6),agreed with the change in t,bo, the surface potential in the aqueous phase, predicted by Eq. (1). This important study suggested that the Gouy-Chapman theory was qualitatively applicable to membranes but received little attention for several years, possibly because other factors (e.g., mobility, dielectric constant) could also have contributed to the change in iodide conductance produced by the indifferent electrolyte and because the mechanism of permeation of iodide ions was not understood at that time. Finkelstein and Cass (1968) next pointed out that iodide ions only permeate the membrane readily when traces of molecular iodine are present. Iodine allows the formation of 1, and I; complexes, which permeate membranes because the
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
85
charge is effectively shared between the atoms of the complex, reducing the Born energy. Iodine thus functions as a carrier for iodide ions in much the same manner that nonactin functions as a carrier for potassium ions. McLaughlin et a l . (1970) then used a variety of carriers to test the predictions of the double layer theory. Nonactin, valinomycin, a cyclic polyether, and the polyiodide system all yielded results qualitatively compatible with diffuse double layer theory [Eq. (l)]when a variety of neutral and negative membranes were examined in monovalent salt concentrations ranging from to 1 M . The divalent ions Ca2+and Mg2+also had the expected effect on the surface potential when they were added in the presence of low concentrations of monovalent ions. Recall that the Gouy relation, Eq. (l),predicts that a concentration of 10+ M divalent ions will exert a larger effect on the surface potential than a concentration of lov3 M monovalent ions when the membrane is formed from negative lipids (w = - 1/60 A2). This striking effect of divalent ions on the surface potential of negative membranes was also tested under more physiological conditions (McLaughlin et al., 1971). Meybranes were formed from the negative lipids phosphatidyl serine ( P F ) or phosphatidyl glycerol (PG-) in a decimolar solution of monovalent ions, and the effects of the alkaline earth cations on the surfacepotential were examined. When the bilayers were formed from PS=, the addition of Sr2+or Ba2+decreased the magnitude of the surface potential as predicted by the theory of the diffuse double layer. In particular, the potential decreased 27 mV for a tenfold increase in concentration in the millimolar-decimolar range, as predicted by Eq. (11A) (or approximately by Eq. (1) with z = 2 and the effect of the monovalent ions ignored). A tenfold increase in the concentration of Ca2+also produced a 27 mV decrease in ~ the potential in this region, which was again due to “ ~ c r e e n i n g , ”but It will be helpful to refer to the nonspecific effect an ion exerts on the surface potential of a charged membrane [i.e., a change in salt concentration C in Eq. (1)l as “screening.” The term “binding” will be reserved for those ions that have the ability to change the surface charge density [i.e., u in Eq. ( l ) ]and thus to affect the surface potential. Some ions are capable of both “screening” and “binding” and, when Ca2+is added to the lo-’ M KCI solution bathing a membrane formed from the negative lipid PS‘, a rather complex sequence of events occurs. When the concentration of Ca2+ in the bulk M), bindingoccurs to the phase is increased from very low values (lod6< [CaZ+] < membrane, which reduces the charge density and the magnitude of the surface poten< [Ca2+l< lo-’ M, the binding remains approximately constant, and tial. When the surface potential is reduced essentially by a screening mechanism. When [Ca*+]> lo-’ M, more binding occurs. The binding remains essentially constant over the intermediate range because the concentration offree calcium ions at the surface of the membrane is essentially independent of the bulk concentration. The following
86
STUART MCIAUGHLIN
it was necessary to invoke some binding to account for the observation that this cation was effective at a lower concentration than Sr2+or Ba2+.One can object to the use of neutral carriers like nonactin as “probes” of changes in the surface potential because the conductance they produce will also respond to changes in membrane fluidity, dielectric constant, etc. Any changes in these parameters, however, shouId cause the conductance produced by both positive and negative species to change in the same direction. When symmetrical effects are seen in opposite directions with carriers of cations and anions (McLaughlin et al. 1970, 1971), it is difficult to envision any factor other than the electrostatic potential being of importance. The probes do respond, however, to both the diffuse double layer potential produced by charges and the potential produced by dipoles associated with lipids (Section IV). As discussed by Haydon and Hladky (1972), the results obtained with the probes on membranes formed from negative lipids (Lesslaueret aZ., 1967; McLaughlin et aZ., 1970, 1971) could all have been due to a fortuitous change in the dipole potential with salt concentration. This would be a remarkable coincidence, because experiments reveal that the dipole potential of membranes formed from either zwitterionic (e.g., PE’, Szabo et d., 1972; McLaughlin et al., 1971; and PC’, Hladky and Haydon, 1973) or neutral lipids (e.g., GMOO, Hladky and Haydon, 1973; and GDOO, Szabo et al., 1973) does not vary with a change in either the alkali metal or alkaline earth chloride salt concentration. Furthermore, by measuring the zeta potential, the potential at the hydrodynamic plane of shear, one should be able to distinguish experimentally between a change in dipole and diffuse double Iayer potentials. The zeta potential should respond only to changes in the latter parameter. The relationship between the zeta and surface potentials is discussed in detail by Carroll and Haydon (1975). For low charge densities, the plane of shear is thought to lie within 1 A of the envelope of the head group. argument illustrates this point for a membrane formed from negatively charged lipids. If the concentration of monovalent ions is C+ = lo-’ M and the “intrinsic” dissociation constant for Cazt with the membrane is K = 10 M ,then, to a good approximation, the surface potential predicted by Eq. (1lA)for lo4 < cP+ < lo-’ M may be represented analytically by the Gouy expression for divalent ions alone, Eq. (4) (McLaughlin et al., 1971). By combining this expression, exp - (Be&/kT) = 4Apd/C?+, with the Boltzrnann relation, Cet(0) = C2+exp- (2e&,/kT), we obtain: Cz+(0) 4 A 2 d for < cP+ < lo-’ M.The free concentration of divalent ions at the surPce ofthe membrane, C + ( O ) , is thus approximately independent of the bulk concentration, and the number of bound ions (or charge density) is, therefore, also independent of the bulk concentration.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
87
The zeta potential 5 should, therefore, closely approximate the surface potential t,b0 for the experimental conditions of the studies discussed below.
2. ZETAPOTENTIALMEASUREMENTS MacDonald and Bangham (1972) measured the electrophoretic mobility u of vesicles of known charge density in a variety of different salt solutions. The zeta potential 6 was calculated from the Helmholtz-Smoluchowski equation:
5 = ?W/E,EO
(7)
where 7 is the viscosity, E , the dielectric constant, and e0 the permittivity of free space. Overbeek and Wiersema (1967) discuss, in some detail, the assumptions inherent in the derivation of this equation. Shaw (1970) or Aveyard and Haydon (1973) may be consulted for a less detailed derivation. The value of 5 was determined for vesicles containing a mixture of negative (phosphatidic acid) and zwitterionic M solution of monova(phosphatidyl choline) lipids formed in a lent ions. The agreement with the prediction of the Gouy equation, Eq. (l),was good, at least up to a potential of about - 60 mV. Above this value, the magnitude of the zeta potential did not increase as rapidly as double layer theory predicts the surface potential should increase, but the problem probably lies with the technique rather than the theory (Haydon, 1964). MacDonald and Bangham (1972)also measured the zeta potentials of vesicles formed from brain phospholipids and cholesterol as a function of the salt concentration (lod3to 10-1 M). They observed good agreement between these measurements and measurements of the change in surface potential of a monolayer formed from the same lipid mixture, except at the highest salt concentrations. As they point out, the deviation probably occurs when the Debye length is short because the double layer is extremely compressed, a significant proportion of the counterions are within the shear layer, and the zeta potential is consequentIy smaller than the surface potential. Haydon and Myers (1973) also determined the zeta potentials of charged vesicles and compared these potentials with the surface potentials predicted by Eq. (1). They used an elegant method to estimate the charge density at the surface of the membrane, measuring the change in interfacial tension produced by the adsorption of a charged molecule to a monolayer formed from a neutral lipid (GMO), then calculating the concentration of the absorbed surfactant by means of the
88
STUART MCLAUGHLIN
Gibbs equation. Kezdy (1972) may be consulted for a short, lucid introduction to the Gibbs equation and lipid monolayers. The values of t+bo predicted via Eq. (1) from the known charge density and salt concentration agreed remarkably well with the measured value of (. For example, when the solutions contained the anion dodecyl sulfate at a M , and the concentration of KCI was 0.1 M, concentration of 5 x the predicted value of Jio was - 55 mV, and the measured value of ( was also -55 mV. When the KCl concentration was 0.01, the predicted value of t+bo was - 76 mV, and the measured value of ( was - 73 mV. Similar agreement, except at the highest salt concentrations, was observed between the value of t+bo predicted from Eq. (1) and the experimentally determined value of ( when the vesicles were positively charged due to the adsorption of the cation, dodecyl trimethylammonium. McLaughlin and Harary (1976) also compared the zeta potential of phospholipid vesicles with the surface potential predicted by Eq. (1). All these zeta potential measurements confirm the adequacy of the Gouy equation at low charge densities, the region where discrete charge effects should be most important.
3. TRANSITORY CHANGES IN POTENTIAL ACROSS
A
BLACK
LIPID MEMBRANE MacDonald and Bangham (1972) used an interesting approach to deduce the change in surface potential produced by a change in salt concentration. When a bilayer membrane is formed from a mixturepf 95% phosphatidyl choline (PC?) and 5% phosphatidyl serine (PS=), the charge density is about - 1/1200 Az,When formed, at 20°C, in a solution containing a M concentration of monovalent electrolyte, Eq. (1)predicts that the surface potential should be t+bo = - 100 mV. The potential profile adjacent to the membrane, ignoring, for simplicity, dipole potentials, is illustrated in Fig. 6 (left). When a salt (e.g., KC1) is added to one side of the membrane, the double layer potential will be reduced in magnitude on that side, as illustrated in Fig. 6 (right). This change in surface potential manifests itself as a change in the potential AV between the two aqueous phases that the membrane separates. This potential difference can be measured with electrodes in the bulk solutions but will not be maintained indefinitely. The membrane now separates asymmetrical salt solutions, and AV will decay exponentially towards a steady-state diffusion potential, which will depend on the relative membrane permeabilities of the anion and cation. The resistance of the membrane is very high, about 108 C4 cmz, and the time constant for the decay is thus about 108 C4 * F = 102
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
89
FIG. 6. Illustration of the transitory change in potential, AV, which occurs when salt is added to one side of a membrane with a charge density of u = - 1/1200 Az (MacDonald and Bangham, 1972). The profiles are drawn to scale for a membrane of 50 A thickness and a AV of about 80 mV. See text for details.
sec. By measuring the potential difference between the two solutions a few seconds after adding the salt and extrapolating these measurements back to zero time, AVt+,,, MacDonald and Bangham (1972) were = AVt+ (Fig. 6). able to estimate the change in surface potential For the particular case under consideration, the change in surface po= 80 mV if the salt contential predicted by the Gouy equation is centration is increased from to lo-’ M on one side of the membrane. The actual measured change in potential, AVt+,, was -- 60 mV. Experiments were done on a variety of membranes formed in solutions of different salts. The observation that similar changes in potential occurred with different alkali metal cations (Na+, Li+, K+) is consistent with there being little binding of these ions to negatively charged membranes. As illustrated by one numerical example above, there was “approximate agreement” between the measured values of AVt-o and the potential changes predicted by the Gouy theory.
4. SURFACEPOTENTIALSTUDIES WITH MONOLAYERS Davies (Davies and Rideal, 1963; Fig. 2-18) tested the Gouy equation by forming a monolayer of a long hydrocarbon chain quaternary amine, then measuring the change in surface potential when NaCl was added to the aqueous subphase. Note that Eq. (1) predicts, for high positive values of I,$~, that
90
STUART MCLAUGHLIN
at,bo,,/a log C = -2.3kTle = -59 mV
at 25°C. Davis did in fact observe a - 59 mV change in the surface potential for a tenfold increase in the monovalent salt concentration over the entire experimental range investigated ( to lo-’ M ) . Good agreement with the theory has also been observed by MacDonald and Bangham (1972) in monolayer studies with phospholipids. They formed monolayers from mixtures of phosphatidic acid (PA-) and phosphatidyl choline (PC’) and measured the change in surface potential with an ionizinaair electrode as salt (KCl) was added to the subphase. When the monolayer consisted of 20%PA-, Eq. (1) predicts a change of surface potential of 111 mV when the salt concentration is increased from 1 to 100 mM at 20°C, whereas they observed a change of 95 mV in potential. When the monolayer consisted of 5% PA, the predicted and observed changes were 83 and 79 mV, respectively. There was “approximate agreement between monolayer and bilayer potentials and both of these with potential changes predicted by the Gouy theory.” There is also good agreement between direct surface potential measurements on monolayers and indirect “probe” measurements on bilayers. Figure 2 of Szabo et d .(1972) illustrates a titration curve of the zwitterionic lipid, phosphatidyl ethanolamine (PE’). The changes in surface potential predicted with the aid of Eq. (5)from conductance measurements made with the nonactin-K+ complex on black lipid membranes agree quite well with the change in surface potential measured above PE’ monolayers with an ionizing electrode (Papahadjopoulos, 1968).The curve was not interpreted theoretically, but if done so in terms of equations presented in Section 111, is consistent with an “intrinsic” or surface pK of about 10 for the tertiary m i n e and 1 for the phosphate group, both reasonable values. Haydon and Myers (1973) combined several different techniques to test the Gouy-Chapman theory. They estimated the number of moles of dodecyl trimethylammonium ions adsorbed to glycerol monooleate (GMOO) monolayers or bilayers by applying the Gibbs equation to measurements of the change in interfacial tension. The surface charge density was therefore known, and they calculated the surface potential predicted by double layer theory from Eq. (1).The Gouy equation was then tested by measuring the change in surface potential as a function of charge density. Both compensation potential measurements with a vibrating plate potentiometer in the air above a monolayer and conductance measurements with the nonactin-K+ complex in a black lipid membrane yielded similar results, which agreed almost exactly with the predictions of the Gouy equation. The agree-
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
91
ment was observed at three different ionic strengths and five different values of the charge density.
5. SHIFTS OF THE CONDUCTANCE-VOLTAGE CURVESOF PORE-FORMING ANTIBIOTICS The conductance G produced on a black lipid membrane by the antibiotic monazomycin, in contrast to that produced by a simple carrier like nonactin, depends markedly on the applied voltage V. Monazomycin almost certainly functions by forming pores in membranes, as illustrated in Fig. 7 (Muller and Finkelstein, 1972a,b; Muller and Andersen, 1975; Wanke, 1975; Moore and Neher, 1976). The molecular mechanism by which the channel, or precursor molecules that form the channel, respond(s) to a change in voltage is unknown, but is represented schematically in Fig. 7A as a voltmeter that responds to the transmembrane potential 4. The potential profile predicted by the Gouy-Chapman theory for a membrane formed from cholesterol and the negative lipid phosphatidyl glycerol in a low2M KC1 solution is illustrated in the upper portion of Fig. 7A. The phenomenological
w
Y M KCI
I I A
B
MEMBRANE POTENTIAL, V (mV)
C FIG. 7. The use of monazomycin, a voltage-dependent pore-forming antibiotic, as a “probe” of the surface potential (Muller and Finkelstein, 1972b). See text for details.
92
STUART MCLAUGHLIN
dependence of G, the membrane conductance, on the applied or measurable potential between the two bulk aqueous phases, V, is illustrated in Fig. 7C by the line designated [M$+] = 0. If MgZ+ is now added to one side of the membrane, Eq. (11A) from the theory of the diffuse double layer predicts that the divalent cation will screen the charges more effectively than the monovalent ion and will therefore reduce the magnitude of the surface potential on that side of the membrane (Fig. 7B). As illustrated in Fig. 7B, and discussed qualitatively by Chandler et a l . (1965) and Muller and Finkelstein (1972b) and quantitatively by Nelson et a l . (1975), a change in the surface potential at one interface will not affect, to any significant degree, the surface potential at the other interface. The field is assumed to be constant within the membrane (Neumcke and Lauger, 1970; d e Levie and Moreira, 1972; de Levie et al., 1972, 1974a; de Levie and Seidah, 1974) and the dipole potentials at the membrane-solution interface are ignored for simplicity. The solutions remain symmetrical with respect to the concentration of the permeant ion, 10+ M KC1; hence, there is little measurable potential developed between the two aqueous solutions. The potential difference between the two membrane-solution interfaces, however, has changed, and the “molecular voltmeter” in the membrane will sense this change. The molecules in the membrane have no way of distinguishing between a transmembrane potential that arises from a change in the applied potential and one that arises from a change in the surface potential! The effect of the change in surface potential is, therefore, to shift the conductance vs voltage curve of Fig. 7C along the voltage axis, and this shift should be a measure of the change in surface potential. The shift observed when 1.67 x M Mg2+ was added to one side of the membrane agreed with the change in surface potential predicted by the double layer theory [Eq. (llA)], assuming a charge density of 1/60 .&. Good agreement with the predictions of the theory were noted for all concentrations of divalent and monovalent ions examined.
‘
When a potential V is applied to a bilayer, essentially all the potential drops across the membrane, provided, of course, that the membrane resistance is much higher than the resistance ofthe solution in series with it (Walz et al., 1969). The capacitance ofa bilayer or biological membrane is of the order of lo+’ F/cm2. The capacitive charges will form a diffuse layer, but only when the Debye length becomes extremely large (at a salt concentration of less than 1 mM, for example, where the Debye length will be greater than 100 A) will the capacitance of the diffuse layer become significant with respect to the capacitance of the bilayer and a measurable fraction of the potential fall in the aqueous phase.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
93
6. SUMMARY Five independent experimental tests have demonstrated that the simplest form of the theory of the diffuse double layer is adequate to describe the electrostatic potential produced by charges at a membrane solution interface. Preliminary results obtained with amphiphilic spin labels also support the Gouy-Chapman analysis (Gaffney and Mich, 1976). Equation (1) is capable of predicting, to an accuracy of at worst 20%, the surface potentials observed over the entire range of charge densities (up to about 1/60 A2) and monovalent salt concentrations ( to 1 M ) that would normally be encountered in any biological system. It also is capable of predicting the ability of divalent cations to affect the double layer potential, although it is apparent that some binding of these ions (particularly Ca2+among the alkaline earth cations) may also occur. The claim by Haynes (1974) that the theory is seriously in error when applied to membrane solution interfaces is discussed in Appendix 11. Haydon and Hladky (1972) pointed out, quite correctly, that all the evidence which supports the Gouy-Chapman theory of the diffuse double layer is “circumstantial .” Thoreau pointed out that “some circumstantial evidence is very strong, as when you find a trout in the milk.”
111.
ADSORPTION OF CHARGED MOLECULES TO MEMBRANES
A. Theoretical Description of the Adsorption
A variety of pharmacologically significant molecules are amphipathic in nature and adsorb “hydrophobically” (Tanford, 1973) to phospholipid bilayer membranes. The cationic local anesthetics, for example, change significantly the surface potentials of artificial bilayer membranes (Bangham et d., 1965; McLaughlin, 1975) at the same concentration at which they block nerves. They have also been used to perturb the calcium-induced phase separation (Ohnishi and Ito, 1974; see also Galla and Sackmann, 1975) and temperatureinduced phase transition (Papahadjopoulos et al., 1975) of lipids i n bilayers, the concanavalin-A-induced clustering of intramembranous particles (Ryan et al., 1974; Poste et al ., 1975a,b), the virus-mediated fusion of cell membranes (Poste and Reeve, 1972) and the discharge of mucocysts in Tetrahymena (Satir, 1975), but their mechanism of action on all the biological membranes is unknown. Anions such as the
94
STUART MCLAUGHLIN
salicylates enhance the cationic and depress the anionic conductances ofNauanax neurons (Barker and Levitan, 1971) and black lipid membranes (McLaughlin, 1973) at identical concentrations, but the mechanisms by which these molecules affect the electrical properties of the nerve membrane is a matter for debate (Levitan and Barker, 1972a, McLaughlin, 1973). Fluorescent probes such as l-anilinonaphthalene-8-sulfonate (ANS) and 2-toludinonaphthalene-6sulfonate (TNS) adsorb hydrophobically to artificial bilayer membranes and change their fluorescence in response to a potential applied across the membrane (Conti and Malerba, 1972; for recent reviews see Azzi, 1975; Conti, 1975; Radda, 1975; Waggoner, 1976). These probes have been used to follow action potentials in neurons but would be of more value if the mechanism by which they responded to the change in membrane potential were known. Experimental investigations designed to reveal the mechanism by which the local anesthetics block nerves and the fluorescent probes respond to a change in membrane potential would obviously be facilitated if one could quantitatively describe the adsorption of these molecules to the bilayer portion of nerves and other biological membranes. As suc-
-7
-6
-5
-4
-3
-2
-1
log ,o [A-l (MI
FIG.8. The dependence of u,the number of anions adsorbed toa unit area of a neutral membrane, as a function of [A-1, the concentration of these anions in the bulk aqueous phase. The curve labeled “Langmuir” illustrates the prediction of Eq. ( 8 ) when surface potential effects are ignored. The curve labeled “Stern” illustrates the prediction of Eq. (8)when surface potential effects are taken into account by assuming that the charges are smeared uniformly over the surface of the membrane. The Stern equation is a combination of the Gouy, Boltzmann, and Langmuir relations. The values the maximum number of adsorbed anions per unit area, and K,the dissociation of urnax, constant, were assumed to b e 1/60 A* and lo+ M in both cases. For the Stern equation, the total concentration of monovalent electrolyte was assumed to b e lo-’ M and the temperature 25°C.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
95
cinctly phrased by Scatchard (1949),we wish to answer the following questions about the binding: “How many? How tightly? Where? Why?” Two extreme theoretical approaches to the problem are possible. One is to ignore the change in surface potential produced by the adsorption of the charged molecules. A variety of expressions have been used (e.g., Mohilner, 1966, p. 369; Aveyard and Haydon, 1973) to describe the adsorption of neutral and charged molecules to surfaces, one of the simplest expressions being the Langmuir adsorption isothenn: (+
=
(l/K)(cr”U
-
(+“-I,=,
(8)
where (+ is the number of molecules adsorbed to the membrane per is the maximum number of molecules adsorbed per unit area, urnax unit area, K ( M ) is a desorption or dissociation constant and [A-I,=, is the aqueous concentration of the adsorbing species at the membrane-solution interface, x = 0. When considering a fluid bilayer membrane, there are good theoretical reasons for preferring the use of the Volmer rather than the Langmuir adsorption isotherm, but this expression reduces to the same form as Eq. (8) when (+ << (+“‘ax, the experimental range discussed in this ~ e c t i o n We . ~ consider the adsorbing species to be an anion, A-, for the remainder of this section. If the membrane is initially neutral, and we ignore the change in surface potential produced by the adsorption of the anion, we can assume that the concentration at the membrane-solution interface, [A-I,=,, is equal to the concentration in the bulk aqueous phase, [A-I. The curve in Fig. 8 labeled “Langmuir” illustrates the dependence of (+ on [A-I
’
The Langmuir adsorption isotherm (e.g., Aveyard and Haydon, 1973; pp. 25-27) is derived on the assumption that the adsorption sites are spatially fixed: [A-]/KL = u
/ ( u L ~ ~- U )
(8)
For the hydrophobic adsorption of a molecule to a fluid membrane, it is perhaps more appropriate to use the Volmer isotherm (e.g., Aveyard and Haydon, 1973; pp. 22-24), which is derived on the assumption that the adsorbed molecules are not localized in space:
[A-]/K, = [u/(uFax- u)]. exp[u/(urax - u)] It is easy to show, however, that the Volmer isotherm reduces to the sanie form as the Langmuir isotherm when u << upax.By expanding the right-hand side of this equation in a power series and then taking the [ 1,1] Pad6 approximant (McLaughlin and Harary, 1976) we obtain [A-l/(Kv/2) = u/[(uFax/2) - ul which is of the same form as Eq. (8),provided we equateKV/2= KL anduFax/2 = uLmx.
96
STUART MCLAUGHLIN
predicted by Eq. (8) with this assumption. Note that the shape of the curve is identical to that of a titration curve for the binding of H+ to a weak acid or base. In this example, the value of K was arbitrarily chosen as M and the value of urnax as 1/60 A*, about the area of a phospholipid molecule in a bilayer membrane. The other extreme theoretical approach is to consider the charge on the membrane to be smeared uniformly over the surface and to take into account the surface potential produced by the adsorption of the charged A- species by relating the aqueous concentration at the surface of the membrane, [A-],=,, to the bulk aqueous concentration, [A-1, via the Boltzmann relation: [A-I,=, = [A-lexp (eh/kT)
(9)
If we assume that the membrane is initially neutral, the charge density u is equal to the surface concentration of adsorbed anions. The Gouy equation, Eq. (l),relates the surface potential $, the charge density u,and the salt concentration C. Stern combined Eqs. ( l ) , (8), and (9) (e.g., Bockris and Reddy, 1973; Aveyard and Haydon, 1973), but he also took into account the finite size of the adsorbing ions. We will ignore this aspect of the phenomenon but will, nevertheless, refer to the combination of Eqs. (l),(8),and (9) as a Stern equation. Equations (8)and (9)may be combined to eliminate [A-l,=,, and the resulting expression combined with Eq. (1)to eliminate either $, or u. When 9, is eliminated, one obtains an implicit expression for u in terms of [A-1 which may be solved by a standard iteration technique M, (McLaughlin and Harary, 1976). The result, for K = urnax = 1/60 Az and a salt concentration C = lo-' M , is shown in Fig. 8. Note that the Langmuir and Stern equations predict significantly different results. The former expression ignores completely the surface potential produced by the adsorbed ions, whereas the latter expression assumes that the charges are smeared uniformly over the surface of the membrane and ignores the discrete charge effect (Fig. 4). Experimental data should, therefore, lie somewhere between these two curves. A few other predictions of the Stern equation are now considered. If the binding constant K is changed, then both the curves labeled Langmuir and Stern in Fig. 8 merely shift along the abscissa. If the maxis alimum number of binding sites per unit area of membrane, urnax, lowed to vary, the Langmuir expression predicts the curves will all retain the same shape; the midpoint of the curves (a= d"=/2) will always occur at a concentration [A-1 = K . This can readily be seen by rewriting Eq. (8), assuming [A-I,=, = [A-I, as u = flax
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
97
[ A - ] / ( K + [A-I). The Stern equation predicts a quite different result. As illustrated in Fig. 9, the midpoints of the curves, designated b y the solid circles, shift back towards the value of the binding constant, K = M , as the maximum charge density urnax decreases. That is, the curves resemble more and more the Langmuir adsorption siotherm as f l a xdecreases. This is intuitively reasonable because, as dn" decreases, the magnitude of the surface potential and the deviation of the interfacial concentration, [ A - L , from its bulk value, [A-I, also decrease. The total concentration of monovalent electrolyte in the bulk aqueous phases [C in Eq. (l)] also affects the curves, as shown in Fig. 10. As the value of C decreases, the magnitude of the surface potential obtained for a given charge density increases [Eq. (l)]; this increases the deviation from the Langmuir expression. As the value of C increases, the curves approach the Langmuir expression asymptotically, but there remains a substantial difference between the predictions of the Langmuir (Fig. 8)and Stern (Fig. 10)expressions at salt concentration as high as 1 M . The midpoints (filled circles) of the curves pre-
FIG.9. The effect ofurnax, the maximum number of adsorbed anions per unit area, on the binding curves predicted by the Stern equation. The number of anions adsorbed to a unit area of a neutral membrane, u,is plotted against [A-1, the bulk aqueous concentration of the adsorbing anion. The Stern equation, in contrast to the Langmuir adsorption isotherm, predicts that the shape of the binding curve will change as the value of amax i s varied. The curves are drawn according to a combination of Eqs. (l), (8),and (9). The circles denote the midpoints of the curves. See text for details.
98
STUART MCIAUGHLIN
1.6 (u
0 x
h
1.2
\
0.8 b 0.4 0 -8
-7
-5
-6
log,,
-4
-3
-2
-I
[A-](M)
FIG.10. The effect of salt concentration on the binding curves predicted b y the Stern equation. The number ofanions adsorbed to a unit area of a neutral membrane, u, is plotted against [A-1, the bulk aqueous concentration of the anion, for different values of the total concentration of monovalent electrolyte, C. The circles denote the midof the curves, which were drawn according to the Stern equation, points (cr = umax/2) assuming that K = 10” M and T = 25°C. Note that a decrease in the salt concentration produces an increase in the deviation of the curves from the prediction of the Langmuir expression (Fig. 8).
dicted by the Stern equation shift to about an order of magnitude higher value of [A-1 for each tenfold decrease in salt concentration (Fig. 10). This reflects the prediction of Eq. (4); for a given charge density, a tenfold change in the salt concentration produces a 59 mV change in the surface potential at 25”C, which lowers the interfacial concentration of A- by one order of magnitude [Eq. (9)l.Other properties of the Stem equation will become apparent as various experimental results are interpreted in terms of this theory in Section II1,B. Although the Gouy equation was shown in Section II,B to provide an adequate description of the potential produced by charges at a membrane-solution interface, we cannot assume that the Stem equation will provide an equally good description of the adsorption of charged molecules. As noted by Aveyard and Haydon (1973): “The surface charge has been assumed to be smeared over the surface rather than, as it actually is, in the form of discrete ions and electrons.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
99
The diffuse layer in reality consists of the overlapping ionic atmosphere of each individual surface charge and the potential in a plane parallel to the surface fluctuates from place to place according to the degree of overlap of these atmospheres. The potentials in the Gouy-Chapman theory are thus average potentials. As far as the properties of the diffuse layer are concerned this averaging probably does not introduce much error, but for the specific adsorption of ions, as in the Stem theory, the assumption of smeared charge is thought to be less valid.” Many theoretical treatments of the discrete charge effect illustrated in Fig. 4 are available in the literature. The reader is referred, as a start, to Grahame (1958) and to Levine (1971). The possible relevance of discrete charge effects to the surface potential of biological membranes has been reviewed by Brown (1974), while Nelson and McQuarrie (1975)have recently presented an elegant and quite general procedure for calculating the electrostatic potential due to a fixed, discrete array of charges on a membrane. The objective of Section II,B is to examine the experimental evidence, which suggests that the Stem equation can describe the hydrophobic adsorption of a charged molecule to membranes in a concentration region where the discrete charge effect should be important. 8. Experimental Tests of the Stern Equation
Note that Eqs. (l),(8),and (9) can be combined to eliminate [A-],=, and either u or Jl0. One equation predicts how t,bo should vary as a function of the concentration of the adsorbing anion in the bulk aqueous phase, [A-1, and the other equation predicts how u will vary as a function of [A-I. To test whether the Stem equation is capable of describing the hydrophobic adsorption of charged molecules to membranes, one should measure independently both the charge density u and the surface potential t,bo as the concentration of A- is varied. McLaughlin and Harary (1976)measured the surface potential produced by the adsorption of the TNS to phospholipid membranes, then compared these data, and measurements of the charge density (Huang and Charlton, 1972) with the predictions of the Stem equation. Data are usually not presented in the manner illustrated in Figs. 8-10 but are analyzed in the form of either a Scatchard (1949) or reciprocal plot. The advantages of the Scatchard over the double reciprocal plot are discussed in standard texts (e.g., Edsall and Wyman, 1958, p. 617). In brief, this plot gives a more even, relative, weight to the different points on the curve.
100
STUART MCLAUGHLIN
In a Scatchard plot, the ordinate is u/[A-l and the abscissa is cr. If surface potential effects are ignored and it is assumed that LA-],=, = [A-I, then a Scatchard plot of the Langmuir adsorption isotherm, Eq. (8),yields a straight line with a slope of - 1/K and a y intercept of umax/K. The data obtained by Huang and Charlton (1972) were expressed in terms of F, the number of moles of TNS bound per mole of phospholipid, but this may be converted into a charge density u on the outer surface of the vesicle via the relation u = P/60 k (McLaughlin and Harary, 1976). Figure 11 shows a Scatchard plot of the data obtained by Huang and Charlton (1972) at 25°C. In the absence of other information, Huang and Charlton (1972) descril5ed their data, reasonably enough, in the simplest manner possible; the best fit they obtained to a straight line (e.g., a Langmuir isotherm) yielded K = 3 x M andamax= 1/550 A*. The Stern equation can also be used to describe the data. The computer was instructed to search over a “parameter” space (McLaughlin and Harary, 1976) to find the values of K and umax that would provide the best fit of the Stern equation to the data. For the data obtained at 25”C,these values were K = 2 x lop4M and umax = 1/70 hiz. The best fit of the Stern equation to the data is shown by the curve in Fig. 11. The best fit of the Langmuir expression to the data illustrated in Fig.
-
-
z
3-
B
e
-k I
2-
IS I
FIG.11. A "Scatchard' plot of the data of Huang and Charlton (1972) for the adsorption of TNS anions to vesicles formed from the zwitterionic lipid phosphatidyl choline, PC'.Vis the number of moles of TNS bound per mole of PC'. The curve is the best fit of the Stern equation to the data and was obtained for K = 2 x M and umx= 1/70 & . at T = 25°C. In a conventional Scatchard plot, the Langmuir equation is used to describe the data, which results in a straight line.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
101
11 would be a straight line. Both the Langmuir and Stem equations thus provide a reasonable fit to the experimental data, but there is a substantial difference between the interpretation of the data in the two cases. Both the Langmuir and Stem equations predict the same value for the y intercept, but the Langmuir isotherm predicts a slope of - l / K , whereas the limiting slope in the Stern equation can be shown to be (- 1/K)(1 + 273 umax/fl)at 25°C. In the specific case under consideration, it is apparent that the values of K and umax derived from a best fit of the data to the Stem equation differ by about an order of magnitude from the values derived from a best fit of the same data to a Langmuir expression. Which interpretation is correct? It could be argued that the Stern equation is capable of fitting these (cr,[A-]) data only because it contains two adjustable parameters. A critical test of the Stern equation for this molecule is thus reduced to the question: Does the adsorption of TNS produce a measurable surface potential Jlo, and can one describe these (Jlo,[A-]) data with the Stern equation using the values of K and crmm derived from a fit of the (uJA-1) data? The change in surface potential was estimated by two independent techniques. One approach was to observe the effect of TNS on the conductance of both anion and cation selective membranes. Data are presented in Fig. 12 for membranes formed from dioleoyl phosphatidy1 choline (PC*). The addition of TNS to the aqueous solution bathing the PC' membrane produced an increase in the conductance when the permeant species was a cation and a decrease in the conductance when the permeant species was an anion.8To the extent that the changes in conductance are equal in magnitude and opposite in direction and they are within error, they can be interpreted with the aid of Eqs. ( 5 )and (6)as being due to the adsorption of the TNS anion and a change in the electrostatic potential in the interior of the membrane, AJlo-. derived from the data of Fig. In Fig. 13, the average values of 12 and Eqs. (5) and (6),are plotted as circles. The solid line is the prediction of the Stern equation for the values of K and d"axwhich provided the best fit of the Stem equation to the (u,[A-]) data of Fig. 11.
* Similar results were obtained at pH 5 and 7, which proves that it is the anionic and not the neutral form ofTNS that is producing the change in conductance. Similar results (within a factor of two) were obtained on black lipid membranes formed from both a different zwitterionic lipid, phosphatidyl ethanolamine (PE'), and a neutral lipid, glycerol monooleate, (GMOO). This supports the suggestion of Huang and Charlton (1972) that TNS adsorbs to membranes by essentially hydrophobic forces.
102
STUART MCLAUGHLIN
-4.0 lOgl0
FJSI
(MI
FIG.12. The effects of TNS on the conductance of a cation (open circles) and anion (filled circles) selective membrane. The black lipid membranes were formed from d i e leoyl phosphatidyl choline in a lo-' M KCI, pH = 7.0, solution at 25°C. The heights of the vertical bars are twice the standard deviations of the measurements obtained from five separate experiments. The curves through the points have no theoretical significance.
The fit is satisfactory, but it could be argued that the agreement is fortuitous because these probes also respond to a change in dipole potential, whereas, as discussed in Section IV, we suspect that only a change in double layer potential will cause a change in the concentration of TNS in the aqueous phase at the membrane solution interface. To check that the change in potential measured by the probes was, in fact, due to the production of a diffuse double layer (i.e., that AJlo- = Jlo for the case of TNS), we measured the effect of TNS on the electrophoretic mobility of unsonicated mu1tilaminar vesicles formed from either egg or dioleoyl phosphatidyl choline, and then calculated the zeta potential from Eq. (7). The results are illustrated in Fig. 14. The solid lines are the predictions of the Stern equation for the three values of the salt concentration. There are no adjustable parameters in these curves, the values of K and amax being determined from the fit of the (cr,[A]) data presented in Fig. 11. While the data provide a good test of the Stern equation in the region where we suspect that discrete charge effects might be important (e.g., low charge densities and high ionic strengths), it should be pointed out that the data in Figs. 11-14 do not extend to sufficiently high concentrations to determine either amax or K with any great accuracy (McLaughlin and Harary, 1976).The quotient crmax/K, however, can be determined accurately.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
-5 0
log,,
-40
103
-3.0
mNSl ( M I
FIG.13. The change in surface potential, A&-, produced by TNS on a black lipid membrane formed from phosphatidyl choline. The open circles indicate the values deduced from the conductance measurements of Fig. 12 by means of Eqs. (5)and (6).The curve is the prediction of the Stern equation for the values of K and urn deduced from a M and 1/70 As,respectively. fit to the data of Fig. 11; 2 x
10-3~
/NaCI
Ksz M NaCl
0-1M NaCl
log,, [TNSI (MI
FIG.14. The zeta potentials of phosphatidylcholine vesicles measured as a function of the concentration of TNS in the bulk aqueous phase. The curves are the theoretical deduced from a fit to the predictions of the Stern equation, for the values of K and urnax data of Fig. 11; K = 2 x 10-4M, umax = 1/70 Az at T = 25°C. The heights of the vertical bars through the points are twice the standard errors of the mean for measurements made on 80 different vesicles in eight separate experiments.
104
STUART MCLAUGHLIN
In summary, the analysis confirms that the Stern equation provides a remarkably adequate description of the adsorption of TNS to bilayer membranes formed from phosphatidyl choline. An analysis of the data of Haydon and Myers (1973) indicates that this conclusion, rather than being restricted to one particular adsorbant and membrane, is more generally valid. Their data were obtained on monolayers and membranes formed from glycerol monooleate (GMOO), a neutral lipid, in the presence of either the anion dodecyl sulfate (SDS) or the cation dodecyl trimethylammonium (DTAB). The number of SDS anions adsorbed to a GMOO monolayer was inferred from interfacial tension measurements and the Gibbs equation, then converted into the potential predicted by the Gouy equation (Haydon and Myers, 1973).These predicted potentials are plotted as circles in Fig. 15A. The data obtained at the three different salt concentrations can all be fitted quite well with the Stern equation, assuming a value of mmax = 1/60 A2, and M . Values of the zeta potential a binding constant K = 2.5 x measured at lo-' and 1 t 2M salt concentrations are illustrated by crosses in Fig. 15A, whereas a datum was not obtained at M for technical reasons. As with TNS (Fig. 14), there is a good fit between the experimentally measured value of the zeta potential and the potential predicted by the Stern equation. There was, moreover, good agreement with both these measurements and the values of the change in surface potential calculated from either conductance measurements with the nonactin-K+ species [see Eq. (5)]or from compensation potential measurement on monolayers when the NaCl concentration was or loy3M . In the lo-' M electrolyte solution, however, there was a marked disagreement with the prediction of the Gouy equation, and the measured change in potential, as calculated from either the conductance or the compensation potential measurements. This is almost certainly due, as discussed by Haydon and Myers (1973), to SDS producing a change in the dipole as well as the double layer potential. One would expect the change in dipole potential to increase as the number of SDS molecules adsorbed to the membrane increased. As the number of molecules adsorbed at a given SDS concentration increases with salt concentration (Fig. lo), the change in dipole potential should be greatest in the lo-' M salt solution, in agreement with the observations. For the cation DTAB, the data are plotted in an analogous manner in Fig. 15C. The circles indicate the surface potentials predicted by the Gouy equation from experimentally determined values of the charge density. The zeta potential measurements obtained at lo-' and M salt concentrations are shown as crosses. The agreement with
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
1 05 c.10-3
M
- 100
- dO
C = 10'' M - 60
- 40
- 20
0
log,o [SDS] (MI
FIG. 15. (A) T h e zeta potentials (crosses) and snrface potentials predicted by the Gouy relation, Eq. ( l ) ,from a measurement of the charge density (circles) for the adsorption of the anion dodecyl sulfate to membranes formed from the neutral lipid glycerol monooleate. T h e predictions of the Stern equation are shown for K = 2.5 x M ,urnax = 1/60 A', T = 2VC, and the three indicated values of the monovalent salt concentration, C. Data from Haydon and Myers (1973). (B) Predictions of the Stern equation when the quotient f l a X / K is maintained constant, equal to the value used to fit the data presented in (A). T h e curves for both M and lo-' M salt concentrations refer to values ofurnax= 1/20,1/200, and 1/500A2in descending order. ( C )The z e t a p e tentials (crosses) and surface potentials predicted by the Gouy equation from a measurement ofthe charge density (circles) for the adsorption ofthe cation dodecyl triniethylammonium, DTAB, to a glycerol monooleate membrane. T h e curves illustrate the predictions of the Stern equation for K = 8 x M ,urnax = 1/60Az,T = 2VC, and the three indicated values of the monovalent salt concentration, C. Data from Haydon and Myers (1973).
106
STUART MCUUGHLIN
the theoretical predictions of the Stern equation (curves in Fig. 15C) is not quite as good for DTAB as for either TNS or SDS. The changes in surface potential predicted by the Gouy equation at the three different ionic strengths agree, however, with the values estimated from either the nonactin-K+ conductance or the compensation potential measurements on monolayers (Haydon and Myers, 1973). Thus, this compound, like TNS, does not appear to markedly change the dipole potential. Again, it should be stressed that the values obtained forumaxand for K can be changed within wide limits, provided that one maintains a constant quotient (McLaughlin and Harary, 1976). Figure 15B illustrates the curves predicted for amax = 1/20, 1/200, or 1/500 k when the quotient crmax/K = 6.6 x 102 M-l, the value taken for the curves illustrated in Fig. 15A. A comparison of Fig. 15A with Fig. 15B indicates that the data can be described quite well with amax = 1/20, less well with umax = 1/200, and not at all with umax = 1/500 k in the 1 0 - I M NaCl solution. The deviation between the curves obtained with different values of vmax at a given salt concentration (Fig. 15B) occurs most rapidly in the lo-' M salt solution because the number of ions adsorbed increases with salt concentration (Fig. 10). The experiments discussed above indicate that the theory is remarkably successful in describing the hydrophobic adsorption of ions to uncharged bilayer membranes. Equations ( l ) , (8),and (9) can easily be extended to take into account the hydrophobic adsorption of ions to charged membranes, an important consideration for those interested in biological membranes. [Amines, for example, are widely used to modify the electrical properties of nerve membranes. In a recent study of the adsorption of long-chain amines to bilayer membranes, Hyer et al. (1976) observed that the above equations could accurately describe the binding of the cations and explain the inactivation of the monazomycin-induced, voltage-dependent conductance.] There are, however, some serious contradictions to applying this simple theory to highly charged membranes. McLaughlin (1975) has varied the surface charge of membranes by mixing PG-, a negative lipid, in known quantities with PE', a zwitterionic lipid. The surface potential of these membranes was first estimated with the nonactin-K+ probe [Eq. (5)J.It agreed, within experimental error, with the values predicted by the Gouy equation, on the assumption that the relative composition of the black lipid membrane and membraneforming solution were identical, an assumption consistent with the observations of MacDonald and Bangham (1972). The adsorption of
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
107
the local anesthetic tetracaine, a cation that binds hydrophobically to these black lipid membranes, was studied as a function of the initial value of the surface potential. The local anesthetic did bind more strongly to the negatively charged membranes, but the additional binding agreed quantitatively with the predictions of the Stern equation only when the magnitude of the surface potential was less than 60 mV. When the surface potential was higher than this value, it had much less effect on the adsorption than the theory predicted? In conclusion, we note that most biological membranes contain about 10-20% anionic lipids (White, 1973), enough fixed charge to produce surface potentials of 34-60 mV in a lo-' M solution of monovalent ions [Eq. (l)].We may, therefore, with some confidence apply the Stern equation to the hydrophobic binding of charged molecules such as fluorescent probes (e.g., TNS), pH indicators (e.g., BTB), local anesthetics, salicylates, detergents, etc., to the bilayer portion of most biological membranes. One should, however, be more cautious about applying the Stern equation to the binding of ions to discrete sites on membranes. Although a very good agreement with the theory was obtained by Fromherz and Masters (1974) when they studied the binding of H+ to a negatively charged pH-sensitive dye imbedded in a positively charged monolayer, the agreement with the theory was less good when the monolayer was formed with a percentage of negatively charged lipids: More extensive measurements indicate that this discrepancy is reproducible and extends to measurements with mu1tivalent counterions (P. Fromherz, personal communication). The titration curves of the zwitterionic lipid, PE' (Szabo et al., 1972), may, on the other hand, be described reasonably well by the Stem equation, assuming intrinsic pK values of = 1 for the phosphate and = 10 for the primary amine groups (S. McLaughlin, unpublished). MacDonald et al. (1976) have also considered the titration curves of the negative lipid PS' in terms of these equations. We note finally that there is a formal analogy between Eq. (8),which describes an equilibrium
' Within the framework of the Gouy-Chapman theory of the diffuse double layer, this effect could be partially due to the charged portion of the adsorbing molecule being a few angstroms from the plane of the membrane. For low charge densities, the electrostatic potential falls off exponentially with distance from the membrane. For high charge densities, the potential falls off more rapidly with distance, as illustrated in Fig. 2. The charge on the adsorbing molecule would thus experience a lower fraction of the surface potential as the charge density of the membrane increased. Other factors are probably of importance.
108
STUART MCLAUGHLIN
system, and the Michaelis-Menten equations, which have been developed to describe enzyme kinetics. As stated by Edsall and Wyman (1958), “any advance in the theoretical analysis of one [system] can be readily carried over to the other, provided that due account is taken of the physical significance of the mathematical symbols employed.” Theuvenet and Borst-Pauwels (1976) have demonstrated that surface potential effects are indeed important in describing the kinetics of membrane-bound enzymes that translocate charged molecules and that the Michaelis-Menten equations must be appropriately modified. IV. MOLECULAR DIPOLES AT MEMBRANE-SOLUTION INTERFACES A. Experimental Estimates of the Dipole Potential
Consider a bilayer membrane formed from a neutral (e.g., GMOO) or zwitterionic (e.g., PC*, PE’) lipid. The electrostatic potential in the aqueous phase adjacent to the membrane is equal to the value of the potential in the bulk aqueous phase, which is defined as zero. The experimental basis for this statement is the fact that the electrophoretic mobility, and hence the zeta potential, of vesicles formed from these lipids is zero, irrespective of the concentration of indifferent electrolyte in the aqueous phase (Hanai et a1 ., 1965; Haydon and Myers, 1973; McLaughlin et al., 1975b). This does not, of course, imply that the potential in the interior of the membrane is also zero. The orientation of dipoles in (i) the water molecules adjacent to the membrane, (ii) the polar head group, and (iii) the ester linkages to the glycerol backbone could all produce a potential difference between the interior of the bilayer and the aqueous phase, as illustrated in Fig. 16. One method of estimating the dipole potential associated with lipids is to measure the change in surface potential when a monolayer of the lipid is spread at an air-water interface. Such measurements suggest that the potential in the interior of a membrane formed from the above lipids could be several hundred millivolts, positive with respect to the aqueous solutions (e.g., Hladky and Haydon, 1973; Hladky, 1974). One should be extremely cautious about extrapolating results obtained with monolayers to bilayers, but we may obtain independent information by measuring the conductance of a charged permeant species such as the nonactin-K+ complex, which will depend on this
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
109
FIG.16. Sketch of the potential profile thought to exist within a membrane formed from a neutral or zwitterionic lipid. The available evidence suggests that there is a positive dipole potential, of the order of 0.5 V, at the membrane-solution interface.
surface dipole potential in a predictable manner. To consider a specific example, the difference between the surface potentials of monolayers formed from GMOO and PC' is about - 120 mV (Hladky and Haydon, 1973). If the difference in the surface potentials in the interior of black lipid membranes formed from GMOO and PC' is also A$o- = - 120 mV, then Eq. (5) predicts that
G(lMo/GP,E= exp - (eA$o-/k7') = lo2 That is, the cation conductance of a black lipid membrane formed from GMOO should be two orders of magnitude higher than that of a membrane formed from PC, all other factors being equal-and it is. The surface potential measurements on monolayers, in conjunction with Eq. ( 5 ) , indicate that black lipid membranes formed from PE' should have a slightly lower conductance than membranes formed from PC' when exposed to nonactin, and this is indeed the case (Hladky, 1974; G. Szabo, personal communication). Although estimates from monolayer studies of the difference in dipole potential between bilayers formed from different lipids generally agree with the experimental results obtained with carriers, the origin of the dipole potential remains obscure. As suggested by Haydon and Hladky (1972), the potential could arise from oriented water molecules at the surface of the membrane. There is also evidence that a large portion of the potential could derive from the ester linkages in a normal phospholipid. Replacement of the ester by ether linkages produces changes of up to 200 mV in the surface potentials of condensed monolayers of various phosphatidyl choline molecules (Paltauf et al. 1971). We reiterate that the magnitude, as well as the origin, of the
110
STUART MCLAUGHLIN
electrostatic potential within a membrane (also called the inner or Galvani potential) is not only unknown, but is in principle unmeasurable (Guggenheim, 1929, 1930; Bockris and Reddy, 1970). Only differences in the dipole potential can be estimated with the “probe” measurements on bilayer membranes or with ionizing electrodes on monolayers. As noted by Andersen et al. (1976a), “monolayer surface potentials represent the difference between the potential at a clean air-water interface, and the potential after the monolayer is spread; we are concerned with the absolute value of the potential difference between the interior of the bilayer and the adjacent aqueous phases,” as illustrated in Fig. 16. One interesting attempt has been made to estimate this dipole potential using extrathermodynamic assumptions (Andersen and Fuchs, 1975). When the conductance produced by the lipid-soluble anion tetraphenyl boron is measured immediately after the application of a voltage pulse, it may be described, in the Nernst-Planck formalism, as G- = ( F 2 / d )u-k-C- exp (eJlo-/kT)
(104
where F is the Faraday constant, d the thickness of the membrane, uthe mobility of the ion within the membrane, k- the partition coefficient due to the difference in chemical potential, C- the aqueous concentration of the anion, and q0- the potential illustrated in Fig. 16 produced by the dipoles. The product of the last three terms in this equation is simply the equilibrium concentration of these ions in the membrane. Andersen and Fuchs (1975) or Szabo (1976) may be consulted for a more detailed discussion of this equation. Similarly, the conductance produced by the lipid-soluble cation, tetraphenylarsonium, is given by G+ = (F2/d) u+k+C+exp (- e&-/kT)
(lob)
The ratio of these two conductances, when measured at identical concentrations is G-/G+ = (u-/u+)(k-/k+) exp (2e Jlo-/kT)
(104
If the mobilities and partition coefficients of the two lipid-soluble ions were identical, this ratio would provide an estimate of the dipole potential, $o-. The mobilities are probably very similar, but the partition coefficients, unfortunately, are likely to be quite different (Krishnan and Friedman, 1971). After attempting to take this effect into account, Andersen and Fuchs, (1975) estimated the dipole potential of membranes formed from bacterial phosphatidyl ethanolamine to be + 310 mV, substantially lower than the values of about + 500 mV
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
111
estimated from surface potential measurements made on monolayers of phosphatidyl ethanolamine. As Andersen and Fuchs (1975) point out, quantitative agreement is not to be expected, and the exercise only allows one to conclude that the dipole potential is likely to be large and positive inside a bilayer membrane. B. Molecules that Change the Dipole Potential
As mentioned above, the nonactin-K+ conductance of membranes formed from the zwitterionic lipid phosphatidyl ethanolamine, PE" (Szabo et al., 1972; McLaughlin et al., 1971) and neutral lipids glycerol monooleate, GMOO (Hladky and Haydon, 1973) and glycerol dioleate, GDOO (Szabo et al., 1973) does not depend on the concentration of impermeant electrolyte (e.g., LiCl, CaCl,) in the aqueous phases. Equation (5) thus implies that the dipole potential is independent of salt concentration for these lipids, a conclusion consistent with the observation that the surface potential of a monolayer formed from GMOO is also independent of the electrolyte concentration in the aqueous subphase (Hladky and Haydon, 1973). This is an important point because in Section 11, we assumed that the dipole potential of charged lipids was independent of salt concentration. We now examine the degree to which the dipole potential extends into the aqueous phase. If the dipole potential did extend into the water, it should affect the hydrophobic absorption of anions such as TNS and cations such as the local anesthetics. These ions absorb equally well (McLaughlin, 1975; McLaughlin and Harary, 1976) to membranes with substantially different dipole potentials (PE', P C , and GMOO membranes), which indicates that the dipole potential extends very little, if at all, into the aqueous phase adjacent to the membrane. The dipole potential can change when either neutral molecules, such as cholesterol (Szabo et al., 1972; Szabo, 1975; Szabo, 1976), salicylamide (McLaughlin, 1973), and phloretin (Andersen et al., 1976a), or charged molecules, such as SDS (Haydon and Myers, 1973), intermix with the lipids forming the membrane. Figure 17 illustrates that the adsorption of salicylamide to a black lipid membrane formed from PE' can change the dipole potential. Salicylamide, at a concentration of 10 mM in the aqueous phase, increased the cation and decreased the anion conductance by about a factor of 20. This implies, from Eqs. (5) and (6), that it produced a change in dipole potential of about 75 mV. Identical effects were observed at pH 5 and 7 (pK salicylamide = 8.4), which confirms that it is indeed the neutral form of
112
STUART MCIAUGHLIN
(3
0
8
-1 -5
-4
-3
log ,o [SALICYLAMIDE] ( M)
-2
-a
-7
-6
-5
log,o [DTFB] (MI
FIG.17. The neutral form of salicylamide enhances the cation (left) and depresses the anion (right) conductance of a black lipid membrane formed from the zwitterionic lipid phosphatidyl ethanolamine. The aqueous solutions contain 10-lM KCl buffered to pH 7 with 5 mM potassium phosphate. (A) The solutions contain 5 x lO-'M nonactin. M salicylamide increases the conductance by about a factor of 20. (B) Note that Open circles: no salicylamide; closed circles: M salicylamide. Note that M salicylamide decreases the anion conductance by about a factor of 20. Salicylamide presumably affects the dipole potential. DTFB is a weak acid uncoupler (Cohen et al., 1976).
the drug that is producing the change in surface potential. Also consistent with this claim is the observation that 10 mM salicylamide does not produce any change in the zeta potential of vesicles at these concentrations ( S . McLaughlin, unpublished). At a concentration of 10 d, salicylamide also enhances the cation and depresses the anion conductances of neurons from Nuvanax (Levitan and Barker, 1972b). A parsimonious, albeit highly speculative interpretation of this result is that the conductance of this neuron can respond to a change in the dipole potential (McLaughlin, 1973). An even larger change in the dipole potential is observed with phloretin, the aglycone of phlorhizin (Andersen et al., 1976a). Phloretin and its analogs are of interest because they are potent modifiers of a number of biological transport systems. The molecules are all weak acids, but only the neutral form appears to affect the electrical properties of phospholipid bilayers. The effects of these molecules on the conductance of PE' membranes can all be explained in terms of a change in the dipole potential. Phloretin at a concentration of lo-* M, for example, produces a 103-fold increase in the cation and a 103-fold decrease in the anion conductance. Andersen et al. (1976a) concluded that these changes in bilayer conductance were due to a 180 mV
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
113
change in the dipole potential, as predicted by Eqs. (5) and (6),a conclusion consistent with the observation that phloretin also causes a 200 mV change in the surface potential of a monolayer formed from PE'. The picture is somewhat less clear in cholesterolcontaining PE' bilayers, where phloretin affects the cation much more than the anion conductance. It would appear that phloretin decreases the membrane viscosity as well as the magnitude of the dipole potential. A particularly puzzling observation is that phloretin produces no change in the dipole potential of cholesterolcontaining phospholipid monolayers. In general, there is a good correlation between surface potential changes observed on monolayers and on bilayers (e.g., MacDonald and Bangham, 1972; Szabo et al. 1972; Haydon and Myers, 1973; Szabo, 1976).
V.
ELECTROSTATlC "BOUNDARY" POTENTIALS
A. Theoretical Description of the Model
In Section 11, we considered how charges located at a membrane-solution interface-the negatively charged phosphate moiety in phosphatidyl glycerol (PG-), for example-produced a diffuse double layer in the aqueous phase immediately adjacent to the membrane. In Section 111, we considered how charged compounds, such as the salicylates, local anesthetics, and fluorescent probes, adsorbed to bilayers and changed both the charge density and double layer potential at the membrane-solution interface. All of the molecules we considered in Section I11 were amphipathic, i.e., one end of the molecule possessed a nonpolar region and was therefore hydrophobic, while the other end possessed a charge and was therefore hydrophilic. It was thus reasonable to assume that they adsorbed to the membrane solution interface in a parallel manner to the lipid molecules themselves, the charge being located at the interface and the nonpolar region in the interior of the membrane. We now consider how a compound that has hydrophobic groups arranged symmetrically around a charge (e.g., tetraphenylborate, dipicrylamine) will adsorb to a bilayer membrane. As discussed by Lauger and Neumcke (1973) and by Andersen and Fuchs (1975),a molecule such as tetraphenylborate will adsorb to a bilayer with its charged central region located in the interior of the membrane a few angstroms from the interface. Why? There are essentially two forces acting on a tetraphenylborate ion in a bilayer; the
114
STUART MCLAUGHLIN
B
A
I Aqueous phase
I
x =d
X’O
Membrane
Aqueous phase
x =d
1’0
Membrane
Aqueous phase
FIG. 18. (A) Schematic of the potential energy barrier to the movement of a tetraphenylborate anion across a PE’ black lipid membrane. (B) Schematic of the concentration profile of tetraphenylborate within the bilayer membrane. The full line is drawn for an applied potential of 0 mV, the stippled line for an applied potential difference of about 75 mV when boundary potentials are negligible. The exact location of the energy “wells” is unknown, but the available evidence suggests that they lie either adjacent to or within the region where the dipole potential changes rapidly, as seen in Fig. 16 (from Andersen and Fuchs, 1975).
“image charge” (see Footnote 3, p. 81) or ion dipole force, which repels the ion from the membrane and the “hydrophobic” force which attracts the ion into the hydrocarbon interior of the membrane. If the potential energy curve due to the repulsive force is plotted as a function of distance from the interface, it will be found to rise more slowly towards a maximum at the center of the membrane than the potential energy curve due to the attractive force falls with distance. By adding these two curves, it becomes apparent that there exist potential energy ‘‘wells’’ inside the membrane adjacent to the interfaces, as illustrated in Fig. 18A. We assume, for simplicity, that the ions adsorbed in these “wells” are smeared uniformly over a plane within the membrane, while their counterions, which cannot penetrate the membrane, lie in a parallel plane at the membrane solution interface. The two planes of charge are separated by a dielectric and thus act as a charged capacitor. I n two perceptive papers, which appear not to have come to the attention of other workers in the area, Markin et al. (1971) and Grigor’ev et al. (1972) note that the adsorption of ions will produce a drop in the electrostatic potential, Vco, across this outer region. The “boundary potential,” as they refer to it, will be given by
Vc = Q / C o
(11)
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
115
where Q is the charge density and C, is the specific capacitance of this coulombs/cm2 and outer region. If, for example, Q = - 2 x C, = 20 x 10-6F/cm2, then Vc = - 0.1 V. If the effective dielectric constant of the boundary region is E , = 2.5, a specific capacitance of C, = 20 x 10-6F/cm2 corresponds to a spacing of d = E ~ E ~ = / C1, bi between the two planes of charge; if the effective dielectric constant is 10, then d = 4 bi. The assumption that the counterions are confined to a plane parallel to the membrane is, of course, imprecise. They will, as illustrated in Fig. 19, form a diffuse double layer, and the potential drop in this layer may be calculated from Eq. (1).If Q = - 2 x 10+ coulombs/cm2 = 1 electronic charge per 800 biz and the monovalent salt concentration in the bulk aqueous phase is 1M , then the double layer potential in the aqueous phase at the surface of the membrane is predicted to be - 8.7 mV at 25°C. The important point to note in Fig. 19 is that for high salt concentrations the boundary potential Vcois much larger than the diffuse double layer potential produced in the aqueous phase. For the remainder of this section, we ignore, for simplicity, diffuse double layer effects. The boundary potential will manifest itself in a number of different ways. It will, for example, affect the adsorption of anions such as tetraphenylboron and dipicrylamine to phospholipid bilayer membranes. In terms of the simplest three-capacitor model that can be formulated (Markin et al., 1971; Andersen et al., 1976b), the number of adsorbed anions per unit area of membrane, Q, should vary as:
where K is a constant, [A-] is the concentration of the adsorbing anion in the bulk aqueous phase, and Vco the magnitude of the boundary potential.'O As discussed in the next section, we are testing the selfconsistency of this simple model by independently measuring both Q and Vco as a function of [A-I.
'" Equation (12) is consistent with the fact that both a Langmuir and a Volmer isotherm reduce, for low adsorption, to Henry's Law. We could interpret K = Fernax/& where Q""" i s the inaxinium nurnber ofcharges that can adsorb per unit area and KCIis a dissociation constant. Szabo (1976) may be consulted for an alternative but equivalent interpretation of K in terms of standard chemical potentials. Instead of using a smeared charge capacitor analogy, one should really calculate the time-average electrostatic potential experienced by an adsorbed ion due to all the other ions. Nelson and McQuarrie (1975) have made progress towards this goal with an elegant discrete charge calculation, but they assiuned that the charges were spatially fixed. Since the charges we are considering are almost certainly mobile, we really require an ensemble average over all possible configurations, a difficult calculation which has not, to the best of our knowledge, been attempted.
116
STUART MCLAUGHLIN
c
--
9
7
- ‘- - -I K O
----
I-
FIG. 19. Schematic of the electrostatatic potential adjacent to a bilayer membrane produced by the adsorption ofa lipid-soluble anion. The adsorbed anions produce a diffuse double layer in the aqueous phase Q and a “boundary” potential within the membrane phase V,. See text for details.
B. Experimental Tests of the Model
Bruner (1975) and Andersen and Fuchs (1975) measured, respectively, values of Q produced by the adsorption of dipicrylamine and tetraphenylboron to phospholipid bilayer membranes. They applied a sufficiently large voltage across the membrane to force essentially all the charge from one potential energy well to the other, then determined the charge by integrating the current over time. (The current crossing the membrane-solution interface in the few milliseconds required to make the measurement is negligible.) The data obtained by Bruner (1975) for dipicrylamine are illustrated in Fig. 20 as squares. The lines are the predictions of a combination of Eqs. (11) and (12), Q = FK[A-] exp(- eQ/C&T), drawn with the indicated values ofK and C,. A best fit was obtained with K = 0.48 cm and C, = 21 pF/cm2. This equation predicts that when Q is sufficiently small, V, = Q/C,
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
117
ing the value of K in the above equation by a factor x merely shifts the curve in Fig. 20 along the abscissa by an amount equal to log x , whereas multiplying the value of C, by a factor x shifts the curve along both the ordinate and the abscissa by log x.1 It could be argued that the qualitative agreement of theory and experiment illustrated in Fig. 20 is merely fortuitous and that the real explanation for the deviation from linearity lies elsewhere. Bruner (1975) suggests, for example, that there might be only one binding site for dipicrylamine/1000 Az of bilayer membrane, and Szabo (1976) suggests that the analogous apparent saturation observed with tetraphenylborate might be due merely to the formation of aggregates in the aqueous phase. We can, however, make an independent test of the hypothesis. We (Andersen et a2 ., 1976b) measured the change in electrostatic potential that occurred within the membrane by means of a “probe” molecule discussed in Section II1,B. If Eqs. (11)and (12) are combined to eliminate Q, one obtains Vco = (FK/C.)[A-]exp-eV,/KT. This equation predicts, when one substitutes in the values of K and C, obtained from a best fit of Bruner’s (1975) data illustrated in Fig. 20, that the value of the boundary potential Vco should be -80 mV when [A-] = loe6 M. Conductance measurements with the probe molecule, DTFB, also conducted on bilayer formed from PC’ in a lo-’ M NaCl solution, indicate that
-5 1
-6
i
-0
-9
-8
log,
-7
-6
-5
[A-I (MI
FIG. 20. The number of dipicrylamine anions adsorbed to a bilayer membrane formed from PC’ plotted as a function of the aqueous concentration of the adsorbing anion. The data points are from Bruner (1975). The curves are drawn according to a combination of Eqs. (11) and (12) in the text, with the indicated values of K, the interfacial partition coefficient, and C,, the specific capacitance of the outer region.
118
STUART MCLAUGHLIN
the potential on the inside of the membrane changes by - 70 mV at a M concentration of dipicrylamine. This provides strong evidence that the deviation from linearity observed in Fig. 20 is in fact due to some sort of electrostatic potential. The quantitative interpretation of these data is, however, complicated by the fact that double layer potentials are not negligible when the salt concentration is 10-l M . To further investigate the phenomenon of boundary potentials, we (Andersen et a1 ., 197613) are studying, by means of several independent techniques, the effect of tetraphenylborate on the electrical properties of bilayer membranes. The data that we have obtained from measurements of Q vs [A-1 for tetraphenylborate can, like the data that Bruner (1975) obtained with dipicrylamine, be described to a first approximation (see Footnote 10, p. 115),with the combination of Eqs. (11) and (12). For PE’ bilayers formed in 1 M NaCl, we estimate a value for the capacitance of the outer region of 50 x 1C6F/crn2and a value for K , as defined in Eq. (12), of 4 x lo-’ cm. Independent measurements of V, vs [A-1 with positively and negatively charged “probes” (Section I11,B) indicate that tetraphenylboron does indeed produce approximately the change in electrostatic potential within the membrane predicted by the model. Charge pulse measurements (e.g., Feldberg and Kissel, 1975) also yield data consistent with the three-capacitor model. Finally, we have investigated the changes in boundary potential that occur when a voltage is applied across the membrane. These effects are discussed quantitatively in Appendix 111. We merely note here that the measurable voltages one must apply across the bilayer to move a given fraction of adsorbed tetraphenylborate anions from one “well” to the other (Fig. 18B) increase dramatically, as predicted theoretically, when boundary potentials are present. We suspect that this phenomenon may be of physiological importance in understanding the movement of “gating particles” in electrically excitable biological membranes, as discussed in the next section.
VI.
BIOLOGICAL IMPUCATIONS
A. The Conductance-Voltage Curves of Excitable Membranes
In the early 1950s, Hodgkin and Huxley presented an elegant phenomenological description of the action potential. They suggested that there were specific regions in a nerve membrane, which are now believed to be channels (e.g., Hille, 1970; Armstrong, 1975), that al-
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
119
lowed sodium and potassium ions to penetrate the insulating bilayer framework. The conductances of these channels were shown to depend on voltage and time in a precisely defined manner. The theory withstood all challenges for a decade, but in the early 1960s some apparently contradictory experiments were reported on perfused squid axons. The axons were perfused with solutions in which the concentration of K+ was lowered, while osmolarity was kept constant with either Na+ or sucrose. When K+ was replaced with Na+, the magnitude of the measurable resting potential fell, or the cell “depolarized,” because the resting potential is due essentially to the diffusion of K+ out of the cell. When the magnitude of the resting potential fell below - 30 mV, the nerve would not transmit action potentials, in accordance with the predictions of the Hodgkin-Huxley theory (Baker et al., 1963). When K+ was replaced with sucrose, the magnitude of the resting potential again fell, but action potentials could still be elicited from depolarized membranes (Tasaki and Shimamura, 1962; Baker et al., 1962, 1964; Narahashi, 1963). It appeared that the channels were responding to some factor other than the measurable membrane potential. Chandler et al. (1965)put forth an interesting explanation, which required only a slight modification of the existing theory. They suggested (see Fig. 21) that the inner surface of the squid axon contains fixed charges of density u2= - 1/700 Az.When perfused with a 300 mM KC1 solution, these charges will produce, according to Eq. (l),a potential on the inner surface of the nerve membrane of Jlz = - 17 mV. Replacing K+ by Na+ will not affect Jlz; hence the change in 4, the transmembrane potential, will equal the change in V, the measurable or resting potential. Equation (1)predicts, however, that when K+ is replaced with sucrose the magnitude of Jlz will increase. The increase in the magnitude of J12 compensates for the decrease in the magnitude of V (Fig. 21), and the transmembrane potential 4 remains essentially unchanged. As the voltage-dependent channels within the membrane respond to 4 and not directly to V, the nerve remains excitable. More specifically, a change in the salt concentration produces a change in J12, which manifests itself as a shift in the conductance-voltage curves along the voltage axis in a manner analogous to that illustrated in Fig. 7 (Narahashi, 1963; Baker et al., 1964; Moore et al., 1964). Chandler et al. (1965) were able to deduce a value for the charge density on the inner surface of the squid axon by a quantitative consideration of these shifts. Gilbert and Ehrenstein (1969) deduced the value of the surface potential and charge density on the outer surface of the squid axon (Jl1 and u1,respectively, in Fig. 21) by studying the shifts in the conduc-
120
STUART MCLAUGHLIN
1
”OUTSIDE” 0
I
“INSIDE“
u2
FIG.21. The profile of electric potential in the vicinity of a phospholipid bilayer when the charged lipids in the membrane are allowed to distribute themselves between the two interfaces according to the Boltzmann relation. The membrane is assumed to b e homogeneous in the yz plane; the charge per unit area at the “outer” and “inner” surfaces, u1andu,, is assumed to be distributed uniformly over the surface, and any potential due to dipoles is ignored for simplicity. V is the resting potential, JI1 the double layer potential at the outer surface, JI, the double layer potential at the inner surface, and 4 the potential difference between the two membrane-solution interfaces. If the resting potential V is assumed to b e 75 mV, the concentration of monovalent ions in the bathing solution 0.5 M ,and the percentage of negative lipid in the bilayer 15%, the surface potentials are predicted to b e JI, = - 50 mV and JI, = - 15 mV. The potential profiles are drawn to scale for these values of V, JI1,J12, and a membrane thickness of 50 A (McLaughlin and Harary, 1974).
tance-voltage curves produced by an increase in the concentration of Ca2+in the external medium. Ca2+will reduce the magnitude of $1essentially by a screening mechanism,” increase the magnitude of 9 (Fig. 21), and thus shift the conductance-voltage curves along the voltage axis in a manner exactly analogous to that illustrated in Fig. 7. The potential at the outer surface of the squid axon was calculated to be JI1 = - (45-60) mV (Gilbert and Ehrenstein, 1969; Gilbert, 1971). The conclusion that mu1tivalent ions shift the conductance-voltage curves of squid axons by essentially a screening mechanism can be extended to other excitable membranes; divalent ions produce similar l1 The model predicts some binding of Cast to the nerve membrane, and the intrinsic association constant deduced by Gilbert and Ehrenstein (1969) for squid axons is in good agreement with the value deduced for the binding of Ca2+to negative phospholipids in artificial bilayer membranes (McLaughlin et al., 1971).See Footnote 5 for further discussion of the relation between the screening and binding effects observed with divalent ions.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
121
effects on Myxicola (Schauf, 1975; Begenisich, 1975), as well as toad (Brismar, 1973) and frog myelinated nerves (Hille et al., 1975a). Perhaps the strongest evidence that the shifts in the conductance-voltage curves are indeed due to an electrostatic effect comes from the experiments of Hille et al. (1975a). Equation (11A) in Appendix I predicts that a change in the concentration of monovalent ions in the external solution will produce little shift when the solution contains its normal complement of divalent ions. When the concentration of divalent ions is lowered, changes in concentration of monovalent ions do shift the conductance voltage curves as expected theoretically (Hille et al., 1975a). The above papers may be consulted for additional references and a more detailed discussion of the charge densities thought to exist adjacent to the sodium and potassium channels in a variety of excitable membranes. Rojas and Keynes (1975) discuss how the distribution of the “displacement” particles in nerves could be affected by the existence of asymmetrical surface potentials (Fig. 21). If their suggestion is correct, both the distribution and time constant vs voltage curves obtained for the displacement particles should shift along the voltage axis as the concentration of Ca2+in the external solution is varied. In an equally speculative vein, we note that the “boundary” potentials discussed in Section V could also be relevant to the interpretation of the “gating” or displacement currents observed in excitable membranes (e.g., Bezanilla and Armstrong, 1975; Rojas and Keynes, 1975; Nonner et al., 1975). The charged gating particles are thought to distribute themselves between two energy wells adjacent to the membrane solution interfaces, as illustrated in Fig. 18B. In the steady state, the ratio of the number of particles in the two wells will be given by a Boltzmann distribution. In terms of the model of Fig. 21, the potential difference between the two interfaces, the approximate location of the two wells, is given by = V - (t,b1-t,b2) and is a linear function of the applied voltage V. If, however, the gating charges produce boundary potentials, the potential difference between the two wells will not be a linear function of the applied potential and might be described by the equations developed in Appendix 111.The curves available in the literature that relate the distribution of gating particles in excitable membranes to the measurable membrane potential have been described by the Boltzmann relation, assuming noninteger values for the charge on the gating particle. One can fit these curves (e.g., Nonner et al,, 1975; Fig. 6) equally well with the equations developed in Appendix 111, assuming reasonable values for the outer capacitance and integer values for the charge on the gating particles. There is, however, little point in this exercise unless the decay of the
122
STUART MCLAUGHLIN
gating currents with time can be clearly shown to be a nonexponential process (Bezanilla and Armstrong, 1975). Theory predicts and experiments with tetraphenylborate on bilayer membranes confirm that the decay of current with time is nonexponential when boundary potentials are significant (Andersen et a1 ., 1976b). B. Distribution of Charged lipids in Biological Membranes
A variety of factors will influence the distribution of charged lipids in a biological membrane. If the rate of transverse diffusion (flip-flop) of lipids is faster than the rate of biosynthetic turnover, a membrane potential will cause an unequal distribution of lipids between the two constituent monolayers of the bilayer (McLaughlin and Harary, 1974). The gradient of the potential difference between the two interfaces (Fig, 21) acts as a driving force to move negatively charged lipids from the inner to the outer monolayer. The distribution of charged lipids, and hence surface potentials [Eq. (l)]illustrated in Fig. 21, reflects the equilibrium situation for a membrane comprised of 15%negative lipids, a resting potential V of 75 mV, and a concentration of monovalent salt of 0.5 M . The outer surface potential is predicted to be - 50 mV, the inner surface potential - 15 mV, in excellent agreement with the values deduced for the squid axon from physiological measurements. It must be admitted that the flip-flop process is not rapid in artificial bilayer membranes. The half-time for the transverse movement of spin-labeled phosphatidyl choline in an artificial bilayer is 6.5 hours (Kornberg and McConnell, 1971), but the half-time for unlabeled lipids is much longer, (Johnson et al., 1975; Rothman and Dawidowicz, 1975; Roseman et al., 1975; Hall and Latorre, 1976). If there are, however, a few regions in a biological membrane where flip-flop can occur rapidly, the charged lipids that flip-flop will undergo lateral diffusion. The lateral diffusion constant is D = lo-* cm2/sec (Edidin, 1974); so, in 1 sec the lipids will diffuse d%%= 104 A. The existence of such regions has not been demonstrated. Nor has it been demonstrated that the charges in the vicinity of the “channels” in nerves arise from lipids. All one can state at this time is that flip-flop of the charges in the vicinity of a K+ channel in a squid axon does not occur within Q hour (Ehrenstein et al., 1975). If the rate of lipid flip-flop is faster than the rate of turnover, the geometry of the cell or organelle will also govern the distribution of charged lipids. In a membrane with a low radius of curvature, charged
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
123
lipids will tend to move to the outer monolayer to minimize the electrical free energy (Israelachvili, 1973a). Israelachvili (1973a) approximated the outer and inner diffuse double layers adjacent to the liposome membrane by spherical capacitors, a model that is certainly qualitatively correct, as illustrated in Fig. 1D. Mille and Vanderkooi (1976) have, however, recently made extensive numerical calculations of the electrostatic potential adjacent to a charged spherical particle as a function of surface charge density, concentration of the added salt, and particle size. These calculations should be of interest to both experimentalists and theoreticians considering electrostatic phenomena at surfaces with a low radius of curvature. The geometry of the individual lipid molecules is also likely to be of importance in determining their distribution between the two interfaces when the radius of curvature is low (Litman, 1974; Berden et al., 1974). The theoretical considerations of Israelachvili and Mitchell (1975), based on an analysis of solid angles, provide an important start in approaching this problem. The above considerations were all based on the assumption that flip-flop occurs more rapidly than turnover in a biological membrane, a highly questionable assumption. There is good evidence, however, that lipids in both artificial and natural membranes undergo rapid lateral diffusion in the plane of the membrane. A consideration of elementary electrostatics (e.g., the discussion on pp. 662-663 and Fig. 2 of Israelachvili, 1973a) indicates that, in the absence of other specific interactions, charged lipids in the outer monolayer of a biological membrane will migrate to regions where the radius of curvature is small (e.g., the edges of retinal rod outer segment discs, the cristae of mitochondria, and microvilli). Some proteins are known to immobilize adjacent lipids (Jost et al., 1973). In addition to inducing the formation of such “boundary layers,” which are quite possibly lipid-specific (Jost and Griffith, 1976), a charged macromolecule in a membrane will also affect the distribution of charged lipids in accordance with the Poisson-Boltzmaim relation, Eq. (4A) in Appendix I. That is, a positively charged macromolecule imbedded in a membrane will utilize negatively charged lipids, as well as anions in the aqueous solution, as counterions. Small ions that bind to lipids can also influence their distribution. Calcium, for example, is capable of inducing phase transitions in membranes formed from one lipid (Tiauble and Eibl, 1974; MacDonald et a2 ., 1976)and phase separation in a membrane comprised of a mixture of lipids (Ohnishi and Ito, 1974; Jacobson and Papahadjopoulos, 1975).
124
STUART MCLAUGHLIN
C. Permeation of Charged Molecules through Membranes
When most of the KCl inside a squid axon is replaced with sucrose, the magnitude of surface potential increases and the voltage sensed by the gating mechanism in the voltage dependent “channels” is changed (Chandler et al., 1965). One might also expect the [K+l/[Cl-] ratio in the vicinity of the channel to increase, enhancing the relative permeability of the membrane for potassium ions over chloride ions. This is indeed the case (Chandler et al., 1965), but there are several difficulties in interpreting this result quantitatively, one being that the channels and voltage sensors are probably spatially separate entities. Experiments by Henderson et al. (1974) and Hille et al. (1975b) with tetrodotoxin (TTX) and saxitoxin (STX) provide perhaps the best evidence that the magnitude of the surface potential at the outer mouth of the sodium channel is substantially less than the magnitude of the surface potential at the voltage sensor. Both TTX and STX block sodium channels when added to the extracellular fluid, probably by acting as “plugs” at the channel mouth (Hille, 1975). Because TTX is a monovalent and STX a divalent cation, their relative effectiveness should change in the manner predicted by the Boltzmann relation when the surface potential is changed by increasing the [Ca2+].The relative effectiveness of TTX vs STX did increase with an increase in [Ca2+],but the estimated change in surface potential was significantly less than the change in surface potential sensed by the voltage-dependent “h” and “m” gates of the sodium channel. This aspect of the effect of surface potential on the permeation of ions through channels remains to be elucidated by further experimentation on single channels in both biological (Neher and Sakmann, 1976) and artificial membranes. Finkelstein and Holz (1973) discuss the effect of surface potentials on the conductance of pores formed by the antibiotic nystatin in artificial black lipid membranes. Nystatin pores, like sodium channels, appear to sense only a fraction of the electrostatic potential that exists at the surface of the bilayer in which they are situated. The conductance produced by a charged molecule (e.g., a nonactin-K+ complex) which passes through the bilayer portion of a biological membrane should, on the other hand, depend on the surface potential in exactly the manner predicted by Eqs. (5)and (1).Indeed, the conductances produced by certain of these molecules were used on artificial bilayers as “probes” of the surface potential. Certain weak acids, the classic example being 2,4-dinitrophenol (DNP), act as spe-
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
125
cific carriers of H+ ions across bilayer membranes, and their effectiveness as carriers depends on the surface potential via Eq. (6) (Hopfer et al., 1970; McLaughlin, 1972; Foster and McLaughlin, 1974). These weak acids uncouple oxidative and photosynthetic phosphorylation in mitochondria, bacteria, and chloroplasts, probably by virtue of their ability to dissipate an electrochemical gradient of H+ (Mitchell, 1966; Greville, 1969; Skulachev, 1971; Harold, 1972). Many bacteria (e.g., Micrococcus 1ysodeikticus) contain a much higher percentage (up to 80%) of negatively charged phospholipids than do mitochondria (20%). Equation (1) predicts that the surface potentials of these bacterial membranes will be more negative than those of mitochondria1 membranes. If all other factors are equal, and if Mitchell is correct about the way in which uncouplers function, weak acids (e.g., DNP) should be less effective and weak bases (e.g., local anesthetics) more effective in uncoupling these bacteria than mitochondria. A claim has been made that the surface potential will modify the uptake of weak acids (e.g., benzoic acid) by cells. A bilayer is essentially impermeable to the anionic form of this weak acid (S. McLaughlin, unpublished observation). The neutral form of the weak acid has a partition coefficient into oil of k = 5 (Leo et al., 1971),the thickness of the membrane is d = 50 A, and the diffusion constant in a fluid membrane must be of the order of 0 = 10%m2/sec, so the permeability is P = kD/d = 0.1 cm/sec, over a million times higher than the permeability of the charged form. The uptake of this and other weak acids by biological cells is indeed consistent with the neutral form being the only permeant species; as discussed by Rubery and Sheldrake (1973), the uptake vs pH curves resemble titration curves with maximum uptake at low values of the pH. A detailed examination of the uptake of benzoic acid by Proteus vulgaris reveals, however, that the curve is broadened and appears to be displaced to about 1 pH unit above the pK. This displacement is quite general. Rubery and Sheldrake (1973) suggest that this effect could be due to the possession by the membranes of a negative surface potential and the pH adjacent to the plasma membrane being “lower than the bulk pH, resulting in an increase in the apparent pK.” This argument is incorrect. It has long been known that the effective pK of a molecule adsorbed to a membrane will be affected by the surface potential (Hartley and Roe, 1940; Montal and Gitler, 1973; Fromherz, 1973; Fromhelz and Masters, 1974), but the concentration of a neutral molecule in the aqueous phase adjacent to a membrane will not be so affected, and it is this concentration which determines the rate of permeation. The effect
126
STUART MCLAUGHLIN
discussed by Rubery and Sheldrake (1973) is probably due to the existence of unstirred layers adjacent to the membranes.12 D. Fluorescent Probes
Much evidence has now accumulated that fluorescent probes such as ANS and TNS adsorb to the bilayer component of biological mem-
branes and that the surface potential of the membrane exerts a controlling influence on this adsorption (Azzi, 1973; Feinstein and Felsenfeld, 1975).As discussed in Section 111, the adsorption of the probes to the bilayer component of the membrane is due mainly to the “hydrophobic” or entropic forces that tend to sequester the aromatic portion of these molecules away from water. X-ray evidence (Lesslauer e t al., 1972) confirms that, for phosphatidyl choline bilayers, the sulfonate group of ANS lies in the plane of the polar head groups and the aromatic residue protrudes a short distance into the fatty acid side chain layer. As demonstrated in Section 111, the adsorption of a fluorescent probe, such as TNS, to an artificial bilayer membrane depends on the surface potential. It is not an unreasonable extrapolation to assume that the adsorption of these probes to the bilayer component of a biological membrane will also depend on the surface potential and that this adsorption will be described, at least qualitatively, by the Stern equation. Most biological membranes have a negative surface potential because of a preponderance of negative lipids. Any factor that decreases the magnitude of this potential will tend to increase the adsorption of the anionic probes. The cationic local anesthetics, for example, adsorb hydrophobically to negatively charged bilayer membranes (Bangham et al., 1965; McLaughlin, 1975) and reduce the magThe resistance of the membrane is in series with the resistance of the adjacent aqueous “unstirred layers.” For a molecule with a membrane permeability of 0.1 cm/sec, an aqueous unstirredlayer of only l O O O A will provide an equivalent resistance to that of the membrane. (For an introduction to unstirred layers and references, see McLaughlin and Eisenberg, 1975.)As pointed out by Gutknecht and Tosteson (1973), the flux of the neutral forin of a weak acid through a membrane will depend not only on its concentration, but also on the concentration of the anion, A-. In brief, the reaction H+ + A- = HA will be at equilibrium throughout the unstirred layer; as the concentration of HA tends to fall near the membrane, A- can combine with H+, provided the system is well buffered, and thus facilitate the flux of HA through the unstirred layer towards the cell. This effect would explain why the uptake of a weak acid by a cell does not fall as rapidly as predicted when the pH is increased. The concentration of HA falls, but the concentration ofA- rises, and this facilitates the flux of HA through the unstirred layers.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
127
nitude of the negative surface potential. This almost certainly explains why they enhance the binding of ANS to myelin (Feinstein and Felsenfeld, 1975) and other biological membranes (Vanderkooi and Mar1970). Other factors that reduce the magtinosi, 1971; Feinstein et a,?,, nitude of the negative surface potential include H+ ions, which can neutralize the surface charges and reduce the magnitude of u in Eq. (l),and cations, particularly mu1tivalent cations, which increase the value of C in Eq. (1). H+ ions and multivalent cations do enhance the binding of fluorescent probes to both artificial (e.g., Vanderkooi and Martinosi, 1969; McLaughlin et al., 1971; Flanagan and Hesketh, 1973; Haynes and Staerk, 1974; Haynes, 1974) and biological membranes (e.g., Feinstein and Felsenfeld, 1975).The binding of these anionic probes to biological membranes will also modify the surface potential. The production of a more negative surface potential on the exterior of the membrane might be expected to directly influence the permeability of the membrane to anions and cations. The adsorption of molecules to the outer surface of a membrane can produce other, less direct, effects. One of the more interesting of these effects is the “bilayer couple” mechanism discussed by Sheetz and Singer (1974). Fortes and Ellory (1975) have speculated that it might be the bilayer couple effect, the expansion of the outer half of the erythrocyte membrane, rather than the direct production of a more negative surface potential, which depresses the permeability of erythrocyte membranes to anions when ANS is added to the bathing medium. E. Photochemical Reactions
Trissl (1975), in an interesting paper that extends the technique of MacDonald and Bangham (1972) discussed in Section 11, points out that photochemical reactions may be studied by making surface potential measurements. If pigment molecules, either chlorophylls or carotenoids for example, are located at one interface of an artificial bilayer membrane, a photochemical reaction of the pigment with a substrate dissolved in the aqueous phase should lead to a change in the charge of both the pigment and the substrate. If the pigment remains bound to the membrane and the substrate remains in the aqueous phase, the charge density of the membrane will change. A change in the charge density produces a change in the surface potential, as predicted by Eq. (1). This will lead to a transitory change in the measurable potential between the two aqueous phases separated by the membrane, in a manner analogous to that illustrated in Fig. 6.
128
STUART MCLAUGHLIN
F. Osmolarity of Solutions in Small Vesicles
Many subcellular organelles have a high surface-to-volume ratio. A synaptic vesicle from a frog or snake nerve terminal, for example, is approximately spherical and has an internal radius of about 200 A. A synaptic vesicle contains about 15%negative lipids (White, 1973). If we assume that the charged lipids are evenly distributed between the two surfaces and that a lipid occupies an area of 60 A2,it follows that the inner monolayer bears (0.15)4 r (200)*/60 1200 fixed negative charges. The vesicle must therefore contain 1200 cations to act as counterions to these fixed charges. I n spite of claims to the contrary in the biological literature, these counterions must be considered as part of the thermodynamically defined membrane phase and will not induce an osmotic flow of water across the membrane. A simple thermodynamic argument demonstrates this point. Consider, as shown in Fig. 1, the semi-infinite aqueous phase adjacent to a charged surface. At equilibrium, the chemical potential of water in this aqueous phase will be a constant equal to its value in the bulk (x + m).).Now consider the charged surface to be a bilayer membrane, one side of the membrane being comprised of charged lipids, the other of neutral lipids. If the bulk aqueous phases are identical on both sides of the membrane, the chemical potentials of water will also be identical, and there will be no flow of water through the membrane. Such a flow would violate the second law of thermodynamics. Thus, the excess of counterions and deficit of coions that occur in the diffuse double layer adjacent to the charged lipids must be considered as part of the thermodynamically defined membrane phase. Within this phase they d o exert an osmotic pressure, ~ ( x ) in , excess of the bulk osmotic pressure, r ( m ) , by an amount: v ( x ) - r ( m ) = kTn(w){exp-[xeJl(x)/kTJ + exp[zeJl(x)/kTI - 2). This effect must, however, be exactly counterbalanced by some other factor (A. Mauro, 1962, personal communication) such as a pressure induced by the electric field (e.g., Frank, 1955; Rice and Nagasawa, 1961). Could this phenomenon have any physiological significance? If a synaptic vesicle of inner radius r were formed from neutral molecules and the fluid it contained were in osmotic equilibrium with the extracellular fluid of concentration C = 0.15 M, the vesicle could contain, at most, 4rr3 CN = 3000 cations and anions. The vesicle could contain, therefore, only 3000 acetylcholine molecules. Some evidence (Kuffler and Yoshikami, 1975) indicates, however, that a synaptic ves-
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
129
icle might contain more than this number of acetylcholine molecules. The existence of negative lipids within the membrane would allow the accumulation of an additional 1200 acetylcholine molecules within the vesicle. Charged lipids might also play a role in sequestering Ca2+in the sarcoplasmic reticulum and retinal rod outer segment discs, where the surface-to-volume ratio is also very high.
0. Other Effects
A surface potential could affect, in a variety of ways, the activity of an enzyme located in a membrane. Dawson (1968) and Goldhammer et al. (1975) review the evidence that surface charge is a factor in determining the susceptibility of lipids to phospholipase-catalyzed hydrolysis. Negative surface potentials appear to affect both the activity of the phosphatidylcholine exchange protein (Wirtz, 1974) and the active uptake of ions by plant cells (Theuvenet and Borst-Pauwels, 1976). Gingell(1971),in a provocative essay, suggests that a change in surface potential can initiate pinocytosis in free-living amebae. Muller and Finkelstein (1974) have developed a simple, quantitative and, in this reviewer’s opinion, very reasonable model to explain the inhibition of transmitter release by Mg2+ at the frog neuromuscular junction. Mg3+ was postulated to decrease the magnitude of the surface potential on the outside of the presynaptic membrane by a nonspecific “screening” effect in the diffuse double layer. This will reduce the interfacial concentration of Ca2+ and thus inhibit transmitter release. Hall and Simon (1976) discuss a possible mechanism whereby the entry of calcium ions into a nerve terminal could alter the surface charge on the presynaptic membrane and lead to a fusion of synaptic vesicles with this membrane. The removal of calcium ions could lead to the budding-off of vesicles. The surface potential, in conjunction with the van der Waals force (Israelachvili, 1973b) is obviously of great importance in determining the interaction between membranes. Evidence for long-range electrostatic repulsion between biological membranes exists for both erythrocytes (Brooks and Seaman, 1973; Jan and Chien, 1973; Luner et al., 1975) and HeLa cells (Deman and Bruyneel, 1974). As discussed by Parsegian (1974) and Parsegian and Gingell(1973), the juxtaposition of such cells will, conversely, change the surface potentials adjacent to the membranes.
130
STUART MCLAUGHLIN
When monitored by means of electrophoretic mobility measurements, the cell surface charge of normal but not transformed 3T3 cells shows a decrease when the cells are stimulated to undergo cell division (Adam and Adam, 1975).The effect of concanaval in A on the surface charge density of normal and transformed cells is discussed by Milito and Todd (1976). The cell surface charge of undifferentiated neuroblastoma cells is about 30%more negative than that of differentiated cells from the same culture (Elul et al., 1975). The significance, if any, of these observations is unknown. The examples cited in this section are obviously speculative in nature and are discussed merely to indicate the variety of biological phenomena that could be affected by surface potentials. As techniques develop to change the charge density and surface potential of biological membranes in a controlled manner, it will be possible to test these speculations more rigorously. Bakker et al. (1975),for example, varied the charge density and surface potential of the bilayer portion of mitochondrial membranes, then examined the binding of uncouplers of oxidative phosphorylation to the membranes. Although the experiments can be critized on technical grounds,I3 they do illustrate how an understanding of the existence of surface charges can be used to test a particular hypothesis. The results provide indirect support for the Mitchell hypothesis, which predicts that uncouplers act by carrying H+ ions across the bilayer component of mitochondria1 membranes rather than by acting on specific proteins (Hanstein, 1976). As the detailed kinetic mechanism of action of the uncouplers becomes known on artificial bilayers (Cohen et al., 1976), the comparison with biological membranes of varying charge densities will become more fruitful.
APPENDIX I
The Poisson equation for a planar surface is given by
dW)/dX2 =
-P(X)/ErGI
l3 The results presented in Fig. 3 of Bakker et al. (1975) demonstrate an apparent increase in the binding of weak acid uncouplers to phospholipids in vesicles as the pH is lowered. This result, however, is probably due to the permeability of multilaminar, nonsonicated liposomes to the neutral but not to the charged form of the weak acids. As the pH i s lowered, more uncoupler penetrates to the inner layers of the liposomes, and more surface area is available for binding both the neutral and charged forms.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
131
where +(x) is the electrostatic potential at a distance x from the membrane, +(m) = 0, p ( x ) is the charge density at the distance x, el. is the dielectric constant of the solution, and e0 is the permittivity of free space. We assume that the dielectric constant does not change until the surface of the membrane is reached. The charge density in the aqueous phase at any distance x from the membrane is, by definition, p ( x ) = ze[n+(x) - n-(x)I
(2.4)
where z is the valence, e is the electronic charge, and n+(x)and n-(x) are the numbers of cations and anions per unit volume. We assume that the electrolyte is symmetrical and that ions are point charges. The Bolkmann relation predicts that: n+(x)= n ( w ) exp - [ze+(x)/kT] n-(x) = n ( w ) exp [ze+(x)/kTl
(3.4)
The Bolkmann relations follow from the equilibrium condition that the electrochemical potential of an ion must be the same at all distances x from the membrane, provided we assume that the charges are “smeared” uniformly over the surface of the membrane and that neither the standard chemical potential nor the activity coefficient varies with distance. The combination of Eqs. (1A)-(3A) results in the Poisson-Boltzmann relation: d2+(x)/dx2= [ 2 z e n ( w ) / ~ sinh ~ ~ ~ (zet,h(x)/kT) I
(4.4)
It must be admitted that there is a fundamental inconsistency in the use of Eq. (4A), the nonlinear Poisson-Bolkmann relation. The inconsistency was apparently first discussed by Fowler (1927, 1936, pp. 261-274) and arises, in brief, because the electrostatic potential, +(x), appearing in the right-hand side of Eq. (4A) is the potential of the average force acting on an ion at a distance x from the membrane, whereas the electrostatic potential appearing in the left-hand side of Eq. (4A) is the average potential at a distance x from the membrane. These two potentials are not identical: among other reasons, the ion creates its own atmosphere (e.g., Loeb, 1951; Kirkwood and Poirier, 1954; Levine and Bell, 1966; Olivares and McQuarrie, 1976). Onsager (1933) and Kirkwood (1934) may be consulted for an extended discussion of this fundamental inconsistency. The appropriate boundary conditions are that: +(x) = $07
=
d+(x)/dx = +(x) = 0, x = 03
(5.4)
The solution to Eq. (4A) that satisfies these boundary conditions is
132
STUART MCIAUGHUN
1n 1 + (Y exp ( - K X ) 1 - a exp ( - K X )
=
@A)
where a!=
exp (zeJlo/2kT) - 1 exp (zeJlo/2kT) + 1
and
The constant 1 / is ~ defined as the Debye length. The potentials $(x) predicted by Eq. (6A) are plotted as a function of distance from the membrane in Figs. 2 and 3, which illustrate, respectively, the dependence of potential profiles on initial surface potential (or charge density) and on salt concentration. Note that for small potentials we may linearize the exponent, exp (ze3r,/2kT) = 1 + zeJlo/2kT, and Eq. (6A) reduces to $ = Jlo exp -
(4
(9A) Thus, for small potentials, +(x) falls to l/e its value at the membrane solution interface in a distance of 1 / ~ . ' To relate the charge density on the surface of the membrane, u,to the surface potential u=
-
I
$o,
we note that electroneutrality implies
01
p(x)dx. By substituting in Eq. (lA), (5A), and then the first
0
integration of Eq. (4A), we obtain the Gouy equation: u = (8n(w)~&T)'~sinh (zeJlo/2kT)
(ION
By following through the derivation without restricting ourselves to a symmetrical electrolyte (Grahame, 1947; Delahay, 1965; Aveyard and Haydon, 1973), we obtain the more general relation which must be used for solutions of mixed electrolytes:
APPBNDIX II
Haynes (1974) measured the fluorescence produced by the adsorption of l-anilino-8-naphthalenesulfonate(ANS) to bilayer membranes
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
133
and assumed that the fluorescence was an accurate measure of the number of ANS molecules adsorbed. The number of adsorbed ANS molecules should be a function of the electrostatic potential at the surface of the membrane and, in the simplest case, should be described by a combination of the Gouy, Langmuir, and Boltzmann relations. When Haynes attempted to fit his data with these equations, he observed serious discrepancies that he interpreted (see pp, 58-62 of his paper) in terms of a discrete charge effect. However, no measurements were made of the surface potentials produced by ANS. When the zeta potentials of PC= vesicles exposed to solutions containing various concentrations of ANS and NaCl are measured, the results do agree with the predictions of double layer theory. The data obtained are very similar to those presented in Fig. 14 for TNS (S. McLaughlin, unpublished experiments), Although we do not claim to understand the discrepancy between the predictions of the Gouy-Chapman theory and the fluorescence results obtained by Haynes with ANS, we note that it is almost certainly not due to a discrete charge effect. Haynes (1974) quotes a theoretical expression that indicates, quite correctly, that the discrepancy should be largest at low charge densities and high ionic strengths. The discrepancy he observed, however, became larger as the charge density increased and the ionic strength decreased. M. Eisenberg and S. McLaughlin are attempting to resolve the problem by directly measuring the change in charge density, surface potentid ,and fluorescence produced by the addition of ANS to a solution containing PC’ membranes. APPENDIX 111
We describe here, in terms of a three-capacitor model, the “boundary” potentials discussed in Section V. As a crude first approximation, we (Andersen et al., 1976b) assume that the energy wells illustrated in Fig. 18A divide the membrane into three regions. The two outer regions are defined as having a specific capacitance of C,, the inner region a specific capacitance of C , . If C , is the specific capacitance of the membrane, 1/C, = l/Ci + 2/C,. Implicit in this model are assumptions that hydrophobic ions such as tetraphenylborate adsorb in a plane and that they may be considered to be uniformly smeared over that plane. Andersen et al. (1976b) discuss this assumption in more detail. If, when no voltage is applied to the membrane, the charge adsorbed to the wells illustrated in Fig, 18A has a density q , the voltage drop across the left-hand boundary layer will be given by V& = q /Co. The voltage drop across the right-hand boundary layer will be given
134
STUART MCLAUGHLIN
by V'& = - q / C , . As illustrated in Fig. 19, no potential falls across the inner region, V , = 0. Now let a voltage V , be applied across the membrane. At any time after the application of the voltage, let the charge delivered by the voltage clamp to the surface of the membrane be qo the charge in the left hand well be 9', and the charge in the right-hand well be q". By applying Gauss' Law from left to right across the three regions of the membrane, we obtain
K O= (9 + qc)/C,,V
, = (q
+ qc - q ' ) / C 1 ,and v'&= (qc - q ) / C , .
We now define Aqc as the charge that the voltage clamp delivers to the outer surface of the membrane to maintain V, a constant as the internal charges move between the two wells (Fig. 18B).This charge is measured by integrating the current transient over time, ignoring the charge moved during the capacitive transient. As discussed by Andersen and Fuchs (1975), essentially no charge crosses the membrane-solution interface in the time required to make the measurement. It follows that Aqc/Cm = ( 9 ' - q ) / C 1 .We define Aqc,, as the limiting charge moved to the outside of the membrane when V , becomes very large. In terms of our model, Aqc,, = b q where b = C , / C l . A combination of the above expressions and the Boltzmann relation leads to the following equation:
+
[& [
11
Aqc, Aqc = exp bV, + (b - l)Aqc/CM Pqclnax - Aqc Note that, as b approaches unity, the equation reduces to the Boltzmann expression. As qc approaches zero and boundary potentials become negligible, the expression reduces to an equation derived previously by Andersen and Fuchs (1975). The expression derived here predicts, and experiments with tetraphenylborate confirm (Andersen et al., 1976b), that as more ions adsorb to the membrane and boundary potentials become larger, an increasingly larger voltage V, must be applied to the membrane to move a given fraction of the charge from one well to the other. We estimate, from these measurements, that the value of b is 0.95-0.98. Equivalently, the value of the outer capacitance C , is 20-60 x F/cm2. De Levie et al. (1974b), however, derived a value of about 5 x F/cm2 for the value of C, using a different analytic technique (de Levie et al., 1974a). ACKNOWLEDGMENTS This work was supported by Grant PCM76-04363from the National Science Foundation. I thank 0. Andersen, H. Friedman, S. Hladky, R. MacDonald, and A. Mauro for valuable discussions and correspondence about the topic discussed in Section V1,F.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
135
REFERENCES Abraham-Shrauner, B. (1975). Generalized Gouy-Chapman potential of charged phospholipid membranes with divalent cati0ns.J. Math. B i d . 2,333-339. Adam, G., and Adam, G. (1975).Cell surface charge and regulation of cell division of 3T3 cells and transformed derivatives. E x p . Cell Res. 93,71-78. Andersen, 0. S., and Fuchs, M. (1975).Potential energy barriers to ion transport within lipid bilayers. Biophys. J . 15, 795-830. Andersen, 0.S., Finkelstein, A., Katz, I., and Cass, A. (1976a).The effect ofphloretin on the permeability of thin lipid membranes.J. Gen. Physiol. 67,749-771. Andersen, O., Feldberg, S., Nakadomari, H., Levy, S., and McLaughlin, S. (197613).A new type ofelectrostatic boundary potential in lipid bilayers. Biophys. J . 16, 194a. Armstrong, C. M. (1975). Ionic pores, gates, and gatingcurrents. Quart. Rev. Biophys. 7 , 179-2 10. Aveyard, R.,and Haydon, D. A. (1973). “An Introduction to the Principles of Surface Chemistry.” Cambridge Univ. Press, London and New York. Azzi, A. (1975). T h e application offluorescent probes in membrane studies. Quart. Reo. Biophys. 8,237-316. Baker, P. F., Hodgkin, A. L., and Shaw, T. I. (1962).The effects of changes in internal ionic concentrations on the electrical properties of perfused giant axons. J . Physiol. (London) 164,355-374. Baker, P. F., Hodgkin, A. L., and Meves, H. (1963). Internal potassium concentration and the action potential of giant axons of Loligo. J . Physiol. (London) 167, 56P-57P. Baker, P. F., Hodgkin, A. L., and Meves, H. (1964).The effect of diluting the internal solution on the electrical properties of a perfused giant ax0n.J. Ph ysiol. (London)170, 54 1-560. Bakker, E. P., Arents, J. C., Hoebe, J. P. M., and Terada, H. (1975). Surface potential and the interaction of weakly acidic uncouplers of oxidative phosphorylation with liposomes and mitochondria. Biochitn. Biophys. Acta 387,491-506. Bangham, A. D., Standish, M. M., and Miller, N. (1965).Cation permeability of phospholipid model membranes; effect of narcotics. Nature (London)208, 1295-1297. Bangham, A. D., Hill, M. W., and Miller, N. G. A. (1974).Preparation and use of lipsomes as models of biological membranes. Methods Membr. B i d . 1,l-68. Barker, J. L., and Levitan, J. (1971). Salicylate: Effect on membrane permeability of molluscan neurons. Science 172, 1245-1247. Barlow, C. A., Jr. (1970). The electrical double layer. In “Physical Chemistry, an Advanced Treatise” (H. Eyring, ed.), pp. 167-246. Academic Press, New York. Begenisich, T. (1975). Magnitude and location of surface charges on Myxicola giant axons. J . Gen. Physiol. 66,47-65. Berden, J. A., Cullis, P. R., Hoult, D. I., McLaughlin, A. C., Radda, G. K., and Richards, R.E. (1974). Frequency dependence of31P NMR linewidths in sonicated phosphe lipid vesicles: Effects of chemical shift anisotropy. FEBS Lett. 46,55-58. Bezanilla, F., and Armstrong, C . M . (1975).Kinetic properties and inactivation of the gating currents of sodium channels in squid axon. Phil. Trans. Roy. SOC. London, Ser. B 270,449-458. Bockris, J. O’M., and Reddy, A. K. N. (1973). “Modern Electrochemistry,” Plenum, New York. Bradshaw, R. W., and Robertson, C. R. (1975). Effect of ionic polarizability on electre diffusion in lipid bilayer membranes. J . Membr. Biol. 25,93-114. Branton, D., and Deamer, D. W. (1972). “Membrane Structure.” Springer-Verlag, Berlin and New York.
136
STUART MCLAUGHLIN
Bretscher, M. S. (1973). Membrane structure: Some general principles. Science 181, 622-629. Brismar, T. (1973). Effects of ionic concentration of permeability properties of nodal membrane in myelinated nerve fibers of Xenopus laeois. Potential clamp experiments. Acta Physiol. Scand. 87,474-484. Brooks, D. E., and Seaman, G . V. F. (1973).The effect of neutral polymers on the electrokinetic potential of cells and other charged particles.]. Colloid Interface Sci. 43, 670-686. Brown, R. H. (1974).Membrane surface charge: Discrete and uniform modelling. Progr. Biophys. Mol. Biol. 28,343-370. Bruner, L. J. (1975).The interaction of hydrophobic ions with lipid bilayer membranes. J . Membr. Biol. 22, 125-141. Carroll, B. J., and Haydon, D. A. (1975).Electrokinetic and surface potentials at liquid interfaces. J . Chem. SOC., Faraday Trans. 1 71,361-377. Chandler, W. K., Hodgkin, A. L., and Meves, H. (1965).The effect of changing the internal solution on sodium inactivation and related phenomena in giant axons. J . Physiol. (London) 180,8214336. Chapman, D. L. (1913). A contribution to the theory of electrocapillarity. Philos. Mag. 25,475-481. Ciani, S . (1965). A rate theory analysis of steady diffusion in a fixed charge membrane. Biophysik 2,368-378. Cohen, F., Eisenberg, M., and McLaughlin, S. (1976). Kinetic studies on black lipid membranes with uncouplers of oxidative phosphorylation. Biophys. J . 16, l l l a . Conti, F. (1975).Fluorescent probes in nerve membranes. Annu. Reu. Biophys. Bioeng. 4,287-310. Conti, F., and Malerba, F. (1972). Fluorescence signals in ANS-stained lipid bilayers under applied potentials. Biophysik 8,326-332. Davies, J. T., and Rideal, E. K. (1963). “Interfacial Phenomena.” Academic Press, New York. Dawson, R. M. C. (1968). The nature of the interaction between protein and lipid during the formation of lipoprotein membranes. In “Biological Membranes” (D. Chapman, ed.), pp. 203-232. Academic Press, New York. Delahay, P. (1965).“Double Layer and Electrode Kinetics.” Wiley, New York. d e Levie, R., and Moreira, H. (1972). Transport of ions of one kind through thin membranes. I. General and equilibrium considerations.]. Membr. Biol. 9,241-260. de Levie, R., and Seidah, N. G . (1974).Transport of ions of one kind through thin membranes. 111. Current-voltage curves for membrane soluble ions. J . Membr. Biol. 16, 1-16. de Levie, R., Seidah, N. G., and Moreira, H. (1972). Transport of ions of one kind through thin membranes. 11. Nonequilibrium steady-state behavi0r.J. Membr. Biol. 10, 171-192. de Levie, R., Seidah, N. G., and Moreira, H. (1974a). Transport of ions of one kind through thin membranes. IV. Admittance for membrane-soluble ions. J . Membr. Biol. 16, 17-42. de Levie, R., Seidah, N. G . , and Larkin, D. (1974b). Tetraphenylborate adsorption onto an artificial ultrathin membrane. Electroanal. Chem. Interface Electrochem. 49, 153-1 59. Deman, J. J., and Bruyneel, E. A. (1974). Evidence for long-range electrostatic repulsion between HeLa cells. E x p . Cell Res. 89,206-216. Edidin, M. (1974). Rotational and translational diffusion in membranes. Annu. Reu. Biophys. Bioeng. 3, 179-201.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
137
EdsaI1, J. T., and Wyman, J. (1958). “Biophysical Chemistry.” Academic Press, New York. Ehrenstein, G., Gilbert, D. L., and Lipicky, R. J. (1975). Does phospholipid flip-flop affect axon potassium channels? Biophys. J . 15,847-849. Einstein, A. (1956). “Investigations on the Theory of the Brownian Movement” (R. Furth, ed., A. D. Cowper, trans].). Dover, New York. Elul, R., Brons, J., and Kravitz, K. (1975). Surface charge modifications associated with proliferation and differentiation in neuroblastoma cultures. Nature (London) 258, 616-617. Feinstein, M. B., and Felsenfeld, H. (1975). Reactions of fluorescent probes with normal and chemically modified myelin. Biochemistry 14,3041-3048. Feinstein, M. B., Spero, L.,and Felsenfeld, H. (1970).Interaction ofa fluorescent probe with erythrocyte membrane and lipids: Effects of local anesthetics and calcium. F E B S Lett. 6,245-248. Feldberg, S. W., and Kissel, C . (1975). Charge pulse studies of bansport phenomena in bilayer membranes. J . Membr. B i d . 20,269-300. Fettiplace, R., Andrews, D. M., and Haydon, D. A. (1971). The thickness, composition and structure of some lipid bilayers and natural membranes. J. Membr. BioZ. 5, 277-296. Feynnian, R. P., Leighton, R. B., and Sands, M. (1964). “The Feynman Lectures on Physics,” Vol. 2. Addison-Wesley, Reading, Massachusetts. Finkelstein, A., and Cass, A. (1968). Permeability and electrical properties of thin lipid membranes. J . Gen. Physiol. 52, 145s-172s. Finkelstein, A., and Holz, R. (1973). Aqueous pores created in thin lipid membranes by the polyene antibiotics nystatin and amphotericin B. In “Membranes” (G. Eisennian, ed.), Vol. 2, pp. 377-408. Dekker, New York. Flanagan, M. T., and Hesketh, T. R. (1973). Electrostatic interactions in the binding of fluorescent probes to lipid membranes. Biochim. Biophys. Acta 298,535-545. Fortes, P. A. G., and Ellory, J. C. (1975). Asymmetric membrane expansion and modification of active and passive cation permeability of human red cells b y the fluorescent probe 1-anilino-8-naphthalene sulfonate. Biochim. Biophys. Acta 413,65-78. Foster, M., and McLaughlin, S. (1974). Complexes between uncouplers of oxidative phosphorylation. J . Membr. Biol. 17, 155-180. Fowler, R. H. (1927). Strong electrolytes in relation to statistical theory, in particular the phase integral of Gibbs. Trans. Faraday SOC.23,434-443. Fowler, R. H. (1936). “Statistical Mechanics.” Cambridge Univ. Press, London and New York. Frank, H. S . (1955). Thermodynamics of a fluid substance in the electrostatic field.]. Chem. Phys. 23,2023-2032. Fromherz, P. (1973). A new m e t h d for investigation of lipid assemblies with a lipoid pH indicator in monomolecular films. Biochim. Biophys. Acta 323,326-334. Fromherz, P., and Masters, B. (1974). Interfacial pH at electrically charged lipid monolayers investigated by the lipoid pH indicator method. Biochim. Biophys. Acta 356, 270-275. Gaffney, B. J., and Mich, R. J. (1976). A new measurement of surface charge in model and biological lipid membranes. J . Am. Chem. Soc. 98, 3044-3045. Galla, H. J., and Sackmann, E . (1975). Chemically induced lipid phase separation in model membranes containingcharged lipids: A spin label study. Biochim. Biophys. Acts 401,509-529. Gilbert, D. L. (1971). Fixed surface charges. In “Biophysics and Physiology of Excitable Membranes” (W. J. Adelman, ed.), pp. 359-378. Van Nostrand-Reinhold, Princeton, New Jersey.
138
STUART MCLAUGHLIN
Gilbert, D. L., and Ehrenstein, G. (1969).Effect of divalent cations on potassium conductance of squid axons: Determination of surface charge. Bi0phys.J. 9 , 4 4 7 4 6 3 . Gingell, D. (1971).Cell membrane surface potential as atransducer. In “Membranes and Ion Transport” (E. Bittar, ed.), Vol. 3, pp. 317-357. Wiley (Interscience), New York. Goldhammer, A. R., Jain, M. K., and Cordes, E. H. (1975).Phospholipases. 111. Effects of ionic surfactants on the phospholipase-catalyzed hydrolysis of unsonicated egg lecithin liposomes. /. Membr. Biol. 23, 293-304. Gouy, M. (1910).Sur la constitution de la charge e‘lectrique a la surface d’un e‘lectrolyte. J. Phys. (Paris) 9,457-468. Grahame, D. C. (1947).The electrical double layer and the theory of electrocapillarity. Chem. Reo. 41,441-501. Grahame, D. C. (1958).Discreteness-of-chargeeffects in the inner region of the e l e c trical double layer. Z. Elektrochem. 62,264-274. Greville, G . D. (1969).A scrutiny of Mitchell’s chemiosmotic hypothesis of respiratory chain and photosynthetic phosphorylation. Curr. Top. Bioenerg. 3, 1-78. Grigor’ev, P. A., Yermishkin, L. N., and Markin, V. S. (1972). Direct passage of ions across lipid membranes-11. Experimental. Biofizika 17,788-793. Guggenheim, E . A. ( 1929). The conceptions of electrical potential difference between two phases and the individual activities of i0ns.J. Phys. Chem. 33,842-849. Guggenheim, E . A. (1930).On the conception of electrical potential difference between two phases. 1I.J. Phys. Chem. 34,1540-1543. Gutknecht, J., and Tosteson, D. C. (1973).Diffusion of weak acids across lipid bilayer membranes: Effects of chemical reactions in the unstirred layers. Science 182, 1258-126 1. Hall, J. E., and Latorre, R. (1976).Nonactin-K+ complex as a probe for membrane asymmetry. Biophys. J. 16, 99-103. Hall, J. E., and Simon, S. A. (1976).A simple model for calcium-induced exocytosis. Biochim. Biophys. Acta 436,613-616. Hanai, T., Haydon, D. A., and Taylor, J. (1965). Polar group orientation and the e l e c trical properties of lecithin bimolecular 1eaflets.J. Theor. Biol. 9,278-296. Hanstein, W. G. (1976). The role of the mitochondria1 uncoupler binding site in uncoupling of oxidative phosphorylation. Bioph ys. J. 16, 133a. Harold, F. M. (1972). Conservation and transformation of energy by bacterial membranes. Bacteriol. Rev. 36, 172-230. Hartley, G. S., and Roe, J. W. (1940).Ionic concentrations at interfaces. Trans. Faraday SOC.36, 101-109. Haydon, D. A. (1964).The electrical double layer and electrokinetic phenomena. Recent Prog. Surface Sci. 1,94-158. Haydon, D. A., and Hladky, S. B. (1972).Ion transport across thin lipid membranes: A critical discussion of mechanisms in selected systems. Quart. Rev. Biophys. 5, 187-282. Haydon, D. A., and Myers, V. B . (1973). Surface charge, surface dipoles and membrane conductance. Biochim. Biophys. Acta 307,429443. Haynes, D. (1974). 1-anilino-8-naphthalenesulfonate:A fluorescent indicator of ion binding and electrostatic potential on the membrane surface.]. Membr. Biol. 17, 34 1-367. Haynes, D. H., and Staerk, H . (1974). I-anilino-8-naphthalenesulfonate: A fluorescent probe of membrane surface structure, composition and mobi1ity.J. Membr. Biol. 17, 3 13-340. Henderson, R., Ritchie, J. M., and Strichartz, G. R. (1974). Evidence that tetrodotoxin and saxitoxin act at a metal cation binding site in the sodium channels of nerve membrane. Proc. Natl. Acad. Sci. U.S.A. 71,3936-3940.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
139
Hille, B. (1970). Ionic channels in nerve membranes. Progr. Biophys. Mol. Biol. 21, 1-32. Hille, B. (1975). The receptor for tetrodotoxin and saxitoxin. Biophys.]. 15,615-619. Hille, B., Woodhull, A. M., and Shapiro, B. I. (1975a). Negative surfice charge near sodium channels of nerve: Divalent ions, monovalent ions and pH. Phil. Trans. Roy. Soc. London, Ser. B 270,301-318. Hille, B., Ritchie, J . M., and Strichartz, G. R. (197Fjb).The effect ofsurhce charge on the nerve membrane on the action of tetrodotoxin and saxitoxin in frog myelinated nerve. 1.Physiol. (London) 246,36P-37P. Hladky, S. B. (1974).The energy barriers to ion transport by nonactin across thin lipid membranes. Biochim. Biophys. Acta 352, 71-85. Hladky, S. B. (1977).Ctcrr. Top. Membr. Trunsp. 10. (In preparation.) Hladky, S . B., and Haydon, D. A . (1973).Membrane conductance and surface potential. Biochim. Biophys. Acta 318,464-468. Hopfer, U., Lehninger, A. L., and Lennarz, W. J. (1970).The effect ofthe polar moiety of lipids on bilayer conductance induced by uncouplers of oxidative phosphorylation. 1.Membr. B i d . 3, 142-155. Huang, C., and Charlton, J. P. (1972).Interactions ofphosphatidylcholine vesicles with 2-p-toluidin ylnaphthalene-6sulfonate. Biochemistry 11,735-740. Hyer, E. J., Muller, R. U., and Finkelstein, A. (1976). Inactivation of monazomycin induced voltage dependent conductance; inactivation by long chain quaternary ammonium conipounds. ]. Gen. Physiol. 67, 703-729. Israelachvili, J. N . (1973a).Theoretical considerations on the asymmetric distribution of charged phospholipid molecules on the inner and outer layers of curved bilayer membranes. Biochim. Biophys. Actu 323,659-663. Israelachvili, J. N. (197313). Van der Waals forces in biological systems. Qzcurt. Reo. Biophys. 6 , 341-387. Israelachvili, J. N., and Mitchell, D. J. (1975).A model for the packing of lipids in bilayer membranes. Biochim. Biophys. A c t a 389, 13-19. Jacobson, K., and Papahadjopoulos, D. (1975).Phase transitions and phase separations in phospholipid membranes induced by changes in temperature, pH, and concentration of bivalent cations. Biochemistry 14, 152-161. Jan, K., and Chien, S . (1973).Role of surface electric charge i n red blood cell interactions.]. Gen. Physiol. 61, 638-654. Johnson, L. W., Hughes, M. E., and Zilversmit, D. B. (1975).Use of phospholipid exchange protein to measure inside-outside transposition in phosphatidylcholine liposomes. Biochim. Biophys. Acta 375, 176-185. Jost, P. C., and Griffith, 0. H. (1976). Polar head group specificity in lipid-protein interactions. Biophys. ]. 16, 48a. Jost, P. C., Griffith, 0. H., Capaldi, R. A,, and Vanderkooi, G. (1973). Evidence for boundary lipid in membranes. Proc. Nutl. Acad. Sci. U . S A . 70,480-484. Ke‘zdy, F. J. (1972). Lipid monolayers. In “Membrane Molecular Biology” (C. F. Fox, ed.), pp. 123-145. Sinauer, Stamford. Kirkwood, J. G. (1934).On the theory of strong electrolyte so1utions.J. Chem. Phys. 2, 767-78 1. Kirkwood, J. G., and Poirier, J. C. (1954). The statistical mechanical basis of the Debye-Huckel theory of strong electrolytes. J . Phys. Chem. 58, 591-596. Kornberg, R. D., and McConnell, H. M. (1971). Inside-outside transitions of phospholipids in vesicle membranes. Biochemistry 10, 1111-1120. Krishnan, C. V., and Friedman, H. L. (1971). Solvation enthalpies of electrolytes in methanol and dimethylformamide. ]. Phys. Chem. 75,3606-3612. Kuffler, S . W., and Yoshikami, D. (1975). T h e number of transmitter molecules in a
140
STUART MCIAUGHLIN
quantum: An estimate from ionotophoretic application of acetylcholine at the neuromuscular synapse. J. Physiol. (London) 251,465-482. Lauger, P., and Neumcke, B. (1973).Theoretical analysis ofion conductance in lipid bilayer membranes. In “Membranes” (G. Eisenman, ed.), Vol. 2, pp. 1-59. Dekker, New York. Leo, A., Hansch, C., and Elkins, D. (1971).Partition coefficients and their uses. Chem. Reo. 71,525-616. Lesslauer, W., Richter, J., and Lauger, P. (1967). Some electrical properties of bimolecular phosphatidyl inositol membranes. Nature (London) 213, 1224-1226. Lesslauer, W., Cain, J. E., and Blasie, J. K. (1972).X-Ray diffraction studies of the lecithin bimolecular leaflet with incorporated fluorescent probes of biological membranes. Proc. Natl. Acad. Sci. U S A . 69, 1499-1503. Levine, S. (1971). Adsorption isotherms in the electric double layer and the discreteness-of-charge effect. j~ Colloid Znterface Sci. 37, 619-634. Levine, S., and Bell, G. M. (1966). Modified Poisson-Boltsmann equation and free energy of electrical double layers in hydrophobic colloids. Discuss. Faraday SOC. 42, 69-80. Levine, Y. K., and Wilkins, M. H. F. (1971).Structure oforiented lipid bilayers. Nature (London),New Biol. 230, 69-72. Levitan, H., and Barker, J. L. (1972a). Membrane permeability: Cation selectivity reversibly altered by salicylate. Science 178, 63-64. Levitan, H., and Barker, J. L. (1972b). Effect of non-narcotic analgesics on membrane permeability of molluscan neurones. Nature (London),New Biol. 239,55-57. Litman, B. J, (1974). Determination of molecular asymmetry in the phosphatidylethanolamine surface determination in mixed phospholipid vesicles. Biochemistry 13, 2844-2848. Loeb, A. L. (1951).An interionic attraction theory applied to the diffuse layer around colloid paxticles. J. Colloid Sci. 6,75-91. Luner, S . J., Sturgeon, P., Szklarek, D., and McQuiston, D. T. (1975). Effects of proteases and neuraminidase on RBC surface charge and agglutination. Vox Sang. 28, 184-199. MacDonald, R. C., and Bangham, A. D. (1972).Comparison of double layer potentials in lipid monolayers and lipid bilayer membranes.J. Membr. Biol. 9,361-372. MacDonald, R. C., Simon, S. A., and Baer, E . (1976).Ionic influences on the phase transition of dipalmitoylphosphatidylserine.Biochemistry 15, 885-891. McLaughlin, S . (1972).The mechanism ofaction of DNP on phospholipid bilayer membranes. J. Membr. Biol. 9,361-372. McLaughlin, S . (1973). Salicylates and phospholipid bilayer membranes. Nature (London)243,234-236. McLaughlin, S. (1975). Local anesthetics and the electrical properties of phospholipid bilayer membranes, In “Molecular Mechanisms of Anesthesia” (B. R. Fink, ed.), pp. 193-220. Raven, New York. McLaughlin, S., and Eisenberg, M. (1975).Antibiotics and membrane biology. Annu. Rev. Biophys. Bioeng. 4,335-366. McLaughlin, S . , and Harary, H. (1974). Phospholipid flip-flop and the distribution of surface charges in excitable membranes. Biophys. J . 14,200-208. McLaughlin, S., and Harary, H. (1976).The hydrophobic adsorption of charged molecules to bilayer membranes: A test of the applicability of the Stern equation. Biochemistry 15, 1941-1948. McLaughlin, S . G. A., Szabo, G., Eisenman, G., and Ciani, S. M. (1970). Surface charge and the conductance of phospholipid membranes. Proc. Natl. Acad. Sci. U.S.A. 67, 1268-1275.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION INTERFACES
141
McLaughlin, S. C . A., Szabo, G.,and Eisenman, G. (1971). Divalent ions and the surface potential of charged phospholipid membranes.J. Cen. Physiol. 58, 667-687. McLaughlin,A. C., Cullis, P. R., Hemminga, M. A., Hould, G. I.,Radda, G. K., Ritchie, G. A., Seeley, P. J., and Richards, R. E. (1975a). Application of 31P NMR to model and biological membrane systems. F E B S Lett. 57,213-218. McLaughlin, S., Bruder, A., Chen, S., and Moser, C. (1975b).Chaotropic anions and the surface potential of bilayer membranes. Biochim. Biophys. Acta 394,304413. Markin, V. S., Crigor’ev, P. A., and Yermishkin, L. N. (1971). Forward passage of ions across lipid membranes-I. Mathematical model. Biofizika 16, 1011-1018. Milito, R. P., and Todd, P. (1976). Surface charge density of rat cells treated with concanavalin A. Biophys. J. 16,218a. Mille, M., and Vanderkooi, G. (1976). Electrochemical properties of spherical polyelec trolytes. I. Impermeable sphere model. J . Colloid Interface Sci. (In press.) Mitchell, P. ( 1966). Chemiosmotic coupling in oxidative and photosynthetic phosphe rylation. Biol. Rev. Cambridge Phil. S O C . 41, 445-502. Mohilner, D. M. (1966). The electrical double layer Part I . Elements of double-layer theory. Electroanal. Chem. I, 241-409. Montal, M., and Gitler, C. (1973). Surface potential and energy-coupling in bioenergy-conserving membrane systems. Bioenergetics 4,363-382. Moore, J. W., Narahashi, T., and Ulbricht, W. (1964). Sodium conductance shift in an axon internally perfused with a sucrose and low potassium solution. J. Physiol. (London) 172, 163-173. Moore, L. E., and Neher, E . (1976). Fluctuation and relaxation of analysis of monazw mycin conductance channels in black lipid membranes. Biophys. J. 16,80a. Moore, W. J. (1972). “Physical Chemistry.” Prentice-Hall, Englewood Cliffs, New Jersey. Mueller, P., Rudin, D. O., Tien, H. T., and Wescott, W. C. (1963). Methods for the formation of single bimolecular lipid membranes in aqueous so1ution.J. Phys. Chem. 67,534-535. Muller, R. U. (1971). Voltage dependent conductance induced in thin lipid membranes by monazomycin. Ph.D. Thesis. Albert Einstein College of Medicine, New York. Muller, R., and Andersen, 0. S. (1975). Single monazomycin channels. Int. Biophys. Congr. (Abstr., 111.) Muller, R. U., and Finkelstein, A. (1972a). Voltage-dependent conductance induced in thin lipid membranes by monazomycin. J . Gen. Physiol. 60,263-284. Muller, R. U., and Finkelstein, A. (1972b). The effect of surface charge on the voltage-dependent conductance induced in thin lipid membranes by monazOr mycin. J. Gen. Physiol. 60,285-306. Muller, R. U., and Finkelstein, A. (1974). The electrostatic basis of Mg++inhibition of transmitter release. Proc. Natl. Acad. Sci. U S A . 71,923-926. Narahashi, T. (1963). Dependence of resting and action potentials on internal potassium in perfused squid giant ax0ns.J. Physiol. (London) 169,91-115. Neher, E., and Sakmann, B. (1976). Agonist-induced discrete conductance changes in frog muscle. Biophys. J. 16, l54a. Nelson, A. P., and McQuarrie, D. A. (1975). The effect of discrete charges on the electrical properties of a membrane. I.J. Theor. Biol. 55, 13-27. Nelson, A. P., Colonomos, P., and McQuarrie, D. A. (1975). Electrostatic coupling across a membrane with titratable surface gr0ups.J. Theor. B i d . 50,317-325. Neumcke, B. (1970).Ion flux across lipid bilayer membranes with charged surface. Biophysik 6,231-240. Neumcke, B., and Lauger, P. (1969). Nonlinear electrical effects in lipid bilayer membranes. Biophys. J. 9, 1160-1170.
142
STUART MCLAUOHLIN
Neumcke, B., and Lauger, P. (1970). Space charge-limited conductance in lipid bilayer membranes.J. Membr. Biol. 3,54-66. Nonner, W., Rojas, E., and Stampfli, R. (1975). Displacement currents in the node of ranvier. Pfluegers Arch. 354, 1-18. Ohnishi, S., and Ito, T. (1974). Calcium induced phase separations in phosphatidylserine-phosphatidylcholinemembranes. Biochemistry 13,881-887. Olivares, W., and McQuarrie, D. A. (1976). On the theory of ionic solutions. Onsager, L. (1933).Theories of concentrated electrolytes. Chem. Reo. 1 3 , 7 3 4 9 . Overbeek, J. Th. G., and Wiersema, P. H. (1967). The interpretation of electrophoretic mobilities. In “Electrophoresis” (M. Bier, ed.), Vol. 2, pp. 1-52. Academic Press, New York. Paltauf, F., Hauser, H., and Phillips, M. C. (1971). Monolayer characteristics of some l,e-diacyl, 1-alkyl-2-acyl and 1,2-dialkyl phospholipids at the air-water interface. Biochim. Biophys. Acta 249,539-547. Papahadjopoulos, D. (1968). Surface properties of acidic phospholipids: Interaction of monolayers and hydrated liquid crystals with uni- and bivalent metal ions. Biochim. Biophys. Acta 163,240-254. Papahadjopoulos, D., Jacobson, K., Poste, G., and Shepherd, G. (1975). Effects of local anesthetics on membrane properties I. Changes in the fluidity of phospholipid bilayers. Biochim. Biophys. Acta 394,504-519. Parsegian, A. (1969). Energy of an ion crossing a low dielectric membrane: Solutions to four relevant electrostatic problems. Nature (London) 221,844-846. Parsegian, V. A. (1974). Possible modulation of reactions on the cell surface by changes in electrostatic potential that accompany cell contact. Ann. N.Y. Acad. Sci. 283, 362471. Parsegian, V. A., and Gingell, D. (1973).A physical force model ofbiological membrane interaction. In “Recent Advances in Adhesion” (L. Lee, ed.), pp. 153-190. Gordon and Breach, New York. Poste, G., and Reeve, P. (1972). Inhibition of cell fusion by local anesthetics and tranquillizers. Exp. Cell Res. 72,556-560. Poste, G., Papahadjopoulos, D., and Nicholson, G. L. (1975a). Local anesthetics affect transmembrane cytoskeletal control of mobility and distribution of cell surface receptors. Proc. Natl. Acad. Sci. U.S.A. 72,4430-4434. Poste, G., Papahadjopoulos, D., Jacobson, K., and Vail, W. J. (197%). Effects of local anesthetics on membrane properties 11. Enhancement of the susceptibility of mammalian cells to agglutination by plant lectins. Biochim. Biophys. Acta 394,520-539. Radda, C.K. (1975).Fluorescent probes in membrane studies. Methods Membr. Biol. 4, 97- 188. Rice, S. A., and Nagasawa, M. (1961).“Polyelectrolyte Solutions.” Academic Press, New York. Rojas, E., and Keynes, R. D. (1975).On the relation between displacement currents and activation of the sodium conductance in the squid giant axon. Phil. Trans. Roy. SOC. London, Ser. B 270,459-482. Roseman, M., Litrnan, B. J., and Thompson, T. E. (1975). Transbilayer exchange of phosphatidylethanolamine for phosphatidylcholine and N-acetimidoylphosphatidylethanolamine in single-walled bilayer vesicles. Biochemistry 14, 4826-4830. Rothman, J. E., and Dawidowicz, E. A. (1975). Asymmetric exchange of visicle phospholipids catalyzed by the phosphatidylcholine exchange protein. Measurement of inside-outside transitions. Biochemistry 14,2809-2815. Rubery, P. H., and Sheldrake, A. R. (1973). Effect ofpH and surface charge on cell uptake of auxin. Nature (London),New Biol. 244,285-288.
ELECTROSTATIC POTENTIALS AT MEMBRANE-SOLUTION
INTERFACES
143
Ryan, G. B., Unanue, E. R.,and Karnovsky, M. J. (1974).Inhibition of surface capping of macromolecules by local anesthetics and tranquillizers. Nature (London) 250, 56-57. Saffman, P. G., and Delbruck, M. (1975). Brownian motion in biological membranes. P ~ o c .Natl. Acad. S c i . U.S.A.72,3111-3113. Satir, B. (1975).The final steps in secretion. Sci. Amer. 2 3 3 , 2 8 4 7 . Scatchard, G. (1949). The attractions of proteins for small molecules and ions.Ann. N.Y. Acad. Sci. 51,660-672. Schauf, C. L. (1975). The interactions of calcium with Myxicola giant axons and a description in terms of a simple surhce charge model. J. Physiol. (London) 248, 613-624. Shaw, D. J. (1970). “Introduction to Colloid and Surface Chemistry.” Butterworth, London. Sheetz, M. P., and Singer, S . J. (1974). Biological membranes as bilayer couples. A molecular mechanism of drug-erythrocyte interactions. Proc. Natl. Acad. Sci. U S A . 71,4457-4461. Singer, S . J. (1971). The molecular organization of biological membranes. In “Structure and Function of BioIogical Membranes” (L. I. Rothfield, ed.), pp. 145-222. Academic Press, New York. Singer, S. J., and Nicolson, G. L. (1972). T h e fluid mosaic model of the structure of cell membranes. Science 175,720-731. Skulachev, V. P. (1971). Energy transformations in the respiratory chain. Curr. Top. Bioenerg. 4,127-190. Stoekenius, W., and Engelman, D. M. (1969).Current models for the structure ofbiological membranes. J . Cell Biol. 42,613-646. Szabo, G. (1975). Dual mechanism for the action of cholesteroi on membrane perrneability. Nature (London) 2 5 2 , 4 7 4 9 . Szabo, G. (1976). The influence of dipole potentials on the magnitude and the kinetics of ion transport in lipid bilayer membranes. In “Extreme Environment; Mechanism of Microbial Adaption” (M. R. Heinrich, ed.), pp. 321-348. Academic Press, New York. Szabo, G., Eisenman, G., McLaughlin, S. G. A,, and Krasne, S. (1972). Ionic probes of membrane structures. Ann. N.Y. Acad. Sci. 195,273-290. Szabo, G., Eisenman, G., Laprade, R., Ciani, S. M.,and Krasne, S. (1973). Experimentally observed effects of carriers on the electrical properties of bilayer membranes-equilibrium domain. In “Membranes” (G. Eisenman, ed.), Vol. 2, pp. 179-328. Marcel Dekker, New York. Tanford, C. (1961). “Physical Chemistry of Macromolecules.” Wiley, New York. Tanford, C. (1973). “The Hydrophobic Effect: Formation of Micelles and Biological Membranes.” Wiley, New York. Tasaki, I., and Shimamura, M. (1962). Further observations on restingand action potential of intracellularly perfused squid axon. Proc. Natl. Acad. Sci. U.S.A. 48, 1571-1577. Theuvenet, A. P. R., and Borst-Pauwels, G. W. F. H. (1976).The influence of surface charge on the kinetics of ion-translocation across biological membranes. J. Theor. BioZ. 57,313-329. Trauble, H., and Eibl, H . (1974).Electrostatic effects on lipid phase transitions: Membrane structure and ionic environment. Proc. Natl. Acad. Sci. U S A . 71, 214-219. Trauble, H., and Overath, P. (1973). The structure ofEscherichia coli membranes studied by fluorescence measurements of lipid phase transitions. Biochirn. Biophys. Acta 307,491-512. Trissl, H. W. (1975). A theoretical consideration how to study biochemical interfacial
144
STUART MCLAUGHLIN
photoreactions. 2.Nuturforsch. 30q 124-126. Vanderkooi, J., and Martonosi, A. (1969). Sarcoplasmic reticulum. VII. Use of 8aniline 1-naphthalene’sulfonate as conformational probe of biological membranes. Arch. Biochem. Biophys. 133, 153-163. Vanderkooi, J., and Martonosi, A. (1971). Sarcoplasmic reticulum XII. The interaction of 8-anilinel-naphthalene sulfonate with skeletal muscle microsomes. Arch. Bioc h m . Biophys. 144,87-98. Verwey, E. J. W., and Overbeek, J, Th. G . (1948). “Theory of the Stability of Lyophobic Colloids.” Elsevier, Amsterdam. Waggoner, A. (1976). Fluorescent Probes of Membranes. In “The Enzymes of Biological Membranes” (A. Martinosi, ed.), Vol. 1, pp. 119-137. Plenum, New York. Walz, D., Bamberg, E., and Lauger, P. (1969). Non linear electrical effects in lipid bilayer membranes. 1. Ion injection. Biophys. J . 9,1150-1 159. Wanke, E. (1975). Monazomycin and nystatin channel noise. Int. Biophys. Congr. (Abstr., 111, 112.) White, D. A. (1973). The phospholipid composition of mammalian tissues. In “Form and Function of Phospholipids” (G. B. Ansell, J. N. Hawthorne, and R. M. C. Dawson, eds.), pp. 441482. Elsevier, Amsterdam. Wirtz, K. W. A. (1974). Transfer of phospholipids between membranes. Biochim. Biophys. Actu 344,95-117. Zwolinski, B. J,, Eyring, H., and Reese, C. E. (1949). J . Phys. CoUaid Chem. 53, 1426-1453.
A Thermodynamic Treatment of Active Sodium Transport S. ROY CAPLAN AND ALVlN ESSlG Department of Membrane Research Weizmann Institute of Science Rehovot, lsrael and Department of Physiology Boston University School of Medicine Boston, Massachusetts
I . Introduction . . . . . 11. Theory of the None
............... 111.
...................................... ...............
IV. V. VI . VII. VIII. IX.
Theory o f the Equivalent Circuit Model ..................... Experimental Evaluation of the Equivalent Circuit Model .... Utility of the Thermodynamic Affinity A . . . . .
Some General Comments . . . . . ................ Conclusions ......................... .............. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.
145 147 147 147 149 149 150 150 162 165 165 169 170 173 173
INTRODUCTION
Thermodynamics provides a basis for the systematic investigation of the kinetic and energetic factors determining transport under all circumstances. In approaching the study of epithelial active sodium transport from a thermodynamic point of view, we have kept two considerations in mind. (i) Studies in epithelial tissues are relevant to the study of active sodium transport in all the many tissues in which it is observed; epithelial membranes serve as useful model systems for the thermodynamic analysis of transport since they permit the ready control of bath concentrations and electrical forces determining rates of 145
146
S. ROY CAPLAN AND ALVIN ESSIG
transport and metabolism. (ii) The understanding of active sodium transport must contribute to an understanding of all active transport processes, both ionic and non-ionic. It is useful to consider briefly the classical approaches still commonly used to evaluate the energetic and kinetic factors determining transport. Studies of energetics are frequently based on the use of the flux ratio; in analogy to passive transport in simple systems, it is presumed that the ratio of the unidirectional fluxes of a given species provides a measure of the forces promoting its transport. The fundamental flaws in this approach have been discussed in detail elsewhere (Kedem and Essig, 1965); briefly, the flux ratio can evaluate energetic parameters only if the flows are entirely by way of the active pathway and if tracer flows are uninfluenced by flows of other isotopes or other chemical species. Another classical approach to these issues is based on the equivalent circuit model (Ussing and Zerahn, 1951), which comprises an active conductance IF‘, an electromotive force of sodium transportENa,and a parallel passive conductance KP. I n this model, the K’S are considered to represent the kinetic factors, whereas ENais commonly taken as the “driving force” for the transport process. However, fundamental theoretical considerations as well as experimental results indicate that EN, incorporates both kinetic and energetic factors (Civan et al., 1966; Essig and Caplan, 1968; Hong and Essig, 1976). The classical point of view leads to generally accepted intuitions about the characteristics of the active transport process. By analogy with stoichiometric chemical reactions, it was long assumed that active sodium transport is associated with a unique ratio of sodium transported to oxygen consumed (Zerahn, 1956; Ussing, 1960).Again, both theoretical considerations and experimental results show this not to be the case (Essig and Caplan, 1968; Vieira et al., 1972a). Furthermore, it was considered that the energy available for transport can be evaluated by the “calorific value,” i.e., the heat released in the oxidation of glucose in a bomb calorimeter (Zerahn, 1956; Ussing, 1960). However, the pertinent quantity is not the heat released (enthalpy) but rather the free energy of the reaction under in vivo conditions. In the present article, we present a point of view based on nonequilibrium thermodynamics (NET). This enables us to analyze the behavior of a tissue systematically under a wide variety of operating conditions, and also the mode of action of substances regulating transport. In particular, it enables us to distinguish between effects on energetic as against kinetic factors. We then consider briefly the equivalent circuit representation, pointing out its limitations and the extent to which it and NET provide complementary information.
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
147
II. THEORY OF THE NONEQUlLlBRlUM THERMODYNAMIC (NET) APPROACH
A. Background
Many investigators have employed NET in order to analyze the modes of utilization of metabolic energy. Jardetzky and Snell (1960) presented a general theoretical analysis of transport and various metabolic processes utilizing the notation of NET, but requiring neither linearity nor the validity of the Onsager reciprocal relations. Kedem (1961) gave a formal description of active transport based on linear NET, assuming the validity of the Onsager relations, and showed how the formalism permits the correlation of different types of measurements in two-flow systems. Hoshiko and Lindley (1967) emphasized the importance of a clear operational definition of active.transport and extended the methods of Kedem and Katchalsky (1958)to the active transport of single salt and bi-ionic systems. Procedures were outlined to evaluate the requisite 10 or 15 phenomenological coefficients. Heinz (1974) has analyzed sodium-linked amino acid transport in an NET formulation based on that of Rapoport (1970). In order to distinguish between coupled and uncoupled processes, a “quasi-chemical” notation was introduced which treats all uncoupled events in terms of an intrinsically stoichiometric chemical reaction complicated by leakage. B. Model for the Active Tmnsport of a Single Ion
We have preferred to restrict ourselves to systems with active transport of only one ion, uncoupled to the flows of other species (Essig and Caplan, 1968). The theoretical basis for our work is an extension of that of Kedem. In contrast to the treatments of Rapoport and Heinz, we treat the coupling between transport and metabolism quite generally, admitting the possibility of intrinsic incomplete coupling of this process. In analyzing the energetics of active sodium transport, it is necessary to start with a simple model. In frog skins and toad bladders of appropriate species, there appears to be only one significant active transport process, that of sodium, and thus only one significant output for our thermodynamic system. We assume, despite the great complexity of biological tissues, that we can isolate one metabolic process which “drives” active transport. This model is represented in Fig. 1. Here one input process, the metabolism of substrate, is linked to one
148
S. ROY CAPIAN A N D ALVIN ESSlG OUTPUT
mM
+ nN
PP + q Q
INPUT
FIG. 1. General scheme for the coupling of metabolism to sodium transport (Essig and Caplan, 1968).
output process, the transport of sodium. In this representation, the consumption of M and N to produce P and Q provides the free energy that brings about the active transport of sodium across the membrane. Similarly, we take a simplified view of the histology of the system (Fig. 2). The process of active transport takes place in the rectangular box. Although it is not necessary for our analysis, in accord with many experimental observations we represent the outer or apical region as a simple passive barrier across which sodium moves down its electrochemical potential gradient. At the inner or basolateral surface is the mechanism responsible for active sodium transport, the so-called sodium “pump.” Since the active transport system transports only sodium ions, whereas the tissue as a whole reabsorbs sodium chloride, it is necessary for there to be a pathway across which chloride can move. This is represented as a simple passive channel in parallel with the active transport pathway, which is presumably accessible more or less to all of the ions in the bathing solutions. Since we have considered active transport to be a two-flow process, for there are two pertinent flow equations to be considered-one
GI-
>
FIG.2. Model of composite transport system (Essig and Caplan, 1968).
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
149
sodium transport in the active pathway, which we represent byJNaa, and one for metabolic reaction, which we represent by Jr:
Here XNa is the negative electrochemical potential difference of sodium, and A is the affinity of a metabolic reaction which is driving sodium transport. Under the usual experimental conditions, the affinity is equivalent to the negative Gibbs free energy change - AG of the driving reaction, as yet undefined in biochemical terms. The L’s are phenomenological coefficients. For simplicity, we assume linearity. JNaa, the rate of active sodium transport, is of course a function of the negative electrochemical potential difference, XNa, but to the extent that it is coupled to metabolism, it must also be a function of the affinity A. Jr, the rate of metabolism, here taken as suprabasal oxygen consumption, is of course a function ofA, but to the extent that it is linked to transport it must also be a function of XNa.By analogy with a variety of transport processes in nonliving systems (Miller, 1960; Blumenthal e t al., 1967), the validity of the Onsager reciprocal relation is assumed, i.e., the cross-coefficients in the two equations are set equal. The appropriateness of the simple thermodynamic formulation can be defended on various theoretical grounds, but we instead consider here the experimental evidence bearing on this point. Fortunately, very convenient systems are available for this purpose.
111.
EXPERIMENTAL EVALUATION OF THE NET APPROACH
A. Appropriate Tissues
Thanks to the pioneering work of Ussing and Zerahn (1951) with frog skin and Leaf et al. (1958) with toad bladder, two relatively simple epithelial membranes are available for use as experimental model systems. When either of these membranes is exposed to physiological saline solutions at each surface, an electrical potential difference is generated. This is the consequence of active sodium transport from the outer surface of the frog skin (or urinary surface of the toad bladder) across the tissue into the body fluid. As mentioned above, in appropriate species, essentially only sodium is transported actively. Consequently, in the absence of a transinembrane electrical potential difference, the magnitude of the electrical current is equiva-
150
S. ROY CAPIAN AND ALVIN ESSIG
lent to the rate of active sodium transport. Hence the effect of experimental manipulations, for example, the addition of drugs or hormones, can be observed promptly by monitoring the “short-circuit” current. A combination of voltage clamp and tracer isotope techniques allows an evaluation of the rate of active sodium transport at various settings of the transmembrane electrical potential. B. Stoichiometry
From a classical biochemical point of view, by analogy with coupled reactions in uitro, it is reasonable to expect that the transport of a given quantity of sodium ion would be associated with the oxidation of substrate in a fixed stoichiometric ratio. Indeed, this was described in early studies in frog skin (Zerahn, 1956). From the viewpoint of NET, on the other hand, it is only necessary that the rate of input of metabolic energy exceed the rate of performance of electroosmotic work. Hence, in the short-circuited state, i.e., in the absence of either a concentration difference or an electrical potential difference across the membrane, on a priori thermodynamic grounds, there is no limitation on Na/O, ratios, and these might well differ in various tissues. Recent studies of oxygen consumption employing Clark electrodes with vigorous stirring have demonstrated suprabasal Na/O, ratios varying from 7.1 to 30.9 in short-circuited frog skins (Vieira et al., 1972a). Similarly, in determinations of COz production in shortcircuited toad bladders, Al-Awqati et al. (1975) found extensive varia(Fig. 3). On the other hand, the ratio in a given bility of dJNa/dJCO1 bladder was characteristic, differing in paired tissues by only 0.92 k 0.94 (mean f SD, n = 8 pairs). The basis for different apparent stoichiometric ratios is as yet unknown. These relationships become still more complex with variation of the sodium concentration or electrical potential difference. Under these circumstances, the suprabasal Na/O, ratio will remain constant only if transport and metabolism are completely coupled (Essig and Caplan, 1968; Lahav et al., 1976; Lang et al., 1977). C. Relationship between Flows and FoRes
1. ELECTROCHEMICAL POTENTIALDIFFERENCE
We first consider the behavior of the transport process. With the use of identical solutions at each surface, there is no concentration difference across the membrane. Thus, the electrochemical potential dif-
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
151
in 28 toad bladders (mean 2 1 SD) (Al-Awqati et al., RG.3. Values of (dJNa/dJco,) 1975).
ference for sodium is given by FA$, where F is the Faraday constant and A$ is the electrical potential difference. If the rate of active sodium transport is indeed a linear function of the forces promoting transport, one might expect to find linear current-voltage relationships. Such relationships have, in fact, been described for epithelial tissues, but their significance has been unclear because of the possibility that an appreciable fraction of transport was by way of leak pathways. Obviously, to the extent that this is the case, linear current-voltage relationships are of little interest, since they are to be expected in dilute aqueous electrolyte solutions. For our purpose, it was important to study tissues in which transport by way of leak pathways was minimal. This was accomplished by carefully avoiding edge damage in mounting the tissues and by the use of tracer isotopes for the measurement of passive ion fluxes. This permitted the choice of tissues in which a high fraction of total conductance was by way of the active pathway. The results of such studies in the toad bladder (Saito et al., 1974) are shown in Fig. 4.For the seven tissues selected, an
152
S. ROY CAPLAN AND ALVIN ESSIG
I
Ka/K=0.596 k 0 . 0 5 9
.-200 FIG. 4. Normalized current-voltage relationship in the toad bladder (Saito et al., 1974).
average of some three-fifths of the total conductance was attributable to the active pathway. The finding of a steady-state linear normalized current-voltage relationship in these tissues strongly indicates that the rate of active sodium transp~rt],,~was in fact a linear function of the electrical potential difference. In order to demonstrate such linearity, the tissues were observed under quasi-steady-state conditions some 15-30 sec after the perturbation of A+. The observation of transients, on the one hand, or perturbations for many minutes, on the other hand, may well give different results (Mandel and Curran, 1972). The demonstration of linearity of the active transport process is consistent with the use of Eq. (l),and indicates that the phenomenological coefficients and the affinity are unaffected by perturbations of A$ of the magnitudes and duration employed. Corresponding studies were carried out of the rate of metabolism, here taken as oxygen consumption. For this purpose, we employed an adaptation of the Ussing-Zerahn chamber incorporating oxygen electrodes, as shown in Fig. 5. First, studies were carried out in the frog
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
153
FIG.5 . Apparatus for the simultaneous measurement of electrical parameters and oxygen consumption. OE, oxygen electrode; PC, polarographic circuit; R, recorder (Vieira et ol., 1972a).
skin (Vieira e t al., 1972b). Again there was striking linearity, with steady-state J r being a linear function of A$ over a range, in this instance, of + 160 to - 160 mV (Fig. 6). The finding of linearity ofJ, in A$ is consistent with the validity of Eq. (2) and again indicates constancy of phenomenological coefficients and invariance of the affinity with the perturbations of A$ employed. Later studies showed similar behavior in the toad bladder (Lang et al., 1976).Our conclusion that the dependence of the rate of oxygen consumption on the potential reflects the function of the active transport system is supported by the absence of such dependence following abolition of sodium transport by M ouabain, a specific inhibitor of the sodium pump (Fig. 7) (Vieira et al., 1972b; Saito et al., 1973). The studies just described examined the influence of only electrical driving forces on transport and metabolism. Recently, Danisi and Vieira ( 1974) have reported the effects of concentration driving forces. In studies of toad skins in which the transmembrane electrical potential difference was nullified, it was found that the rates of both active transport and oxygen consumption were linear functions of the chemi-
154
S. ROY CAPLAN A N D ALVIN ESSlG J,
t -160 -120
("mole
sec-' cm-'1
50
-80 -40 0 40 A\J ( m V )
80
120
160
FIG.6. Dependence of the rate of oxygen consumptionJ, on the electrical potential difference A$ in the frog skin (Vieira et al., 197213).
I
0
= OUABAIN ABSENT
0 8 OUABAIN
PRESENT
FIG.7. Dependence of the rate of O2consumptionJ, on the electrical potential difference AJI in the aldosterone-treated frog skin; influence of ouabain (lo+ M ) (Saito et al., 1973).
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
155
FIG.8. (A) Short-circuit current l o as filnction of the chemical potential difference of sodium (Danisi and Vieira, 1974). (B) Rate of suprabasal oxygen consumption as fiinction of the chemical potential difference of sodium (Danisi and Vieira, 1974).
156
S. ROY CAPLAN A N D ALVIN ESSIG
cal potential difference of sodium across the membranes (Fig. 8). Thus, we see that whatever the effects of external conditions on the intracellular concentration and electrical potential profiles, Eqs. (1) and (2) appear to be applicable.
2. AFFINITY a. Theoretical Considerations. The experimental results to date are consistent with the validity of a linear NET approach, but, of course, the experiments cited have tested only the effects of variation of the electrochemical potential difference. The experimental techniques available do not permit a systematic study of the effects of variation of the affinity. Here, however, work in other systems is pertinent. It might seem that linearity in A would obtain only over a limited range, since for an isolated chemical reaction, linearity requires A << RT, where R is the gas constant and T the absolute temperature. However, as was pointed out years ago by Prigogine (1961), biochemical reactions with large A often consist of a large number of elementary reactions in series. In such cases, the A’s of the elemental reactions may be sufficiently small that these reactions show linearity. Since in the steady state all series reactions occur at the same rate, J r may then be linear in the overall affinity, which is the sum of the affinities of the individual reactions. A further consideration is the fact that under certain circumstances, the kinetics of enzymatic reactions are compatible with linear NET equations over an appreciable range (Caplan, 1971; Rottenberg, 1973). Also, complex metabolic reactions may include a number of diffusional steps for which flows are more highly linear in chemical potential differences than is the case for chemical reactions. Although the question of linearity of active transport inA cannot yet be tested experimentally, it is quite conceivable that with the advent of appropriate model systems this may be possible. One such model system may be a lipid bilayer impregnated with Na,K-ATPase in such a manner as to ensure a well-defined polarity. Fixing the transmembrane electrical potential and the Na+, ATP, ADP, and Pi concentrations in the bathing solutions, as well as other important constituents, should permit a complete test of Eqs. (1) and (2). b. Experimental Considerations. Rottenberg (1973) has studied these matters from both a theoretical and experimental point of view. He has shown that in mitochondria rates of both oxidation and phosphorylation are linear functions of both the affinity of oxidation and the affinity of phosphorylation in regions quite far from equilibrium. Furthermore, his system provided a striking example of the Onsager reciprocal relation (Fig. 9).
157
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT Oxidation Affinity, kcol/md TC
'E
-
35.5
36.0
I
36.5 I
I
37.0 I
37.5
38.0
I
x
-
E4
I -10.5
-11.0
-10.0
I -9.5
I -9.0
I
-8.5
Phosphorylolion Affinity,kcol/m01
FIG.9. The rates of oxidative phosphorylation as a function of the reaction affinities. The open circles show the rate of the phosphorylation as a function of the phosphorylation affinity, and the filled circles show the rate of this reaction as a function of the oxidation affinity. T h e open triangles show the rate of oxygen consumption as a function of the phosphorylation affinity, while the closed triangles show the rate of oxygen consumption as a function of the oxidation affinity (Rottenberg, 1973).
Given the above support for linear phenomenology from two theoretical points of view and from Rottenberg's experimental results, it seems reasonable to pursue the possibility that active transport in epithelial tissues might indeed show linearity not only in the electrochemical potential difference of sodium but also in the affinity of the metabolic driving reaction. If so, we can evaluate the affinity by use of the two phenomenological equations [Eqs. (1) and (2)l:
A
=
-Zo/(dJJdA$)
(A constant)
(3)
Here the numerator represents the current measured in the absence of both an electrical potential difference and a concentration difference across the membrane, the short-circuit current, and the denominator is the slope of the plot relating the rate of metabolismJ, (oxygen consumption) to the electrical potential difference A$. The affinity A represents the free-energy change (per mole of 0,) for a characteristic region of the metabolic chain for whichA remains constant on pertur-
158
S. ROY CAPIAN AND ALVIN ESSIG
bation of A$. Admittedly, for the present this affinity is a rather vague quantity, being evaluated only in abstract thermodynamic terms. Nevertheless, it is of physiological interest, since it must reflect the substrate-product concentration ratio of some critical reaction in the metabolic pool which supports active transport. This is in contradistinction to mean cell concentration ratios of various substrates and products, including nucleotides such as ATP and ADP, and creatine phosphate and creatine (Handler et al., 1969). Although attempts have been made to study the driving forces for transport by such measurements, mean concentration ratios may well depend importantly on tissue functions other than transepithelial transport. Attempts to evaluate cytoplasmic ATP/(ADP x Pi) (Veech et al., 1970) also involve theoretical and experimental difficulties. Clarification of the significance of the affinity calculated by the above thermodynamic method must await experiments correlating thermodynamic studies with a variety of biochemical procedures. Meanwhile, several experimental studies appear to support the validity of attempting to evaluate the affinity by the means described. The first, a phenomenon studied by Vieira in the frog skin, has been termed a “memory” effect (Vieira et al., 1972b). This involves the observation of the short-circuit current Zo and the concomitant rate of oxygen consumption JrO, before and after perturbation of the electrical potential. If we perturb the potential so as much to enhance sodium transport for an extended period, we find that, on return to the short-circuit state, both Zo andJm are less than initially. These results are shown by the open circles of Fig. 10. It is seen that after permitting the skin to function at A$ = - 160 mV for some 15-20 min, which substantially enhances the rates of transport and metabolism, on return to the short-circuit state Zo andJro have decreased (i.e., AZo and AJro are negative). The solid circles show the converse effect, noted after temporarily setting A$ at + 160 mV, thereby depressing transport and metabolism. Both of these sets of observations are consistent with the interpretation that, although brief perturbations of A$ have relatively little influence on tissue metabolite levels, sufficiently long perturbations of significant magnitude change substrate and product concentrations appreciably. The resultant change in the substmte-product ratio, and thus presumably in the affinity, manifests itself on return to the short-circuit state by alterations in the magnitudes of I. and Jro. A second study employs the diuretic, amiloride. This pharmacological agent increases the excretion of salt by depressing the reabThe affinityhas been designated by the symbols A, “A,” and A,, (i.e.,apparent affinity) in different publications.
159
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
O
t-20
+, after
+160mV
o after
-160mV
0
SKIN 2
20
A10
FIG.10. Effect of electrical potential perturbations on subsequent values of shortcircuit current Zo and oxygen consumption J m in the frog skin (Vieira et al., 197213).
sorption of sodium from the kidney filtrate. Studies in the toad bladder demonstrate that amiloride depresses active sodium transport by interfering with the passive entry process at the outer (urinary) surface (Bentley, 1968; Handler et d.,1972). In our view of the mechanisms influencing the metabolic pools, we would anticipate that marked depression of sodium entry would cause a continuing accumulation of the intermediary metabolites that drive transport, with a gradual increase in the value of the affinity. This could account for the commonly observed “overshoot” of s hort-circuit current on removal of the drug. Figure 11 shows this result again in our studies (Essig, 1975). Here the open circles represent “control” tissues, and the solid circles represent paired “experimental” tissues. For each pair, the control and experimental tissues are derived from the same animal, assuring good matching of function prior to administration of the drug. Before the administration of amiloride, the magnitude of Zo is the same in the paired tissues. Following initial observations, amiloride was applied to the outer surface in concentrations sufficient to depress Zo to about a third of the initial level for 4 hours. On subsequent removal of the amiloride, lo in the treated tissues rose to a level signifi-
160
S. ROY CAPLAN A N D ALVIN ESSlG
3 CONTROL AMILORIDE
120-
( MEAN
*
SE, 17.7)
100-
I
h
7
5
80-
a
5
AMILORIDE REMOVED
AMILORIDE
I
[‘I
60-
0
H
40-
20
0’
3
€
-
1
0
1
2
3
4
TIME ( HOURS)
FIG.11. Effect of amiloride on short-circuit current I,, in the frog skin (Saito et al., 1973).
cantly higher than in the initial period and higher than observed simultaneously in the control tissues. Whether this behavior can be attributed to effects on the affinity is considered in Fig. 12. Before the administration of amiloride, A was the same in the control and experimental tissues, just as was the case for Zo. One hour after the administration of amiloride, A was not demonstrably affected. Four hours after the administration of amiloride, however, A in the treated tissues was significantly greater than initially and significantly greater than simultaneously in the paired control tissues. A third study uses the cardiac glycoside, ouabain, an inhibitor of the sodium-potassium-ATPase generally identified with the sodium pump. Since ouabain should have no direct effect on tissue metabolite levels, it would not be expected to have a prompt effect on the affinity. This prediction was tested by Owen et al. (1975b) in frog skin. In order to apply the above formulation for the evaluation of the affinity, it is necessary to depress Zo and (-dJ,/dA$) substantially, but not completely, so that the quantity A in Eq. (3) will remain determinate. This was accomplished by the use of a low concentration of ouabain, lo-’ M. Within 24 hours of exposure to this concentration of ouabain,
161
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
10! 20
OJ -1
0
1
2
3
4
TIME (HOURS)
FIG.12. Effect of arniloride on the affinity A in the frog skin (Saito et al., 1973).
the short-circuit current in the experimental hemiskin had gradually fallen to a level about half that in the paired control hemiskin. This was accompanied by a significant depression of the sensitivity of oxygen consumption to perturbation of A$, as is shown in Fig. 13, representing a typical result in nine experiments. As might be expected, incomplete inhibition of sodium transport for a relatively short period was not associated with a significant effect on the affinity (Fig. 14). (It is to be anticipated that on prolonged exposure to ouabain, as in the case of amiloride, depression of sodium transport would eventually result in enhancement of the affinity.) In a fourth study, Owen et al. (1975b) examined the effect of 2deoxy-D-glucose (2DG), an inhibitor of carbohydrate metabolism. In view of its effects in other tissues, 2DG might be expected to depress the affinity of the metabolic reaction driving sodium transport. Preliminary studies in frog skin showed that 2DG inhibits short-circuit current and that glucose interferes with this effect. When analyzed according to the Michaelis-Menten model, the kinetics appear to be competitive (Owen et al., 1975a). I n ten studies of frog skins exposed to 1 mM glucose, a concentration of 16 mM 2DG depressed active sodium transport, as measured by the short-circuit current, to an average of 58% of the control level. As expected, this was associated with a significant decrease in the affinity, in this case, to 53% of control level (Fig. 15). The various lines of evidence just discussed give us some tentative
162
S. ROY CAPLAN AND ALVIN ESSlG I
I
1
Exmcrimolal picornole Or 'r rec-crnz
picornole 0 2 'r sec-cmz
T
T
I
I
-
I
I
1 1011 1111 1 -1 50 50 + - 50110 11 50 + AYmV
Jr
picomole 9 sec crn2
AYrnV
Con
-
T l120 6 O I 10'33.3tA
'O
t
-rw+%i+ AY rnV
50
0 OYrnV
50
+
FIG.13. Effect of ouabain on the dependence of the rate of oxygen consumptionJ, on the electrical potential difference AS (Owen et al., 197513).
confidence in our interpretation of the nature of the metabolic pools driving transport and the significance of the affinity. On this basis, it seems appropriate to use the thermodynamic formalism to investigate the mechanisms of action of substances that influence transport by unknown means. First, however, we consider a classical approach.
IV.
THEORY OF THE EQUIVALENT CIRCUIT MODEL
For the analysis of active transepithelial sodium transport, many have used an equivalent circuit model (Ussing and Zerahn, 1951).In this model, the active pathway consists of a conductance element which we call K~ in series with the electromotive force of sodium
163
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
a
CONTROL
I OUAIAIN(107M) (MEAN i S.E.,
n=S
+
OUABAIN
TT
-
2
0
2
1
6
TIME (h)
FIG.14. Effect of ouabain on the affinityA in the frog skin (adapted from Owen et al., 197513).
P L -
1
0
1
2
3
TIME (h]
FIG.15. Effect of 2-deoxy-~-glucoseon the affinity A in the frog skin (adapted from Owen et al., 1975b). Note drop in the affinity of the experimental hemi-skins as compared to the control hemi-skins after addition of 2DG.
164
S. ROY CAPLAN AND ALVIN ESSIG
&,=Ka K =
EN^
K P + K ~
KP
FIG. 16. Equivalent circuit model for active sodium transport.
transport ENa? The parallel passive pathway may be represented by the conductance K~ (Fig. 16).The short-circuit current lo is then given by
(41
10 = KBEjqa
ENa is commonly considered to be the energetic factor of the active sodium transport system. Theoretical arguments, however, suggest that ENa is not a purely energetic quantity (Civan et al., 1966; Essig and Caplan, 1968).This is readily seen by relating the parameters of the equivalent circuit and thermodynamic representations. Thus, KB =
-(d/dA$)(FJNaa)= F2LNa
(A constant)
(5)
and 10 =
(FJNaa)Atb=O
= FLNad
(6)
Combining Eqs. (4),(5), and (6) shows that
ENa
=(
)(LNar/LNa)A
(7)
Thus, it appears that, in contradistinction to the thermodynamic affinity, the electromotive force of sodium transport comprises both permeability and energetic factors, even in the simplest conceivable equivalent circuit model. For more elaborate models, still more complex relationships between ENa and A would be anticipated. The dependence of ENaon permeability and energetic factors is confirmed by experimental findings. In accordance with the model of Fig. 2, for purposes of analysis, we assume that entry of sodium across the apical barrier is passive and that the electromotive force of sodium transportENais associated with the function of a sodium pump at the basolateral membrane. Any electrochemical potential difference that may appear across the apical membrane does not constitute an independent contribution to EN8,since it is a consequence of the operation of the pump.
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
V.
1 65
EXPERIMENTAL EVALUATION OF THE EQUlVALENT CIRCUIT MODEL
Many investigators have evaluated ENa using a variety of techniques (e.g., Ussing and Zerahn, 1951; Yonath and Civan, 1971; Larsen, 1973; Saito et al., 1974). We present here our results, using a simple and rapid means of evaluating Ka and thus EN, (Hong and Essig, 1976). For a system with both active and passive transepithelial pathways, the total conductance K is given by K = P $ K P
(8)
Since the diuretic amiloride inhibits active sodium transport by depressing entry of sodium into the active pathway, the residual conductance persisting after the abolition of short-circuit current by amiloride may be taken as KP. Combining Eqs. (4)and (8), ENais then given by the relation (9) ENa = Z,J(K - K ~ ) Thus, it is possible to define the dynamic effects of various substances on Ka and ENa b y means of frequent determinations of Zo and K , followed by the determination of KP at the end of the experiment. [It has been shown that with appropriate care, ~ P m a ybe maintained constant over extended periods (Saito and Essig, 1973).] This technique was used to study the effects of antidiuretic hormone (ADH), ouabain, amiloride, and 2DG in toad bladder. The results are shown in Table I (Hong and Essig, 1976).ADH (100 mU/ml) enhanced Ka markedly and depressed ENa slightly. Ouabain M) depressed both Ka and ENa. Amiloride (5 x 1 O - W ) depressed P and enhanced Of particular interest were the effects of the metabolic inhibitor 2DG. These are shown in Fig. 17. It is seen that 7.5 x 10h3M 2DG depressed K~ promptly without affecting ENa. The fact that a potent metabolic inhibitor markedly depresses active sodium transport without significant depression of ENa confirms our theoretical conclusion that ENa is not a purely energetic parameter. VI.
UTILITY OF THE THERMODYNAMIC AFFINITY A
For the reasons discussed above, we do not feel thatENacan be used to evaluate energetic factors. On the other hand, it may be helpful to This effect on ENadisagrees with the results of Yonath and Civan (1971) but agrees with the findings of Larsen (1973) in toad skins. Possibly the difference in our results was methodological.
166
S. ROY CAPLAN AND ALVIN ESSlG
TABLE I EFFECTS O F ANTIDIURETIC HORMONE, OUABAIN, AMILORIDE,AND 2-DEOXY-D-GLUCOSE ON 20, KB, AND ENa"
~~
ADH (100 mU/ml) Control Ouabain Control
M)
Amiloride (5 x lO-'M) Control 2DG (7.5 x Control
M)
7 7
17 17
30 30
8 8
0.41 f 0.06" 0.99 -e 0.02
0.54 1.04
5 5
6 6
60 60
7 7
2.25 1.04
f 0.11* f 0.01
2.53 f 0.19" 1.08 f 0.02
0.89 0.97
f 0.03e &
0.02
0.07* 0.03
0.76 0.95
k
2
0.07' 0.03
0.41 f 0.04" 1.00 + 0.01
0.21 f 0.05b 1.02 f 0.02
1.92 0.98
5
0.33d
0.33 f 0.03b 1.05 f 0.05
0.38 f 0.03" 1.11 f 0.08
0.89 f 0.05 0.94 f 0.05
&
2
f 0.02
~~
~
" For 2DC and ouabain, t
was the time at the end of the observation period when response was maximal. For amiloride and ADH, t was the time for peak response. The data are expressed as mean x r / x c = o f SEM. * Significant difference between control and experimental tissues: p < 0.001. Significant difference between control and experimental tissues: p < 0.02. Significant difference between control and experimental tissues: p < 0.005.
use A for this purpose. An appropriate substance for first studies is the hormone aldosterone. Aldosterone promotes substantial and stable enhancement of the rate of active sodium transport for extended periods. Because of its considerable importance, much study has been directed at the mechanism of its action. Three main possibilities that have been considered are pictured in Fig. 18, adapted from Fanestil et uZ. (1968). Mechanism 1is facilitation of sodium movement across the outer passive permeability barrier, as has been suggested by Sharp and Leaf (1966), CrabbC and Ehrlich (1968), and others. Mechanism 2 is facilitation of the linkage between oxidative metabolism and the phosphorylation of ADP, resulting in enhancement of the ATP/ADP ratio. In our terms, enhancement of the ATP/ADP ratio (or the ratio of some other pair of appropriate reactants) corresponds to an increase in A. Mechanism 3 involves a direct effect of aldosterone on the active transport mechanism, resulting in the more rapid pumping of sodium. The results of studies of Saito et al. (1973) concerning the mechanism of aldosterone action in the frog skin are shown in Figs. 19 and 20. Prior to the administration of aldosterone, the short-circuit current in paired control and experimental tissues was the same. Following overnight exposure to physiological concentrations of aldosterone, I 0
$
Control Exporimontal
0 0
0 0
0
0
0
0
P
P
P P P
P
40
50
60
m i
10
I
0.6
I
0.4
I i
I
0
10
20 Tim.
30 (min)
FIG. 17. Effect of2-deoxy-~-glucoseon short-circuit current I,,, active conductance and electromotive force of sodium transport EN, in the toad bladder (Hong and Essig, 1976).
K’,
OUTSIDE
CELL
INSIDE
FIG.18. Model of the active sodium transport system; possible mechanisms ofregulation of transport (Essig, 1975, adapted from Fanestil et al., 1968).
168
S. ROY CAPLAN AND ALVIN ESSIG
p CONTROL f 70-
Y E
W-
i
50ALDOSTERONE
0
- 401
I[
3
H '
ALDOSTERONE
I MEAN f SE, n . 8 )
30
!
20-
I
I
DAY I
DAY 2
FIG. 19. Effect ofaldosterone on short-circuit current lo in the frog skin (Saito et al., 1973).
in the control tissues was much the same as initially, whereas Zo in the treated tissues was appreciably greater (Fig. 19).These results are essentially the same as described by many others. Figure 20 shows the effect of aldosterone on A. As with Zo, the value of A was initially the same in control and experimental tissues. Following overnight exposure to aldosterone, A was appreciably greater in the treated tissues
p CONTROL
4
ALDOSTERONE
I MEAN f SE. n.8)
140-
120-
OJ
I
DAY
I
I
DAY
lI
FrG. 20. Effect of aldosterone on the affinity A in the frog skin (Saito et al., 1973).
169
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
than in the untreated tissues. These results suggest that one means whereby aldosterone may enhance the rate of active sodium transport is by increasing the affinity of the driving metabolic reaction, as in mechanism 2 above. [Alternatively, if indeed aldosterone selectively stimulates mitochondria-rich cells (Scott and Sapirstein, 1975), it is possible that the observations reflect activation of cells of high affinity.] This does not appear, however, to be the sole means by which aldosterone can influence transport. By combining electrical conductance and appropriate tracer isotope measurements, it is possible to . results of such studies by evaluate the effect of aldosterone on K ~ The Saito and Essig (1973) in toad bladder are shown in Fig. 21. As is seen, paired control and experimental tissues showed closely similar values of K~ prior to the administration of aldosterone. Some 2 hours after the administration of aldosterone, at a time when stimulatory effects on the current were observed, Ka of the aldosterone-treated tissues had increased comparably. These increments in current and Ka increased progressively during the final 6 hours of the experiment.
EXPERIMENTAL COMPARISON OF
VII.
ENa
AND A
In view of fundamental differences between EN,and A on theoretical grounds, it is of interest to compare the effects of various agents on these two parameters. Ideally such a comparison should be made simultaneously in a single tissue. This is not as yet possible. However, as mentioned, for several of the agents discussed above, we have
-
0:
5
2E
4 CONTROL
.5.
4 ALDOSTERONE
ALDOSTERONE
( MEAN* SE. n.1z-141
I
.4-
.3-
T
-
E
-., Y
.2.I
I
s
0
1
2
3
4
5
6
7
8
TIME ( HOURS)
FIG.21. Effect ofaldosterone on the active conductance I? of the toad bladder (Saito and Essig, 1973).
170
S. ROY CAPIAN AND ALVIN ESSIG
determinations of A in frog skin and determinations of ENain toad bladder. As is seen in Table 11, four agents appear to have discrepant effects on ENa and A. (Present techniques do not permit a determination ofA during the transient response to ADH.) It is appreciated that no definitive conclusions can be deduced from studies in different tissues under different conditions. Nevertheless, the results support the theoretical inference that ENa is not simply an energetic parameter, but reflects kinetic factors as well. All of this does not imply that ENais without physiological significance. Since ENa is the maximal electrochemical potential difference that can be achieved by the sodium pump, its modulation may be of importance in the regulation of intracellular sodium concentration. Although dramatic changes in ENa were observed in this study, the demonstration that the metabolic inhibitor 2DG rapidly depresses P , but not ENa,suggests dynamic interaction between permeability and energetic factors so as to protect E N a . Similarly, enhancement by aldosterone of K~ and A, but not ENa,may represent a mechanism for stimulation of transepithelial sodium transport without disturbance of intracellular sodium levels (Lipton and Edelman, 1971).
VIII.
SOME GENERAL COMMENTS
At this point, it is useful to look back at what has been accomplished in order to see where we stand. Clearly most experimentalists have a need for conceptual models as a basis for the planning and interpretation of experiments. In the study of epithelial transport, this need has been fulfilled by the equivalent circuit model. Whatever its shortcomings, it indicated the importance of both kinetic and energetic factors and has led to experiments attempting to differentiate between the two. However, the description of the system in these terms fails to take explicitly into consideration the existence of two flows. As such, the output parameter, ENa,must necessarily fail to give a precise value for the energetic parameter of prime interest-the free energy of the metabolic input process driving transport. The availability of NET permits an appropriate approach to two-flow, or for that matter n-flow, systems. This being the case, we would suggest that NET should now replace the equivalent circuit model as a basis for experimental design and interpretation. We make this suggestion despite the fact that the applicability of linear NET to biological systems remains to be completely tested, taking the view that a theoretical construct may
-4
I
rn
g D
a4
TABLE I1
EFFECTSOF VAHOUS AGENTS ON EN, AND A"
vl
EN, Agent Ouabain Amiloride 2DG Aldosterone
A
Dose
t
EN,
Source
Dose
10-'M
5-30 min 1-60 min
J
5 x lo-' M
Hong and Essig (1976) Hong and Essig (1976)
7.5 X 1 W M 5 x IO-'M
5-60min 1-6 hr
-
10-7 M 10-'-10-5 M 10-7-10-5 M 1.6 x lo-* M 5x M
f
-
Hong and Essig (1976) Saito and Essig (1973)
t (hr)
2.5 1 4 1 14-18
A
Source
-
Owen et al (197513) Saito et al. (1973) Saito et al. (1973) Owen et al. (1975b) Saito et al. (1973)
f
J. f
'ENa was determined in toad bladders,A was determined in frog skins. Dash signifies no change; upward arrow, a significant increase; downward arrow, a significant decrease.
8 E 2 8 7
172
S. ROY CAPIAN A N D ALVIN ESSlG
often be particularly useful before its range of validity and its molecular basis are fully known. In order to analyze the system in terms of linear NET, it is necessary, as emphasized above, to carry out experiments according to appropriate well-defined protocols. For example, to demonstrate linearity of sodium transport in the electrical potential difference, it is necessary to perturb A$ symmetrically and for appropriate periods. It is appreciated that others employing different protocols have often demonstrated nonlinearity. While such nonlinearity may well be of biological significance, it is not readily amenable to full analysis with the existing formulations. Hence, our emphasis on the study of the system under conditions of linearity such that a self-consistent analysis is possible. For the present, experimental limitations have necessitated the study of transport and metabolism under conditions that are not always identical. Thus, whereas a reliable rate of oxygen consumption cannot be determined in less than 6 min with our current technique^,^ the dependence of the rate of active sodium transport on A$ is more conveniently determined by perturbing AJI at intervals of seconds. Since both types of protocol result in quasi-steady states we have combined the various data for purposes of analysis. It is possible that different protocols would lead to different numerical results, and perhaps even different conclusions, reflecting the response of the system to its experimental constraints. The solution of this difficulty must await more sophisticated techniques currently under development. Granting the present limitations in the application of NET, the determination of all the L’s and A provides a complete thermodynamic characterization of the system in different states of interest. Such data are potentially of use in attempts to elucidate mechanisms in that they impose constraints on proposed models. Of particular interest is the thermodynamic affinity A , which as discussed presumably reflects some critical substrate-product ratio, which remains constant on brief perturbation of A$. Such constancy might imply either large metabolic pools or some regulatory mechanism. Whatever the case, the affinity is a characteristic property of the system whose study is of fundamental importance in the analysis of energetics. The rate of oxygen consumption is determined by evaluating the slope of the plot of oxygen tension vs time 4 to 6 min after perturbingA+. Approximately 2 min after each perturbation of A+ the slope becomes appreciably constant and remains so during the subsequent 4 min.
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
IX.
173
CONCLUSIONS
1. Classical approaches to the energetics of active transport invoke doubtful and invalid assumptions. 2. The classical equivalent circuit model is inappropriate, since it is insufficient to describe incompletely coupled two-flow processes. On theoretical grounds, the “electromotive force of sodium transport” ENacomprises both kinetic and energetic factors. 3. Linear nonequilibrium thermodynamics (NET), although incompletely tested, appears to provide a valid framework for the design and analysis of experiments. In contrast to ENa,the thermodynamic affinity A represents a purely energetic quantity. 4. There is no unique stoichiometric ratio that relates the rates of active sodium transport and oxygen consumption under all circumstances. I n the short-circuited state, suprabasal Na/Oz may vary from animal to animal. For incompletely coupled flows, Na/Oz varies with the transmembrane electrical potential difference. 5. Rates of active sodium transport and oxygen consumption show the postulated linear dependence on the electrochemical potential difference of sodium. Although it has not yet been possible to test for linearity in A, various considerations suggest its likelihood. 6. Whereas A behaves appropriately as an energetic parameter in model experiments, EN, does not. 7 . The NET approach should facilitate understanding of mechanisms altering sodium transport. For example, aldosterone appears to modify both permeability and energetic factors. ACKNOWLEDGMENTS This work was supported by the USPHS (Grant HL 14322 to the Harvard-MIT Program in Health Sciences and Technology), the U.S.-Israel Binational Science Foundation, Jerusalem, Israel, the National Kidney Foundation, the Medical Foundation, Boston, Massachusetts, and NSF Grants GB 24697 and 40704. REFERENCES Al-Awqati, Q., Beauwens, R., and Leaf, A. (1975).Coupling of sodium h-ansport to respiration in the toad bladder.]. Membr. Biol. 22,91. Bentley, P. J. (1968).Amiloride: A potent inhibitor of sodium transport across the toad bladder. J. Physiol. (Londvn) 195, 317. Blumenthal, R., Caplan, S. R., and Kedem, 0. (1967). The coupling of an enzymatic reaction to transmembrane flow of electric current in a synthetic “Active Transport” system. Biophys. J. 7, 735. Caplan, S. R. (1971). Nonequilibrium thermodynamics and its application to bioenergetics. Cztrr. T o p Bivenerg. 4, 1.
174
S. ROY CAPIAN AND ALVIN ESSIG
Civan, M. M., Kedem, O., and Leaf, A. (1966).Effect of vasopressin on toad bladder under conditions of zero net sodium transport. Am. J . Physiol. 211,569. CrabbB, J., and Ehrlich, E. N. (1968).Amiloride and the mode of action of aldosterone on sodium transport across toad bladder and skin. Pflueger’s Arch. 304,284. Danisi, G., and Vieira, F. L. (1974).Nonequilibrium thermodynamic analysis of the coupling between active sodium transport and oxygen consumption. J . Gen. Physiol. 64,372. Essig, A. (1975).Energetics of active transport processes. Biophys. J . 15,651. Essig, A., and Caplan, S. R. (1968).Energetics of active transport processes. Biophys. J . 8, 1434. Fanestil, D. D., Herman, T. S., Fimognari, G. M., and Edelman, I. S. (1968).Oxidative metabolism and aldosterone regulation of sodium transport. In “Regulatory Functions of Biological Membranes” (J. Jarnefelt, ed.), p. 177.Elsevier, Amsterdam. Handler, J. S., Preston, A. S., and Orloff, J. (1969).The effect of aldosterone on glycolysis in the urinary bladder of the toad. J . Biol. Chem. 244,3194. Handler, J. S., Preston, A. S., and Orloff, J. (1972).Effect ofADH, aldosterone, ouabain, and amiloride on toad bladder epithelial cells. Am. J . Physiol. 222, 1071. Heinz, E. (1974).Coupling and energy transfer in active amino acid transport. Curr. Top. Membr. Transp. 5, 137. Hong, C. D., and Essig, A. (1976).Effects of 2-deoxy-~-glucose,amiloride, vasopressin, and ouabain on active conductance and ENain the toad bladder. J . Membr. Biol. 28, 121. Hoshiko, T., and Lindley, B. D. (1967).Phenomenological description of active transport of salt and water. J . Gen. Physiol. 50, 729. Jardetzky, O.,and Snell, F. M. (1960).Theoretical analysis oftransport processes in living systems. Proc. Natl. Acad. Sci. U S A . 46,616. Kedem, 0. (1961).Criteria of active transport. In “Membrane Transport and Metabolism” (A. Kleinzeller, and A. Kotyk, eds.), p. 87.Academic Press, New York. Kedem, O., and Essig, A. (1965).Isotope flows and flux ratios in biological membranes. J . Gen. Physiol. 48, 1047. Kedem, O., and Katchalsky, A. (1958).Thermodynamic analysis of the permeability of biological membranes to non-electrolytes. Biochim. Biophys. Acta 27,229. Lahav, J., Essig, A., and Caplan, S. R. (1976).The thermodynamic degree of coupling between metabolism and sodium transport in frog skin. Biochim. Biophys. A C ~ Q 448,389. Lang, M. A,, Caplan, S. R., and Essig, A. (1977).Sodium transport and oxygen consumpQ tion in toad bladder-a thermodynamic approach. Biochim. Biophys. A C ~464,571. Larsen, E. H.(1973).Effect of amiloride, cyanide and ouabain on the active transport pathway in toad skin. In “Transport Mechanisms in Epithelia” (H. H. Ussing, and N. A. Thorn, eds.) p. 131.Academic Press, New York. Leaf, A., Anderson, J., and Page, L. B. (1958).Active sodium transport by the isolated toad bladder. J . Gen. Physiol. 41,657. Lipton, P., and Edelman, I. S. (1971). Effects ofaldosterone and vasopressin on electrolytes of toad bladder epithelial cells. Am. J . Physiol. 221,733. Mandel, L. J., and Curran, P. F. (1972). Response of the frog skin to steady-state voltage clamping. 1. The shunt pathway.J. Gen. Physiol. 59,503. Miller, D. G.(1960).Thermodynamics of irreversible processes. The experimental verification of the Onsager reciprocal relations. Chem. Reo. 60, 15. Owen, A., Caplan, S. R., and Essig, A. (1975a).The interaction of 2-deoxy-Dglucose and glucose: Effects on the short-circuit current of frog skin. Biochim. Biophys. Acta 389,407.
THERMODYNAMICS OF ACTIVE SODIUM TRANSPORT
1 75
Owen, A., Caplan, S. R., and Essig, A. (1975b).A comparison of the effects of ouabain and 2-deoxy-D-glucose on the thermodynamic variables of the frog skin. Biochim.
Biophys. Actu 394,438. Prigogine, I. (1961). “Thermodynamics of Irreversible Processes.” Wiley, New York. Rapoport, S. I. (1970).The sodium-potassium exchange pump: relation of metabolism to electrical properties of the cell. I. Theory. Biophys. J. 10, 246. Rottenberg, H. (1973).The thermodynamic description of enzyme-catalyzed reactions.
Biophys. J. 13, 503. Saito, T., and Essig, A. (1973).Effect ofaldosterone on active and passive conductance and EN,in the toad b1adder.J. Membr. Biol. 13, 1. Saito, T., Essig, A., and Caplan, S. R. (1973).The effect ofaldosterone on the energetics of sodium transport in the frog skin. Biochim. Biophys. Actu 318,371. Saito, T., Lief, P. D., and Essig, A. (1974).Conductance of active and passive pathways in the toad bladder. Am. J. Physiol. 226, 1265. Scott, W. N., and Sapirstein, V. S. (1975). Identification of aldosterone-induced proteins in the toad’s urinary bladder. Proc. Natl. Acad. Sci. U.S.A. 72,4056. Sharp, G. W. G., and Leaf, A. (1966). Mechanism of action of aldosterone. Physiol. Rev. 46,593. Ussing, H. H. (1960).“The Alkali Metal Ions in Biology.” Springer-Verlag, Berlin and New York. Ussing, H. H., and Zerahn, K. (1951).Active transport of sodium as the source of electric current in the short-circuited isolated frog skin. Actu Physiol. Scund. 23, 110. Veech, R. L., Raijam, L., and Krebs, H. A. (1970).Equilibrium relations between the cytoplasmic adenine nucleotide system and nicotinamide-adenine nucleotide system in rat liver. Biochem. J . 117,499. Vieira, F. L., Caplan, S. R., and Essig, A. (1972a).Energetics of sodium transport in frog skin. I. Oxygen consumption in the short-circuited state.J. Gen. Physiol. 59,60. Vieira, F. L., Caplan, S. R., and Essig, A. (197213).Energetics of sodium transport in frog skin. 11. The effects of electrical potential on oxygen consumpti0n.J. Gen. Physiol.
59, 77. Yonath, J., and Civan, M. M. (1971). Determination ofthe driving force ofthe Na+ pump in toad bladder by means of vas0pressin.j. Membr. Biol. 5,366. Zerahn, K. (1956). Oxygen consumption and active sodium transport in the isolated and short-circuited frog skin. Actu Physiol. Scand. 36, 300.
This Page Intentionally Left Blank
Anaerobic Electron Transfer and Active Transport in Bacteria W l L N . KONlNGS AND J O H A N N E S BOONSTRA Department of Microbiology Biological Center University Groningen Haren, The Netherlands
.......................................................
on Transfer Systems .................................. A. Nitrate Respiration ............................................... B. Furnarate Reduction ....................................... 111. Phosphorylation Coupled to Electron Transfer .................. A. Introduction ..................................................... B. ATP Synthesis Coupled to Nitrate Respiration ..................... C. ATP Synthesis Coupled to Fumarate Reduction .................... IV. Anaerobic Active Transport ........................................... A. Introduction ..................................................... B. Anaerobic Active Transport Coupled to Nitrate Respiration ...................................................... C. Anaerobic Active Transport Cou Reduction ................... ............................... D. Involvement of ATP .............................................. E. Involvement of a Proton-Motive Force ............................. References ...........................................................
177 180 180 190 195 195 197 198 199 199 204 208 211 213 219
I. INTRODUCTION
The energy present in an oxidizable substrate can be released through a graded series of reversible oxidation-reduction reactions, with reducing equivalents transferred to a terminal electron acceptor via the electron carriers of an electron transfer system. The electron transfer systems of bacteria are basically similar to those of higher organisms, but the bacterial systems are much more varied with respect to electron carriers and terminal electron acceptors. In higher organisms, only oxygen can function as terminal electron acceptor, but 1 77
178
WIL N. KONINGS AND JOHANNES BOONSTRA
in bacteria several inorganic and organic compounds, in addition to oxygen, may perform this function. Bacterial electron transfer systems vary from very simple to very complex systems (for review, see White and Sinclair, 1971), and a wide diversity exists in the electron carriers that participate in these systems. Among them are dehydrogenases, quinones, nonheme iron proteins, flavins, several types of cytochromes, and terminal oxidases. In some bacteria, these electron carriers are arranged in a linear chain of electron transfer intermediates; in others, branched electron transfer systems occur (White and Sinclair, 1971). In obligately aerobic bacteria, oxygen is the only terminal electron acceptor and cytochromes of the a , d , and o types can function as terminal oxidases. The electron transfer system of these bacteria, the respiratory chain, resembles most closely the electron transport chain in mitochondria. In facultative anaerobes, the situation with regard to terminal electron acceptors is more complex. Under aerobic conditions, these organisms contain a functional respiratory chain, whereas under anaerobic conditions electron transfer systems that are coupled to electron acceptors other than oxygen may be present. Strictly anaerobic bacteria are defined as organisms that can never use oxygen as terminal electron acceptor. In many of these organisms, the presence of anaerobic electron transfer systems has been demonstrated. A wide variety of terminal electron acceptors may be used in facultative and strict anaerobes, but detailed information about the role of only a few compounds is available. Among them are the nitrogen compounds, nitrate and nitrite; the sulfur compounds, sulfate, sulfite (Postgate, 1965; Barton et al., 1972; LeGall and Postgate, 1973), thiosulfate, and tetrathionate (De Groot and Stouthamer, 1970b); and the organic compounds, fumarate and carbonate. In Fig. 1, a generalized scheme of bacterial electron transfer systems is presented as a linear series of electron carriers. In many cases, more than one sequence of carriers between oxidizable substrates and terminal electron acceptors can be operative, and these different pathways of electron flow may be interconnected to various degrees. In some strict anaerobes there are very simple systems, that do not contain cytochromes as electron carrier. In addition to the respiratory chain and anaerobic electron transfer systems, cyclic electron transfer systems occur in phototrophic bacteria (Parson, 1974). Electrons transferred through these pathways are derived from reduced bacteriochlorophyll in a light-dependent process. They are then transferred to acceptors, whose nature still is a point of discussion, and thereupon via quinones and cytochromes
179
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
nitrite
substrates-dehydrogenases-
quinones ~ c y t o c h r o m e -terminal s
fum r a te
FIG.1. Generalized scheme of electron transfer systems in bacteria.
back to oxidized chlorophyll. These systems may be operative under aerobic as well as anaerobic conditions. All these electron transfer systems are membrane-bound, and the free energy (AG) released by the electron flow through these systems can be used for several energy-dependent processes in the membrane. The role of the membranes in energy metabolism has been studied extensively in mitochondria and chloroplasts, and the reader is directed to several excellent review articles (Lardy and Ferguson, 1969; Pressman, 1970; Racker, 1970; Skulachev, 1971; Slater, 1971; Van Dam and Meyer, 1971; Harold, 1972) for detailed information. In bacterial cytoplasmic membranes, the same energy-dependent processes occur: (i) synthesis of ATP (oxidative phosphorylation); (ii) accumulation of substrates and ions against concentration gradients; (iii) reversal of the direction of oxidation; (iv) transhydrogenation, the reduction of NADP by NADH (Fig. 2). These processes have been studied extensively in bacteria grown under aerobic conditions, and the central role of the respiratory chain in these processes has been clearly demonstrated. In contrast to the many studies that have been done on electron
-
NADP Tronshvdroaenation
-
/ 1-
Dissipation by Uncouplers
ADPtPi
Cotion, Anion Translocalion
DCCD
+ ATP
FIG.2. Relation between respiratory chain and energy-dependent processes. From Harold (1972), with permission.
180
WIL N. KONINGS AND JOHANNES BOONSTRA
transfer in the respiratory chain, the information available on anaerobic electron transfer systems is often very limited. The situation with regard to processes dependent on the energy supplied by anaerobic electron transfer systems is even worse, and until recently attention was paid almost exclusively to the synthesis of ATP. This rather restricted information might have led to an underestimation of the role of electron transfer systems in the metabolic machinery of anaerobically grown bacteria. It is the aim of this discussion to summarize the information available on some anaerobic electron transfer systems and to present evidence that these systems are coupled to active transport of metabolites in a way similar to that already demonstrated for the respiratory chain. Inasmuch as information about anaerobic active transport coupled to electron flow is at present available only for the anaerobic electron transfer systems in which nitrate or fumarate functions as a terminal electron acceptor, we will focus our attention only on those systems. II. ANAEROBIC ELECTRON TRANSFER SYSTEMS A. Nitrate Respiration
Anaerobic electron transfer to nitrate as terminal electron acceptor is called nitrate respiration (Taniguchi et al., 1956). The terminal oxidase involved in this electron transfer system is nitrate reductase, which catalyzes the reduction of nitrate to nitrite. Nitrate respiration has been demonstrated in strictly aerobic bacteria (Pseudomonas),as well as in facultatively and strictly anaerobic bacteria (for definitions, see Section I). The reduction product of nitrate respiration, nitrite, is highly toxic. In a few organisms, nitrite can be further reduced to the nontoxic free nitrogen. This process, termed denitrification, involves, in addition to nitrate respiration, another electron transfer system in which nitrite reductase functions as terminal oxidase. In addition to this dissimilatory nitrate reduction, nitrate can also be reduced by a process termed assimilatory nitrate reduction (Pichinoty, 1960; Chang and Morris, 1962; Forget and Pichinoty, 1964; Van’t Riet et al., 1968). Through this reduction process, nitrate, which is reduced to ammonia, can serve as a source of cell nitrogen. A few organisms can carry out both assimilatory nitrate reduction and nitrate respiration; others can perform only one of these processes. The first step in assimilatory nitrate reduction is mediated also by a nitrate reductase,
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
181
and the possible involvement of the same enzyme in both pathways is still a point of discussion. (Pichinoty, 1960, 1965; Forget and Pichinoty, 1964; Hadjipetrou and Stouthamer, 1965; Van’t Riet et al., 1968; Showe and DeMoss, 1968; Kodama et al., 1969; Lam and Nicholas, 1969b; De Groot and Stouthamer, 1970a,c; Van Hartingsveldt and Stouthamer, 1974). The respiratory nitrate reductase forms a complex with other components of the electron transfer system (Taniguchi and Itagaki, 1960; Itagaki et al., 1961b; Ruiz-Herrera and DeMoss, 1969; Ruiz-Herrera et al., 1969). This complex is membrane-bound and can be isolated in a particulate fraction (Itagaki et al., 1961b, 1962; Naik and Nicholas, 1966; Lam and Nicholas, 1969b,c; Konings and Kaback, 1973a; Konings et al., 1975; Boonstra et aE., 1975a). The components of the electron transport pathway to nitrate reductase vary from organism to organism. The electrons are donated to this pathway via primary dehydrogenases in the same way as in the aerobic respiratory chain. In general, the dehydrogenases are inducible enzymes, and the nature of the best electron donor depends, therefore, to a large extent on the growth conditions. The involvement of cytochromes has been demonstrated in nitrate respiration of many organisms. The cytochromes involved are usually of the b type (Itagaki et al., 1961b; Ruiz-Herrera and DeMoss, 1969; Ruiz-Herrera et aZ., 1969; De Groot and Stouthamer, 1970a; De Vries et al., 1974), but, in some organisms, cytochromes of the c type seem to function as electron carriers (Fewson and Nicholas, 1961b). Information on the role of other electron carriers in nitrate respiration is available for only a few organisms. In Escherichia coli (Nicholas and Nason, 1955), flavoproteins have been shown to mediate the transfer of electrons from NADH-dehydrogenase to nitrate reductase, but this electron carrier does not seem to play a role in the transfer from formate dehydrogenase to nitrate reductase (Linnane and Wrigley, 1963). A role of quinones in nitrate respiration was clearly demonstrated in the transfer of electrons from NADH to cytochrome b in Klebsiella aerogenes (Knook and Planta, 1971, 1973). In addition to dehydrogenases, flavoproteins, quinones, cytochromes, and nitrate reductase, other components, such as lipid factors, seem to be required in some organisms for the formation of a functional nitrate respiration complex (Itagaki et al., 1961a,b). Genetic studies contributed to a large extent to our understanding of nitrate respiration inE. coli and other organisms. Mutants defective in nitrate respiration have been selected for their ability to grow anaerobically on a nutrient medium containing chlorate, a compound that can be reduced by nitrate reductase to the toxic compound chlo-
182
WIL N. KONINGS AND JOHANNES BOONSTRA
rite (Piiichaud et al., 1967). These mutants map at distinct positions of the E. coli chromosome and are designated as chl A to chl G (Adhya et al., 1968; Venables and Guest, 1968; Guest, 1969; Casse, 1970; Glaser and DeMoss, 1972; Stouthamer, 1967a,b, 1969; De Groot and Stouthamer, 1969, 1970~). Ruiz-Herrera et al. (1969) have isolated nitrate reductase mutants by direct screening of colonies derived from cells surviving after nitrosoguanidine treatment, for their ability to utilize formate as an electron donor for nitrate reduction. Mutants obtained by this technique are designated NR- (Venables and Guest, 1968) or nar (Van Hartingsveldt et al., 1971; Van Hartingsveldt and Stouthamer, 1973). Several of the mutants described above have been characterized biochemically, and in some cases the defects have been located in specific components of the nitrate respiration system. The following section will be concerned with the properties of nitrate respiration in E . coli because this system has been studied most extensively, and at present, is best understood. 1. ESCHERZCHZA COLZ
In E. coli grown anaerobically in the presence of nitrate, formate serves as the most effective electron donor for nitrate respiration (Taniguchi and Itagaki, 1960; Wimpenny and Cole, 1967; Cole and Wimpenny, 1968; Ruiz-Herrera and DeMoss, 1969; Lester and DeMoss, 1971). The oxidation of this substrate can occur by two inducible formate dehydrogenases, which are involved in two distinct enzyme systems: formate dehydrogenase (N) which is involved in the nitrate respiration system, and formate dehydrogenase (H) which is a coniponent of the formate-hydrogen-lyase pathway. In this pathway formate dehydrogenase (H) is coupled to hydrogenase via some unidentified electron carriers (Gest and Peck, 1955; Gray and Gest, 1965; RuizHerrera and Alvarez, 1972),and electrons from formate are transferred to H+ (Fig. 3). The formate dehydrogenase activities are influenced differently by the growth conditions. Formate dehydrogenase (N) activity is regulated by the level of oxygen and by the presence of nitrate in the growth medium (Itagaki et al., 1962; Ruiz-Herrera and DeMoss, 1969; Ruiz-Herrera and Alvarez, 1972; Ruiz-Herrera et al., 1972). Formate dehydrogenase (H) is induced under anaerobic conditions in a medium of low pH that contains formate. The enzyme activity is repressed by oxygen and by nitrate in the growth medium (Pichinoty, 1962; Gray et al., 1966; Wimpenny and Cole, 1967; Ruiz-Herrera and
183
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA FORMATE-NITRATE REDUCTASE PATHWAY HCOOH
1 1-ytochrome
CO2
lormale dehydrogenasc
NO,
bsSsnitrate rcduclasc
Lo2
FORM ATE-HY DROGENLY ASE PATHWAY
HCOOH
co*
formate dehydrogenaw
IX I l--lX*l-
hydrogenasc
(:”
FIG. 3. Scheme for the formate-nitrate reductase pathway and the formate-hydrogenlyase pathway of Escherichia coli. From Glaser and DeMoss (1971), with permission.
Alvarez, 1972; Ruiz-Herrera et al., 1972). The activities of both enzymes can be measured separately because the enzymes transfer electrons to different redox dyes. Formate dehydrogenase (N) transfers electrons to redox dyes, which accept two electrons, like methylene blue and phenazine methosulfate (Itagaki et al., 1962; Gray et al., 1966; Ruiz-Herrera and DeMoss, 1969; Ruiz-Herrera and Alvarez, 1972; Ruiz-Herrera e t al., 1972); formate dehydrogenase (H) reacts with benzylviologen, an electron acceptor with a low redox potential, which is reduced by one electron (Ruiz-Herrera et al., 1972). In addition to these two formate dehydrogenase activities, formate oxidase activity is also found in E . coli. Evidence has been presented that this enzyme activity is due to the presence of an auto-oxidizable cytochrome b , in the electron transfer system from formate dehydrogenase (N) to nitrate reductase (Ruiz-Herrera and DeMoss, 1969; Itagaki et al., 1961b, 1962). Both formate dehydrogenase (N) and formate dehydrogenase (H)are membrane-bound and can be solubilized in an active form by treatment with a non-ionic detergent (RuizHerrera et al., 1972). The properties of membrane-bound and partially purified enzymes have been determined, and distinct differences are observed in the kinetic properties of formate dehydrogenass(N) and formate dehydrogenase (H) (Ruiz-Herrera et al., 1972). These observations indicate that formate dehydrogenase (N) and formate dehydrogenase (H) are different enzymes. On the other hand, genetic evidence has been presented which suggests that a single formate dehydrogenase is involved in these enzyme activities (Casse, 1970; Glaser and DeMoss, 1972; O’Hara et al., 1967; Venables et al., 1968; RuizHerrera et al., 1969; Guest, 1969). The formation of formate dehydrogenase (N), formate dehy-
184
WIL N. KONINGS AND JOHANNES BOONSTRA
drogenase (H), and also formate oxidase is stimulated by the addition of molybdate and selenite to the growth medium (Pinsent, 1954; Fukuyama and Ordal, 1965; Lester and DeMoss, 1971; Enoch and Lester, 1972). It is not clear whether both elements are essential components of the enzymes themselves or are merely required for the formation of active enzyme complexes. It has been demonstrated that molybdenum is an essential component of nitrate reductase (Taniguchi and Itagaki, 1960; MacGregor et al., 1974) and that the addition of molybdate to the growth medium stimulates the formation of both nitrate reductase and cytochrome b , (Lester and DeMoss, 1971).Additional supplementation of the growth medium with selenite is required for the formation of the enzyme system that permits formate to serve as an effective electron donor for nitrate reduction, indicating that selenite plays a specific role in formate dehydrogenase (N) itself (Lester and DeMoss, 1971; Enoch and Lester, 1972; Shum and Murphy, 1972). In addition to formate, other substrates such as lactate, ~ - a glycerol-phosphate, and NADH can function as electron donors for nitrate reduction in cells grown under appropriate conditions (Nicholas and Nason, 1955; Taniguchi and Itagaki, 1960; Cole and Wimpenny, 1968; Lester and DeMoss, 1971; Boonstraet al., 1975a).FAD seems to function as electron carrier in the transfer of electrons from NADH to nitrate reductase (Nicholas and Nason, 1955). In a particulate fraction from E . coli, the anaerobic oxidation of NADH in the presence of nitrate was stimulated markedly by catalytic amounts of FAD, and chemically reduced FAD served as an electron donor for the reduction of nitrate in this fraction (Nicholas and Nason, 1955). A similar involvement of FAD has not been demonstrated in the formatedehydrogenase-nitrate-reductase system (Taniguchi and Itagaki, 1960; Linnane and Wrigley, 1963). Electrons are transferred from the dehydrogenases via cytochromes to nitrate reductase. A role of cytochrome b , in the formatedehydrogenase-nitrate-reductase complex was suggested by Sat0 and Egami in 1949. More evidence was obtained from studies with a partially purified formate-dehydrogenase-nitrate-reductase complex (Itagaki et al., 1961b, 1962). This particulate fraction contained cytochrome b,. The latter, in turn, was fully reduced under anaerobic conditions in the presence of formate and vitamin K, or a natural lipid factor. Subsequent addition of a small amount of nitrate resulted in rapid and complete oxidation of reduced cytochrome, with the absorption spectrum returning to that of the aerobic preparation. At the same time, the accumulation of nitrite was detected in the reaction mixture.
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
185
Moreover, formate dehydrogenase purified from this particulate fraction also contained cytochrome b,; this same cytochrome was found in nitrate reductase isolated by heat treatment of this preparation (Itagaki et al., 1961a). In addition, MacGregor (1975b) demonstrated the presence of cytochrome b , apoprotein in purified nitrate reductase. Genetic evidence has also been presented for the involvement of cytochrome b in the formate-dehydrogenase-nitrate-reductase complex. Several NR- mutants, isolated by Ruiz-Herrera et al. (1969), contained diminished levels of cytochrome bw5. Ruiz-Herrera and DeMoss (1969) studied the reduction of cytochromes in wild types and NR- mutants of E . coli and presented evidence for the participation of two distinct cytochromes of the b type with different redox potentials. From cytochrome b , the electrons are transferred to nitrate reductase. T h e synthesis of nitrate reductase in anaerobically grown E . coli can be increased 20-fold by the addition of nitrate to the growth medium (Showe and DeMoss, 1968). The enzyme is membranebound (Taniguchi e t al., 1956; Taniguchi and Itagaki, 1960; Showe and DeMoss, 1968; MacGregor and Schnaitman, 1971) and has been solubilized (MacGregor and Schnaitman, 1971; Taniguchi and Itagaki, 1960; MacGregor et al., 1974; MacGregor, 1975a) and purified to homogeneity (MacGregor, 1975a). The enzyme appears to be spherical and has a molecular weight (MW) of about lo6daltons (Taniguchi and Itagaki, 1960; MacGregor et al., 1974; MacGregor, 1975a,b). It is composed of three different subunits: subunit A with MW of 142,000 daltons, subunit B with a MW of 58,000 daltons, and subunit C, which was identified as the cytochrome b , apoprotein. This apoprotein has a MW of 19,500 daltons. The subunits are present in a ratio of A: B : C of 1: 1:2. The most likely structure of the enzyme is a 16-subunit structure containing four subunits of A, four subunits of B, and eight subunits of C (MacGregor, 1975a,b). Four moles of molybdenum are bound per mole of enzyme (MacGregoret al., 1974). Molybdenum can by itself slowly reduce nitrate to nitrite, and studies on other molybdenum-containing enzymes have indicated that molybdenum is directly involved in electron transfer and undergoes a change in valence during this process (Guymon and Spence, 1966; Forget and Dervartanian, 1972; Stiefel, 1973). In addition to molybdenum, the enzyme contains about 40 nonheme iron atoms per molecule (Taniguchi and Itagaki, 1960), the function of which has not been elucidated. This involvement of heavy metals in nitrate reductase explains the inhibitory effect of chelating agents such as cyanide and azide (Taniguchi et al., 1956; Taniguchi and Itagaki, 1959; Forget, 1974). In
186
WIL N. KONINGS AND JOHANNES BOONSTRA
contrast to earlier suggestions (Nicholas and Nason, 1955), the enzyme does not seem to contain flavins and therefore is not a metalloflavoprotein (Taniguchi and Itagaki, 1960; Forget, 1974). The properties of several of the chl mutants can be explained by deletions in one or more components of the nitrate reductase enzyme. Chl A mutants, for instance, contain normal amounts of subunits A and B and diminished amounts of subunit C, but lack a molybdenumcontaining cofactor (Mo-X). Chl B mutants also lack Mo-X (MacGregor, 1975c), but as it accumulates in the cytoplasm of chl B mutants, the defect must lie in the attachment or insertion of Mo-X into the enzyme. Recently, Riviere et al. (1975) isolated the product of the chl B gene from chl A mutants ofE. coli. The purified protein, which has been termed "FA-factor," has a MW of 35,000 daltons, and evidence was presented that this protein is required for a functional reconstitution of the components of the nitrate reductase complex. Chl C, too, has a defective enzyme, and evidence has been presented that chl C is the locus for one of the structural genes of nitrate reductase (Guest, 1969), most likely of subunit A (MacGregor, 1975~). The defect in chl E also involves one of the structural genes of the enzyme. Since these mutants do not make cytochrome b , (MacGregor, 1975c), they presumably contain subunit C. Table I shows the distribution of nitrate reductase components in mutants and wild-type strains of E. coli. Addition of missing components to cytoplasmic fractions of these mutants results in reconstitution of nitrate reductase activity, as demTABLE I DISTRIBUTION OF NITRATE REDUCTASE COMPONENTS IN MUTANT AND WILDTYPE STRAINSOF Escherichia coli" Nitrate reductase subunits
Strain WildType chi A
chi B chi C chi E
Cytoplasm
-
Membrane
+ + + -
-
Mo-X cofactor cytoplasm
-
-
++ -
" From MacGregor (1975c), by permission.
Membrane
+
-
-
Association factor Cytoplasm
+ + + +
Membrane
-
-
-
187
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
TABLE I1
RECONSTITUTION OF NITRATEREDUCTASEACTIVITY AND FORMATIONO F MEMBRANEPARTICLES FROM CYTOPLASMIC EXTRACTSOF CHLORATE-RESISTANT MUTANTS OF Escherichia COP Nitrate reductase activity in
Cytoplasmic fractions combinedb
chl A chl A chl C chl B chl B chl B
+ chl C
+ chl E + chl E + chl A + chl C + chl E
Particulate fraction from Incubated incubated mixturec mixturesd (nmoles/mg protein x min)
(0.1 <0.1
<0.1 41 38 28
Protein in particulate fraction (mg)
3.0 3.6 4.1 3.2 3.1 4.2
From MacCregor and Schnaitman (1973),by permission. Cytoplasmic fractions were obtained by breakage of the cells in a French pressure cell and removal of particulate components by 1-hour centrifugation at 200,000 g. None of the cytoplasmic fractions contained detectable nitrate reductase activity prior to mixing. The cytoplasmic fractions contained 12-20 mg protein/ml. Each mixture consisted of 2 ml of each cytoplasmic fraction. The mixtures were M MgCl,. This incubation incubated for 2 hours at 32°C in the presence of 1.25 x was carried out in a Thunberg tube in an atmosphere of H2. The particulate fractions were obtained from the incubated mixtures by centrifugation for 1 hour at 200,000 g.
onstrated by the elegant experiments of Azoulay and collaborators (Azoulay et al., 1969, 1972, 1975; Riviere and Azoulay, 1971; Mutaftchiev and Azoulay, 1973; Riviere et al., 1975) and of MacGregor and Schnaitman (1973). When cytoplasmic fractions of chl B mutants are mixed with cytoplasmic fractions of chl A, chl C, or chl E mutants, reconstitution of nitrate reductase activity occurs and membranelike particles are formed (Table 11).Formation of membranelike particles also occurred when the mutant cytoplasmic fractions were incubated alone, but mixing of these incubated cytoplasmic fractions did not result in an active nitrate reductase complex (MacGregor and Schnaitman, 1973). For reconstitution the following components appeared to be required: (i) membrane-bound nitrate reductase protein subunits, (ii) a molybdenum-containing cofactor (MO-X), (iii) a soluble association factor, (iv) phospholipids (Azoulay et al., 1975).
188
WIL N. KONINGS AND JOHANNES BOONSTRA
2. AEROBIC AND FACULTATIVE BACTERIA Nitrate respiration has been demonstrated in aerobic bacteria such as Pseudomonas (Fewson and Nicholas, 1961b; Kodama et al., 1969; Kodama, 1970) and in many facultative anaerobes such as Staphylococcus aureus (Sasarman et al., 1971, 1974), Rhixobium japonicum (Cheniae and Evans, 1959; Lowe and Evans, 1964),Salmonella typhimurium (Stouthamer, 1969), Thiobacillus denitrificans (Adams et a1 ., 1971; Aminuddin and Nicholas, 1973, 1974), and Haemophilus
species (White, 1962,1963, 1966; Sinclair and White, 1970).This electron transfer system has been studied in these organisms in some detail. The properties of the enzyme complex are, in general, very similar to those described for the E . coli system, and the reader is directed to the literature for more detailed information. In this section, we discuss only a few organisms that may supply additional information leading to a better understanding of this electron transfer system. Klebsiella (Aerobacter) aerogenes is closely related to E . coli. It possesses the ability to perform nitrate respiration anaerobically and also to assimilate nitrate both aerobically and anaerobically (Pichinoty, 1960, 1965; Forget and Pichinoty, 1964; Hadjipetrou and Stouthamer, 1965; Van’t Riet et al., 1968). The best electron donor for the respiratory nitrate reductase is NADH (Hadjipetrou and Stouthamer, 1965; Knook and Planta, 1971). In addition, a high NADH oxidase activity is present. It was demonstrated in this organism that, in addition to ubiquinone and cytochrome b (Knook and Planta, 1971,1973), flavoproteins are involved in the transfer of electrons from NADH to nitrate and inhibitors at the level of the flavoproteins [e.g., rhein (Kean et al., 1971) and rotenone] inhibit NADH nitrate reductase activity (Knook and Planta, 1973; Knook, 1972; Knook et al., 1973). Another typical denitrifying bacterium is Micrococcus denitrificans, which uses nitrate for both assimilation (Chang and Morris, 1962) and respiration (Fewson and Nicholas, 1961a; Naik and Nicholas, 1966). In this organism, the best electron donors for the reduction of nitrate to nitrite are NADH and succinate (Scholes and Smith, 1968; Lam and Nicholas, 1969b), but molecular hydrogen can also be used as electron donor (Verhoeven et al., 1954). The respiratory nitrate reductase is associated with cell membranes (Lam and Nicholas, 1969b), but the assimilatory nitrate reductase seems to be nonparticulate (Lam and Nicholas, 1969~). A particulate fraction that possesses respiratory nitrate reductase activity also contains NADH oxidase activity. The affinity constants
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
189
(K,) for NADH oxidation with nitrate and oxygen as acceptors are similar (1.8-3.0 x lop5M ) . This indicates the involvement of the same primary dehydrogenase in both electron transfer systems (Lam and Nicholas, 1969a,b). In the NADH-nitrate reductase pathway of M . denitrificans, only cytochrome b seems to function as an electron carrier (Scholes and Smith, 1968; John and Whatley, 1970).The content of cytochrome c in cells grown anaerobically in the presence of nitrate is higher than in aerobically grown cells (Porra and Lascelles, 1965; Scholes and Smith, 1968). It is very unlikely, however, that cytochrome c plays a role in the nitrate reductase pathway because the particulate fraction does not oxidize exogenous reduced cytochrome c under anaerobic conditions with nitrate. However, rapid oxidation occurred when oxygen was introduced into the system. Nitrate reductase was solubilized and purified 100-fold by Lam and Nicholas (1969b). The purified preparation does not contain flavins or cytochromes, but does contain molybdenum. An enzymatic role for molybdenum and iron has been indicated (Lam and Nicholas, 196913; Forget and Dervartanian, 1972). 3. STRICTLYANAEROBICBACTERIA
Anaerobic electron transfer has not received as much attention in strict anaerobes (for definition, see Section I) as in facultatively anaerobic bacteria. Hence, available information is limited. Veillonella alcalescens and Selenomonas ruminantium are two strict anaerobes that perform nitrate respiration when grown in the presence of nitrate (Inderlied and Delwiche, 1973; De Vries et al., 1973, 1974). Both organisms contain substantial amounts of cytochrome b and small amounts of other cytochromes; the levels of cytochrome b in these organisms are not very different from those found in certain facultative organisms. Dual wavelength experiments with crude membrane fractions of these organisms showed that cytochrome b is involved in the transfer of electrons to nitrate and also to fumarate (see Section 11,B73).The best electron donor in S . ruminantium, grown anaerobically on lactose, is NADH; in V. alcalescens, grown on lactate in the presence of nitrate, the best electron donor is NADH too, but aglycerol phosphate, L-lactate, formate, and L-malate are also effective (De Vries et al., 1974; Konings et al., 1975). The nitrate reductase system of V . alcalescens has been shown to be membrane-bound (Inderlied and Delwiche, 1973; Konings et aZ., 1975) and has characteristics of both assimilatory and dissimilatory nitrate reductases.
190
WIL N. KONINGS AND JOHANNES BOONSTRA
B. Fumamte Reduction
Anaerobic electron transfer systems in which fumarate functions as the final electron acceptor have not been studied as much as the nitrate respiration systems. Electron transfer-linked fumarate reduction has been demonstrated in several facultative organisms. In contrast to nitrate respiration, fumarate reduction has been found also in many strict anaerobes. This might indicate that, in these organisms, electron transfer to fumarate is more common than to nitrate. In most organisms, the system is induced by growth under anaerobic conditions in the presence of fumarate. Other electron acceptors, such as oxygen and nitrate, repress the formation of the anaerobic electron transfer system. The terminal oxidase of fumarate reduction is fumarate reductase, which catalyzes the reduction of fumarate to succinate. The other components of this electron transfer system vary from organism to organism. In some organisms, like Streptococcus faecalis, very simple systems are present in which dehydrogenases, flavins, quinones, and nonheme iron proteins participate; in other organisms, cytochromes, usually of the b type, are also electron transfer intermediates. Depending on the growth conditions, several substrates, such as ~ - a glycerol phosphate, NADH, L-malate, formate, lactate, and molecular hydrogen can donate electrons to this electron transfer system. The components of fumarate reduction have been found in the particulate fraction of cell extracts, and the system can be isolated as a functional complex from this &action.
1. ESCHERICHIA COLI Growth of E . coli under anaerobic conditions with glycerol as carbon source and fumarate as electron acceptor results in the induction of anaerobic L-a-glycerol phosphate dehydrogenase and fumarate reductase. These two enzymes constitute a functional complex that is membrane-bound (Miki and Lin, 1973; Konings and Kaback, 1973a; Boonstra et al., 1975a), and which catalyzes the dehydrogenation of L-a-glycerol phosphate at the expense of fumarate without any added cofactors (Miki and Lin, 1973). In E . coli, the dissimilation of both L-a-glycerol phosphate and glycerol is mediated by one of two L-a-glycerol phosphate dehydrogenases (Cozzarelli et al., 1968; Kistler et al., 1969; Kistler and Lin, 1971) (Fig. 4).Glycerol enters the cells by diffusion and is converted by glycerol kinase to L-a-glycerol phosphate under aerobic conditions inside the cell. L-a-glycerol phosphate itself can be taken
191
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
ME DlUM CELL Foci I i ta t e d
Glycerol
, k Glycerol
Diffusion
Kinarc
Active
0 3P
yy? Transport
I
Fructose 1,6 dip
I A
G3P
*
Anaerobic
,
Aerobic
,
DHAP
GAP
Dchydrojenases
FIG.4. The pathways for glycerol and L-a-glycerol phosphate dissimilation in Escherichia coli. G3P, L-a-glycerol phosphate; DHAP, dehydroxyacetone phosphate; GAP, D-glyceraldehyde 3-phosphate. From Freedberg and Lin (1973),with permission.
up directly from the medium by a specific active transport system (Lin et al., 1962; Hayashi et al., 1964).Aerobically, L-a-glycerol phosphate is oxidized by a membrane-associated aerobic L-a-glycerol phosphate dehydrogenase (Koch et al., 1964; Kistler and Lin, 1971) to dihydroxyacetone phosphate (Kistler and Lin, 1972). Anaerobically, in the presence of fumarate as electron acceptor, it is converted to dihydroxyacetone phosphate (Kistler and Lin, 1972) by an anaerobic ~ - a glycerol phosphate dehydrogenase (Kistler and Lin, 1971). Studies of the growth properties, on glycerol and L-a-glycerol phosphate, of mutants that lack the aerobic or the anaerobic L-a-glycerol phosphate dehydrogenases and of double mutants indicate that the anaerobic enzyme is distinct from the aerobic L-a-glycerol phosphate dehydrogenase. This conclusion is supported by differences in biochemical properties of the purified enzymes and by the demonstration that the enzymes map at distinct loci on the chromosome (Cozzarelli et al., 1968; Kistler et al., 1969; Kistler and Lin, 1971, 1972). The aerobic and anaerobic L-a-glycerol phosphate dehydrogenase are affected differently by the growth conditions. The activity of the anaerobic L-a-glycerol phosphate dehydrogenase is about the same in cells grown either anaerobically with nitrate or aerobically, but, under anaerobic conditions with fumarate as terminal electron acceptor, the level is elevated severalfold (Kistler and Lin, 1971). The ratio of aerobic to anaerobic enzyme is therefore high when molecular oxygen or nitrate serves as electron acceptor and low when fumarate plays this role (Freedberg and Lin, 1973).
192
WIL N. KONINGS AND JOHANNES BOONSTRA
Fumarate reduction, too, can be catalyzed in E . coli by two distinct enzymes that also catalyze the oxidation of succinate (Hirsch et al., 1963). One of these, succinate dehydrogenase, has a predominantly oxidative function as a participant of the tricarboxylic acid cycle. It is a membrane-bound flavoprotein and serves as a direct electron donor to the respiratory electron transfer chain leading to oxygen (Spencer and Guest, 1973,1974b). The enzyme is present only under aerobic conditions and completely repressed under anaerobic conditions (Spencer and Guest, 1973). Succinate dehydrogenase oxidizes succinate more rapidly than it reduces fumarate and has a lower K , for succinate than for fumarate. Mutants that lack succinate dehydrogenase activity are unable to grow on succinate, but growth on fumarate is unimpaired (Hirsch et al., 1963; Spencer and Guest, 1974a). The second enzyme with fumarate-reducing activity is fumarate reductase which is also membrane-bound (Peck et al., 1957). The synthesis is repressed under aerobic conditions and derepressed anaerobically and to some extent aerobically in the presence of glucose (Hirsch et al., 1963; Spencer and Guest, 1973). In contrast to succinate dehydrogenase, this enzyme oxidizes succinate at about the same rate as it reduces fumarate, even though the K , for fumarate is much lower than for succinate (Hirsch et al., 1963). Fumarate reductase functions as a terminal oxidase during anaerobic growth, but it can also provide succinate for biosynthesis when the tricarboxylic acid cycle enzymes are repressed (Amarasingham and Davis, 1965; Gray et al., 1966). Mutants have been isolated which lack functional fumarate reductase and are unable to use fumarate as an anaerobic electron acceptor (Spencer and Guest, 1973). The complex of anaerobic L-a-glycerol phosphate dehydrogenase and fumarate reductase is present in the particulate fraction of cell extracts and in isolated membrane vesicles (Konings and Kaback, 1973a; Boonstra et al., 1975a). It catalyzes the anaerobic oxidation of ~ - a glycerol phosphate in the presence of fumarate as terminal electron acceptor. No stimulation of the coupled activity is observed upon the addition of FAD or FMN, and the complex is probably saturated with flavins (Miki and Lin, 1973). Recently, Singh and Bragg (1975) demonstrated in a cytochrome-deficient (hem A-) mutant of E . coli that electron transfer from NADH or a-glycerol phosphate to fumarate does not require the participation of cytochromes, but involves menaquinone and, most likely, nonheme iron proteins. Anaerobic growth of E . coli in the presence of fumarate results in the induction of the anaerobic electron transfer system to fumarate. It is of interest that these cells contain, in addition, the respiratory chain and the nitrate respiration system. Studies with isolated membrane
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
1 93
vesicles, which will be presented in the “active transport” section, demonstrate that these three electron transfer systems can be linked to the same dehydrogenase, such as L-a-glycerol phosphate dehydrogenase (Boonstra et al., 1975a). This indicates that some electron carriers may be shared by all three electron transfer systems.
2. OTHER FACULTATIVE BACTERIA Evidence for anaerobic electron transfer with fumarate as terminal electron acceptor has been presented for only a few facultative anaerobes. Detailed studies in Klebsiella aerogenes (Ruch et al., 1974) demonstrated that anaerobic growth of this organism on L-a-glycerol phosphate as carbon source requires exogenous electron acceptors such as fumarate. Under these conditions, L-a-glycerol phosphate is converted by an (anaerobic) flavin-linked a-glycerol phosphate dehydrogenase to dihydroxyacetone phosphate. The membrane-bound terminal oxidase, fumarate reductase, is induced by anaerobic growth in the presence of fumarate; oxygen strongly represses its synthesis (Pichinoty and Coudert, 1962). Quinones and cytochromes function as electron carriers in the electron transfer to fumarate in Bacillus inegaterium (Kroger and Dadak, 1969) and Haemophilus influenxae (White, 1966; Sinclair and White, 1970). Detailed studies have dealt with the role of quinones in electron transfer in Proteus rettgeri (Kroger et al., 1971; Kroger, 1974). This organism contains ubiquinone and menaquinone, which have considerably different redox potentials. The functional positions of these quinones with respect to the cytochromes and dehydrogenases was elucidated (Fig. 5). Menaquinone, the lower-potential quinone, serves only as an electron donor to fumarate reductase. In addition, the oxidation of both NADH and formate is linked to fumarate reduction, and the coupled activities of the dehydrogenases with fumarate reductase parallel the content of menaquinone. Ubiquinone, the higher-potential quinone, is involved in the respiratory pathway of both succinate and formate. It was postulated that a b type cytochrome forms a link between formate dehydrogenase and the pathways to oxygen and fumarate. The oxidation of NADH by oxygen can occur also through ubiquinone, but this NADH dehydrogenase appears to be different from the menaquinone-linked dehydrogenase.
3. STRICTLYANAEROBICBACTERIA Fumarate reduction linked to electron transfer has been demonstrated in several strict anaerobes (for definition, see Section I).
194
WIL N. KONINGS AND JOHANNES BOONSTRA
-
0.1
-
-P
0 -
r.
0
b -0.1 3
-0.32 I I II
-0.45
Fordate
FIG. 5. Pathways of electron transfer of Proteus rettgeri, taking into account the appropriate redox potentials. a,, Cytochrome a,; a2,cytochrome a*; b,, cytochrome b,; b,, cytochrome b,; Fo-DH, formate dehydrogenase; Fu-R, fumarate reductase; HQNO, 2-n-heptyl-4-hydroxyquinoline-N-oxide; MK, menaquinone; NADH-DHMK, NADH dehydrogenase linked to menaquinone; NADH-DH,, NADH dehydrogenase linked to ubiquinone; o, cytochrome o; Su-DH, succinate dehydrogenase; Q, ubiquinone. From Kroger et al. (1971), with permission.
Energy-dependent membrane functions have been studied intensively in Streptococcus faecalis because it does not contain ironporphyrin enzymes or a functional tricarboxylic acid cycle. However, it has been demonstrated that it contains an active membrane-bound fumarate reductase (Gunsalus, 1947; Jacobs and Vandemark, 1960) which is coupled to the oxidation of NADH; the enzyme appears to be constitutive (Aue and Deibel, 1967). Flavins, nonheme iron, and naphtoquinone have all been implicated as electron carriers in the electron transfer system to fumarate (Jacobs and Vandemark, 1960; Baum and Dolin, 1965; Faust and Vandemark, 1970) (Fig. 6). Fumarate reductase can accept electrons from reduced FMN. The enzyme is inhibited by succinate and malonate and is sensitive to air oxidation; chelating agents (cyanide, azide, EDTA) have no effect on the enzyme’s activity (Deibel and Kvetkas, 1964; Aue and Deibel, 1967). In other strict anaerobes, cytochromes are intermediates in electron transfer to fumarate. In Propionibacterium arabinosum, electrons from the electron donors NADH, L-a-glycerol phosphate, and lactate, are transferred via flavoproteins, nonheme iron, b type cytochromes,
1 95
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
and fumarate reductase to fumarate (Sone, 1972). Cytochromes of the b type are also involved in the electron transfer system with fumarate reductase as terminal oxidase in Veillonella alcalescens and in Selenomonas rurninantium, organisms that also perform nitrate respiration (De Vries et al., 1974). Bacteroides melaninogenicus forms a membrane-bound electron transfer system, provided that protoheme is supplied in the growth medium. This chain includes a CO-binding pigment, cytochrome c , and possibly flavoproteins; these pigments can be reduced by NADH and oxidized by fumarate or by oxygen (Rizza et al., 1968), although oxygen is lethal for this organism. Desulfouibrio spp. grows strictly anaerobically, with sulfate as terminal electron acceptor. Several species, such as Desulfovibrio gigas, can also grow anaerobically in sulfate-free media provided that fumarate is present (Miller and Wakerley, 1966). Cells grown in the presence of fumarate contain a membrane-bound electron transport system to fumarate reductase in which cytochromes of the b type and possibly menaquinone-6 are involved (Hatchikian and LeGall, 1972; Hatchikian, 1974). Hydrogen (Barton et al., 1970) and pyruvate (Hatchikian and LeGall, 1970a,b) are able to serve as electron donors for fumarate reduction. Electron transfer-linked fumarate reductase has also been reported in Bacteroides ruminicola, one of the most important species of ruminal bacteria. In this system, NADH is the physiological electron donor, and electrons can be accepted by fumarate and also by oxygen, COe, oxaloacetate, and malate; a cytochrome of the b type and a flavoprotein are involved (White et al., 1962).
111.
PHOSPHORYLATION COUPLED TO ELECTRON TRANSFER
A. Introduction
The energy required for the phosphorylation of ADP to ATP can be supplied by electron flow in the electron transfer chain if the dif-
NADH
NAD+
+
H+
AT P
Fuma r a t e
Succinate
FIG.6. Fumarate reduction pathway of Streptococcusfaecalis. Fp, flavoprotein; Fe, nonheme iron; NQ, naphtoquinone. From Faust and Vandemark (1970), with permission.
1 96
WIL N. KONINGS AND JOHANNES BOONSTRA
TABLE I11 STANDARD OXIDATION-REDUCTION POTENTIALS OF ELECTRONDONORSAND ACCEPTORS INVOLVED IN NITRATERESPIRATION AND FUMARATEREDUCTION"
Oxygen/water Nitrate/nitrite Fumarate/succinate Oxaloacetate/malate Pyruvate/lactate Dihydroxyacetone phosphate/a-glycerol phosphate Riboflavin, ox/red NAD+/NADH Carbon dioxide/formate H+/H,
815 421 31 - 166 - 185 - 190 -219 -320 -420 -420
" EL is standard oxidation-reduction potential against the H electrode at pH 7.0 and 25°C. Data are taken from Handbook of Biochemistry (1973) (H. A. Sober and R. A. Harte, eds.), 2nd ed., pp. 7-33. The Chemical Rubber Co., Cleveland, Ohio.
ference in free energy (AGL) between two adjacent redox carriers is at least - 9 kcal; this means that, in a two-electron transfer, the difference in redox potentials at pH 7 and 25°C (AE b) must be at least 220 mV. Table I11 lists the redox potentials of electron donors and acceptors functional in nitrate respiration and fumarate reduction. From these data it can be calculated that the AE b between several electron donors and the acceptors nitrate or fumarate is sufficiently large that the synthesis of at least one ATP per two electrons transferred becomes energetically possible. In bacteria, phosphorylation of ADP to ATP is catalyzed by membrane-bound Ca2+-and M$+-dependent ATPases (Bogin et al., 1970; Butlin et al., 1971; Gutnick et al., 1972). The coupling of ATPase to the respiratory chain has been studied extensively, and the mechanism of this phosphorylation in aerobically grown bacteria seems to be essentially the same as in mitochondria (Gel'man et al., 1967; Brodie and Gutnick, 1972). Several hypotheses have been advanced to explain the mechanism by which the free energy liberated by oxidation-reduction reactions in the electron transfer chain can be utilized for the synthesis of ATP or energy-dependent processes. Only one of these theories, the chemiosmotic coupling theory, will be discussed in relation to active transport processes (see Section IV,A), and
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
197
the reader is directed for more information to the excellent review of Harold (1972). ATP synthesis coupled to the respiratory chain is now a well-known phenomenon. This seems not to be the case for anaerobic electron transfer-linked phosphorylation, and statements like “the enzymatic machinery of oxidative phosphorylation is only produced under aerobic conditions” can be found in the literature (Harold, 1972). However, the evidence currently available demonstrates that the energy needed for the membrane-bound synthesis of ATP and for active transport processes can also be supplied by anaerobic electron transfer systems. Under certain anaerobic growth conditions, this synthesis of ATP may make an important contribution to the cellular ATP pool. Evidence has been presented that under certain conditions ATP itself may supply the energy for active transport. Before turning to active transport processes under anaerobic growth conditions, it seems, therefore, appropriate to discuss briefly the available evidence for ATP synthesis linked to anaerobic electron transfer systems. 6. ATP Synthesis Coupled to Nitrate Respiration
Indirect evidence has been presented for a coupling of ATP synthesis to nitrate reduction in several Pseudomonas species (Yamanaka et al., 1962; Ohnishi, 1963; Ishaque et al., 1973). Ota et al. (1964) reported such a coupling of oxidative phosphorylation to nitrate reduction, with NADH as electron donor, in extracts ofE. coli. These observations confirmed the earlier observations of Takahashi et al. (1957). Extensive studies on oxidative phosphorylation coupled to electron transfer in various Enterobacteriaceae have been performed (Hadjipetrou and Stouthamer, 1965; Stouthamer and Bettenhausen, 1972, 1973; Stouthamer, 1973). In many microorganisms, the cell yield in grams of dry weight is directly proportional to the ATP yield, averaging about 10 g of cells per mole of ATP generated ( Y A T p = 10). With this value, the ATP production associated with nitrate reduction in Klebsiella aerogenes has been calculated (Hadjipetrou and Stouthamer, 1965). For this organism, YATpwas calculated to be 10.2 from the anaerobic molar growth yieId on glucose, corrected for acetate production. In the presence of nitrate, the molar growth yield for glucose was strongly increased. By dividing the growth yield per mole of nitrate reduced by YATP,the yield of ATP per two electrons transferred to nitrate was obtained. The results indicated that 3 moles of ATP were produced per mole of nitrate reduced. The formation of ATP was probably coupled only to the reduction of nitrate to nitrite
1 98
WIL N. KONINGS AND JOHANNES EOONSTRA
because nitrite was excreted, and no increase in growth yield was obtained with nitrite as terminal electron acceptor (Hadjipetrou and Stouthamer, 1965; Stouthamer and Bettenhausen, 1973). Proteus mirabilis showed a similar increase in growth yield under anaerobic conditions in the presence of nitrate, and P/NO,- values of 1.48 have been measured (Stouthamer and Bettenhausen, 1972). In Micrococcus denitrificans nitrate-dependent ATP synthesis has been demonstrated by Naik and Nicholas (1966) and later by John and Whatley (1970). Particles isolated from cells grown anaerobically in the presence of nitrate catalyzed ATP synthesis with NADH or succinate and either oxygen or nitrate as electron acceptor. Addition of ADP and Pi to the reaction mixture caused increased oxygen uptake or nitrate reduction with NADH as electron donor. This indicates that ATP synthesis is coupled very tightly to electron transfer. C. ATP Synthesis Coupled to Fumamte Reduction
Proteus rettgeri can grow anaerobically with fumarate as sole carbon and energy source. Growth yield experiments indicate that, under these conditions, for every 7 moles of fumarate consumed, 1 mole of ATP is formed by substrate level phosphorylation and 5 moles of ATP by fumarate reduction (Kroger, 1974). The ATP yield found experimentally is somewhat lower than that calculated, but it is nonetheless clear that ATP is formed by both substrate level phosphorylation and by electron transfer-coupled phosphorylation. Analogous discrepancies between calculated and experimentally found ATP yields have been reported in other organisms, but no explanation can be offered at this time (Payne, 1970; Stouthamer, 1973). Evidence for a coupling of ATP synthesis to electron transfer with fumarate as terminal electron acceptor has also been presented for strictly anaerobic bacteria. In Propionibacterium freudenreichii, the high molar growth yield on glucose was explained by the formation of two ATP molecules in electron transfer from NADH to fumarate (De Vries et al., 1973). Such an explanation may also be valid for the high molar growth yield on fructose of Selenomonas ruminantium or on glucose of Anaerovibrio lipolytica (Hobson, 1965; Hobson and Summers, 1967; De Vries et al., 1973). An interesting organism with respect to electron transfer-coupled phosphorylation is Streptococcus faecalis. Until recently, it was thought that this organism could only generate ATP by substrate level phosphorylation. However, Faust and Vandemark (1970) demonstrated in cell-free preparations of s.faecalis, that phosphorylation occurred as a result of coupling to the oxidation of NADH with fumarate
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
199
as terminal electron acceptor. Both NADH oxidation and ATP formation were inhibited by the flavoprotein inhibitor, quinacrine HC1, and by the iron chelator, bathophenanthroline disulfonate. It is concluded that this phosphorylation is coupled to the flavin region of the electron transfer system from NADH to fumarate (see Fig. 6).
IV.
ANAEROBIC ACTIVE TRANSPORT
A. introduction
Transport of substrates across the cytoplasmic membrane occurs by several mechanisms that will be defined in order to clarify the subsequent discussion. 1. In passive diffusion, substrate crosses a membrane as a result of random molecular motion; no specific interactions are thought to occur with molecular species in the membrane. 2. In facilitated diffusion, the transported substrate combines reversibly with a specific carrier in the membrane. The carrier or carrier-substrate complex oscillates between the inner and outer surfaces of the membrane, releasing and binding molecules on either side. Neither passive diffusion nor facilitated diffusion requires metabolic energy, and neither leads to concentration against a gradient. 3. By active transport, the substrate is accumulated against an electrochemical or osmotic gradient. This mechanism requires metabolic energy as well as a specific carrier molecule in the membrane. In “classic” active transport, no chemical modification occurs during substrate accumulation. 4. Group translocation. This transport mechanism is also dependent on metabolic energy, but in this process the transported substrate is subjected to a chemical modification in such a way that the reaction itself results in the passage of the molecule through the diffusion barrier. An important group translocation system in bacteria is the phosphoenolpyruvate phosphotransferase system in which various carbohydrates are phosphorylated during their passage through the membrane. In this section we will be concerned mainly with active transport processes across the cytoplasmic membranes of anaerobically grown cells. Our current concept of active transport in bacteria has been developed largely by studies under aerobic conditions in whole cells or in membrane vesicles isolated according to the procedure of Kaback (1971, 1974a). Membrane vesicles, which consist of closed cytoplasmic membrane
200
WIL N. KONINGS AND JOHANNES BOONSTRA
sacs, have proved to be a very useful model system for the study of active transport processes and other membrane-related phenomena. They can be isolated by osmotic lysis of spheroplasts from gramnegative organisms and of protoplasts from gram-positive organisms (Kaback and Hong, 1973; Kaback, 1974b).The vesicles have a continuous surface, and, if gentle isolation procedures are applied, the vesicle membrane has the same orientation as the cytoplasmic membrane of whole cells (Kaback, 1972; Kaback and Hong, 1973; Konings et al., 1973; Altendorf and Staehelin, 1974; Konings, 1975; Shortet al., 1974). In membrane vesicles of several gram-positive and gram-negative organisms, active transport of amino acids, sugars, carboxylic acids, and several other metabolites has been shown to be coupled to electron transfer in the membrane-bound respiratory chain (Kaback and Hong, 1973; Kaback, 1974b). In E. coli membrane vesicles, the transport systems are energized primarily by the oxidation of D-laCtate or reduced phenazine methosulfate (Barnes and Kaback, 1970; Kaback and Milner, 1970; Kaback and Barnes, 1971; Konings et al., 1971; Hong and Kaback, 1972; Kaback and Hong, 1973), with oxygen as terminal electron acceptor. The energy-coupling site for transport in E. coli is localized in a segment of the respiratory chain between Dlactate dehydrogenase and cytochrome b In membrane vesicles from other bacteria, transport is coupled to the oxidation of other physiological electron donors, such as NADH (Konings and Freese, 1971,1972). Several lines of evidence indicate that the energization of transport in these membrane vesicles does not involve the synthesis of stable energy-rich phosphate intermediates such as ATP (Kaback and Milner, 1970; Barnes and Kaback, 1970; Kerwar et al., 1972; Klein and Boyer, 1972; Konings and Freese, 1972; Prezioso et al., 1973). Although these studies show that ATP does not play an essential role in active transport in membrane vesicles, evidence obtained from studies in whole cells, however, indicates that ATP can drive active transport under anaerobic conditions (Abrams and Smith, 1971; Klein and Boyer, 1972; Schairer and Haddock, 1972; Berger, 1973; Or et al., 1973; Parnes and Boos, 1973; Van Thienen and Postma, 1973). Therefore, active transport may be coupled in several ways to the cell’s metabolic machinery. The nature of these linkages is unknown; general models have been proposed that involve a direct chemical coupling of the transport carriers to ATP or to intermediates of oxidative phosphorylation (Boyer and Klein, 1972), a direct coupling to the respiratory chain (Kaback and Hong, 1973) or chemiosmotic coupling to ion gradients (Mitchell, 1963, 1970, 1973). Most of the evidence
20 1
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
available to date supports one of the two latter models, and these are therefore briefly discussed. A model has been presented by Kaback and Hong (1973) that proposes a direct coupling of the carriers to specific sites of the respiratory chain (Fig. 7 ) .According to this model, the transport proteins, the carriers, possess high affinity for their transport substrates only in the oxidized (disulfide) form, whereas the reduced (sulfhydryl) form has low substrate affinity. Active uptake of a particular sugar or amino acid is associated with the reduction of the appropriate carrier by the electron donor. In Fig. 7 , the electron transport chain is given from E . coli with D-lactate as electron donor. Upon reduction, the high affinity form of the carrier undergoes a conformational change that results in the translocation of bound substrate from the outer surface of the membrane to the inner surface. The resulting low-affinity (sulfhydryl) form of the carrier then releases the substrate, and the carrier is reoxidized. By alternative oxidation and reduction of the carrier, substrate is transferred from the outside to the inside against a concentration gradient until the internal concentration is sufficient to saturate the reduced form of the carrier. At that point the rate of efflux will equal the rate of influx, and a steady state will be achieved. In the reduced state, the carrier is mobile in the membrane and can mediate facilitated dif-
OUT
IN
PYR
D-LAC
-
REDUCED
HIGH K ,
REDUCED
-
FIG. 7. Conceptual working model for D-lactic dehydrogenase-coupled transport systems. D-LAC, D-lactate, PYR, pyruvate; fp, flavoprotein; cyto b , , cytochrome b,; ox, oxidized; red, reduced. OUT signifies the outside surface ofthe membrane; IN signifies the inside surface. The spheres located between fp and cyto b , represent the carrier; w, a high-affinity binding site a n d w , a low-affinity binding site. The remainder of the cytochrome chain from cytochrome b , to oxygen has been omitted. From Kaback (1973), with permission.
202
WIL N. KONINGS AND JOHANNES BOONSTRA
fusion. This model postulates that carriers of different substrate specificity are coupled to specific sites in the electron transfer chain, thereby conferring functional heterogeneity on otherwise identical electron transfer chains. The attractive feature of this model is that it does not require any special characteristics from the membrane other than it function as a diffusion barrier. The model explains the absence of any correlation between the rates of oxidation of various electron donors and their relative effects on transport. However, the model fails to adequately explain the behavior of electron transfer-coupling mutants (Hong and Kaback, 1972) and other mutants (Simoni and Shallenberger, 1972; Rosen, 1973a,b; Yamamoto et al., 1973) that exhibit normal electron transfer properties, but are defective in Dlactate dependent transport. The model also does not account for the inhibitory action of uncouplers. A very different type of mechanism for active transport is visualized by the chemiosmotic coupling hypothesis. This hypothesis, developed by Mitchell (1961, 1963, 1966, 1973), rests upon the following postulates: 1. The cytoplasmic membrane is essentially impermeable to most ions and in particular to OH- and H+. 2. The respiratory chain is an alternating sequence of hydrogen and electron carriers, arranged across the membrane in loops. The oxidation of a substrate results in the translocation of protons from one side of the membrane to the other; in any one loop, two protons pass across. Translocation of protons is equivalent to the movement of OHin the opposite direction, so that oxidation of a substrate results in the distribution of H+ and OH- on opposite sides of the membrane (Fig. 8).Both a pH gradient and an electrical potential are therefore established across the membrane, and the sum of these forces constitutes the proton-motive force: Ap = A$-
ZApH
Ap is the proton-motive force, A$ is the electrical potential across the membrane, ApH is the pH difference between interior and exterior, Z = 2/3 RTIF, in which R is the gas constant, T is the absolute temperature, and F is the Faraday constant. Z has a numerical value of about
60 mV at 25°C. 3. The proton-motive force generated by the respiratory chain reverses the direction of ATPase so as to bring about net synthesis of ATP (Fig. 8).On the other hand, ATPase itself can function as a proton translocator, and the hydrolysis of intracellular ATP leads to the efflux
203
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
R e i p i r a t o r v ch ain
AlPare
Reversible ATPare poked by A p n and
A$
onFIG.8. Chemiosmotic hypothesis in principle: extrusion of protons by the respiratory chain, generation of ApH and A$, and the poising of ATPase by the proton-motive force. From Harold (1972),with permission.
of protons into the medium and consequently establishes a protonmotive force. According to the chemiosmotic coupling model, the proton-motive force is the driving force for active transport of substrates (Mitchell, 1963, 1966, 1970, 1973). Neutral substrates, such as lactose, will be transported via a coupled movement with protons. It is postulated that the transport proteins, the carriers, have affinity for both the substrate and the protons; the pH gradient and the electrical potential will drive the movement of protons and charge and consequently the active transport and accumulation of substrate (i.e., symport). Anions, such as
204
WI1 N. KONINGS AND JOHANNES BOONSTRA
phosphate, will also be transported by a proton symport system. However, this transport will be electroneutral and influenced only by the pH gradient. Transport of positively charged substrates, such as lysine or potassium (in the presence of valinomycin) will be driven by the membrane potential only. This movement will be electrogenic and does not involve protons (i.e., uniport). The electrical potential is also the driving force for the transport of lipophilic cations such as triphenylmethylphosphonium (TPMP+) and dibenzyldimethylammonium (DDA+),but this transport does not involve specific membrane proteins. The attractive feature of the chemiosmotic coupling model is that the proton-motive force is visualized as the common factor for the synthesis of ATP, transport, and other energy-linked functions of the membrane. In addition, this model offers an explanation for the inhibitory action of uncoupling agents on transport. It is proposed that these compounds are soluble in the membrane and act as circulating carriers, conducting protons across the membrane, thereby shortcircuiting the proton-motive force. Studies in whole cells of microorganisms, as well as in membrane vesicles, have supplied evidence in favor of a chemiosmotic type of energy coupling, and it appears to be beyond dispute that the proton-motive force plays an essential role in the mechanism of transport under both aerobic and anaerobic conditions (Mitchell, 1966, 1973; Harold, 1972,1974; Boos, 1974; Kaback, 1974b; Hamilton, 1975; Lombardi et al., 1974; Ramos et al., 1976; West, 1970; West and Mitchell, 1972, 1973). B. Anaerobic Active Tmnsport Coupled to Nitrate Respiration
Anaerobic active transport of lactose has been studied in whole cells of E . coli ML 308-225, a strain which is constitutive for the M protein, the lactose permease. Cells grown on glucose in the presence of nitrate, i.e., conditions that induce formate dehydrogenase and nitrate reductase, exhibit markedly increased lactose transport in the presence of formate and nitrate (Konings and Kaback, 1973a). On the other hand, cells grown anaerobically on glucose in the absence of an electron acceptor fail to show an increase in lactose transport upon the addition of formate and nitrate. These data indicate a coupling of lactose transport to the electron transfer system formatedehydrogenase-nitrate-reductase.More evidence for such a coupling has been obtained from studies in membrane vesicles from E. coli grown anaerobically on glucose in the presence of nitrate.
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
2 05
In order to demonstrate anaerobic transport coupled to electron transfer in membrane vesicles, a modified isolation procedure was required (Konings and Kaback, 1973a). Components of anaerobic electron transfer systems are apparently loosely bound to membranes and are removed by more drastic isolation procedures. The vesicle preparation consists of closed sacs, with no recognizable internal structure, surrounded by a single 65-70-A membrane (Konings and Kaback, 1973a) (Fig. 9). In the absence of electron donors and acceptors, lactose and amino
FIG. 9. Electron micrograph of membrane vesicles isolated honi Escherichin coli ML 308-225 grown anaerobically in the presence of glycerol and fiiniarate. Micrograph was obtained by Dr. M. Boublik and Mr. F. Jenkins ofthe Roche Institute of Molecular Biology, Nutley, N. J. x 67,800. From Konings and Kaback (1973a), with permission.
206
WI1 N. KONINGS AND JOHANNES BOONSTRA
minutes
FIG.10. Uptake of amino acids under anaerobic conditions in membrane vesicles from Escherichia coli ML 308-225 grown anaerobically in the presence of glucose and nitrate. Sodium formate and potassium nitrate were added to the reaction mixture at final concentrations of 10 mM. (U-14C)amino acids mixture (57 mCi/mAtom C) was added at a final concentration of 35 pAtoms C. Transport assays were carried out at 25°C. A, No additions; A, KNO,; 0, formate; 0, formate nitrate. The reaction mixture contained 50 mM potassium phosphate, pH 6.6,lO mM magnesium sulfate, and 0.05 mg membrane protein. From Boonstra and Konings (1976), with permission.
+
acid uptake at a relatively high endogenous rate was observed with these membrane vesicles. This indicates that these vesicles are not as depIeted of endogenous energy sources as those prepared by the original procedure (Kaback, 1971).The membrane vesicles have high formate dehydrogenase and nitrate reductase activities and reduce nitrate rapidly in the presence of formate. This formate-dehydrogenase nitrate-reductase electron transfer system is coupled to the anaerobic transport of lactose and amino acids, as demonstrated by the marked stimulation of uptake in the presence of both the elec-
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
207
tron donor formate and the electron acceptor nitrate (Fig. 10). Moreover, a strong stimulation of amino acid uptake is observed with chlorate as electron acceptor. Ferricyanide, which most likely accepts electrons from the electron transfer system at a level beyond cytochrome b in the nitrate respiration system, can also replace nitrate (Boonstra et al., 1976) (Fig. 11). Further evidence for the involvement of electron transfer in anaerobic transport has been obtained from studies with electron
FIG. 1. Uptake of amino acids under anaerobic conditions in membrane vesicles from Escherichia coli ML 308-225grown anaerobically in the presence of glucose and nitrate. Transport assays were carried out as described in legend to Fig. 11. (U-14C) amino acid mixture (57 mCi/mAtom C) was added at a final concentration of 35 pAtoms C. V, No additions; A, Na chlorate (10 mM); 0,K ferricyanide (10 mM); 4 Na formate (10 mM) + Na chlorate (10 mM); 0, Na formate (10 mM) + K ferricyanide (10 mM). From Boonstra et al. (1976), with permission.
208
WI1 N. KONINGS AND JOHANNES BOONSTRA
transfer inhibitors. The formate-plus-nitrate-dependent transport of amino acids and lactose is almost completely inhibited by 2-nheptyl-4-hydroxyquinoline-N-oxide (HQNO), an inhibitor at the level of cytochrome b, and by cyanide, an inhibitor of nitrate reductase itself (Konings and Kaback, 1973a,b; Boonstra et al., 1976). The membrane vesicles contain a functional respiratory chain, and transport of amino acids and lactose can also be energized by electron transfer with oxygen as terminal electron acceptor. Effective electron donors are NADH and the artificial electron donor system, ascorbate plus phenazine methosulfate (Asc-PM S). Formate can also effectively energize transport under aerobic conditions; under these conditions, the addition of nitrate has no significant effect on the rate of uptake. The electron donors NADH and Asc-PMS, however, fail to stimulate transport under anaerobic conditions in the presence of nitrate. This indicates that, in these membrane vesicles, only formate dehydrogenase is coupled effectively to nitrate reductase (Konings and Kaback, 1973a; Boonstra et al., 1975a). Under different growth conditions, however, other electron donors also can function in this electron transfer system. In membrane vesicles from E . coli grown anaerobically on glycerol in the presence of nitrate, L-a-glycerol phosphate plus nitrate stimulates amino acid transport, but the extent of stimulation is lower than with formate plus nitrate (Boonstra et al., 1975a). A similar coupling between anaerobic transport and the electron transfer system with nitrate as terminal acceptor has been demonstrated in strictly anaerobic organisms. Membrane vesicles from the strict anaerobe Veillonella alcalescens, grown on lactate in the presence of nitrate, catalyze active transport of L-glutamate and other amino acids under anaerobic conditions in the presence of the electron donor L-lactate and the electron acceptor nitrate (Fig. 12). Llactate alone or nitrate alone have hardly any effect on L-glutamate uptake. L-lactate could be replaced by NADH, L-a-glycerol phosphate, formate, or L-malate. This indicates that, in these membrane vesicles, several dehydrogenases are coupled effectively to nitrate respiration. None of these electron donors could energize transport under aerobic conditions as expected, since V . alcalescens does not contain a functional respiratory chain (Konings et al., 1975). C. Anaerobic Active T m m p o ~Coupled i to Fumamte Reduction
A coupling of anaerobic transport to the electron transfer system with fumarate as terminal electron acceptor has been suggested by up-
209
ANAEROBIC ACTIVE TRANSPORT I N BACTERIA
D
C
B 3
/ TIME
(minutes)
FIG.12. Anaerobic transport of L-glutamate by membrane vesicles ofveillonellu alcalescens. Transport assays were carried out as described in legend to Fig. 11. L-(U-'IC) Li glutamate (265 mCi/mmole) was added at a final concentration of 7.6 p M . (A) 0, L-lactate (10 m M ) + KNO, (10 mM); 0, Li L-lactate (10 mM); 4KNO, (10 mM); V, without electron donor or acceptor added. (B) 0, NADH (10 mM) KNO, (10 mM); 0, NADH (10 mM). (C) Q Na-L-a-glycerol phosphate (10 mM) + KNO, (10 mM); 0 , Na Na formate (10 m M ) + KN03(10mM); 0, Na L-a-glycerol phosphate (10 mM). (D) 0, formate (10 mM). From Konings et al. (1975), with permission.
+
take experiments in whole cells. Butlin (1973) and Rosenberg et al. (1975) demonstrated that mutants of E. coli deficient in Ca2+- and Mg2+-stimulated ATPase (unc A) can catalyze active transport of serine and phosphate under anaerobic conditions with fumarate as electron acceptor. In whole cells ofE. coli ML 308-225, grown anaerobically on glycerol in the presence of fumarate, a marked stimulation of lactose uptake is observed upon the addition of L-a-glycerol phosphate plus fumarate. Under these conditions, L-a-glycerol phosphate dehydrogenase and fumarate reductase are induced. Such a stimulatory effect of L-a-glycerol phosphate plus fumarate is not observed in cells grown anaerobically on glucose alone, on glucose in the presence of nitrate, or in cells grown aerobically on glycerol. More evidence for a coupling between active transport and the anaerobic electron transfer to fumarate has been obtained with membrane vesicles from cells grown on glycerol in the presence of fumarate. These membrane vesicles, isolated with the same procedure as used for vesicles from glucose-nitrate grown cells, have a high endogenous rate of lactose uptake, and the addition of the electron donor L-a-glycerol phosphate alone or of fumarate alone causes significant stimulation of lactose uptake (Fig. 13). In the presence of both ~ - a -
210
WI1 N. KONINGS AND JOHANNES BOONSTRA
FIG. 13. Uptake of lactose anaerobic conditions in membrane vesicles isolated from Escherichia coli ML 308-225 grown anaerobically in the presence of glycerol and fumarate. Transport assays were carried out as described in legend to Fig. ll. (U-14C)lactose (20 mCi/mole) added at a final concentration of0.5 mM. A,No additions; A, Na fumarate (10 mM); 0, Na L-a-glycerol phosphate (10 mM); 0, Na-L-a-glycerol phosphate (10 mM) + Na furnarate (10 mM). From Boonstra and Konings (1976), with permission.
glycerol phosphate and fumarate, however, a stimulation of amino acid and lactose uptake is observed which is significantly higher than the sum of the stimulations exerted by the electron donor or acceptor alone (Konings and Kaback, 1973a; Boonstra et d.,1975a). In agreement with these observations, the membrane vesicles have high activities of anaerobic L-a-glycerol phosphate dehydrogenase, and fumarate reductase and fumarate reduction occurs at a high rate in the presence of L-a-glycerol phosphate (Boonstra et al., 1975a). Further evidence for the involvement of electron transfer to fumarate is provided by the observation that HQNO inhibits by more than 70% transport energized by L-a-glycerol phosphate plus fumarate (Konings and Kaback, 197313). Anaerobic transport in the presence of fumarate in these membrane vesicles is also stimulated to some extent by D-lactate. This indicates that L-a-glycerol phosphate dehydrogenase and D-lactate dehydrogenase are coupled to fumarate reductase. It is of interest that
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
21 1
these membrane vesicles reduce nitrate at a high rate in the presence of formate and that formate plus nitrate catalyzes transport of lactose even better than does L-a-glycerol phosphate plus fumarate; formate plus fumarate, however, did not stimulate transport to a significant extent. The electron donor, L-a-glycerol phosphate, in these vesicles is also coupled to nitrate reductase, and the stimulation observed with this electron donor in the presence of nitrate is even higher than with fumarate (Boonstra et al., 1975a). These observations indicate that, in membrane vesicles from cells grown anaerobically on glycerol in the presence of fumarate, two anaerobic electron transfer systems are present, both of which are coupled to anaerobic active transport. The data obtained from the uptake experiments suggest that these electron transfer systems have some common electron transfer intermediates. Moreover, these membrane vesicles contain a functional respiratory chain, and active transport can be obtained with the electron donors Asc-PMS, succinate, NADH, and D-lactate with oxygen as terminal electron acceptor (Konings and Kaback, 1973a; Boonstra et al., 1975a). D. Involvement of ATP
In membrane vesicles from E . coli grown anaerobically on either glucose plus nitrate, glycerol plus fumarate, or glucose done, all attempts to demonstrate transport with externally supplied ATP as energy source have been unsuccessful (Konings and Kaback, 1973a; Boonstraet al., 1975a). In addition, no transport was obtained in membrane vesicles prepared in the presence of an ATP-generating system, consisting of ADP, creatine phosphate and creatine phosphate kinase. These observations point to a direct coupling of electron transfer to transport, without the involvement of ATP, similar to what has been demonstrated for membrane vesicles from aerobically grown cells (Barnes and Kaback, 1970; Konings and Freese, 1972; Klein and Boyer, 1972; Prezioso et al., 1973; Or et al., 1973).Observations made with whole cells of E . coli mutants deficient in Ca2+- and Mg2+dependent ATPase (unc A) support this conclusion. These mutants perform active transport of serine and phosphate anaerobically in the presence of fumarate (Butlin, 1973; Rosenberg et al., 1975) and active transport of proline in the presence of nitrate (Yamamoto et al., 1973). Moreover, active transport of proline was observed under anaerobic conditions in membrane vesicles from the E . coli mutants DL 54 and NR 70, which lack ATPase activity. Membrane vesicles from these mutants grown anaerobically in the presence of nitrate (Fig. 14) dem-
212
WIL N. KONINGS AND JOHANNES BOONSTRA
TIME (ninuta)
FIG. 14. Uptake of L-proline under anaerobic and aerobic conditions in membrane vesicles from Escherichia coli ML 308225 (panel A) and Escherichia coli DL 54 (panel B), grown anaerobically in the presence of glucose and nitrate. Transport assays were carried out as described in legend to Fig. 11. L-(U-IIC)proline (232 mCi/mmole) was added at a final concentration of 16.6 p M . V,no additions; 0, K formate (10 mM) + K nitrate (10 mM); 0, K formate (10 m M ) oxygen; A, Li ascorbate (10 mM) + PMS (0.1 mM) + oxygen. From Boonstra et al. (1975b), with permission.
+
onstrate a high stimulation of proline uptake under anaerobic conditions, with formate as electron donor and nitrate as acceptor, and under aerobic conditions with formate or ascorbate-PMS as energy sources (Boonstra et al., 1975b). In contrast, vesicles from aerobically grown E . coli DL54 demonstrate low transport activity under aerobic conditions with D-lactate as energy source (Altendorf et al., 1974; Berger and Heppel, 1974). Such decreased transport activity under aerobic conditions was also observed in other ATPase mutants of E . coli (Schairer and Haddock, 1972; Simoni and Shallenberger, 1972; Van Thienen and Postma, 1973; Rosen and Adler, 1975). Observations in these mutants led to the hypothesis that the membrane-bound ATPase accomplished a structural role in the transport process (Rosen, 1973a; Altendorf et al., 1974; Rosen and Adler, 1975). According to this hypothesis, removal or deformation of ATPase in some mutations exposes a proton channel through the membrane. This leads to an increased proton permeability of the membrane and, consequently, to a collapse of the membrane potential. The increased proton permeability, as well as the transport defect, can be cured by the addition of carbodiimides such as DCCD (Nieuwenhuis et al., 1973; Van Thienen and Postma, 1973; Rosen, 1973a, 197313; Altendorf et al., 1974; Tsuchiya and Rosen, 1975). However, the results shown
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
21 3
in Fig. 14 indicate that this explanation does not hold for membrane vesicles from anaerobically grown ATPase mutants. These membrane vesicles have normal transport activity but lack ATPase; the addition of DCCD has no effect (Boonstra et aZ., 197513).The transport role of ATPase, therefore, seems to be unresolved at this time. There is no question, however, that ATPase activity and active transport are closely linked processes. There is evidence that, in the absence of electron flow, the energy for active transport can be supplied by ATP upon hydrolysis by the membrane-bound ATPase. For instance, mutants of E. coli that have defective heme synthesis and therefore possess a defective respiratory chain (Devor et al., 1974; Singh and Bragg, 1974) or whose ubiquinone synthesis is disturbed (Parnes and BOOS,1974),actively transport lactose, P-galactosides, and amino acids. DCCD, which acts as an inhibitor of the ATPase, strongly inhibits transport in these mutants. Observations made with Streptococcus faecalis and Streptococcus lactis demonstrate that ATP can supply the energy for active transport via the generation of a proton-motive force (see Section IV,A). Glycolyzing cells of these organisms generate a pH gradient and a membrane potential by the extrusion of protons via ATPase (Harold et al., 1970b; Harold and Papineau, 1972a,b; Kashket and Wilson, 1972). The generation of this proton-motive force is inhibited by the ATPase inhibitor DCCD (Baron and Abrams, 1971; Ashgar et al., 1973). This proton-motive force can provide the energy for active transport of nutrients, as discussed below. The preceding experiments provided evidence for a link between ATP and active transport via the generation of a proton-motive force. Such a proton-motive force does not seem to be an essential intermediate in all active transport processes. Evidence obtained in aerobically grown E . coli indicates that ATP itself is the obligatory energy source for substrate transport via shock-sensitive transport systems that are associated with periplasmic binding proteins (Berger, 1973; Berger and Heppel, 1974; Cowell, 1974; Curtis, 1974). E. Involvement of a Proton-Motive Force
The generation of a membrane potential has been demonstrated in both aerobically (Scholes and Mitchell, 1970; Jeacocke et al., 1972; Hirata et aZ., 1973; Lawford and Haddock, 1973) and anaerobically (Harold et aE., 1970a; Harold and Papineau, 1972a; Kashket and Wilson, 1974) grown cells of several microorganisms. Evidence has also been presented for proton extrusion by membrane vesicles of
214
WIL N. KONINGS AND JOHANNES BOONSTRA
aerobically grown cells during substrate oxidation via the respiratory chain (John and Hamilton, 1971; Reeves, 1971; Hirata et al., 1973, 1974; Hertzberg and Hinckle, 1974; Schuldiner and Kaback, 1975) and membrane vesicles of phototropically grown cells upon illumination (Hellingwerf et al., 1975). Proton extrusion under anaerobic conditions might be accomplished, according to Mitchell (1970), by oxidation-reduction of pairs of substrates via an oxidation chain arranged across the membrane in a loop or at the expense of glycolytic ATP by the membrane-bound ATPase. The generation of such a proton-motive force by the action of membrane-bound ATPase was demonstrated in anaerobically grown Staphylococcus aureus (Jeacocke et al., 1972) and Streptococcus faecalis (Harold et al., 1970a; Harold and Papineau, 197213);estimates of the proton-motive force in these experiments were of the order of 200 mV (interior alkaline and negative) (Harold and Papineau, 1972a). Currently available information on anaerobic electron transfer systems makes these systems, like the respiratory chain in aerobically grown cells, excellent candidates for proton extrusion. Experiments carried out recently in our laboratory demonstrate that electron flow in the nitrate reductase system results in the generation of a membrane potential (Fig. 15; Boonstra and Konings, 1976). A high rate of uptake of the lipophilic cation triphenylmethylphosphonium (TPMP+)is observed upon addition of the electron donor formate and the acceptor nitrate. Either electron donor or acceptor alone has hardly any effect. In membrane vesicles from E . coli ML 308-225, grown anaerobically in the presence of glycerol and fumarate, TPMP+ transport was stimulated under anaerobic conditions by electron flow from L-a-glycerol phosphate to fumarate (Fig. 16) and also from ~ - a glycerol phosphate or formate to nitrate. Again, hardly any stimulation of uptake was found with the electron donors or acceptors alone. At steady state levels of accumulation and membrane potential can be calculated by means of the Nernst equation. Membrane vesicles from E . coli, grown anaerobically in the presence of nitrate, generate under anaerobic conditions in the presence of formate and nitrate, a membrane potential of about - 80 mV. Under aerobic conditions with ascorbate-PMS as energy source, a membrane potential is generated of the same magnitude. In membrane vesicles from E . coli, grown anaerobically in the presence of fumarate, a membrane potential of about -76 mV is generated in the presence of glycerol-l-P and fumarate under anaerobic conditions (Boonstra and Konings, 1976). Similar results are obtained in membrane vesicles from aerobically grown E . coli (Schuldiner and Kaback, 1975).
215
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
-
25
.-uC
I
g Y
f
20
€ E
!f . =E m
15
W
5 +
10
n I n I-
S
1
2
3
4 MINUTES
5
FIG. 15. Uptake of txiphenylmethylphosphonium ion under anaerobic conditions in membrane vesicles isolated from Escherichia coli ML 308-225 grown anaerobically in the presence of glucose and nitrate. Transport assays were carried out as described in legend to Fig. 1 1 . ,H-TPMP-bromide (113.5 mCi/mmole) was added at a final concentration of 2 X 10-'M. 0,No additions; 4 KNO, (10 mM ); A,Na formate (10mM); 0,Na formate (10 mM) + KN03 (10mM). From Boonstra and Konings (1976), with pennission.
Another line of evidence for proton translocation by anaerobic electron transfer systems that use nitrate or fumarate as terminal electron acceptors comes from measurements on energy-dependent quenching from atebrin fluorescence in membrane particles from anaerobically grown E . coli that possess an inside-out orientation with respect to the original cell (Haddock and Kendall-Tobias, 1975). These studies indicate that NADH-, formate-, D-lactate-, and D,L-a-glycerol-
216
WI1 N. KONINGS AND JOHANNES BOONSTRA
-2 ._20 2
e
P
t n F E E
315
E
v
W
ak
3
a +
z+
/
/
/
/
10
I
I
5
0.
I
I
I
I
I
1
2
3
4
5
MINUTES
FIG. 16. Uptake of triphenylmethylphosphonium ion under anaerobic conditions in membrane vesicles from Escherichia coli ML 308-225,grown anaerobically in the presence of glycerol and fumarate. Transport assays were carried out as described in legend to Fig. 11. (3H)Triphenylmethylphosphonium bromide (113.5mCi/mmole) was added at a final concentration of 2 x M . 0, No additions; 4 K fumarate (10mM); A,Na L-a-glycerol phosphate (10 mM); 0, Na L-a-glycerol phosphate (10mM) + K fumarate (10mM). From Boonstra, and Konings (1976), with permission.
phosphate-dependent nitrate reduction and NADH- and formatedependent fumarate reduction generate a transmembrane pH gradient. It was found that the stoichiometry of proton translocation associated with nitrate reduction, the H+/N03- ratio, was higher than 2 in spheroplasts of E . coli grown anaerobically in the presence of nitrate. This proton translocation is sensitive to uncouplers, such as CCCP and azide (Garland et aZ., 1975). Furthermore, Singh and Bragg (1975) demonstrated with the same technique that, in membrane particles
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
217
from a cytochrome-deficient mutant of E . coli K 12, a pH gradient is generated by NADH-dependent fumarate reduction. In whole cells and membrane vesicles from aerobically grown E . coli the transmembrane pH gradient has been determined by measuring the distribution of weak acids across the membrane (Rottenberg, 1975; Padan et al., 1976; Ramos et al., 1976). In membrane vesicles from E . coli grown anaerobically in the presence of nitrate, the ApH is determined by measuring the distribution of acetate across the membrane with the flow-dialysis method (Ramos et a1., 1976). Under aerobic conditions with ascorbate-PMS as electron donor, a ApH is generated of about 1.5 pH units. However, under anaerobic conditions with formate as electron donor and nitrate as acceptor, no ApH appears to be generated. Formate itself is a weak acid and is permeable only in the undissociated form (Garland et al., 1975). At the high concentration of formate (10 mM) used in the transport studies, the accumulation of formic acid rapidly dissipates the pH gradient. The total proton-motive force in membrane vesicles under anaerobic conditions with formate-nitrate is therefore mainly composed of the membrane potential and has a magnitude of about - 80 mV (Boonstra and Konings, 1976). Studies of the relationship between a proton-motive force and nutrient transport have been done in Streptococcus faecalis and Streptococcus Zactis for the reason that these anaerobic organisms appear to lack cytochromes and do not carry out electron transfer-coupled phosphorylation. Instead, they rely entirely upon glycolysis (Harold and Papineau, 1972a; Asghar et al., 1973).This statement seems questionable, since evidence has been presented that under certain growth conditions these organisms contain anaerobic electron transfer systems (see Section II,B,3) and even cytochrome-linked electron transfer systems (Bryan-Jones and Whittenbury, 1969). They also perform electron transfer-coupled phosphorylation. Kashket and Wilson (1973) have shown that the addition of protons to the external medium of S. Zactis, which has been deprived of fermentable energy sources, provides the energy for active transport of thiomethylgalactoside (TMG) to a concentration of more than 20 times that of the external medium; a quantitative relationship exists between the proton-motive force and the degree of accumulation of TMG. The lipid-soluble DDA+ and TPMP+ are taken up also by S. faecalis in response to an electrical potential (interior negative) by the extrusion of protons or indirectly Na+ (Harold and Papineau, 1972a). An artificial membrane potential, generated by valinomycin-induced K+ efflux, also stimulated anaerobic serine and TPMP+ uptake in membrane vesicles from anaerobically grown E . coli cells (Fig. 17) (Boonstra and Kon-
21 8
WIL N. KONINGS AND JOHANNES BOONSTRA
-.C
a
0.7
C a
n e E
;0.6 E"
2
-E
0.5
Y 3
0.4-
W
z a W
v)
0.3
-
o0.1
.
2
b
,
3
4
o o 0
I
1
2
5
1
1
f
2 3 MINUTES
I
1
4
5
FIG.17. Uptake of L-serine and triphenylmethylphosphoniumion under anaerobic conditions in membrane vesicles from Escherichia coli M L 308-225,grown anaerobically in the presence of glucose and nitrate. 2 pl of membrane vesicles (about 20 mg protein/ml) in 0.1M K phosphate (pH 6.6)and valinomycin (1 nmole/mg membrane protein) were diluted into 50 pI Na phosphate (50mM) of pH 6.6(0----@or K phosphate (50mM), pH 6.6(O---O), containing L-(U-'~C)serine (156mCi/mmole) or CH) triphenylmethylphosphonium bromide (113.5mCi/mmole). L-serine concentration: 0.064 mM, TPMP+ concentration: 0.4 m M . Both the vesicle suspension and the diluting medium were equilibrated at 25°C prior to the start of the experiment. At the times indicated, the reactions were terminated and the samples assayed as described in the legend to Fig. 11. From Boonstra and Konings (1976),with permission.
ings, 1976). This stimulation was similar to that demonstrated in membrane vesicles from aerobically grown cells (Hirata et al., 1973; Schuldiner and Kaback, 1975). To a large extent, the evidence for a link between a proton-motive force and transport is based on the mode of action of uncouplers of oxidative phosphorylation. These uncouplers also inhibit transport under
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
219
anaerobic conditions (Ashgar et al., 1973). In S. faecalis and in anaerobically grown E . coli, a series of'uncouplers have been found to block transport of K+, phosphate, sucrose, several amino acids, and thiomethylgalactoside (Pavlasova and Harold, 1969). Moreover, in membrane vesicles, anaerobic transport is inhibited by uncouplers such as dinitrophenol (DNP) and carbonylcyanide rn-chlorophenylhydrazone (CCCP) (Konings and Kaback, 1973b). Additional information of a link between the proton-motive force and active transport under anaerobic conditions in membrane vesicles from E . coli comes from studies with the ionophores nigericin and valinomycin. Nigericin catalyzes an electrically neutral exchange of potassium for protons and thus collapses ApH; valinomycin specifically increases the potassium permeability and thus dissipates the A$. Glutamate uptake, energized anaerobically by formate-nitrate, is inhibited to a small extent by nigericin, while the inhibition by valinomycin is almost complete (Boonstra and Konings, 1976). ACKNOWLEDGMENT The authors would like to express their appreciation to Drs. R. N. Campagne, W. Harder, M. Knight, J. C. Kuenen, H. Veldkamp, and P. A. M. Michels for their constructive criticism of the manuscript and their valuable suggestions. Drs. E. Azoulay, C. H. MacGregor, and B. A. Haddock kindly supplied manuscripts prior to publication. We thank Mrs. J. W. Schroder-ter Avest and Miss R. C. Kalsbeek for their help in the preparation of this manuscript. REFERENCES Abrams, A., and Smith, J. B. (1971).Increased membrane ATPase and K+ transport rates in Streptococcus faecalis induced by K+ restriction during growth. Biochem. Biophys. Res. Comm. 44,1488-1495. Adams, C . A., Warnes, C. M., and Nicholas, D. J. D. (1971).A sulphite-dependent nitrate reductase from Thiobacillus hnitrijicans. Biochim. Biophys. Acta 235,
398-406. Adhya, S.,Claery, P., and Campbell, S. (1968).A deletion analysis of prophage lambda and adjacent genetic regions. Proc. Natl. Acad. Sci. U.S.A.61,956-962. Altendorf, K. H., and Staehelin, L. A. (1974).Orientation of membrane vesicles from Escherichia coli as detected by freeze-cleave electron microscopy. J. Bacteriol. 117,
888-899. Altendorf, K., Harold, F. M., and Simoni, R. D. (1974).Impairment and restoration of the energized state in membrane vesicles of a mutant of Escherichia coli lacking adenosine triphosphatase, J . Biol. Chem. 249,4587-4593. Amarasingham, C . R.,and Davis, B. D. (1965).Regulation of a-ketoglutarate dehydrogenase formation in Escherichia coli. J. Biol. Chem. 240,3554-3568. Aminuddin, M., and Nicholas, D. J. D. (1973).Sulphide oxidation linked to the r e d u e tion of nitrate and nitrite in Thiobacillus denitrificans. Biochim. Biophys. Acta 325,
81-93. Aminuddin, M., and Nicholas, D. J. D. (1974).Electron transfer during sulphide and sulphite oxidation in Thiobacillus denitrificans. J. Gen. Microbiol. 82,115-123. Asghar, S . S., Levin, E., and Harold, F. M. (1973).Accumulation of neutral amino acids
220
WIL N. KONINGS AND JOHANNES BOONSTRA
by Streptococcus faecalis. Energy coupling by a proton-motive force. J. Blol. Chem. 248,5225-5233. Aue, B. J., and Deibel, R. H. (1967). Fumarate reductase activity of Streptococcus faecalis. J . Bacteriol. 93, 1770-1776. Azoulay, E., Puig, J., and Couchoud-Beaumont, P. (1969). Etude des mutants chlorate-resistants chez Escherichia coli K-12. 1. Reconstitution in vitro de I’activit6 nitrate reductase particulaire chez Escherichia coli K-12. Biochim. Biophys. A C ~ 171, U 238-252. Azoulay, E., Couchoud-Beaumont, P., and Lebeault, J. M. (1972). Etude des mutants chlorate-resistance d’Escherichia coli KI2. Biochim. Biophys. Acta 256,670-680. Azoulay, E., Pommier, J., and Riviere, C. (1975). Membrane reconstitution in chl-r mutants ofEscherichia coli K-12.Z. Part played by phospholipids in the complementation process. Biochim. Biophys. Acta 389,236-251. Barnes, E. M., and Kaback, H. R. (1970).P-Galactoside transport in bacterial membrane preparations. Energy coupling via membrane-bound D-lactic dehydrogenase. Proc. Natl. Acad. Sci. U.S.A. 66, 1190-1198. Baron, C., and Abrams, A. (1971). Isolation of a bacterial membrane protein, nectin, essential for the attachment of adenosine triphosphatase. J. Biol. Chem. 246, 1542- 1544. Barton, L.L., LeGall, J., and Peck, H. D. (1970).Phosphorylation coupled to oxidation of hydrogen with fumarate in extracts of the sulfate reducing bacterium Desulfovibrio gigas. Biochem. Biophys. Res. Comm. 41, 1036-1042. Barton, L. L., LeGall, J., and Peck, H. D. (1972). Oxidative phosphorylation in the obligate anaerobe Desuljouibrio gigas. In “Horizons of Bioenergetics” (A. S . Pietro, and H. Gest, eds.), pp. 33-51. Academic Press, New York. from Baum, R. J., and Dolin, M. I. (1965). Isolation of 2-solanesyl-1,4-naphtoquinone Streptococcus faecalis, 1OC1.J . Biol. Chem. 240,3425-3433. Berger, E. A. (1973). Different mechanisms ofenergy coupling for the active transport of proline and glutamine in Escherichia coli. Proc. Natl. Acad. Sci. U S A . 70, 1514-1518. Berger, E. A., and Heppel, L. A. (1974). Different mechanisms of energy coupling for shock-sensitive and shock-resistant amino acid permease of Escherichia coli. 1. Biol. Chem. 249,7747-7755. Bogin, E., Higashi, T., and Brodie, A. F. (1970). Interchangeability of coupling factors from bacterial and mammalian origin. Biochem. Biophys. Res. Comm. 38,478-483. Boonstra, J., and Konings, W. N. (1976).Eur. J . Biochem. Submitted. Boonstra, J., Huttunen, M. T., Konings, W. N., and Kaback, H. R. (1975a). Anaerobic transport in Escherichia coli membrane vesicles. J. Biol. Chem. 250,6792-6798. Boonstra, J., Gutnick, D. L., and Kaback, H. R. (1975b). Physiological suppression of a transport defect in mutants of Escherichia coli deficient in Cat+, Mg++-stimulated adenosine triphosphatase.]. Bacteriol. 124, 1248-1255. Boonstra, J., Sips, H. J., and Konings, W. N. (1976). Active transport by membrane vesicles from anaerobically grown Escherichia coli energized by electron transfer to ferricyanide and chlorate. Eur. J. Biochem. 69,35-44. Boos, W. (1974). Bacterial Transport. Ann. Rev. Biochem. 43, 123-146. Boyer, P. D., and Klein, W. L. (1972). Energy coupling mechanisms in transport. In “Membrane Molecular Biology” (C. F. Fox, and A. D. Keith, eds.), pp. 323-344. Sinauer Associates, Stanford, Connecticut. Brodie, A. F., and Gutnick, D. L. (1972). Electron transport and oxidative phosphorylation in microbial systems. In “Electron and Coupled Energy Transfer in Biological
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
221
Systems” (T. E. King, and M. Klingenberg, eds.), Vol. lB, pp. 599. Dekker, New York. Bryan-Jones, D. G., and Whittenbury, R. (1969). Haematin-dependent oxidative phosphorylation in Streptococcus faecalis. J. Gen. Microbiol. 58,247-260. Butlin, J. D. (1973). Energy transformation in Escherichia coli K-12. Ph. D. thesis. Australian Natl. Univ., Canberra City, Australia. Butlin, J. D., Cox, G. B., and Gibson, F. (1971). Oxidative phosphorylation in Escherichia coli K-12. Mutations affecting magnesium ion or calcium stimulated adenosine triphosphatase. Biochem. 1. 124,75-81. Casse, F. (1970). Mapping of the gene chl B controlling membrane-bound nitrate reductase and formic hydrogen-lyase activities in Escherichia coli K-12. Biochem. Biophys. Res. Comm. 39,429-436. Chang, J. P., and Morns, J. G. (1962). Studies on the utilization of nitrate by Micrococcus denitrificans. ]. Gen. Microbiol. 29,301-310. Cheniae, G., and Evans, H. J. (1959). Properties of a particulate nitrate reductase from the nodules of the soybean plant. Biochim. Biophys. Acta 35, 140-153. Cole, J. A., and Wimpenny, J. W. T. (1968). Metabolic pathways for nitrate reduction in Escherichia coli. Biochim. Biophys. Acta 162,39-48. Cowell, J. L. (1974). Energetics of glycylglycine transport in Escherichia coli. J. Bacteriol. 120, 139-145. Cozzarelli, N. R., Freedberg, W. B., and Lin, E . C. C. (1968). Genetic control of the L-a-glycero-phosphate system in Escherichia coli. J. Mol. B i d . 31,371-387. Curtis, S . J. (1974). Mechanisms of energy coupling for transport of D-ribose in Escherichia coli. J . Bacteriol. 120, 295-303. De Groot, G. N., and Stouthamer, A. H. (1969). Regulation of reductase formation in Proteus mirabilis. 1. Formation of redudases and enzymes of the formic hydrogenlyase complex in the wild type and in chlorate-resistent mutants. Arch. Mikrobiol. 66,220-233. De Groot, G. N., and Stouthamer, A. H. (1970a). Regulation of reductase formation in Proteus mirabilis. 11. Influence of growth with azide and of haeni deficiency on nitrate reductase formation. Biochim. Biophys. Acta 208,414-427. De Groot, G. N., and Stouthamer, A. H. (1970b). Regulation of reductase formation in Proteus mirabilis. 111. Influence of oxygen, nitrate and azide on thiosulfate reductase and tetrathionate reductase formation. Arch. Mikrobiol. 74,326-329. De Groot, G. N., and Stouthamer, A. H. ( 1 9 7 0 ~ )Regulation . of reductase formation in Proteus mirabilis. IV. Influence of different growth conditions on the formation of various electron transport enzymes in a chlorate-resistant mutant. Arch. Mikrobiol. 74,340-349. Deibel, R. H., and Kvetkas, M. J. (1964). Fumarate reduction and its role in the diversion of glucose fermentation by Streptococcus faecalis. ]. Bacteriol. 88,858-864. Devor, K. A., Schairer, H. U., Renz, D., and Overath, P. (1974). Active transport of pgalactosides by a mutant of Escherichia coli defective in heme synthesis. Eur. J. Biochem. 45,451-456. De Vries, W., Van Wijck-Kapteyn, W. M. C., and Stouthamer, A. H. (1973). Generation of ATP during cytochrome-linked anaerobic electron transport in propionic acid bacteria.]. Gen. Microbiol. 76, 3141. De Vries, W., Van Wijck-Kapteyn, W. M. C., and Oosterhuis, S. K. H. (1974). The presence and function of cytochromes in Selenomonas ruminantium, Anaerouibrio lipolytica and Veillonella alcalescens. J. Gen. Microbiol. 81,69-78. Enoch, H. G., and Lester, R. L. (1972). Effects of molybdate, tungstate and selenium
222
WI1 N. KONINGS AND JOHANNES BOONSTRA
compounds on formate dehydrogenase and other enzyme systems in Escherichia coli. J . Bacteriol. 110,1032-1040. Faust, P. J., and Vandemark, P. J. (1970).Phosphorylation coupled to NADH oxidation with fumarate in Streptococcus faecalis 1OC1. Arch. Biochem. Biophys. 137, 392-398. Fewson, C. A., and Nicholas, D. J. D. (1961a).Respiratory enzymes in Micrococcus denitrljicans. Biochim. Biophys. Acta 48,208-210. Fewson, C. A., and Nicholas, D. J. D. (1961b).Nitrate-reductase from Pseudomonas aeruginosa. Biochim. Biophys. Acta 49,335-349. Forget, P. (1974).The bacterial nitrate reductases. Eur. J . Biochem. 42,325-332. Forget, P., and Dervartanian, D. V. (1972).The bacterial nitrate reductases: EPR studies on nitrate reductase A from Micrococcus denitrificans. Biochim. Biophys. Acta 256,600-606. Forget, P., and Pichinoty, F. (1964).Influence de la respiration anabrobie du nitrate et du fumarate sur le mbtabolisme fermentaire d’Aerobacter aerogenes. Biochim. Biophys. Acta 82,441444. Freedberg, W. B., and Lin, E. C. C. (1973).Three kinds of controls affecting the expression of the glp regulon in Escherichia coli. J . Bacteriol. 115,816-823. Fukuyama, T., and Ordal, E. J. (1965).Induced biosynthesis offormic hydrogenlyase in iron deficient cells of Escherichia coli. J . Bacteriol. 90,673-680. Garland, P. B., Downie, J. A,, and Haddock, B. A. (1975).Proton translocation and the respiratory nitrate reductase of Escherichia coli. Biochem. J . 152,547-559. German, N. S., Lukoyanova, M. A., and Ostrovskii, D. N. (1967).In “Respiration and Phosphorylation of Bacteria” (G. B. Pinchot, ed.), pp. 90-110.Plenum, New York. Gest, H., and Peck, H. D. (1955).A study ofthe hydrogenlyase reaction with systems derived from normal and anaerogenic coli-aerogenes bacteria. J. Bacteriol. 70, 326-334. Glaser, J. H., and DeMoss, J. A. (1971).Phenotypic restoration by molybdate of nitrate reductase activity and chl D mutants ofEscherichia coli.J . Bacteriol. 108,854-860. Glaser, J. H.,and DeMoss, J. A. (1972).Comparison of nitrate reductase mutants ofEscherichia coli selected by alternative procedures. Mol. Gen. Genet. 116,1-10. Gray, C. T., and Gest, H. (1965).Biological formation of molecular hydrogen. Science 148,186-192. Gray, C. T., Wimpenny, J. W. T., Hudges, D. E., and Mossman, M. R. (1966).Regulation of metabolism in facultative bacteria. I. Structural and functional changes in Escherichia coli associated with shifts between the aerobic and anaerobic states. Biochim. Biophys. Acta 117,22-32. Guest, J. R. (1969).Biochemical and genetic studies with nitrate reductase c-gene mutants ofEscherichia coli. Mol. Gen. Genet. 105,285-297. Gunsalus, I. C. (1947).Products of anaerobic glycerol fermentation by Streptococcus faecalis. J . Bacteriol. 54,239-244. Gutnick, D. L., Kanner, B. I., and Postma, P. W. (1972).Oxidative phosphorylation in mutants of Escherichia coli defective in energy transduction. Biochim. Biophys. Acta 283,217-222. Guymon, E. D., and Spence, J. T. (1966).The reduction of nitrate by molybdenum (V), J. Phys. Chem. 70,1964-1969. Haddock, B. A,, and Kendall-Tobias, M. W. (1975).Functional anaerobic electron transport linked to the reduction ofnitrate and fumarate in membranes from Escherichia coli as demonstrated by quenching of atebrin fluorescens. Biochem. J . 152, 655-659.
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
223
Hadjipetrou, L. P., and Stouthamer, A. H. (1965). Energy production during nitrate respiration by Aerobacter aerogenes.J. Gen. Microbiol. 3 8 , 2 9 3 4 . Hamilton, W. A. (1975). Energy coupling in microbial transport. Ado. Microbiol. Physiol. 12, 1-53. Harold, F. M. (1972). Conservation and transformation of energy by bacterial membranes. Bacteriol. Rev. 36, 172-230. Harold, F. M. (1974). Chemiosmotic interpretation of active transport in bacteria. Ann. N . Y. Acad. Sci. 227,297-311. Harold, F. M., and Papineau, D. (1972a).Cation transport and electrogenesis by Streptococcus jaecalis. I. The membrane potential. J. Membr. Biol. 8 , 2 7 4 4 . Harold, F. M., and Papineau, D. (1972b). Cation transport and electrogenesis by Streptococcus faecalis. 11. Proton and sodium extrusion. J , Membr. Biol. 8,45-62. Harold, F. M., Pavlasova, E., and Baarda, J. R. (1970a). A transmembrane pH gradient in Streptococcus faecalis: Origin and dissipation by proton conductors and N,N’dicyclohexylcarbodiimide. Biochim. Biophys. Acta 196,235-244. Harold, F. M., Baarda, J. R., and Pavlasova, E. (1970b). Extrusion of sodium and hydrogen ions as the primary process in potassium ion accumulation by Streptococcus faecalis. J . Bacteriol. 101, 152-159. Hatchikian, E. C. (1974). On the role of menaquinone-6 in the electron transport of hydrogen: Fumarate reductase system in the strict anaerobe Desulfooibrio gigas. J. Gen. Microbiol. 81,261-266. Hatchikian, E. C., and LeCall, J. (1970a). Etude du m6tabolisme des acides dicarboxyliques e t du pyruvate chez les bactkries sulfato-rkductrices. Ann. Znst. Pasteur, Paris 118,125-142. Hatchikian, E. C., and LeCall, J. (1970b). Etude du m6tabolisme des acides dicarboxyliques et du pyruvate chez les bactkries sulfato-rkductrices. Ann. Znst. Pasteur, Paris 118,288-301. Hatchikian, E. C., and LeGall, J. (1972). Evidence for the presence of a b-type cytochrome in the sulfate reducing bacterium Desulfooibrio gigas, and its role in the reduction of fumarate by molecular hydrogen. Biochim. Biophys. Acta 267, 479-4 84. Hayashi, S., Koch, J. P., and Lin, E. C. C. (1964). Active transport of L-aglycerophosphate in Escherichia coli. J. Biol. Chem. 239,3098-3105. Hellingwerf, K. J., Michels, P. A. M., Dorpema, J. W., and Konings, W. N. (1975). Transport of amino acids in membrane vesicles of Rhodopseudomonas spheroides energized by respiratory and cyclic electron flow. Eur. J . Biochem. 55,397-406. Hertzberg, E. L., and Hinckle, P. C. (1974). Oxidative phosphorylation and proton translocation in membrane vesicles prepared from Escherichia coli. Biochem. Biophys. Res. Comm. 58, 178-184. Hirata, H., Altendorf, K., and Harold, F. M. (1973). Role of an electrical potential in the coupling of metabolic energy to active transport by membrane vesicles ofEscherichia coli. Proc. Natl. Acad. Sci. U.S.A.70, 1804-1808. Hirata, H., Altendorf, K., and Harold, F. M. (1974). Energy coupling in membrane vesicles of Escherichia coli. I. Accumulation of metabolites in response to a n electrical potential. J. Biol. Chem. 249, 2939-2945. Hirsch, C. A., Rasminsky, M., Davis, B. D., and Lin, E . C. C. (1963).A fumarate reductase in Escherichia coli distinct from succinate dehydrogenase. J . Biol. Chem. 238, 3770-3774. Hobson, P. N. (1965). Continuous culture of some anaerobic and facultatively anaerobic lumen bacteria. J. Gen. Microbiol. 38, 167-180.
224
WIL N. KONINGS AND JOHANNES BOONSTRA
Hobson, P. N., and Summers, R. (1967). The continuous culture ofanaerobic bacteria.J . Gen. Microbiol. 47,53-65. Hong, J-S., and Kaback, H. R. (1972). Mutants ofSalmonella typhimurium andEscherichia coli pleiotrophically defective in active transport. Proc. Natl. Acad. Sci. U S A . 69,3336-3340. Inderlied, C. B., and Delwiche, E. A. (1973).Nitrate reduction and the growth ofveillonella alcalescens. J . Bacteriol. 114, 1206-1212. Ishaque, M., Donawa, A., and Aleem, M. I. H. (1973). Energy-coupling mechanisms under aerobic and anaerobic conditions in autotrophically grown Pseudomonas saccharophila. Arch. Biochem. Biophys. 159,570-579. Itagaki, E., Fujita, T., and Sato, R. (1961a). Solubilization and some properties of formic dehydrogenase from Escherichia coli. Biochem, Biophys. Res. Comm. 5,30-34. Itagaki, E., Fujita, T., and Sato, R. (1961b). Cytochrome b,-nitrate reductase interaction in a solubilized system from Escherichia coli. Biochim. Biophys. Acta 51, 390-392. Itagaki, E., Fujita, T., and Sato, R. (1962).Solubilization and properties of formate dehydrogenase and cytochrome b, from Escherichia coli. J . Biochem. 52,131-141. Jacobs, N. J., and Vandemark, P. J. (1960).Comparison of the mechanism of glycerol oxidation in aerobically and anaerobically grown Streptococcus faecalis. J . Bacteriol. 79,532-538. Jeacocke, R. E., Niven, D. F., and Hamilton, W. A. (1972).The proton motive force in Staphylococcus aureus. Biochem. J . 127,57P. John, P., and Hamilton, W. A. (1971). Release ofrespiratory control in particles from M i crococcus denitrificans by ion-translocating antibiotics. Eur. J . Biochem. 23, 529-532. John, P., and Whatley, F. R. (1970). Oxidative phosphorylation coupled to oxygen and nitrate reduction in Micrococcus denitrijicans. Biochim. Biophys. Acta 216, 342-352. Kaback, H. R. (1971). Bacterial membranes. Methods Enzymol. 22,99-120. Kaback, H. R. (1972).Transport across isolated bacterial cytoplasmic membranes. Biochim. Biophys. Acta 265,367-416. Kaback, H. R. (1973). Active transport in bacterial cytoplasmic membrane vesicles. Symp. Soc. E x p . Biol., 27th, pp. 145-174. Kaback, H. R. (1974a).Transport in isolated baderial membrane vesicles. Methods Enzymol. 31,698-709. Kaback, H. R. (197413).Transport studies in bacterial membrane vesicles. Science 186, 882-892. Kaback, H. R., andBarnes, E. M. (1971). Mechanism ofactive transport in isolated membrane vesicles. 11. The mechanism of energy coupling between D-lactic dehydrogenase and P-galactoside transport in membrane preparations from Escherichia coli. J. Biol. Chem. 246,5523-5531. Kaback, H. R., and Hong, J-S. (1973).Membranes and transport. CRC Crit. Reu. Microbiol. 2, 333-376. Kaback, H. R., and Milner, L. S. (1970). Relationship o f a membrane-bound D-lactic dehydrogenase to amino acid transport in isolated bacterial membrane preparations. Proc. Natl. Acad. Sci. U.S.A. 66, 1008-1015. Kashket, E. R., and Wilson, T. H. (1972).The role ofmetabolic energy in the transport of /3-galactosides by Streptococcqs lactis 7962. J . Bacterial. 109,784-789. Kashket, E. R., and Wilson, T. H. (1973).Proton-coupled accumulation of galactoside in Streptococcus faecalis. Proc. Natl. Acad. Sci. U S A . 70,2866-2869. Kashket, E. R., and Wilson, T. H. (1974). Protonmotive force in fermenting Strepto-
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
225
coccus lactis 7962 in relation to sugar accumulation. Biochem. Biophys. Res. Comm. 59,879-886. Kean, E. A., Gutman, M., and Singer, T. P. (1971). Studies on the respiratory chainlinked nicotinamide adenine dinucleotide dehydrogenase. J . Biol. Chem. 246, 2346-2353. Kerwar, G. K., Gordon, A. S., and Kaback, H. R. (1972).Mechanisms of active transport in isolated membrane vesicles. IV. Galactose transport by isolated membrane vesicles from Escherichia coli. J . Biol. Chem. 247,291-297. Kistler, W. S., and Lin, E. C. C. (1971). Anaerobic L-wglycerol-phosphate dehydrogenase of Escherichia coli: Its genetic locus and its physiological role. J. Bacteriol. 108, 1224-1234. Kistler, W. S., and Lin, E. C. C. (1972). Purification and properties of the flavinestimulated anaerobic La-glycerophosphate dehydrogenase of Escherichia coli. J . Bacteriol. 112,539-547. Kistler, W. S., Hirsch, C. A., Cozzarelli, N. R., and Lin, E. C. C. (1969).Second pyridine nucleotide-independent La-glycerophosphate de hydrogenase in Escherichia coli K-12. J . Bacteriol. 100, 1133-1135. Klein, W. L., and Boyer, P. D. (1972). Energization of active transport by Escherichia coli. J . Biol. Chem. 247,7257-7265. Knook, D. L. (1972).The electron transport system ofKlebsiella aerogenes. Ph.D. thesis. Vrije Univ., Amsterdam. Knook, D. L., and Planta, R. J. (1971).Function ofubiquinone in electron transport from NADH to nitrate and oxygen in Aerobacter aerogenes, J . Bacteriol. 105,483-488. Knook, D. L., and Planta, R. J. (1973). The function of ubiquinone in Klebsiella aerogenes. Arch. Mikrobiol. 93, 13-22. Knook, D. L., Van't Riet, J., and Planta, R. J. (1973).The participation of cytochromes in the process of nitrate respiration in Klebsiella aerogenes. Biochim. Biophys. Acta 292,237-245. Koch, J. P., Hayashi, S-I., and Lin, E. C. C. (1964).The control of dissimilation of glycerol and L-a-glycerophosphate in Escherichia coli.J . Biol. Chern. 239,3106-3108. Kodama, T. (1970).Effects of growth conditions on formation of cytochrome system of a denitrifying bacterium, Pseudomonas stutzeri. Plant Cell Physiol. 11,231-239. Kodama, T., Shimada, K., and Mori, T. (1Q69).Studies on anaerobic biphasic growth of a denitrifying bacterium, Pseudomonas stutzeri. Plant Cell Physiol. 10,855-865. Konings, W. N. (1975).Localization ofmembrane proteins in membrane vesicles ofBacillus subtilis. Arch. Biochem. Biophys. 167,570-580. Koninys, W. N., and Freese, E. (1971).Lserine transport in membrane vesicles ofBacillus subtilis energized by NADH or reduced phenazine methosulfate. FEBS Lett. 14,65-68. Konings, W. N., and Freese, E. (1972).Amino acid transport in membrane vesicles of Bacillus subtilis. J . Biol. Chem. 247, 2408-2418. Koninys, W. N., and Kaback, H. R. (1973a). Anaerobic transport in Escherichia coli membrane vesicles. Proc. N a t l . Acad. Sci. U.S.A. 70,3376-3381. Konings, W. N., and Kaback, H. R. (197313).(Unpublished results.) Konings, W. N., Barnes, E. M., and Kaback, H. R. (1971). Mechanisms ofactive transport in isolated membrane vesicles. 111. The coupling of reduced phenazine methosulfate to the concentrative uptake of P-galadosides and amino acids.]. Biol. Chem. 246, 5857-5861. Konings, W. N., Bisschop, A., Veenhuis, M., and Vermeulen, C. A. (1973). New procedure for the isolation of membrane vesicles ofBacillus subtilis and an electron microscopy study of their ultrastructure. J. Bacteriol. 116, 1456-1465.
226
WIL N. KONINGS AND JOHANNES BOONSTRA
Konings, W. N., Boonstra, J., and De Vries, W. (1975). Amino acid transport in membrane vesicles of the obligately anaerobic Veillonella alcalescens.]. Bacteriol. 122, 245-249. Kroger, A. (1974). Electron transport phosphorylation coupled to fumarate reduction in anaerobically grown Proteus rettgeri. Biochim. Biophys. Acta 347,273-289. Kroger, A,, and Dadik, V. (1969). On the role of quinone in bacterial electron transport. The respiratory system of Bacillus megatherium. Eur. 1. Biochem. 11,328-340. Kroger, A., Dadtik, V., Klingenberg, M., and Diemer, F. (1971).On the role of quinones in bacterial electron transport. Differential roles of quinone and menaquinone in Proteus rettgeri. Eur. 1.Biochem. 21,322-333. Lam, Y., and Nicholas, D. J. D. (1969a). Aerobic and anaerobic respiration in Micrococcus denitrificans. Biochim. Biophys. Acta 172,450461. Lam, Y., and Nicholas, D. J. D. (1969b). A nitrate reductase from Micrococcus denitrificans. Biochim. Biophys. Acta 178,225-234. Lam, Y., and Nicholas, D. J. D. ( 1 9 6 9 ~ )A. nitrite reductase with cytochrome oxidase activity fiom Micrococcus denitrificans. Biochim. Biophys. Acta 180,459-472. Lardy, H. A., and Ferguson, S. M. (1969). Oxidative phosphorylation in mitochondria. Ann. Rev. Biochem. 38,991-1034. Lawford, H. G., and Haddock, B. A. (1973). Respiration-driven proton translocation in Escherichia coli. Biochem. 1. 136,217-220. LeGall, J., and Postgate, J. R. (1973).The physiology of sulphate reducing bacteria.Adu. Microbial Physiol. 10, 81-133. Lester, R. L., and DeMoss, J. A. (1971). Effects of molybdate and selenate on formate and nitrate metabolism in Escherichia coli. 1.Bacteriol. 105,1006-1014. Lin, E. C. C., Koch, J. P., Chused, T. M., and Jorgensen, S. E. (1962). Utilization of L-cY-glycerophosphate by Escherichia coli without hydrolysis. Proc. Natl. Acad. Sci. U.S.A. 48, 2145-2150. Linnane, A. W., and Wrigley, G. W. (1963). Fragmentation of the electron transport chain of Escherichia coli. Preparation of a soluble formate dehydrogenasecytochrome b, complex. Biochim. Biophys. Acta 77,408-418. Lombardi, J. F., Reeves, J. P., Short, S. A., and Kaback, H. R. (1974).Evaluation of the chemiosmotic interpretation of active transport in bacterial membrane vesicles. Ann. N.Y. Acad. Sci. 227,312-327. Lowe, R. H., and Evans, H. J. (1964). Preparation and some properties o f a soluble nitrate reductase from Rhizobium japonicum. Biochim. Biophys. Acta 85,377-389. MacGregor, C. H. (1975a). Solubilization of Escherichia coli nitrate reductase by a membrane bound protease.]. Bacteriol. 121, 1102-1110. MacCregor, C. H. (1975b). Anaerobic cytochrome b, in Escherichia coli: Its association with and regulation of nitrate reductase.]. Bacteriol. 121, 1111-1116. MacGregor, C. H. (197%). Synthesis of nitrate reductase components in chlorate resistant mutants of Escherichia coli. 1.Bacteriol. 121, 1117-1121. MacGregor, C. H., and Schnaitman, C. A. (1971). Alterations in the cytoplasmic membrane of various chlorate-resistant mutants of Escherichia coli. I . Bacteriol. 108, 564-570. MacGregor, C. H., and Schnaitman, C. A. (1973). Reconstitution of nitrate reductase activity and formation of membrane particles from cytoplasmic extracts of chlorate-resistant mutants of Escherichia coli. 1.Bacteriol. 114,1164-1 176. MacGregor, C. H., Schnaitman, C. A., Normansell, D. E., and Hodgins, M. G. (1974). Purification and properties of nitrate reductase fiom Escherichia coli K-12.1. Biol. Chem. 249,5321-5327. Miki, K., and Lin, E. C. C. (1973). Enzyme complex which couples glycerol-3-P dehydrogenation to fumarate reduction in Escherichia coli. 1.Bacteriol. 114,767-771.
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
227
Miller, J. D. A., and Wakerley, D. S . (1966).Growth of sulfate-reducing bacteria by fumarate dismutation. J. Gen. Microbiol. 43, 101-107. Mitchell, P. (1961).Coupling of phosphorylation to electron and hydrogen transfer by a chemiosmotic type o f mechanism. Nature (London)191, 144-148. Mitchell, P. (1963). Molecule, group, and electron translocation through natural membranes. Biochem. SOC. Symp. 22,142-168. Mitchell, P. (1966).Chemiosmotic coupling in oxidative and photosynthetic phosphorylation. Biol. Reu. Cambridge Phil. SOC.41,445-502. Mitchell, P. (1970). Membranes of cells and organelles: Morphology, transport and metabolism. Symp. SOC. Gen. Microbio2. 20, 121-166. Mitchell, P. (1973). Performance and conservation of osmotic work by proton-coupled solute porter systems. Bioenergetics 4, 63-91. Mutaftchiev, S., and Azoulay, E. (1973). Membrane reconstitution in chl-r mutants of Escherichia coli K-12. VI. Morphological study of membrane assembly during complementation between extracts of chl-r mutants. Biochim. Biophys. Acta 307, 525-540. Naik, M. S., and Nicholas, D. J. D. (1966).Phosphorylation associated with nitrate and nitrite reductase in Micrococcus denitrificans and Pseudomonas dmitrificans.Biochim. Biophys. Acta 113,490-497. Nicholas, D. J. D., and Nason, A. (1955). Diphosphopyridine nucleotide-nitrate reductase from Escherichia coli. J. Bacteriol. 69,580-583. Nieuwenhuis, F. J. R. M., Kanner, B. I., Gutnick, D. L., Postma, P. W., and van Dam, K. (1973).Energy conservation in membranes of mutants ofEscherichia coli defective in oxidative phosphorylation. Biochim. Biophys. Acta 325,62-71. O'Hara, J., Gray, C. T., Puig, J., and Pichinoty, F. (1967).Defects on formate hydrogenlyase in nitrate reductase-negative mutants ofEscherichia coli. Biochem. Biophys. Res. Comm. 28,951-957. Ohnishi, T. (1963). Oxidative phosphorylation coupled with nitrate respiration with cell-free extracts of Pseudomonas denitrificans. J. Biochem. 53,71-79. Or, A., Kanner, B. I., and Gutnick, D. L. (1973).Active transport in mutants of Escherichia coli with alterations in the membrane adenosine triphosphatase complex. FEBS Lett. 35,217-219. Ota, A., Yamanaka, T., and Okunuki, K. (1964).Oxidative phosphorylation coupled with nitrate respiration. 11. Phosphorylation coupled with anaerobic nitrate reduction in a cell-free extract o f Escherichia coli. Biochem. J . 55,131-135. Padan, E., Zilberstein, D., and Rottenberg, H. (1976).The proton electrochemical gradient in Escherichia coli cells. Eur. J . Biochem. 63,533-541. Parnes, J. R., and Boos, W. (1973).Energy coupling of the a-methylgalactoside transport system of Escherichia coli. J. Biol. Chem. 248,4429-4435. Parson, W. W. (1974). Bacterial photosynthesis. Ann. Rev. Microbiol. 28,41-59. Pavlasova, E., and Harold, F. M. (1969). Energy coupling in the transport o f /3galactosides by Escherichia coli: Effect of proton conductors. J. Bacteriol. 98, 198-204. Payne, W. J. (1970).Energy yields and growth of heterotrophs. Ann. Reo. Microbiol. 24, 17-52. Peck, H. D., and Gest, H. (1957). Formic dehydrogenase and the hydrogenlyase enzyme complex in coli-aerogenes bacteria. J. Bacteriol. 73,706-721. Peck, H. D., Smith, 0. H., and Gest, H. (1957).Comparative biochemistry ofthe biological reduction of fumaric acid. Biochim. Biophys. Acta 25, 142-147. Pichinoty, F. (1960).Rbduction assimilative du nitrate par les cultures akrobies d'Aerobacter aerogenes. Influence de la nutrition azot6e sur la croissance. Folia Microbiol. (Prague) 5, 165-170.
228
WIL N. KONINOS AND JOHANNES BOONSTRA
Pichinoty, F. (1962). Inhibition par l’oxygbne de la biosynthese et de I’activitb d e l’hydrogenase et de hydrogen-lyase chez les bactbries anaerobies facultatives. Biochim. Biophys. Acta 64, 111-124. Pichinoty, F. (1965). L’effet oxygbne de la biosynthbse des enzymes d’oxydorkduction bactkriens. Cent. Natl. Rech. Sci. Symp. 124,507-522. Pichinoty, F., and Coudert, G. (1962).Action rkpressive de l’oxygbne sur la biosynthbse de la fumarique-rkductase d’derobacter aerogenes. Experientia 18,257-259. Pikchaud, M., Pichinoty, F., Azoulay, E., and Le Minor, L. (1967). Mutations affectant la nitrate-rkductase A et d’autres enzymes bactkriennes d’oxydorbduction. Etude prkliminaire. Ann. Inst. Pasteur, Paris 112,24-37. Pinsent, J. (1954). The need for selenite and molybdate in the formation offormic dehydrogenase by members of the coli-aerogenes group of bacteria. Biochem. J . 57, 10-16. Porra, R. J., and Lascelles, J. (1965). Haemproteins and haem synthesis in facultative photosynthetic and denitrifying bacteria. Biochem. J . 94,120-126. Postgate, J. R. (1965). Recent advances in the study of the sulfate-reducing bacteria. Bacteriol. Reu. 29,425-442. Pressman, B. C. (1970). Energy-linked transport in mitochondria. In “Membranes of Mitochondria and Chloroplasts” (E. Racker, ed.), pp. 213-250. Van Nostrand-Reinhold, Princeton, New Jersey. Prezioso, C., Hong, J-S., Kerwar, G. K., and Kaback, H. R. (1973).Mechanisms ofactive transport in isolated bacterial membrane vesicles. XII. Active transport by a mutant of Escherichia coli uncoupled for oxidative phosphorylation. Arch. Biochem. Biophys. 154,575-582. Racker, E. (1970). Function and structure of the inner membrane of mitochondria and chloroplasts. In “Membranes of Mitochondria and Chloroplasts” (E. Racker, ed.), pp. 127-171. Van Nostrand-Reinhold, Princeton, New Jersey. Ramos, S., Schuldiner, S., and Kaback, H. R. (1976). The electrochemical gradient of protons and its relationship to active transport in Escherichia coli membrane vesicles. Proc. Natl. Acad. Sci. U.S.A. 73, 1892-1896. Reeves, J. P. (1971).Transient pH changes during D-lactate oxidatioaby membrane vesicles. Biochem. Biophys. Res. Comm. 45,931-936. Riviere, C., and Azoulay, E. (1971). Reconstitution in uitro of membranous particles by complementation between extracts of chl-r mutants in Escherichia coli K-12. Biochem. Biophys. Res. Comm. 45,1608-1614. Riviere, C., Giordano, G., Pommier, J., and Azoulay, E. (1975). Membrane reconstitution in chl-r mutants of Escherichia coli K-12. VIII. Purification and properties of the FAfactor, the product of the chl B gene. Biochim. Biophys. Acta 389,219-231. Rizza, V., Sinclair, P. S., White, D. C., and Cuorant, P. R. (1968). Electron transport of the protohaem requiring anaerobe Bacteroides melaninogenicus. J . Bacteriol. 96, 665-67 1. Rosen, B. P. (1973a). Restoration of active transport in an Mg2+-adenosine triphosphatase-deficient mutant of Escherichia coli. J . Bacteriol. 116,1124-1129. Rosen, B. P. (1973b). P-galactoside transport and proton movement in an adenosine triphosphatase-deficient mutant of Escherichia coli. Biochem. Biophys. Res. Comm. 53,1289-1296. Rosen, B. P., and Adler, L. W. (1975). The maintenance of the energized membrane state and its relation to active transpofi inEscherichia coli. Biochim. Biophys. Acta 387,23-36. Rosenberg, H., Cox, C. B., Butlin, J. D., and Cutowski, S. J . (1975). Metabolic transport in mutants of Escherichia coli K-12 defective in electron transport and coupled phosphorylation. Biochem. J . 146,417-423.
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
229
Rottenberg, H. (1975). The measurements of transmembrane electrochemical proton gradients. J. Bioenerg. 7, 61-74. Ruch, F. E., Lengeler, J., and Lin, E . C. C. (1974).Regulation ofglycerol catabolism in Klebsiella aerogenes. 1.Bacteriol. 119,50-56. Ruiz-Herrera, J., and Alvarez, A. (1972). A physiological study of formate dehydrcgenase, formate oxidase and hydrogenlyase from Escherichia coli K- 12. Antonie oan Leeuwenhoek; J . Microbiol. Serol. 38,479-491. Ruiz-Herrera, J., and DeMoss, J. A. (1969). Nitrate reductase complex of Escherichia coli K-12: Participation of specific formate dehydrogenase and cytochrome b, components in nitrate reduction. J. Bacteriol. 99, 720-729. Ruiz-Herrera, J., Showe, M. K., and DeMoss, J. A. (1969).Nitrate reductase complex of Escherichia coli K-12; isolation and characterization of mutants unable to reduce nitrate.J. Bacteriol. 97, 1291-1297. Ruiz-Herrera, J., Alvarez, A., and Figueroa, I. (1972). Solubilization and properties of formate dehydrogenases from the membrane of Escherichia coli. Biochim. Biophys. Acta 289,254-261. Sasarman, A., Surdeanu, M., Portelange, V., Dobardzic, R., and Sonea, S. (1971). Classification of vitamin K-deficient mutants of Staphylococcus aureus, J. Gen. Microbiol. 65, 125-130. Sasarman, A., Purvis, P., and Portelange, V. (1974). Role of menaquinone in nitrate respiration in Staphylococcus aureus. J. Bacteriol. 117,911-913. Sato, R., and Egami, F. (1949). Studies on nitrate reductase. Bull. Chem. Soc.Jpn. 22, 137- 143. Schairer, H. U., and Haddock, B. A. (1972). P-galactoside accumulation in a Mg2+-, Ca2+-activated adenosine triphosphatase deficient mutant of Escherichia coli. Biochem. Biophys. Res. Comm. 48,544-551. Scholes, P. B., and Mitchell, P. (1970).Respiration driven proton translocation in Micrococcus aknitrijcans. Bioenergetics 1,61-72. Scholes, P. B., and Smith, L. (1968). Composition and properties of the membranebound respiratory chain system of Micrococcus denitrificans. Biochim. Biophys. Acta 153,363-375. Schuldiner, S., and Kaback, H. R. (1975). Membrane potentials and active transport in membrane vesicles from Escherichiu coli. Biochemistry 14,5451-5461. Short, S . A,, Kaback, H. R., Kaczorowski, G., Fisher, J., Walsh, C. T., and Silverstein, S. C. (1974). Determination of the absolute number ofEscherichia coli membrane vesicles that catalyze active transport. Proc. Natl. Acad. Sci. U.S.A. 71,5032-5936. Showe, M. K., and DeMoss, J. A. (1968). Localization and regulation of synthesis of nitrate reductase in Escherichia coli.]. Bacteriol. 95, 1305-1313. Shum, A. C., and Murphy, J. C. (1972). Effects of selenium compounds on formate metabolism and coincidence of selenium-75 incorporation and formic dehydrogenase activity in cell-free preparations of Escherichia coli. J. Bacteriol. 110, 447-449. Simoni, R. D., and Shallenberger, M. K. (1972). Coupling of energy to active transport of amino acids in Escherichiu coli. Proc. Natl. Acad. Sci. U.S.A. 69,2663-2667. Sinclair, P. P., and White, D. C. (1970).Effect ofnitrate, fumarate and oxygen on the formation of the membrane-bound electron transport system of Haernophilus paruinjluenzae. J . Bacteriol. 101,365-372. Singh, A. P., and Bragg, P. D. (1974).Energization of phenylalanine transport and energy dependent transhydrogenase by adenosine triphosphate in cytochromedeficient Escherichia coli K-12. Biochem. Biophys. Res. Comm. 57, 1200-1206. Singh, A. P., and Bragg, P. D. (1975). Reduced nicotinamide adenine dinucleotide dependent reduction of fumarate coupled to membrane energization in a cy-
230
WIL N. KONINGS AND JOHANNES BOONSTRA
tochrome deficient mutant of Escherichia coli K-12.Biochim. Biophys. Acta 396, 229-241. Skulachev, V. P. (1971).Energy transformation in the respiratory chain. Curr. Top. Bioenerg. 4, 127-190. Slater, E. C. (1971).The coupling between energy-yielding and energy-utilizing r e a c tions in mitochondria. Quant. Reu. Biophys. 4,35-72. Sone, N. (1972).The redox reactions in propionic acid fermentation. I. Occurrence and nature of an electron transfer system in Propionibacterium arabinosum. J. Biochem. 71,931-940. Spencer, M. E., and Guest, J. R. (1973).Isolation and properties o f fumarate reductase mutants of Escherichia coli. J . Bacteriol. 114,563-570. Spencer, M.E., and Guest, J. R. (1974a).Proteins of the inner membrane ofEscherichia coli: Identification of succinate dehydrogenase by polyacryl gel electrophoresis with sdh amber mutants. J . Bacteriol. 117,947-953. Spencer, M.E., and Guest, J. R. (1974b).Proteins of the inner membrane ofEscherichia coli: Changes in composition associated with anaerobic growth and h a r a t e reductase amber mutation. ]. Bacteriol. 117,954-959. Stiefel, E. I. (1973).Proposed molecular mechanism for the action of molybdenum in enzymes: Coupled proton and electron transfer. Proc. Natl. Acad. Sci. U.S.A. 70, 988-992. Stouthamer, A. H. (1967a).Nitrate reduction in Aerobacter aerogenes. I. Isolation and properties of mutant strains blocked in nitrate assimilation and resistant against chlorate. Arch. Mikrobiol. 56,68-75. Stouthamer, A. H.(196%). Nitrate reduction in Aerobacter aerogenes. 11. Characterization of mutants blocked in the reduction of nitrate and chlorate. Arch. Mikrobiol. 56,76-80. Stouthamer, A. H.(1969).A genetical and biochemical study of chlorate-resistant mutants of Salmonella typhimurium. Antonie van Leeuwenhoek; J. Microbiol. Serol. 35,505-521. Stouthamer, A. H. (1973).A theoretical study on the amount of ATP required for synthesis of microbial material. Antonie van Leeuwenhoek; J . Microbiol. Serol. 39, 545-565. Stouthamer, A. H., and Bettenhausen, C. (1972).Influence of hydrogen acceptors on growth and energy production of Proteus mtrabilis. Antonie van Leeuwenhoek; J . Microbiol. Serol. 38,Bl-90. Stouthamer, A. H., and Bettenhausen, C. (1973).Utilization of energy for growth and maintenance in continuous and batch cultures of microorganisms. Biochim. Biophys. Acta 301,53-70. Takahashi, H., Taniguchi, S., and Egami, F. (1957).Nitrate reduction in aerobic bacteria and that in Escherichia coli coupled in phosphorylation. J. Biochem. (Tokyo)43, 223-230. Taniguchi, S., and Itagaki, E. (1959).Solubilization and purification of particulate nitrate reductase of anaerobically grown Escherichia coli. Biochim. Biophys. Acta 31, 294-295. Taniguchi, S., and Itagaki, E. (1960).Nitrate reductase of nitrate respiration type from Escherichia coli. I. Solubilization and purification from the particulate system with molecular characterization as a metalloprotein. Biochim. Biophys. Acta 44, 263-279. Taniguchi, S., Sato, R., and Egami, F. (1956).The enzymatic mechanism of nitrate and nitrite metabolism in bacteria. In “Inorganic Nitrogen Metabolism” (W. D. McElroy, and B. Glass, eds.), pp. 87-108.Johns Hopkins Press, Baltimore, Maryland.
ANAEROBIC ACTIVE TRANSPORT IN BACTERIA
231
Tsuchiya, T., and Rosen, B. P. (1975). Restoration of active calcium transport in vesicles of an Mg2+-CaZ+-ATPasemutant of Escherichia coli by wild type Mg*+-ATPase. Biochem. Biophys. Res. Comm. 63,832-838. Van Dam, K., and Meyer, A. J. (1971).Oxidation and energy conservation by mitochondria. Ann. Rev. Biochem. 40, 115-160. Van Hartingsveldt, J., and Stouthamer, A. H. (1973). Mapping and characterization of mutants ofPseudomonas aeruginosa affected in nitrate respiration, i n aerobic or anaerobic growth. J. Gen. Microbiol. 74, 97-106. Van Hartingsveldt, J., and Stouthamer, A. H. (1974).Properties of a mutant of Pseudomonas aeruginosa affected in aerobic growth. J . Gen. Microbiol. 8 3 , 3 0 3 4 1 0 . Van Hartingsveldt, J., Marinus, M. G., and Stouthamer, A. H. (1971). Mutants ofpseudomonas aeruginosa blocked in nitrate or nitrite dissimilation. Genetics 67,469-482. Van Thienen, G., and Postma, P. W. (1973).Coupling between energy conservation and active transport of serine in Escherichia coli. Biochim. Biophys. Acta 323,429-440. Van 't Riet, J., Stouthamer, A. H., and Planta, R. J. (1968).Regulation of nitrate assimilation and nitrate respiration i n Aerobacter aerogenes. J. Bacteriol. 96,1455- 1464. Venables, W. A., and Guest, J. R. (1968).Transduction of nitrate reductase loci of Escherichia co2i by phages P1 and A. Mol. Gen. Genet. 103,127-140. Venables, W. A., Wimpenny, J. W. T., and Cole, J. A. (1968).Enzymic properties of a mutant of Escherichia coli K-12 lacking nitrate reductase. Arch. Mikrobiol. 63, 117-121. Verhoeven, W., Koster, A. L., Van Nievelt, M. C. A. (1954).Studies on true dissimilatory nitrate reduction. Antonie van Leeuwenhoek; J . Microbiol. Serol. 20,273-284. West, I. C. (1970). Lactose transport coupled to proton movements inEscherichia coli. Biochem. Biophys. Res. Comm. 41,655-661. West, I. C., and Mitchell, P. (1972). Proton coupled P-galactoside translocation in non-metabolizing Escherichia coli. J. Bioenerg. 3,445-462. West, I. C., and Mitchell, P. (1973). The stoichiometry of lactose-proton symport across the plasma membrane of Escherichia coli. Biochem. J . 132,587-592. White, D. C. (1962). Cytochrome and catalase patterns during growth of Haemophilus parainfluenzae. J . Bacteriol. 83,851-859. White, D. C. (1963).Respiratory systems in the hemin-requiring Haemophilus species. J . Bacteriol. 85, 84-96. White, D. C. (1966).The obligatory involvement ofthe electron transport system in the catabolic metabolism of Haemophilus parainfluenzae. Antonie van Leeuwenhoek; J . Microbiol. Serol. 32, 139-158. White, D. C., and Sinclair, P. R. (1971).Branched electron-transport systems in bacteria. Adv. Microbial Physiol. 5, 173-211. White, D. C., Bryant, M. P., and Caldwell, D. R. (1962). Cytochrome-linked fermentation in Bacteroides ruminicola. J. Bacteriol. 84,822-828. Wimpenny, J. W. T., and Cole, J. A. (1967). The regulation of metabolism in facultative bacteria. 111. The effect of nitrate. Biochim. Biophys. Acta 148, 233-242. Yamamoto, T. H., MBvel-Ninio, M., and Valentine, R. C. (1973). Essential role of membrane ATPase or coupling factor for anaerobic growth and anaerobic active transport in Escherichia coli. Biochim. Biophys. Acta 314,267-275. Yamanaka, T., Ota, A., and Okunuki, K. (1962). Oxidative phosphorylation. I. Evidence for phosphorylation coupled with nitrate reduction in a cell-hee extract of Pseudomonas aeruginosa. J . Biochem. (Tokyo) 51,253-258.
This Page Intentionally Left Blank
Protein Kinases and Membrane Phosphorylation M . MARLENE HOSEY AND MARZANO TAO Department of Biological Chemistry University of Illinois at the Medical Center Chicago. Illinois
I . Introduction .......................................................... I1. Protein Kinases ....................................................... A . Introduction ..................................................... B . Cyclic AMP-Dependent Protein Kinases ........................... C . Cyclic AMP-Independent Protein Kinases ......................... D . Membrane-Bound Protein Kinases ................................ 111. Membrane Phosphorylation . . . . . . . . . . . ............................ A . Introduction ..................................................... B . Erythrocyte Membrane Phosphorylation ........................... C. Rhodopsin Phosphorylation ....................................... D . Muscle Membrane Phosphorylation ............................... E . Synaptic Membrane Phosphorylation .......................... F . Myelin Phosphorylation . . . . . . . . . . . . .......................... G . Microtubule Phosphorylation ..................................... H . Phosphorylation of Other Membranes ............................. IV . Membrane-Bound Phosphoprotein Phosphatases ....................... A . Introduction ..................................................... B . Erythrocytes ..................................................... C . Corpus Luteum .................................................. D . Cardiac Muscle .................................................. E . Synaptosomes ..... ............................................ F . Conclusion . . . . . . . ............................................ V . Concluding Remarks ..................................................
........................................................
.
1
233 237 237 237 251 254 258 258 259 271 276 286 291 293 297 299 299 300 301 301 302 303 303 304
INTRODUCTION
In recent years. we have witnessed an explosion of interest in the study of protein kinases and of protein phosphorylation . Two major factors that have undoubtedly contributed to this increased interest 233
234
M. MARLENE HOSEY AND MARIANO TAO
are the observations that (i) protein kinase plays an important role in the action of cyclic AMP; and that (ii) protein phosphorylation is a widely occurring phenomenon. Also, the realization that protein phosphorylation-dephosphorylation represents a powerful mechanism for the regulation of cellular processes may have provided further impetus. The hormonal control of glycogen metabolism, which has been the subject of many review articles (Lamer and Villar-Palasi, 1971; Walsh and Krebs, 1973), provides a classical example of this type of regulation. Figure 1 illustrates the opposing effects of protein phosphorylation on the synthesis and the degradation of glycogen. In this scheme, a series of phosphorylation reactions, involving cyclic AMPdependent protein kinase and phosphorylase kinase, activate glycogen phosphorylase which in turn catalyzes the phosphorolysis of glycogen, Concurrently, the synthesis of glycogen is halted by the phosphorylation of glycogen synthetase, which converts the enzyme into a less active form that depends on glucose-6-phosphate for full activity. The purpose of this review is to attempt to relate protein phosphorylation to various membrane functions. The possibility that protein phosphorylation may play a role in the regulation of membrane functions is suggested by the observation that endogenous membrane phosphorylation occurs widely and by the finding that phosphate acceptor proteins, protein kinase, and phosphoprotein phosphates are HORMONE
n
V
ADENYLATE CYCLASE ATP
1 . a
C Y C L I C AMP
PROTEIN K I N A S E
PHOS b K I N A S E (INACTIVE) ATP
-k:: GLYCOGEN SYNTHETASE
I
GLYCOGEN SYNTHETASE D
PHOS h ATP
H
O
S a
ADP
FIG. 1. The phosphorylation and regulation of glycogenolytic enzymes.
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
235
inherent components of many cell membranes. These components of the phosphorylation-dephosphorylation reaction have been partially characterized in some membrane systems, whereas in others, the study of membrane phosphorylation is still emerging from the descriptive phase. In considering the physiological significance of membrane phosphorylation, conclusive evidence is scarce and speculation abounds. A major factor that has undoubtedly contributed to our current state of confusion is our lack of a thorough knowledge of the composition, organization, dynamics, and activities of the cell membrane. The significance of membrane phosphorylation has been studied using a number of different approaches. Attempts have been made to assess the function of cyclic AMP-dependent membrane phosphorylation by extrapolating back to the role of cyclic A M P in that particular cell. This is possible for those systems where the effects of cyclic AMP are known. However, the interpretation relies on the supposition that all cyclic AMP actions are mediated by protein kinase. Although the activation of protein kinases represents a major mechanism by which cyclic AMP carries out its function as a second messenger in the transmission of hormonal signals, it does not appear to be the only mechanism of cyclic AMP action. For example, cyclic ,4MP regulates gene expression in bacteria by enhancing mRNA synthesis through the action of a binding protein that is devoid of protein kinase activity (Perlman and Pastan, 1971). In addition, sevreal recent reports have suggested tha't other mediators of cyclic AMP action may exist. Yuh and Tao (1974) have purified to homogeneity two protein factors from rabbit erythrocytes that have high affinity for cyclic AMP but are unrelated to the cyclic AMP-dependent protein kinases. A somewhat similar protein factor has been isolated by Tsuzuki and Kiger (1975) from Drosophilu m e h o g a s t e r . Dgiskeland and Ueland (1975) have reported that mouse liver cystol contains a cyclic AMP receptor whose binding capacity is enhanced by Mg-ATP. Although the role of these receptors has not been defined, it seems likely that they may also mediate the action of cyclic AMP. Comparative studies of the phosphorylation of normal and abnormal cell membranes can be fruitful in elucidating the functional significance of membrane phosphorylation. If the membrane defect is known and protein phosphorylation is altered, a correlation can be made between the two phenomena. A possible drawback to this approach is the availability of suitable systems or materials. However, this type of study has been used to correlate spherocytosis to abnormal membrane phosphorylation, but with conflicting results (Greenquist and Shohet, 1974; Zail and van den Hoek, 1975).
236
M. MARLENE HOSEY AND MARIANO TAO
In spite of the difficulties encountered in ascertaining the significance of membrane phosphorylation, at least one membrane-related phenomenon, ion permeability, has been implicated in almost all types of membrane phosphorylation. That permeability is altered by phosphorylation has been suggested for synaptic membranes (Greengard, 1975), rod outer segments (Weller et al., 1975a; Miller et al., 1975), cardiac sarcoplasmic reticulum (Tada et al., 1974, 1975a; LaRaia and Morkin, 1974), toad bladder membranes (Greengard, 1975), and .avian erythrocyte membranes (Rudolph and Greengard, 1974), to mention a few. The possibility that other membrane-related phenomena, such as cell morphology (Greenquist and Shohet, 1974), enzyme activity (Constantopoulos and Najar, 1973; Tria et al., 1974), hormone release processes (Lemay et al., 1974), and platelet aggregation (Steiner, 1975), may be altered by phosphorylation has also been suggested. In this review, we will attempt to logically synthesize what is known about membrane phosphorylation and to relate this information to physiologically relevant processes, This review consists of three main parts. The first deals with the general properties of protein kinases. In this regard, we have restricted our discussion to those kinases that catalyze the phosphorylation of seryl and/or threonyl residues of proteins. Considerable attention will be directed towards those protein kinases found in the soluble fraction since these enzymes have been characterised extensively. It is hoped that the information obtained from studying the soluble enzymes will be useful in analyzing and understanding the nature of the membrane-bound kinases. In addition, attempts will be made to distinguish those protein kinases, either soluble or membrane bound, which are activated by cyclic nucleotides from those which are not. In the second part, membrane phosphorylation will be analyzed in detail by considering a number of different membrane systems. This is necessary since not all membranes have the same composition, structure, and function. Evidence will be presented, together with speculation, concerning the role of protein phosphorylation in the regulation of membrane activities. Finally, the last part will contain some fragmentary information regarding the dephosphorylation reaction and phosphoprotein phosphatases. Although phosphoprotein phosphatases are no less important than protein kinases in the regulatory process, these enzymes have received little attention recently. Admittedly, many parts of this review will, of course, be incomplete due to lack of data and/or of suitable systems to approach the problems. It is hoped that the reader will understand these limitations
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
237
in the formulation of models that attempt to illustrate certain aspects of membrane phosphorylation. II. PROTEIN KINASES A. Introduction
Protein kinase refers to a class of relatively nonspecific enzymes that catalyze the phosphorylation of more than one type of protein. The overall reaction may be written in general form as Protein
+ NTP 5Protein-PO, + NDP
(1)
A divalent cation, mainly M$+, is required for the reaction. In most instances, ATP serves as the phosphoryl donor, although in some cases, GTP has also been shown to be a substrate as well. The phosphate is incorporated into the hydroxyl group of the serine residue to form a phosphomonoester linkage. However, reactions resulting in the formation of phosphothreonyl linkage have also been demonstrated. The first observation of protein kinase activity was made about 20 years ago by Burnett and Kennedy (1954) who described a liver enzyme that catalyzes the phosphorylation of casein. Yet, it was not until 1968, when Walsh et al. (1968) showed that certain protein kinases are activated by cyclic AMP, that the study of protein kinases attracted many investigators. This is to be expected since there is already a great deal of interest in the role of cyclic AMP as a regulatory agent. After the discovery of cyclic AMP-dependent protein kinase, it became necessary to classify protein kinases into cyclic AMP-dependent and -independent enzymes in order to distinguish between hormone related and unrelated processes. However, this type of classification can be ambiguous unless the properties of the protein kinase in question are well characterized because cyclic AMP-dependent protein kinases can be converted into independent forms by dissociation of their subunits. B. Cyclic AMP-Dependent Protein ffinases
1. DISTRIBUTION AND MULTIPLICITY Cyclic AMP-dependent protein kinase was first isolated from rabbit skeletal muscle by Krebs and colleagues (Walsh e t al., 1968). The en-
238
M. MARLENE HOSEY AND MARIANO TAO
zyme catalyzes the phosphorylation and acfivation of phosphorylase kinase. In addition, it was found that other proteins, such as casein and protamines, can also serve as substrates, a finding that has markedly facilitated assay and study of the enzyme. Following the initial report from Krebs’ laboratory, Langan (1968) showed that a protein kinase isolated from calf liver which catalyzes the phosphorylation of histones and protamines is also activated by cyclic AMP. Thereafter, the enzyme has been found to be present in every animal tissue examined (Kuo and Greengard, 1969), including the nonnucleated red blood cell (Tao et al., 1970). The widespread occurrence of cyclic AMP-dependent protein kinases prompted Kuo and Greengard (1969) to propose that the vaned effects of cyclic AMP are mediated by protein kinases. However, the sweeping generalization is not universally valid for reasons mentioned earlier. Multiple forms of cyclic AMP-dependent protein kinases have been found in many tissues (Tao et al., 1970; Chen and Walsh, 1971; Tao and Hackett, 1973; Corbin et al., 1975). Tao et al. (1970), employing DEAE-cellulose chromatography, were the first to separate the rabbit red blood cell enzymes into two active peaks, I and I1 (Fig. 2). The fraction I1 shown in Fig. 2 was further resolved into two peaks (IIa and IIb) of cyclic AMP-dependent protein kinase activity on QAESephadex ion exchange. The molecular weights (MW) of the three
6ol
0
I
rO.3bM
X
-
2
x
: w
- 0
FRACTION
NUMBER
FIG.2. Separation of rabbit erythrocyte cyclic AMP-dependent protein kinase fractions I and I1 by DEAE-cellulose chromatography.Data from Tao and Hackett (1973).
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
239
enzyme fractions differ. The MW of kinases I, IIa, and IIb have been estimated by gel filtration to be 170,000, 120,000, and 240,000, respectively. Chen and Walsh (1971) have separated the rat liver protein kinases into three active fractions. Two of these fractions are dependent on cyclic AMP for activity and have similar sedimentation coefficients (6.8 S). The third fraction, 4.0 S, is not stimulated by cyclic AMP, but may be derived from the first two. The separation of rat liver cyclic AMP-dependent protein kinases into two active fractions has also been reported by Yamamura et al. (1971). They showed that the two fractions were not artifacts due to proteolysis during purification. Rabbit skeletal muscle contains at least three fractions of cyclic AMP-dependent protein kinases, with sedimentation coefficients of 6.8 S, 4.9 S, and 4.8 S (Reimann et al., 1971). Three cyclic AMP-dependent protein kinases are also found in bovine epididymal spermatozoa, with MWs of 120,000, 78,000, and 56,000 (Garbers et al., 1973). Multiple forms of cyclic AMP-dependent protein kinases with MWs of 280,000, 140,000, and 90,000 have been found in bovine heart muscle (Rubin et al., 1972), but the two lower molecular weight species may be dissociation or degradation products of the largest species. Why cyclic AMP-dependent protein kinases are heterogenous is unknown. They may represent artifacts arising from degradation, dissociation, or association during isolation (Rubin et al., 1972; Corbin et al., 1972), but it seems fairly certain that there are at least two cyclic AMP-dependent protein kinase species, which correspond to the two major fractions, I and I1 (Fig. 2), eluted from DEAE-cellulose columns (Corbin et al., 1975; Hofmann et al., 1975). Various tissues contain either one or the other, or a mixture of both (Corbin et a1 ., 1975). The possibility that one type may be produced from the other by dissociation of subunits or by proteolytic digestion has been ruled out (Corbin et al., 1975). Corbin et al. (1975) have attempted to characterize the two types on the basis of the ability of histone or 0.5M NaCl to convert them into cyclic AMP-independent forms, Type I enzyme dissociates rapidly at 30°C in the presence of 0.5 M NaCl or 0.7 mg/ml histone, whereas the type I1 enzyme dissociates only slightly. Furthermore, the type I enzyme is generally more dependent on cyclic AMP than the type I1 enzyme. This suggests that the type I1 holoenzyme may be partially active in the absence of cyclic AMP. Based on these distinctions, Corbin et al. (1975) showed that rat adipose tissue contains almost exclusively the type I1 enzyme. The kinases from rat brain and pig stomach mucosa also chromatograph as predominantly type I1 en-
240
M. MARLENE HOSEY AND MARIANO TAO
zymes, although in each case some type I enzymes are detected as well. Rat heart, on the other hand, contains both types of enzyme, with 75% of the activity being that of type I. A recent report by Costa et al. (1976) indicates that the expression of types I and 11cyclic AMP-dependent protein kinases in Chinese hamster ovary cells is cell-cycle specific. They showed that the type I kinase activity was high in mitosis and remained constant throughout the cell cycle. On the othdr hand, the activity of the type I1 enzyme seems to be related to the onset of DNA synthesis where a rapid increase in activity occurs at the GUS phase boundary. This increase in type I1 kinase activity represents an increase in d e n o w enzyme synthesis. It is of interest to note that the cellular cyclic AMP level in Chinese hamster ovary cell also fluctuates with the cell cycle. Cyclic AMP increases during the cell cycle coordinate temporally with increases in type I1 kinase activity. In a similar study employing Tetrahymena pyriformis, Majumder e t aE. (1975)also show that the level of cyclic AMP and the specific activity of cyclic AMP-dependent protein kinase I1 are maximal in organisms undergoing transition from G1 to S phase. These studies point to a functional difference between kinase I and I1 and suggest that the latter enzyme may play a role in the coordination of cell cycle by regulating DNA, RNA, and/or protein synthesis.
2. SUBUNIT STRUCTURE AND MODE OF ACTIVATION Studies of the interaction between cyclic AMP and cyclic AMPdependent protein kinase have elucidated the molecular mechanism by which the cyclic nucleotide activates the enzyme (Gill and Garren, 1970; Tao et al., 1970; Kumon et al., 1970; Brostrom et al., 1970; Tao, 1971a). Equation (2) shows the effect of cyclic AMP on the quaternary structure of cyclic AMP-dependent protein kinase: RC
+ cyclic AMP
R-cyclic AMP
+C
( 2)
The inactive enzyme (RC) is constructed of two dissimilar functional subunits: a regulatory (R) or inhibitory subunit that binds cyclic AMP and a catalytic (C) subunit. The binding of cyclic AMP to the regulatory subunit destabilizes the complex and releases the catalytic moiety for enzymatic activity. At present, there is no indication that the regulatory component also possesses enzymatic activity. The above mechanism has found general acceptance and appears to apply to all cyclic AMP-dependent protein kinases isolated from various sources. A similar mode of activation has also been proposed for the
24 1
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
cyclic GMP-dependent protein kinases isolated from lobster tail muscle (Miyamoto et al., 1973) and rat pancreas (Leemput-Coutrez et al., 1973). In contrast, the activation of the silkworm cyclic GMPdependent protein kinase by cyclic GMP does not lead to the dissociation of the enzyme into catalytic and regulatory subunits (Takai et al., 1976). Work with highly purified enzymes from bovine heart (Rosen and Erlichman, 1975; Hofmann et al., 1975) and rabbit skeletal muscle (Hofmann et a1 ., 1975) has shown that two catalytic and two regulatory subunits aggregate to form the inactive holoenzyme. In the presence of cyclic AMP, the holoenzyme dissociates into two monomeric catalytic subunits and one dimeric regulatory subunit that bind two molecules of cyclic AMP. The sedimentation coefficients and molecular weights of bovine heart and skeletal muscle cyclic AMP-dependent protein kinases and their corresponding subunits are shown in Table I. The physical properties of the catalytic components of the two enzymes are the same. On the other hand, the regulatory subunit of the skeletal muscle enzyme is slightly smaller than that of the heart enzyme (Hofmann et d., 1975). In addition, there are other distinguishable differences between these enzymes. The chromatographic property of the purified bovine heart muscle enzyme on DEAETABLE I THE SEDIMENTATION COEFFICIENTS AND MOLECULARWEIGHTSOF BOVINE CARDIAC AND RABBIT SKELETAL MUSCLE CYCLIC AMP-DEPENDENT PROTEINKINASES S*O,W
Bovine cardiac muscle Holoenzyme Catalytic subunit Regulatory subunit
7" 3.4'; 3.6b 4.3"; 4.6b
Rabbit skeletal muscle Holoenzyme Catalytic subunit Regulatory subunit
7" 3.4" 5"
Hofmann et al. (1975). Erlichman et al. (1973). ' Huang and Huang (1975). Corbin et al. (1972).
a
Molecular weight 174,000b 40,000"; 38,000b 55,000a
226,000'; 123,00Od 40,000' 48,000'
242
M. MARLENE HOSEY AND MARIANO TAO
cellulose resembles that of a type I1 enzyme, whereas rabbit skeletal muscle resembles that of a type I enzyme. The rabbit skeletal muscle protein kinase is dissociated into its subunits by cyclic AMP more readily than the bovine heart muscle kinase (Beavo et al., 1974). Erlichman et al. (1974) have reported that the bovine cardiac muscle cyclic AMP-dependent protein kinase can catalyze the autophosphorylation of its regulatory component, resulting in the incorporation of 2 moles of phosphate from ATP per mole of holoenzyme. The phosphoryl acceptor site in each of the regulatory monomers appears to be a seryl residue (Rosen and Erlichman, 1975). Phosphorylation of the heart kinase is reversible (Rosen and Erlichman, 1975), the reverse reaction having a sharp pH optimum at 5.5, whereas the forward reaction has a broad pH activity profile with a maximum between pH 6 and 8.5. Since autophosphorylation of the cardiac enzyme occurs readily, Rangel-Aldao and Rosen (1976) suggest that the phosphoenzyme predominates under physiological conditions. Both phosphoand dephosphoprotein kinase bind cyclic AMP with approximately the same affinity and dissociate rapidly and to the same extent in the presence of the cyclic nucleotide (Rangel-Aldao and Rosen, 1976). However, the dephospho-R reassociates at least five times faster than the phospho-R with the catalytic subunit to regenerate the cyclic AMP-dependent holoenzyme. Thus, phosphorylation of the regulatory subunit may prolong the action of the catalytic subunit following the removal of cyclic AMP. Autophosphorylation of the regulatory subunit of a bovine brain cyclic AMP-dependent protein kinase has also been reported (Maeno e t al., 1974). The kinase used in the study of Maeno e t al. (1974) corresponds to the peak I1 enzyme eluted from DEAE-cellulose column. Contrary to what is found in bovine heart and brain, autophosphorylation of the regulatory subunit of cyclic AMP-dependent protein kinase does not occur in rabbit skeletal muscle (Hofmann e t a1 ., 1975). However, the rabbit skeletal muscle, but not the bovine heart, enzyme binds Mg-ATP with high affinity (Haddox et al., 1972; Hofmann et a1 ., 1975).The binding of Mg-ATP to the skeletal muscle protein kinase increases the dissociation constant for cyclic AMP about ten fold. Furthermore, Mg-ATP is required for the reassociation of the isolated subunits of skeletal muscle protein kinase, but not of heart muscle protein kinase. The differences between the rabbit skeletal muscle and the bovine heart muscle protein kinases described above provide a further distinction between the type I and the type I1 cyclic AMP-dependent protein kinases, and may hold true for cyclic AMP-dependent protein
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
243
kinases found in other tissues with similar elution properties on DEAE-cellulose.
3. OTHERFACTORS INFLUENCING CYCLICAMP-DEPENDENT PROTEINKINASEACTIVITY Since the interaction between the regulatory and the catalytic subunits of cyclic AMP-dependent protein kinases involves weak chemical forces, dissociation of the subunits can also be achieved by chemical perturbations. Huang and Huang (1975) have reported that 0.2-0.3 M NaSCN or KSCN converts rabbit skeletal muscle protein kinase from the cyclic AMP-dependent to the independent form. At O'C, the addition of up to 4.5M urea to cyclic AMP-dependent protein kinase causes a similar conversion and activation. The enzyme treated with either thiocyanate or urea regains its cyclic AMP dependency when the dissociating agent is removed. Dissociation of cyclic AMP-dependent protein kinases can also be effected by their protein substrates through protein-protein interactions. The bovine brain cyclic AMP-dependent protein kinase has been reported to be dissociated and converted by a cyclic AMPindependent form in the presence of histone (Miyamoto e t al., 1971). A similar enzyme from rabbit erythrocytes is also dissociated in the presence of protamine as a result of the formation of a complex between protamine and the regulatory subunit (Tao, 1972). The ability of certain protein substrates to dissociate the enzyme may account, in part, for the basal activity observed in the absence of cyclic AMP and the variations in degree of cyclic AMP stimulation with different substrates. A heat-stable protein inhibitor specific for cyclic AMP-dependent protein kinases has been isolated from boiled muscle extracts by Walsh e t al. (1971). The inhibitor forms a complex with the catalytic subunit of the protein kinase. The complex is catalytically inactive and cannot recombine with the regulatory subunit (Ashby and Walsh, 1972). However, a somewhat more complicated role of the protein inhibitor has been suggested by Donnelly et al. (1973a,b). These workers show that a lobster tail heat-stable protein factor with properties similar to those of the inhibitor described by Walsh e t al. (1971) acts by altering the substrate specificity both of cyclic AMPdependent and cyclic GMP-dependent protein kinases, increasing the phosphorylation of certain protein substrates while decreasing that of others. Furthermore, using arginine-rich histone as substrate, the protein inhibitor stimulates the catalytic subunit of cyclic GMP-
244
M. MARLENE HOSEY AND MARIANO TAO
dependent protein kinase while it inhibits that of cyclic AMPdependent protein kinase. These workers suggest that the protein inhibitor may function as a modulator in regulating the substrate specificities of the cyclic nucleotide-dependent protein kinases. From the above studies, it would appear that protein kinases may be regulated by factors other than the cyclic nucleotides. Conceivably, a change in the ionic conditions of the cell could lead to a change in protein kinase activity. In the cell nucleus, both histone and protamine could regulate protein kinase activity by binding to the regulatory subunit. The possibility exists that certain hormones, such as the polypeptide variety, may also directly interact with the cyclic AMPdependent protein kinases leading to inhibition or activation of these enzymes. Finally, the protein kinase activity may be modulated by the level of protein inhibitor in the cell.
4. NONENZYMESUBSTRATES AND NATURE O F THE PHOSPHORYL ACCEPTORSITE
Cyclic AMP-dependent protein kinase exhibits a high degree of specificity for ATP and does not utilize GTP as the phosphoryl donor (Labrie et a1 ., 1971; Tao and Hackett, 1973). Inhibition studies indicate that UTP, CTP, and dTTP fail to compete with ATP in the kinase reaction, suggesting that these nucleotides are not substrates of the enzyme (Miyamoto et al., 1969; Tao, 1971b). However, it appears that dATP can also serve as substrate, but not as effectively as ATP (Miyamot0 et a,!., 1969). A number of proteins are known to be substrates of cyclic AMPdependent protein kinases (Langan, 1973; Rubin and Rosen, 1975). Certain of these proteins have known enzymatic functions that are altered by phosphorylation (see a recent review by Rubin and Rosen, 1975). However, many investigators have found it more expedient to employ protein substrates such as casein, protamines, or histones, even though the significance of phosphorylation of these proteins is unknown. Of these substrates, histones prove to be best and are widely used (Langan, 1973). Among the different histone fractions, f l , f2a, f2b, and f3 are phosphorylated by cyclic AMP-dependent protein kinases at varying rates (Langan, 1968; Chen and Walsh, 1971), with f2b generally the best. Kemp et ul. (1975), studying the substrate specificity of rabbit skeletal muscle cyclic AMP-dependent protein kinase with the aid of genetic variants of P-casein, have found that P-casein B is phosphorylated about 70 times faster than the most common variant, p-casein
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
245
A2. Other &casein variants such as A', A3, and C are phosphorylated at a rate comparable to that of A2. The enhanced phosphorylation of P-casein has been attributed to the presence of an arginine in the place of a serine residue in position 122. The need for an arginine residue in close proximity to the acceptor site has been confirmed by Daile e t ul. (1975), who have used various synthetic peptides equivalent to amino acid residues 106-113 of the basic protein of human myelin as substrates. Their studies and those of others (Kemp et al., 1975; Bylund and Krebs, 1975) show that the phosphoryl acceptor site contains the general sequence Arg-X-YSer-Z. The position of the basic amino acid in the active sequence is not firm and may extend from two to five residues from the amino terminal of serine (Kemp e t al., 1975). In addition to the primary sequence, the tertiary structure of the acceptor also determines substrate specificity. Thus, Bylund and Krebs (1975) show that native chicken egg white lysozyme must first be heat-denatured or Scarboxymethylated before it can be phosphorylated.
5. PHOSPHORYLATION AND REGULATIONO F PHOSPHORYLASE KINASE AND
GLYCOGEN SYNTHETASE
The role of cyclic AMP-dependent protein kinase in the regulation of glycogen metabolsim has been extensively investigated (Lamer and Villar-Palasi, 1971; Walsh and Krebs, 1973). Cyclic AMPdependent protein kinase catalyzes the phosphorylation of both phosphorylase kinase (Walsh e t al., 1968) and glycogen synthetase (Soderling et ul ., 1970). a. Phosphorylase Kinase. The kinetics of phosphorylase kinase phosphorylation have been examined in great detail by Hayakawa e t al. (1973)and by Cohen (1973).Phosphorylase kinase is constructed of three nonidentical subunits, a,p, and y , and has the structure a4P4y4 (Cohen, 1973). The y subunit is the catalytic component and is not phosphorylated. Both a and p subunits are phosphorylationted by cyclic AMP-dependent protein kinase, but at different rates. The phosphorylation of subunitp is rapid and precedes that of subunit a,which occurs at a slower rate. The phosphorylation of the /3 subunit closely parallels the increase in phosphorylase kinase activity. The significance of subunit a phosphorylation in relation to phosphorylase kinase activation is not clear. Hayakawa e t al. (1973), having shown that an increase in phosphorylase kinase activity occurred even after the reaction involving the /Isubunit was completed, suggested that subunit a phosphorylation might also be involved in the activation
246
M. MARLENE HOSEY AND MARIANO TAO
process. Cohen (1973),however, did not observe a similar increase in phosphorylase kinase activity. The reason for this discrepancy is not known. In vivo regulation of phosphorylase kinase activity has been studied by Yeaman and Cohen (1975).They show that the enzyme is activated five- to ten-fold following an intravenous injection of adrenalin into the marginal ear vein of a rabbit and that the tryptic phosphopeptides prepared from a rabbit muscle kinase phosphorylated in vitro are the same as the phosphopeptides obtained from the in vivo phosphorylated kinase. Hence, the activation of phosphorylase kinase by cyclic AMP-dependent protein kinase appears to constitute one of the events of hormonally stimulated glycogenolysis. b. Glycogen Synthetase. Glycogen synthetase exists in two interconvertible molecular forms, I and D (Friedman and Lamer, 1963). Synthetase I is the active form, whereas D, the phosphorylated form, exhibits a minimal level of activity. However, synthetase D becomes fully active in the presence of high concentrations of glucose-6phosphate. Cyclic AMP-dependent protein kinase catalyzes the conversion of glycogen synthetase I form to D form (Schlender et al., 1969; Soderling et al., 1970). Soderling (1975) reported that the complete conversion of skeletal muscle glycogen synthetase from the I to the D form required the incorporation of 8 moles of phosphate per mole of enzyme (400,000 daltons). However, a somewhat different result emerged from the analysis of alkali-labile phosphate contents of purified enzyme preparations. Roach et al. (1976) reported that purified glycogen synthetase D contained as much as 3.5 moles of phosphate per subunit (90,000 daltons). The reason for the discrepancy between the two studies is not known, but the latter estimate may represent a more physiologically significant description of the phosphorylation state of glycogen synthetase. Glycogen synthetase can be further phosphorylated by a slow, cyclic AMP-independent protein kinase-catalyzed reaction (Soderling, 1975). This phosphorylation has no effect on the synthetase activity. Nimmo and Cohen (1974, 1975) similarly reported the phosphorylation of rabbit skeletal muscle glycogen synthetase by a cyclic AMPindependent protein kinase. The significance of this phosphorylation reaction remains unknown. It does not appear to convert glycogen synthetase I to D. In contrast, Huang et al. (1975) and Schlender and Reimann (1975) independently showed that glycogen synthetase can be phosphorylated and inactivated by a cyclic AMP-independent protein kinase. The phosphorylation of glycogen synthetase by either cyclic AMPdependent or -independent protein kinase results only in partial inac-
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
247
tivation of enzyme activity. As shown by Huang et al. (1975), a more complete conversion of rabbit skeletal muscle synthetase I to D occurred when phosphorylation was carried out first by cyclic AMPindependent protein kinase and then by the cyclic AMP-dependent protein kinase, but not when the order was reversed. Incubation of synthetase with the cyclic AMP-independent protein kinase led to the incorporation of 1 mole of phosphate per mole of synthetase subunit. A second mole of phosphate was incorporated when further incubated with cyclic AMP-dependent protein kinase. Incubation with cyclic AMP-dependent protein kinase alone resulted in the incorporation of 2 moles of phosphate per mole of synthetase subunit; no further phosphorylation of glycogen synthetase was observed when cyclic AMPindependent protein kinase was subsequently added. Huang et al. (1975) have speculated that each glycogen s ynthetase subunit contains three phosphoryl acceptor sites, one specific for cyclic AMPindependent protein kinase, whereas the other two for cyclic AMPdependent protein kinase. Although three sites are available, only two are phosphorylated. The third is excluded once phosphates are incorporated into two of the acceptor sites. In contrast to the observation of Huang et al. (1975), Schlender and Reimann (1975) found that the glycogen synthetase of rabbit kidney medulla was inactivated to the same extent by the two kinases added either simultaneously or sequentially. The reason for this discrepancy is not known. Perhaps the cyclic AMP-independent protein kinases isolated by these laboratories are not the same.
AND REGULATION OF OTHERENZYMES 6. PHOSPHORYLATION
a. Hormone-Sensitive Lipase and Cholesterol Esterase. Rizack (1964) first reported that a hormone-sensitive lipase from adipose tissue homogenates was activated by ATP and cyclic AMP. The possibility that cyclic AMP-dependent protein kinase may be responsible for the activation of the hormone-sensitive lipase was examined by Corbin and Krebs (1969) and by Huttunen e t al. (1970). In a crude homogenate, the activation of lipase is blocked by the addition of the protein inhibitor of cyclic AMP-dependent protein kinase. The activation process is restored upon the addition of excess exogenous cyclic AMP-dependent protein kinase (Corbin and Krebs, 1969; Corbin et a1 ., 1970). A more direct demonstration of the requirement for ATP, cyclic AMP, and cyclic AMP-dependent protein kinase in the activation of lipase is provided by studies employing partially purified lipase preparations (Huttunen et d., 1970; Huttunen and Steinberg,
248
M. MARLENE HOSEY AND MARIANO TAO
1971; Steinberg, 1973). Using a 100-fold-purified lipase preparation, good correlation between lipase activation and protein phosphorylation by cyclic AMP-dependent protein kinase is observed. In a study using a similar approach, Khoo e t al. (1976) reported that rat and chicken adipose tissue cholesterol esterases are also activated by cyclic AMP-dependent protein kinase. They further showed that the cholesterol esterase activities in these tissues are activated to the same extent as the hormone-sensitive lipase. The two enzyme activities are not separated during limited purificaiton, and they exhibit the same pattern of reversible deactivation, pH activity profile, and responses to prior treatment of the tissue with glucagon. Based on these data, Khoo et al. (1976) suggested that the two hydrolases might represent different activities of a single enzyme protein. b. Phosphorylase Phosphatase. A partially purified phosphorylase phosphatase from rabbit skeletal muscle has been found to be inactivated by cyclic AMP-dependent protein kinase (Huang and Glinsmann, 1975). The kinase catalyzes the phosphorylation of an inhibitor protein associated with the phosphatase. The phosphorylated phosphatase can be reactivated by the addition of Mn2+or trypsin or by the dissociation of the inhibitor protein from the catalytic component. The activation by Mn2+ suggests that this enzyme and a relatively nonspecific phosphoprotein phosphatase isolated earlier may be the same (Katoand Bishop, 1972; Zieve and Glinsmann, 1973).The existence of an inhibitor-enzyme complex in phosphorylase phosphatase has been reported by Brandt et al. (1975). c. Pyruvate Kinase. Cyclic AMP-dependent protein kinase also catalyzes the phosphorylation of pig liver type L pyruvate kinase (Engstrom et al., 1974). The phosphorylation of the enzyme resulted in an inhibition of its activity when measured at low phosphoenolpyruvate concentrations. At higher concentrations of phosphoenolpyruvate, the inhibition becomes less pronounced, and at 1 mM phosphoenolpyruvate, the enzyme is fully active. The inhibition of pyruvate kinase activity may play a key role in the stimulatory effect of cyclic AMP on gluconeogenesis (Exton and Park, 1968). d. Carbonic Anhydrase. Two isoenzymes of carbonic anhydrase have been isolated from bovine erythrocytes (Narumi and Miyamoto, 1974). One of these has been purified to homogeneity and is activated by cyclic AMP-dependent protein kinases from hog muscle or bovine brain. The cyclic AMP-dependent protein kinases catalyze the phosphorylation of both carbonic anhydrase isoenzyme preparations. However, the phosphorylation of the less pure carbonic anhydrase isoenzyme did not result in enzyme activation. Possibly in this prepa-
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
249
ration, the contaminating protein rather than the carbonic anhydrase isoenzyme was the phosphoryl acceptor. Narumi and Miyamoto (1974)suggest that the stimulation of gastric juice secretion by cyclic AMP may be due to a similar activation of .carbonic anhydrase by cyclic AMP-dependent protein kinase in gastric mucosa. e. Phenylalanine Hydroxylase. Phenylalanine hydroxylase isolated from rat liver is a phosphoprotein and contains about 0.3 mole of protein-bound phosphate per mole of subunit (50,000daltons). When the enzyme is incubated in the presence of ATP and cyclic AMPdependent protein kinase, additional phosphate is incorporated, bringing the total phosphate content of the enzyme up to about 1 mole per mole of subunit (Milstein et al., 1976).The phosphorylation of the enzyme in vitro is accompanied by a 2.6-fold increase in hydroxylase activity. Barrengeret al. (1972)have separated rat liver phenylalanine hydroxylase into three different isoenzymes by chromatography on calcium phosphate columns, It is possible that these isoenzymes actually represent a single enzyme but contain different amounts of bound phosphate. Since phenylalanine hydroxylase is a dimer of two slightly different 50,000-dalton subunits, partial phosphorylation could give rise to a heterogeneous population of enzyme molecules separable by calcium phosphate chromatography. f. Troponin. Troponin, a protein that regulates the Ca2+sensitivity of actomyosin ATPase, can be phosphorylated by either phosphorylase kinase (Stull e t al., 1972; Cole and Perry, 1975) or cyclic AMPdependent protein kinase (Reddy e t a1., 1973; England et a1., 1974). Phosphorylation of rabbit skeletal muscle troponin by either kinase resulted in the incorporation of phosphate into both TN-I and TN-T subunits of troponin. The kinases also catalyzed the phosphorylation of isolated TN-I subunit, but the phosphorylation of the isolated TN-T subunit was observed only with phosphorylase kinase. The phosphorylated troponin show a marked stimulation of ATPase activity in the presence of Ca2+.The significance of troponin phosphorylation is further substantiated by the finding that purified troponin preparations of rabbit skeletal (England et a1 ., 1974) and cardiac (Cole and Perry, 1975) muscles contain protein-bound phosphates. g. RNA Polymerase. In addition to the above studies where several enzyme activities have been found to be altered by cyclic AMPdependent protein phosphorylation, the possibility that other functional proteins may be similarly altered has also been implicated in a number of less defined systems. Martelo et al. (1970) showed that both rabbit skeletal muscle and reticulocyte cyclic AMP-dependent protein kinases catalyzed the phosphorylation of Escherichia coli
250
M. MARLENE HOSEY A N D M A R I A N O TAO
RNA polymerase. The increase in polymerase activity by protein kinase was correlated with the phosphorylation of the (T component of the RNA polymerase. The significance of this finding remains unknown since a similar study employing homologous kinase has not been carried out. Jungmann et al. (1974) have reported the activation of calf ovarian nuclear RNA polymerase by cyclic AMP-dependent protein kinase. The activity of the RNA polymerase I1 was increased ninefold after phosphorylation. A 2.5-threefold stimulation of RNA synthesis by the phosphorylated polymerases Ia and Ib was also observed. Since these RNA polymerase preparations are not homogenous, it is not known whether the activity changes are associated with phosphate incorporated into the enzymes. Nevertheless, RNA synthesis in ovarian nuclei appears to be regulated by protein phosphorylation. h. Ribosomes. Incubation of rabbit reticulocytes with cyclic AMP or N6,02-dibutyrylcyclic AMP caused an increase in the incorporation of phosphate into a ribosomal protein of MW 27,500 (Krystosek et a1 ., 1974; Cawthon et al., 1974) located on the smaller subunits. On the other hand, a considerably greater number of phosphoproteins resulting from incubation of ribosomes with cyclic AMP-dependent protein kinase has been reported by Traut’s laboratory (Traugh et al., 1973; Traut et al., 1974).A protein in the 40 S subunit and six different proteins in the 60 S subunit have been identified as substrates for the kinase. In rat liver, at least four proteins of the 40 S subunit and ten proteins of the 60 S subunit are phosphorylated in vitro by the cyclic AMP-dependent protein kinase (Eil and Wool, 1973a). Cyclic AMPdependent phosphorylation of corpus luteum ribosomes and ribosomal subunits has been investigated by Azhar and Menon (1975a). At least nine proteins of 80 S ribosomes and 12 proteins of the 60 S subunits are found to be phosphorylated. However, only one major and four minor bands are phosphorylated in the case of 40 S ribosomal subunits. The phosphorylation of ribosomal proteins appears to have little, if any, effect on the function of ribosomes in protein synthesis (Eil and Wool, 1973b; Krystosek et al., 1974). i . Tyrosine Hydroxylase. Morgenroth et al, (1975) have initially reported that cyclic AMP-dependent protein kinase regulates the activity of rat brain tyrosine hydroxylase. Both cyclic AMP and cyclic AMP-dependent protein kinase cause tyrosine hydroxylase activity to increase severalfold in a high-speed supernatant preparation obtained from rat brain. In addition to cyclic AMP and protein kinase, the activation of tyrosine hydroxylase required the presence of ATP and M$+. Lloyd and Kaufman (1975), on the other hand, showed that tyro-
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
25 1
sine hydroxylase activation was not due to the incorporation of phosphate into the enzyme. Rather, the activation may be due to the formation of a phosphoprotein which then interacts with tyrosine hydroxylase as a polyanion, in a manner similar to that observed with phosphatidyl-L-serine. The phospholipid has previously been shown to stimulate the activity of a highly purified bovine caudate tyrosine hydroxylase (Lloyd and Kaufman, 1974). C. Cyclic AMP-Independent Protein Kinases
1. INTRODUCTION Although the occurrence of cyclic AMP-independent protein kinases is also widespread, these enzymes, in general, have not received as much attention as the cyclic AMP-dependent kinases. The study of these kinases is somewhat complicated by the fact that the cyclic AMP-dependent protein kinases can also exist in a form that is independent of the cyclic nucleotide. To what extent the number of protein kinases described prior to the discovery of cyclic AMPdependent protein kinase (Walsh e t al., 1968) are truly independent of cyclic AMP is difficult to ascertain. Three basic criteria can generally be applied to the study of the cyclic AMP-independent protein kinases in order to distinguish them from the cyclic AMP-dependent enzymes. (i) The enzyme must be neither activated by cyclic AMP nor inhibited by the regulatory subunit of the cyclic AMP-dependent protein kinase. (ii) The enzyme activity must not be inhibited or altered by the protein inhibitor of the cyclic AMP-dependent protein kinase. (iii) The substrate specificity of the enzyme must be sufficiently different from that of the cyclic AMP-dependent protein kinase. The last criterion is difficult to define since the cyclic AMPindependent and the cyclic AMP-dependent protein kinases are known to phosphorylate some common substrates. However, certain information can be gained by examining the relative activity of an enzyme toward various protein substrates. In general, the cyclic AMPdependent protein kinases isolated from various tissues all exhibit a greater activity towards histones, whereas most of the cyclic AMPindependent protein kinases favor casein and phosvitin. The study of differences in protein substrate specificities can be further extended to include those enzymes known to be activated or inactivated by phosphorylation. Although these biological substrates would enable
252
M. MARLENE HOSEY AND MARIANO TAO
us to better distinguish the cyclic AMP-dependent protein kinase from the cyclic AMP-independent protein kinase, equivocal results may also arise. It is also possible that both types of protein kinases are capable of catalyzing the phosphorylation of the same enzyme substrate. An example of this situation is found in the study of glycogen synthetase as discussed earlier. In addition to the protein substrates, the two types of kinases may also exhibit differences in specificities with respect to the phosphoryl donor. Most of the cyclic AMP-independent protein kinases are known to utilize both ATP and GTP as phosphoryl donors, whereas the cyclic AMP-dependent protein kinases utilize only ATP (Kumar and Tao, 1975). However, the ability to employ both ATP and GTP as phosphoryl donors is probably not a general property of all the cyclic AMP-independent enzymes. For example, the T7 bacteriophageinduced protein kinase does not utilize GTP (Pai et al., 197513). In this section, we shall describe a heterogeneous population of protein kinases that have properties sufficiently different from those of the cyclic AMP-dependent protein kinases. However, we must emphasize that the grouping of the kinases under cyclic AMPindependent enzymes is not absolute. It is possible that further studies may reveal that certain of these enzymes are actually related to the cyclic AMP-dependent protein kinases.
2. DISTRIBUTIONAND GENERAL PROPERTIES Cyclic AMP-independent protein kinases have been isolated from a wide variety of sources. Among these enzymes, some have broader specificities than others. One of the best characterized enzymes is phosphorylase b kinase. Phosphorylase b kinase was originally thought to be a specific enzyme that catalyzes the phosphorylation and activation of phosphorylase b. However, subsequent studies indicate that it may have a broader specificity and can also catalyze the phosphorylation of casein (DeLange et al., 1968), troponin (Stull et al., 1972; Cole and Perry, 1975), and a 95,000-dalton sarcoplasmic reticulum protein (Schwartz et al., 1976). The role of phosphorylase b kinase in the regulation of glycogen degradation and some of its properties have been described earlier. Bingham and Farrel (1974) have isolated a protein kinase from the Golgi apparatus of the lactating mammary gland. This enzyme seems to have a high preference for dephosphorylated casein although it also catalyzes, to a limited extent, the phosphorylation of P-lactoglobulin, a-lactoalbumin, and fat globule membrane proteins. Calf brain con-
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
253
tains at least three different fractions of phosvitin kinase (Walinder, 1973). All three enzymes catalyze the phosphorylation of phosvitin and casein, but not of histones. Similar enzymes have been isolated from ox brain (Rodnight and Lavin, 1964)and from rooster liver (Goldstein and Hasty, 1973). The latter enzyme has been purified 8000-fold. The kinase has a sedimentation coefficient of about 7.6 S and a MW of about 160,000. The ox brain and rooster liver kinases can utilize GTP as well as ATP as a phosphoryl donor. In addition to the cyclic AMP-dependent protein kinases, rabbit red blood cells contain at least two other kinases (Traugh and Traut, 1974; Kumar and Tao, 1975). These kinases catalyze the phosphorylation of casein and phosvitin, utilizing either ATP or GTP as the phosphoryl donor. None of these kinase activities is affected by cyclic nucleotides. Based on sucrose density gradient centrifugation, MWs of 9.5 X 105 and 1.4 x 106 have been estimated for the two kinases (Kumar and Tao, 1975). In the presence of NaSCN, these kinases are converted into smaller molecular weight species that are fully active; this suggests the possible existence of subunit structures (Kumar and Tao, unpublished observation). Interestingly, both kinase activities are inhibited by 2,3-diphosphoglyceric acid, a metabolite that is present in high concentration in red blood cells. Mu1tiple forms of cyclic AMP-independent protein kinases have also been demonstrated in rabbit renal cortex (Reimann and Schlender, 1976). Based on ion exchange chromatography, two major fractions are obtained. These kinases are neither activated by cyclic AMP nor inhibited by the protein inhibitor of cyclic AMPdependent protein kinase. Interestingly, glycogen synthetase can serve as substrate for these kinases, suggesting that the synthetase activity may also be regulated by processes unrelated to cyclic AMP. Dogfish skeletal muscle contains a protein kinase that catalyzes the phosphorylation of an 18,000-dalton protein that copurifies with parvalbumin (Demaille et al., 1975). The dogfish kinase also phosphorylates protamine and, to a limited extent, phosvitin and histone, but not phosphorylase kinase, phosphorylase b, or casein. The properties of the dogfish kinase are different from those of phosphorylase kinase and cyclic AMP-dependent protein kinase (Blum et aZ., 1974). Recently, a large number of enveloped animal viruses have been found to possess a cyclic AMP-independent protein kinase activity in their virions (Strand and August, 1971; Randall et aZ., 1972; Downeret al., 1973; Tan, 1975). The enzyme catalyzes the phosphorylation of endogenous viral proteins and is stimulated by dithiothreitol and protamine (Strand and August, 1971; Downer et d.,1973). Kleiman and
254
M. MARLENE HOSEY AND MARIANO TAO
Moss (1975a, 1975b) have purified a protein kinase and two phosphoryl acceptors from vaccinia virions. The protein kinase has a MW of 62,000 and the phosphoryl acceptors, 38,500 and 11,700. The phosphorylation of the vaccinia viral associated proteins is dependent on Mg-ATP and protamine or other basic proteins. Protamine appears to function as an enzyme activator and is not appreciably phosphorylated. The role of protein kinase in viral infection, replication, and assembly remains unknown. However, it is interesting to point out that the structural proteins of a number of animal viruses are phosphoproteins (Tan and Sokol, 1972; Sokol and Clark, 1973; Pal and RoyBurman, 1975; Krystal et al., 1975). Although protein phosphorylation does not appear to occur in E . coli, a protein kinase is induced following bacteriophage T7 infection (Rahmsdorf et al ., 1974). The T7 bacteriophage gene-encoded protein kinase has been purified 5000-fold (Pai et al., 1975a). The best substrates for the kinase are lysozyme and histone, in that order (Pai et al., 197513). In addition, the kinase also catalyzes the phosphorylation of the p’ subunit of E . coli DNA-dependent RNA polymerase. This phosphorylation reaction is inhibited at high ionic strength and by cyclic AMP. Phosphorylation of host RNA polymerase seems to be required for the proper termination of transcription at the end of the early T7 region. In the absence of a functional kinase gene, E . coli polymerase does not recognize the termination signal (Zillig et al., 1975). D. Membrane-Bound Protein Kinarer
1. INTRODUCTION
Many cell membrane preparations contain an autophosphorylation system (Lemaire et al., 1971; Majumder and Turkington, 1972; Guthrow et al., 1972; Hosey and Tao, 1976a). Both protein kinase and protein substrates are found to be intrinsic components of the cell membranes. The protein kinase activities associated with the membranes have been measured by determining the total amount of phosphate incorporated into membrane proteins (autophosphorylation) or added exogenous protein substrates. In the autophosphorylation reaction, the number and the nature of the phosphopeptides have been analyzed by sodium dodecyl s ul fate-polyacrylamide gel electrophoresis followed either by autoradiography or by determination of the radioactivity in each gel slice. By examining the phosphorylation patterns of membrane polypeptides obtained under various conditions, it has
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
255
been possible to gain insight into the nature and multiplicity of the membrane-bound protein kinases and to distinguish between cyclic AMP-dependent and -independent processes. I n some systems, the membrane-bound protein kinases have been solubilized and partially characterized.
2. CYCLICNUCLEOTIDE-DEPENDENT PROTEINKINASES Studies of membrane autophosphorylation in bovine anterior pituitary (Lemay et al., 1974), human erythrocytes (Guthrow et al., 1972; Rubin and Rosen, 1973; Avruch and Fairbanks, 1974; Fairbanks and Avruch, 1974; Hosey and Tao, 1976a), and human platelets (Steiner, 1975) have shown that cyclic AMP selectively stimulates the phosphorylation of certain membrane components but not others. The resul ts indicate that cyclic AMP-dependent and -independent protein kinases and their respective substrates are present in these systems. In contrast, the endogenous phosphorylation of chick brain microtubule (Sloboda et al., 1975), rat brain synaptic (Ueda et al., 1973), and rat adipocyte (Chang et al., 1974) membrane components is observed only in the presence of cyclic AMP. I n the absence of the cyclic nucleotide, no specific protein is selectively phosphorylated. A similar observation indicates that only cyclic nucleotide-dependent enzymes are associated with the plasma membranes of smooth muscle. Casnellie and Greengard (1974) showed that cyclic GMP stimulated the endogenous phosphorylation of two proteins in isolated membrane fractions from mammalian organs rich in smooth muscle, including ductus deferens, uterus, and small intestine. The phosphorylation of a third protein found in these preparations was stimulated by cyclic AMP. Although cyclic GMP-dependent protein kinase has been isolated from a number of sources (Kuo, 1975; Miyamoto et al., 1973; Nakazawa and Sano, 1975), this is the first report demonstrating the specific phosphorylation of membrane proteins by the kinase. Protein kinase activity fully responsive to cyclic AMP has been solubil ized from membrane preparations of bovine anterior pituitary (Lemay et al., 1974), bovine corpus lutem (Azhar and Menon, 1975b), and human erythrocyte (Rubin, 1975) membranes. Treatment of anterior pituitary membranes with either 1.0 M NH&l or 0.5% Triton X-100 solubilizes about 80% of the membrane-bound kinase activity. The soluble enzyme catalyzes the phosphorylation of histones and is activated by cyclic AMP. Azhar and Menon (197513) have extensively investigated the properties of the solubil ized kinases from two plasma membrane fractions,
256
M. MARLENE HOSEY AND MARIANO TAO
F I and FII, of corpus luteum. These kinases are strongly associated with the cell membranes, Solubilization of these enzymes is achieved with either ionic or nonionic detergents, but not with monovalent salts. The solubilized F I protein kinase and cyclic AMP-binding activities cosediment as a 6.3s component in sucrose density gradient. Cyclic AMP dissociated the enzyme into a 4.8s catalytic subunit and two cyclic AMP-binding subunits with sedimentation coefficients of 8.1s and 6.7s. The solubilized FII protein kinase appears to be less homogeneous and sediments as two components, 7.7s and 5.5s. The cyclic AMP-binding activity also sediments as two components, 6.7s and 5.5s. However, in the presence of cyclic AMP, a 4.8s catalytic subunit and a 6.3s cyclic AMP-binding subunit were obtained. The differences in the molecular sizes of the catalytic and the binding subunits, as observed under different experimental conditions may be due to aggregation, disaggregation, or proteolysis. A cyclic AMP-dependent protein kinase associated with human erythrocyte membranes has been solubilized with 0.5%Triton X-100 in 56 mM sodium borate, pH 8 (Rubin, 1975). About 90% of the enzyme activity was released in soluble form. The solubilized protein kinase was stimulated by cyclic AMP and retained most of its cyclic AMP-binding activity. These observations substantiate previous studies of Shimomura et al. (1974) on partially purified kinase from erythrocyte membranes. They showed that the kinetic and catalytic properties of the enzyme extracted from human red cell membranes closely resembled those of the soluble cyclic AMP-dependent protein kinases obtained from various sources such as rat liver, rat brain, and rabbit skeletal muscle. From the above studies, it would appear that cyclic AMPdependent protein kinases are true components of some membrane systems. The kinases are tightly bound to the membranes, and their solubilization requires methods that are generalIy employed for the dissociation of membrane-bound proteins. Hence, it seems unlikely that these enzymes represent soluble kinases which have become attached to the cell membranes during fractionation. The relationship of membrane-bound cyclic AMP-dependent protein kinases to those found in the soluble fraction remains to be determined. Now that certain membrane-bound kinases have been solubilized, information regarding this relationship, if any, should be forthcoming.
3. CYCLICAMP-INDEPENDENT PROTEINKINASES In contrast to the membrane systems described above, cyclic nucleotides are essentially without effect on the protein kinase activities
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
257
associated with guinea pig fat cell membranes (Chang et al., 1974), certain rat skeletal muscle membrane fractions (Andrew et a1 ., 1975), rabbit erythrocyte membranes (Hosey and Tao, 1976a), and rod outer segments prepared from frog or ox retinas (Welleret al., 1975a; Miller and Paulsen, 1975). The rabbit erythrocyte membranes contain at least two independent phosphorylation reactions that can be differentiated based on phosphoryl donor requirements. One reaction specifically requires ATP as phosphoryl donor, while the other is less specific and requires either ATP or GTP (Hosey and Tao, 1976a). The phosphorylation pattern of membrane proteins in the presence of ATP is clearly different from that of GTP (Fig. 3). Since GTP is not a substrate for the cyclic AMP-dependent protein kinases of rabbit erythrocytes (Tao and Hackett, 1973),this rules out the possibility that the membrane-bound GTP : protein kinase may be derived from the cyclic AMP-dependent enzymes. The nature of the erythrocyte membrane-bound ATP :protein kinase remains unknown. As pointed out earlier, not all the proteins of human erythrocyte membranes that are phosphorylated are responsive to cyclic AMP.
FIG.3. Autoradiograph depicting the endogenous phosphorylation of rabbit erythrocyte membrane polypeptides using either [y-3zP]ATPor [y-SzP]GTPas the phosphoryl donor. Each pair represents the phosphopeptides formed in the presence (right) or absence (left) of NaF.
25 8
M. MARLENE HOSEY AND MARIANO TAO
Avruch and Fairbanks (1974) suggested that there might be more than one kinase associated with these membranes. They have proposed that a 215,000-dalton polypeptide of the spectrin complex is phosphorylated by a specific monovalent and divalent cation-activated protein kinase. Although the enzyme appears to be quite specific in the phosphorylation of red cell membrane protein, its substrate specificity is not absolute since it also catalyzes the phosphorylation of casein. A protein kinase has been extracted from bovine (Weller e t al., 1975a; Frank and Buzney, 1975) and frog (Miller and Paulsen, 1975) rod outer segments. The enzyme specifically catalyzes the phosphorylation of photo-bleached rhodopsin. Interestingly, the enzyme prepared from bovine rod outer segments (ROS) seems to prefer GTP as the phosphoryl donor (Chader et al., 1975). The rate of ROS phosphorylation was greater in the presence of GTP than of ATP. Frog rhodopsin kinase also utilizes GTP, but to a lesser extent than ATP (Miller and Paulsen, 1975). Thus, it seems fairly certain that cyclic AMP-independent protein kinases are present in erythrocyte ghosts and rod outer segments. Whether similar enzymes are also found in other membrane preparations remains to be determined. The inability to demonstrate an effect of cyclic AMP on membrane phosphorylation does not necessarily indicate that the membrane-bound kinase is distinct from the cyclic AMP-dependent enzyme. As pointed out earlier, cyclic AMPdependent protein kinase can exist as an independent form by losing its regulatory component. Accordingly, the tightly bound cyclic AMP-independent protein kinase found in purified mouse mammary gland cell membranes may be similar to the cytosolic cyclic AMPdependent protein kinase. They have the same substrate specificity (Majumder and Turkington, 1972). 111.
MEMBRANE PHOSPHORYLATlO N
A. Introduction
This section deals with the specifics of phosphorylation of several different types of membranes. As mentioned earlier, only phosphorylation of seryl and of threonyl residues will be discussed. Phosphorylation studies have been carried out on membranes derived from various types of subcellular fractions from many types of normal and
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
259
abnormal cells. Our purpose is not to make this an all-inclusive “listing” but rather to concentrate on a few systems in which membrane phosphorylation has been extensively examined. We will be concerned with the characterization of each system with respect to cyclic AMP dependency, regulation by ionic conditions, phosphoryl donor specificity, identification of membrane substrates, and if known, the functional imp1ications of the specific phosphorylation. From the information provided, we hope that the reader can grasp a working knowledge of the relationship between protein phosphorylation and membrane activities as it stands today. 6. Erythrocyte Membrane Phoophorylation
1. INTRODUCTION Perhaps the most extensively studied membrane is that of the mammalian erythrocytes. Unlike the membranes of more complex tissues, those of mature mammaIian erythrocytes can be isolated in a comparatively pure form by relatively simple techniques, making them a popular model for study of membrane phenomena. Furthermore, the polypeptide content of human erythrocyte membranes has been characterized (Fairbanks et al., 1971) with the aid of sodium dodecyl sulfate-pol yacrylamide geI electrophoresis. Based on Coomassie Blue staining profiles, seven major polypeptides have been identified. They range in MW from 240,000 to 29,000. These polypeptides are consecutively numbered with arabic numerals from the highest to lowest molecular weight (Fig. 4). In addition, many minor components have also been detected and are designated with decimals of the number of the nearest major protein band of higher molecular weight (Steck, 1972). The human erythrocyte membrane also contains four identifiable sialoglycoproteins that stain positive with periodic acidSchiff’s reagent and are designated as PAS 1-PAS 4. The organization of erythrocyte membrane proteins has been reviewed by Steck (1974). Both the human (Fairbanks and Avruch, 1974; Avruch and Fairbanks, 1974; Rubin, 1975; Hosey and Tao, 1976a) and rabbit (Hosey and Tao, 1976a) erythrocyte membranes possess multiple protein kinase activities. These enzymes can be distinguished by their response to various agents, such as cations and cyclic AMP, and b y their phosphoryl donor and acceptor specificities.
260
M. MARLENE HOSEY AND MARIANO TAO
FIG. 4. Resolution of the polypeptides of rabbit (R) and human (H) erythrocyte membranes by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Data from Hosey and Tao (197%).
2. CYCLIC AMP-DEPENDENT PHOSPHORYLATION O F ERYTHROCYTE MEMBRANEPROTEINS One of the membrane-bound kinases of the human erythrocyte is cyclic AMP-dependent (Rubin et al., 1972; Guthrow et al., 1972; Roses and Appel, 1973; Rubin and Rosen, 1973; Fairbanks and Avruch, 1974; Shimomura et al., 1974; Hosey and Tao, 1976a). In fact, Rubin et at. (1972) show that over 70% of the total cyclic AMPdependent protein kinase of the human erythrocyte is membranebound. The distribution is in contrast to that in rabbit erythrocytes where all the cyclic AMP-dependent protein kinase activities are found in the cytoplasmic fraction (Tao and Hackett, 1973; Hosey and Tao, 1976a). Both the catalytic and the cyclic AMP-binding components of the cyclic AMP-dependent protein kinase appear to be localized on the inner surface of the plasma membrane (Kant and Steck, 1973; Rubin et al., 1973). This topographic orientation of the kinase may contribute to its accessibility to membrane substrates in the intact erythrocytes. The pH optimum of the human erythrocyte membrane cyclic
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
261
AMP-dependent protein kinase was initially reported by Guthrow et al. (1972) to be 6.5. However, our own studies indicate that the substrates of this enzyme are phosphorylated equally well at basic pH. The autophosphorylation patterns of human erythrocyte membranes in the presence and absence of cyclic AMP is shown in Fig. 5. The principal substrates of the cyclic AMP-dependent membrane phosphotransferase are two minor polypeptides: one with an apparent MW of 200,000 (polypeptide 2.1) and another with an apparent MW of 50,000-52,000. This latter phosphopeptide has been variously designated as IVc (Rubin and Rosen, 1973; Rubin, 1975), 4.5 (Fairbanks and Avruch, 1974), 32P-A(Guthrow et al., 1972), and 4.8 (Hosey and Tao, 1976a),but will be referred to as 4.8 hereafter. In addition to 2.1 and 4.8,the phosphorylation of several other minor components is also stimulated by cyclic AMP except to a lesser extent. These include polypeptides 2.2, 2.3, 2.4, and 2.9 as reported by Fairbanks and
FIG.5. Autoradiographofthe phosphopeptides of human (H) and rabbit (R) erythrocyte membranes. Autophosphorylation of erythrocyte membranes was carried out in the presence of [y-3ZPlATPwith and without cyclic AMP. Data from Hosey and Tao (1976a).
262
M. MARLENE HOSEY AND MARIANO TAO
Avruch (1974) and 2.2, 2.3,4.1, and 4.5 (MW 58,000) as demonstrated by Hosey and Tao (1976a). Rubin and Rosen (1973) have also reported some stimulation of the phosphorylation in the area of component 3 by cyclic AMP. These variations are probably due to differences in phosphorylation and electrophoretic conditions. Since autoradiography gives better resolution than counting of gel slices, the use of autoradiography may account for the greater number of phosphoproteins reported in recent studies. The phosphorylation of band 4.5 and of band 4.8 is interesting, since there is indication that these proteins may represent the regulatory subunits of the cyclic AMP-dependent protein kinases. Haley (1975) has photolabeled two human erythrocyte membrane proteins in the presence of [32P]8-azidoadenosine3’,5’-monophosphate. One of these proteins exhibits a MW of 55,000 and the other, 49,000. The 55,000-dalton protein is photolabeled at nanomolar concentration of the cyclic AMP derivative. In contrast, the incorporation of label into the lower molecular weight protein occurs only after the concentration of the cyclic nucleotide is raised to 0.25 pM. In similar studies, Guthrow et al. (1973) and Rubin (1975) show that [3H]Ns-(ethyl 2diazomalonyl>cyclic AMP can be incorporated into a protein component with an apparent MW of 50,000. These results, together with the phosphorylation data, suggest that the membrane-bound cyclic AMP-dependent protein kinase may catalyze the autophosphorylation of its regulatory subunit in a reaction analogous to that reported by Rosen and Erlichman (1975)and Hofmann et al. (1975) for the soluble enzymes. In spite of these coincidences, there is evidence that the photolabeled peptide( s) and the phosphopeptide( s) probably represent distinct proteins. In rabbit erythrocyte membranes, the phosphorylation of components 4.5 and 4.8 is not stimulated by cyclic AMP unless exogenous cyclic AMP-dependent protein kinase is added (Hosey and Tao, 1977). Consistent with this observation, cyclic AMP binding activity is not detected in rabbit erythrocyte membranes. Recently, Rubin (1975) has separated the cyclic AMP binding moiety and kinase activity from the cyclic AMP-stimulated phosphoprotein by extraction with Triton X-100. Thus, the phosphorylated membrane proteins, 4.5 and 4.8, are probably not the regulatory subunits of the membrane-bound cyclic AMP-dependent protein kinase. 3. CYCLIC AMP-INDEPENDENT PHOSPHORYLATION OF ERYTHROCYTE MEMBRANEPROTEINS
At least two cyclic AMP-independent protein kinase activities are present in rabbit erythrocyte membranes. This conclusion is reached,
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
263
in part, by examining the differences in phosphorylation patterns of erythrocyte membranes using two different phosphoryl donors. AS shown in Fig. 3, a lesser number of rabbit erythrocyte membrane proteins are phosphorylated in the presence of GTP than of ATP. Polypeptides in the area of 2-2.1 and 2.9-3 are phosphorylated both by ATP and by GTP, whereas polypeptides 4.1, 5.5, and several other minor components are phosphorylated only by ATP. While the reaction in the presence of GTP exhibits a pH optimum at 8.5, that in the presence of ATP gives a complex pH activity profile. However, in both instances, different pH’s favor the phosphorylation of different polypeptides. For example, polypeptides 4.1 and 5.5 are phosphorylated b y ATP to a greater extent at slightly acidic pH, while polypeptide 3 is phosphorylated to a greater extent at basic pH (Fig. 6). The difference in the phosphorylation pattern exhibited b y ATP and GTP suggests that rabbit erythrocyte membranes contain two protein kinases that differ in phosphoryl donor and in phosphoryl acceptor specificities. The labeling of polypeptides 2-2.1 and 2.9-3 both b y ATP and by GTP may be explained by assuming either that ATP is also a substrate of the GTP-utilizing enzyme or that these polypeptides can serve as phosphoryl acceptor for both kinases. The data presently available do not permit us to make a distinction between these two possibilities. The effect of pH on the phosphorylation of membrane polypeptides by ATP and by GTP may also be amenable to several interpretations. It is possible that pH may directly alter the structure of membrane proteins in such a way as to modify their acceptor capacity. Altematively, the results may be also explained by the presence of several membrane kinases and that the change in the degree of phosphorylation of certain protein(s) with pH simply reflects the activity of one or more of these kinases. The phosphorylation pattern of human erythrocyte membranes in the presence of GTP is similar to that of rabbit erythrocyte membranes. However, a slight difference in phosphorylation patterns occurs when ATP is used as substrate. As shown in Fig. 5, polypeptide 4.1, which is a major phosphopeptide in rabbit erythrocyte membranes, is not appreciably labeled in the human erythrocyte membranes. Furthermore, there is also considerably less labeling of bands 2.5 and 5.5 in the human than in the rabbit erythrocyte membranes. The phosphorylation of human erythrocyte membrane proteins is also affected by pH. The phosphorylation of polypeptides 2.9-3, 4.1, 4.5, and 4.8 by ATP is significantly enhanced at pH 8.5. Furthermore, the studies of Avruch and Fairbanks (1974) show that mono- and divalent cations preferentially stimulate the phosphorylation of polypep-
264
M. MARLENE HOSEY AND MARIANO TAO
(-1
RELATIVE
MOBILITY
(+I
FIG.6. Effect of pH on the autophosphorylation of rabbit erythrocyte membrane components in the presence of [y-SzPIATP.(A) Tris-acetate, pH 6.0; (B)Tris-HC1, pH 6.5; ( C )Tris-HC1, pH 7.5; (D) Tris-HC1, pH 8.5; (E) glycine-NaOH, pH 8.5. Data from Hosey and Tao (1976a).
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
265
tide 2. They attributed this effect to the action of a specific protein kinase. In summary, there is compelling, although indirect, evidence for the presence of cyclic AMP-independent protein kinases in red cell membranes. Recently, we have been able to solubilize the protein kinases from rabbit erythrocyte membranes and separate these enzymes into two active fractions (unpublished observations). One fraction specifically utilizes ATP as the phosphoryl donor to phosphorylate casein, while the other utilizes both ATP and GTP as substrates. These preliminary studies appear to bear out our conclusion derived from the analyses of membrane autophosphorylation.
4. PHOSPHORYLATION OF ERYTHROCYTE MEMBRANESB Y SOLUBLEPROTEINKINASES Both human and rabbit erythrocyte membranes can be phosphorylated by soluble cyclic AMP-dependent and -independent protein kinases isolated from rabbit erythrocyte lysates (Hosey and Tao, 1977). The substrate specificity of the three purified cyclic AMPdependent protein kinases from rabbit erythrocytes (Tao and Hackett, 1973) is more or less similar to that of the membrane-bound enzyme from human erythrocytes. The phosphorylation of components 2.1, 2.3,4.5,and 4.8 of the rabbit erythrocyte membranes is stimulated by cyclic AMP in the presence of these enzymes (Hosey and Tao, 1977). As compared to the phosphorylation of human erythrocyte membranes, the phosphorylation of rabbit erythrocyte membranes is stimulated to a greater. extent by the soluble cyclic AMP-dependent protein kinases, presumably because the human erythrocyte membrane already contains a similar enzyme activity. The phosphorylation of rabbit and human erythrocyte membrane components can also be catalyzed by the casein kinases purified from rabbit erythrocyte lysate (Kumar and Tao, 1975). One of the principal substrates of the casein kinase is a component with an electrophoretic mobility similar to 4.8 (Hosey and Tao, 1977), but whether this phosphopeptide is the same species as the substrate phosphorylated by the cyclic AMP-dependent protein kinase awaits further experimentation.
5. PHYSIOLOGICAL ROLE O F MEMBRANEPHOSPHORYLATION a. Human Erythrocytes. In spite of the relatively simple organization of the mammalian erythrocytes, the pattern of membrane phosphorylation obtained through the action of both membrane-bound and
266
M. MARLENE HOSEY AND MARIANO TAO
soluble kinases is quite complex. One can speculate about the functional significance of the various phosphoproteins on these membranes only if their functions are known. However, with the exception of bands 2 and 3, little information is available concerning the identity or function of other phosphorylated protein components of the red cell membranes. Band 2 is a part of the spectrin complex (Marchesi and Steers, 1968), which is a complex of actomyosin-like fibrillar proteins. Recent evidence indicates that the spectrin complex forms a submembranous matrix to which transmembrane and cytoplasmic surface membrane components are attached (Elgsaeter et al., 1973, 1976). That such a network is formed, is based, in part, on results of experiments using agents to modify spectrin structure (Elgsaeteret al., 1973, 1976) and of cross-linking studies (Wang and Richards, 1974). The spectrin network is probably responsible for the characteristic biconcave cell shape, susceptibility to deformation and other physical properties of the intact cell (Jacob et a1 ., 1972, Steck, 1974). As a result of this peripheral protein network, the free diffusion of surface proteins (Peters et al., 1974), the ligand-induced movement of antigens (Lee and Feldman, 1964; Nicolson et al., 1971; Loor et al., 1972), and the mobility of the membrane-intercalated particles (Pinto da Silva, 1972; Elgsaeter and Branton, 1974) are impeded. Hence, there is considerably less fluidity in the intact red cell membrane than other membrane systems (Nicolson, 1976). The spectrin complex contains a Ca2+-dependent ATPase activity (Rosenthal et al., 1970). This Ca2+-ATPase is insensitive to ouabain and inhibited by M$+. In view of the low affinity of this ATPase for Ca2+, and since Ca2+ transport requires M$+, Schatzmann (1975) suggested that this ATPase is probably unrelated to transport. However, the spectrin complex has properties similar to actomyosin and may be involved in the morphological behavior of red cells (Rosenthal et al., 1970; Schatzmann, 1975; Avissar et al., 1975). It has been suggested that the spectrin complex, polypeptides 1 and 2, constitute the myosin component, whereas polypeptide 5, the actin component (Guidotti, 1972; Steck, 1974; Avissar et al., 1975). Having considered the spectrin structure and its possible function in red cell membranes, one can begin to formulate a working hypothesis for the role of polypeptide 2 phosphorylation. It is possible that the phosphorylation of polypeptide 2 is required for maintaining the normal biconcave disc structure and the deformability of red cells. Dephosphorylation of phosphopeptide 2 of the spectrin complex results in disc to sphere transformation. This reasoning is in line with the observation that depletion of ATP causes disc to sphere trans-
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
267
formation, and that the biconcave disc shape of the red cell is restored after regeneration of ATP (Weed and LaCelle, 1969; Nakao, 1974). Conceivably, the phosphorylation of polypeptide 2 may have two major effects: (i) on the structure of the spectrin and (ii) on the Ca2+ ATPase activity of the spectrin complex. That Ca2+-ATPase activity may be altered seems likely since various studies indicate that the phosphorylation of myosin in a number of systems in an alteration of actomyosin ATPase activity (Adelstein and Conti, 1975; Gorecka et al., 1976). Clearly, more data are needed to establish the validity of this working hypothesis. It remains to be determined whether polypeptide 2 isolated from normal red cell membranes contains bound phosphate; and whether phosphopeptide 2 is dephosphorylated during disc to sphere transformation. The study of 32Pincorporation and turnover in intact cells under conditions where red cell shape transformation occurs may provide valuable clues regarding the role of polypeptide 2 phosphorylation. The effect of phosphorylation on the ATPase activity of the spectrin complex is another area that needs to be investigated. The area designated as band 3 is a glycoprotein(s) with a mass of approximately 90,000 daltons, (Furthmayr e t al., 1976). This component contains 8% carbohydrate (Ho and Guidotti, 1975) and represents 25-30% of the total erythrocyte membrane proteins (Steck, 1974). However, component 3 appears to be heterogeneous and may consist of a family of closely related polypeptides (Nicolson, 1976). There is sufficient evidence to indicate that band 3 penetrates and traverses the lipid bilayer so that parts of this protein are exposed on both the outer and the inner surfaces of the membrane (see recent reviews by Steck, 1974; and Nicolson, 1976). Furthermore, band 3 appears to exist as a dimer in the membrane, and in conjunction with glycophorin (a major sialoglycoprotein), makes up the membraneintercalated particles revealed by freeze-fracture deep-etch electron microscopy. Other inner surface membrane components, such as 4.1, 4.2, 6 (glyceraldehyde-3-phosphate dehydrogenase), and 7 are also probably associated with this complex. Pinto da Silva and Nicholson (1974) called this complex a “permeaphore” because of its role in transport processes and water movement. The role of component 3 in anion (sulfate)transport has been examined by Cabantchik and Rothstein (1974). They showed that anion transport is inhibited by the covalent linkage of 4,4’-diisothiocyano2’-ditritiostilbene disulfonic acid to a membrane protein fraction of 95,000 MW (presumably component 3).Because of the linear relationship of binding to inhibition and the unique architecture of the site, it
268
M. MARLENE HOSEY A N D MARIANO TAO
is suggested that the nonpenetrating fluorescent marker-binding site of component 3 is the substrate-binding site of the anion transport system. Kahlenberg (1976), on the other hand, studied the transport of glucose in human erythrocyte ghosts exposed to a variety of agents capable of selectively extracting various membrane proteins and to agents capable of cross-linking band 3. He concluded from his studies that glucose transport is associated with the proteins of band 3. I n another study, Ho and Guidotti (1975) show that band 3 is also involved in phosphate transport. Both anion (Rothstein et al., 1976; Ho and Guidotti, 1975) and glucose transports have been suggested to occur through a water-filled channel formed by specific subunit aggregation of the transport proteins in the erythrocyte membrane, a process distinct from the classical mobile carrier theory. Thus, the phosphorylation of band 3 could very likely influence the transport of one or more of these solutes by altering the interaction of the proteins of band 3. Because of the heterogeneity, both in composition and in function, of band 3, better separation techniques are needed to resolve the various components of band 3 in order to ascertain which of the band 3 polypeptide(s) is phosphorylated under different transport conditions. The fact that cyclic AMP can stimulate the phosphorylation of selected components of erythrocyte membranes by either soluble or membrane-bound protein kinases suggests a role for this nucleotide in red cell function. Hornlone-sensitive adenylate cyclase has been demonstrated in rabbit (Hosey and Tao, 1975) and rat (Sheppard and Burghardt, 1969) erythrocyte membranes. Furthermore, recent reports seem to indicate that endogenous activators of adenylate cyclase can stimulate phosphoprotein formation in intact mammalian and avian red cells. Duffy and Schwarz (1974) have demonstrated that membrane phosphorylation can be moderately stimulated in intact human red cells by norepinephrine and prostaglandin E2,and in intact rat red cells by norepinephrine, prostaglandin El, and prostaglandin EP, but are unable to correlate phosphorylation with changes in Na+ transport and Na+, K+-ATPase activity. However, these are preliminary studies and the question regarding the interrelationship between cyclic AMP and ion transport in mammalian erythrocytes remains unresolved. In avian erythrocytes (Rudolph and Greengard, 1974; Aurbach, 1975) and other membrane systems (Schwartz e t al., 1976), cyclic AMP-dependent protein phosphorylation has been implicated to regulate Na+ transport. I n addition, Braughler and Corder (1976) recently showed that the kinetic property of human renal Na+,K+-ATPase was modified by cyclic AMP-dependent protein
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
269
kinase. In view of these studies, the effect of phosphorylation on Na+K+-ATPase activity and cation transport in red cells needs to be reexamined in greater detail. b. Abnormal Erythrocytes. Several attempts have been made to determine whether altered membrane phosphorylation underlies or promotes various disease states in which there is a known or suspected membrane aberration. In a few instances, erythrocyte membranes have been chosen for the phosphorylation studies although they may not necessarily be the site of expression of altered physiological function in the disease. The phosphorylation of unfractionated erythrocyte membranes of patients with cystic fibrosis was found to be no different from those of normal subjects, both in the presence and absence of cyclic AMP (Dufi and Schwarz, 1973).This is not surprising since the membrane defect in this disease is an alteration in Na+ transport that appears to be confined to the exocrine glands. It has yet to be demonstrated whether the membranes of the affected glands contain altered protein kinase activity. In contrast, studies dealing with phosphorylation of erythrocyte membranes from patients with myotonic muscular dystrophy have lent support to the theroy that an underlying membrane defect may be responsible for many of the physical manifestations of this disease. Roses and Appel (1975)have demonstrated that a protein (designated as component a, 90,000-100,000 daltons) migrating with the band 3 complex of human erythrocyte membranes undergoes significantly less phosphorylation in erythrocytes from myotonic patients than from normal subjects. In contrast, in membranes obtained from patients with Duchenne muscular dystrophy, the phosphorylation of band 3, and of band 2, is increased over controls and myotonic membranes (Roses et al., 1975).Whether or not the kinase or the substrates are altered in the dystrophic membranes has not been resolved. Two laboratories have investigated the phosphorylation of erythrocytes from patients with hereditary spherocytosis, a disease characterized by an abnormal erythrocyte shape. Greenquist and Shohet (1974) initially reported that the phosphorylation of bands 1,2, and 3 was decreased in several patients with hereditary spherocytosis. This observation tends to support the premise that phosphorylation of spectrin is necessary to maintain the biconcave shape of red cells. Unfortunately, Zail and van den Hoek (1975) have been unable to confirm these results. In the latter study, no defect in the protein kinase activity was found in the membranes of the hereditary spherocytosis erythrocytes. Phosphorylation of membrane substrates and exogenous substrates was not significantly different from controls. In addition, the binding
2 70
M. MARLENE HOSEY AND MARIANO TAO
and dissociation of cyclic AMP to and from erythrocyte membranes were the same in normal preparations and those from patients with hereditary spherocytosis. The reasons for the discrepancy between these two studies are unknown. However, Zail and van den Hoek (1975) studied an early phase of phosphorylation, while Greenquist and Shohet (1974) assayed preparations phosphorylated for 60 min at 37°C. In considering the time course of erythrocyte membrane phosphorylation (Hosey and Tao, 1967a),other factors such as substrate depletion and phosphatase action may play a role in producing altered phosphorylation profiles a f p r a 60-min incubation. The above studies suffer from a lack of understanding of the nature of the protein kinases associated with the red cell membranes. Previous investigators made no attempt to examine the effect of monovalent and divalent cations on the phosphorylation of red cell membrane proteins, nor have they attempted to utilize GTP as the phosphoryl donor. As pointed out by us (Hosey and Tao, 1976a) and by Avruch and Fairbanks (1974), the red cell membranes contain multiple kinase activities, which are distinguishable by their phosphoryl donor specificity and cation requirement. Therefore, the phosphorylation of abnormal red cell membranes must be reexamined in the light of this new information. It is possible that entirely different results may be obtained from these studies. We have investigated the phosphorylation of human erythrocyte membranes from normal individuals and from patients with sickle cell disease (Hosey and Tao, 1976b). The autophosphorylation of the sickle cell membranes in the presence of GTP differs significantly from controls. One or more polypeptides in the area of bands 4.5-4.8 incorporate a significant amount of 32Pin the sickle cell membranes but not in the normal membranes. Furthermore, the phosphorylation of bands 2-2.1 is decreased in the membranes obtained from patients with sickle cell disease in comparison with controls. The phosphorylation of both types of membrane in the presence of ATP with or without cyclic AMP does not differ appreciably. We have yet to ascertain whether abnormalities in the phosphorylation of the diseased membranes are due to changes in the polypeptide substrates or to alteration in the enzyme(s) utilizing GTP. Although sickle cell disease is well defined as being due to the presence of an abnormal hemoglobin, membrane aberrations do occur. Sickle cells are leaky in terms of cation content and transport (Kuranstin-Mills e t al., 1974) and exhibit a higher degree of rigidity (Chien et al., 1970). Whether these or other altered membrane functions are in any way related to the altered patterns of membrane phosphorylation remains to be determined.
PROTEIN KINASES A N D MEMBRANE P H O S P H O R Y U T I O N
271
c. Avian Erythrocytes. In contrast to enucleated mammalian red cells, avian erythrocytes have an extremely active cateholaminesensitive adenylate cyclase (Davoren and Sutherland, 1963; Bilezikian and Aurbach, 1974). In addition, several studies (see Aurbach, 1975) have shown that isoproterenol and other catecholamines stimulate the bidirectional flux of Na+ and K+ across the plasma membranes of these cells. These effects of the catecholamines on ion transport are directly related to their effects on increasing the cyclic AMP concentrations in the intact erythrocytes. A short communication from Rudolph and Greengard (1974) indicates that in intact turkey erythrocytes, isoproterenol stimulates the phosphorylation of a membrane polypeptide with an apparent MW of 240,000. They suggested that this protein may correspond to band 2 found in human and rabbit erythrocytes. However, this appears unlikely since the phosphorylation of band 2 in human and rabbit erythrocyte membranes is not stimulated by cyclic AMP (Avruch and Fairbanks, 1974; Hosey and Tao, 1977).In any case, positive correlations between protein phosphorylation, cyclic AMP concentrations and Na+ uptake in turkey erythrocytes have been observed. C. Rhodopsin Phosphorylation
1. RHODOPSIN The phosphorylation of the visual pigment rhodopsin b y a kinase endogenous to rod outer segments (ROS) has been studied extensively in several laboratories. Before discussing the phosphorylation studies, it may be useful to review briefly basic facts about rhodopsin. The photosensitive rod outer segments, which project from the retina, are large cylindrical organelles. They consist of a plasma membrane that surrounds a stack of 1000-2000 discs, the number depending on the species. Each intracellular disc consists of a double layer of cell membrane composed of rhodopsin and phospholipid. These membranous particles have attracted much attention since they are easy to isolate in sufficient quantities in a biologically active state, i.e., they are physiologically excitable and will adapt to darkness in vitro. Rod outer segments consist of 40% lipid and 60% protein, 80% of which is rhodopsin. Rhodopsin itself is an apoprotein, composed of the protein opsin and the prosthetic group ll-cis retinal (retinene, retinealdehyde). Upon exposure to light, rhodopsin isomerizes into an all-trans configuration known as lumirhodopsin, which rapidly decays
272
M. MARLENE HOSEY AND MARIANO TAO
to metarhodopsin. In the presence of H 2 0 ,the latter compound dissociates into opsin (protein) and retinal (chromophore) moieties. This entire light-initiated process is known as bleaching. Opsin and retinal can be recycled back to rhodopsin in a two-step reaction. The first step requires the enzymatic conversion of the retinal from the all-trans form to the 11-cis isomer. This is followed by a spontaneous recombination of the opsin and retinal to yield the photosensitive rhodopsin. Rhodopsin can also be regenerated in vitro b y the addition of 11-cis retinal (or 9-cis retinal) to the bleached ROS. 2. CHARACTERISTICS O F RHODOPSINPHOSPHORYLATION AND DEPHOSPHORYLATION
The phosphorylation of rhodopsin has been demonstrated in the retinas of living frogs (Kiihn, 1974), in isolated bovine (Kiihn and Bader, 1976) and frog retinas (Miller and Paulsen, 1975; Kiihn and Bader, 1976), in frog (Bownds et al., 1974; Miller and Paulsen, 1975) and bovine (Kiihn et al., 1973; Frank et aZ., 1973; Weller et al., 1975a; Chader et al., 1975, 1976) rod outer segments, and in preparations of purified rhodopsin (Schichi et al., 1974; Weller et a2 ., 1975a, 1976). A unique aspect of rhodopsin phosphorylation is its light activation. Although light activates the phosphorylation reaction, the literature is unclear as to whether a direct relation exists between the extent of bleaching and the extent of phosphorylation. Several investigators have demonstrated a linear relationship between the amount of 32P incorporated into bovine rhodopsin and the percentage of bleached rhodopsin present (Kiihn et al., 1973; Frank et al., 1973; Weller et al., 1975a).However, Bownds et al. (1974) feel that this relationship holds only when the amount of bleaching is greater than 10%. These investigators have studied frog ROS and found that when bleaching is less than 5%, more 32Pis incorporated per mole of bleached rhodopsin than when bleaching is greater. Miller and Paulsen (1975) have demonstrated that bleached frog ROS membranes which have been regenerated 80-95% by addition of 11-cis retinal can be phosphorylated without further exposure to light. In contrast to these findings, Schichi et aZ(1974) have reported that, after regenerating bleached rhodopsin with 9-cis retinal, phosphorylation does not proceed without further reexposure to light. Whether the differences reported represent differences in techniques, such as ROS preparation or regeneration, the type of light used (white or orange), or whether they result from species variations is yet to be determined. Several laboratories (Kiihn et ul ., 1973; Miller and Paulsen, 1975; Weller et a2 ., 1975a) have found
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
2 73
that 3’P-phosphorylated opsin can be reconstituted to rhodopsin without loss of radioactivity. Dephosphorylation of rhodopsin in ROS has been observed to occur at a much slower rate than the phosphorylation reaction and is not affected by light (Kuhn et al., 1973; Frank et al., 1973; Weller et al., 1975a). Miller and Paulsen (1975) have reported that the rate of dephosphorylation is greater in crude preparations of ROS than in purified preparations. These results suggest that the phosphatase for ROS or a cofactor of the phosphatase may be removed during purification. 3. OPSINKINASE
The phosphorylation of rhodopsin can be demonstrated by incubating either intact retinas with 32P-orthophosphate (Miller and Paulsen, 1975), or membrane fractions with either [y-32P]ATP or [y-32P]GTP(Miller and Paulsen, 1975; Chader et al., 1975, 1976). Utilizing frog ROS, Miller and Paulsen (1975) showed that the phosphate transfer from GTP is less than that from ATP. However, Chader et al. (1975,1976) observe a greater maximum increase in light-stimulated phosphorylation of intact bovine ROS in the presence of GTP (46-fold increase) than in the presence of ATP ($fold increase). Using highpressure liquid chromatography, these latter workers found no transfer of the 32Pgroup from GTP to ATP. In contrast, Weller et al. (1976) report that opsin kinase obtained from highly purified ROS phosphorylates purified bleached rhodopsin much less efficiently in the presence of GTP than ATP. Neither cyclic AMP nor cyclic GMP has any effect on the light-stimulated phosphorylation of rhodopsin (Weller et al., 1975a, 1976; Frank and Buzney, 1975; Chader et al., 1975). Further characterization of the phosphorylation of isolated, intact ROS in the presence of ATP and GTP showed that differential regulation of phosphorylation occurs, depending on the phosphoryl donor used (Chader et a1 ., 1976). The evidence, thus, seems to indicate that more than one kinase is involved in phosphorylation of intact ROS. It appears less likely that there is only one enzyme, which uses either ATP or GTP, that is subject to separate physiological control. The phosphorylation of rhodopsin is catalyzed only by the kinase present in ROS. Frank and Buzney (1975) have shown that rhodopsin is not phosphorylated by kinases obtained from bovine brain, skeletal muscle, or cardiac muscle. The enzyme that catalyzes the phosphorylation of rhodopsin, “opsin kinase” (Weller et al., 1975a), can be easily extracted from ROS by homogenization (Weller et al., 1975a), hypotonic lysis (Miller and Paulsen, 1975), or sonication (Kuhn et al.,
274
M. MARLENE HOSEY AND MARIANO rAo
1973). This ease of extractability suggests that the enzyme is loosely associated with the interdiscal membranes. Neither the extracted membranes nor the solubilized enzyme show any appreciable amount of self-phosphorylation (Miller and Paulsen, 1975; Weller et al., 1975a). The solubilized enzyme can be added back to the extracted ROS fraction and phosphorylation will occur in a light-activated reaction. Exposure of the extracted membranes to light alone, followed by the addition of the unilluminated opsin kinase, but not the reverse, results in an activated reaction (Miller and Paulsen, 1975). Frank and Buzney (1975) have proposed that the mechanism of light activation of rhodopsin phosphorylation involves a conformational change in the rhodopsin moiety, which causes the appropriate phosphate acceptor site to become exposed. It is quite clear that light does not activate the kinase directly (Frank and Buzney, 1975; Miller and Paulsen, 1975). Interestingly, Weller et aZ. (1975a) found that if the rod outer segments are exposed to light before extraction of the kinase, both the quantity of protein and the kinase activity that can be extracted is significantly decreased. This indicates that the kinase is bound more tightly to bleached than to unbleached rhodopsin. The opsin kinase prefers endogenous substrate of the ROS over exogenous substrates such as histones (Weller et a1 ., 1975a, 1976). However, histones are also phosphorylated to some extent by the ROS kinase, and this reaction is stimulated by light (Kiihn et aZ., 1973). Interestingly, the phosphorylation of histones is stimulated by cyclic AMP even though cyclic AMP and cyclic AMP-dependent protein kinase have no effect on rhodopsin phosphorylation (Frank and Buzney, 1975). A similar observation has been made for myelinassociated kinase (Miyamoto and Kakiuchi, 1974; Camegie et a2 ., 1974). Two phosphodiesterase inhibitors, SQ 20,009 and theophylline, appear to inhibit directly the opsin kinase (Weller et al., 1975a). This effect, seen with the aid of solubilized enzyme to phosphorylatepurified rhodopsin, seems to be unrelated to the phosphodiesteraseinhibiting properties of these compounds. 4.
SITE(S) OF PHOSPHORYLATION
The phosphorylation site(s) of rhodopsin, as determined from a papain digest, are probably localized on the interdiscal surface of the ROS membranes, in a region distinct from the chromophore binding site (Virmaux et al., 1975). Papain removes about 36% of the rhodopsin-associated polypeptides from the interdiscal surface of the ROS, leaving a membrane-bound core that maintains all spectral prop-
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
2 75
erties and the retinaldehyde-binding site. Approximately 75% of the 3zP incorporated into rhodopsin is recovered in the papain soluble fraction, whereas less than 20% of the 32Pbound is found in the chromophore-binding fraction (Virmaux et al., 1975).
5.
PHYSIOLOGICAL
ROLE
O F PHOSPHORYLATION
Several studies have dealt with the role of phosphorylation in the response of rhodopsin to light. As rhodopsin phosphorylation and dephosphorylation are relatively slow processes (Kuhn, 1974; Kuhn and Bader, 1976), they cannot be implicated in the visual process itself. Rather, studies on the function of rhodopsin phosphorylation have centered on this process as a mechanism to regulate the sensitivity of the retina during light and dark adaptation. Weller et al. (1975b, 1975c) have studied the effects of light and phosphorylation on the permeability of rod outer segments to Caz+ and have provided evidence to indicate that both the efflux of 45Caand the entry of 45Caare greater in ROS exposed to white light than in those kept in the dark (dim red light). Furthermore, the phosphorylation of bleached rhodopsin results in a decrease in Ca2+permeability, such that the values of Ca efflux and influx are similar to those of ROS kept in the dark. These studies have been interpreted to implicate phosphorylation of rhodopsin as a means of adaptation of the retina to light. In this scheme (Fig. 7), light would cause an increase in permeability to Caz+, which may be directly or indirectly responsible for causing the known light-induced hyperpolarization of the ROS plasma membrane resulting in decreased permeability to Na+. In any event, light produces an increase in bleached rhodopsin that results in an increase in the phosphorylation of rhodopsin, which in turn causes a decrease in the permeability to Ca2+. Weller e t al. (197510, 1975c) feel that the phosphorylation mechanism may serve as a process to “switch-off’ photoreceptors activated by background illumination, causing the retina to be more sensitive to small increases in light. Similarly, they implicate dephosphorylation in the process of dark adaptation. Kuhn (1974) and Kuhn and Bader (1976) support this hypothesis in that the slow time courses of phosphorylation and dephosphorylation are compatible with the time courses of light and dark adaptation. Miller e t al. (1975)have similarly linked phosphorylation to the sensitivity of the retina to light. Their theory is supported by rather indirect evidence that indicates that phosphorylation inhibitors increase the sensitivity of the retina to the decrease in permeability caused by
276
M. MARLENE HOSEY AND MARIANO TAO
LIGHT
FIG.7. Diagrammatic representation of the possible role of phosphorylation in photoreceptors.
exposure to light. The use of rather high concentrations of adenosine as the inhibitor of protein kinase in these studies makes the data hard to interpret. D. Muscle Membrane Phosphorylation
1. INTRODUCTION The phosphorylation of muscle membrane proteins has been examined in various subcellular preparations of vertebrate cardiac, skeletal, and smooth muscle (see below), and invertebrate (Higgins and Greenberg, 1974) cardiac muscle. In each instance, attempts have been made to associate phosphorylation with contractile processes. Since characteristic differences have been found in the different types of muscle, each will be discussed separately. Frequently, the study of membrane phosphorylation is hampered by the difficulty in determining the subcellular source of the membrane proteins under consideration. This is especially true for muscle when attempts are made
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
2 77
to attribute functional significance to phosphorylation of specialized membranes, e.g., sarcoplasmic reticulum (SR). Calcium metabolism is regulated by a complex system of membrane in muscle cells, and it seems reasonable to attempt to correlate membrane phosphorylation with the regulation of calcium fluxes. However, caution must be exercised in analyzing data obtained from impure membrane preparations.
2. CARDIAC MUSCLE a. Membrane Preparations. The phosphorylation of cardiac muscle membranes has been studied in two types of microsomal fractions. One preparation is supposedly enriched in cell-surface membranes (Matsui and Schwartz, 1966), while the other is enriched in sarcoplasmic reticulum (Katz and Repke, 1967; Harigaya and Schwartz, 1969). Representative preparations of the cell-surface enriched preparations (Krause et al., 1973) and SR-enriched membranes (Wray e t a1 ., 1973) are judged by electron microscopy to be homogeneous vesicles and to be essentially free of mitochondria as assessed by enzymatic markers. On the other hand, Katz and co-workers (Katz and Repke, 1973; Katz et al., 1975; Kirchberger e t al., 1972; Tada e t al., 1974, 1975a) who utilize the method of Harigaya and Schwartz (1969) to prepare SR-enriched microsomes, admit that these membranes are contaminated with mitochondria and plasma membranes (Katz and Repke, 1973). The difficulties involved in ascribing purity to microsomal preparations has been discussed by Katz and Repke (1973). The phosphorylation of plasma membrane fractions has been studied in preparations obtained from porcine (Krause e t al., 1973, 1975) and guinea pig (Hui e t al., 1976) myocardium. The two preparations are both enriched in Na+,K+-ATPase, but differ with respect to Ca2+ metabolism. Vesicles obtained from porcine heart are poor in M$+dependent Ca2+-ATPase and do not accumulate Ca2+in the presence of oxalate (Krause et al., 1973). In contrast, the plasma membrane preparation from guinea pig ventricle contains both Ca2+-ATPase and Ca2+-dependentM$+-ATPase activities. These vesicles can accumulate Ca2+in the presence of oxalate (Hui et al., 1976). Phosphorylation studies in SR-enriched cardiac microsomes have been carried out in several laboratories (Kirchberger et al., 1972, 1974; Wray e t al., 1973; LaRaia and Morkin, 1974; Tada e t al., 1974, 1975a; Schwartz e t al., 1976). These vesicles are capable of accumulating large amounts of Ca2+in the presence of oxalate. Ca2+accumulation in the presence of oxalate is commonly referred to as Ca2+uptake, while that in the absence of oxalate is referred to as Ca2+binding.
278
M. MARLENE HOSEY AND MARIANO TAO
b. Cyclic AMP-Dependent Phosphorylation of Cardiac Muscle Membranes. Cyclic AMP-dependent protein kinase is present both in plasma membrane-enriched (Krause e t a1 ., 1973; Hui et a1 ., 1976) and SR (Wray et al., 1973; Kirchberger et al., 1974; Gibson and Newcombe, 1975) vesicles and can catalyze the autophosphorylation of these preparations. However, further phosphorylation is observed in either preparation by the addition of a soluble cardiac cyclic AMPdependent protein kinase. The phosphorylation of each type of membrane is dependent on M$+, and can be inhibited by high concentrations (greater than 1 mM) of Ca2+(Wray et al., 1973; Krause et a1 ., 1973, 1975). Studies of the cell-surface membrane fraction of porcine heart indicate that histone is not a substrate for this membrane-bound kinase, but it can be phosphorylated by a Triton extract of these microsomes. I n contrast, the membrane-bound kinase of the guinea pig plasma membranes (Hui et a1 ., 1976) and of SR fractions (Wray et a1 ., 1973) can phosphorylate histone in a reaction enhanced by the addition of cyclic AMP. Wray et al. (1973) initially reported that the phosphorylation of endogenous substrate by the SR protein kinase is unaffected by the heat stable inhibitor of cyclic AMP-dependent protein kinases. However, this is not supported by a recent report of Schwartz et al. (1976) who indicate that the inhibitor is effective in preventing cyclic AMP-dependent phosphorylation of their SR fraction. Data on comparable studies with plasma membrane-enriched membranes have not appeared. Although the cell surface-enriched and the SR-enriched microsomes possess hormone-sensitive adenylate cyclase, the ability of this enzyme system to influence phosphorylation has been tested only in SR. Studies of Kirchberger et al, (1974) demonstrated that addition of 10 yM epinephrine activated adenylate cyclase and caused phosphorylation of SR to increase, whether or not exogenous protein kinase was present. Adenylate cyclase activity is unaffected by the presence or absence of protein kinase. That these effects of epinephrine are inhibited by 20 yM propranolol indicates that they are mediated by &receptor activation. I n related studies, SR-enriched microsomes prepared from canine hearts 2 hours after a single dose of isoproterenol exhibited a three-fold increase in membrane phosphorylation (Fedelesova and Ziegelhoffer, 1975). Under these conditions, added cyclic AMP resulted in only a small increase in phosphorylation. In both studies, the endogenous activation of protein kinase by catecholamines is presumably due to the cyclic AMPelevating effect of these hormones. c. Substrate of the Cyclic AMP-Dependent Reaction. The most
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
279
striking similarity of the cyclic AMP-dependent protein kinasemediated phosphorylation of plasma membrane-enriched and SRenriched microsomes is that a protein of low MW appears to be the major substrate. This protein has an apparent MW of 20,000-22,000 in canine (Tada et al., 1975a; Schwartz et al., 1976) and rabbit SR (LaRaia and Morkin, 1974) and 24,000 in porcine cell-surface membranes (Krause et a1 ., 1973).Tentatively named phospholamban (Tada et al., 1974, 1975a), this protein is distinct from ATPase, has the characteristics of a stable phosphoester, and is insensitive to extraction by lipid solvents (Tada et a1 ., 1975a). The canine SR protein correlates with a protein component on SDS gels, which stains poorly with Coomassie Blue and which is not stained by the periodic acid-Schiffs reagent. Whether phospholamban bears any relation to the 20,000 MW low affinity Ca-binding protein described by Ostwald and MacLennan (1974) is not known. The similarities of protein phosphorylation in the cell-surface-enriched and SR-enriched membranes are striking. Conceivably, a similar protein exists on both plasma and SR membranes which can be phosphorylated and acts on Ca2+ flux at two different sites. The phosphorylation of the 22,000-dalton component of canine cardiac SR is reversible (Tada et a1 ., 1975b). Dephosphorylation of the protein is catalyzed b y SR-membrane bound or by cytoplasmic phosphoprotein phosphatase. The occurrence of this enzyme satisfies the requirement that phosphoproteins undergo rapid turnover in order to have any physiological significance. d . Phosphorylation of SR by Phosphorylase Kinase. A recent study by Schwartz et al. (1976) has demonstrated that SR membranes also undergo phosphorylation mediated by endogenous or exogenous phosphorylase b kinase. The phosphorylation of SR membranes by phosphorylase b kinase differs from that catalyzed by the cyclic AMP-dependent protein kinase in several ways. (i) The reaction mediated by phosphorylase b kinase is not inhibited b y the heat stable inhibitor of protein kinase, but is blocked by EGTA. (ii) The protein substrate of SR membranes which is phosphorylated by phosphorylase b kinase has a MW of 95,000. The phosphorylation of the 22,000-dalton component of SR is not mediated by phosphorylase b kinase. The identity of the 95,000-dalton protein is not known; however, Schwartz et al. (1976) have suggested that it may be Ca2+ ATPase, phosphorylase, the kinase itself, or perhaps an unknown component. e. Physiological Significance of Cardiac Muscle Phosphorylation. The contractile activity of cardiac muscle is regulated by Ca2+.Ca2+
2 80
M. MARLENE HOSEY A N D M A R I A N O TAO
activates the myofilaments by binding to troponin C, thereby preventing the troponin-tropomyosin complex from inhibiting the interactions of the contractile proteins, actin and myosin (Ebashi and Endo, 1968; Katz, 1970). An increase in the intracellular level of Ca2+ leads to contraction; conversely, relaxation follows as Ca2+is removed by uptake and extrusion mechanisms. Both extracellular and intracellular Ca2+contribute to the increases in Ca2+levels required to support contraction (Langer, 1973). However, extracellular Ca2+ that enters the cell does not directly reach the contractile machinery, but rather, is sequestered into the sarcoplasmic reticulum (Katz et al., 1975). From there, Ca2+is released to the cardiac contractile proteins. Myocardial Contractility is enhanced by agents such as epinephrine which activate adenylate cyclase. According to Katz et al. (1975), the positive inotropic effects of these agents may be mediated by cyclic AMP-dependent phosphorylation of a specific membrane protein, phospholamban (Fig. 8). The phosphorylation of phospholamban in sarcoplasmic reticulum leads to an increase in Ca2+uptake (Tada e t al., 1974). Presumably, a similar protein is phosphorylated in the plasma membranes. The phosphorylation of plasma membranes either prolongs or enhances the influx of Ca2+(Hui et al., 1976). The end result is to increase the storage of Ca2+within the sarcoplasmic reticulum, thus, increasing the quantity available for release during contraction. Conversely, the enhanced rate of Ca2+uptake into the sarcoplasmic reticulum may increase the rate of relaxation of cardiac muscle, an effect that is observed with agents such as epinephrine. On the other hand, Schwartz et al. (1976) object to emphasis put on phospholamban in the regulation of the Caz+pump and muscle contraction. Their objection is based on the finding that the phosphorylation of a 95,000 MW component of sarcoplasmic reticulum by phosphorylase kinase produces similar enhancement of Ca2+uptake. They suggest that catecholamines augment contraction and accelerate the rate of relaxation by increasing the intracellular concentrations of cyclic AMP and Ca2+. Both Ca2+ and cyclic AMP (through cyclic AMP-dependent protein kinase) can activate phosphorylase b kinase, which in turn stimulates glycogenolysis and Ca2+uptake into the sarcoplasmic reticulum. This hypothesis suggests that contractility may be tightly coupled to glycogen metabolism, from which energy is derived. Schwartz et al. (1976) and Katz et al. (1975) place different emphasis on the enzyme actually involved in the stimulation of Ca2+ uptake. However, there seems to be no compelling argument against the possibility that cyclic AMP-dependent protein kinase and phosphorylase b kinase both may be involved in the regulation of Caz+up-
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
28 1
take, perhaps one potentiating or complementing the action of the other. In addition to modifying the Ca2+pump, it is also possible that both protein kinases may regulate muscle contraction by altering the activity of troponin. As discussed earlier, both cyclic AMP-dependent protein kinase and phosphorylase kinase can catalyze the phosphorylation of troponin I (Cole and Perry, 1975) of the troponin complex. Thus, phosphorylation could regulate the inhibitory action of troponin I on the actomyosin ATPase. The cyclical phosphorylation-dephosphorylation of troponin could also account for the effects of catecholamines on the contractile cycle of the myocardium.
HORMONE
n
I
ATP
EXTERNAL
Ca2+
FIG. 8. Role of membrane phosphorylation in the regulation of the contractile process in cardiac muscle. Unfilled arrows indicate stimulation.
2 82
M. MARLENE HOSEY AND MARIANO TAO
3. SKELETALMUSCLE
a. Plasma Membranes. Plasma membranes of rat diaphragm (Pinkett and Perlman, 1974) and rabbit hind leg skeletal muscle (Sulakhe and Drummond, 1974) appear to contain endogenous protein kinase activity that is independent of cyclic AMP. A microsomal fraction of rat diaphragm rich in Na+-K+-ATPase and 5’-nucleotidase activities contains protein kinase activity that phosphorylates histone and a minor membrane with an apparent MW of 51,000 (Pinkett and Perlman, 1974). The phosphorylation of this fraction is independent of cyclic AMP and is inhibited by caffeine and theophylline. As observed earlier with rhodopsin phosphorylation (Weller et al., 1975a), the mechanism of the inhibitory effects of these compounds is unknown but appears to be unrelated to their actions as phosphodiesterase inhibitors. Sulakhe and Drummond (1974) studied the phosphorylation of rabbit skeletal muscle sarcolemma by bovine heart cyclic AMPdependent protein kinase. The phosphorylation of these membranes, prepared according to Sulakhe et al. (1973), was increased twofold in the presence of the exogenous protein kinase alone and six- to sevenfold by the kinase plus cyclic AMP. The membrane substrate(s) of the cyclic AMP-dependent protein kinase have not been identified. Both Ca2+-stimulatedM$+-ATPase and Ca2+-uptakewere increased in the phosphorylated membranes. Ca2+binding was unchanged. b. Sarcoplasmic Reticulum Membranes. In contrast to results obtained with cardiac membranes, Katz et al. (1975) have reported that they have been unsuccessful in attempts to phosphorylate SRenriched microsomes prepared from rabbit fast skeletal muscle. Phosphorylation of these membranes did not occur in the presence of a cyclic AMP-dependent protein kinase prepared from the soluble fraction of either skeletal muscle or cardiac muscle. Furthermore, the addition of protein kinase to the skeletal muscle microsomes did not result in increased Ca2+uptake. However, recent studies of Schwartz et al. (1976) have demonstrated that SR fragments prepared from slow (soleus) and mixed fast (tibialis) skeletal muscles of the cat do undergo phosphorylation. In SR obtained from slow skeletal muscle, the phosphorylation pattern is analogous to that observed for cardiac muscle. A 20,000-dalton component is phosphorylated in the presence of exogenous or endogenous cyclic AMP-dependent protein kinase, while a 95,000-dalton component is phosphorylated by endogenous or exogenous phosphorylase b kinase. In contrast, SR fragments derived from fast (tibialis) skeletal
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
283
muscle contain only one phosphoprotein with an apparent MW of 95,000. Although this latter reaction is stimulated by both exogenous cyclic AMP-dependent protein kinase and phosphorylase b kinase, characterization of the reaction indicates that the cyclic AMPdependent protein kinase effect is most likely due to activation of endogenous phosphorylase b kinase in the SR fraction. In both slow and fast muscle, phosphorylation of SR protein is accompanied by an increase in Ca2+transport. c. Other Membrane Fractions. Additional studies of skeletal muscle phosphorylation have been carried out by Andrew et al. (1973, 1975). Utilizing fractions isolated from rat hind limb muscle, these investigators found the active membrane-bound phosphotransferase activity localized in a light microsomal fraction that did not enter a 15-36% sucrose gradient. The fraction did not coincide with enzyme markers for SR or plasma membranes, but was extensively iodinated by lactoperoxidase, indicating the presence of outer surface membranes. The origin of the membranes containing this kinase activity is unknown, but may represent outer sarcolemma or vesicles of transverse tubule membranes. The kinase is not regulated by cyclic nucleotides and predominately phosphorylates a single membranebound polypeptide estimated to be 28,000 daltons by SDS gel electrophoresis. Other components are also phosphorylated, but to a lesser degree. Whether or not exogenous cyclic AMP-dependent protein kinase could phosphorylate these membranes was not reported. The phosphorylation of the 28,000-dalton substrate was studied in membranes isolated from both normal and denervated muscle. The content of the 28,000 MW protein was decreased in fractions of denervated muscle to 40% of the control, A corresponding decrease in phosphorylation accompanied the decrease in protein, although it is not clear if the amount of phosphate incorporated per mole of substrate was also affected. The phosphorylation of crude membrane fractions derived from human skeletal muscle (quadriceps femori) from normal subjects and from patients with myotonic muscular dystrophy has been studied (Roses and Appel, 1974). Although there is little difference in the polypeptide profiles of normal and diseased membranes, the phosphorylation of the myotonic membranes was approximately 64% of the control when assayed at pH 6.5. The phosphorylation of two components, MW 50,000 and MW 30,000, was consistently less in the myotonic membranes at pH 6.5. However, when assayed at pH 7.5, no differences in phosphorylation were detected; the significance of these observations is as yet unclear.
284
M. MARLENE HOSEY AND MARIANO TAO
d . Physiological Significance of the Phosphorylation of Skeletal Muscle Membranes. The evidence discussed above suggests that phosphorylation of skeletal muscle membranes occurs at multiple sites. Protein substrates of exogenous and endogenous phosphotransferases appear to be located on plasma and SR membranes and possibly in association with membranes of the transverse-T-tubule system. Although the physiological significance of these phosphorylations is not clear, it is tempting to relate these processes to the control of calcium fluxes and contractility. The phosphorylation of SR proteins of slow skeletal muscle occurs in a manner analogous to that in cardiac muscle SR, i.e., a 20,000dalton component is phosphorylated by a cyclic AMP-dependent protein kinase and a 95,000-dal ton component is phosphorylated by phosphorylase b kinase. However, in membranes derived from SR of fast skeletal muscle, only the reaction catalyzed by phosphorylase b kinase occurs. I n both fast and slow skeletal muscle, phosphorylation of SR membrane proteins is accompanied by increases in Ca2+transport. The observation that increased Ca2+ transport occurs in fast skeletal muscle in the absence of the cyclic AMP-mediated phosphorylation of a 22,000-dalton component (phospholamban) appears to emphasize the importance of phosphorylase b kinase, and perhaps the entire gl ycogenolytic process, in the regulation of the excitationcontraction coupling processes. Similarly, the cyclic AMP-mediated process of phospholamban phosphorylation is deemphasized, as it becomes evident that this reaction is not essential to the regulation of all contractile processes. These results are not surprising since the contractility of fast skeletal muscle is not as sensitive to regulation b y epinephrine and other agents that act via cyclic AMP as cardiac and slow skeletal muscles. One would not, therefore, expect cyclic AMPmediated processes to play as much of a role in the regulation of excitation-contraction coupling in fast skeletal muscle as it may in slow skeletal and cardiac muscle. The mechanism whereby phosphorylation of the two SR proteins may regulate Ca2+ fluxes has been discussed in the section on the phosphorylation of cardiac muscle. That the phosphorylation of skeletal muscle plasma membranes by exogenous cyclic AMP-dependent protein kinase is accompanied by increases in Ca2+uptake and Ca2+-dependentM$+-ATPase (Sulakhe and Drummond, 1974) suggests that cyclic AMP may serve as a regulatory effector of Ca2+fluxes at the plasma membrane. Similar observations have been made in studies of cardiac muscle (see above). This hypothesis gains further support from the observation that these sarcolemma fractions contain an active adenylate cyclase system (Severson
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
285
et al., 1972). The origin, i.e., fast or slow skeletal muscle, of these fractions is not specified; we assume both types were included. The lack of an endogenous cyclic AMP-dependent protein kinase in sarcolemmal preparations (Sulakhe and Drummond, 1974; Pinkett and Perlman, 1974) does not eliminate this hypothesis since skeletal muscle does contain multiple forms of soluble cyclic AMP-dependent protein kinases. However, whether the soluble kinases are accessible to the plasma membrane substrates in the intact cell has not been determined. Although Pinkett and Perlman (1974)were unable to stimulate endogenous sarcolemmal protein kinase with cyclic AMP, they did observe that membranes isolated from rats previously injected with isoproterenol or epinephrine contained increased protein kinase activity. This suggests that the sarcolemmal membranes may contain an active cyclic AMP-dependent protein kinase whose regulatory subunit is lost during isolation procedures. Thus, the data seem to indicate that cyclic AMP may play a role in the regulation of plasma membrane phosphorylation and Ca2+transport processes. That phosphorylation of membrane proteins may play a regulatory role in the contractile process in skeletal muscle is also suggested by the studies of Andrew e t al. (1975) of normal and denervated muscle. Denervated skeletal muscle is characterized b y many changes in membrane-related processes, one of which is spontaneous contractions or fibrillations. Whether or not the specific loss of the 28,000dalton protein kinase substrate from denervated muscle membrane fractions can be correlated in any way with the onset of spontaneous activity is not known. Andrew et al. (1975) favor the hypothesis that the membrane fraction used in their studies represents T-tubule fragments. However, it is unsettling that the amount of the major substrate for phosphorylation present in these fractions is significantly decreased after denervation, whereas the T system is known to proliferate after denervation (Andrew et al., 1975). Further studies should delineate both the origin and the significance of the phosphorylation of 28,000-dalton in normal and denervated skeletal muscle.
4. SMOOTHMUSCLE Numerous studies relating cyclic nucleotides and smooth muscle contraction have been performed (Schultz and Hardman, 1975; Andersson et al., 1975). However, only one report has appeared concerning cyclic nucleotide-mediated phosphorylation of smooth muscle membranes. Casnellie and Greengard (1974) have reported that the phosphorylation of two proteins in undefined microsomal
286
M. MARLENE HOSEY AND MARIANO TAO
fractions of several smooth muscle-rich tissues is stimulated by cyclic GMP, with ATP as the phosphoryl donor. Whether or not GTP can be used as the phosphoryl donor was not reported. These phosphoproteins, MW 130,000 and MW 100,000, were found in smooth muscle preparations isolated from rabbit small intestine and guinea pig ductus deferens and uterus. The apparent K , for cyclic GMP was approximately one-tenth that for cyclic AMP. The phosphorylation of a third protein with an apparent MW of 50,000 was regulated by cyclic AMP. The significance of these observations is difficult to assess since it is not known whether the phosphorylation occurs in the membranes of smooth muscle parenchymal cells or in the membranes of some minor cell type present in the preparation. In addition, the subcellular origins of the membranes studied is not known. Cyclic AMP has been implicated in the relaxation of smooth muscle (Andersson, 1972; Bar, 1974). In some smooth muscle, stimulation of P-adrenergic receptors results in an increase in cyclic AMP concentration, which precedes relaxation. The Ca2+-ATPase and Ca2+-binding activities of various smooth muscle microsomal fractions are stimulated by cyclic AMP (Andersson and Nilsson, 1972; Nilsson, 1973). Conceivably, this stimulatory process may be similar to that which occurs in the heart. The role of cyclic GMP in the contraction-relaxation cycle is less certain. Although a number of agents that stimulate contraction also stimulate cyclic GMP formation, this increase in cyclic GMP appears to be a secondary event brought about by the increase in intracellular Ca2+concentration associated with contraction (Andersson et al., 1975). However, cyclic GMP may participate in the release of Ca2+by blocking the stimulatory action of cyclic AMP on Ca2+binding to smooth miiscle microsomes. It is possible that this process is due to a second site phosphorylation mediated by the cyclic GMP-dependent protein kinase. The role of cyclic AMP-dependent and cyclic GMP-dependent protein kinases in smooth muscle contraction-relaxation must be reexamined using well-defined membrane preparations. E. Synaptic Membrane Phorphorylation
1. CYCLICAMP-DEPENDENT PHOSPHORYLATION
Evidence suggesting a role for cyclic AMP in neuronal transmission (see Greengard, 1975 and Bloom et al., 1975) has given added impetus to the study of the phosphorylation of synaptic membrane proteins. Subcellular fractions of rat brain characterized by electron microscopy
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
287
as being rich in synaptic membrane fragments have been shown to be substrates for both endogenous and exogenous cyclic AMPdependent protein kinases (Johnson et al., 1971). Greengard and co-workers (Ueda et al., 1973; Maeno et aZ., 1975)and Weller and Rodnight (1971, 1973) have characterized the synaptic membrane phosphorylation reaction in greater detail. Two proteins, I and 11, of synaptic membrane fractions of rat brain have been identified by SDS pol yacrylamide gel electrophoresis and autoradiography as substrates for endogenous cyclic AMP-dependent protein kinase (Ueda e t al., 1973). Phosphoprotein I (MW 86,000) appears to be characteristic of synaptic proteins, since it is found only in synapse-containing neural tissue (Ueda et a1 ., 1973). In contrast, protein I1 (MW 49,000) appears to be more wideIy distributed, since its cyclic AMP-stimulated phosphorylation is observed in many membranes of neural and nonneural origin (Ueda et al., 1973). Although protein I may be specific to synapses, Rubin e t al. (1972) have reported that a protein of similar molecular weight is also phosphorylated by a cyclic AMP-dependent protein kinase in human erythrocyte membranes. The rapid time course of the cyclic AMP-stimulated phosphorylation and dephosphorylation of synaptic fragments from bovine brain (Weller and Rodnight, 1973) and especially of protein I1 of rat brain synaptic membranes (Ueda et al., 1973) appears to be characteristic of these membranes. The stimulation of phosphorylation of bovine brain synaptic membranes by cyclic AMP is 100% at 1 min but less than 20%at 10 min (Weller and Rodnight, 1973). Previous studies by these investigators (Weller and Rodnight, 1971) had shown that the ratelimiting step of membrane phosphorylation is in the action of a membrane phosphatase. If the bovine brain synaptic membranes are preincubated under dephosphorylation conditions and then phosphorylated, the stimulation of phosphorylation by cyclic AMP is greatly enhanced. Weller and Rodnight (1971) concluded that in this preparation the effect of cyclic A M P is limited by the state of phosphorylation of the membrane and that cyclic AMP does not appear significantly to alter the rate of phosphatase activity. The maximum stimulation by cyclic AMP of phosphorylation of proteins I and I1 of rat brain synaptic membranes occurs at 5 sec. After 5 sec protein I1 is rapidly dephosphorylated in the presence of cycIic AMP, with control levels reached within 2 min. Maeno et al. (1975) have attributed these rather unique kinetic properties to the presence of cyclic AMP-stimulated protein phosphatase activity. This phosphatase does not appear to require divalent cations for activity as it is active in the presence of 0.1 mM EDTA (Maeno et al., 1975). The half-maximal concentrations of cyclic AMP required for stimulation
288
M. MARLENE HOSEY AND MARIANO TAO
of phosphorylation and dephosphorylation of protein I1 were about 9x M (Ueda et al., 1973) and 7 x M (Maeno et al., 1975), respectively. Protein I does not appear to be a substrate for the phosphatase, which acts on protein 11. The mechanism of cyclic AMPstimulated dephosphorylation of protein I1 has not been delineated.
2. REGULATIONO F SYNAPTIC MEMBRANEPHOSPHORYLATION BY
DIVALENT CATIONS
The phosphorylation of synaptic membrane proteins I and I1 differs in response to Znz+(Ueda et al., 1973, 1975). In the absence of cyclic AMP and M$+, ZnZf stimulated the phosphorylation of protein 11. This effect of Znz+ was not observed in the presence of cyclic AMP, nor was a stimulatory effect of Znz+observed for protein I. The physiological significance of the effect of Znz+is unknown; however, Ueda et al. (1975) have recently reported that the effect of 10 mM Zn2+on the phosphorylation of protein I1 can be overcome by 10 mM M$+. Whether or not the differential effects of ions and phosphatase on the phosphorylation of proteins I and I1 is due to intrinsic differences in the proteins themselves or to the action of separate kinases is unknown. DeLorenzo (1976) has shown that the phosphorylation of several proteins in synaptic membrane-enriched preparations from rat cerebral cortex is regulated by calcium. The phosphorylation of two proteins, with MWs of 60,000-63,000 and 49,000-52,000, is highly dependent on the presence of calcium. The concentration of calcium necessary to produce half-maximal stimulation of these reactions is 30-80 p M . Other proteins in the synaptic membrane fractions also incorporate more phosphate in the presence of calcium, however, the effects are less marked. The role of cyclic AMP in the calciummediated phosphorylations has not been reported. Whether or not the effect of calcium is mediated directly or indirectly has yet to be determined. It is conceivable that calciurfi causes the release of norepinephrine, which stimulates the production of cyclic AMP; the latter, in turn, might stimulate phosphorylation. However, it is also possible that calcium directly stimulates a synaptic membrane kinase in a manner analogous to the stimulation of phosphorylase kinase.
3. SOLUBILIZATION OF SYNAPTIC MEMBRANEPROTEIN I N A S E AND
SUBSTRATES
Recent studies of synaptic membrane phosphorylation have shown that protein I1 and its associated protein kinase and phosphatase can be extracted from synaptic membranes by Triton X-100 (0.25%) or
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
2 89
NH4Cl (0.1 M ) (Ueda et a2 ., 1975). Although some of the properties of the phosphorylation-dephosphorylation system observed in the intact membranes were preserved in the extracts, others were altered. The phosphorylation of Triton-solubilized protein I1 was stimulated b y cyclic AMP in the presence of Mg2+ and inhibited by cyclic AMP in the presence of Zn2+.I n contrast, in the NH4Cl extract, the phosphorylation of protein I1 was inhibited by cyclic AMP in the presence of either M$+ or Zn2+.The phosphorylation of protein I was observed in the NH4C1extract, but was not seen in the Triton extract. In both extracts, cyclic AMP stimulated the dephosphorylation of protein I1 in the absence of divalent cations. Kinetic studies of the solubilized system have led Greengard and his associates to postulate that protein 11, its kinase, and its phosphatase exist as a complex in the intact membrane (Maeno et al., 1975). The role of the phosphorylation of proteins I and I1 in neuronal transmission remains to be determined.
4. ROLE OF PHOSPHORYLATION
IN
NEURONALTRANSMISSION
The cyclic AMP-dependent phosphorylation of synaptic membranes has been implicated in neuronal transmission. This speculation is based on electrophysiological ,pharmacological, and biochemical evidence to the effect that cyclic AMP may mediate the action of certain putative neurotransmitters, such as norepinephrine, dopamine, octopamine, and serotonin, both in the central and autonomic systems (see Bloom et al., 1975; Greengard, 1975). A simplified version (Fig. 9) of this hypothesis is that the transmitter is released presynaptically, interacts with its specific receptor, and activates adenylate cyclase. The resultant increase in cyclic AMP activates the protein kinase located in the postsynaptic membrane. The phosphorylation of specific proteins in the postsynaptic membrane is thought to result in ion permeability changes, which in turn give rise to the postsynaptic potential. Although cyclic AMP-dependent phosphorylation of proteins in synaptic membrane preparations reaches a maximum at 5 sec (the earliest time period studied; Ueda e t al., 1973), it is well known that the generation of postsynaptic potentials occurs in the course of a few hundred milliseconds. Obvious technical difficulties have as yet made it impossible to correlate specific membrane protein phosphorylation and postsynaptic potential generation. Nevertheless, the fact that phosphorylation appears to be maximal at the earliest time period that can be tested and that at least one membrane protein undergoes rapid dephosphorylation b y a membrane-bound phosphatase (Maeno et al., 1975) argue in favor of membrane phosphorylation as causing the postsynaptic potential.
290
M. MARLENE HOSEY AND MARIANO TAO
IONS
n
FIG.9. Diagrammatic representation correlating protein phosphorylation with synaptic membrane processes. The presence and/or functions of c$+,the enzymes and the phosphoproteins in the p r e and postsynaptic terminals, as illustrated, is purely speculative as discussed in the text. Unfilled arrows indicate stimulation.
Another problem that needs consideration is one frequently encountered in studies of membrane phenomena, namely, the origin of the proteins in the synaptic membrane preparation undergoing phosphorylation. Since proteins I and I1 might be located either pre- or postsynaptically, it may not be unreasonable to consider a presynaptic role for membrane phosphorylation, such as transmitter release, etc. DeLorenzo (1976) has proposed that the calcium-mediated phosphorylation of synaptic-membrane fragments (specifically of proteins of -60,000 and 50,000 daltons) may play a role in such processes. It is
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
291
argued (DeLorenzo, 1976) that calcium causes the release of norepinephrine from axon terminals, which results in the generation of the post-tetanic potential. Diphenylhydantoin, an agent commonly used in epileptic disorders, antagonizes the calcium-mediated release process and prevents the generation of the posttetanic potential. DeLorenzo (DeLorenzo, 1976; DeLorenzo and Glaser, 1976) claims that diphenyl hydantoin prevents the calcium-stimulated phosphorylation of rat brain proteins in homogenates and in synaptic fragments. Although the theory is attractive, further investigation is necessary to ascertain the role of cyclic AMP in these Ca2+-mediatedprocesses (for reasons discussed earlier). Furthermore, it becomes increasingly more apparent that attempts be made to ascribe a pre- or postsynaptic origin to phosphorylated synaptic membrane fragments before ascertaining their physiological role in neuronal transmission. F. Myelin Phosphorylation
Myelin membranes are composed mainly of lipid with few major protein components. The phosphorylation of myelin membranes by endogenous and exogenous protein kinases has been demonstrated in several laboratories (Johnson et al., 1971; Camegie et al., 1973, 1974; Miyamoto and Kakiuchi, 1974; Steck and Appel, 1974). The substrate is a small basic protein termed “myelin basic protein.” Other myelin proteins are not phosphorylated (Carnegie et al., 1974; Steck and Appel, 1974). Both bovine and human myelin contain only one basic protein (MW 18,000), whereas rat myelin contains two basic proteins (MW 18,000 and 15,000).All of these appear to undergo phosphorylation. Endogenous phosphorylation of myelin membranes has been demonstrated in extensively washed myelin fractions from rat brain (Steck and Appel, 1974; Miyamoto and Kakiuchi, 1974) and bovine spinal cord (Carnegie et al., 1974). SDS polyacrylamide gel electrophoresis of either phosphorylated whole myelin (Steck and Appel, 1974; Carnegie et al., 1974) or its acid extract (Steck adn Appel, 1974; Miyamoto and Kakiuchi, 1974) has identified myelin basic protein as the substrate of the endogenous kinase. Cyclic nucleotides do not alter the activity of the myelin protein kinase (Carnegie et al., 1974; Miyamoto and Kakiuchi, 1974), nor can GTP replace ATP as the phosphoryl donor (Camegie et al., 1974). The phosphorylation of intact myelin can be further increased by the addition of exogenous cyclic AMP-dependent protein kinases isolated from rabbit muscle (Carnegie et a1 ., 1973) and bovine brain (Miyamoto and Kakiuchi, 1974). This additional phosphorylating activity is also unaffected by cyclic
292
M. MARLENE HOSEY AND MARIANO TAO
AMP (Miyamoto and Kakiuchi, 1974). However, when myelin protein is extracted and purified from the myelin membranes, phosphorylation of the purified proteins by exogenous protein kinases becomes cyclic AMP dependent (Miyamoto and Kakiuchi, 1974; Carnegie et al., 1974). Although the phosphorylation of intact myelin by endogenous protein kinase activity is unaffected by cyclic AMP, a cyclic AMPdependent enzyme has been extracted from myelin membranes with 0.2% Triton-X-100 (Miyamoto, 1975). The phosphorylation of histone by the solubilized enzyme is stimulated by cyclic AMP, whereas the phosphorylation of myelin basic protein is not. At present, it is difficult to reconcile these observations and to determine the significance of the cyclic AMP-dependent phosphorylation of isolated myelin basic protein. That the phosphorylation of myelin basic protein occurs in vivo has been demonstrated in rats injected intracranially with 32Porthophosphate (Steck and Appel, 1974; Miyamoto and Kakiuchi, 1974). Steck and Appel (1974) were unable to detect significant labeling of myelin 5 min after the injection, but incorporation of phosphate was detected 30 min later (Miyamoto and Kakiuchi, 1974). Acid extracts of rat myelin phosphorylated in vivo revealed that both basic proteins were phosphorylated. The sites of phosphorylation of myelin basic proteins by both exogenous and endogenous protein kinases have been identified (Carnegie et d . , 1974; Daile et d . , 1975). Among the 26 serine residues found in myelin basic protein, only two are phosphorylated. The endogenous myelin protein kinase phosphorylates serine-55, while the exogenous kinases preferentially phosphorylate serine-110. Phosphorylation of serine-110 by the exogenous protein kinases is identical regardless of the source of kinase, whether bovine brain, heart, or rabbit skeletal muscle (Daile et al., 1975). Deibler et al. (1975) have presented evidence which also indicates that only two sites on any polypeptide chain are phosphorylated in native myelin basic protein. A recent study by Smith et al. (1976) indicates that phosphorylation of myelin basic protein also occurs on basic amino acids. Their results suggest that the formation of rapidly turning over, acid-labile, basestable nitrogen phosphates on histidine and arginine residues is catalyzed by an endogenous protein kinase (Smith et al., 1976). In addition, Smith et al. (1976) suggest that the base-stable phosphates account for the greatest percentage of the total phosphoprotein formation in myelin basic protein.
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
293
Myelin is more or less a passive membrane, which acts principally as a permeability barrier. Since the function of myelin basic protein is unknown, it is difficult to ascertain the physiological significance of the phosphorylation of this protein. As the phosphorylation does appear to occur in vivo, further experimentation may capitalize on this finding in an attempt to determine the role of the basic protein in myelin function. G. Microtubule Phosphorylation
1. INTRODUCTION Although microtubules are not themselves considered to be membranous proteins, in some instances, much of the microtubule protein is associated with cell membranes. This appears to be especially true of synaptic membranes (Lagnado et al., 1971). A discussion of microtubule phosphorylation, therefore, seems pertinent. The major component of microtubules is the tubulin dimer with a subunit molecular weight of 55,000-56,000 dal tons. Although this protein has the tendency to form large aggregates, the inclusion of GTP during tubulin purification appears to prevent aggregation (Weisenberg et al., 1968). The tubulin dimer binds colchicine, and this property is used to assay for the presence of tubulin. Colchicine does not bind to tubulin aggregates (Weisenberg et al., 1968). In 1970, Goodman et al. observed that the major protein component of a bovine brain microtubule preparation served as a substrate for an endogenous cyclic AMP-dependent protein kinase and for a soluble brain cyclic AMP-dependent protein kinase. Since that time, this phenomenon has been the somewhat controversial subject of numerous investigations. The problems that have been addressed most concern the nature of the microtubule-associated kinase activity, and whether or not tubulin itself is a substrate for protein kinases. OF MICROTUBULE PROTEINS 2. PHOSPHORYLATION
Phosphorylation of tubulin has been reported to occur in preparations obtained from rat (Murray and Froscio, 1971; Eipper, 1972, 1974a,b; Rappoport et al., 1975),bovine (Goodman et al., 1970), chick (Piras and Piras, 1974; Sloboda et al., 1975), and porcine (Soifer e t al., 1972, 1975; Lagnado and Tan, 1975) brains; porcine thyroid (Rappoport e t al., 1972, 1975); chick embryonic muscle (Piras and Piras,
2 94
M. MARLENE HOSEY AND MARIANO TAO
1974); and He-La cells (Piras and Piras, 1974). Cyclic AMP markedly stimulates the endogenous phosphorylation of microtubules obtained from bovine (Goodman e t al., 1970) and chick brains (SIoboda et al., 1975), porcine thyroid, and rat brains (Rappoport et al., 1972, 1975), but not from chick embryonic muscle (Piras and Piras, 1974). I n rat brains, stimulation by cyclic AMP was not observed by Eipper (1974a). The reason for these discrepancies probably reflects the loss or inactivation of a regulatory subunit of a cyclic AMP-dependent protein kinase during the tubulin-purification procedure. Various methods have been used to purify tubulin. Differences in results have prompted Letterier et al. (1974) to suggest that the procedure used to purify microtubules may determine whether or not a cyclic AMPdependent protein kinase co-purifies with tubulin. Soifer (1975) has noted that, after lyophilization of tubulin, the associated kinase becomes independent of cyclic AMP. It is clear that the use of different procedures to purify tubulin leads to differences in the phosphorylation profile (Rappoport et al., 1975, 1976). However, in situations where the cyclic nucleotide is without effect, it was not reported whether addition of an exogenous cyclic AMP-dependent protein kinase can stimulate the phosphorylation of the microtubule preparations. Whether or not the tubulin dimer is phosphorylated is a very controversial issue. Eipper (1972, 1974a,b) demonstrated the presence of seryl phosphates in the P-subunit of the tubulin dimer, after either in uiuo or in vitro phosphorylation. In contrast, Rappoport et al. (1976) detected no 32Pin purified thyroid tubulin isolated from rats injected with 32Pand 14C-1eucine.Similarly, Sloboda et al. (1975) found no %P associated with the tubulin dimer isolated from brains of chicks injected with SzP-orthophosphate.These and other conflicting observations have prompted Rappoport et al. (1976) to propose that tubulin, in its native state, is not a substrate for the microtubule-associated protein kinase. Their evidence indicates that the procedure used to obtain purified tubulin determines whether or not tubulin will be phosphorylated, either in uivo or in uitro. Tubulin prepared by several cycles of polymerization-depolymerization (Shelanski et al., 1973) is not a good substrate for phosphorylation, whereas that prepared under more drastic conditions, such as vinblastine precipitation or the Weisenberg procedure (Weisenberg et al., 1968) may be phosphorylated (see Rappoportet al., 1976, for discussion). It may be assumed that tubulin becomes partially denatured during isolation, making it a better substrate for phosphorylation. Recent studies (Lagnado and Tan, 1975; Sloboda et al., 1975; Rap-
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
295
poport et al., 1975, 1976) have demonstrated that one or more nontubulin microtubule-associated proteins are effective substrates for the microtubule-associated cyclic AMP-dependent protein kinase. Sloboda et al. (1975) report that a 300,000-dalton protein (MAP,) and a 350,000-dalton component (MAP,) co-purify with chick brain tubulin through successive cycles of assembly-diassembly. The phosphorylation of MAP,, tubulin, and another protein of MW 73,000 is markedly stimulated by cyclic AMP, however, MAP, is 650 times more effective as a substrate than is tubulin. Cyclic GMP was without effect. Although only MAP, was phosphorylated in vitro, both MAP, and MAP, appeared to be phosphorylated in uiuo following intracerebral injection of 32P-~rth~phosphate. In these experiments, no label was found in the tubulin. Similar observations have been made by Lagnado and Tan (1975) with microtubule fractions prepared from slices of guinea pig cerebral cortex. Analysis of SDS-polyacrylamide gels of preparations preincubated with 32P-~rthophosphaterevealed that two-thirds of the radioactivity was found in minor proteins of high MW located near the origin of the gels. In addition, Rappoport e t al. (1975, 1976) have observed that the principal phosphorylated species of rat brain and thyroid microtubules prepared by the Shelanski procedure (Shelanski e t a1 ., 1973) is a component of similarly high MW. Thus it appears that minor high molecular proteins are the preferred substrates of the cyclic AMP-dependent protein kinase associated with purified microtubules from several species. The function of these proteins is not yet known; whether or not they represent dynein-like proteins is questionable (Rappoport e t al., 1976). However, they apparently account for a filamentous coating on assembled neurotubules (Dentler e t al., 1975). 3. MICROTUBULE-ASSOCIATEDPROTEINKINASE
The microtubule-associated protein kinase has a pH optimum of approximately 6.6 for casein (Soifer et al., 1972) and 6.8 for the microtubule-associated non-tubulin protein (Sloboda et a2 ., 1975, see below). The microtubule-associated kinase phosphorylates histone, protamine, and casein. The phosphorylation of histone and protamine are cyclic AMP-dependent, whereas the phosphorylation of casein is not (Soifer, 1975). Microtubule-associated protein kinase of chick muscle (Piras and Piras, 1974) can use GTP almost as effectively as ATP. This had also been reported for the porcine brain tubulin-associated kinase (Soifer et al., 1972). However, in a later report, Soifer et al. (1975) state that
2 96
M. MARLENE HOSEY A N D M A R I A N O T A O
GTP cannot replace ATP as the phosphoryl donor. Such inconsistencies are disturbing, but may suggest the presence of more than one kinase in microtubule preparations. It is well known that GTP is not used as a phosphoryl donor for cyclic AMP dependent kinases (Labrie et al., 1971; Tao and Hackett, 1973). Whether or not GTP can be used as a phosphoryl donor in other microtubule preparations has not been reported. Such a finding would seem to be of potential importance since certain procedures (Weisenberg et al., 1968) employ high concentrations of GTP (0.1-1.0 mM) to prevent aggregation during tubulin purification. Since the tubulin dimer binds GTP, Rappoport et al. (1976) have proposed that the phosphorylation of the tubulin dimer is an artificial event arising from the use of bound GTP as a substrate for a kinase that phosphorylates denatured tubulin. Considerable effort has gone into determining whether tubulin itself possesses protein kinase activity or whether the protein kinase represents a separate entity that is closely associated with the purified microtubule preparations. Soifer et al. (1972) reported that casein is phosphorylated by preparations possessing only the tubulin dimer and containing no other protein components, as seen in polyacrylamide gels stained with Coomassie Blue. More recent studies by Soifer et al. (1975) also failed to detect a separation of tubulin-associated protein kinase from tubulin. That tubulin retains an associated protein kinase through several treatments, such as precipitation by vinblastine, elution from DEAE-cellulose, and sedimentation with reassembled microtubules implies that the kinase is intrinsic to the tubulin itself (Soifer et al., 1975). However, evidence presented by Eipper (1974a)and Rappoport et al. (1975,1976) seems to clearly indicate that the kinase and tubulin are separate entities. The protein kinase and tubulin activities have been partially separated from each other by gel filtration, sucrose density gradient centrifugation, and vincristine precipitation (Eipper, 1974a).In gel filtration, protein kinase elutes at the void volume, while the tubulin dimer is retarded by the column. The protein kinase activity at the void volume is contaminated with aggregated tubulin but has a specific activity 80 times greater than that which trails into the tubulin dimer fraction. Rappoport et al. (1975) have raised the question whether the kinase that co-purifies with tubulin is simply a contaminant of the soluble brain cyclic AMPdependent protein kinase. However, based on a number of properties, such as K, for cyclic AMP and the inhibitory effects of calcium, it appears that the kinase associated with microtubules is distinct from the soluble cyclic AMP-dependent kinases.
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
297
4. PHYSIOLOGICAL SIGNIFICANCE OF TUBULINPHOSPHORYLATION Although it was thought at one time that phosphorylation of tubulin might be involved in the process of tubulin polymerization, recent studies indicate that this is not the case. Rappoport et al. (1976) have presented evidence indicating that tubulin is a poor substrate for phosphorylation and that phosphorylation does not affect the polymerization process. The rate and extent of tubulin polymerization is the same in untreated tubulin as in tubulin treated with ATP and cyclic AMP or with the heat-stable inhibitor of protein kinase (Rappoport et al., 1976). The function of the high molecular weight microtubule proteins is not well established. Some investigators feel that these proteins may be motile proteins which play a role in axoplasmic transport processes (Dentler et al., 1975; Sloboda et al., 1975). A regulatory role of cyclic AMP-dependent phosphorylation in such processes presents an attractive hypothesis worthy of further experimentation. If similar proteins are associated with microtubules of synaptic vesicles and other secretory cells, it may be that phosphorylation of these proteins plays a role in the regulation of release processes. H. Phosphorylation of Other Membranes
1. FAT CELLS In addition to the studies described above, membrane phosphorylation has also been conducted in a number of interesting but less extensively studied systems. Chang et d.(1974) have investigated the autophosphorylation of purified membrane fractions from rat adipocytes and have identified two phosphoproteins of about 22,000 and 16,000 daltons. The phosphorylation of these proteins is dependent on cyclic AMP. The same two proteins are phosphorylated when intact fat cells are exposed to low concentrations of ATP. Chang e t al. (1974) feel that the phosphorylation of these proteins correlates with the suppression of insulin-stimulated glucose transport. To further substantiate this interpretation, Chang e t al. (1974) extended their studies to guinea pig adipocytes, cells that are naturally resistant to the glucose oxidizing effects of insulin. Cyclic AMP had no effect on the phosphorylation of guinea pig fat cell membranes, and no phosphopeptides correspond to the two cyclic AMP-dependent phosphoproteins in rat adipocyte membranes. These findings correlate well with the insulin insensitiv-
298
M. MARLENE HOSEY AND MARIANO TAO
ity of guinea pig adipocytes and with the inability of ATP to inhibit the slight stimulation of glucose transport by insulin. Chang et ul. (1974) suggested that the specific membrane phosphorylation reactions may serve as a mechanism by which insulin regulates glucose transport through changes in the local concentration of cyclic AMP. However, a recent study by Avruch et al. (1976) indicates that this might not be the only action of insulin. They showed that insulin consistently and selectively stimulated the formation of a 123,000-dalton phosphopeptide found in the cytoplasm and endoplasmic reticulum of rat fat cells. The phosphorylation of this polypeptide is accentuated when fat cells are exposed to both insulin and epinephrine. On the other hand, insulin appears to inhibit the formation of other phosphopeptides whose formation is stimulated by epinephrine, presumably by lowering the cyclic AMP level in the cell or by preventing of the stimulation of adenylate cyclase by epinephrine. Thus, these data suggest that insulin inhibits the phosphorylation of certain polypeptides while it concurrently enhances the phosphorylation of a 123,000-dalton protein. 2. TOADBLADDER Antidiuretic hormone regulates the transport of sodium and water in toad bladders by a process mediated by cyclic AMP (Orloff and Handler, 1967). Exposure of toad bladder to either antidiuretic hormone or monobutyryl cyclic AMP caused a decrease in the phosphorylation of a specific protein (protein D, 50,000 daltons) present in a crude membrane fraction (DeLorenzo et al., 1973). DeLorenzo et ul. (1973) suggest that cyclic AMP acts at the mucosal surface of toad bladders to increase the effective permeability of the apical surface of the mucosal epithelium to Na+. In a subsequent report, DeLorenzo and Greengard (1973) show that cyclic AMP activates a membranebound phosphoprotein phosphatase, which causes a rapid dephosphorylation of protein D. The exact molecular mechanism of this activation process is unknown, but several possibilities have been suggested. It is possible that cyclic AMP might directly interact with the phosphatase, or that it could bind to protein D, making protein D more accessible to the phosphatase. On the other hand, cyclic AMP could activate a protein kinase, which in turn may activate and phosphorylate the phosphatase. 3. SECRETORY CELLS
Plasma membranes isolated from bovine anterior pituitaries contain an endogenous cyclic AMP-dependent protein kinase that catalyzes
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
299
the phosphorylation of endogenous and exogenous proteins (Lemay et al., 1974). Since cyclic AMP appears to play a role in the action of the hypothalamic-releasing hormones in the anterior pituitary gland (see monograph b y Robison et al., 1971), it suggests that protein phosphorylation may be involved in hormone secretion. Mouse mammary cells in organ culture incorporate 32Pinto 19 specific plasma membrane proteins (Majumder and Turkington, 1972). These proteins are also phosphorylated in purified plasma membranes by an endogenous cyclic AMP-independent protein kinase or by added exogenous cyclic AMP-dependent protein kinase. I n intact cells, the phosphorylation of these proteins is stimulated by the synergistic action of insulin and prolactin. This stimulatory effect has been attributed to the induction of cyclic AMP-dependent protein kinase by these hormones. However, this study is difficult to reconcile with the observation of Rillema (1976) who showed that agents known to elevate intracellular levels of cyclic A M P abolish the stimulatory effect of prolactin on casein synthesis. Therefore, cyclic AMP and cyclic AMP-dependent protein kinase are probably not mediators of prolactin action as suggested by Majumder and Turkington (1972). In contrast, Rillema (1976) implicated cyclic GMP as a possible mediator of prolactin action. In any event, the role of membrane phosphorylation in the secretory process of mammary cells remains unknown.
IV. MEMBRANE-BOUND PHOSPHOPROTEIN PHOSPHATASES A. Introduction
If phosphorylation is to serve as an effective regulatory process, mechanisms for dephosphorylation must exist within the system regulated. This suggests that phosphoprotein phosphatase may play an important regulatory role. Therefore, in a system regulated by phosphorylation-dephosphorylation, protein kinase provides the “on” signal while the phosphatase the “off” signal. In contrast to our fairly well-defined knowledge of the phosphorylating enzymes and their role in control processes, the phosphoprotein phosphatase has not been investigated extensively. In the study of phosphoprotein phosphatases, there are two questions that are of major interest. The first is the question of specificity and multiplicity. Since there are a number of protein kinases catalyzing the phosphorylation of various phosphoryl acceptors, it becomes obvious to inquire whether multiple species of phosphatase exist, each catalyzing the de-
300
M. MARLENE HOSEY AND MARIANO TAO
phosphorylation of a specific phosphoprotein, or a group of related phosphoproteins. Studies of the dephosphorylation of phosphorylase, phosphorylase kinase, glycogen synthetase, and histone seem to indicate that the dephosphorylation reaction is probably less specific than the phosphorylation reaction and that a single phosphatase can act on all these substrates (Nakai and Thomas, 1974; Killilea et d., 1976). The second question regarding the study of phosphatases pertains to the regulation of the phosphatase activity or the dephosphorylation process. Recent studies indicate that phosphatases may be regulated in a manner somewhat similar to that of the cyclic AMP-dependent protein kinase. Brandt et aZ. (1974, 1975) observed that liver phosphorylase phosphatase can exist in a less active form consisting of an inhibitor-enzyme complex. What regulates the interconversion of the phosphatase between active and inactive forms remains unknown, although Killilea et al. (1976) have suggested that this may be brought about by hormones. Huang and Glinsmann (1975) have recently implicated cyclic AMP-dependent protein kianse as a possible modulator of skeletal muscle phosphorylase phosphatase activity. Phosphoprotein phosphatase activities associated with membrane preparations have been demonstrated in a number of systems. Some of these activities have been solubilized and partially characterized. However, in general, these enzymes have not been extensively investigated. We know little about the substrate specificities of these phosphatases and less about their regulation. B. Erythrocytes
Rabbit erythrocyte membrane preparations contain a nonspecific acid phosphatase (Berry and Hochstein, 1969). The enzyme activity, which appears to be localized on the outer surface of the cell membranes, is markedly inhibited by NaF. A membrane-bound phosphatase has also been reported in human, porcine, and bovine erythrocytes (Heller and Hanahan, 1972). Both phosphatases of human and porcine membranes are activated by K+; Ca2+inhibits this activation. The study of these phosphatases in general employs p-nitrophenyl phosphate as substrate. Whether phosphoproteins are also substrates of these enzymes remains unknown. The possibility that erythrocyte membranes may contain a phosphoprotein phosphatase is implicit in the kinetics of membrane autophosphorylation (Hosey and Tao, 1976a).An examination of the time course of membrane phosphorylation shows that a maximum is reached, whereupon there is a decrease in the amount of phosphate incorporated. The latter half of the reac-
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
301
tion is probably due to a cessation of further incorporation of phosphatase resulting from a depletion of phosphoryl donor and the ensuing hydrolysis of phosphoprotein linkages by the action of phosphatase. Phosphoprotein phosphatase activities have been demonstrated in hemolysates of rabbit reticulocytes (Lightfoot et al., 1975). At least two phosphatase activities were separated by chromatography on DEAE-cellulose. These phosphatases catalyze the dephosphorylation of 40s ribosomal proteins previously phosphorylated by cyclic AMP-dependent protein kinase. The possibility that membranebound phosphoproteins may also serve as substrates for these soluble phosphatases has not been investigated. C. Corpus Luteum
Phosphatase activity towards phosphoproteins has been demonstrated in two fractions, I and 11, of bovine corpus luteum plasma membranes (Azhar and Menon, 1975~).An examination of the substrate specificities of these phosphatases showed that the fraction I plasma membrane enzyme was more active towards phosphoprotamine than phosphohistones or phosphocasein. In contrast, the fraction I1 enzyme exhibits little preference between phosphoprotamine and phosphohistones and dephosphorylates both substrates with equal efficiency. That these phosphatases may play a role in plasma membrane dephosphorylation is suggested by the observation that maximally phosphorylated membranes are partially dephosphorylated upon further incubation. Solubilization of these phosphatases has been achieved by extracting the plasma membrane with sodium deoxycholate (0.1%),Lubrol-PX (0.2%),or Triton X-100 (0.1%). Phosphoprotein phosphatase activities of fraction I and I1 sediment as two distinct molecular species at 6.7 S and 4.8 S, respectively. A comparison of the kinetic properties and substrate specificities of the membrane-bound and solubilized enzymes has not been carried out. D. Cardiac Muscle
Recently, Tada e t al. (1975b) showed that dog cardiac microsomes, consisting mainly of fragmented sarcoplasmic reticulum, contained phosphoprotein phosphatase activity. The enzyme catalyzes the dephosphorylation of phosphohistones and may be involved in the dephosphorylation of “phospholamban.” As discussed earlier, phospholamban has been implicated in the mediation of the cyclic AMP-
302
M. MARLENE HOSM AND MARIANO TAO
dependent protein kinase-stimulated Ca2+-ATPaseactivity and Ca2+ transport process in cardiac muscle. Dephosphorylation of phospholamban by the phosphatase could reverse these effects of cyclic AMP-dependent protein kinase. Conceivably, the same phosphatase could also catalyze the dephosphorylation of the 95,000-dalton phosphoprotein described by Schwartz et al. (1976). Hence, the phosphatase, together with cyclic AMP-dependent protein kinase and/or phosphorylase kinase could constitute an “off-on” switch of the Ca2+ pump. In addition to the membrane-bound phosphatase, two soluble phosphoprotein phosphatases have been isolated from extracts of dog myocardium (Tada et al., 1975b). The soluble phosphatases also catalyze the dephosphorylation of phosphohistones and phospholamban. Similarly, Nakai and Thomas (1974) have isolated from bovine heart a relatively nonspecific soluble phosphoprotein phosphatase with activity towards glycogen synthetase, phosphorylase a, phosphorylase kinase, phosphohistone, and phosphocasein. It is possible that these soluble phosphatases may also play a role in the dephosphorylation of membrane phosphoproteins. E. Synaptoromer
The phosphorylation and dephosphorylation of a protein found in soluble and particulate fractions of many vertebrate tissues is regulated by cyclic AMP (Malkinson et al., 1975). Synaptic membranes contain a similar molecular weight protein that can undergo endogenous phosphorylation and dephosphorylation (Maeno et d., 1975; Ueda et al., 1975). Both reactions are regulated by cyclic AMP. The phosphorylation-dephosphorylation system appeared to be intact when extracted from the cell membranes with either Triton X-100 or NH,CI. Based on kinetic data, Ueda et al. (1975) suggested that the 49,000-dalton protein, its cyclic AMP-dependent protein kinase, and its cyclic AMP-dependent phosphatase may exist as a complex in synaptic membranes or in solubilized form. In a toad bladder membrane preparation, cyclic AMP similarly caused a decrease in the phosphorylation of a specific membrane protein (protein D; DeLorenzo and Greengard, 1973). Incubation of =P-labeled membranes in the presence of cyclic AMP caused an increase in the rate of hydrolysis of phosphate from this specific protein. On the basis of these studies, Maeno et al. (1975) suggested that the dual action of cyclic AMP is stimulating phosphorylation and dephosphorylation may be a mechanism for generating a transient regulatory signal.
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
303
F. Conclusion
From the studies described above, it is clear that a mechanism for dephosphorylation exists within certain cell membranes. Although it is attractive to hypothesize that membrane-bound phosphatases are an integral part of the regulatory system, the significance of these enzymes in the regeneration of membrane proteins into their nonphosphorylated state remains to be established. However, there is no compelling reason to assume that the dephosphorylation of membranebound phosphoproteins must also involve a membrane-associated phosphatase. It is equally possible for the cytosolic phosphatases to catalyze the dephosphorylation reaction. I n view of the many regulatory mechanisms operating in the phosphorylation reaction, the possibility that dephosphorylation reaction may be similarly regulated seems likely. However, with the exception of phosphorylase phosphatase (Huang and Glinsmann, 1975; Brandt et al., 1975),the regulation of other phosphatases, in particular, those associated with the cell membranes, have not been explored extensively. In addition, the nature of the effect of cyclic AMP on the toad bladder and synaptic membrane phosphatases requires further characterization. V.
CONCLUDING REMARKS
A considerable amount of information basic to the understanding of the significance of membrane phosphorylation has emerged from the analysis of different membrane autophosphorylating systems. These include the nature and the regulation of the protein kinases and the membrane protein substrates of these enzymes. Fragmentary information regarding the physiological implication of membrane phosphorylation is also available. The evidence suggesting that Ca2+uptake in cardiac and skeletal muscle is regulated by protein phosphorylation seems convincing. It is also fairly certain that protein phosphorylation plays a role in the light and dark adaptation of the retina. In other systems, such as erythrocytes, synaptosomes, myelin, etc., the functional implications of phosphorylation have not been clearly defined. Although the studies reviewed in this article represent an impressive array of data dealing with membrane phosphorylation, it is clear that the data have not provided a detailed description of the molecular mechanism of the role of membrane phosphorylation. At present, a direct causal relationship has not been established between the phosphorylation of a polypeptide and the alteration of a specific function. Therefore, a shift in emphasis toward the identification of the biologi-
304
M. MARLENE HOSEY AND MARIANO TAO
cal activities of the phosphoproteins is essential. This entails the exceedingly difficult tasks of isolating and purifying the various phosphoproteins and identifying their functional activities. Finally, a physiologically active system must be reconstituted from the individual components. The possibility that the phosphoprotein may not contain any enzymatic activity but may serve as a regulator of other enzymes must also be considered. In this regard, a knowledge of the topographic relationship between the phosphoprotein and its neighboring proteins seems relevant. Previous studies have focused on the effects of protein phosphorylation on transport processes. In view of the diverse receptor activities associated with the plasma membranes, investigations into the possible modifications of these receptor activities by phosphorylation might yield new insights regarding other aspects of membrane phosphorylation. Although our knowledge of membrane structure and of the function of the various components is limited, with the rapid advances in membrane technology, the elucidation of the precise role of protein phosphorylation in membrane processes seems feasible and may be forthcoming. ACKNOWLEDGMENTS This work was supported in part by the American Cancer Society (BC-65C) and by the National Institutes of Health (5 R01 CA17036-02,-03). M. Hosey is a recipient of a National Research Service Fellowship (1-F32-AM 05077-01,-02) from the National Institutes of Health. M. Tao is an established Investigator of the American Heart Association. REFERENCES Adelstein, R. S., and Conti, M. A. (1975). Phosphorylation of platelet myosin increases actin activated myosin ATPase activity. Nature (London)256,597-598. Andersson, R., and Nilsson, K. (1972). Cyclic AMP and calcium in relaxation in intestinal smooth muscle. Nature (London),New Biol. 238, 119-120. Andersson, R., Nilsson, K., Wikberg, J., Johnsson, S., Mohme-Lundholm, E., and Lundholm, L. (1975). Cyclic nucleotides and the contraction of smooth muscle.Ado. C y clic Nucleotide Res. 5,491-518. Andersson, R. G. G. (1972). Cyclic AMP and calcium ions in mechanical and metabolic responses of smooth muscles; influence of some hormones and dmgs. Acta Physiol. Scand. S u p p l . 382, 1-59. Andrew, C. G., Roses, A. D., Almon, R. R., and Appel, S. H. (1973). Phosphorylation of muscle membranes: Identification of a membrane-bound protein kinase. Science 182,927-929. Andrew, C. G., Almon, R. R., and Appel, S. H. (1975). Macromolecular characterization of muscle membranes. Endogenous protein kinase and phosphorylated protein substrate from normal and denervated muscle. J . Biol. C h e m . 250, 3972-3980.
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
305
Ashby, C. D., and Walsh, D. A. (1972).Characterization of the interaction of a protein kinase. J . Biol. Chem. inhibitor with adenosine 3’,5‘-monophosphate-dependent 247,6637-6642. Aurbach, G . D. (1975).Beta-adrenergic receptors, cyclic AMP and ion transport in the avian erythrocyte. Adu. Cyclic Nucleotide Res. 5, 117-132. Avissar, N., deVries, A., Ben-Shaul, Y., and Cohen, I. (1975).Actin-activated ATPase from human erythrocytes. Biochim. Biophys. Acta 375,35-43. Avruch, J., and Fairbanks, G. (1974).Phosphorylation of endogenous substrates by erythrocyte membrane protein kinases. I. A monovalent cation-stimulated reaction. Biochemistry 13,5507-5514. Avruch, J., Leone, G. R., and Martin, D. B. (1976).Effects ofepinephrine and insulin on phosphopeptide metabolism in adipocytes. J . Biol. Chem. 251, 1511-1515. Azhar, S., and Menon, K. M. J. (1975a).Adenosine 3’,5‘-monophosphate dependent phosphorylation of ribosome and ribosomal subunits from bovine corpus luteum. Biochim. Biophys. Acta. 392,64-74. Azhar, S., and Menon, K. M. J. (197513).Adenosine 3’ : 5‘monophosphate-dependent and plasma-membrane-associated protein kinase from bovine corpus luteum. Solubilization and properties of solubilized enzyme. Eur. J. Biochem. 58,105-115. Azhar, S.,and Menon, K. M. J. (1975~). Solubilization and characterization of phosphoprotein phosphatase(s) from bovine corpus-luteum plasma membranes. Eur.]. Biochem. 55,487-495. Bar, H. -P. (1974).Cyclic nucleotides and smooth muscle. Ado. Cyclic Nucleotide Res. 4,195-237. Barrenger, J. A., Geiger, P. J., Huzino, A,, and Bessnian, S. P. (1972).Isozymes of phenylalanine hydroxylase. Science 175,903-905. Beavo, J. A., Bechtel, P. J., and Krebs, E. G. (1974). Preparation of homogeneous cyclic AMP-dependent protein kinase(s) and its subunits from rabbit skeletal muscle. Methods Enzymol. 38C, 299-308. Berry, D.H., and Hochstein, P. (1969).Membrane acid phosphatase in rabbit erythrocytes. Arch. Biochem. Biophys. 131, 170-176. Bilezikian, J. P., and Aurbach, G . D. (1974). The effects ofnucleotides on the expression of/?-adrenergic adenylate cyclase activity in membranes from turkey erythrocytes. J . Biol. Chem. 249,157-161. Bingham, E. W., and Farrell, H. M., Jr. (1974).Casein kinase from the golgi apparatus of lactating mammary gland. J. Biol. Chem. 249,3647-3651. Bloom, F. E., Siggins, G. R., Hoffer, B. J,, Segal, M., and Oliver, A. P. (1975).Cyclic nucleotides in the central synaptic actions of catecholamines. Ado. Cyclic Nucleotide Res. 5,603-618. Blum, H.E.,Pocinwong, S., and Fischer, E. H. (1974).Phosphorylation of a dogfish skeletal muscle protein related to parvalbumin. In “Metabolic Interconversion of Enzymes” (E. H. Fischer, E. G. Krebs, H. Neurath, and E. R. Stadtman, eds.), pp. 197-208. Springer-Verlag, Berlin. Bownds, D., Brodie, A., Robinson, W. E., Palmer, D., Miller, J., and Shedlovsky, A. (1974).Physiology and enzymology of frog photoreceptor membranes. E x p . Eye Res. 18,253-269. Brandt, H., Killea, S. D., and Lee, E. Y. C. (1974). Activation of phosphorylase phosphatase by a novel procedure: Evidence for a regulatory mechanism involving the release of a catalytic subunit from enzyme-inhibitor complex(es) of higher molecular weight. Biochem. Biophys. Res. Commrrn. 61,598-604. Brandt, H., Lee, E . Y. C., and Killea, S. D. (1975).A protein inhibitor of rabbit liver phosphorylase phosphatase. Biochem. Bioph ys. Res. Commrrn. 63, 950-956.
306
M. MARLENE HOSEY AND MARIANO TAO
Braughler, J. M., and Corder, C. N. (1976). Non-linearity of human renal (Na+ + K+)ATPase: Elimination by purification and correlation with cyclic AMP-dependent protein kinase activity. Biochem. Biophys. Res. Commun. 71,408-412. Brostrom, M. A,, Reimann, E. M., Walsh, D. A., and Krebs, E. G. (1970). A cyclic 3’,5’-AMP-stimulated protein kinase from cardiac muscle. Adu. Enzyme Regul. 8, 191-203. Burnett, G., and Kennedy, E. P. (1954). The enzymatic phosphorylation of pr0teins.J. Biol. Chem. 21,969-980. Bylund, D. B., and Krebs, E. G. (1975). Effect of denaturation on the susceptibility of proteins to enzymatic phosphorylation. J. Biol. Chem. 250, 6355-6361. Cabantchik, Z. I., and Rothstein, A. (1974). Membrane proteins related to anion permeability of human red blood cells. I. Localization of disulfonic stilbene binding sites in proteins involved in permeati0n.J. Menibr. Biol. 15,207-226. Carnegie, P. R., Kemp, B. E., Dunkley, P. R., and Murray, A. W. (1973).Phosphorylation of myelin basic protein by an adenosine 3’:5‘-cyclic monophosphate-dependent protein kinase. Biochem. J. 135, 569-572. Carnegie,P. R., Dunkley, P. R., Kemp, B. E., and Murray,A. W. (1974). Phosphorylation of selected serine and threonine residues in myelin basic protein by endogenous and exogenous protein kinases. Nature (London) 249, 147-149. Casnellie, J. E., and Greengard, P. (1974). Guanosine 3’ :5’-monophosphate-dependent phosphorylation of endogenous substrate proteins in membranes of mammalian smooth muscle. Proc. Natl. Acad. Sci. LISA. 71, 1891-1895. Cawthon, M. L., Bitte, L. F., Krystosek, A., and Kabat, D. (1974).Effect of cyclic adenosine 3’ :5‘-monophosphate on ribosomal protein phosphorylation in reticu1ocytes.J. Biol. Chem. 249,275-278. Chader, G. J., Fletcher, R. T., and Krishna, G. (1975).Light-induced phosphorylation of rod outer segments by guanosine triphosphate. Biochem. Biophys. Res. Commun. 64,535-538. Chader, G. J., Fletcher, R. T., O’Brien, P. J., and Krishna, G. (1976). Differential phosphorylation by GTP and ATP in isolated rod outer segments of the retina. Biochemistry 15, 1615-1620. Chang, K. J., Marcus, N. A., and Cuatrecasas, P. (1974). Cyclic adenosine monophosphate dependent phosphorylation of specific fat cell membrane proteins by an endogenous membrane-bound protein kinase. Possible involvement in the regulation of insulin-stimulated glucose transport. J. Biol. Chem. 249,6854-6865. Chen, L. J., and Walsh, D. A. (1971). Multiple forms of hepatic adenosine 3’:5’monophosphate dependent protein kinase. Biochemistry 10,36143621. Chien, S . , Usami, S., and Bertles, J. F. (1970). Abnormal rheology of oxygenated blood in sickle cell anemia.J. Clin. Invest. 49,623-634. Cohen, P. (1973). The subunit structure of rabbit-skeletal-muscle phosphorylase kinase, and the molecular basis of its activation reactions. Eur. J. Biochem. 34, 1-14. Cole, H. A., and Perry, V. S . (1975). The phosphorylation of troponin I from cardiac muscle. Biochem. J. 149,525-533. Constantopoulos, A., and Najjar, V. (1973). The activation of adenylate cyclase: 11. The presence of (A) adenylate cyclase in a phospho (inhibited) form (B) a dephospho (activated) form with a cyclic adenylate stimulated membrane protein kinase. Biochem. Biophys. Res. Commun. 53,794-799. Corbin, J. D., and Krebs, E. G. (1969). A cyclic AMP-stimulated protein kinase in adipose tissue. Biochem. Biophys. Res. Commun. 36,328336. Corbin, J. D., Reimann, E. M., Walsh, D. A., and Krebs, E. C. (1970). Activation of adipose tissue lipase by skeletal muscle cyclic adenosine 3’,5‘-monophosphatestimulated protein kinase.J. Biol. Chem. 245,4849-4851.
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
307
Corbin, J. D., Brostrom, C. D., King, C. A., and Krebs, E. G. (1972). Studies on the adenosine 3',5'-monophosphate-dependentprotein kinases of rabbit skeletal muscle. J . Biol. Chem. 247, 7790-7798. Corbin, J. D., Keely, S. L., and Park, C. R. (1975). The distribution and dissociation of cyclic adenosine 3' :5'-monophosphate-dependent protein kinases in adipose, cardiac, and other tissues. ]. Biol. Chem. 250, 218-225. Costa, M., Gerner, E. W., and Russell, D. H. (1976). Cell cycle-specific activity oftype I and type I1 cyclic adenosine 3' :5'-monophosphate dependent protein kinases in Chinese hamster ovary cells. J . Biol. Chem. 251, 3313-3319. Daile, P., Carnegie, P. R., and Young, J. D. (1975). Synthetic substrate for cyclic AMP-dependent protein kinase. Nature (London) 257,416-418. Davoren, P. R., and Sutherland, E. W. (1963). The effects of 1-epinephrine and other agents on the synthesis and release of adenosine 3',5'-phosphate by whole pigeon erythrocytes. J. Biol. Chem. 238,3009-3015. Deibler, G. E., Martenson, R. E., Kramer, A. J., Kies, M. W., and Miyamoto, E. (1975). The contribution of phosphorylation and loss of COOH-terminal arginine to the microheterogeneity of myelin basic protein. ]. Biol. Chem. 250,7931-7938. DeLange, R. J., Kemp, R. G., Riley, W. D., Cooper, R. A., and Krebs, E. G . (1968).Activation of skeletal muscle phosphorylase kinase by adenosine triphosphate and adenosine 3',5'-monophosphate. J. Biol. Chem. 243,2200-2208. DeLorenzo, R. J. (1976). Calcium-dependent phosphorylation of specific synaptosomal fraction proteins: Possible role of phosphoproteins in mediating neurotransmitter release. Biochem. Biophys. Res. Commun. 71,590-597. DeLorenzo, R. J., and Glaser, G. H. (1976). Effect of diphenyhydantoin on the endogenous phosphorylation of brain protein. Bruin Res. 105,381-386. DeLorenzo, R. J., and Greengard, P. (1973). Activation by adenosine 3':5'monophosphate of a membrane-bound phosphoprotein phosphatase from toad bladder. Proc. Natl. Acud. Sci. U.S.A. 70, 1831-1835. DeLorenzo, R. J., Walton, K. G., Curran, P. F., and Greengard, P. (1973). Regulation of phosphorylation of a specific protein in toad-bladder membrane by antidiuretic hormone and cyclic AMP, and its possible relationship to membrane permeability changes. Proc. Natl. Acad. Sci. U.S.A. 70,880-884. Demaille, J., Dutruge, E., Baron, G., PechCre, J. F., and Fisher, E. H. (1975). A reappraisal ofthe relationship ofphosphate-acceptor protein to parvalbumins. Biochem. Biophys. Res. Commun. 67, 1034-1040. Dentler, W. L., Granett, S., and Rosenbaum, J. L. (1975). Ultrastructural localization of the high molecular weight proteins associated with in uitro-assembled brain micrctubules. J. Cell Biol. 65,237-241. Donnelly, T. E., Jr., Kuo, J. F., Reyes, P. L., Liu, Y. P., and Greengard, P. (1973a). Protein kinase modulator from lobster tail muscle. I. Stimulatory and inhibitory effects of the modulators on the phosphorylation of substrate proteins by guanosine 3',5'-monophosphatedependent and adenosine 3',5'-monophosphate-dependent protein kinases. J . Biol. Chem. 248, 190-198. Donnelly, T. E., Jr.. Kuo, J. F., Miyamoto, E., and Greengard, P. (1973b). Protein kinase modulator from lobster tail muscle. 11. Effects of the modulator on holoenzyme and catalytic subunit of guanosine 3',5'-monophosphate-dependent and adenosine 3',5'-monophosphate dependent protein kinases. J . Biol. Chem. 248, 199-203. Dldskeland, S . O., and Ueland, P. M. (1975). A CAMPreceptor from mouse liver cytosol whose binding capacity is enhanced by M$+-ATP. Biochem. Biophys. Res. Commun. 66,606-613. Downer, D. N., Rogers, H. W., and Randall, C. C. (1973). Endogenous protein kinase and phosphate acceptor proteins in vaccinia virus. Virology 52, 13-21.
308
M. MARLENE HOSEY AND MARIANO TAO
D u e , M., and Schwarz, V. (1973). Cyclic AMP stimulated phosphorylation of erythrocyte membranes from cystic fibrosis and control subjects. Clin. Chem. Acta 49, 397-399. Duffy, M., and Schwarz, V. (1974). Hormone and adenosine 3‘,5’ cyclic monophosphate stimulated phosphorylation of human erythrocyte membranes. Biochern. Pharmacol. 23,2464-2466. Ebashi, S., and Endo, M. (1968).Calcium ion and muscular contraction. Progr. Biophys. Mol. Biol. 18, 123-183. Eil, C., and Wool, I. G. (1973a). Phosphorylation of liver ribosomal proteins. Characterization of the protein kinae reaction and studies of the structure of phosphorylated ribosomes.]. Biol. Chem. 248,5122-5129. Eil, C., and Wool, I. R. (1973b). Function of phosphorylated ribosomes. The activity of ribosomal subunits phosphorylated in vitro by protein kinase.]. Biol. Chem. 248, 5130-5 136. Eipper, B. A. (1972). Rat brain microtubule protein: Purification and determination of covalently bound phosphate and carbohydrate. Proc. Natl. Acad. Sn’. U.S.A. 69, 2283-2287. Eipper, B. A. (1974a). Rat brain tubulin and protein kinase activity.J. Biol. Chem. 249, 1398-1406. Eipper, B. A. (1974b). Properties of rat brain tubu1in.J. Biol. Chem. 249,1407-1416. Elgsaeter, A., and Branton, D. (1974). Intramembrane particle aggregation in erythrocyte ghosts. I. The effects of protein removal. J. Cell B i d . 63, 1018-1030. Elgsaeter, A., Shotton, D., and Branton, D. (1973).Control ofprotein distribution in the erythrocyte membrane.J. Cell Biol. 59,89a. Elgsaeter, A., Shotton, D. M., and Branton, D. (1976). Intramembrane particle aggregation in erythrocyte ghosts. 111. The influence of spectrin aggregation. Biochim. Biophys. Acta 426, 101-122. England, P. J., Stull, J. T., Huang, T. S., and Krebs, E. G. (1974). Phosphorylation and dephosphorylation of skeletal muscle troponin. In “Metabolic Interconversion of Enzymes” (E. H. Fischer, E. G. Krebs, H. Neurath, and E. R. Stadtman, eds.), pp. 175-184. Springer-Verlag, Berlin. Engstrom, L., Berglund, L., Bergstrom, G., Hjelmquist, G., and Ljunstrom, 0. (1974). Regulatory phosphorylation of purified pig liver pyruvate kinase. I n “Lipmann Symposium: Energy, Regulation and Biosynthesis in Molecular Biology” (D. Richter, e d ) , pp. 192-204. de Gruyter, Berlin. Erlichman, J., Rubin, C. S., and Rosen, 0. M. (1973). Physical properties of a purified cyclic adenosine 3’ :5’-monophosphate-dependent protein kinase from bovine heart musc1e.J. Biol. Chem. 248, 7607-7609. Erlichman, J., Rosenfeld, R., and Rosen, 0. M. (1974). Phosphorylation of cyclic adenosine 3‘ :5’-monophosphatedependent protein kinase from bovine cardiac muscle. 1.B i d . Chem. 249,5000-5003. Exton, J. H., and Park, C. R. (1968). Control of gluconeogenesis in liver. 11. Effects of glucagon, catecholamines and adenosine 3‘,5’-monophosphate on gluconeogenesis in the perfused rat 1iver.J. B i d . Chem. 243,4189-4196. Fairbanks, G., and Avruch, J. (1974). Phosphorylation of endogenous substrates by erythrocyte membrane protein kinases. 11. Cyclic adenosine monophosphatestimulated reactions. Biochemistry 13,5514-5521. Fairbanks, G., Steck, T. L., and Wallach, D. F. H. (1971). Electrophoretic analysis of the major polypeptides of the human erythrocyte membrane. Biochemistry 10, 2606-26 17. Fedelesova, M., and Ziegelhoffer, A. (1975). Enhanced calcium accumulation related to
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
309
increased protein phosphorylation in cardiac sarcoplasmic reticulum induced by cyclic 3’,5‘-AMP or isoproterenol. Experientia 31,518-520. Frank, R. N., and Buzney, S . M. (1975). Mechanism and specificity of rhodopsin phosphorylation. Biochemistry 14,5110-5117. Frank, R. N., Cavanagh, H. D., and Kenyon, K. R. (1973). Light-stimulated phosphorylation of bovine visual pigments by adenosine triphosphate. J. Biol. Chem. 248, 596-609. Friedman, D. L., and Larner, J. (1963). Studies on UDPGa-glucan transglucosylase. 111. Interconversion of t w o forms of muscle UDPG-cu-glucan transglucosylase by a phosphorylation-dephosphorylation reaction sequence. Biochemistry 2,669-675. Furthmayr, H., Kahane, I., and Marchesi, V. T. (1976). Isolation of the major intrinsic transmembrane protein of the human erythrocyte membrane. J , Membr. Biol. 26, 173-187. Garbers, D. L., First, N. L., and Lardy, H. A. (1973). Properties of adenosine 3’,5’monophosphate-dependent protein kinases isolated from bovine epididymal spermatozoa. /. Biol. Chem. 248,875-879. Gibson, K., and Newcombe, M. (1975). Myocardial microsomal protein kinase from the rat and the rabbit. Biochem. SOC. Trans. 3,547-548. Gill, G. N., and Garren, L. D. (1970). A cyclic3’,5‘-adenosine monophosphate dependent protein kinase from adrenal cortex: Comparison with a cyclic AMP binding protein. Biochem. Biophys. Res. Commun. 39,335-343. Goldstein, J. L., and Hasty, M. A. (1973). Phosvitin kinase from the liver of the rooster. Purification and partial characterization. J. Biol. Chem. 248,6300-6307. Goodman, D. B. P., Rasmussen, H., DiBella, F., and Guthrow, C. E. (1970). Cyclic phosphorylation of isolated neurotuadenosine 3’ :5’-monophosphate-stimulated bule subunits. Proc. Natl. Acad. Sci. U S A . 67, 652-659. Gorecka, A., Aksoy, M. O., and Hartshorne, D. J. (1976).The effect of phosphorylation of gizzard myosin on actin activation. Biochem. Biophys. Res. Commun. 71, 325331. Greengard, P. (1975). Cyclic nucleotides, protein phosphorylation, and neuronal f u n c tion. Ado. Cyclic Nucleotide Res. 5,585-601. Greenquist, A. C., and Shohet, S. B. (1974). Defective protein phosphorylation in membranes of hereditary spherocytosis erythrocytes. FEES Lett. 48,133-135. Guidotti, G. (1972). Membrane proteins. Ann. Rev. Biochem. 41,731-752. Guthrow, C. E., Allen, J. E., and Rasmussen, H. (1972). Phosphorylation of an endogenous membrane protein by an endogenous membrane-associated cyclic adenosine 3’,5’-monophosphatedependentprotein kinase in human erythrocytes. J . Biol. Chem. 247,8145-8153. Guthrow, C. E., Rasmussen, H., Brunswick, D. J., and Cooperman, B. S. (1973).Specific photoaffinity labeling of the adenosine 3’:5’-monophosphate receptor in intact ghosts from human erythrocytes. Proc. Natl. Acad. Sci. U.SA. 70,3344-3346. Haddox, M. K., Newton, N. E., Hartle, D. K., and Goldberg, N. D. (1972).ATP(M$+) induced inhibition of cyclic AMP reactivity with a skeletal muscle protein kinase. Biochem. Biophys. Res. Commun. 47,653-661. Haley, B. E. (1975).Photoaffinity labeling of adenosine 3’,5’-monophosphate binding sites of human red cell membranes. Biochemistry 14,3852-3857. Harigaya, S., and Schwartz, A. (1969). Rate of calcium binding and uptake in normal animal and failing human cardiac muscle. Membrane vesicles (relaxing system) and mitochondria. Circ. Res. 25,781-794. Hayakawa, T., Perkins, J. P., and Krebs, E. G. (1973). Studies on the subunit structure of rabbit skeletal muscle phosphorylase kinase. Biochemistry 12,574-580.
310
M. MARLENE HOSEY AND MARIANO TAO
Heller, M., and Hanahan, D. J. (1972). Erythrocyte membrane-bound enzymes. ATPase, phosphatase, and adenylate kinase in human, bovine, and porcine erythrccytes. Biochim. Biophys. Acta 255,239-250. Higgins, W. J., and Greenberg, M. J. (1974). Intracellular actions of 5hydroxytryptamine on the bivalve myocardium. I. Cyclic nucleotide-dependent protein kinases and microsomal calcium uptake. J . E x p . 2001.190,305-316. Ho, M. K., and Guidotti, G. (1975).A membrane protein from human erythrocytes involved in anion exchange. ]. B i d . Chern. 250,675-683. Hofmann, F.,Beavo, J. A., Bechtel, P. J., and Krebs, E. G. (1975).Comparison of adenosine 3‘:5’-monophosphate-dependent protein kinases from rabbit skeletal and bovine heart muscle.]. Biol. Chem. 250,7795-7801. Hosey, M. M., and Tao, M. (1975).Effect of 5‘-guanylylimidodiphosphate on prostaglandin- and fluoride-sensitive adenylate cyclase of rabbit red blood cells. Biochem. Biophys. Res. Commun. 64,1263-1269. Hosey, M. M., and Tao, M. (1976a). An analysis ofthe autophosphorylation of rabbit and human erythrocyte membranes. Biochemistry 15,1561-1568. Hosey, M. M., and Tao, M. (1976b).Altered erythrocyte membrane phosphorylation in sickle cell disease. Nature (London) 263,424-425. Hosey, M. M., and Tao, M. (1977).Phosphorylation of rabbit and human erythrocyte membranes by soluble adenosine 3’:5’-monophosphate-dependent and independent protein kinases.]. Biol. Chem. 252, 102-109. Huang, F. L., and Glinsmann, W. H. (1975).Inactivation of rabbit muscle phosphorylase phosphates by cyclic AMP-dependent kinase. Proc. Natl. Acad. Sci. U.S.A. 72,
3004-3008. Huang, K.P., Huang, F. L., Glinsmann, W. H., and Robinson, J. C. (1975).Regulation of glycogen synthetase activity by two kinases. Biochem. Biophys. Res. Commun. 65,
1163-1 169. Huang, L. C., and Huang, C. H. (1975).Rabbit skeletal muscle protein kinase. Conversion from CAMP dependent to independent form by chemical pertubations. Biochemistry 14, 18-24. Hui, C. -W., Drummond, M., and Drummond, G. (1976).Calcium accumulation and cyclic AMP-stimulated phosphorylation in plasma membrane-enriched preparations of myocardium. Arch. Biochem. Biophys. 173,415-427. Huttunen, J. K., and Steinberg, D. (1971).Activation and phosphorylation of purified adipose tissue hormone-sensitive lipase by cyclic AMP-dependent protein kinase. Biochim. Biophys. Acta 239,411-427. Huttunen, J. K.,Steinberg, D., and Mayer, S. E. (1970).ATP-dependent and cyclic AMP-dependent activation of rat adipose tissue lipase b y protein kinase from skeletal muscle. Proc. Natl. Acad. Sci. U S A . 67,290-295. Jacob, H., Amsden, T., and White, J. (1972).Membrane microfilaments oferythrocytes: Alteration in intact cells reproduces the hereditary spherocytosis syndrome. Proc. Natl. Acad. Sci. U S A . 69,471-474. Johnson, E. M., Maeno, H., and Greengard, P. (1971).Phosphorylation of endogenous proteins of rat brain by cyclic adenosine 3‘,5‘-monophospahte-dependentprotein kinase. J . Biol. Chem. 246,7731-7739. Jungmann, R. A., Hiestand, P. C., and Schweppe, J. S . (1974).Adenosine 3‘:5’monophosphate-dependent protein kinase and the stimulation of ovarian nuclear ribonucleic acid polymerase activities. J . Blol. Chem. 249, 5444-5451. Kahlenberg, A. (1976).Partial purification of a membrane protein from human erythrocytes involved in glucose transport. J . Biol. Chem. 251,1582-1590. Kant, J. A., and Steck, T. L. (1973).Adenosine 3’,5’-monophosphate binds only to t h e
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
31 1
inner surface of human erythrocyte membranes. Biochem. Biophys. Res. Commun. 54, 116-122. Kato, K., and Bishop, J. S. (1972).Glycogen synthetase-D phosphatase. I. Some new properties of the partially purified enzyme from rabbit skeletal muscle. J. Biol. Chem. 247,7420-7429. Katz, A. M. (1970).The contractile proteins of the heart. Physiol. Reo. 50,63-158. Katz, A. M., and Repke, D. I. (1967).Quantitative aspects of dog cardiac microsomal calcium binding and calcium uptake. Circ. Res. 21, 153-162. Katz, A. M., and Repke, D. I. (1973).Calcium-membrane interactions in the myocardium: Effect of ouabain, epinephrine, and 3‘,5’-cyclic adenosine monophosphate. Amer. J. Cardiol. 31, 193-201. Katz, A. M., Tada, M., and Kirchberger, M. A. (1975). Control of calcium transport in the myocardium by the cyclic AMP-protein kinase system. Ado. Cyclic Nucleotide Res. 5,453-473. Kemp, B. E., Bylund, D. B., Huang, T. S., and Krebs, E. G. (1975).Substrate specificity of the cyclic AMP-dependent protein kinase. Proc. N a t l . Acad. Sci. U.S.A. 72, 34483452. Khoo, J. C., Steinberg, D., Huang, J. J., and Vagelos, P. R. (1976).Triglyceride diglyceride, monoglyceride and cholesterol ester hydrolases in chicken adipose tissue activated by adenosine 3’:5‘-monophosphate-dependent protein kinases. J. Biol. Chem. 251,2882-2890. Killilea, S . D., Brandt, H., and Lee, E . Y. C. (1976).Modulation of protein function by phosphorylation: The role of protein phosphatase(s). Trends Biochem. Sci. 1, 30-33. Kirchberger, M. A., Tada, M., Repke, D. I., and Katz, A. M. (1972).Cyclic adenosine 3‘,5‘-monophosphate-dependentprotein kinase stimulation of calcium uptake by canine cardiac microsomes. J. Mol. Cell Cardiol. 4,673-680. Kirchberger, M. A., Tada, M., and Katz, A. M. (1974).Adenosine 3’:5‘-monophosphate dependent protein kinase-catalyzed phosphorylation reaction and its relationship to calcium transport in cardiac sarcoplasmic reticulum. J . Biol. Chem. 249, 6166-6 173. Kleiman, J. H., and Moss, B. (1975a). Purification ofa protein kinase and two phosphate acceptor proteins from vaccinia virions. J. Biol. Chem. 250, 2420-2429. Kleiman, J. H., and Moss, B. (197%). Characterization of a protein kinase and two phosphate acceptor protein from vaccinia virions.]. Biol. Chem. 250,2430-2437. Krause, E. -G., Will, H., Pelouch, V., and Wollenberger, A. (1973).Cyclic AMPdependent protein kinase activity in a cell-membrane enriched subcellular fraction of pig myocardium. Acta B i d . Med. G e m . 31, K37-43. Krause, E . 4..Will, H., Schirpke, B., and Wollenberger, A. (1975).Cyclic AMPenhanced protein phosphorylation and calcium-binding in a cell membraneenriched fraction from myocardium. Ado. Cyclic Nucleotide Res. 5,473-490. Krystal, G . , Winn, P., Millward, S . , and Sakuma, S. (1975).Evidence for phosphopre teins in rheovirus. Virology 64,505-512. Krystosek, A., Bitte, L. F., and Kabat, D. (1974).Phosphorylation of ribosomal proteins in higher organisms. In “Metabolic Interconversion of Enzymes” (E. H. Fischer, E. G . Krebs, H. Neurath, and E. R. Stadtman, eds.), pp. 165-173. Springer-Verlag, Berlin. Kiihn, H. (1974).Light-dependent phosphorylation of rhodopsin in living frogs. Nature (London) 250,588-590. Kiihn, H., and Bader, S. (1976).The rate of rhodopsin phosphorylation in isolated retinas of frog and cattle. Biochim. Biophys. Acta. 428, 13-18.
312
M. MARLENE HOSEY AND MARIANO TAO
Kiihn, H., Cook, J. H., and Dreyer, W. J. (1973). Phosphorylation of rhodopsin in bovine photoreceptor membranes. A dark reaction after illumination. Biochemistry 12, 2495-2 502, Kumar, R., and Tao, M. (1975). Multiple forms of casein kinase from rabbit erythrocytes. Biochim. Biophys. Acta 410,87-98. Kumon, A., Yamamura, H., and Nishizuka, Y. (1970). Mode of action of adenosine 3’,5’-cyclic phosphate on protein kinase from rat liver. Biochem. Biophys. Res. Commun. 41,1290-1297. dependKuo, J. F. (1975). Changes in relative levels of guanosine3‘-5‘-monophosphate protein kinases in lung, heart, ent and adenosine 3’-5’-monophosphate-dependent and brain of developing guinea pigs. Proc. Natl. Acad. Sci. USA. 72,2256-2259. Kuo, J. F., and Greengard, P. (1969). Cyclic nucleotide -dependent protein kinases. IV. Widespread occurrence of adenosine 3’,5’-monophosphate dependent protein kinase in various tissues and phyla of the animal kingdom. Proc. Natl. Acad. Sci. U . S A . 64, 1349-1355. Kuranstin-Mills, J., Kudo, M., and Addae, S. K. (1974). Cation content and transport characteristics of the sickle-cell erythrocyte and their relationship with structural changes in the membrane. Clin. Sci. Mol. Med. 46,679-692. Labrie, F., Lemaire, S., and Courte, C. (1971).Adenosine 3’,5’-monophosphate dependent protein kinase from bovine anterior pituitary gland. I. Properties. ]. Biol. Chem. 246,7293-7302. Lagnado, J,, and Tan, L. P. (1975). The in situ phosphorylation of microtubular protein in brain cortex slices and related studies on the phosphorylation isolated brain tubulin preparations. Ann. N . Y. Acad. Sci. 253,577-597. Lagnado, J. R., Lyons, C., and Wickremasinghe, G. (1971).The subcellular distribution of colchicine binding protein (‘microtubule protein’) in rat brain. FEBS Lett. 15, 254-258. Langan, T. A. (1968). Histone phosphorylation: Stimulation by adenosine 3‘,5‘monophosphate. Science 162,579-581. Langan, T. A. (1973). Protein kinases and protein kinase substrates.Adu. Cyclic Nucleotide Res. 3,99-153. Langer, G. A. (1973). Heart: Excitation-contraction coupling. Ann. Rev. Physiol. 35, 55-86. LaRaia, P. J., and Morkin, E. (1974). Adenosine 3’,5‘-monophosphatedependentmembrane phosphorylation. A possible mechanism for the control of microsomal calcium transport in heart muscle. Circ. Res. 35,298-306. Lamer, J., and Villar-Palasi, C. (1971). Glycogen synthetase and its control. Curr. Top. Cell Regul. 3,196-236. Lee, R. E., and Feldman, J. D. (1964).Visualization of antigenic sites ofhuman erythrocytes with ferritin-antibody conjugates. 1.Cell Biol. 23,396401. Leemput-Coutrez, M., Camus, J., and Christophe, J. (1973). Cyclic nucleotidedependent protein kinases of the rat pancreas. Biochem. Biophys. Res. Commun. 54,182-190. Lernaire, S., Pelletier, G . , and Labrie, F. (1971). Adenosine 3’,5’-monophosphatedependent protein kinase from bovine pituitary gland. 11. Subcellular distribution. 1.Biol. Chem. 246,7303-7310. Lemay, A., Deschenes, M., Lemaire, S.,Poirier, G., Poulin, L., and Labrie, F. (1974). Phosphorylation of adenohypophyseal plasma membranes and properties of assw ciated protein kinase.]. Biol. Chem. 249,323-328. Letterier, J. F., Rappoport, L., and Nunez, J. (1974). Neurotubulin polymerization and phosphorylation reactions catalyzed by “associated” protein kinase. FEBS Lett. 46, 285-288.
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
31 3
Lightfoot, H. N., Mumby, M., and Traugh, J. A. (1975).Dephosphorylation of40S ribosomal subunits by phosphoprotein phosphate activity from rabbit reticulocytes. Biochem. Biophys. Res. Commun. 66,1141-1146. Lloyd, T., and Kaufman, S. (1974). The stimulation of partially purified bovine caudate tyrosine hydroxylase by phosphatidyl-L-serine. Biochem. Bioph ys. Res. Commun. 59, 1262-1269. Lloyd, T., and Kaufman, S. (1975). Evidence for the lack of direct phosphorylation of bovine caudate tyrosine hydroxylase following activation b y exposure to enzymatic phosphorylating conditions. Biochem. Biophys. Res. Commun. 66,907-913. Loor, F., Fomi, L., and Pernis, B. (1972). The dynamic state of the lymphocyte membrane. Factors affecting the distribution and turnover of surface immunoglobulins. Eur. J. Zmmunol. 2,203-212. Maeno, H., Reyes, P. L., Ueda, T., Rudolph, S. A., and Greengard, P. (1974). Autophosphorylation of adenosine 3’,5’-monophosphate-dependentprotein kinase from bovine brain. Arch. Biochem. Biophys. 164,551-559. Maeno, H., Ueda, T., and Greengard, P. (1975). Adenosine 3’ :5‘-monophosphatedependent protein phosphataseactivity in synaptic membrane fractions. 1.Cyclic Nucleotide Res. 1,37-48. Majumder, G. C., and Turkington, R. W. (1972). Hormonedependent phosphorylation of ribosomal and plasma membrane proteins in mouse mammary gland in uitro.]. Biol. Chem. 247,7207-7217. Majumder, G. C., Shrago, E., and Elson, C. E. (1975). Changes in cyclic AMPdependent protein kinase activity in Tetrahymena pyrifonnis during the growth cycle. Biochim. Biophys. Acta 384,399-412. Malkinson, A. M., Krueger, B. K., Rudolph, S. A., Casnellie, J. E., Haley, B. E., and Greengard, P. (1975). Widespread occurrence of a specific protein in vertebrate tissues and regulation by cyclic AMP of its endogenous phosphorylation and dephosphorylation. Metabolism 24,331-341. Marchesi, V. T., and Steers, E., Jr. (1968).Selective solubilization of a protein component of the red cell membrane. Science 159,203-204. Martelo, 0 .J., Woo, S. L. C., Reimann, E. M., and Davie, E. W. (1970). Effect of protein kinase on ribonucleic acid polymerase. Biochemistry 9,4807-4813. Matsui, H., and Schwartz, A. (1966). Purification and properties of a highly active ouabain-sensiti ve Na+,K+-dependentadenosine-triphosphatase from cardiac tissue. Biochim. Biophys. Acta 128,380-390. Miller, J. A., and Paulsen, R. (1975). Phosphorylation and dephosphorylation of frog outer segment membranes as part of the visual process. J. Biol. C h e n . 250, 44274432, Miller, J. A., Brodie, A. E., and Bownds, D. (1975). Light activated rhodopsin phosphorylation may control light sensitivity in isolated rod outer segments. F E B S Lett. 59,20-23. Milstein, S., Abita, J, -P., Chang, N., and Kaufman, S. (1976). Hepatic phenylalanine 4-monooxygenase is a phosphoprotein. Proc. Natl. Acad. Sci. U S A . 73,1591-1593. Miyamoto, E. (1975). Protein kinases in myelin of rat brain: Solubilization and characterization. J . Nerrrochem. 24,503-512. Miyamoto, E., and Kakiuchi, S. (1974).Z n oitro and in oioo phosphorylation of myelin basic protein by exogenous and endogenous adenosine 3’ :5’-monophosphatedependent protein kinases in brain. J. Biol. Chem. 249,2769-2777. Miyamoto, E., Kuo, J. F., and Greengard, P. (1969).Cyclic nucleotide-dependent protein kinases. 111. Purification and properties of adenosine 3’,5‘-monophosphatedependent protein kinase from bovine brain.]. Biol. Chem. 244,6395-6402. Miyamoto, E., Petzold, G. L., Harris, J. S., and Greengard, P. (1971).Dissociation and
314
M. MARLENE HOSEY AND MARIANO TAO
concomitant activation of adenosine 3’,5’-monophosphate-dependentprotein kinase by histone. Biochem. Biophys. Res. Commun. 44,305312. Miyamoto, E., Petzold, G. L., Kuo, J. F., and Greengard, P. (1973). Dissociation and activation of adenosine 3’,5’-monophosphate-dependent and guanosine 3 ‘ 3 monophosphate-dependent protein kinases by cyclic nucleotides and by substrate proteins. J. Biol. Chem. 248,179-189. Morgenroth, V. H., 111, Hegstrand, L. R., Roth, R. H., and Greengard, P. (1975). Evidence for involvement of protein kinase in the activation by adenosine 3’:5’monophosphate ofbrain ty-rosine 3-monooxygenase. J. Biol. Chem. 250,1946-1948. Murray, A. W., and Froscio, M. (1971). Cyclic adenosine 3’ :5’-monophosphate and microtubule function: Specific interaction of the phosphorylated protein subunits with a soluble brain component. Biochem. Biophys. Res. Commun. 44,1089-1094. Nakai, C., and Thomas, J. A. (1974). Properties of a phosphoprotein phosphatase from bovine heart with activity on glycogen synthase, phosphorylase, and histone. J. Biol. Chem. 249,6459-6467. Nakao, M. (1974). ATP-requiring phenomena in red-cell membranes. In “Cellular and Molecular Biology of Erythrocytes” (H. Yoshikawa and S. M. Rappoport, eds.), pp. 35-54. Univ. Park Press, Baltimore, Maryland. Nakazawa, K., and Sano, M. (1975). Partial purification and properties of guanosine 3’,5’-monophosphatedependent protein kinase from pig lung. J. Biol. Chem. 250, 74 15-74 19. Narumi, S., and Miyamoto, E. (1974). Activation and phosphorylation of carbonic anhydrase by adenosine 3’,5’-monophosphate-dependentprotein kinases. Biochim. Biophys. Acta 350,215-224. Nicolson, G. L. (1976). Transmembrane control of the receptors on normal and tumor cells. I. Cytoplasmic influence over cell surface components. Biochim. Biophys. Acta 457,57-108. Nicolson, G. L., Masouredis, S. P., and Singer, S . J. (1971). Quantitative twodimensional ultrastructural distribution of Rh (D) antigenic sites on human erythrocyte membranes. Proc. Natl. Acad. Sci. U.S.A. 68, 1416-1420. Nilsson, K. ( 1973). Ca2+-binding and release in smooth muscle sarcoplasmic reticulum in relation to cyclic AMP. Acta Physiol. Scand. Suppl. 396,8. Nimmo, H. G., and Cohen. P. (1974). Glycogen synthetase kinase 2 (GSK-2)the identification of a new protein kinase in skeletal muscle. F E B S Lett. 47, 162-166. Nimmo, H. G., and Cohen, P. (1975). Glycogen synthetase kinase 2 and adenosine 3‘.5‘-monophosphateindependentprotein kinase activities in rabbit skeletal muscle. Biochem. SOC. Trans. 3,85-86. Orloff, J., and Handler, J. (1967). The role of adenosine 3‘,5‘-phosphate in the action of antidiuretic hormone. Amer. J. Med. 42, 757-768. Ostwald, T. J., and MacLennan, D. H. (1974). Isolation of a high affinity calciumbinding protein from sarcoplasmic reticulum. /. Biol. Chem. 249,974-979. . Pai, S. H., Ponta, H., Rahmsdorf, H. J., Hirsch-Kaufmann, M., Herrlich, P., and Schweiger, M. (1975a). Protein kinase of bacteriophage T7. I. Purification. Eur. I . Biochem. 55,299-304. Pai, S. H., Rahmsdorf, H. J., Ponta, H., Hirsch-Kaufniann, M., Herrlich, P., and Schweiger, M. (1975b). Protein kinase of bacteriophage T7. 11. Properties, enzyme synthesis in oitro and regulation of enzyme synthesis and activity in oioo. Eur. J. Biochem. 55,305-314. Pal, B. K., and Roy-Burman, P. (1975). Phosphoproteins: Structural components on oncornaviruses. ]. Virol. 15,540-549. Perlman, R. L., and Pastan, I. (1971): The role ofcyclic AMP in bacteria. Ctcrr. Top. Cell Regtil. 3, 117-134.
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
315
Peters, R., Peters, J., Tews, K. H., and Bahr, W. (1974).A microfluorimetric study of translational diffusion in erythrocyte membranes. Biochim. Biophys. Acta 367,
282-294. Pinkett, M. O., and Perlman, R. L. (1974). Phosphorylation of muscle plasma membrane protein by a membrane-bound protein kinase. Biochim. Biophys. Acta 372,
379-387. Pinto da Silva, P. (1972).Translational mobility of the membrane intercalated particles of human erythrocyte ghosts. J. Cell Biol. 53,777-787. Pinto da Silva, P., and Nicholson, G. L. (1974).Freeze-etch localization ofconcanavalin A receptors to the membrane intercalated particles of human erythrocyte ghost membranes. Biochim. Biophys. Acta 363,311419. Piras, M. M., and Piras, R. (1974).Phosphorylation of vinblastine-isolated microtubules from chick-embryonic muscles. Eur. J . Biochem. 47,443-452. Rahmsdorf, H.J., Pai, S. H., Ponta, H., Herrlich, P., Roskoski, R., Jr., Schweiger, M., and Studier, F. W. (1974).Protein kinase induction in Escherichia coli by bacteriophage T7. Proc. Natl. Acad. Sci. U.S.A. 71,586-589. Randall, C. C., Rogers, H. W., Downer, D. N., and Gentry, C.A. (1972).Protein kinase activity in equine herpesvirus. J . Virol. 9,216-222. Rangel-Aldao, R.,and Rosen, 0. M. (1976).Dissociation and reassociation of the phosphorylated and nonphosphorylated forms of adenosine 3’:5‘-monophosphate dependent protein kinase from bovine cardiac muscle. J . Biol. Chem. 251,
3375-3380. Rappoport, L., Letterier, J. L., and Nunez, J . (1972). Nonphosphorylation in uitro ofthe 6 S tubulin from brain and thyroid tissue. F E B S Lett. 26,349-352. Rappoport, L.,Letterier, J. L., and Nunez, J. (1975).Protein-kinase activity, in uitro phosphorylation and polymerization of purified tubulin. Ann. N. Y. Acad. Sci. 253, 6 11-629. Rappoport, L., Letterier, J. L., Virion, A., and Nunez, J. (1976).Phosphorylation of microtubule-associated proteins. Eur. J . Biochem. 62,539-549. Reddy, Y. S.,Ballard, D., Giri, N. Y., and Schwartz, A. (1973). Phosphorylation of cardiac native tropomyosin and troponin: Inhibitory effect of actomyosin and possible presence of endogenous myofibrillar-located cyclic AMP-dependent protein kinase. J .
Mol. Cell. Cardiol. 5,461-471. Reiniann, E. M.,and Schlender, K. K. (1976).Multiple forms of glycogen synthase kinase: Isolation of forms which are independent of cyclic AMP.J. Cyclic Nocleo-
tide Res. 2,3946. Reimann, E. M.,Walsh, D. A,, and Krebs, E. G . (1971).Purification and properties of rabbit skeletal muscle adenosine 3’,5’-nionophosphate-dependentprotein kinase. J . B i d . Chem. 246,1986-1995. Rillema, J. A. (1976). Possible interactions of cyclic nucleotides with the prolactin stimulation of casein synthesis in mouse mammary gland explants. Biochim. Biophys. Acts 432,348-352. Rizack, M. A. (1964).Activation of an epinephrine-sensitive lipolytic activity from adipose tissue by adenosine 3‘,5’-phosphate.J. Biol. Chem. 239,392-395. Roach, P. J., Takeda, Y., and Lamer, J. (1976).Rabbit skeletal muscle glycogen synthase. I. Relationship between phosphorylation state and kinetic properties. J .
Biol. Chem. 251,1913-1919. Robison, G. A., Butcher, R. W., and Sutherland, E. W. (1971).“Cyclic AMP.” Academic Press, New York. Rodnight, R., and Lavin, B. E. (1964).Phosvitin kinase from brain: Activation by ions and subcellular distribution. Biochem. J . 93,84-91. Rosen, 0. M., and Erlichman, J. (1975).Reversible autophosphorylation of a cyclic
316
M. MARLENE HOSEY A N D MARIANO TAO
3’ :5I-AMP-dependent protein kinase from bovine cardiac muscle. J . Biol. Chem. 250,7788-7794, Rosenthal, A. S., Kregenow, F. M., and Moses, H. L. (1970). Some characteristics of a CaZ+-dependent ATPase activity associated with a group of erythrocyte membrane proteins which form fibrils. Biochim. Biophys. Acta 196,254-262. Roses, A. D., and Appel, S. H. (1973). Erythrocyte protein phosphorylation. J. Biol. Chem. 248,1408-1411. Roses, A. D., and Appel, S. H. (1974). Muscle membrane protein kinase in myotonic muscular dystrophy. Nature (London) 250,245-247. Roses, A. D., and Appel, S. H. (1975). Phosphorylation of component a of the human erythrocyte membrane in myotonic muscular dystrophy. J. Membr. Biol. 20, 51-58. Roses, A. D., Herbstreith, M. H., and Appel, S. H. (1975). Membrane protein kinase alteration in Duchenne muscular dystrophy. Nature (London) 254,350351. Rothstein, A., Cabantchik, Z. I., and Knauf, P. (1976). Mechanism of anion transport in red blood cells: Role of membrane proteins. Fed. Proc., Fed. Am. Soc. Erp. Biol. 35, 3-10. Rubin, C. S. (1975). Adenosine 3’:5’-monophosphatereglated phosphorylation of erythrocyte membrane proteins. Separation of a cyclic AMP-dependent protein kinase from its endogenous substrates in erythrocyte plasma membranes. J . Biol. Chem. 250,9044-9052. Rubin, C. S., and Rosen, 0. M. (1973).The role of cyclic AMP in the phosphorylation of proteins in human erythrocyte membranes. Biochem. Biophys. Res. Commun. 50, 421-429. Rubin, C. S.,and Rosen, 0. M. (1975). Protein phosphorylation.Ann. Rev. Biochem. 44, 831-887. Rubin, C. S., Erlichman, J., and Rosen, 0. M. (1972). Cyclic adenosine 3‘,5’monophosphate-dependent protein kinase of human erythrocyte membranes. I . Biol. Chem. 247,6135-6139. Rubin, C. S.,Rosenfeld, R. D., and Rosen, 0. M. (1973). Studies on the orientation ofcyclic AMP-dependent protein kinase in human erythrocyte membranes. Proc. Natl. Acad. Sci. U S A . 70,3735-3738. Rudolph, S. A., and Greengard, P. (1974). Regulation of protein phosphorylation and membrane permeability by p-adrenergic agents and cyclic adenosine 3’ :5’monophosphate in the avian erythrocyte.]. Biol. Chem. 249,5684-5687. Schatzmann, H. J. (1975). Active calcium transport and Cat-activated ATPase in human red cells. In “Current Topics in Membranes and Transport” (F. Bronner and A. Kleinzeller, eds.), Vol. 6, pp. 126-168. Academic Press, New York. Schlender, K. K., and Reimann, E. M. (1975). Isolation of a glycogen synthase I kinase that is independent of adenosine 3’ :5’-monophosphate. Proc. Natl. Acad. Sci. U S A . 72,2197-2201. Schlender, K. K., Wei, S. H., and Villar-Palasi, C. (1969). UDP-glucose: Glycogen a4-glucosyltransferase I kinase activity of purified muscle protein kinase. Cyclic nucleotide specificity. Biochim. Biophys. Acta 191,272-278. Schultz, G., and Hardman, J. G. (1975).Regulation of cyclic GMP levels in the ductus deferens of the rat. Adv. Cyclic Nuckotide Res. 5,339351. Schwartz, A., Entman, M. L., Kanike, K., Lane, L. A., Van Winkle, W. B., and Bornet, E. P. (1976). The rate of calcium uptake into sarcoplasmic reticulum of cardiac muscle and skeletal muscle. Effects of cyclic AMP-dependent protein kinase and phosphorylase b kinase. Biochim. Biophys. Acta 426,57-72. Severson, D. L., Drummond, G. I., and Sulakhe, P. V. (1972). Adenylate cyclase in
PROTEIN KINASES A N D MEMBRANE PHOSPHORYLATION
317
skeletal muscle. Kinetic properties and hormonal stimulation. J. Biol. Chem. 247,2949-2958. Shelanski, M. L., Gaskin, F., and Cantor, C. R. (1973).Microtubule assembly in t h e absence of added nucleotides. Proc. Natl. Acad. Sci. U S A . 70, 765-768. Sheppard, H., and Burghardt, C. (1969).Adenyl cyclase in non-nucleated erythrocytes of several mammalian species. Biochem. Phannacol. 18,2576-2578. Shichi, H., Somers, R. L., and O'Brien, P. J. (1974).Phosphorylation ofrhodopsin: Most rhodopsin molecules are not phosphorylated. Biochem. Biophys. Res. Commun. 61, 2 17-221. Shimomura, R.,Matsumura, S., and Nishizuka, Y. (1974).Membraneassociated a d e n e sine 3',5'-monophosphate-dependent protein kinase and its substrate in human erythrocytes. I . Biochem. 75, 1-10. Sloboda, R. D., Rudolph, S. A., Rosenbaum, J. L., and Greengard, P. (1975). Cyclic AMP-dependent endogenous phosphorylation of a microtubule-associated protein. Proc. Natl. Acad. S c i . U S A . 72,177-181. Smith, C. H., Brown, N. E., and Larner, J. (1971).Molecular characteristics o f t h e totally dependent and independent forms of glycogen synthetase of rabbit skeletal muscle. 11. Some chemical characteristics of the enzyme protein and of its change on interconversion. Biochim. Biophys. Acta 242,81-88. Smith, L. S., Kern, C. W., Halpern, R. M., and Smith, R. A. (1976).Phosphorylation on basic amino acids in myelin basic protein. Biochem. Biophys. Res. Commun. 71, 45 9 4 6 5 . Soderling, T. R . (1975). Regulation of glycogen synthetase. Specificity and stoichiometry of phosphorylation of the skeletal muscle enzyme by cyclic 3' :5'-AMPdependent protein kinsae. J. Biol. Chem. 250,5407-5412. Soderling, T. R., Hickenbottom, J. P., Reimann, E. M., Hunkeler, F. L., Walsh, D. A., and Krebs, E. G. (1970).Inactivation of glycogen synthetase and activation ofphosphorylase kinase by muscle adenosine 3',5'-rnonophosphatedependent protein kinase. ]. Biol. Chem. 245,6317-6328. Soifer, D. (1975). Enzymatic activity in tubulin preparations: Cyclic-AMP-dependent protein kinase activity of brain microtubule protein.]. Neurochem. 24,21-33. Soifer, D., Laszlo, A. H., and Scotto, J. M. (1972).Enzymatic activity in tubulin preparations. I. Intrinsic protein kinase activity in lyophilized preparations of tubulin from porcine brain. Biochim. Biophys. Acta 271, 182-192. Soifer, D., Laszlo, A., Mack, K., Scotto, J., and Siconolfi, L. (1975).The association o f a cyclic AMP-dependent protein kinase activity with microtubule protein. Ann. N.Y. Acad. Sci. 253,598-610. Sokol, F., and Clark, H. F. (1973). Phosphoproteins, structural components of rhabdoviruses. Virology 52,246-263. Steck, A. J., and Appel, S. H. (1974). Phosphorylation of myelin basic pr0tein.J. Biol. Chem. 249,5416-5420. Steck, T. L. (1972).Cross-linking the major proteins of the isolated erythrocyte membrane.]. Mol. Biol. 66,295-305. Steck, T. L. (1974). The organization of proteins in the human red cell membrane.]. Cell. Biol. 62, 1-19. Steinberg, D. (1973). Hormonally regulated enzymes in adipose tissue linked to cyclic AMP-dependent protein kinase. Miami Winter Symp. 5,47-65. Steiner, M. (1975).Endogenous phosphorylation of platelet membrane proteins. Arch. Biochem. Biophys. 171,245-254. Strand, M., and August, J. T. (1971). Protein kinase and phosphate acceptor proteins in Rauscher murine leukaemia virus. Nature (London),New Biol. 233,137-140.
31 8
M. MARLENE HOSEY AND MARIANO TAO
Stull, J. T., Brostrom, C. O., and Krebs, E. G. (1972).Phosphorylation of the inhibitor component of troponin by phosphorylase kinase. J . Biol. Chem. 247,5272-5274. Sulakhe, P. V., and Drummond, G. I. (1974).Protein kinase-catalyzed phosphorylation of muscle sarcolemma. Arch. Biochem. Biophys. 161,448-455. Sulakhe, P. V., Drummond, G. I., and Ng, D. C., (1973). Calcium binding by skeletal muscle sarco1emma.J. Biol. Chem. 248,4150-4163. Tada, M., Kirchberger, M. A., Repke, D. I., and Katz, A. M. (1974). The stimulation of calcium transport in cardiac sarcoplasmic reticulum by adenosine 3’ :5’monophosphate-dependent protein kinases. J . Biol. Chem. 249,6174-6180. Tada, M., Kirchberger, M. A., and Katz, A. M. (1975a). Phosphorylation of a 22,OOG dalton component in the cardiac sarcoplasmic reticulum by adenosine 3’ :5’monophosphate-dependent protein kinase. J . Biol. Chem. 250,2640-2647. Tada, M., Kirchberger, M. A., and Li, H. C. (197%). Phosphoprotein phosphatasecatalyzed dephosphorylation of the 22,000 dalton phosphoprotein of cardiac sarcoplasmic reticulum. J . Cyclic Nucleotide Res. 1,329-338. Takai, Y., Nakaya, S., Inoue, M., Kishimoto, A., Nishiyma, K., Yamamura, H., and Nishizuka, Y. (1976). Comparison of mode of activation of guanosine 3‘:5’monophosphate-dependent and adenosine 3’ :5’-monophosphatedependent protein kinases from si1kworm.J. Biol. Chern. 251, 1481-1487. Tan, K. B. (1975). Comparative study of the protein kinase associated with animal viruses. Virology 64, 566-570. Tan, K. B., and Sokol, F. (1972). Structural proteins of simian virus 4 0 Phosphoproteins. J . Virol. 10,985-994. Tao, M. (1971a). Mechanism of activation of a rabbit reticulocyte protein kinase by adenosine 3‘,5’-monophosphate-dependentprotein kinase. Ann. N.Y. Acad. Sci. 185,227-231. Tao, M. (1971b). Further studies on the properties of the rabbit reticulocyte adenosine 3‘,5’-cyclic monophosphate-dependent protein kinase I. Arch. Biochem. Biophys. 143,151-157. Tao, M. (1972). Dissociation of rabbit red blood cell cyclic AMP-dependent protein kinase I by protamine. Biochem. Biophys. Res. Commun. 46,56-61. Tao, M.,and Hackett, P. (1973).Adenosine cyclic 3‘ :5‘-monophosphate-dependent protein kinase from rabbit erythrocytes. Purification and characterization of multiple forms. J . Biol. Chem. 248,5324-5332. Tao, M., Salas, M. L., and Lipmann, F. (1970). Mechanism of activation by adenosine 3’,5‘-cyclic monophosphate of a protein phosphokinase from rabbit reticulocytes. Proc. Natl. Acad. Sci. U.S.A. 67,408-414. Traugh, J. A., and Traut, R. R. (1974). Characterization of protein kinases from rabbit erythrocytes.]. Biol. Chem. 249, 1207-1212. Traugh, J. A., Mumby, M., and Traut, R. R. (1973).Phosphorylation of ribosomal proteins by substrate-specific protein kinase from rabbit reticulocytes. Proc. Natl. Acad. Sci. U S A . 70,373-376. Traut, R. R., Howard, G. A,, and Traugh, J. A. (1974). Ribosomal proteins from rabbit reticulocytes: Numbers, molecular weights, relative amounts and phosphorylation by protein kinases. In “Metabolic Interconversion of Enzymes” (E. H. Fischer, E. G. Krebs, H. Neurath, and E. R. Stadtman, eds.), pp. 155-164. Springer-Verlag, Berlin. Tria, E., Luly, P., Tomasi, V., Trevisani, A., and Barnabei, 0.(1974).Modulation by cyclic AMP in oitro of liver plasma membrane (Na+-K+)-ATPaseand protein kinases. Biochim. Biophys. Acta 343,297-306. Tsuzuki, J., and Kiger, J. A., Jr. (1975). Cyclic AMP binding proteins in early embryos of Drosophila melanogaster. Biochim. Biophys. Acta 393,225-235.
PROTEIN KINASES AND MEMBRANE PHOSPHORYLATION
319
Ueda, T., Maeno, H., and Greengard, P. (1973).Regulation ofendogenous phosphorylation of specific proteins in synaptic membrane fractions from rat brain by adenosine 3’,5’-monophosphate.J. Biol. Chem. 248,8295-8305. Ueda, T., Rudolph, S . A., and Greengard, P. (1975).Solubilization of a phosphoprotein and its associated cyclic AMP-dependent protein kinase and phosphoprotein phosphatase from synaptic membrane fractions, and some kinetic evidence for their existence as a complex. Arch. Biochem. Biophys. 170,492-503. Virmaux, N.,Weller, M., and Mandel, P. (1975).Localization of the major site of light stimulated phosphorylation in a region of rhodopsin distinct from the chromophore binding site. FEBS Lett. 53,320-323. Walinder, 0.(1973).Calfbrain phosvitin kinase. 11. Purification and characterization of three different fractions of phosvitin kinase. Biochirn. Biophys. Acta 293,140-149. Walsh, D.A., and Krebs, E.G. (1973).Protein kinases. The Enzymes 8,555-581. Walsh, D. A., Perkins, J. P., and Krebs, E. G. (1968). An adenosine 3’,5‘monophosphate-dependent protein kinase from rabbit skeletal muscle. J. Biol. Chem. 243,3763-3765. Walsh, D. A,, Ashby, C. D., Gonzalez, C., Calkins, D., Fischer, E. H., and Krebs, E.G. (1971).Purification and characterization of a protein inhibitor of adenosine 3’,5’monophosphate-dependent protein kinases. J. Biol. Chem. 246, 1977-1985. Wang, K.,and Richards, F.M. (1974).An approach to the nearest neighbor analysis of membrane proteins. J . Biol. Chem. 249,8005-8018. Weed, R. I., and LaCelle, P. L. (1969).ATP dependence of erythrocyte membrane deformity. In “Red Cell Membrane Structure and Function” (G. A. Jamieson and T. J. Greenwalt, eds.), pp. 318-338. Lippincott, Philadelphia, Pennsylvania. The colchicine-binding proWeisenberg, R. C., Borisy, G. C., and Taylor, E. W. (1968). tein of mammalian brain and its relation to microtubules. Biochemistry 7, 4466-4479. Weller, M., and Rodnight, R. (1971).Turnover of protein-bound phosphorylserine in membrane preparations from ox brain catalyzed by intrinsic kinase and phosphatase activity. Biochem. J. 124, 393406. Weller, M., and Rodnight, R. (1973).Protein kinase in membrane preparations from ox brain. Biochem. J . 132,483492. Weller, M., Virmaux, N., and Mandel, P. (1975a).Light-stimulated phosphorylation of rhodopsin in the retina. The presence of a protein kinase that is specific for photobleached rhodopsin. Proc. Natl. Acad. Sci. U . S A . 72,381-385. Weller, M., Virmaux, N., and Mandel, P. (1975b).Role of light and rhodopsin phosphorylation in control of permeability of retinal rod outer segment disks to Ca++ Nature (London) 256,68-70. Weller, M., Goridis, C., Virmaux, N., and Mandel, P. (1975~). A hypothetical model for the possible involvement of rhodopsin phosphorylation in light and dark adaptation in the retina. E x p . Eye Res. 21,405408. Weller, M., Virmaux, N., and Mandel, P. (1976).The relative specificity ofopsin kinase towards ATP and GTP and the lack of effect of cyclic nucleotides on the activity of the enzyme. E x p . Eye Res. 23,67-70. Will, H.,Schirpke, B., and Wollenberger, A. (1973).Binding of calcium to a cell membrane-enriched preparation from pig myocardium: Increase in calcium affinity upon membrane protein phosphorylation enhanced by a membrane-bound cyclic AMP-dependent protein kinase. Acta Biol. Med. Germ. 31, K45-52. Wray, H. L., Gray, R. R., and Olsson, R. A. (1973).Cyclic adenosine 3‘,5’monophosphate-stimulated protein kinase and a substrate associated with cardiac sarcoplasmic reticulum. J. Biol. Chem. 248, 1496-1498.
320
M. MARLENE HOSEY AND MARIANO TAO
Yamamura, H., Kumon, A., Nishiyama, K., Takeda, M., and Nishizuka, Y. (1971).Characterization of two adenosine 3’,5’-monophosphatedependentprotein kinases from rat liver. Biochem. Biophys. Res. Commun. 45,1560-1566. Yeaman, S . J., and Cohen, P. (1975).The hormonal control of activity of skeletal muscle phosphorylase kinase. Phosphorylation of the enzyme at two sites in oioo in response to adrenalin. Eur. J . Biochem. 51,93-104. Yuh, K. M., and Tao, M. (1974).Purification and characterization of adenosine adenosine cyclic 3’,5‘-monophosphate binding protein factors from rabbit erythrocytes. Biochemistry 13,5220-5226. Zail, S . S.,and van den Hoek, A. K. (1975).Studies of protein kinase and the binding of adenosine 3’,5’-monophosphate by membranes of hereditary spherocytosis erythrocytes. Biochem. Biophys. Res. Commun. 66, 1078-1086. Zieve, F.J., and Glinsmann, W. H. (1973).Activation ofglycogen synthetase and inactivation ofphosphorylase kinase by the same phosphoprotein phosphatase. Biochem. Biophys. Res. Commun. 50,872-878. ZiUig, W., Fujiki, H., Blum, W., Janekovic, D., Schweiger, M.,Rahmsdorf, H.-J.,Ponta, H., and Hirsch-Kauffmann, M. (1975).In uioo and in oitro phosphorylation of DNA-dependent RNA polymerase of Escherichia coli by baderiophage-Tinduced protein kinase. Proc. Natl. Acad. Sci. U.S.A. 72,2506-2510.
Mechanism and PhysiologicaI Significance of Calcium Transport across Mammalian Mitochondrial Membranes LEENA M E L A Departments of Surgery and Biochemistry and Biophysics University of Pennsylvania Philadelphia. Pennsylvania
I . Introduction .......................................................... 322 I1. Early Experiments Leading to the Discovery of Mitochondrial Ability to Accumulate CrP+ Ions ...................................... 322 111. T h r e e s t e p Mechanism of Mitochondrial Ca2+Accumulation . . . . . . . . . . . . 326 A . Energy-Independent Binding .................................... 327 B . Carrier-Mediated Accumulation of Ca2+ . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 C Accumulation of Ca2+in the Matrix .......................... 335 IV . Role of Mitochondria in the Physiological Ca2+ Concentration ................................................... 337 A Kinetics of Mitochondria1 Ca2+Accumulation ................... 338 B . Mitochondfial Affinity for Ca2+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340 C . Mitochondrial Capacity to Accumulate CaZ+ ....................... 341 D . Release of CrP+ from Mitochondria under Physiological Conditions ................ ................................ 342 V . Physiological Significance of Mito rial Caz+Accumulation in Different Tissues .................................................. 344 A . Heart ........................ ............................... 345 B . Liver and Kidney ................................................ 346 C . Brain and Nervous Tissue ........................................ 347 D . Smooth Muscle ........................................... 348 E . Calcifying Tissue ................................................ 349 VI . Some Aspects of the Pathophysiology of Mitochondrial CrP+ Accumulation ................................................... 350 A . C 2 + Accumulation by Tumor Cell Mitochondria ................... 351 B . Effect of Ischemic Cell Injury on Mitochondrial Ca2+Accumulation . 352 354 VII . Summary ............................................................ References .......................................................... 354
.
.
321
322
LEENA MELA
1.
INTRODUCTION
The role of mitochondria in cellular energy conservation is well documented. The formation of high-energy compounds utilizing energy accumulated in the course of electron transfer from Krebs' cycle intermediates to oxygen via the respiratory chain undoubtedly constitutes the most important function of these organelles. Mitochondrial Caz+accumulation, another prominent function of the mitochondrial membrane system, often is considered to be of secondary importance. This judgment probably arises from the limited and sometimes questionable evidence concerning the purpose of mitochondrial Caz+accumulation. However, the evidence for the existence of an active mechanism for Caz+ accumulation by the mitochondrial membrane system seems overwhelming. Moreover, the reaction appears to be favored over phosphorylation of ADP to ATP. Mitochondrial calcium accumulation appears to be compatible with active transport across the inner mitochondrial membrane. The process requires energy and occurs against a concentration gradient. Mitochondria are capable of lowering the free extramitochondrial calcium concentration to lC7M. The transport against an electrochemical gradient is less well documented. Although seldom used to specify mitochondrial calcium accumulation, the term calcium pump may be applicable. The problem of mitochondrial Caz+accumulation is important both physiologically and pathologically. My mission in writing this article is (i)to point out what is known about the mechanism and kinetics of mitochondrial Caz+ accumulation and (ii) to discuss the evidence in support of the notion that mitochondrial Ca2+transport functions in the regulation of intracellular free Ca2+concentration. I also wish to elaborate on the effects of some unphysiological or disease states on the ability of mitochondria to accumulate Caz+.This article does not attempt to review completely the literature on mitochondrial Caz+accumulation reviewed by Harris et al., 1966; Lehninger et al., 1967; Lehninger, 1970; Carafoli and Rossi, 1971; Chance and Montal, 1971; and Scarpa, in press. Rather, it will emphasize the physiological significance of the phenomenon. II. EARLY EXPERIMENTS LEADING TO THE DISCOVERY OF MITOCHONDRIAL ABILITY TO ACCUMULATE CA2+ IONS
It was first reported by Lehninger in 1949 that Caz+added to a suspension of isolated liver mitochondria caused uncoupling of respiration. In 1953, Slater and Cleland observed that heart mitochondria
MITOCHONDRIAL CALCIUM TRANSPORT
323
(sarcosomes), were able to bind CaZ+tightly in uitro. Also Siekevitz and Potter (1955)and Siekevitz et al. (1953) reported an irreversible stimulation of mitochondrial respiration by an addition of Caz+.Irreversible stimulation of mitochondrial respiratory activity by Ca2+was also reported by Lindberg and Ernster in 1954. The first evidence of Ca2+-induced reversible respiratory bursts and of simultaneously measured redox changes of cytochromes were shown by Chance in 1956. The early authors neither measured Ca2+accumulation nor considered that it could have been responsible for the observed mitochondrial respiratory stimulation. Lehninger (1949),for instance, included Caz+in the list of other known uncoupling agents of unknown mechanisms. Chance (1956) realized that the respiratory cycles induced by Ca2+closely resembled those caused by ADP. Only in 1961 and 1962 Vasington and Murphy, and DeLuca and Engstrom (1961), found that isolated mitochondria accumulated large quantities of added Caz+ ions in a respiration-dependent process. This was the discovery that, for the first time, clearly linked Caz+accumulation and increased respiratory activity, and, thus, opened a new field for further investigation. Vasington and Murphy (1961, 1962) used isolated rat kidney mitochondria and measured Ca2+ accumulation by means of radioactive “Ta. They concluded as follows: (i) Mitochondria can accumulate large quantities of Caz+,up to 2 pmoles/mg of mitochondrial protein. (ii) Maximal “binding” requires the presence of a respiratory substrate, ATP, inorganic phosphate, and Mg2+ ions. ADP can partially substitute for ATP. (iii) The accumulation of Ca2+is inhibited by respiratory inhibitors and uncoupling agents. According to DeLuca and Engstrom (1961), the accumulation of Ca2+was dependent on ATP and M$+ ions, and on the presence of an oxidizable substrate, but not on inorganic phosphate. They concluded that Caz+accumulation was not directly dependent on oxidative phosphorylation nor on the operation of the entire respiratory chain. Ca2+accumulation was not inhibited by ol igomycin. At that time, the requirement of both ATP and Mg2+ and of an oxidizable substrate caused some confusion. The fact that substrate oxidation alone, without ATP and M$+, did not provide sufficient support for Caz+ accumulation indicates that the mitochondrial preparations available to the investigators of those times were not of a sufficiently good quality. When better coupled preparations of intact mitochondria became available, due to improved techniques of isolation, new facts were discovered about the requirements for optimal Ca2+accumulation.
324
LEENA MELA
In 1963 Lehninger et al. showed that inorganic phosphate was accumulated in a definite molar ratio together with accumulated Ca2+.Uncouplers and cyanide blocked the uptake of both phosphate and Ca2+. Oligomycin did not inhibit the accumulation of Ca2+ if respiratory substrate was present. They also showed that, in the presence of a high concentration of ATP, respiratory substrate was not needed to support Ca2+ accumulation by mitochondria. In 1963 and 1964, Brierley (1963) and Brierley et al. (1963, 1964), working on isolated heart miochondria, supported Lehninger’s (1962) view that ATP in the absence of added respiratory substrates was able to support Caz+ accumulation that occurred parallel with phosphate uptake. They also showed that the ATP-supported Ca2+accumulation was inhibited by oligomycin, whereas substrate-supported accumulation was not oligomycin sensitive. Both ATP and substrate-driven Ca2+ accumulation were uncoupler sensitive. Taken together, these findings showed that isolated liver or heart mitochondria were able to accumulate large quantities of Ca2+ions in an energy-dependent process. The energy required was provided either by substrate oxidation or by ATP hydrolysis. When inorganic phosphate was present it was accumulated with Ca2+.However, some Caz+ accumulation occurred also in the absence of inorganic phosphate. At this time, several speakers at the first Johnson Foundation Colloquium presented evidence for the accumulation of various divalent cations in mitochondria. Chance (1963) pointed out that small amounts of Ca2+added to mitochondria caused transient cyclic oxidation of mitochondrial NADH and cytochrome b, paralleling a rapid burst of respiratory activity. He suggested a stepwise mechanism for Ca2+ accumulation: (i) surface binding, followed by (ii) an energydependent accumulation in the absence of added inorganic phosphate and (iii) a fully completed uptake after an addition of phosphate. Under these conditions, Caz+would precipitate as a Ca phosphate in the mitochondrial matrix space. He felt that an effective energy donor for the Ca2+ accumulation would be the high-energy intermediate formed at the first phosphorylation site (DPNH - I). Likewise Brierley et al. (1963, 1964) had postulated that Ca2+ accumulation could utilize either ATP or the high-energy intermediate X I as an energy source. Simultaneously, it was shown by Chappell et al. (1963a,b)that other divalent cations, such as SP+ and Mn2+,could be accumulated in mitochondria through a process similar to Ca2+ accumulation in the absence or presence of inorganic phosphate. Later C a d o l i et al. (1965)
-
MlTOCHONDRIAL CALCIUM TRANSPORT
325
and Drahota et al. (1969) verified the ability of mitochondria to accumulate Sf+. Maynard and Cotzias (1955) demonstrated that mitochondria accumulated over 40% of the total amount of Mn2+reaching the liver after an in vivo injection. In 1963, both Brierley et al. (1963) and Saris (1963 a,b) reported that H+ ions were extruded from mitochondria when Ca2+was accumulated. Saris (1963 a,b) investigated the stoichiometry of H+ ion extrusion. The molar ratio of H + extruded/Ca2+added was 0.8 in the presence of added inorganic phosphate, and it exceeded 1.0in the absence of phosphate. The ratio of ATP hydrolyzed/Ca2+ accumulated was 1.4, according to Saris (1963 a). Rossi and Lehninger (1964) and Bielawski and Lehninger (1966) showed that in the absence of phosphate 1.9 molecules of Ca2+were accumulated for each energy-conserving site activated. Rossi and Lehninger (1964) also verified that mitochondria were able to accumulate only limited amounts of Ca2+in the absence of added phosphate (100-150 nmoles/mg protein was found to constitute maximal uptake). In the presence of added phosphate and ATP, massive amounts of Ca2+ were accumulated by liver mitochondria (Lehninger et al., 1963; Rossi and Lehninger, 1964). Greenawaltet al. (1964) illustrated by electron microscopy that Ca2+ and phosphate were deposited as insoluble, amorphous, electron-dense hydroxyapatite (presumably) within the inner compartment of mitochondria (matrix). The molar ratio of 1.67 of accumulated Ca2+ and phosphate suggested this. At the same time, Peachey (1964) and Greenawal t and Carafoli (1966) presented electron microscopic evidence in isolated mitochondria and in mitochondria in whole cells that Ca2+,S P , and Ba2+were accumulated in “granules” localized in the mitochondrial matrix. It was shown by Rasmussen et al. (1965) and by Chance and Yoshioka (1965) that acetate could also act as a permeant anion and move across the mitochondrial membrane as a counterion with Ca2+. The swelling of mitochondria induced by Ca2+was first noted by Slater and Cleland (1953) and by Tapley (1956). On the other hand, Bartley and Amoore (1958) showed that Mn2+,when taken up by mitochondria, displaced about 50%of the endogenous mitochondrial Ca2+, while siniulianeously displacing H+ ions. Tapley (1956) and Raablauf (1953) showed that Mn2+was able to inhibit mitochondrial swelling. Thus, about 10 years ago it had been established that mitochondria can accumulate large amounts of various divalent cations against a concentration gradient. The uptake process was energy dependent, utilizing either substrate oxidation or ATP as an energy source. “Massive loading” of mitochondria with divalent cations (up to 2
326
LEENA MELA
pmoles/mg protein) induced an irreversible stimulation of respiration, resembling the effect of uncoupling agents. If smaller amounts of Ca2+were added, reversible bursts of respiratory activity were noted. Caz+accumulation occurred in the presence or absence of permeant anions, such as phosphate or acetate, and in exchange with H+ ions. Ca2+ but not Mn2+ accumulation induced mitochondrial swelling. There was electron microscopic evidence indicating that the accumulated Caz+,S?+, or Baz+localized in matrix granules. Although no clear mechanism of Caz+accumulation was proposed at that time, it was noted by Vasington and Murphy (1962) that the total number of respiratory carriers and phospholipids of the mitochondrial membrane was too small to account for the binding of such large amounts of divalent cations. These authors proposed that the respiratory enzymes and high-energy intermediates acted in a “catalytic manner” to activate cation transport, but other substances would have to act as binding sites for the cations.
111.
THREE-STEP MECHANISM OF MITOCHONDRIA1 CAz+ ACCUMULATION
When Chance (1963) first proposed a three-step mechanism for mitochondrial Caz+ accumulation, he had only fragmentary evidence to support his conclusions. Many new techniques have been introduced to measure Ca2+accumulation. Particularly kinetics of the interaction of Caz+with mitochondria have become important. Spectrophotometric measurements of various Caz+ indicators were adapted for use in mitochondrial suspensions (Mela and Chance, 1968; Scarpa, 1972; Scarpa, 1975). Precise measurement of free Caz+concentrations were also was found necessary. This need introduced the utilization of Ca2+ buffers (Reed and Bygrave, l974,1975a7b;CarafoIi, 1975a; Carafoli et al., 197513).When CaZ+buffers were used, Ca2+indicator dyes were not suitable for the measurements of uptake kinetics. As a result, new quenching techniques, utilizing inhibitors of respiration and Ca2+ transport, were adapted to make kinetic measurements of Ca?+ accumulation possible (Reed and Bygrave, l974,1975a,b; Carafoli, 1975b; Carafoli et al., 1975b). In the following, I will summarize the results of these attempts to explain the mitochondrial Ca2+transport mechanism in terms of three independent steps. 1. Binding. The first step in the interaction of Caz+ with mitochondria is its binding to the membranes. This process is energy inde-
MITOCHONDRIAL CALCIUM TRANSPORT
327
pendent. Two types of binding sites have been described, high- and low-affinity sites. 2. Carrier-Mediated Accumulation of Ca2+. The second step involves the actual energy-dependent transport of Ca2+into the mitochondrial membrane. This is now known to occur via a specific divalent cation carrier. 3. Accumulation in the Matrix. Removal of Ca2+ from the membrane-bound carrier molecule occurs only in the presence of certain permeant anions. This step completes the Ca2+ accumulation process and allows Ca2+to be deposited within the matrix space either as a precipitate or in soluble form.
A. Energy-Independent Binding
The energy-independent binding of divalent cations in mitochondria was first reported by Slater and Cleland (1953) and by Chappell et al. (1963). In their studies of the accumulation of Mn2+by liver mitochondria, Chappell et d . (1963)used the enhancement of the proton relaxation rate of water as an indicator of Mn2+bound to macromolecules. From these data, they concluded that Mn2+,“accumulated” in liver mitochondria, appeared in three different forms: about 30 nanoatoms/mg of protein were bound to the surface; 200-300 nanoatoms/mg were accumulated in a respiration-dependent process in the absence of phosphate; and more than 2000 natoms/mg were accumulated when phosphate was added. In studies with rat liver mitochondria, incubated with &CaC12in the absence of respiration or ATP, Rossi et al. (1967a) concluded that energy-independent binding of Ca2+in mitochondria was pH sensitive and was also affected competitively by K+ and Na+ ion concentrations. Either H+ or K+ was released to the medium during Ca2+binding (Carafoli et al., 1965a). In an equilibrium study, Rossi et al. (1967a) found that at pH 8.0, 35 nmoles Ca2+/mgprotein and at pH 6.5, 16.3 nmoles Ca2+/mgprotein were bound energy independently. Rossi et al. suggested that, besides “surface” binding sites, some sites deeper in the mitochondrial membrane might be involved in energy-independent Ca2+binding. Competition between monovalent and divalent cations for the same mitochondrial binding sites was shown by Scarpa and Azzi (1968). Gear and Lehninger (1968), however, argue that mono- and divalent cations do not bind to the same sites in the membrane (i) because, according to their results, monovalent cation binding is accompanied by
328
LEENA MELA
a release of H+ ions, whereas Ca2+binding is not; and (ii)because they did not observe any release of bound Ca2+during monovalent cation binding. Other investigators (Wenner and Hackney, 1967; Hollander et al., 1968) also failed to demonstrate ejection of H+ ions during energy-independent Ca2+binding. Phospholipids have been proposed as binding sites for divalent cations in mitochondrial membranes (Chappell et al., 1963). By using phospholipiddepleted mitochondria and submitochondrial particles, Scarpa and Azzi (1968) showed that Ca2+binding was reversibly lowered. They also found that local anesthetics competitively inhibited mono- and divalent cation binding with a Ki of 0.2 mM. This finding also points to membrane phospholipids as Ca2+-bindingsites. HIGH-AND LOW-AFFINITYBINDING Lehninger and his collaborators (Lehninger, 1969; Reynafarje and Lehninger, 1969; Lehninger et al., 1969; Lehninger and Carafoli, 1970) first demonstrated the existence of so-called high- and lowaffinity energy-independent Ca2+-bindingsites in the mitochondrial membranes. Small amounts of Ca2+ added to rotenone and to antimycin A-inhibited mitochondria, in the absense ofATP, were taken up immediately. If the ratio of Ca2+/protein was in the order of 1-5 nmoles/mg, all the added Ca2+was taken up in a few seconds. Larger amounts of Ca2+were taken up in a slower process until saturation was reached at 50-60 nmoles Ca2+/mg protein. A Scatchard plot (Scatchard, 1949) of the data indicated two binding sites: one with a small number of sites 0.6-6 nmoles/mg protein and a very high affinity constant K, s lO-'M; and another with sites up to 50-60 nmoles/mg protein and a K , ranging from 50 to 200 p M . The number of low affinity sites correlates well with those found by Rossi et al. (1967a). The high-affinity sites, which Lehninger and his co-workers named the high-affinity, energy-independent Ca2+-bindingsites, also had other important characteristics. S?+ and Mn2+ were obviously bound at the same sites, but less efficiently (Reynafarje and Lehninger, 1969). Binding of Ca2+to these sites was inhibited by lanthanides, local anesthetics, and uncouplers such as DNP and C1-CCP (Reynafarje and Lehninger, 1969; Lehninger, 1969; Lehninger et al., 1969; Lehninger and Carafoli, 1970). The inhibition by lanthanides and local anesthetics is not surprising. However, the inhibition by uncouplers in an energy-independent reaction is harder to understand. Since endogenous Ca2+,which amounts to 4-6 nmoles/mg protein, was also released by uncouplers, Lehninger et al. (1969) concluded
329
MITOCHONDRIAL CALCIUM TRANSPORT
that endogenous Ca2+is bound to the high-affinity binding sites, and that lanthanides accelerate the release of endogenous Ca2+(Carafoli and Rossi, 1971). The low-affinity Ca2+binding sites appear to be phospholipids of the mitochondrial membrane (Carafoli and Rossi, 1971; Lehninger et al., 1969; Scarpa and Azzone, 1968), and more specifically polar headgroups of phospholipids (Reed and Bygrave, 1974). The nature of the high-affinity binding site is uncertain. Recent independent evidence from two laboratories does not support the existence of the energy-independent high-affinity Ca2+binding site in mitochondria. Akerman et al. (1974) were unable to demonstrate the existence of high-affinity Ca2+binding when the “energized state” of mitochondria had first been relaxed by treatment with carefully titrated amounts of inhibitors, rotenone and antimycin A. Moreover, they found that high-affinity binding was lost in the presence of the Ca+ ionophore X537A. The authors concluded that the so-called high-affinity binding actually represents energy-dependent transport of Ca2+to the inner membrane. Reed and Bygrave (1974) came to the same conclusion by different experimentation. They separated the external Ca2+binding (“surface” binding) from “transport” to the internal phase by the use of EGTA washes after incubation with 45Ca2+.They found that the “highaffinity” binding (inaccessible to EGTA) was not inhibited by respiratory inhibitors, but was inhibited by uncouplers in the presence of respiratory inhibitors and also by La3+and ruthenium red. They, thus, concluded that the high-affinity binding was not energy-independent, but represented Ca2+transported to the internal phase of mitochondria. The energy supply of this transport comprises the energy reserve of inhibited mitochondria and is, therefore, limited (Chance et al., 1969a; Azzi and Chance, 1969). CA~+-BINDING PROTEINS
AND
THEIRPOSSIBLE FUNCTION
Recently, a few laboratories have independently isolated Ca2+binding factors from the mitochondrial membranes. Lehninger (1971) first reported the isolation of a water-soluble fraction from liver mitochondria by means of mild osmotic shock. This fraction was capable of binding Ca2+with high affinity. The component was heat labile, and its binding capacity was inhibited by lanthanum ions. In 1972, Gomez-Puyou et al., studying the soluble protein fraction first isolated by Lehninger, found that it contained carbohydrates. Carbohydrates were also found to be part of Ca2+-bindingproteins isolated by
330
LEENA MELA
Tashmukhamedov et al. (1972) and Kimura et al. (1972). Sottocasa et a2. (1971, 1972) demonstrated the presence of glycoproteins in liver mitochondria. These proteins were extracted and purified (Carafoli et al., 1972; Sottocasa et al., 1972, 1974; Carafoli and Sottocasa, 1974; Carafoliet al., 1975a; Carafoli, 1975b).The glycoprotein has an MW of approximately 33,000 daltons. Ten percent of it is carbohydrate. It contains one sialic acid residue per mole and has phospholipids attached to it. The glycoprotein represents 1% of total mitochondrial protein. The isolated gl ycoprotein was found to exhibit both high- and low-affinity Ca2+binding. A Scatchard plot of Ca2+binding to the beef liver mitochondrial glycoprotein (Carafoli and Sottocasa, 1974) showed 2.6 nmoles of high-affinity sites/mole of protein and a & of 0.15 p M . The number of low-affinity sites was 20 nmoles/mole of protein, with a Kd of 9.1 p M . Ca2+binding to the protein was inhibited completely by La3+ ions and 72% by ruthenium red (Carafoli and Sottocasa, 1974; Carafoli, 197513). Where in the mitochondria this glycoprotein occurs has not been clearly established. Carafoli and Sottocasa (1974) and Carafoli (1975b) indicate that it appears in both inner and outer mitochondrial membranes. Recently, Hackenbrock and Miller (1975) used polycationic ferritin to demonstrate electron microscopically the location of anionic sites on mitochondrial membrane. They found that anionic sites appear in high density on the surface of the inner boundary membrane. It is that part of the outer surface of the inner membrane that directly faces the outer membrane. The cristae membrane surfaces also contained anionic sites, but at lower density. These data are in good agreement with the findings of Brdiczka et a2. (1974) that the carbohydrate-containing proteins of the inner mitochondrial membranes appear to be localized in the inner boundary membrane. It is not certain, however, whether these anionic sites represent the Ca2+-bindingglycoproteins isolated from mitochondria.
ROLE OF ENERGY-INDEPENDENT CA2+BINDINGIN CA2+ACCUMULATION The relationship of Ca2+binding and the Ca2+-bindingproteins to the actual energy-dependent Ca2+accumulation process has not been established. Chance (1965a) and Scarpa and Azzi (1968)proposed that energy-independent Ca2+ binding would constitute the first step in Ca2+ accumulation. Scarpa and Azzone (1968) also put forward the possibility that the initial phospholipid-binding site could deliver
33 1
MITOCHONDRIA1 CALCIUM TRANSPORT
Ca2+to some other site, perhaps a carrier, involved in the actual translocation of Ca2+.They proposed this possibility because of the discrepancies between the K,: s of actual translocation and the surface binding. Similarly, Carafoli (1975b) believes that the membrane glycoprotein would act as a “recognition” site for Ca2+,permitting the cation to reach the actual carrier site, or possibly to be transported across the membrane without a specific carrier in response to the membrane potential. In a “reconstitution” experiment, Carafoli (1975b) added the isolated Ca2+-bindingglycoprotein to lecithin bilayers and measured the electrical conductance GM in the presence of Ca2+. Increased electrical conductance was induced by the glycoprotein, but not by immunoglobulin, horseradish peroxidase, or ovalbumin, which were used as controls. Ruthenium red, added with the glycoprotein, alsd inhibited the increase in conductance. However, Carafoli (197513) was unable to demonstrate any Nernst potentials related to transmembrane movements of Ca2+.Thus, any function of the Ca2+-bindingproteins as Ca2+carriers has so far not been demonstrated, although their high affinity for Ca2+,as well as their La3+and ruthenium red sensitivity, are suggestive of this function. 6. Carrier-Mediated Accumulation of Ca2+
As stated previously, this process is energy dependent and occurs in the absence of added permeant anions. This process can be dealt with as two separate reactions: (i) The Ca2+carrier function; (ii) interaction of Ca2+carrier function with mitochondria1 energy sources. EVIDENCEFOR THE EXISTENCEO F
A
CA2+CARRIER
Chance and co-workers (Chance, 1963; Chance, 1965a; Chance and Yoshioka, 1965) proposed a carrier-mediated Ca2+transport where the high-energy intermediate X I acted as the Ca2+carrier. This proposal was based on the large redox changes of cytochrome b upon addition of Ca2+to respiring mitochondria. Chance (1965a) also stated that there was no “special calcium site” that was involved in ion transport but not in oxidative phosphorylation. This statement was based on the finding that the crossover point in the respiratory chain was between cytochromes b and c in the reaction of the chain with both ADP and Ca2+.Since the energy requirement for Ca2+was measured to be twice that for ADP, Chance (1965a) proposed that there was a common intermediate that had two binding sites for Ca2+and one for ADP and phosphate.
-
332
LEENA MELA
Later experiments on the specific inhibition of Ca2+accumulation, however, provide evidence for a specific and separate Caz+carrier in the mitochondrial membrane. Lanthanides (Mela, 1968a, 1969a,b; Mela and Chance, 1969)and ruthenium red (Moore, 1971)were found to specifically inhibit mitochondria1 Ca2+accumulation at concentrations low enough to have no effect on oxidative phosphorylation or other energy-linked processes. These data indicate that the specific lanthanide- and ruthenium red-sensitive site, which obviously is involved in divalent cation transport, cannot be the same high-energy intermediate involved in oxidative phosphorylation. Based on the specific inhibition of Ca2+transport in mitochondria (monovalent cation transport was not inhibited) by lanthanides at concentrations around 0.07 nmole/mg protein, Mela (1968a, 1969a,b) and Mela and Chance (1969) proposed that the lanthanide-binding site of mitochondrial inner membrane constituted a specific Ca2+ carrier. They calculated that the molar concentration of these sites in the mitochondrial membrane was 0.07 nmole/mg and the average distance between adjacent sites as 300 (Chance et al., 196913). Moore (1971) proposed that the site blocked by ruthenium red constituted the divalent cation carrier and was presumably a glycoprotein. Vainio et al. (1970) showed that the divalent cation specific carrier site was capable of transporting Ca2+,S P Y Mn2+,and Ba2+into mitochondria. Lehninger et al. (Reynafarje and Lehninger, 1969; Lehninger and Carafoli, 1970) proposed that the high-affinity Ca2+ binding site, in actuality, represented the same mitochondrial membrane site as the lanthanide-sensitive Ca2+carrier. As discussed above, the existence of the “high-affinity energy-independent” Caz+ binding site now is in doubt (Hikermanet al., 1974; Reed and Bygrave, 1974). It appears that the “high-affinity binding sites,” in actuality, indicate uncouplersensitive energy reserve of inhibited mitochondria, rather than the number of Ca2+carrier sites. Thus it does not, at least at the present time, seem probable that these two findings have much in common, as was already pointed out b y Mela and Chance in 1969. It is hoped that some of the recently isolated Ca2+-bindingglycoproteins, which also are sensitive to lanthanide and to ruthenium red, will provide more information about the nature of the Ca2+carrier function. Further evidence has since emerged to support the existence of carrier-mediated transport of Ca2+ in mitochondria. Besides (i) the specific inhibition of Ca2+transport system, discussed above, various other criteria of carrier-mediated transport are met in mitochondrial Ca2+accumulation. (ii) The specificity of the carrier has been shown (Vainio et al., 1970). (iii)High affinity of the Ca2+transport system has recently been studied in detail by a number of investigators. Reed and
MITOCHONDRIA1 CALCIUM TRANSPORT
333
Bygrave (1975a) reported that the K , of rat liver mitochondria for Ca2+ was 4 pM and the K , ofLucilia flight muscle mitochondria was 5 pM (Bygrave et al., 1975). Carafoli and co-workers (Carafoli e t al., 1975a,b; Carafoli, 197513; Jacobus et al., 1975) found the K , for heart mitochondria to be in the range of 1p M . (iv) Substrate saturability is another criterion for a carrier system and has been shown to exist in mitochondrial Ca2+transport (Spencer and Bygrave, 1973; Reed and Bygrave, 1975a; Vinogradov and Scarpa, 1973; Rossi et a1 ., 1974; Alexandre et at., 1974). Although no direct evidence has, as yet, appeared to support the existence of a membrane-bound Ca2+ carrier in mitochondria, the many points of indirect evidence, summarized above, emphasize its existence. Several technical difficulties have to be overcome before conclusive evidence will be forthcoming. Some of the critical unanswered problems are the measurement of the exact amounts of free and bound Ca2+in the mitochondrial compartments, the precise concentration of the carrier in the membrane, and ultimately its purification and functional characterization. INTERACTION OF CA2+CARRIER FUNCTION WITH MITOCHONDRIAL ENERGYSOURCES The Ca2+carrier system of the membrane provides the mitochondria a means of accumulating divalent cations from the surrounding medium, with high affinity and specificity, and against a large concentration gradient. Whenever the mitochondrial “energy store” is not rate limiting, Ca2+carrier function is directed from the surrounding medium into the mitochondria (Chance, 1965a; Drahota e t a1 ., 1965; Chance et a1 ., 1969b; Lehninger, 1970; Lehninger et al. 1967). If the energy store is relaxed, or respiration is inhibited in the absence of ATP, accumulated Ca2+is released from the mitochondria and no uptake occurs, even at the level of small amounts with high affinity (Akerman et a1 ., 1974). Recently, Vasington e t al. (1972) and Rossi e t al. (1973,1974) have shown that Ca2+uptake and release occur via the same carrier system. These studies included binding of Ca2+at a specific membrane site during the Ca2+flux, before its completion. The earlier proposal of Chance and collaborators (Chance, 1963, 1965a; Chance and Yoshioka, 1965),that the high-energy intermediate X I acted as a Ca2+carrier explained the response of respiratory enzymes to interactions of Ca2+with mitochondria. Since then, the picture of the energy-linked nature of Ca2+accumulation has been clarified. Were Ca2+to be transported via a mechanism of direct chemical in-
-
334
LEENA MELA
teraction of Ca2+with the high-energy intermediate, the stoichiometry of Ca+: would always be the same. This, however, is not the case. At high monovalent cation concentrations or at high pH, the Ca: ratio can greatly exceed 2, if no permeant anions are present (Carafoli e t al., 1965; Carafoli e t al., 1967; Gear and Lehninger, 1968).At low pH and low monovalent cation concentration, the ratio Ca2+: can be less than 2 (Carafoli et al. 1966). Thus, it appears that the carrier-mediated Ca2+accumulation is electrogenic, as suggested by Lehninger (1970) and by Chance and Montal (1971). Selwyn e t ul. (1970) showed in experiments in the absence of a respiratory energy source that Ca2+ entered the mitochondria on an electrogenic uniporter or in exchange for one K+ ion, This entry was blocked by praseodymium, which also inhibits the Ca2+carrier (Mela, 1968a, 1969a,b). Thus they concluded that Ca2+uptake is governed by the proton pump of Mitchell (1968) and of Mitchell and Moyle (1969), where respiration or ATP was required for the proton translocation. The primary driving force of Ca2+accumulation has been a focus of considerable debate. Basically three different proposals concerning the energetics of mitochondrial calcium transport have been formulated. These proposals, naturally, are offshoots of the different mechanistic views of oxidative phosphorylation, and will remain unresolved until those primary arguments are settled. According to Mitchell (1968), the driving force of ion transport is the proton motive force or the membrane pH gradient, developed by protons being translocated (extruded) during electron flow in the respiratory chain. Thus, the proton pump is primary to the cation pump. Chappell and Grofts (1965) supported the proton pump hypothesis, but argued that it could be driven by the high-energy intermediate x I. Chance and his collaborators (Chance, 1963, 1965a; Chance et al., 1968; Rasmussen et al., 1965), however, proposed that the cation pump was primary, energetically driven by the high-energy intermediate. The H+ ion translocation was secondarily induced by the cation transport. By comparing Rb+ and Ca2+concentration gradients under the influence of various ionophores, such as valinomycin (Moore and Pressman, 1964), Nigericin (Lardy e t al., 1967) and A 23187 and X 537 A (Reed and Lardy, 1972), which induce permeability to mono- or divalent cations, and by assuming that the Rb+ distribution in the presence of valinomycin follows the Nemst potential, Rottenberg and Scarpa (1974) concluded that mitochondrial Ca2+accumulation is an electrogenic process driven by the membrane potential. Operation of
-
-
-
-
MITOCHONDRIA1 CALCIUM TRANSPORT
335
an uncharged carrier for Caz+, with no compulsory exchange with other anions, fits this scheme. When permeant anions, such as acetate or phosphate, are not present, only limited amounts of Ca2+ can be accumulated. Under these conditions, one H+ ion is released when one Ca2+ is accumulated (see earlier discussion above and the review b y Lehninger et al., 1967). Then, during Caz+-activated electron transport, two protons are ejected per pair of electrons being transferred through the chain. Simultaneous alkalinization of the mitochondrial membranes was noted (Mela, 1966; Chance and Mela, 1966; Rossi et al., 1966; Lynn and Brown, 1966a,b; Addanki et al., 1968; Addanki and Sotos, 1969). With increasing alkalinization, shown to be stoichiometric with the number of H+ ions ejected (Rossi et al., 1966, 1967b; Carafoli and Rossi, 1967; Gear et al., 1967)and to amount to a gradient of approximately 1 pH unit between the outside and the mitochondrial membrane, Ca2+accumulation became slower and finally ceased (Chance and Mela, 1966; Mela, 1968a, 1969a,b; Chance and Yoshioka, 1966). Respiration also became inhibited (Chance and Schoener, 1966; Chance and Mela, 1966; Mela, 1968a; Rossi and Azzone, 1968). Thus, this stage of high membrane alkalinization caused by a large uptake of Ca2+(up to 200 nmoles/mg) (Chance and Mela, 1966; Mela, 1968a) in the absence of permeant anions constituted state 6 of respiratory activity, earlier explained b y Chance and Schoener (1966) as the inhibited state of respiration and the state of high oxidation of the respiratory carriers. In this state, cytochrome b exhibited a low absorbance at 560 nm; the K , of this Caz+ reaction was 15 pM (Chance and Schoener, 1966). Chance and Mela (1966) concluded that the inhibited state 6 was due to the high alkalinity of the mitochondrial inner membrane. Thus, it appeared that release of state 6 was necessary for the collapse of the large pH gradient, for reactivation of the respiratory chain, and to initiate further Ca2+accumulation. C. Accumulation of CaZf in the Matrix
When the mitochondrial membrane reaches a limit of Ca2+accumulation in the absence of permeant anions, state 6 is reached (Chance and Schoener, 1966). This state can be released by an addition of permeant anions, such as phosphate or acetate, or by releasing the accumulated Ca2+from the membrane by inhibitors or uncouplers of respiratory activity, or by Caz+ ionophores. Chance and collaborators showed (Chance, 1965a; Chance and Schoener, 1966) that phosphate added to inhibited mitochondria in
336
LEENA MELA
state 6 initiated respiratory activity and returned the highly oxidized respiratory carriers to a more reduced level. He also showed that respiratory inhibition was released by uncoupling agents. Chance and Mela (1966), Gear et al. (1967), Rossi et a,?. (1966), and Ghosh and Chance (1970), by parallel determinations of extra- and intramitochondrial H+ and OH+ ions, demonstrated that both permeant anions and uncoupling agents released the large pH gradient of the mitochondrial membrane. Chance and Azzi (1969) also indicated that certain anions, such as succinate and acetate, promoted the movement of Ca2+ from the membrane compartment where the pH indicator bromothymol blue (BTB) is localized (Chance and Mela, 1966; Ghosh and Chance, 1970) to the pyridine nucleotide responsive space. Uncouplers released Ca2+from both sites. Mela (1969b),by means of parallel measurements of membrane alkalinity by BTB and Ca2+accumulation by murexide, found that permeant anions removed Ca2+from the BTB site, presumably, to the matrix, whereas uncouplers released it to the extramitochondrial medium. Electron microscopic evidence was provided by Hackenbrock and Caplan (1969) of Ca2+accumulation in different forms in the absence and presence of phosphate. In the absence of phosphate, Ca2+accumulated to a maximum of 190 nmoles/mg protein and led to the formation of dense matrix inclusions, which were composed of tightly packed membranes. No ultrastructural expansion of mitochondria was noted. When phosphate was present, up to 200 nmoles of Ca2+/mg protein was found in an osmotically active form in the water phase of the inner compartment. Ultrastructural swelling of the inner mitochondrial compartment occurred. According to the electrogenic model of Ca2+accumulation, Selwyn et al. (1970) indicated that counterion transport, which is an electrically neutral process, neutralizes the pH difference generated by respiration. Reed and Bygrave (1975a) elaborated on this process in a kinetic study of Caz+ accumulation in the presence and absence of phosphate. Using Ca2+/nitriloacetic acid buffers, which enabled them to determine accurately free Ca2+concentrations (Reed and Bygrave, 1975b), they found that the Ca2+concentration/accumulation velocity curve was sigmoidal, with a Hill coefficient of 1.7 and a K, of 4 pM at 0°C and pH 7.4. The Hill coefficient and K , remained relatively constant in the presence of phosphate up to 2 mM. From the logarithm of velocity/K, versus pH plot, these investigators concluded that the initial binding of Ca2+to the carrier had a pK of 7.5 f 0.5 at 0°C. Comparing the initial velocities of Ca2+transport in the absence or presence of varying concentrations of phosphate, Reed and Bygrave found
MITOCHONDRIAL CALCIUM TRANSPORT
337
that the initial velocity had increased with increasing phosphate (up to 2 mM). Since only the velocity, and not the Hill coefficient or K,, was altered by addition of electrically neutral species, such as phosphoric acid or acetic acid, Reed and Bygrave (1975a) concluded that the only way anions can affect the rate of Ca2+transport is by affecting the dissociation of Ca2+from the carrier (pK, 7.5, rate constant for Ca2+ binding, is largest at high pH). Thus, they concluded that this phase of the reaction constitutes the rate-limiting step. Their conclusion was strengthened by the finding that a dissociated species SCN- (which penetrates the inner membrane in the dissociated form) had no effect on the velocity of Ca2+accumulation. Thus, there now appears to be sufficient evidence for a mechanistic scheme of Ca2+ accumulation in mitochondria. The accumulation occurs in three steps. The first step constitutes binding of Ca2+ to inner membrane Ca2+-bindingsites. These sites may include specific carrier sites and/or unspecific Ca2+-bindingsites. The second step, which involves Ca2+ transport mediated by the specific membrane bound carrier, is electrogenic and requires generation of a proton pump via energy derived from mitochondria1 respiration or ATP hydrolysis. The final step in the completed Ca2+transport is the release of Ca2+from the carrier. This step now appears to be rate limiting in Ca2+accumulation. Electroneutral acids, such as acetate or phosphate, can decrease the internal pH sufficiently for the dissociation of Ca2+ from the carrier to occur, thus causing Ca2+accumulation in the mitochondrial matrix (Reed and Bygrave, 1975a).
IV.
ROLE OF MITOCHONDRIA IN THE PHYSIOLOGICAL CONTROL OF CELLULAR CA2+ CONCENTRATION
For important physiological functions of practically all mammalian cells to proceed uninhibited, the intracellular free Ca2+ion concentration must be closely controlled. Essentially, three cellular structures may be responsible for this control: (i) the cell membrane or plasmalemma, (ii) the endoplasmic reticulum, and (iii) the mitochondria. Parallel with the Na pump, the Ca2+pump of the cell membrane provides an important control of intra-extracellular cation distribution in mammalian cells (Coraboeuf and Vassort, 1968; Reuter and Scholz, 1968). The very specialized function of the endoplasmic reticulum of various types of muscle, the sarcoplasmic reticulum, is well established as a Ca2+ regulator in muscle cells, and unquestionably so in skeletal muscle (for a review see Martonosi. 1972). However, the endoplasmic
338
LEENA MElA
reticulum of other mammalian cells has very low Ca2+uptake capacity and, therefore, is unlikely to act as a Ca2+regulator. The role of mitochondria in controlling cellular free Ca2+has been challenged as well as emphasized, and this debate is by no means settled. I n the following, the evidence, which often is indirect, for the control of cellular free Ca2+by mitochondria will be discussed in detail. For the cellular Ca2+controller to be able to function adequately, the following criteria of both Ca2+uptake and release have to be met: (i) Ca2+uptake rates have to be adequate for fast removal of Ca2+;(ii) the affinity for CaZ+has to be high; (iii)the uptake capacity has to meet the requirements of the particular cell; (iv) a rapid release of Ca2+to the cytoplasm has to be accomplished when necessary. All of these points will be discussed separately. A. Kinetics of Mitochondria1 Ca2+ Accumulation
Using rapid mixing apparatus and spectrophotometric recording with short time constants, Chance and his collaborators first showed that Ca2+ reacted with components of the respiratory chain with half-times in the millisecond range (Chance, 1965a; Mela, 1968b; Vainio et al., 1970). Mela showed that reduced cytochrome b, for instance, reacting with Ca2+,reached a more oxidized steady state with a half-time of 20 msec, while its reaction with added oxygen had a half-time of 200 msec (Mela, 1968b). Since then direct measurements of Ca2+ concentration change s in the extramitochondrial medium have become available by the use of various Ca2+indicators (Mela and Chance, 1968; Scarpa, 1972; Vainio et al., 1970; Vinogradov and Scarpa, 1973; Scarpa, 1974; Scarpa, 1975), and kinetic measurements of Ca2+uptake rates have been performed. The experiments of Mela et al. (Mela, 1969b; Vainio et al., 1970) indicated that the initial rates of Ca2+accumulation by liver mitochondria at fairly high Ca2+ concentration (- 200 /AM) were about 4 nmoles/sec/mg protein in the absence of phosphate and about twice that (-8 nmoles/sec/ mg) in the presence of phosphate (Mela, unpublished). Vinogradov and Scarpa (1973) and Scarpa (1974) obtained similar rates in the presence of phosphate at high Ca2+concentrations (-200 p M ) ,but considerably lower initial rates at low Ca2+ concentration, 0.1-0.3 nmoles/sec/mg protein at 5-15 /AM Ca2+ added. Thus, a sigmoidal relationship between the Caz+concentration and uptake velocity was found. Scarpa and Graziotti (1973) showed that the initial rates of Ca2+uptake by heart mitochondria were similar to those of liver at
MITOCHONDRIA1 CALCIUM TRANSPORT
339
high Ca2+ concentration (-200 p M ) but very much lower at 5 pM Ca2+, exhibiting under those conditions uptake rates of only 0.05 nmoles/sec/mg protein. Scarpa (1975) used a new metallochromic indicator, arsenazo 3, which is much more sensitive to, although less specific for Ca2+and can, thus, under appropriate conditions be used to detect smaller Ca2+changes. These studies led Scarpa to similar conclusions. Thus, mitochondria obtained from 1 gm of cardiac muscle can accumulate only 1 nmole of Ca2+during the relaxation time of mammalian myocardium which is 200 mseconds. Scarpa, thus, concluded that mitochondria alone were unable to remove Caz+ from the cytoplasm fast enough for beat-to-beat control of cardiac contractions. All these experiments, described above, have serious shortcomings, however. They were all performed at 26"C, not 37-38°C; the medium used contained high concentrations of sucrose rather than the salt solutions that are characteristic of intracellular fluid; Scarpa's experiments were all done at high Mg2+ concentrations in an attempt to affect the binding of Ca2+to unspecific sites or compounds (such as the indicators themselves). High MgZ+ concentrations are known to inhibit Ca2+accumulation in isolated heart mitochondria (Carafoli et al., 1975b; Jacobus et al., 1975; Sordahl, 1974). The indicator experiments have two other major deficiencies: (i) The exact concentration of free Ca2+ cannot be determined because Ca2+ buffers cannot be used (Reed and Bygrave, 1975b), and (ii) only the energy-linked accumulation can be measured. All the experimental data with Ca2+indicators indicate that murexide or arsenazo 3 (Mela and Chance, 1968; Mela, 1969a,b; Vinogradov and Scarpa, 1973; Scarpa, 1974, 1975) fail to detect any changes occurring during the removal of Ca2+from the extramitochondrial medium to mitochondria1 binding sites, the socalled energy-independent binding. The reason could be that Ca2+ bound to the energy-independent sites on the surface of the membrane is still available to the Ca2+ indicators. Inhibition of the energy-independent binding b y Ca2+indicators also presents itself as a possible problem. The amount of Ca2+bound to the actual carrier sites may be very small (0.05-0.07 nmoles/mg protein), a concentration below the sensitivity of the Ca2+ indicators (Vinogradov and Scarpa, 1973; Scarpa, 1975). The energy-independent binding, which precedes the energy-dependent accumulation of Ca2+,might occur rapidly enough to provide the necessary uptake of small amounts of free cytoplasmic Ca2+to allow relaxation of the heart muscle fibers. It seems unlikely that energy-independent binding would occur so rapidly as to be essentially completed during mixing and, therefore, be missed by the indicator method. Murexide experiments done in the
340
LEENA MELA
presence and absence of inhibited mitochondria in the medium yield similar deflections for equal amounts of Ca2+ added (Mela and Chance, 1968; Vinogradov and Scarpa, 1973), thus excluding the possibility of measurable amounts of undetected Ca2+ bound in an energy-independent process. Thus, although the presently available evidence is by no means conclusive, kinetic studies of mitochondrial Ca2+ accumulation suggest that, in the mammalian heart, mitochondrial Ca2+ uptake is not rapid enough to provide beat-to-beat control of sarcoplasmic free Ca2+.However, until now in vitro experiments have not been performed under optimal conditions and, thus, are not directly applicable to the in vivo situation. B. Mitochondhl Affinity for Ca2+
The use of indicator methods for measurements of mitochondrial Ca2+accumulation and, particularly, of the affinity for Caz+,has been seriously criticized. Scarpa and collaborators (Vinogradov and Scarpa, 1973; Scarpa and Graziotti, 1973; Scarpa, 1974; Scarpa, 1975) found the K , for Ca2+to be in the range of 40-100 pA4 for heart and 50-70 pM for liver mitochondria. Chance had already found in 1963 (Chance, 1963), by measuring the redox changes of the respiratory chain components, such as cytochrome b , that the K , of this reaction for Ca2+was less than 5 pM. Camfoli and Azzi (1972) also used the redox shift of cytochrome b to determine the K , for Ca2+in the presence of Ca-EGTA buffers, and found a value of 3 pM. Bygrave et al. (1971a,b) and Spencer and Bygrave (1973), who by measuring the uptake of Ca2+by atomic absorption and by determining the accumulated radioactivity of 45Cain the presence of ATP, were able to calculate from the known stability constant of Ca ATP2- the added “free” Ca2+,concluded that the affinity for Ca2+was very high, with a K , of about 2 pM. More recently Reed and Bygrave (1975a) who used Ca2+/nitriloaceticacid buffers for more accurate determination of free Ca2+and ruthenium red quenching to stop the reaction, found the K , to be 4 pM at 0°C and pH 7.4. Although the velocity of Ca2+accumulation increased with increasing phosphate concentration up to 2 mM, the K , remained unaltered. Acetate, on the other hand, at 10 mM increased the sigmoidicity of the l/velocity versus l/concentration plot and increased the K , to 6 pM. Jacobus e t al. (1975) reexamined the redox changes of cytochrome b spectrophotometrically and found that the use of CaZ+-EGTAbuffers to stabilize the free Ca2+concentration at low levels dropped the half-saturation constant of 40 pM obtained
MITOCHONDRIA1 CALCIUM TRANSPORT
34 1
in the absence of Ca2+-EGTA buffers to a low K , value of 4.16 x l P M in heart mitochondria, indicating that the K , may be considerably smaller than 1 p M . The recent finding of Carafoli e t al. (197513) that rat heart mitochondria are capable of extracting nearly all of the Ca2+bound to the Ca2+-bindingcomponent of troponin (Drabikowski et a1 ., 1971, 1973, 1974) in an energy-dependent process is an indication of the high affinity of the mitochondrial Ca2+ accumulation system. With the exception of the data of Scarpa and collaborators (1973-1975), all of the experimental evidence reviewed above indicates that the affinity of mitochondria for Ca2+is high enough to make them a possible and attractive candidate for the cellular Ca2+ regulator. Other authors have recently discussed this point (Carafoli, 1975a; Carafoli et al., 1975b; Jacobus et al., 1975) and emphasized the fact that, although none of the existing evidence conclusively supports the role of mitochondria in cellular control of free Ca2+concentration, for instance in the mammalian heart, it is suggestive. Moreover, none of the evodence goes directly against this possibility. C. Mitochondria1 Capacity to Accumulate CaZt
Since the early experiments of Slater and Cleland (1953), Vasington and Murphy (1961, 1962), and DeLuca and Engstrom (1961), it has been known that mitochondria can accumulate large quantities of Ca2+.Energy-independent binding of Ca2+ can amount to about 40 nmoles/mg protein (Rossi e t al., 1967a), while mitochondria can accumulate more in an energy-dependent process. The maximum amount accumulated in the absence of phosphate can reach 100-150 nmoles/mg protein (Chance, 1965a; Chance and Schoener, 1966). This process does not induce mitochondrial swelling or damage to the membrane (Chance and Mela, 1966; Hackenbrock and Caplan, 1969). Even larger quantities can be accumulated in the presence of permeant anions, such as phosphate. The total amount of Ca2+or Mn2+ uptake can reach 2 pmoleslmg protein (Chappell et al., 1963; Lehninger et al., 1963; Rossi and Lehninger, 1964). Under these conditions, massive swelling of mitochondria occurs (Lehninger, 1962; Hackenbrock and Caplan, 1969) and the membranes rupture. Such massive Ca2+accumulation cannot, therefore, occur in vivo without damage to the mitochondria, and cell death. On the other hand, when performing a physiological function of removing free Ca2+from the cytoplasm, mitochondria would only have to utilize a small fraction of their large Ca2+accumulation capacity. For instance, in the mamma-
342
LEENA MELA
lian heart, beat-to-beat control of cytoplasmic free Ca2+,would involve removal of only about 60 nmoles of Ca2+/gm of tissue during relaxation (Ebashi et al., 1968; Kubler and Shinebourne, 1971).Since about 60-80 mg of mitochondrial protein can be isolated per gram of heart tissue (Scarpa and Graziotti, 1973; Mela, unpublished), about 1nmole of Ca2+will have to be accumulated per milligram of mitochondrial protein per beat (-200 msec). This amount constitutes only a small portion of the full mitochondrial capacity to bind (1/40) or accumulate (1/100-1/2000) Ca2+. Thus, both the affinity and capacity of mitochondria for Ca2+,as analyzed in the in vitro experiments, appear adequate for the most demanding cellular circumstances of efficient removal of free Ca2+from the cytoplasm. The only aspect that has not been adequately tested and proven is the kinetic one. All the “fast kinetics” measured by means of metallochromic indicators seem to have shortcomings. Not only is the suitability of murexide and, particularly, arsenazo 3 (high affinity to Ca2+)for these studies questionable, but also other experimental conditions need improvement. None of the experiments of Scarpa and co-workers (Scarpa, 1974, 1975; Scarpa and Graziotti, 1973; Vinogradov and Scarpa, 1973) or of Mela et al. (1968a,b, 1969a,b) and Vainio et al., (1970) utilized a medium adjusted to simulate intracellular conditions, and the temperature was not physiological. Various other faults can be found. The experiments of Mela (1968a,b) indicate that a Ca2+-inducedreaction of cytochromes b and c occurs, when no Ca2+accumulation is indicated by murexide under conditions where La3+partially inhibits energy-dependent Ca2+accumulation. This reaction seems to be complete within 1 second. However, no fast kinetics can be resolved from those experiments. What amount of Ca2+accumulation (if any) at what rate these changes represent, is, of course, impossible to say presently. However, these and several other questions of similar nature still leave the kinetic argument of mitochondrial role in cellular Ca2+control unanswered. D. Release of Ca2+ fmm Mitochondda under Physiological Conditions
This is the area where our present knowledge of the physiological role of mitochondrial control of cellular free Ca2+is most inadequate. Certain agents are known to induce rapid release of Ca2+from mitochondria. Uncoupling agents and ionophores are such compounds (Chance, 1965a; Drahota et al., 1965; Chance and Mela, 1966; Reed
MITOCHONDRIA1 CALCIUM TRANSPORT
343
and Lardy, 1972; Binet and Volfin, 1975). Recently, it has also been shown that the specific inhibitor of mitochondrial Ca2+ uptake, ruthenium red, induces Ca2+release (Rossi et al., 1973). All these agents, however, although potent in releasing Ca2+from mitochondria, are totally unphysiological . Thus, their function as modulators of the release of mitochondrial Ca2+can be ignored in this discussion. We will have to turn our attention to agents that can exist in the cell under physiological conditions. And here we only have some uncertain candidates. Hormonal control of mitochondrial Ca2+ uptake and release has been suggested several times (Rasmussen, 1970, 1971; Rasmussen and Nagata, 1970; Kurokawa and Rasmussen, 1973; Borle, 197213, 1973a). However, the experimental basis of specific Ca2+-releasing factors (calcitonin, parathyroid hormone, vitamin D) is quite uncertain (Rasmussen and Nagata, 1970; Borle, 1973a,b; DeLuca et d., 1961, 1962; Kimberg and Goldstein, 1967; Kimmich and Rasmussen, 1969; Sampson et al., 1970; Matthews et al., 1972). Rasmussen also has proposed, but not proven, that the direct regulator of mitochondrial Ca2+ influx and efflux is cyclic AMP(cAMP) (Rasmussen, 1970; Kurokawa and Rasmussen, 1973). It is obvious that cAMP is intricately involved in the control of various cellular responses involving Ca2+, but whether it has a direct effect on the interaction of Ca2+with the mitochondrial membrane is still uncertain. Cyclic nucleotide-induced Ca2+release from some intracellular stores has been shown to occur in perfused liver (Friedman and Rasmussen, 1970) and in barnacle muscle fibers (Cheng and Chen, 1975). Direct studies indicating release of Ca2+from mitochondria due to cAMP incubation have appeared (Borle, 1973a,b, 1975a). Unfortunately, repeated efforts of other investigators to confirm these observations have been unsuccessful (Haugaard, personal communication; Mela, unpublished observations). One recent report indicates that cyclic AMP and cyclic GMP-induce Ca2+release from isolated mitochondria (Andersson et al., 1975). However, the controversial issue of the role of cyclic nucleotides as modulators of mitochondrial Ca2+release remains unsettled. In a systematic search for physiological components that may modulate mitochondrial Ca2+release, Chudapongse and Haugaard (1973) studied all intermediates of the glycolytic pathway and found that phosphoenol pyruvate at 1mM concentration facilitates the release of Ca2+frommitochondria. Whether this intermediate plays any role as a mitochondrial Ca2+-releasingagent in vivo is not known. Recent experiments by Chance have utilized bacterial lumines-
344
LEENA MELA
cence to measure oxygen tension of a suspension of mitochondria, and to record the uptake or release of Ca2+as the oxygen pressure varied (Chance, 1976). These experiments showed that as the p02fell below lo+ mm Hg, mitochondrial accumulated Ca2+was rapidly released. Parallel experiments (Chance et al. 1974; Chance, 1976),utilizing the perfused heart and measuring myoglobin oxygenation and deoxygenation in situ during contraction-relaxation, indicated that the cellular (myoglobin region) p 0 dropped during the contraction phase to practically zero. It, therefore, seems possible that, in the heart, the mitochondrial release and subsequent uptake of Ca2+maybe regulated by the natural physiological fluctuations of cellular oxygen. These fluctuations may be large enough and reach low enough PO values to lead to mitochondrial Ca2+release. It is also obvious from the work of Chance that, as the Oztension drops, Ca2+releaseoccurs at higher p 0 2 levels than those capable of inducing inhibition of oxidative phosphorylation (Chance, 1976). In summary, we have emphasized above the various characteristics of mitochondrial Ca2+accumulation and release in relation to their physiological role in cellular control of free Ca'+. The high affinity and capacity of the mitochondrial Ca2+accumulation system are factors that favor its cellular importance. However, the kinetic experiments up-to-date indicate inadequacies in the system, particularly in tissues where rapid removal and release of cellular Ca2+is required for undisturbed and continuous physiological function. Also the present evidence for the regulator of rapid release of Ca2+from mitochondria is far from conclusive. Thus, at the present time, it is not possible to determine the extent of mitochondrial involvement in the control of the myocardial cellular Ca2+during the contraction-relaxation cycle. Enough evidence, however, has accumulated to allow us to postulate that, in various other organs and tissues, mitochondria constitute an important factor in controlling the cellular free Ca2+concentration. In the following section, the physiological significance of mitochondrial Ca2+accumulation in these various organs will be discussed in more detail.
V.
PHYSIOLOGICAL SIGNIFICANCE OF MlTOCHONDRlAL CA ACCUMULATION IN DIFFERENT TISSUES
'+
Elegant new techniques localizing or visualizing accumulated Ca2+ in intact cells or organs have enhanced our knowledge of the in uiuo
MITOCHONDRIAL CALCIUM TRANSPORT
345
compartmentalization of cellular Ca2+. Using cells of isolated Chironomus salivary glands microinjected with the Ca2+-sensitive luminescent protein aequorin (Shimomura e t a,?. 1962), Rose and Loewenstein (1975) recently showed that injected Ca2+was confined to a limited area around the injection point. This phenomenon was abolished by cyanide or ruthenium red. Rose and Loewenstein (1975) interpreted their data to indicate that cellular Ca2+immediately after its injection was sequestered, presumably in mitochondria, and, thus, was not able to diffuse throughout the cell. A. Heart
The significance of mitochondria1 CaZ+accumulation in the control of the contraction-relaxation cycle of the myocardium has been dealt with from several points of view in Section IV. Thus, only a few summarizing points will be discussed here. Mitochondria isolated from mammalian myocardium exhibit an energy-dependent carrier-mediated Ca2+transport system with high affinity and capacity (Chance, 1965a; Mela and Chance, 1969; Scarpa and Graziotti, 1973; Jacobus e t aZ., 1975). I n all important aspects, the heart mitochondrial Ca2+accumulation system is similar to mitochondrial systems from other mammalian organs, such as the liver and kidney, although some reports suggesting differences in important features, such as the lack of maximal respiratory stimulation by Ca2+, have appeared (Jacobus et al., 1975). Some of these problems can be attributed to inadequately controlled heart mitochondrial preparations that cannot be compared to better coupled liver mitochondria. The question of the physiological role of heart mitochondrial Caz+ accumulation in controlling the contraction-relaxation cycle is still unsettled. Much of the evidence for the possible mitochondrial role has been discussed in Section IV. Studies on perfused heart have added a few new aspects toward solving this important question. It was shown by Horn et aZ. (1971) that perfused rat hearts exposed to 45Ca accumulated most of the calcium in their mitochondria. These data are in agreement with those of Carafoli (1967) and of Patriarca and Carafoli (1968). The mitochondrial accumulation was enhanced by epinephrine infusion. These authors also made the important observation that a rapid exchange of Ca2+ between the extracellular space and the mitochondria occurred. This was not the case between extracellular space and the microsomes. Several attempts similar to this one have been made to illustrate the role of mitochondrial Ca2+ accumulation in the heart by utilizing specific inhibitors of mitochon-
346
LEENA MELA
drial electron transfer, oxidative phosphorylation, or Ca2+accumulation in vivo or in perfused organs (Mezon and Bailey, 1975; Horn et al., 1969, 1971; Williamson et al., 1973). The data, however, are difficult to interpret because of unspecific binding and because of several parallel effects of the inhibitors. Thus, it is not clear which of the three celtlllar Ca2+controlling factors is affected: plasma membrane transport, mitochondrial, or sarcoplasmic Ca2+sequestration. More experimental work.is needed to solve the relationships of these three control systems in regulating the myocardial free Caz+concentration. B. liver and Kidney
The mitochondria from these two organs exhibit very similar Ca2+ transport systems. Both types of mitochondria are capable of accumulating large quantities of Ca2+(Lehninger et al., 1963; DeLuca and Engstrom, 1961; Vasington and Murphy, 1961, 1962; Chappell et al., 1963) with high affinity (Carafoli and Azzi, 1972; Reed and Bygrave, 1975a; Jacobus et al., 1975). The control kinetics of CaZ+accumulation, as measured by Scarpa et ul. (Vinogradov and Scarpa, 1973; Scarpa, 1974, 1975) by the murexide indicator method, are sigmoidal and exhibit somewhat faster initial rates at low (0.3 nmoles/sec/ mg) as well as at high concentrations (8.2 nmoles/sec/mg) of Ca2+ than were found in heart mitochondria (Scarpa and Graziotti, 1973). Borle (1971, 1972a) has suggested that the main portion of exchangeable cellular Ca2'in the kidney is mitochondrial. Recently, it has been shown that isolated hepatocytes accumulate Ca2+in a process that is dependent on respiratory substrate and 0 utilization (Kleineke and Stratman, 1974; Ontko et al., 1975). This uptake is sensitive to lanthanides and ruthenium red (Kleineke and Stratman, 1974). Ontko et al. (1975) measured the redox changes of pyridine nucleotides during liver cell Ca2+accumulation and found that it exhibited characteristics similar to those in isolated mitochondria. These data on isolated hepatocytes suggest that the Ca2+accumulatedb y the cells is taken up by the mitochondria and emphasize the role of mitochondria as the Ca2+-sequesteringcompartment. The cytoplasmic Ca2+in both liver and kidney is quite low, in the range of 10-6M. The mitochondrial capacity and affinity for Ca2+is quite adequate in controlling the cytoplasmic Ca2+atthis level. There are indications, however, that, under some pathological conditions, this Ca2+-sequestering mechanism fails. Tissue injury induced by complete or partial ischemia damages the mitochondrial Ca2+accumulation mechanism (Nicholas et al., 1974) at an early stage. As a result,
MITOCHONDRIA1 CALCIUM TRANSPORT
347
evidently, cellular free Ca2+concentration rises, inducing more damage to the cell (Trump and Arstila, 1971). These changes will b e discussed in more detail in Section VI.
C. Brain and Nervous Tissue
Mitochondria isolated from mammalian brain are capable of accumulating Ca2+in an energy-dependent process (Moore and Strasberg, 1970; Tjioe et al., 1970; Clark and Nicklas, 1970). Reports on the specific characteristics of brain mitochondrial Ca 2+ accumulation are contradictory. Tjioe et al. (1970) reported that brain mitochondrial Ca2+ accumulation was strictly dependent on ATP, even in the presence of oxidizable substrates. The ATP effect was not oligomycin sensitive. However, Lazarewicz and Hamberger (1975) were able to show that limited amounts )of Ca2+,up to about 30 nmoles/mg protein, were taken up in the absence of added ATP. If rotenone was added in the presence of succinate, nearly all added Ca2+,80 nmoles/mg, was accumulated when succinate was used as substrate. Mg2+and ATP in the presence of succinate supported a similar uptake of Ca2+.The authors proposed that the supportive role of rotenone or ATP was due to their ability to stabilize the mitochondrial membranes against damage during Ca2+accumulation.Thus, it appears that well-preserved preparations of intact brain mitochondria with high respiratory control ratios possess a Ca2+accumulation system characteristically similar to that of liver, kidney, and heart mitochondria. This has been also shown in unpublished experiments in our laboratory on rat and cat brain mitochondria (Mela and Wrobel-Kuhl, unpublished). Utilizing separate cell fractions rich in neuronal and glial cells for mitochondrial isolation, Lazarewicz et al. (1974a) were able to show that both glial and neuronal mitochondria exhibited capabilities of accumulating Ca2+.Similar amounts of Ca2+were accumulated by both fractions amounting to almost 400 nmoles/mg protein in the presence of ATP and phosphate. The initial rates of Ca2+accumulation by glial mitochondria, however, were faster. Half saturation in glial mitochondria was reached in less than 2 minutes; in neuronal mitochondria the time required was about 4 minutes. However, Seung U. Kim and E. Bonilla (personal communication) showed by electron microscopy of mixed neuronal and glial tissue cultures incubated with Sr2+in the absence of Ca2+that cation deposits were found primarily in neuronal mitochondria. Lazarewicz et al. (1974a) also showed that brain microsomes accumulated Ca2+ 15 times less than mitochondria. Thus,
348
LEENA MELA
although an active ATP-dependent uptake of Ca2+b y brain microsomes does occur (Yoshida et al., 1966; Otsuka et al., 1965; Nakamura and Konishi, 1974), the mitochondrial Ca2+-sequestering system might be physiologically more important than the microsomal one in brain. It has also been suggested that the brain mitochondrial Ca2+accumulation process is involved in the maintenance of anesthesia (Sweetman and Esmail, 1975). This conclusion is based on the authors’ findings that anesthetics such as Althesin, hexobarbitone, and halothane induce redistribution of injected 45Ca2+ in vivo in brain subcellular fractions, and that inhibition of brain mitochondrial Ca2+accumulation is induced by anesthetics in vitro. It has also been proposed that mitochondria play an important role in regulating the free Ca2+at nerve endings and, thus, the neurotransmitter release in terminals (Alnaes and Rahamimoff, 1975).Alnaes and Rahamimoff found that inhibitors of mitochondrial energy-linked functions, such as dicoumarol, rotenone, and antimycin A, as well as ruthenium red, increased the frequency of appearance of miniature end plate potentials and augmented the transmitter release. This augmentation also occurred in Ca2+-freeextracellular medium. Based on their calculations of mitochondrial content in motor nerve terminals, they found that the Ca2+uptake capacity of mitochondria in terminals exceeds the amount of free Ca2+enteringthe terminal due to an action potential by several orders of magnitude. Thus, Alnaes and Rahamimoff (1975) feel that mitochondria are strong candidates for the Ca2+ controller, necessary for the regulation of neurotransmission at motor nerve terminals. D. Smooth Muscle
X-ray microanalytical studies have recently identified uptake of divalent cations in association with phosphate into mitochondria in situ in various smooth muscles (Somlyo et al., 1974; Garfield and Somlyo, 1975). Rabbit portal anterior mesenteric vein and main pulmonary artery muscle mitochondria were shown to accumulate Ba2+,Sr2+,and Ca2+in an oligomycin- and anoxia-sensitive process (Somlyo et al., 1974). Cultured guinea pig aortas exhibited granules that were identified to contain Ca2+and phosphate (Garfield and Somlyo, 1975). Studies on isolated smooth muscle mitochondria have revealed an active Ca2+accumulation system. Batra (1972) and Batra and Timby (1971) showed that human myornetrial mitochondria were able to accumulate Ca2+.These studies were extended by Wikstrom et al. (1974, 1975), who found that bovine uterine mitochondria possessed a typical energy-dependent Ca2+ accumulation system characteristic of
MITOCHONDRIA1 CALCIUM TRANSPORT
349
other mammalian mitochondria. Kinetic studies of uterine mitochondrial Ca2+accumulation indicated that the uptake was faster than in liver or heart mitochondria, exhibiting initial rates of about 13 nmoles/sec/mg protein. The kinetics were not sigmoidal, but hyperbolic. This indicates that, at low Ca2+ concentrations (micromolar range), the initial rates would exceed those found in liver and heart mitochondria. The K , obtained with the murexide indicator method (Mela and Chance, 1968; Scarpa, 1972), in the absence of EGTA buffers, was 25 /.&I. Recently Carafoli has obtained values for K , in the range of 1-3 /.&I by utilizing Ca2+-EGTA buffers (Carafoli, personal communication). Zelck et al. (1975) demonstrated an ATPdependent Ca2+uptakeby isolated pig coronary artery and guinea pig ileum muscle mitochondria. Thus, it appears that all types of smooth muscle mitochondria studied so far are capable of accumulating Ca2+ rapidly and with high affinity. They may, therefore, be involved in regulating free Ca2+duringcontraction-relaxation cycles. It should be noted, however, that a Ca2+-sequesteringmicrosomal fraction with adequate capacity has also been identified in smooth muscle (Carsten, 1969; Fitzpatrick et al., 1972). E. Calcifying Tissue
Various cell types associated with bone calcification show ultrastructural dense granules that are located in mitochondria (Gonzales and Kamovsky, 1961; Martin and Matthews, 1969, 1970; Matthews et aZ., 1970; Holtrop, 1972; Brighton and Hunt, 1974; Gay and Schraer, 1975). These granules have been identified as containing Ca2+and phosphate (Sutfin et al., 1971). Similar granules have been found in the mitochondria of the avian shell gland (Schraer and Schraer, 1971; Schraer et aZ., 1973). Schraer and collaborators (Schraer and Schraer, 1971; Schraer et al., 1973) also showed that mitochondrial preparations isolated from the shell gland of the domestic hen exhibited rapid accumulation of large amounts of Ca2+.The shell gland mitochondria were able to accumulate 1.75 m o l e s Ca2+/mgprotein at pH 6.2 and 25°C in the presence of phosphate with a half-time of about 2 minutes. The speed of accumulation by the shell gland mitochondria was almost 10 times that of the hen liver mitochondria. Another interesting feature of the shell gland mitochondrial Ca2+accumulation is its independence of the medium pH in the range of 6.2-7.4. Hen liver mitochondria, on the other hand, exhibit maximal Ca2+transport activity at pH 7.0 and only half-maximal activity at pH 6.2 (Schraer and Schraer, 1971). This unusual finding might be physiologically important, since it has been
350
LEENA MELA
shown by direct pH measurements that the tissue pH of the shell gland of the hen changes considerably during the different phases of calcification (Simkiss and Taylor, 1971). Brighton and Hunt (1974) have proposed an interesting hypothesis for the mechanism of Ca2+transferto the extracellular matrix vesicles (Ali et al., 1970) via mitochondria during calcification. Brighton and Hunt showed histochemically that mitochondria and cell membranes accumulated large amounts of Ca2+in the upper zone of the epiphyseal growth plate. Brightdn and Heppenstall(l971) also showed with direct measurements of tissue PO that this zone had relatively high p02.At the bottom oi'the zone of hypertrophic cells, the mitochondria lose their Ca2+.This is at the level of the growth plate where the matrix vesicles begin to accumulate Ca2+.The zone of hypertrophic cells also exhibits low values of POz.Thus, Brighton and Hunt (1974) postulate that mitochondria accumulate Ca2+in the region where the p 0 is high enough adequately to support aerobic metabolism and energylinked ion transport, and release it to the extracellular matrix vesicles in the calcifying low pOz region, thus literally serving as the Ca2+ transferring compartment in the calcifying tissue. Recently, several attempts have been made by Brighton and collaborators (personal communication) to isolate growth plate mitochondria to be able to study their Ca '+accumulating capabilities in uitro. Preliminary experiments indicate that these mitochondria have an active Ca2+-transporting system (Brighton et al., unpublished). In the above brief summaries of the mitochondrial Caz+accumulation systems in various mammalian tissues, I have tried to show some examples only of the physiological organ functions where mitochondrial Ca2+accumulation is or might be involved. Our knowledge of these systems in proper in vivo situations is still very limited, and only a fragmentary image of the many important physiological functions of mitochondrial Ca2+accumulation can be reflected to the reader at the present time.
VI.
SOME ASPECTS OF THE PATHOPHYSIOLOGY OF MITOCHONDRIA1 CA ACCUMULATION
'+
In this section, I will concentrate on two kinds of pathological conditions affecting mitochondrial Ca2+transport. The first one will deal with tumor cell mitochondria, and the second one with the alterations induced by ischemic cell injury or low tissue oxygenation. There are two reasons for selecting these two areas of pathology: (i) a consider-
MITOCHONDRIAL CALCIUM TRANSPORT
35 1
able amount of experimental evidence has accumulated on mitochondrial Ca2+transport activities under these conditions; (ii) these two pathological conditions, with different etiological backgrounds, provide examples of different responses of the mitochondrial Ca2+accumulation system in disease states. A. Ca 2+ Accumulation by Tumor Cell Mitochondria
Cittadini et al. (1973), using direct measurements of Ca2+bythe metallochromic indicator murexide (Mela and Chance, 1968; Scarpa, 1972), showed that intact ascites tumor cells as well as isolated mitochondria can accumulate Ca2+up to 30 prnoleslg of dry weight. This accumulation was energy-dependent, and was supported by substrate oxidation but not by glycolysis. If succinate in the presence of rotenone was used as the respiratory substrate, the accumulated Ca2+ was retained in the mitochondria. With other respiratory substrates, however, release of the accumulated Ca2+ occurred after a few minutes. Cittadini et al. (1973) also found that tumor mitochondrial Ca2+accumulation was inhibited by Mg2+,La3+,and ruthenium red, but not b y oligomycin. They concluded that the Ca2+accumulated in ascites tumor cells was all taken up by mitochondria. Since they found tumor cell mitochondrial Ca2+transport similar to normal mammalian mitochondrial Ca2+uptake, they emphasized the importance of the studies in these cells as a tool for studying mitochondrial Ca2+accumulation in their natural environment, the cytoplasm, thus bringing the experimental conditions one step closer to the in d u o situation. Bygrave and his collaborators characterized the tumor cell mitochondrial Ca2+transport system more carefully. Thome and Bygrave (1973a,b, 1974a,b) and McIntyre and Bygrave (1974) found that Ca2+ accumulated by ascites tumor cell mitochondria did not induce uncoupling of respiration when ATP was present. N o appreciable ATPase activity or swelling of mitochondria was induced: this made it possible to stimulate respiration repeatedly by successive additions of Ca 2+. Thus, tumor cell mitochondrial Ca2+transport system differs characteristically from normal cell mitochondria. Further studies of Thome and Bygrave (1974a,c, 1975) led to another important discovery: both ADP and ATP translocation were inhibited by accumulated Ca2+. Thus, it appears that not only the Ca2+transport but also the adenine nucleotide transport mechanism of tumor cell mitochondria differs from that in normal mammalian mitochondria. It has also been shown by Pedersen and Morris (1974) that tumor cell mitochondria exhibit
352
LEENA MELA
very low ATPase activities. This might be partially due to a low translocase activity. It is not immediately evident whether and how these subtle differences of the mitochondrial ion transport system affect the life and activity characteristics of the tumors, It is interesting to note, however, that tumor cell mitochondria possess a very active Ca2+transport, which seems to interfere with the abilities of these mitochondria to phosphorylate and, thus, actively participate in ATP formation.
5. Effect of Ischemic Cell Injury on Mitochondtial Ca2+Accumulation
The mechanism of cell death in ischemic injury is still unsettled. Changes in the cell membrane, leading to altered permeability to various cations, have been proposed as primary defects (Trump and Arstila, 1971; Trump et al., 1971). However, several investigators feel that intracellular alterations due to increased anaerobic metabolism, and lack of substrate, together with inefficient removal of metabolites leading to lowered celhlar pH, contribute to the primary insult (Trump et al., 1971, Trump and Arstila, 1971; Silver, 1976). Whatever the mechanism of altered intracellular cationic, pH, or oxygen environment, these changes are associated, primarily or secondarily, with alterations of subcellular organelles, such as lysosomes (Weissman and Thomas, 1962; Alho, 1970; Janoff and Kaley, 1964), mitochondria (Jennings et al., 1969; Trumpet al., 1971; Ozawa et al., 1967), and, in the heart, sarcoplasmic reticulum (for a review see Martonosi, 1972). It is obvious that the inability of mitochondria to phosphorylate and keep the cell supplied with energy would be fatal in a very short time. Thus, whenever ischemia has progressed to such a state that substrate and 0 supplies become critically rate limiting, mitochondrial energy metabolism decreases to a level where lack of energy induces deterioration of cellular function. Inhibition of mitochondrial oxidative phosphorylation has been shown to occur in an irreversible manner in many organs after certain periods of complete or partial ischemia (Jennings et al., 1969; Ozawa et al., 1967; Mela, 1975; Mela et al., 1971, 1972, 1973, 1974.) It is noteworthy, however, that the Ca2+accumulationmechanism of mitochondria is more sensitive to ischemic injury than is oxidative phosphorylation. Mela and collaborators have shown that, before any inhibition of state 3 respiration, adenine nucleotide transport, or uncoupling can be detected after periods of partial ischemia in the liver
MITOCHONDRIAL CALCIUM TRANSPORT
353
or kidney, mitochondrial Ca2+ uptake activity is significantly decreased (Nicholas et al., 1974; Mela, 1975). The sensitivity of mitochondrial Ca2+accumulation to ischemic injury has also been shown for brain (Lazarewicz et al., 197413) and heart (Schwartz et al., 1968; Harigaya and Schwartz, 1969; Lindenmayer et al., 1970; Schwartz et al., 1973; Sordahl and Schwartz, 1974). If we assume that the inability of mitochondria to accumulate Ca2+ constitutes the primary insult in cell injury, several “secondary” pathological alterations would follow. Intracellular free Ca2+concentration would increase; this, in turn, would interfere with important cellular functions dependent on a constant low level of free Ca2+,initiating a chain reaction that would involve not only intracellular organelles, but also cell membrane function and changes in the extracellular space. Lysosomal enzymes obviously play an important role in ischemic injury. (Weissman and Thomas, 1962; Janoff and Kaley, 1964; Alho, 1970). Do the lysosomal enzymes become activated in the cell because of increased free Ca2+concentration? This is not known, but is feasible. It is also possible that the activation and release of lysosoma1 enzymes intracellularly act as primary triggers of other cellular alterations, such as mitochondrial damage and inhibition (Mellors et al., 1967; Nicholas et al., 1972; Mela et al., 1972, 1973). As long as the cellular PO does not drop below a critical level for the reaction of cytochrome oxidase with oxygen (Chance, 1965b; Lubbers, 1968), lack of oxygen does not seem to be a primary problem. Mela et al. (1973) adjusted liver tissue p 0 2 b y lowering the Ozcontent in inspired air precisely to the same pOz level reached in partial ischemia due to hemorrhage. The authors were able to show experimentally that this level of lowered tissue p 0 2 did not induce decreased mitochondrial phosphorylation or Ca2+transport activities seen after ischemia. Instead, increased phosphorylative and Ca2+ transport activity resulted. These adaptive changes of mitochondrial activity to perform energy-linked reactions have been verified under various chronic and acute conditions of tissue hypoxia (Park e t al., 1973; Mela et al., 1975). Our knowledge about the sequence of the various cellular events associated with cell injury is too limited to permit definitive conclusion at present. Whatever the primary and secondary insults may be, it is quite certsin that mitochondrial functional changes play an important role in the injury process. In particular, the inhibition of mitochondrial Ca2+transport seems to provide one of the most sensitive indicators of cellular ischemic injury.
354
LEENA MELA
VII.
SUMMARY
1. The review begins b y summarizing early findings of investigators during the 1950s and 1960s, which led to the notion that intact isolated mitochondria can accumulate Ca2+, Sr2+, and Mn2+in an energy-dependent process, whether or not permeant anions are present. 2. Evidence is presented in support of a three-step mechanism of mitochondrial Ca 2+ accumulation, i.e., energy-independent binding, carrier-mediated accumulation into the membrane, and aniondependent release from the carrier into the mitochondrial matrix. 3. The physiological significance of mitochondrial Ca2+accumulation system in terms of the regulation of cellular free Ca2+, is discussed in detail. In addition, experimental evidence is presented, indicating physiological control of cellular free Ca2+by mitochondria in a variety of mammalian tissues. 4. Examples of pathological alterations of the mitochondrial Ca2+ transport system are presented for two disease states: tumor cells and ischemically injured cells. This review does not attempt to present the entire literature published in the field. Critical attention is paid to publications pertinent to those specific aspects of mitochondrial Ca2+accumulation emphasized in this article. ACKNOWLEDGMENTS Support of the experimental work from the author’s laboratory has come from the following USPHS grants: 5-T01-GM-00957, GM-12202, R01-GM-19867, Pol-NS10939, and K04-GM-50318. I am deeply indebted to many of my close collaborators, particularly Dr. Britton Chance, for helpful discussions and positive criticism. My sincere thanks are due to all my dedicated assistants and students, whose efforts have been instrumental in many of the accomplishments reviewed here. I particularly acknowledge the expert assistance of Ted Alston, Joseph Campbell, Robert Laskowski, Agnes Knee, Kathleen Kraus, Steven Seitz, and Krystyna Wrobel-Kuhl. REFERENCES Addanki, S., and Sotos, J. F. (1969). Observations on intramitochondrial pH and ion transport by the 5,5-dimethyl 2,4-oxazolidinedione (DMO) method. Ann. N. Y. Acad. Sci. 147,756. Addanki, S., Cahill, F. D., and Sotos, J. F. (1968). Determination of intramitochondrial pH and intramitochondrial-extramitochondrialpH gradient of isolated heart mitoI. Changes during reschondria by the use of 5,5-dirnethyl-2,4-oxazolidinedione. piration and adenosine triphosphate-dependent transport of Cat+, Mg++and Zn ++. J . Biol. Chem. 243,2337. Akerman, K. E., Saris, N.-E. L., and Jkvisalo, J. 0. (1974). Mitochondria1 “high-affinity” binding sites for Ca*+-fact or artifact? Biochem. Biophys%Res. Comm. 58,801.
MITOCHO N D RIAL CALClUM TRANSPORT
355
Alexandre, A,, Rossi, C. S., and Rossi, C. R. (1974). Mitochondria] calcium pump: Detec tion ofenzyme-substrate type ofcomplex. In “Calcium Binding Proteins” (W. Drabikowski, H. Strzelecka-Golaszewska, and E. Carafoli, eds.), p. 875. Elsevier, Amsterdam. Alho, A. (1970). Lysosomal enzymes in murine tomique shock. Acta Chir. Scand. 136, 555. Ali, S. Y., Sajdera, S. W., and Anderson, H. C. (1970). Isolation and characterization of calcifying matrix vesicles from epiphyseal cartilage. Proc. Natl. Acad. Sci. U S A . 67, 1513. Alnaes, E., and Rahamimoff, €3. (1975). On the role of mitochondria in transmitter release from motor nerve terminals. J . Physiol. (London) 248,285. Andersson, R., Nilsson, K., Wikberg, J . , Johansson, S., Momhe-Lundholm, E., and Lundholm, L. (1975). Cyclic nucleotides and the contraction of smooth muscle. Adv. Cyclic Nucleotide Res. 5,491. Azzi, A., and Chance, B. (1969). The “energized state” of mitochondria: Life-time and ATP equivalence. Biochim. Biophys. Acta 189, 141. Bartley, W., and Amoore, J. E. (1958). The effect of manganese on the solute content of rat liver mitochondria Biochem. J . 69, 348. Batra, S . A. (1972). The role of calcium binding by subcellular particles in the contrac tion and relaxation of the myometrium. J . Obstet. Gynaecol. Br. Commonw. 112, 851. Batra, S. A., and Timby, L. (1971). ATP requirement in the course of calcium uptake by human myometrial mitochondria. FEBS Lett. 18,238. Bielawski, J., and Lehninger, A. L. (1966). Stoichiometric relationships in mitochondrial accumulation of calcium and phosphate supported by hydrolysis of adenosine triphosphate. J . Biol. Chem. 241,4316. Binet, A., and Volfin, P. (1975). Effect of the A23187 ionophore on mitochondrial membrane Mgz+and Ca2+.FEBS Lett. 49,400. Borle, A. (1971). Calcium transport in kidney cells and its regulation. In “Cellular Mechanisms for Calcium Transfer and Homeostasis” (G. Nichols, and R. A. Wasserman, eds.), p. 151. Academic Press, New York. Borle, A. (1972a). Kinetic analysis ofcalcium movements in cell culture. V. Intracellular calcium distribution in kidney cells. J. Membr. Biol. 10,45. Borle, A. (1972b).Parathyroid hormone and cell calcium. In “Calcium, Parathyroid Hormone and the Calcitonins” (R. V. Talmadge, and P. L. Munson, eds.), p. 484. Excerpta Medica, Amsterdam. Borle, A. (1973a). Calcium metabolism at the cellular level. Fed. Proc., Fed. Am. SOC. E x p . Biol. 32, 1944. Borle, A. (1973b). Cyclic AMP stimulation of calcium efflux from kidney, liver and heart mitochondria. J . Membr. Biol. 16,221. Borle, A. (1975a). Modulation of mitochondrial control of cytoplasmic calcium activity. In “Calcium Transport in Contraction and Secretion” (E. Carafoli, et al., eds.), p. 77. North-Holland Publ., Amsterdam. Borle, A. (197513). Regulation of cellular calcium metabolism and calcium transport by calcitonin. J. Membr. Biol. 21, 125. Brdiczka, D., Dolken, G., Krebs, W., and Hofmann, D. (1974). The inner boundary membrane of mitochondria. Localization and biochemical characterization, possible functions in biogenesis and metabolism. Hoppe-Seyler’s Z. Physiol. Chem. 355, 731. Brierley, G. P. (1963). Ion accumulation in heart mitochondria. In “Energy-Linked Functions of Mitochondria” (B. Chance, ed.), p. 237. Academic Press, New York.
356
LEENA MELA
Brierley, G. P., Murer, E., and Green, D. E. (1963). Participation ofan intermediate of oxidative phosphorylation in ion accumulation by mitochondria. Science 140,60. Brierley, G. P., Murer, E., and Bachmann, E. (1964). Studies on ion transport. 111. The accumulation of calcium and inorganic phosphate by heart mitochondria. Arch. Biochem. Biophys. 105,89. Brighton, C. T., and Heppenstall, R. B. (1971). Oxygen tension in zones of the epiphyseal plate, the metaphysis and diaphysis. An in vitro and in vivo study of rats and rabbits. J . Bone Joint Surg. 53A, 719. Brighton, C. T., and Hunt, R. M. (1974). Mitochondria1 calcium and its role in calcification. Histochemical localization of calcium in electron micrographs of the epiphyseal growth plate with K-pyroantimonate. Clin. Orthop. Rel. Res. 100,406. Bygrave, F. L., Reed, K. C., and Spencer, T. (1971a). Cooperative interactions in energy-dependent accumulation of Ca2+by isolated rat liver mitochondria. Nature (London),New Biol. 230,89. Bygrave, F. L., Reed, K. C., and Spencer, T. (1971b). Sigmoidal kinetics associated with calcium uptake and related ATPase in rat liver mitochondria. In “Energy Transduction in Respiration and Photosynthesis” (E. Quagliariello, S. Papa, and C. S. Rossi, eds.), p. 981. Adriatica Editrice, Ban. Bygrave, F. L., Daday, A. A,, and Doy, F. A. (1975). Evidence for a calcium-iontransport system in mitochondria isolated from flight muscle of the developing sheep blowfly Lucilia cuprina. Biochem. J. 146,601. Carafoli, E. (1967). In vivo effect of uncoupling agents on the incorporation of calcium and strontium into mitochondria and other subcellular fractions of rat 1iver.J. Gen. Physiol. 50, 1849. Carafoli, E. (1975a). Mitochondria, Ca2+transport and the regulation of heart contraction and metabolism. Mol. Cell. Cardiol. 7, 83. Carafoli, E. (197513).The Interaction of Ca2+with mitochondria, with special reference to the structural role of CaZ+in mitochondrial and other membranes. Mol. Cell. Biochem. 3, 133. Carafoli, E., and Azzi, A. (1972). The affinity of mitochondria for Ca++.Experientia 27, 906. Carafoli, E., and Rossi, C. S. (1967). Ca++-dependent movements of H +and K+across the rat liver mitochondrial membrane. Eur. J . Biochem. 2,224. Carafoli, E., and Rossi, C. S . (1971). Calcium transport in mitochondria.Adu. Cytopharmacol. 1,209. Carafoli, E., and Sottocasa, G. (1974). The Ca2+transport system of the mitochondrial membrane and the problem of the Ca2+ carrier. In “Dynamics of EnergyTransducing Membranes” (L. Emster, R. W. Estabrook, and E. C. Slater, eds.), p. 455. Elsevier, Amsterdam. Carafoli, E., Gamble, R. L., and Lehninger, A. L. (1965a). Ktdependent rebounds and oscillations in respiration-linked movements of Ca and H +in rat liver mitochondria. Biochem. Biophys. Res. Comm. 21,488. Carafoli, E., Weiland, S., and Lehninger, A. L. (1965b). Active accumulation of Sr2+by rat liver mitochondria. I. General features. Biochirn. Biophys. Acta 97, 88. Carafoli, E., Gamble, R. L., and Lehninger, A. L. (1966). Rebounds and oscillations in respiration-linked movements of Ca++and H in rat liver mitochondria. J . Biol. Chem. 241,2644. Carafoli, E., Gamble, R. L., Rossi, C. S., and Lehninger,A. L. (1967). Superstoichiometric ratios between ion movements and electron transport in rat liver mitochondria. J . Biol. Chem. 242, 1199. Carafoli, E., Gazzotti, P., Vasington, F. D., Sottocasa, G. L., Sandri, G., Panfili, E., and + +
+
MITOCHONDRIA1 CALCIUM TRANSPORT
357
deBemard, B. (1972). Soluble Ca2+binding factors isolated from mitochondria. In “Biochemistry and Biophysics of Mitochondria1 Membranes” (G. F. Azzone, et al., eds.), p. 623. Academic Press, New York. Carafoli, E., Dabrowska, R., Crovetti, F., Tiozzo, R., and Drabikowski, W. (1975a). An in vitro study of the interaction of heart mitochondria with troponin-bound Ca2+.Biochem. Biophys. Res. Comm. 62,908. Carafoli, E., Malmstiom, K., Capano, M., Sigel, E., and Crompton, M. (197513). Mitochondria and the regulation of cell calcium. In “Calcium Transport in Contraction and Secretion” (E. Carafoli, et al., eds.), p. 53. North-Holland Publ., Amsterdam. Carsten, M. E. (1969). Role of calcium binding by sarcoplasmic reticulum in the contraction and relaxation of uterine smooth muscle. ]. Gen. Physiol. 53,414. Chance, B. (1956). On possible mechanisms for the control of electron transport in the respiratory chain. Proc. Int. Congr. Biochem. 3rd, 1956, p. 300. Chance, B. (1963). Calcium-stimulated respiration in mitochondria. In “Energy-Linked Function of Mitochondria” (B. Chance, ed.), p. 253. Academic Press, New York. Chance, B. (1965a). The energy-linked reaction of calcium with mitochondria. 1.Biol. Chem. 240,2729. Chance, B. (196513).Reaction of oxygen with the respiratory chain in cells and tissues.]. Gen. Physiol. 49, 163. Chance, B. (1976). Pyridine nucleotide as an indicator of the oxygen requirements for energy-linked functions of mitochondria. Proc. Hurry S. Moss Int. Symp. Reg. Cardiac Metabolism, 1976. Circ. Res. 38, supp. 1-31. Chance, B., and Azzi, A. (1969). The response of reduced pyridine nucleotides to calcium-induced alkalinity. Ann. N. Y. Acud. Sci. 147,805. Chance, B., and Mela, L. (1966). Hydrogen ion concentration changes i n mitochondrial membranes. 1.Biol. Chem. 241,4588. Chance, B., and Montal, M. (1971). Ion-translocation in energy-conserving membrane systems. Curr. Top. Membr. Transport 2, 99. Chance, B., and Schoener, B. (1966). High and Low energy states of cytochromes. 111. In reactions with cations. J . Biol. Chem. 241,4577. Chance, B., and Yoshioka, T. (1965). Reaction of C a + +with mitochondria. Fed. Proc., Fed. Am. Soc. Erp. Biol. 24, 1644. Chance, B., and Yoshioka, T. (1966). External Ca*+concentrationsassociated with membrane alkalinization in mitochondria. Biochemistry 5,3224. Chance, B., Mela, L., and Harris, E. J. (1968). Interaction of ion movements and local anesthetics in mitochondrial membranes. Fed. Proc., Fed. Am. SOC. Erp. Biol. 27, 902. Chance, B., Azzi, A., Lee, I. Y., Lee, C. P., and Mela, L. (1969a). The nature of the respiratory chain: Location of energy conservation sites, the high energy store, electron transfer-linked conformation changes, and the “closedness” of submitochondrial vesicles. In “Mitochondria, Structure and Function” (L. Ernster, and Z. Drahota, eds.), p. 233. Academic Press, New York. Chance, B., Azzi, A., and Mela, L. (1969b). Molecular interactions of calcium transport in mitochondrial membranes. In “Molecular Basis of Membrane Function” (D. L. Tostesson, ed.), p. 561. Prentice-Hall, Englewood Cliffs, New Jersey. Chance, B., Tamura, M., Oshino, N., and Salkowitz, I. (1974). Criterion ofcardiac anoxia and bioenergetic activity. Fed. Proc., Fed. Am. Soc. E x p . Biol. 33,425. Chappell, J. B., and Greville, G. D. (1963). Isolated mitochondria and accumulation of divalent metal ions. Fed. Proc., Fed. Am. SOC.Exp. Biol. 22, 526. Chappell, J. B., and Crofts, A. R. (1965). Gramicidin and ion transport in isolated liver mitochondria. Biochern. 1.95,393.
35 8
LEENA MELA
Chappell, J. B., Cohn, M., and Greville, G. D. (1963).The accumulation of divalent ions by isolated mitochondria. In “Energy-Linked Functions of Mitochondria” (B. Chance, ed.), p. 219.Academic Press, New York. Cheng, S. C., and Chen, S . S. (1975).Stimulation by cyclic nucleotides ofcalcium efflux in barnacle muscle fibers. Life Sci. 16, 1711. Chudapongse, P.,and Haugaard, N. (1973).The effect of phosphoenolpyruvate on calcium transport by mitochondria. Biochim. Biophys. Acta 307,599. Cittadini, A., Scarpa, A., and Chance, B. (1973). Calcium transport in intact Ehrlich Ascites tumor cells. Biochim. Biophys. Actu 291,246. Clark, J. B., and Nicklas, W. J. (1970).The metabolism of rat brain mitoch0ndria.J. Biol. Chem. 245,4724. Coraboeuf, E., and Vassort, G. (1968).Effects of some inhibitors of ionic permeabilities on ventricular action potential and contraction of rat and guinea pig hearts../. Ebctrocurd. 1, 19. DeLuca, H. F., and Engstrom, G. W. (1961).Calcium uptake by rat kidney mitochondria. Proc. Nutl. Acad. Sci. U.S.A. 47, 1744. DeLuca, H.F., Engstrom, G. W., and Rasmussen, H. (1962).The action of vitamin D and parathyroid hormone in vitro on calcium uptake and release by kidney mitochondria. Proc. Natl. Acad. Sd.U.S.A. 48, 1604. Drabikowski, W., Dabrowska, R., and Barylko, B. (1971).Separation and characterization of the constituents of troponin. FEBS Lett. 12, 148. Drabikowski, W.,Dabrowska, R., and Barylko, B. (1973).Properties of troponin and its constituents. Acta Biochim. Pol. 20, 181. Drabikowski, W., Barylko, B., Dabrowska, R., Nowak, E., and Szpacenko, A. (1974). Studies on the properties of TN-C component of troponin and on its effect on the interaction between the constituents of thin filament. In “Calcium Binding Proteins” (W. Drabikowski, H. Stnelecka-Golaszewska, and E. Carafoli, eds.), p. 69.Elsevier, Amsterdam. Drahota, Z., Carafoli, E., Rossi, C. S., Gamble, R. L., and Lehninger, A. L. (1965).The steady state maintenance of accumulated Ca++in rat liver mitochondria. J. Biol. Chem. 240,2712. Drahota, Z . , Gazzotti, P., Carafoli, E., and Rossi, C. S . (1969).A comparison of the effects of different divalent cations on a number of mitochondria1 reactions linked to ion translocation. Arch. Biochem. Biophys. 130,267. Ebashi, S., Kodama, A., and Ebashi, F. (1968).Troponin: I. Preparation and physiological function. J. Biochem. 64,465. Fitzpatrick, D. F., Landon, E. J., Debbas, G., and Hunvitz, L. (1972).A calcium pump in vascular smooth muscle. Science 176,305. Friedman, N., and Rasmussen, H. (1970).Calcium, manganese and hepatic gluconeogenesis. Biochim. Biophys. Acta 222,41. Garfield, R. E., and Somlyo, A. P. (1975).Electron probe analysis and ultrastructure of cultured, freeze dried vascular smooth muscle. Proc. 33rd Ann. EMSA Meet., p. 558. Gay, C., and Schraer, H. (1975).Frozen thin-sections of rapidly forming bone: Bone cell ultrasturcture. Calcif. Tiss. Res. 19,39. Gear, A. R. L., and Lehninger, A. L. (1968).Rapid, respiration-independent binding of alkali metal cations by rat liver mitochondria. J. B i d . Chem. 243,3953. Gear, A. R. L., Rossi, C. S., Reynafaqje, B., and Lehninger, A. L. (1967).Acid-base exchanges in mitochondria and suspending medium during respiration-linked accumulation of bivalent cations. J. Biol. Chem. 242,3403. Ghosh, A. K.,and Chance, B. (1970).Kinetic and equilibrium studies on the reversal of calcium-induced intramitochondrial alkalinity by permeant anions. Arch. Biochem. Biophys. 138,483.
MITOCHONDRIA1 CALCIUM TRANSPORT
359
Gomez-Puyou, A., Gomez-Puyou, M. T., Becker, G., and Lehninger, A. L. (1972).An insoluble Caz+-bindingfactor from rat liver mitochondria. Biochem. Biophys. Res. Commun. 47,814. Gonzales, F., and Kamovsky, M. J. (1961).Electron microscopy of osteoclasts in healing fractures of cat bone. J. Biophys. Biochem. Cytol. 9,299. Greenawalt, J. W., and Carafoli, E . (1966).Electron microscope studies on the active accumulation of Sr++by rat liver mitochondria. J . Cell Biol. 29,37. Greenawalt, J. W., Rossi, C. S., and Lehninger, A. L. (1964).Effect of active accumulation ofcalcium and phosphate ions on the structure of rat liver mitochondria.J . Cell Biol. 23,21. Hackenbrock, C. R., and Caplan, A. I. (1969).Ion-induced ultrastructural transformations in isolated mitochondria. The energized uptake of calcium. J. Cell Biol. 42,
221. Hackenbrock, C.R., and Miller, K. J. (1975).The distribution ofanionic sites on the surfaces of mitochondria1 membranes. Visual probing with polycationic femtin.J . Cell Biol. 65,615. Harigaya, S . , and Schwartz, A. (1969).Rate of calcium binding and uptake in normal animal and failing human cardiac muscle. Membrane vesicles (relaxing system) and mitochondria. Circ. Res. 25,781. Harris, E. J., Judah, J. D., and Ahmed, K. (1966).Ion transport in mitochondria. Curr. Top. Bioenerg, 1,255. Hollander, P., Nordenbrand, K., and Emster, L. (1968).Dissociation of energydependent C a + + uptake from H + efflux and respiratory stimulation in oligomycin-treated mitochondria. Meet. FEBS, Prague, Sth, 1968, p. 247. Holtrop, M. E. (1972).The ultrastructure of the epiphyseal plate. 11. The hypertrophic chondrocyte. Calcif, Tissue Res. 9, 140. Horn, R. S.,Levin, R.,and Haugaard, N. (1969).The influence of oligomycin on the a c tions of epinephrine and theophylline upon the perfused rat heart. Biochem. Pharmacol. 18, 503. Hom, R. S., Fyhn, A., and Haugaard, N. (1971).Mitochondria1 calcium uptake in the perfused contracting rat heart and the influence of epinephrine on calcium exchange. Biochim. Biophys. Acta 226,459. Jacobus, W. E.,Tiozzo, R., Lugli, G., Lehninger, A. L., and Carafoli, E. (1975).Aspects of energy-linked calcium accumulation by rat heart mitochondria. J. Biol. Chem.
250,7863. Janoff, A., and Kaley, G. (1964).Studies on lysosomes in tolerance, shock and local injury induced by endotoxin. In “Bacterial Endotoxins” (M. Landy, and W. Braun, eds.), p. 631.Rutgers Univ. Press, New Brunswick, New Jersey. Jennings, R. B., Sommers, H. M., Herdson, P. B., and Kaltenbach, J. P. (1969).Ischemic injury of myocardium. Ann. N. Y. Acad. Sci. 156,61. Kimberg, D. V.,and Goldstein, S. A. (1967).Binding of calcium by liver mitochondria: An effect of steroid hormones in vitamin D-depleted and parathyroidectomized rats. Endocrinology 80,89. Kimmich, G. A., and Rasmussen, H. (1969).Regulation of pyruvate carboxylase activity by calcium in intact rat liver mitochondria. J . Biol. Chem. 244, 190. Kimura, T., Chu, J.-W., Mukai, K., Ishizuka, I., and Yamakawa, T. (1972).Some properties of a glycoprotein isolated from adrenal cortex mitochondria. Biochem. Biophys. Res. Commun. 49, 1678. Kleineke, J., and Stratman, F. W. (1974).Calcium transport in isolated rat hepatocytes. FEBS Lett. 43,75. Kubler, W., and Shineboume, E. A. (1971).Calcium and the mitochondria. In “Calcium and the Heart” (P. Harris and L. H. Opie, eds.), p. 93.Academic Press, New York.
360
LEENA MELA
Kurokawa, K., and Rasmussen, H. (1973). Ionic control of renal gluconeogenesis. 111. Effect of extracellular phosphate concentration. Biochim. Biophys. Acta 313, 59. Lardy, H. A., Graven, S . N., and Estrada-0, S . (1967). Specific induction and inhibition of cation and anion transport in mitochondria. Fed. Proc., Fed. Am. SOC.E x p . Biol. 26, 1355. Lazarewicz, J. W., and Hamberger, A. (1975). Respiration-linked calcium accumulation in brain mitochondria. Bulleein Acad. Pol. Sci. 23, 143. Lazarewicz, J. W., Haljamae, H., and Hamberger, A. (1974a). Calcium metabolism in isolated brain cells and subcellular fractions. J. Neurochem. 22,33. Lazarewicz, J. W., Strosznajder, J., and Dabrowiecki, 2 . (1974h). Effect of cerebral ischemia on calcium transport in isolated brain mitochondria. Proc. rnt. Congr. Neuropathol., 7th. (In press.) Lehninger, A. L. (1949).Esterification of inorganic phosphate coupled to electron transport between diphosphopyridine nucleotide and oxygen (11). J . Biol. Chem. 178, 625. Lehninger, A. L. (1962). Water uptake and extrusion by mitochondria in relation to oxidative phosphorylation. Physiol. Rev. 42,467. Lehninger, A. L. (1969). Acid-base changes in mitochondria and medium during energy-dependent and energy-independent binding of Ca+’+.Ann. N . Y. Acad. S c i . 147,816. Lehninger, A. L. (1970). Mitochondria and calcium ion transport. Bi0chem.J. 119,129. Lehninger, A. L. (1971). A soluble, heat-labile, high-affinity Ca2+-bindingfactor extracted from rat liver mitochondria. Biochem. Biophys. Res. Commun. 42,312. Lehninger, A. L., and Carafoli, E. (1970). Evidence for a membrane-linked Ca ’++ carrier in rat liver and kidney mitochondria. In “Biochemistry of the Phagocytic Process” (J.Schultz, ed.), p. 9. North-Holland Publ., Amsterdam. Lehninger, A. L., Rossi, C. S . , and Greenawalt J. W. (1963). Respiration-dependent accumulation of inorganic phosphate and Ca++hy rat liver mitochondria. Biochem. Biophys. Res. Commun. 10,444. Lehninger, A. L., Carafoli, E., and Rossi, C. S . (1967). Energy-linked ion movements in mitochondria1 systems. Adv. Enzymol. Relat. Areas Mol. B i d . 29,259. Lehninger, A. L., Rossi, C. S., Carafoli, E., and Reynafaje, B. (1969). Energyindependent and energy-dependent interactions of cations with mitochondria. In “Mitochondria, Structure and Function” (L. Emster, and Z. Drahota, eds.), p. 369. Academic Press, New York. Lindberg, O . , and Emster, L. (1954). Manganese, a co-factor of oxidative phosphorylation. Nature (London) 173, 1038. Lindenmayer, G. E., Harigaya, S., Bajusz, E., and Schwartz, A. (1970). Oxidative phosphorylation and calcium transport of mitochondria isolated from cardiomyopathic hamster hearts. J. Mol. Cell. Cardiol. 1,249. Lubbers, D. W. (1968).The oxygen pressure field of the brain and its significance for the normal and critical oxygen supply of the brain. In “Oxygen Transport in Blood and Tissue” (D. W. Lubbers, U. C. Luft, G. Thews, and E. Witzleb, eds.), p. 124. Thieme, Stuttgart. Lynn, W. S . , and Brown, R. H. (1966a).Accumulation of divalent organic anions by mitochondria. Arch. Biochem. Biophys. 114,260. Lynn, W. S., and Brown, R. H. (1966h). Effects of uncoupling agents on respiration, cation exchange and phosphorylative efficiency in mitochondria. Arch. Biochem. Biophys. 114, 271. McIntyre, H. J., and Bygrave, F. L. (1974). Retention of calcium by mitochondria isolated from Ehrlich ascites tumor cells. Arch. Biochem. Biophys. 165, 744.
MITOCHONDRIAL CALCIUM TRANSPORT
361
Martin, J. H., and Matthews, J. L. (1969). Mitochondrial granules in chondrocytes. Cabi f. Tissue Res. 3, 184. Martin, J. H., and Matthews, J. L. (1970). Mitochondrial granules in chondrocytes, 0steoblasts and osteocytes: An ultrastructural microincineration study. Clin. Orthop. 68,273. Martonosi, A. (1972). Biochemical and clinical aspects of sarcoplasmic reticulum f u n c tion. Curr. Top. Membr. Transport 3, 83. Matthews, J. L., Martin, J. H., Sampson, H. W., Kunin, A. S . , and Roan, J. H. (1970). Mitochondrial granules in the normal and rachitic rat epiphysis. CalciJ Tissue Res. 5, 91. Matthews, J. L., Martin, J. H., Collins, E . J., Kennedy, J. W., and Powell, E . L. (1972). Immediate changes in the ultrastructure of bone cells following thyrocalcitonin administration. In “Calcium, Parathyroid Hormone and the Calcitonins” (R. V. Talmadge, and P. L. Munson, eds.), p. 375. Excerpta Medica, Amsterdam. Maynard, L. S., and Cotzias, G. C. (1955). The partition of manganese among organs and intracellular organelles of the rat. 1.Biol. Chem. 214,489. Mela, L. (1966). Intramitochondrial pH changes. Fed. Proc.,Fed. Am. Soc. E x p . Biol. 25, 414. Mela, L. (1968a).Interaction of La3+andlocal anesthetic drugs with mitochondrial C a + + and Mn uptake. Arch. Biochem. Biophys. 123,286. Mela, L. (1968b). La + + + as a specific inhibitor of divalent cation-induced mitochondrial electron transfer. Fed. Proc., Fed. Am. Soc. E x p . Biol. 27,828. Mela, L. (1969a). Reaction of lanthanides with mitochondrial membranes. Ann. N. Y. Acad. Sci. 147,824. Mela, L. (196913).Inhibition and activation of calcium transport in mitochondria. Effect of lanthanides and local anesthetic drugs. Biochemistry 8,2481. Mela, L. (1975). Mitochondrial metabolic alterations in experimental circulatory shock. In “Gram-Negative Bacterial Infections and Mode of Endotoxin Actions” (B., Urbaschek, R. Urbaschek, and E. Neter, eds.), p. 288. Springer-Verlag, Berlin. Mela, L., and Chance, B. (1968). Spectrophotometric measurements of the kinetics of C a + +and M n + +accumulation in mitochondria. Biochemistry 7,4059. Mela, L., and Chance, B. (1969).Calcium carrier and the “high affinity calcium binding site” in mitochondria. Biochem. Biophys. Res. Commun. 35,556. Mela, L., Bacalzo, L. V., and Miller, L. D. (1971).Defective oxidative metabolism of rat liver mitochondria in hemorrhagic and endotoxin shock. Am. J . Phydol. 220, 571. Mela, L., Miller, L. D., and Nicholas, G. G. (1972). Influence of cellular acidosis and altered cation concentrations on shock-induced mitochondrial damage. Surgery 72, 102. Mela, L., Miller, L. D., Bacalzo, L. V., Olofsson, K., and White, R. R. (1973).Role of intracellular variations of lysosomal enzyme activity and oxygen tension in mitochondrial impairment in endotoxemia and hemorrhage in the rat. Ann. Surg. 178,727. Mela, L., Nicholas, G. G., and Miller, L. D. (1974). Inhibition of mitochondrial energy metabolism in hypovolemic and endotoxic shock. In “Steroids and Shock” (T. M. Glenn, ed.), p. 301. Univ. Park Press, Baltimore, Maryland. Mela, L., Goodwin, C. W., and Miller, L. D. (1975). Correlation of mitochondria1 cytochrome concentration and activity to oxygen availability in the newborn. Biochem. Biophys. Res. Commun. 64,384. Mellors, A., Tappell, A. L., Sawant, P. L., and Desai, I. D. (1967). Mitochondrial swelling and uncoupling of oxidative phosphorylation by lysosomes. Biochim. Biophys. Acta 143, 299. Mezon, A., and Bailey, L. E. (1975). Prevention of relaxation by lanthamum in the kitten heart. J . Mol. Cell. Cardiol. 7, 417. +
+
362
LEENA MELA
Mitchell, P. (1968). “Chemiosmotic Coupling and Energy Transduction.” Glynn Research, Bodmin., Cornwall, England. Mitchell, P., and Moyle, J. (1969). Translocation of some anions, cations and acids in rat liver mitochondria. EurJ. Biochem. 9, 149. Moore, C. L. (1971). Specific inhibition of mitochondrial Ca++transport by ruthenium red. Biochem. Biophys. Res. Commun. 42,298. Moore, C. L., and Pressman, B. C. (1964). Mechanism of action of valinomycin on mitochondria. Biochem. Biophys. Res. Commun. 15,562. Moore, C. L., and Strasberg, P. M. (1970). Cytochromes and oxidative phosphorylation. Hand. Neurochem. 3,53. Nakamura, K., and Konishi, K. (1974).Mechanism of adenosine triphosphate-dependent Ca2+uptake of brain microsomes.J. Biochem. 75, 1129. Nicholas, G. G., Mela, L., and Miller, L. D. (1972). Shock-induced alterations of mitochondrial membrane transport: Effects of endotoxin and lysosomal enzymes on calcium transport. Ann. Surg. 176,579. Nicholas, G. G., Mela, L., and Miller, L. D. (1974). Early alterations in mitochondrial membrane transport during endot0xemia.J. S u e . Res. 16,375. Ontko, J. A., Otto, D. A., Oshino, N., and Chance, B. (1975). Calcium-mediated alterations in the oxidation-reduction state of pyridine-nucleotides in isolated liver cells. FEBS Lett. 53,297. Otsuka, A., Ohtsuki, I., and Ebashi, S. (1965). A T P dependent Ca binding of brain microsomes. J. Biochem. 58, 188. Ozawa, K., Seta, K., Araki, H., and Handa, K. (1967). Effect of ischemia on mitochondrial metabolism. J . Biochem. 61, 512. Park, C. D., Mela, L., Wharton, R., Reilly, J., Fishbein, P., and Aberdeen, E. (1973). Cardiac mitochondrial activity in acute and chronic cyanosis. J. Surg. Res. 14, 139. Patriarca, P., and Carafoli, E. (1968).A study in the intracellular transport of calcium in rat heart. J . Cell. Physiol. 72,29. Peachey, L. D. (1964). Electron microscopic observations on the accumulation of divalent cations in intramitochondrial granules. J . Cell Biol. 20,95. Pedersen, P. L., and Morris, H. P. (1974). Uncoupler stimulated adenosine triphosphatase activity. Deficiency in intact mitochondria from Moms hepatomas and Ascites tumor cells. J . Biol. Chem. 249,3327. Piccinini, I., Meloni, S.,Chiarra, A., and Villani, F. P. (1975). Uptake of lanthanum by mitochondria. Pharmacol. Res. Commun. 7,429. Raablauf, J. (1953). Die Schwellung isolierter Leberzellmitokondrien und ihre physikalisch-chemische Beeinflussbarkeit. Helo. Physiol. Phannacol. Acta 11,142. Rasmussen, H. (1970). Cell communication, calcium ion and cyclic adenosine monophosphate. Science 170,404. Rasmussen, H. (1971). Ionic and hormonal control of calcium homeostasis. Am. J . Med. 50, 567. Rasmussen, H., and Nagata, N. (1970). Renal gluconeogenesis: Effects of parathyroid hormone and dibutyryl3’,5’-AMP. Biochim. Biophys. Acta 215,17. Rasmussen, H., Chance, B., and Ogata, E. (1965).A mechanism for the reactions of calcium with mitochondria. Proc, Natl. Acad. Sci. U.S.A. 53, 1069. Reed, K. C., and Bygrave, F. L. (1974). A re-evaluation of energy-independent calcium-ion binding by rat liver mitochondria. Biochem. J. 142,555. Reed, K. C., and Bygrave, F. L. (1975a). A kinetic study of mitochondrial calcium transport. Eur. ]. Biochem. 55,497. Reed, K. C., and Bygrave, F. L. (1975b). Methodology for in oitro studies of Cal’transport. Anal. Biochem. 67,44.
MITOCHONDRIA1 CALCIUM TRANSPORT
363
Reed, P. W., and Lardy, H. (1972).A 23187:A divalent cation ionophore.]. Biol. Chem.
247,6970. Reuter, H., and Scholz, H. (1968).Der Einfluss von Membranpotentiaranderungenund Calcium auf die Aktivierung der Kontraktion isolierter Herzpfaparate. Pflugers Arch. 300,87. Reynafarje, B., and Lehninger, A. L. (1969).High affinity and low affinity binding of C a + +by rat liver mitochondria. J . Biol. Chem. 244,584. Rose B.,and Loewenstein, W. R. (1975). Calcium ion distribution in cytoplasm visualized by Aequorin: Diffusion in cytosol restricted by energized sequestering. Science 190,1204. Rossi, E., and Azzone, G . F. (1968).Ion transport in liver mitochondria. IV. T h e relationship between ion translocation and electron transp0rt.J. Biol. Chem. 243,1514. Rossi, C. S . , and Lehninger, A. L. (1964).Stoichiometry of respiratory stimulation, accumulation of C a + + a n d phosphate, and oxidative phosphorylation in rat liver mitechondria. ]. Biol. Chem. 239,3971. Rossi, C. S., Bielawski, J., and Lehninger, A. L. (1966).Separation of H + a n d OH-in the extramitochondrial and mitochondrial phases during Ca++-activatedelectron transport. ]. Biol. Chem. 241, 1919. Rossi, C.S., Azzi, A., and Azzone, G. F. (1967a).Ion transport in liver mitochondria. I. Metabolism independent C a + +binding and H+release.J. Biol. Chem. 242,951. Rossi, C. S.,Azzone, G. F., and Azzi, A. (1967b). Ion transport in livermitochondria. 111. Stoichiometry of proton release during aerobic calcium ion translocation. Eur. J . Biochem. 1, 141. Rossi, C. S., Vasington, F. D., and Carafoli, E. (1973).The effect of ruthenium red on the uptake and release of Caz+by mitochondria. Biochem. Biophys. Res. Commun. 50,
846. Rossi, C. S., Alexandre, A., and Rossi, C. R. (1974). Interaction of calcium with mitochondria during calcium flux. FEBS Lett. 43,349. Rottenberg, H., and Scarpa, A. (1974).Calcium uptake and membrane potential in mitochondria. Biochemistry 13,4811. Sampson, H. W., Matthews, J. L., Martin, J. H., and Kunin, A. S . (1970).An electron microscopic localization of calcium in the small intestine of normal, rachitic and vitamin-D-treated rats. Calci-f Tissue Res. 5,305. Saris, N. E. (1963a).T h e calcium pump in mitochondria. Dissertation. Univ. of Helsinki. SOC. Scient. Fennica Comm. Physico-Mathem. 28, 11. Saris, N.E. (1963b). Studies on active calcium transport in mitochondria. Acta Chem. Scand. 17,882. Scarpa, A. (1972). Spectrophotometric measurement of C a + + b ymurexide. Methods Enzymol. 24,343. Scarpa, A. (1974).Kinetic studies of Ca2+transport in mitochondria. I n “Dynamics of Energy-Transducing Membranes” (L. Emster, R. E. W. Estabrook, and E. C. Slater, eds.), p. 471.Elsevier, Amsterdam. Scarpa, A. (1975).Kinetics and energy-coupling of C a + +transport in mitochondria. In “Calcium Transport in Contraction and Secretion” (E. Carafoli, ed.), p. 65. North-Holland Publ., Amsterdam. Scarpa, A. (1977).Transport across mitochondrial membranes. In “Biological Transport” (H. Ussing, and D. C. Tosteson, eds.). Springer-Verlag, Berlin. (In press.) Scarpa, A., and Azzi, A. (1968).Cation binding to submitochondrial particles. Biochim. Biophys. Acta 150,473. Scalpa, A., and Azzone, G. F. (1968).Ion transport in liver mitochondria. VI. The role of surface binding on aerobic C a + +trans1ocation.j. Biol. Chem. 243,5132.
364
LEENA MELA
Scarpa, A., and Graziotti, P. (1973). Mechanisms for intracellular calcium regulation in heart. I. Stopped-flow measurements of Ca++uptake by cardiac mitochondria. J. Gen. Physiol. 62, 756. Scatchard, G. (1949). The attractions ofproteins for small molecules and ions. Ann. N.Y. Acad. Sci. 51, 660. Schraer, H., and Schraer, R. (1971). Calcium transfer across the avian shell gland. In “Cellular Mechanisms for Calcium Transfer and Homeostasis” (G. Nichols, and R. H. Wasserman, eds.), p. 351. Academic Press, New York. Schraer, R., Elder, J. A., and Schraer, H. (1973). Aspects of mitochondrial function in calcium movement and calcification. Fed. Proc., Fed. Am. SOC.Exp. Biol. 32, 1938. Schwartz, A., Lindenmayer, G. E., and Harigaya, S. (1968).Respiratory control and calcium transport in heart mitochondria from the cardiomyopathic Syrian hamster. Trans. N . Y. Acad. Sci. 30,951. Schwartz, A., Wood, J. M., Allen, J. C., Bomet, E. P., Entman, M. L., Goldstein, M. A., Sordahl, L. A., and Suzuki, M. (1973).Biochemical and morphologic correlates of cardiac ischemia. Am. J. Cardiol. 32,46. Selwyn, M. J., Dawson, A. P., and Dunnett, S. J. (1970).Calcium transport in mitochondria. FEBS Lett. 10, 1. Shimomura, O., Johnson, F. H., and Saiger, Y. (1962). Extraction, purification and properties of aequorin, a bioluminescent protein from the luminous hydromedusan Aequorea. J. Cell. Comp. Physiol. 59,223. Siekevitz, P., and Potter, V. R. (1955).Biochemical structure of mitochondria. I. Intramitochondrial components and oxidative phosphory1ation.J. Biol. Chem. 215,221. Siekevitz, P., Potter, V. R., and Simonson, H. C. (1953). Latent adenosinetriphosphatase activity in resting rat liver mitochondria. ]. Biol. Chem. 205,893. Silver, I. A. (1977). Ion fluxes in hypoxic tissues. Microuasc. Res. (In press.) Simkiss, K., and Taylor, T. G. (1971).Shell formation. In “Physiology and Biochemistry of the Domestic Fowl” (D. J. Bell, and B. M. Freeman, eds.), p. 1331. Academic Press, New York. Slater, E. C., and Cleland, K. W. (1953). The effect of calcium on the respiratory and phosphorylative activities of heart-muscle sarcosomes. Biochem. J. 55,566. Somlyo, A. P., Somlyo, A. V., Devine, C. E., Peters, P. D., and Hall, T. A. (1974). E l e c tron microscopy and electron probe analysis of mitochondrial cation accumulation in smooth musc1e.J. Cell B i d . 61,723. Sordahl, L. A. (1974).Effects of magnesium, ruthenium red and the antibiotic ionophore A-23187 on initial rates of calcium uptake and release by heart mitochondria. Arch. Biochem. Biophys. 167, 104. Sordahl, L. A., and Schwartz, A. (1974).Mitochondria1 calcium metabolism in ischemic and failing myocardium. Recent Ado. Studies Card. Structure Metabol. 4, 535. Sottocasa, G. L., Sandri, G., Panfili, E., and DeBemard, B. (1971). A glycoprotein located in the intermembrane space of rat liver mitochondria. FEBS Lett. 17, 100. Sottocasa, G., Sandri, G., Panfili, E., DeBernard, B., Gazzotti, P., Vasington, F. D., and Carafoli, E. (1972). Isolation of a soluble Ca2+binding glycoprotein from ox liver mitochondria. Biochem. Biophys. Res. Commun. 47,808. Sottocasa, G. L., Sandri, G., Panfili, E., Gazzotti, P., and Carafoli, E. (1974).The calcium binding glycoprotein from animal mitochondria. In “Calcium Binding Proteins” (W. Drabikowski, H. Strzelecka-Golaszewska, and E. Carafoli, eds.), p. 855. Elsevier, Amsterdam. Spencer, T., and Bygrave, F. L. (1973). The role of mitochondria in modifying the cellular ionic environment: Studies of the kinetic accumulation of calcium by rat liver mitochondria. J. Bioenerg. 4,347.
MITOCHONDRIAL CALCIUM TRANSPORT
365
Sutfin, L. V., Holtrop, M. E., and Ogilvie, R. E. (1971). Microanalysis ofindividual mitochondrial granules with diameters less than 1000 angstroms. Science 174,947. Sweetman, A. J., and Esmail, A. F. (1975). Evidence for the role ofcalcium ions and mitochondria in the maintenance of anesthesia in the rat. Biochem. Biophys. Res. Commun. 64,885. Tapley, D. F. (1956). The effect of thyroxine and other substances on the swelling of isolated rat liver mitoch0ndria.J. Biol. Chem. 222,325. Tashmukhamedov, B. A., Gagelgans, A. I., Mamatkulov, K., and Makhmudova, E. M. (1972). Inhibition of Ca2+transport in mitochondria by selective blockade of membrane mucopolysaccharides by hexamine cobaltichloride. FEES Lett. 28,239. Thome, R. F. W., and Bygrave, F. L. (1973a). Interaction ofcalcium with mitochondria isolated from Ehrlich ascites tumour cells. Biochem. Biophys. Res. Commun. 50, 294. Thome, R. F. W., and Bygrave, F. L. (1973b). Energy-linked functions of tightly coupled mitochondria isolated from Ehrlich ascites tumor cells. Cancer Res. 33,2562. Thome, R. F. W., and Bygrave, F. L. (1974a). The role of mitochondria in modifying the cellular ionic environment. Calcium-induced respiratory activities in mitochondria isolated from various tumour cells. Biochem. J . 144,551. Thome, R. F. W., and Bygrave, F. L. (1974b). Calcium does not uncouple oxidative phosphorylation in tightly-coupled mitochondria from Ehrlich ascites tumour cells. Nature (London) 248,349. Thome, R.F. W., and Bygrave, F. L. (1974~).Inhibition by calcium of adenine nucleotide translocation in mitochondria isolated from Ehrlich ascites tumour cells. F E E S Lett. 41, 118. Thome, R. F. W., and Bygrave, F. L. (1975). Kinetic evidence for calcium-ion and phosphate-ion transport systems in mitochondria from Ehrlich ascites tumour cells. FEES Lett. 56, 185. Tjioe, S., Bianchi, C. P., and Haugaard, N. (1970). The fundion ofATP in Ca++uptake by rat brain mitochondria. Biochim. Biophys. Acta 216,270. Trump, B. F., and Arstila, A. V. (1971). Cell injury and cell death. In “Principles of Pathobiology” (M. F. LaVia, and R. B. Hill, eds.), p. 9. Oxford Univ. Press, London. Trump, B. F., Croker, B. P., and Mergner, W. J. (1971). The role of energy metabolism, ion and water shifts in the pathogenesis ofcell injury. In “Cell Membranes, Biological and Pathological Aspects” (G. W. Scarpelli, ed.), p. 84. Williams & Wilkins, Baltimore, Maryland. Vainio, H., Mela, L., and Chance, B. (1970). Energy dependent bivalent cation t r a n s b cation in rat liver mitochondria. Eur. J . Biochem. 12,387. Vasington, F. D., and Murphy, J. V. (1961). Active binding of calcium by mitochondria. Fed. Proc., Fed. Am. SOC. E x p . Biol. 20, 146. Vasington, F. D., and Murphy, J. V. (1962). Ca++uptakeby rat kidney mitochondria and its dependence on respiration and phosphorylation. J . B i d . Chem. 237,2670. Vasington, F. D., Gazzotti, P., Tiozzo, R., and Carafoli, E. (1972). The effect of ruthenium red on Ca2+transport and respiration in rat liver mitochondria. Biochitn. Biophys. Acta 256,43. Vinogradov, A,, and Scarpa, A. (1973). The initial velocities of calcium uptake by rat liver mitochondria. J . Biol. Chem. 248,5527. Weissman, G., and Thomas, L. (1962). Studieson lysosomes. I. The effects ofendotoxin, endotoxin tolerance and cortisone on the release of acid hydrolase from a granular fraction of rabbit 1iver.J. E x p . Med. 116,433. Wenner, C. E., and Hackney, J. H. (1967). H+changes associated with divalent cation uptake by mouse liver mitochondria. J . Biol. Chem. 242, 5053.
366
LEENA MELA
Wikstrom, M., Ahonen, P., and Luukkainen, T. (1974). A possible regulatory role of mitochondrial calcium uptake in uterine contractions. I n “Physiology and Genetics of Reproduction” (E. M. Countinho and F. Fuchs, eds.), Part B, p. 177. Plenum, New York. Wikstrom, M., Ahonen, P., and Luukkainen, T. (1975). The role of mitochondria in uterine contractions. FEBS Lett. 56, 120. Williamsoxi, J. R.,Schaffer, S. W., Scarpa, A., and Safer, B. (1973). Investigation of the calcium cycle in perfused rat and frog hearts. Recent Ado. Studies Curd. Structure Metabol. 4,375. Yoshida, H., Kadota, K., and Fujisawa, H. (1966). Adenosine triphosphate dependent calcium binding of microsomes and nerve endings. Nature (London) 212, 291. Zelck, F., Kamstedt, U., and Albrecht, E. (1975). Calcium uptake and calcium release by subcellular fractions of smooth muscle. 11. Kinetics of calcium uptake by microsomes and mitochondria from pig coronary artery and guinea pig ileum. Actu Biol. Med. Ger. 34,981.
Thyroidal Regulation of Active Sodium Transport F . ZSMAIL-BEZGZ Department of Internal Medicine and Pahlavi Medical Research Unit Pohlaui Uniuersity School of Medicine Shiraz. Iran
I . Introduction ......................................................... A. Earlier Theories ................................................. B . Sodium Transport Hypothesis .................................... I1. Thyroid Status and Sodium Transport-Dependent Respiration (QO.(t)) ... A . Qh(t) in Hypothyroid. Euthyroid. and Hyperthyroid States ......... B . Time Course of Effect of Thyroid Hormone on Q&) .............. I11. Possible Pathways of Thyroid Hormone-Induced Increase in Q&) . . . . . . IV. Thyroid Status and Transmembrane Electrochemical Potential Differences of Na+ and K C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . Intracellular Sodium and Potassium Concentration ................. B. Membrane Potential ............................................. C . Sodium Efflux Rate Constant ..................................... D . Comparison of Changes in Transmembrane Electrochemical Potential Differences and QQ(t) .................................. V . Thyroid Status and Membrane NaK-ATPase Activity ................... A . NaK-ATPase and Other Membrane-Bound Enzymes ............... B . Time Course of Effect of Thyroid Hormone on NaK-ATPase Activity ......................................................... C. Mechanism of Stimulation of NaK-ATPase Activity ................ VI . Thyroid Status and Tissue Adenine Nucleotide Content ................ VII. Summary and Conclusions ............................................ References ..........................................................
.
1
367 367 368 369 369 371 372 375 375 376 377 377 379 379 381 382 383 384 385
INTRODUCTION
.
A Earlier Theories
Over 80 years ago. Magnus-Levy (1895) made the fundamental observation that the rate of oxygen consumption was abnormally low in 367
368
F. ISMAlLBElGl
myxedematous and higher in hyperthyroid patients. Rohrer (1924) and Foster (1927) demonstrated that tissues isolated from hyperthyroid or hypothyroid animals and studied in vitro exhibited altered respiratory rates. The first concrete hypothesis on the mechanism of thyroid thermogenesis was put forward by Plummer and Boothby (1922) who demonstrated that hyperthyroid patients required larger than normal amounts of calories per unit muscular work. They reasoned that thyroid hormone produces metabolic inefficiency. In 1951, two laboratories proposed an explicit biochemical theory on the mechanism of thyroid thermogenesis, namely, that analogous to the effect of dinitrophenol (Loomis and Lipmann, 1948), thyroid hormones cause an uncoupling of mitochondrial oxidative phosphorylation. Lardy and Feldott (1951) showed that mitochondria isolated from T,-treated rats had lowered phosphorylation-oxidation (P/O) ratios. They also observed an uncoupling effect by addition of T 4to mitochondria in vitro, but the addition also markedly depressed respiration. Martius and Hess (1951) found that injection of high doses of T4(4-12 mg/100 gm body weight in 72 hours) caused a fall in rat mitochondrial P/O ratio. Furthermore, they demonstrated an in vitro uncoupling effect of T4 at concentrations of 5 x 10-5M or higher. These results were confirmed by Hoch and Lipmann (1954). Not withstanding the continued interest in this mechanism (Hoch, 1968), the “uncoupling” theory was not entirely satisfactory, even at the outset, since the respiratory rate of mitochondria isolated from Tptreated animals was not elevated despite the lowP/O ratios (Hoch and Lipmann, 1954). Moreover, mitochondria isolated from liver and skeletal muscle of rats treated with lower doses of T4(doses enough to cause hyperthyroidism) exhibited normal respiratory control and a normal P/O ratio (Fairhurst et aZ., 1959; Fletcher et al., 1962; Tata et a1 ., 1962). Coupled oxidative phosphorylation was also demonstrated in mitochondria isolated from hyperthyroid patients by Stocker et al. (1968). Information provided by Tata et aZ. (1963), and Gustafsson et al. (1965) indicated that T4increases the respiratory and phosphorylating capacity of mitochondria. Thus, the uncoupling hypothesis was not upheld and alternate mechanisms of thyroid thermogenesis were sought. 6. Sodium Transport Hypothesis
In intact mammalian cells, the process of ATP production (largely produced through mitochondrial respiration) is tightly linked to ATP Abbreviations: T, is 3,3’,5,5’-tetraiodo-~-thyronine; T3is 3,3’,5,-triiodo-~-thyronine.
THYROIDAL REGULATION OF ACTIVE SODIUM TRANSPORT
369
t ATPases
I
CO,
J
+ H20
FIG.1. Model of the cycle of ATP synthesis and hydrolysis. The synthesis of ATP from ADP + P, is coupled to oxidation of substrates in mitochondria.ATP hydrolysis is shown as coupled either to Na+ transport (on the right) or to all other ATP-utilizing pathways (on the left), such as protein synthesis, secretion, or contraction, and designated only as “ATPases.” From Edelman and Ismail-Beigi (1974).
utilization (Fig. 1). ATP is continuously dephosphorylated b y a variety of energy utilizing processes, and the generated ADP (and AMP) are rephosphorylated at the expense of substrate oxidation. In steady-state conditions, with an adequate supply of oxygen and substrate, the availability and activity of ADP (or “energy charge”) controls respiratory and metabolic activity (Chance and Williams, 1956; Hess and Chance, 19.59; Klingenberg, 1968; Atkinson, 1969). Under conditions of tight coupling between ATP production and ATP utilization and with a normal P/O ratio, a sustained increase in respiration and ATP production can only be brought about and maintained by a concomitant increase in ATP utilization. Of the ATP-utilizing processes shown in Fig. 1,active sodium transport is unique in that: (i) it is common to all target cells, and (ii) the process accounts for 20-45% of the total energy expenditure of various resting cells (Whittam, 1964). Therefore, active sodium transport deserved study as the possible “sink” in thyroid thermogenesis.
II. THYROID STATUS AND SODIUM TRANSPORTDEPENDENT RESPIRATION (Qo2(t)) A. QQ(t) in Hypothyroid, Euthyroid, and Hyperthyroid States
Sodium transport-dependent respiration (Qoz(t))can be estimated by measurement of tissue QQ in standard solutions and compared to Q02
370
F. ISMAIL-BEIGI
in similar solutions either devoid of Na+ or containing inhibitors of sodium pump activity such as cardiac glycosides (Whittam, 1964). The difference between the two respiratory rates is taken as Q&(t).In this article, unless otherwise specified, Qoz(t)denotes ouabain-inhibitable respiration. Utilizing this method, estimation of Qoz(t)of various tissues in different thyroid states have been made. 1 . Liver. With the use of ouabain, Ismail-Beigi and Edelman (1970, 1971) showed that Q&(t)accounts for 29%and 34%of hypothyroid and euthyroid Qonof rat liver slices, respectively. Injection of T3 (50 pg/lOO gm body weight on alternate days x 3) produced a large increase in Qonof liver slices in both groups of animals. In the transition from the euthyroid to the hyperthyroid state, 92%of the increase in Qon could be accounted for by the increase in QOn(t).Similarly, 100%of the increase in Qozwhen passing from the hypothyroid to the hypothyroid+T3 state was due to the increase in Qos(t).Estimation of Q02(t)of rat liver slices in sucrose-Ringer’s solution yielded similar results. Addition of ouabain to sucrose-Ringer’s buffer caused no further inhibition of tissue respiration, a finding that supports the specificity of cardiac glycosides action (Blond and Whittam, 1964). Israel et al. (1973), using ouabain, have confirmed that increases in Qo2(t) account for a major fraction of thyroid-induced augmentation of rat liver respiration. 2. Skeletal Muscle, Injection of T 3to hypothyroid and euthyroid rats stimulates Qoz(t)of the diaphragm (Ismail-Beigi and Edelman, 1970).Increased energy utilization by the sodium pump accounted for 43% of the augmented respiration in the transition from hypothyroid to euthyroid state and for 91% of the effect from euthyroid to hyperthyroid state. These results were confirmed by Asano et al. (1976) who also studied the effect of various doses of T3on QOnand the Q&(t) of rat diaphragm (Table I). Qonand Qoz(t)increased linearly as a h n c tion of logarithm of the T3 dose. Furthermore, they presented data in support of the assumption that changes in intracellular Na+ and K+ brought about by ouabain or sodium-free media do not, of themselves, alter oxidative metabolism. 3. Kidney. QOz(t),as measured by ouabain, accounts for 31% and 36% of the Qo, of kidney slices from hypothyroid and euthyroid rats (Ismail-Beigi and Edelman, 1971). Increase in Qo,(t) as a result of T3 injection accounted for 46% and 29%of the increase in Qo, in euthyroid and hypothyroid kidney slices, respectively. The magnitude of *Thisterm may require further definition inasmuch as it is now known (Kleinzeller,
1972) that in a variety of mammalian tissues a sizable proportion of the sodium extrusion system is ouabain-independent.
371
THYROIDAL REGULATION OF ACTIVE SODIUM TRANSPORT
TABLE I DOSE-RESPONSE RELATIONSHIPS OF RAT DIAPHRAGM AFTER A SINGLEINJECTION OF TQad
T3 (&lo0 gm body weight)
0
10
50
250
Qo,
5.3 f 0.2 2.4 f 0.2
6.6 t 0.2 3.0 f 0.2
7.9 f 0.4 3.7 f 0.3
8.3 f 0.3 4.1 f 0.2
Q"*(t)
From Asano et a!. (1976). Rats were injected with a single dose of T3, 1 week after surgical thyroidectomy and assayed 48 hours later. Qon and Qo,(t) are expressed in pl/mg dry weight/hour. Mean f SE. a
the effect of thyroid hormone on sodium transport-dependent respiration of the kidney deserves further study, since in various tissues, including the kidney, there exists a significant amount of ouabaininsensitive sodium transport (Kleinzeller and KnotkovL, 1964). 4. Bruin. Although adult mammalian brain does not respond thermogenically to thyroid hormone (Barker, 1951), the developing brain does so in both thermogenic and morphogenetic terms (Himwich and Fazekas, 1941; Tyler and van Harreveld, 1942; Valcana and Timiras, 1969). Injection of T3into adult rats produced no significant effect on Qo2or QQ(t)of cerebral slices, despite the animal's hyperthyroid state (Ismail-Beigi and Edelman, 1971). 5. Other Tissues. The Qoz response of various target tissues to thyroid hormone has been summarized by Barker (1964). Thyroidal stimulation of QQ(t)cited above does not always explain the increase in Q%, and in some target cells (e.g., adipose tissue), stimulation of sodium transport as measured with the use of ouabain plays little part in the thermogenic response (Fain and Rosenthal, 1971). I n brown fat, 60% of the thermogenic response to norepinephrine can be blocked by ouabain (Horwitz, 1973). It is possible that thyroid hormone and norepinephrine have synergistic effects in brown fat thermogenesis. Thus, further studies on thyroidal regulation of Qol(t)in various target tissues and on the interrelationship between various hormones would be of interest. 6. Time Course of Effect of Thyroid Hormone on QQW
In order to determine the temporal and quantitative relationships between changes in Qo2and Qol(t), the time course of effects of a
372
F. ISMAlLBElGl
single or of repeated doses of T30n QOnand QOp(t) of liver slices were examined in hypothyroid and euthyroid rats (Ismail-Beigi and Edelman, 1974). After injection of a single dose of 50 pg T3/100 gm body weight into hypothyroid and euthyroid rats, the Qo, and QOZ(t) rose simultaneously and reached peak values at 48 hours and then declined and almost reached baseline values by day 6. Following three repeated injections of T3(50pg/lOO gm body weight every other day) into hypothyroid and euthyroid rats, it was found that the liver Qo, reached the new steady state after 2 to 3 days. QOn(t)reached the highest values on days 4-6, at which time the increase in Q&(t)accounted for 90%and 80% of the rise in Qoain hypothyroid and euthyroid rat liver slices, respectively. Asano et a2. (1976) found that injection every other day of T3in rats for 2 weeks was necessary for the QO2 of diaphragm to reach a new steady state, indicating that tissues may take variable times to reach a new metabolic steady state. It is possible that increased energy expenditure for RNA and protein synthesis, or for other phenomena account for stimulation of Na+ transportindependent respiration. Suko (1971) has found stimulation of Ca2+ uptake of rabbit cardiac sarcoplasmic reticulum as a result of thyroid hormone. Further studies of acute and steady-state thyroidal stimulation of sodium-transport-independent respiration in transient and steady states would be of importance. 111. POSSIBLE PATHWAYS OF THYROID HORMONE-INDUCED INCREASE IN Qo,(t)
Figure 1 shows a schematic cell with ATP as the proximate energy donor for ion transport. The cell maintains a constant transmembrane electrochemical potential difference for sodium and potassium b y a steady input of energy. The “transducing machinery,” which translocates Na+ and K + ions at the expense of the free energy of ATP, is operationally defined as the “Na+pump.” There is now extensive information to indicate that the Na+ plus K+-Mg2+-activated ATPase (NaK-ATPase) enzyme system of the cell membrane is an expression of or constitutes an integral part of the ‘“a+ pump” (Skou, 1965; Glynn and Karlish, 1975). Based on the above model, thyroid hormone-induced activation of energy utilization by the Na+ pump in target tissues can be brought about and sustained by four theoretically distinct pathways:
TABLE I1
HYFWTHETICAL MECHANISMSOF THYROIDHORMONE-INDUCED INCREASE IN Qdt)” Predicted changes Hypothetical mechanism
a
(Na+)P
(K+)+Xb
Membrane potential‘
( 1 ) Denotes a decrease, ( T ) an increase, and (-) no change. (Na+),and (K+)I denote intracellular sodium and potassium concentrations. ( t ) in membrane potential denotes hyperpolarization. k denotes the eftlux rate constant of intracellular sodium.
kd
NaK-ATF’ase
ATPIADP
374
F. ISMAIL-BEIGI
1. A primary increase in mitochondria1 ATP synthesis resulting in increased cytoplasmic ATP and ATP/ADP levels (Table 11, mechanism 1).If the Na+pump is operating below saturation with respect to ATP, then the pump might b e stimulated to work at higher rates, thus generating more ADP. There should be a fall in intracellular Na+and a rise in intracellular K + concentrations, with hyperpolarization of the membrane. The increased transmembrane electrochemical potential difference for sodium would increase passive sodium influx and sustain the effect. The Na+ efflux rate constant would increase, and the V,, or the K , of NaK-ATPase would not change. 2. A primary activation of the Na+pump induced by an increase in the number of transport units (V,,, effect), or by a reduction of the K , for ATP or intracellular Na+ (Table 11, mechanism 2). The increased activity of the Na+ pump would lower intracellular Na+; the resultant increase in N a + gradient would sustain an increased inward leak. Other expected changes would be a fall in ATP/ADP ratio, changes in NaK-ATPase activity or kinetics, hyperpolarization of the membrane, and an increase in Na+ efflux rate constant. 3. A primary increase in plasma membrane permeability to sodium (PNa+) leading to increased intracellular sodium concentration and a stimulation of energy expenditure by the Na+pump (Table 11, mechanism 3). The predicted changes could be a fall in cell ATP/ADP ratio, depolarization of the membrane potential, and no change in the V,,, of NaK-ATPase. The Na+ efflux rate constant would either show no change or decrease. 4. A primary change in the stoichiometry of the Na+ pump ( “uncoupling”) leading to increased ATP utilization for a given rate of sodium transport (Table 11, mechanism 4).This mechanism should produce higher intracellular Na+with a lower K + concentrations, and a fall in ATP/ADP ratio. One would expect depolarization of the membrane with no change (or a fall) in Na+efflux rate constant. The NaK-ATPase activity might be increased, although uncoupling of the Na+pump might also be manifested as a rise in Mg2+-ATPase rather than NaK-ATPase. The thyroid-induced effect could be the result of one or of a combination of the above possibilities. Indeed, a single molecular event or alteration could be expressed by two or more of the above pathways. However, if one examines the direction of change, some of the postulates can be excluded as the sole or predominant mechanism of thyroid hormone action.
THYROIDAL REGULATION
3 75
OF ACTIVE SODIUM TRANSPORT
IV. THYROID STATUS AND TRANSMEMBRANE ELECTROCHEMICAL POTENTIAL DIFFERENCES OF Na+ AND K+
The predictions of possible changes in intracellular Na+ and K + concentrations made in Section I11 were based on the principle that the transmembrane electrochemical potential difference for Na+ (A/lNa) reflects the activity of the Na+ pump. Thus: AbNa = RT In (Na,+/Nal+) - F
*
AV
(1)
where R is the gas constant, T is the absolute temperature, Na,+and Nal +the extracellular and intracellular sodium activities, respectively, and F is the Faraday constant. Since in many tissues K + approaches electrochemical equilibrium across the plasma membrane, the following relationship is approximately correct (Goldman, 1943; Hodgkin and Horowicz, 1959):
F
*
AV
=
-RT In (Ki+/K,,+)
(2)
Combining Eqs. (1) and (2) yields:
A@Na = -RT In (Nai+/Kl+)+ RT In (Na,+/K,,+)
( 3)
Since extracellular Na+ and K + remain invariant with respect to thyroid status (U. Liberman, Y. Asano, and S. Wendelin, quoted in Ismail-Beigi and Edelman, 1973), the Nai+/Kl+ratiocan serve as an index of change in A/lNa. A. lntracellular Sodium and Potassium Concentration
The effect of thyroid hormone on intracellular potassium was studied by Elliot and Cheek (1968),who found that the hormone produced a significant increase in Ki of skeletal muscle in hypothyroid children. Valcana and Timiras (1969) reported that thyroid deprivation resulted in an increase in the Na+and a decrease in K+and Mg2+content of neonatal rat brain. El Shahawy et al. (1971) found that daily administration of T 4 t o normal dogs increased total body K+content by 29%, while serum K concentration remained unchanged. The effects of thyroid hormone on intracellular Na+/K+ ratio of various tissues were reported by Ismail-Beigi and Edelman (1973), and the results are summarized in Table 111. It can be seen that T, proin the various target tissues, leading to duced a decrease in Nal +/&+ an increase in AjiNa. It may be concluded that thyroidal “activation” of the Na+ pump (mechanisms 1 or 2, Table II), rather than changes in +
+
376
F. ISMAIL-BEIGI
TABLE I11 A SUMMARY OF THE EFFECTS OF THYROIDHORMONEON THE INTRACELLULAR Na+/K+ RATIO
Source Ismail-Beigi and Edelman (1970) Ismail-Beigi and Edelman (1973)
Thyroid status Euthyroid -+ T3
Tissue
% Change in Na+/K+ratioa
Liver slicesb
-43 (Sig)
Euthyroid
2
T3
Diaphragm'
- 19 (Sig)
Euthyroid
f
T3
Diaphragmd
-21 (NS)
Head Diaphragmd Head
-19 (NS) -20 (Sig) -34 (Sig)
*
Euthyroid T3 Hypothyroid -C Ts Hypothyroid -C T3
Sig denotes statistical significance ( p c O M ) , and NS denotes not significant. These data were obtained after in oitro incubation of liver slices at 37°C for 40 min with I4C-inulin as the e.c.f. marker. These data were obtained in unanesthetized rats following constant infusion of 14C-sucroseas the e.c.f. marker. These results were obtained after 5 hours of equilibration in oioo with 3H-inulin as the e.c.f. marker following bilateral nephrectomy. a
P,,t or uncoupling of the Na' pump, dominates the Na+ transportdependent increase in respiration. B. Membrane Potential
S . Sampson and F. Ismail-Beigi (quoted in Ismail-Beigi, 1971) studied the membrane potential of gastrocnemius muscle of hypothyroid and euthyroid rats treated with T3.Administration of T3to thyroidectomized rats produced a 1.1-mV hyperpolarization that was not statistically significant. In euthyroid rats, injection of T3 significantly changed the muscle membrane potential from -83.4 -C 0.6 mV to -87.4 2 1.0 mV. In a study on resting membrane potentials of patients, Cunningham et d . (1971) recorded high membrane potentials in patients with thyrotoxic myopathy as compared to other severe illnesses. Freedberg et al. ( 1970)measured atrial intracellular potentials of rabbits in various thyroid states and were unable to demonstrate a significant change either in the resting or the action potential. Edmonds et al. (1970)reported a significant hyperpolarization effect of T3 on the hypothyroid rat colon when the animals were sodium depleted. They showed that stimulation of sodium transport by aldosterone in the rat colon is dependent on the presence of thyroid hormones.
THYROIDAL REGULATION OF ACTIVE SODIUM TRANSPORT
377
The data summarized above demonstrate that membrane potentials are either unchanged or slightly increased as a result of thyroid hormone action. These findings suggest that mechanism 3 or 4 (Table 11) are unlikely pathways of thyroidal regulation of sodium transport. C. Sodium Efflux Rate Constant
As indicated above, the thyroid-induced fall in intracellular Na+/K+ ratio implies stimulation of Na+ transport in target cells. To further test this inference, Ismail-Beigi and Edelman (1973)studied the effect of thyroid hormone on the sodium efflux rate constant in liver slices (inv i t r o ) of euthyroid and hypothyroid rats injected with Ts.Rat liver slices that were preincubated in 22Nawere transferred to isotope-free media in successive steps. The efflux of sodium was resolved into two logarithmic components. The more rapid efflux rate constant, k,, can be considered to represent diffusion of z2Nafrom the extracellular space of the slice into the medium, and the slow rate constant, kz,as reflecting the efflux of intracellular sodium ions into the extracellular space and medium. Thyroid status had no effect on k,. In contrast, injection of T 3increased k significantly in hypothyroid and euthyroid rats. D. Comparison of Changes in Transmembrane Electrochemical Potential Differences and Qo,(t)
The results summarized above indicate that the magnitude of the ionic and electrical changes produced by T3are modest as compared to the increase in tissue respiration devoted to Na+transport. In intact cells, the increase in QQ(t) may be a result of combined effects of T3 on the intrinsic activity of the N a + pump and on Na+ permeability. Tosteson and Hoffman (1960), in their studies on low and high K+ sheep erythocytes, found that active Na+transport and passive penneability pathways are intimately related. If thyroid hormone-induced thermogenesis is mediated by simultaneous effects on active and passive Na+transport, then the increase in Qo,(t) would b e larger than the increment produced in the transmembrane electrochemical potential difference. Mathematical approaches to the problem of the magnitude of energy expenditure for the maintenance of transmembrane electrochemical potential differences have been developed. These are based on thermodynamic theories of irreversible processes (Katchalsky and Curran, 1965; Essig and Caplan, 1968).By assuming that the pump is
378
F. ISMAIL-BElGl
driven by ATP, and that the pump exchanges one Na+for one K + ion, the following dissipation function can b e written (Katchalsky and Sprangler, 1968): @
=J~(PATP + P H ~ O - PADP - P P , )
- /hao)
(4) - (pK' - kO)l where CP is the dissipation of free energy per unit time,], is the rate of the phosphorylation-dephosphorylation reaction, Jexch is the rate of exchange between Na+andK+, and pxrepresents the chemical potential of species x with the superscripts i and o referring to inside and outside the cell, respectively. The parenthetical expression (pATP + pHPo- p A D p - CLP,)can be identified as A,, i.e., the affinity of the reaction driving the pump, and +Jexeh[(PNai
- pia)- (A- &)I = ACNa - A& = RT In (Na°Ki/Na'Ko) can be identified as Xexch, i.e., the force of ion exchange. Insertion of the above definitions into Eq. (4)leads to @ =JrAr
+ Jexchxexch
(5)
The coupling between the forces and flows shows that Jexch =
L1lXexch
-k Lldb
JP= LZlXexch + LzzAr
(7)
whereL llrelates flow of ions per unit time across the membrane with transmembrane electrochemical potential difference, and L 22 relates the flow of chemical reaction devoted to the pump to the electrochemical potential difference produced. According to Onsager's Law (Katchalsky and Curran, 1965), the coupling coefficient L 12isequal to L2l.
In resting apolar cells, net Na+and K+ transport does not occur and henceJexchis zero. Thus, combining Eqs. (6) and (7) leads to: - LllL22)/~121Xexch
(8) whereJ, is the flow of the chemical reaction whose affinity appears in the above equations. The effect of thyroid hormone on Na+ transport-dependent respiration has been determined by oxygen consumption measurements i.e., Qol(t) (see Section 11).However, since as mentioned above (Tata, 1966), there is a constant stoichiometrical relationship between Qoz and ATP synthesis in the various thyroid states (i-e., mitochondria1 P/O ratio of 3 with], = 6JATP), 1/6 Qon(t) may be substituted in place of], in Eq. (8). The effect of thyroid hormone action on tissue electrolyte distribu.I]
=
[(L12'
THYROIDAL REGULATION OF ACTIVE SODIUM TRANSPORT
379
tion as summarized in Section IV and examination of Eq. (8)show that thyroid hormone causes a disproportionate rise inJ,, as compared to the increase in Xexchof target tissues. A fall in the composite proportionality constant would necessitate larger increases in Q@(t)for a given increment in transmembrane electrochemical potential difference. Indeed, coefficients LIZand Lzl would be drastically changed if thyroid hormone had an effect on the amount of the transport enzyme, or on the enzyme’s affinity for the various reactants. V.
THYROID STATUS AND MEMBRANE NaK-ATPase ACTIVITY
A. NaK-ATPare and Other Membrane-Bound Enzymes
A stimulatory effect of thyroid hormone on NaK-ATPase activity has been reported by several investigators. Kawada et a2. (1969) reported a threefold increase in tadpole epidermal NaK-ATPase activity during thyroid-induced metamorphosis, Valcana and Timiras (1969) demonstrated an increase in neonatal rat brain NaK-ATPase activity as a result of thyroid hormone replacement in thyroidectomized rats. The effect of thyroid hormone on skeletal muscle, liver, kidney, and cerebral cortex NaK-ATPase and Mg-ATPase activity of hypothyroid and euthyroid rats is summarized in Table IV. Injection of T S produced a significant rise in NaK-ATPase activity of the three target tissues tested, while it had no demonstrable effect on the enzyme activity of cerebral cortex. The increase in Mg-ATPase activity was either insignificant or slight. The stimulatory effect of thyroid on liver NaK-ATPase activity has also been shown by Israel et al. (1973). Thyroid hormone stimulation of renal NaK-ATPase has been confirmed (Katz and Lindheimer, 1973), although the effect was interpreted to b e due to increased Na+ delivery to the nephrons. This explanation is controversial (Michael et al., 1972). I n a dose-response study of renal cortical NaK-ATPase activity, it was found that single injection of 10, 50, and 250 p g T,/100 gm body weight increased NaK-ATPase activity by 17,43, and 66%, respectively, while the injections had no significant effect on Mg-ATPase (C. s. Lo, T. August, and I. S. Edelman, quoted in Edelman, 1975). Nevertheless, since renal NaK-ATPase is known to increase due to a variety of stimuli, including aldosterone and corticosteroids (Jqjrgensen, 1969; Katz and Lindheimer, 1973) and osmolality (Alexander and Lee, 1970), further studies on the mechanism of thyroidal regulation of renal NaK-ATPase would be of interest.
TABLE IV THE EFFECT OF THYROID STATUS ON ATPASESOF RAT MUSCLE, LIVER, KIDNEY, AND CEREBRAL CORTEP ~~
~
~
~~~
Tissue
Thyroid status
Mg-ATPasd
Skeletal'
Thyroidectomized Thyroidectomized + T3 Euthyroid Euthyroid + T3
9.7 f 0.6 10.7 f 0.8 7.9 f 0.8 9.0 f 0.8
Thyroidectomized Thyroidectomized + T3 Eutbyroid Euthyroid + T3
6.5 f 0.2 8.3 f 0.4 6.1 f 0.3 6.9 f 0.4
Thyroidectomized Thyroidectomized + T3 Euthyroid Euthyroid + T3
18.0 2 1.2 19.7 f 1.2 16.8 f 0.8 18.6 f 0.5
Euthyroid Euthyroid
14.1 f 0.5 15.2 f 0.5
~
Increase (%)
~
~~
NaK-ATPase*
~~~
Increase (%)
~
Liverd
Kidneyd
Cerebrald
+ T3
10.3 (NS) 13.9 (NS) 27.7 (Sig) 13.1 (NS)
8.9 f 0.6 15.6 f 1.0 13.1 f 0.8 16.4 2 1.1 0.82 f 0.11 1.26 f 0.11 0.62 2 0.05 1.12 f 0.05
75.3 (Sig) 25.2 (Sig) 53.7 (Sig)
80.6(Sig)
10.7 (NS)
7.3 2 0.6 12.3 f 0.6 10.0 2 0.5 12.1 f 0.7
21.0 (Sig)
7.8 (NS)
16.9 f 0.6 15.9 f 0.4
-6.0 (NS)
9.4 (NS)
68.5 (Sig)
a Thyroidectomized or euthyroid rats were injected with 50 pg T$lOO g m body weight on 3 alternate days. The enzymes were assayed in crude homogenates of liver, kidney, and cerebral cortex and in a partially purified membrane fraction of diaphragm. Mean f SE. Sig denotes p value <0.05, and NS denotes not significant. Mg-ATPase and NaK-ATPase are expressed as pMP,/mg proteinhour. From Asano et al. (1976). From Ismail-Beigi and Edelman (1971).
38 1
THYROIDAL REGULATION OF ACTIVE SODIUM TRANSPORT
TABLE V CHANCE IN ENZYMEACTIVITIES OF LIVERPLASMA MEMBRANEFRACTION FROM HYFOTHYROID AND EUTHYROID RATS (.CT$ Thyroid status
NaK-ATPase
Mg-ATPase
5’-Nucleotidase
Hypothyroid 5 T3* Euthyroid 5 Tad
+91%(<0.005)e +6W0 (<0.001)
-20% (NS) +0.7%(NS)
+26% (NS) -22% (NS)
Rats were injected with 50 pg TJ100 gm body weight or with vehicle on 3 alternate days and sacrificed 48 hours after the last injection. * From Edelman and Ismail-Beigi (1974). Level of statistical significance, NS = not significant. From Ismail-Beigi and Edelman (1971).
Asano et al. (1976) have demonstrated stimulation of skeletal muscle NaK-ATPase as a result of T I injection in hypothyroid and euthyroid rats. The increase in activity of NaK-ATPase correlated closely with stimulation of muscle Qoa(t).Furthermore, in thyroidectomized rats injected with single doses of 10, 50, and 250 ,ug T , Qoz(t) increased linearly with NaK-ATPase activity. The specificity of the effect of T 3 on NaK-ATPase was tested in liver plasma membrane fractions (Ismail-Beigi and Edelman, 1971; Edelman and Ismail-Beigi, 1974). The data summarized in Table V show the effect of T 3 on NaK-ATPase, Mg-ATPase, and 5’nucleotidase activity of partially purified plasma membranes. Injection of T ,had a significant stimulatory effect on NaK-ATPase activity, whereas no clear effect was demonstrable on the other two enzymes tested. The activity of adenyl cyclase in liver homogenates of rats in various thyroid states was reported by Jones et al. (1972) who found no significant change in the activity of this enzyme (as assayed in presence of NaF). This finding is in keeping with other reports that have failed to show a change in adenyl cyclase activity or cyclic AMP content of cardiac muscle from hyperthyroid animals (Sobel et al., 1969; Frazer et al., 1969). Thus it appears that thyroidal stimulation is specific for NaK-ATPase. B. Time Course of Effect of Thyroid Hormone on NaK-ATPare Activity
The time course of thyroidal stimulation of liver NaK-ATPase activity was measured after a single or after repeated injections of T3to hypothyroid and euthyroid rats (Ismail-Beigi and Edelman, 1974). With
382
F. ISMAIL-BEIGI
three doses of T3, the rise in NaK-ATPase activity was temporally in phase with thyroidal stimulation of sodium transport-dependent respiration (Section 11,B). Mg-ATPase activity also rose and, following its peak value at 2 to 3 days, declined close to baseline values by day 6 despite repeated T 3injections. Following a single injection of T3 to euthyroid rats, liver NaK-ATPase rose, peaked at 48 hours, and declined to baseline values by day 6. The changes in NaK-ATPase activity also paralleled the variations in QOl(t)summarized in Section I1,B. It may be concluded that thyroid-induced changes in NaK-ATPase and Qol(t)are temporally in phase. These findings are consistent with the inference that thyroidal stimulation of the Na+pumpmediates the increase in Qol(t)(i-e., mechanism 2, Table 11). C. Mechanism of Stimulation of NaK-ATPare Activity
Thyroidal stimulation of NaK-ATPase in target cells can be brought about either by (i) activation of the Na+pumpvia biosynthesis, stimulation, or removal of inhibition from Na+ pump units (VmaXeffect); or (ii) a change in the kinetics of the enzyme (K, effect); or (iii) a combination of the above. Asano et al. (1976)have shown that injection of T 3 produces a rise in skeletal muscle NaK-ATPase activity as assayed under Vmax conditions, while no change in the K, of the enzyme for ATP was observed (Table VI). They also quote unpublished observations that T increases the number of Na+ pump units in kidney and
TABLE VI THE EFFECTOF Ts ON KINETICS OF NaK-ATPAsE OF SKELETAL MUSCLE^.* ACTIVITY
~~
~
~~~~~~
Hypothyroid Hypothyroid + T3
A P
~
0.35 f 0.2 0.40 f 0.5
6.6 2 0.3 9.2 2 0.7
+0.05 NS
+2.6 <0.005
From Asano et al. (1976). Hypothyroid rats were given a single injection of T8 (250 pgl100 gm body weight) or diluent 1 week after surgical thyroidectomy. Gastrocnemius muscle NaK-ATPase was assayed 48 hours after injection in an ATP regenerating system
THYROIDAL REGULATION
OF ACTIVE SODIUM TRANSPORT
383
intestinal niucosa as estimated by specific binding of H-ouabain and by incorporation of 32Pinto Na+-dependent phosphorylated intermediate from AT3*P- y (C. S. Lo, U. A. Liberman, and I. S. Edelman, quoted in Edelman, 1975). The above data suggest that the mechanism of thyroidal stimulation of NaK-ATPase is an increase in the number of Na+pump units. Further studies are needed to determine whether the increase results from the synthesis of new Na+ pumps (or elements of the pumps) or from the unmasking of latent pump sites.
VI. THYROID STATUS AND TISSUE ADENINE NUCLEOTIDE CONTENT
Thyroid hormones have been reported to depress cell ATP and creatine phosphate content (Chilson and Sacks, 1959; Fletcher e t al., 1962). Ismail-Beigi et al. ( 1973),utilizing techniques of rapid freezing of tissues in situ, determined the adenine nucleotide content of rat liver in various thyroid states. In thyroidectomized rats, treatment with T 3 tended to lower ATP, ADP, and AMP content and the ATP/ADP ratio, but these changes were not statistically significant. In euthyroid rats, T3produced a 20% fall in ATP, a 34% rise in ADP, and a 38% fall in the ATP/ADP ratio; these changes were statistically significant. The overall tendency for depression of cell ATP/ADP ratio as a result of thyroid hormone treatment appears to be against a primary increase in ATP production, leading to stimulation of sodium transport-dependent respiration (i.e., mechanism 1, Table 11). This interpretation is based on the assumption that intracellular ATP and ADP pool sizes do not change as a result of T 3treatment. A fall in the cytoplasmic ATP/ADP ratio per se can stimulate glycolysis and mitochondria1 oxidative reactions (Hess and Chance, 1959; Klingenberg, 1968; Atkinson, 1969). However, a T3-induced decrease in the cell ATP/ADP ratio may not be enough t o sustain the large increase in liver QO2.Thus, T 3 might cause concomitant stimulation of ATP production and utilization. This conclusion is consistent with the available data, which indicate that thyroid hormones act to increase the activity of a number of regulatory glycolytic enzymes and mitochondrial oxidative capacity (Drabkin, 1950; Lee et al., 1959; Bronk, 1963; Lee and Lardy, 1965; Tata, 1966; Kubista et al., 1971; Werner and Berry, 1974), as well as stimulation of carrier-mediated ADP transport into mitochondria (Babior et al., 1973).
384
F. ISMAIL-BEIGI
VII.
SUMMARY AND CONCLUSIONS
In homeothermic vertebrates, thyroid hormones (TIand T3) regulate oxygen consumption and thus the rate of heat production by the organism. Evidence has been summarized in support of the hypothesis that thyroid hormones stimulate energy expended in active sodium transport in target tissues. This pathway is not the only mechanism involved in thyroid thermogenesis, since in some tissues a significant fraction of the increase in respiration is not dependent on sodium transport (e.g., adipose tissue). On the other hand, thyroidal stimulation of liver respiration can be explained almost wholly as having resulted from an increase in energy expenditure by the sodium pump. The implication that thyroid hormones stimulate respiration through increased energy utilization for sodium transport is supported by data utilizing a number of independent techniques for measuring the activity of the sodium pump. Thyroidal activation of the Na+pump is expressed by (i) an increase in Qon(t);(ii) an increase in the transmembrane electrochemical potential difference for Na+ and K+, resulting in a fall in the Na+i/K+iratio; (iii) an increase in NaK-ATPase activity of target tissues; and (iv) an increase in intracellular sodium efflux rate constant. The molecular events and mechanism that mediate thyroidal stimulation of active sodium transport are not yet known, but may involve RNA induction and protein synthesis (Tata, 1966). The data summarized in this article demonstrate the importance of the Na+ pump as an energy-utilizing system in the energy balance of cells and tissues. It is possible that regulation of energy utilization by the Na+pump may also contribute to changes in metabolic rate seen in a variety of physiological states such as adaptation to cold, fever, and catecholamine hypermetabolism. Thyroidal regulation of active sodium transport may be of importance in the evolutionary transition from the poikilothermic to homeothermic vertebrates (Stevens, 1973). Finally, although the physiological significance of increased transmembrane electrochemical potential difference for N a + and K + is unclear, thyroidal stimulation of NaK-ATPase activity and the resultant increase in intracellular K + concentration may produce the necessary milieu for proper cellular growth and differentiation. This would then link the thermogenic and morphogenetic actions of thyroid hormones in mammals.
THYROIDAL REGULATION OF ACTIVE SODIUM TRANSPORT
3 85
REFERENCES Alexander, J. C., and Lee, J. B. (1970).Effect of osmolality on Na+-K+-ATPasein outer renal medulla. Am. J. Physiol. 219, 1742-1745. Asano, Y.,Liberman, U. A., and Edelman, I. S. (1976). Thyroid thermogenesis: Relationship between Natdependent respiration and Na+ + Ktadenosine triphosphatase activity in rat skeletal muscle.]. Clin. Invest. 57,368-379. Atkinson, D. E. (1969). Regulation of enzyme function. Ann. Reu. Microbiol. 23,47-68. Babior, B. M., Creagan, S., Ingbar, J. H., and Kipnes, R. S. (1973). Stimulation of mitochondrial adenosine diphosphate uptake by thyroid hormones. Proc. Natl. Acad. S c i . U.S.A. 70, 98-102. Barker, S. B. (1951). Mechanism of action of thyroid hormone. Physiol. Rev. 31, 205-243. Barker, S . B. (1964). Physiological activity of thyroid hormone and analogues. In “The Thyroid Gland” (R. Pitt-Rivers, and W. R. Trotter, eds.), Vol. I, p. 199. Butterworth, London. Blond, D. M., and Whittam, R. (1964).The regulation of kidney respiration by sodium and potassium ions. Biochem. J. 92, 158-167. Bronk, J. R. (1963). Thyroid hormones: Control of terminal oxidation. Science 141, 816-818. Chance, B., and Williams, G. R. (1956). The respiratory chain and oxidative phosphorylation. Advan. Enzymol. Relat. Subj. Biochem. 17,65-134. Chilson, 0. P., and Sacks, J. (1959).Effect of hyperthyroidism on distribution of adenosine phosphates and glycogen in liver. Proc. SOC.E x p . B i d . Med. 101,331-335. Cunningham, J. N., Carter, N. W., Rector, F. C., and Seldin, D. W. (1971).Restingtransmembrane potential difference of skeletal muscle in normal subjects and severly ill patients. I. Clin. Invest. 50,49-59. Drabkin, D. L. (1950). Cytochrome c metabolism and liver regeneration. Influence of thyroid gland and thyroxine. J . Biol. Chem. 182,335-348. Edelman, I. S. (1975). Thyroidal regulation of renal energy metabolism and (Na+ + K+)-activated adenosine tnphosphatase activity. Med. Clin. N . Am. 59, 605-614. Edelman, I. S., and Ismail-Beigi, F. (1974). Thyroid thermogenesis and active sodium transport. Recent Progr. H o r n . Res. 30,235-257. Edmonds, C. J,, Thompson, B. D., and Marriott, J. (1970). Interrelationship of the effects of aldosterone and thyroid hormones on sodium transport and electrical properties of rat colon. ]. Endocrinol. 48, 189-197. Elliott, D. A., and Cheek, D. B. (1968). Muscle electrolyte patterns during growth. In “Human Growth” (D. B. Cheek, ed.), pp. 260-273. Lea & Febiger, Philadelphia, Pennsylvania. El Shahawy, M., Tucker, R., Wahner, H., and Smith, R. E. (1971). Hyperthyroidism and potassium. J . Am. Med. Ass. 217, 969. Essig, A., and Caplan, S . R. (1968). Energetics ofactive transport processes. Biophys. 1. 8,1434-1457. Fain, J. N., and Rosenthal, J. W. (1971). Calorigenic action of triiodothyronine.on white fat cells: Effects of ouabain, oligomycin and cathecolamines. Endocrinology 89, 1205-12 1 1.
386
F. ISMAIL-BElGl
Fairhurst, T. A. S.,Roberts, J. C., and Smith, R. E. (1959).Effects ofphysiological levels of thyroxin in vivo on respiration and phosphorylation in rat liver fractions. Am. 1. Physiol. 197,370-376. Fletcher, K., Myant, N. B., and Tyler, D. D. (1962). The influence of thyroid hormone upon the metabolism of adenosine triphosphate in rat liver. J . Physiol. (London). 162,345-357. Foster, C . L. (1927). A note on tissue respiration in relation to thyroidectomy. Proc. S O C . E x p . Biol. Med. 24,334-337. Frazer, A., Hess, M. E., and Shanfeed, J. (1969). The effects of thyroxine on rat heart adenosine, 3’, 5’-monophosphate, phosphorylase b kinase, and phosphorylase a activity. J . Phamacol. E x p . Ther. 170, 10-16. Freedberg, A. S., Papp, J. G., and Williams, E. M. V. (1970). The effect of altered thyroid status on atrial intracellular potentials. J . Physiol. (London) 207, 357-369. Glynn, I. M., and Karlish, S. J. D. (1975). The sodium pump. Ann. Reu. Physiol. 37, 13-55. Goldman, D. E. (1943). Potential, impedance, and rectification in membranes. J . Gen. Physiol. 27, 37-60. Gustafsson, R., Tata, J. R., Lindberg, O., and Emeter, L. (1965). The relationship between the structure and activity of rat skeletal muscle mitochondria after thyroidectomy and thyroid hormone treatment. ]. Cell Biol. 26, 555-578. Hess, B., and Chance, B. (1959). Phosphorylation efficiency of the intact cell. I. Glucose-oxygen titration in ascites tumor cells. J . Biol. Chem. 234,3031-3035. Himwich, H. E., and Fazekas, J. F. (1941). Comparative studies ofthe metabolism ofthe brain of the infant and adult dogs. Am. J . Physiol. 132,454-459. Hoch, F. L. (1968). Biochemistry of hyperthyroidism and hypothyroidism. Postgrad. Med. J . 44,347-362. Hoch, F. L., and Lipmann, F. (1954). The uncoupling of respiration and phosphorylation by thyroid hormones. Proc. Natl. Acad. Sci. U 3 . A . 40, 909-921. Hodgkin, A. L., and Horowicz, P. (1959). The influence of potassium and chloride ions on the membrane potential of single muscle fibers. J. Physiol. (London) 148, 127-152. Horwitz, B. A. (1973).Ouabain-sensitive component of brown fat thermogenesis. Am. J . Phgsiol. 224,352-355. Ismail-Beigi, F. (1971). Mechanism of the thermogenic action of thyroid hormones: Role of active sodium transport. Ph. D. Thesis. Univ. of California, San Francisco. Ismail-Beigi, F., and Edelman, I. S. (1970). The mechanism of thyroid calorigenesis: Role of active sodium transport. Proc. Natl. Acad. Sci. U.S.A. 67, 1071-1078. Ismail-Beigi, F., and Edelman, I. S.(1971). The mechanism of the calorigenic action of thyroid hormone: Stimulation of Na+ + K f activated adenosine-triphosphatase activity. J . Gen. Physiol. 57, 710-722. Ismail-Beigi, F., and Edelman, I. S. (1973).Effect of thyroid status on electrolyte distribution in rat tissues. Am. J. Physiol. 225, 1172-1177. Ismail-Beigi, F., a9$ Edelman, I. S. (1974). Time-course of the effects of thyroid hormone on respiration and Na+ + K+-ATPaseactivity in rat liver. Proc. SOC. E x p . Biol. Med. 146,983-988. Ismail-Beigi, F., Salibian, A., Kirsten, E., and Edelman, I. S. (1973).Effects of thyroid hormone on adenine nucleotide content of rat liver. Proc. SOC. E x p . Biol. Med. 144, 471-474. Israel, W., Videla, L., and MacDonald, A. (1973). Metabolic alterations produced in the liver by chronic ethanol administration. Comparison between the effects produced by ethanol and by thyroid hormones. Biochem. J . 134,523-529.
THYROIDAL REGULATION OF ACTIVE SODIUM TRANSPORT
387
Jones, J. K., Ismail-Beigi, F., and Edelman, I. S. (1972).Rat liver adenyl cyclase activity in various thyroid states.]. Clin. Invest. 51, 2498-2501. Jergensen, P. L. (1969). Regulation of the (Na+ + K+)-activated ATP hydrolyzing enzymes system in rat kidney. 11. The effect of aldosterone on the activity in kidneys of adrenalectomized rats. Biochim. Biophys. Acta. 192,326-334. Katchalsky, A., and Curran, P. F. (1965). “Nonequilibrium Thermodynamics in Biophysics.” Harvard Univ. Press, Cambridge, Massachusetts. Katchalsky, A., and Sprangler, R. (1968). Dynamics of membrane processes. Quart. Reu. Biophys. 1,127-162. Katz, A. I., and Lindheimer, M. D. (1973).Renal sodium and potassium-activated adenosine triphosphatase and sodium reabsorption in the hypothyroid rat. 1.Clin. Invest. 52,796-804. Kawada, J., Taylor, R. E., and Barker, S. B. (1969).Measurement of NaK-ATPase in separated epidermis of Rana catesbeiana frogs and tadpoles. Comp. Biochem. Physiol. 30,965-969. Kleinzeller, A. (1972). Cellular transport of water. In “Metabolic Pathways” (D. M. Greenberg, ed), Vol. 6, pp. 91-131. Academic Press, New York. Kleinzeller, A., and Knotkovi, A. (1964).T h e effect of ouabain on the electrolytes and water transport in kidney cortex and liver s1ices.J. Physiol. (London).175,172-192. Klingenberg, M. (1968).The respiratory chain, In “Biological Oxidations” (T. P. Singer, ed.), p. 3., Wiley (Interscience), New York. KubiGta, V., KubiStova, J., and Pette, D. (1971).Thyroid hormone induced changes in the enzyme activity pattern of energy-supplying metabolism of fast (white), a slow (red), and heart muscle of the rat. Eur. J. Biochem. 18,553-560, Lardy, H., and Feldott, G . (1951). Metabolic effects of thyroxine in vitro. Ann. N.Y. Acad. Sci. 54,636-644. Lee, G., and Lardy, H. (1965). Influence of thyroid hormone on L-a-glycerophosphate dehydrogenase and other dehydrogenases in various organs of the rat. J . Biol. Chem. 240, 1427-1436. Lee, Y. P., Takemori, A. E., and Lardy, H. (1959). Enhanced oxidation of aglycerophosphate by mitochondria of thyroid fed rats. ]. Biol. Chem. 234, 3051-3054. Loomis, W. F., and Lipmann, F. (1948). Reversible inhibition of the coupling between phosphorylation and oxidation. 1. Biol. Chem. 173,807-808. Magnus-Levy, A. (1895). Uber den respiratovischen gewechsel unter dem einfluss der thyroiden sowie unter versehiedenen pathologischen zustanden. Berlin. Klin. Wochenschr. 32,650-652. Martius, C., and Hess, B. (1951). The mode of action of thyroxine. Arch. Biochem. Biophys. 33,486-487. Michael, U. F., Barenberg, R. L., Chavez, R., Vaamonde, C. A,, and Papper, S. (1972). Renal handling of sodium water in hypothyroid rats. Clearance and micropuncture studies. J . Clin. Inuest. 51, 1405-1412. Plummer, H . S., and Boothby, W. M. (1922). The cost of work in exophthalmic goiter. Am. J. Physiol. 63,406-410. Rohrer, A. (1924).Vergleich des sauerstoff verbrauchs uber lebender sugetierorgane im normalen zustande und nach futterung mit schilddrusen-hormon. Biochern. Z. 145, 154- 159. Skou, J. C. (1965).Enzymatic basis for active sodium transport of Na+and K+across cell membranes. Physiol. Rev. 45, 596-617. Sobel, B. E., Dempsey, P. J., and Cooper, T. (1969). Normal myocardial adenyl cyclase activity in hyperthyroid cats. Proc. SOC.E x p . Biol. Med. 132,6-9.
388
F. ISMAIL-BEIGI
Stevens, E. D. (1973).The evolution of end0thermy.J. Theor. Biol. 38,597-611. Stocker, W. W.,Samaha, F. J., and De Groot, L. J. (1968).Coupled oxidative phosphorylation in muscle of thyrotoxic patients. Am. J . Med. 44, 900-909. Suko, J. (1971).Alterations of Ca ++-activatedATPase of cardiac sarcoplasmic reticulum in hyper- and hypothyroidism. Biochim. Biophys. Acta. 252,324-327. Tata, J. R. (1966). The regulation of mitochondrial structure and function by thyroid hormones under physiological conditions. In “Regulation of Metabolic Processes in Mitochondria” (J. M. Tager, S . Papa, E. Quagliariello, and E. C. Slater, eds.), pp. 489-507.Elsevier, Amsterdam. Tata, J. R., Emster, K., and Lindberg, 0. (1962).Control of basal metabolic rate by thyroid hormones and cellular function. Nature (London). 193,1058-1060. Tata, J. R., Emster, L., Lindberg, O., Arrhenius, E., Pederson, S., and Hedman, R. (1963).The action of thyroid hormones at the cell level. Biochem. J . 86,408-428. Tosteson, D. C., and Hoffman, J. F. (1960).Regulation of cell volume by active cation transport in high and low potassium sheep red cells. J. Gen. Physiol.44, 169-194. Tyler, D. B., and van Harreveld, A. (1942).The respiration of developing brain. Am. J . Physiol. 136,600-606. Valcana, T., and Timiras, P. S. (1969).Effect of hypothyroidism on ionic metabolism and NaK-activated ATP phosphohydrolase activity in the developing rat brain.]. Neurochem. 16,935-943. Werner, H. V., and Berry, M. N. (1974).Stimulatory effects of thyroxine administration on reducing-equivalent transfer from substrate to oxygen during hepatic metabolism of sorbitolktnd glycerol. Eur. J. Biochem. 42,315-324. Whittam, R. (1964).The interdependence of metabolism and active transport. In “The Cellular Functions of Membrane Transport” (J. F. Hoffman, ed), p. 139. Prentice-Hall, Englewood Cliffs, New Jersey.
A
Bacteriodes melaninogenicus, fumarate reduction in, 195 Bacteroides ruminicola, fumarate reduction in, 195 Bacteria anaerobic electron transfer and active transport in, 177-229 fumarate reduction in, 190-195 nitrate respiration in, 180-189 phosphorylation coupled to electron transfer in, 195-199 “Bilayer couple” mechanism, 127 “Binding” effects, on surface potential, 85 Biological membranes, charged lipid distribution in, 122-123 Birds, erythrocyte membrane phosphorylation in, 271 Black lipid membranes, 73,84 transitory changes in potential across,
Active transport in bacteria, 177-229 chemiosmotic hypothesis of, 203 facilitated diffusion in, 199 group translocation in, 199-200 nitrate respiration coupled to, 204-208 passive diffusion in, 199 Adenine nucleotide, in tissue, thyroid hormone effects on, 383 Adenocarcinoma cells, water in, NMR spectroscopy of, 41 Adsorption of charged molecules to membranes, 93-108 theory, 93-99 Aerobic bacteria, nitrate respiration in, 188-189 Alkali cations, in intracellular fluids, 88-89 NMR spectroscopy, 1-69 Amebae, surface potential effects on pino- Bladders of toads cytosis in, 129 phosphorylation in, 297-298 Anaerovibrio lipolytica, fumarate reducin sodium transport studies, 149 tion in, 198 l-Anilino-8-naphthalenesulfonate, see Bolkmann relation, 131 “Boundary” potentials ANS electrostatic, 113-118 ANS, adsorption of, to bilayers, 126-127, experimental tests of, 116-118, 133132-133 134 Antibiotics, pore-forming, conductanceBrain voltage curves of, 91 mitochondrial calcium transport in, Apotransferrin, in solution, NMR spec347-348 troscopy of, 29 sodium-transport respiration in, 371 ATP in anaerobic active transport, 211-213 synthesis, nitrate respiration coupled C to, 197-198 Calcifying tissue, mitochondrial calcium transport in, 349-350 B Calcium transport Bacillus megatherium, fumarate r e d u c across mitochondrial membranes, 321tion in, 193 366 389
390
SUBJECT INDEX
in brain and nervous tissue, 347-348 D in calcifying tissue, 349-350 carrier-mediated calcium accumula- Debye-Huckel theory of weak electrolytes, 80 tion of, 331-337 Debye length, 79,82, 132 early studies, 323-326 Desulfooibrio spp., fumarate reduction in, energy-independent binding, 327195 331 Difhse double layer, theoretical descripin heart, 345-346 tion of, 73-82 in ischemic cell injury, 352-353 Dipicrylamine, adsorption to bilayer kinetics of, 338-340 membrane, 113 in liver and kidney, 346-347 Dipole potential pathophysiology of, 350-353 experimental estimates of, 108-11 1 role in cellular calcium control, molecules affecting, 111-113 337-344 Dipoles, molecular, at membrane-soluin smooth muscle, 348-349 tion interfaces, 108-113 three-step mechanism, 326-337 in tumor cells, 351-352 Cancer cells. (See also Tumor cells) E water in, NMR spectroscopy of, 41 Carbonic anhydrase Electron-probe X-ray microanalysis, in phosphorylation of, 248-249 cell study, 4 in solution, NMR spectroscopy of, 29 Electron transport Cardiac muscle anaerobic, in bacteria, 177-229 membrane phosphorylation in, 277-281 phosphorylation coupled to, in bacteria, phosphoprotein phosphatases in, 301195-199 302 Electrostatic potentials Ceruloplasmin, in solution, NMR specat membrane-solution interfaces, 71troscopy of, 29 144 Chemiosmotic hypothesis, of active transbiological implications, 118-130 port, 203 Enzymes, in membranes, surface potenCholesterol, effects on dipole potential, tial effects on, 128-129 111 Equivalent circuit model Cholesterol esterase, phosphorylation of, theory of, in sodium transport, 162-164 247-248 experimental evaluation, 165 Concanavalin A, in solution, NMR spec- Erythrocytes troscopy of, 29 membrane phosphorylation in, 259Conductance-voltage curves, of excitable 27 1 membranes, 118-122 abnormalities, 269 Corpus luteum, phosphoprotein phosphain birds, 271 tases in, 301 physiological role, 265-271 Cyclic AMP-dependent protein kinases by soluble protein kinases, 265 distribution and multiplicity of, 237NMR spectroscopy of, 34-35 240 phosphoprotein phosphatases in, 300nonenzyme substrates of, 244-245 301 in phosphorylation, 245-247 Eschertchia coli subunit structure and mode of activafumarate reduction in, 190-193 tion of, 240-243 nitrate respiration in, 181, 182-187 Cystic fibrosis, erythrocyte membrane Excitable membranes, conductance-voltDhosDhowlation in. 269 - _ . age curves of, 118-122
391
SUBJECT INDEX
F Facilitated diffusion, in active transport,
199 Fat cells, phosphorylation in, 297-298 Fluorescent probes, adsorption of, to bilayers, 126-127 Formate-hydrogenlyase pathway, in bacteria, 183 Formate-nitrate reductase pathway, in bacteria, 183 Fumarate reduction in bacteria, 190-195 anaerobic active transport coupled to, 208-21 1 ATP synthesis coupled to, 198-199
0 Glycogen synthetase, phosphorylation by,
246-247 Glycogenolytic enzymes, phoshorylation and regulation of, 234 Group translocation, in active transport,
199-200 Gouy-Chapman theory of diffuse double layer, 72-83 Gouy equation, 77, 78,80 experimental tests of, 82-93
H Haemophilus sp., nitrate respiration in, 188 Haemophilus injluenzae, fumarate reduction in, 193 Heart membrane phosphorylation in muscle of, 277-281 mitochondrial calcium transport in,
345-346 Helmholtz-Smoluchowski equation, zeta potential calculation from, 87 Hemocyanin, in solution, NMR spectroscopy of, 29-31 Hemoglobin, in solution, NMR spectroscopy of, 29
I Inkacellular fluids, NMR spectroscopy of,
1-69
Iodide, as conductance mediator, 84-85 Ion-selective microelectrodes, intracelM a r recording with, 4
K Kidney mitochondrial
calcium transport in,
346-347 sodium-transport respiration in, 370-
37 1 Klebsiella aerogenes fumarate reduction in, 193 nitrate respiration in, 181
1 Langmuir adsorption isotherm, 95 Lipase, hormone-sensitive, phosphorylation by, 247-248 Lipids in bilayers and biomembranes, 71-72 translational motion of, 81 charged, in biological membranes,
122-123 Liver mitochondrial
calcium transport in,
346-347 sodium-transport respiration in, 370 Lysozyme, in solution, NMR spectroscopy of, 29, 31
M Mammary cells, phosphorylation of, 299 Maxwell equations, 77 Membrane phosphorylation in erythrocytes, 259-271 in fat cells, 297-298 in muscle, 276-286 in microtubules, 283-297 in myelin, 291-293 protein kinases and, 233-320 in rhodopsin, 271-276 in secretory cells, 298-299 in synaptic membranes, 286-291 in toad bladder, 298 Membrane-solution interfaces electrostatic potentials at, 71-144 fixed charges at, 73-93
392
SUBJECT INDEX
Membranes charged molecule adsorption to, 93-108 of mitochondria, calcium transport across, 321-366 permeation of charged molecules through, 124-126 Mg-ATPase, in membranes, thyroid effects on, 379-383 Mtcmcoccus denitrijicans
nitrate-dependent ATP synthesis in,
198 nitrate respiration in, 188-189 Microtubules, phosphorylation in, 293297 Mitchell hypothesis of uncoupler action, 130 Mitochondria calcium accumulation by, 341-342 calcium release by, 342-344 calcium transport across membranes of, 321-366 Molecular dipoles, at membrane-solution interfaces, 108-113 Monazomycin, as surface-potential “probe,” 91 Monolayers, surface potential studies with, 89-91 Muscle membrane phosphorylation in, 276-286 mitochondrial calcium transport in, 348-349 sodium-transport respiration in, 370 water in, NMR spectroscopy of, 38 Muscular dystrophy, erythrocyte membrane phosphorylation in, 269 Myelin, phosphorylation in, 291-293
N NaK-ATPase, in membranes, thyroid effects on, 379-383 Nervous tissue, mitochondrial calcium transport in, 347-348 Neuronal transmission, phosphorylation role in, 289-291 Nitrate respiration active transport coupled to, 204-208 in bacteria, 180-189 ATP synthesis coupled to 197 NMR spectroscopy, of intracellular fluids, 1-69
alkali ions, 41-58 in intracellular fluids, 48-58 in model systems, 44-48 comparison with other methods, 5 principles of, 6-17 techniques of, 17-19 water, 19-41 in ordered model systems, 21-27 Nonactin, as “molecular voltmeter,” 82 Nonequilibrium thermodynamic (NET) approach experimental evaluation, 149-162 theory of, 147-149
0 Opsin kinase, in rhodopsin phosphorylation, 273-274 Osmolarity, of solutions in small vesicles, 128-129
P Passive dausion, in active transport, 199 Phenylalanine hydroxylase phosphorylation of, 249 Phloretin, effects on dipole potential, 111 Phosphatidylcholine exchange protein, surface potential effects on, 129 Phospholipid bilayer, electric potential of, 120 Phosphoprotein phosphatases, membrane-bound, 299-303 Phosphorylase kinase, phosphorylation by, 245-246 Phosphorylase phosphatase, phosphorylation of, 248 Phosphorylation, of membranes, see Membrane phosphorylation Photochemical reactions, surface potential studies on, 127 Photoreceptors, phosphorylation in, 276 Pigments, in studies of photochemical reactions by surface potential, 127 Pinocytosis, surface potential effects on, 129 Pituitary plasma membranes, phosphorylation of, 298-299 Plant cells, ion uptake by, surface potential effects on, 129 Poisson-Boltzmann relation, 77,80,131
393
SUBJECT INDEX
Poisson equation, 77, 130 Potassium NMR spectroscopy of, 42 in thyroid sodium-transport regulation, 375-379 Propionibacterium arabinosum, fumarate reduction in, 194 Propionibacterium freudenreichii, fumarate reduction in, 198 Protein kinases membrane phosphorylation and, 233320 b y cyclic AMP-dependent enzymes, 237-251 b y cyclic AMP-independent enzymes, 251-252 membrane bound, 254-258 microtubule-associated, 295-296 Protein solutions, water in, NMR spectroscopy of, 27-34 Proteins, calcium-binding, in mitochondrial transport, 329-330 Proteus mirabilis, nitrate respiration in, 197-198 Proteus rettgeri fumarate reduction in, 193-194 ATP synthesis coupled to, 198-199 Proton-motive force, in anaerobic active transport, 213-219 Pseudomonas sp. nitrate respiration in, 180 ATP synthesis coupled to, 197 Pyruvate kinase, phosphorylation of, 248
R Respiratory chain, energy-dependent processes and, 179 Rhizobium japontcum, nitrate respiration in, 188 Rhodopsin phosphorylation of, 271 physiological role, 275-276 sites of, 274-275 Ribosomes, phosphorylation of, 250 RNA polymerase, phosphorylation of, 249-250
S Salicylamide, effects on dipole potential, 111-112
Salmonella typh ymurium, nitrate respiration in, 188 Sarcoplasmic reticulum, membrane phosphorylation in, 282-283 Saxitoxin, permeation of, through membranes, 124 “Screening” effects, on surface potential, 85 SDS, effects on dipole potential, 111 Secretory cells, phosphorylation in, 298-299 Selemonas ruminatium fumarate reduction in, 195 ATP synthesis coupled to, 198-199 nitrate respiration in, 189 Serum albumin in solution, NMR spectroscopy of, 29 Sickle cell disease, erythrocyte membrane phosphorylation in, 270 Skeletal muscle, membrane phosphorylation in, 282-285 Smooth muscle, membrane phosphorylation in, 285-286 Sodium ion, NMR spectroscopy of, 41-44 Sodium transport (active) affinity in, 156-162, 165-169 electrochemical potential difference in, 150-156 equivalent circuit model in analysis of, 162-164 experimental evaluation, 165 flows and forces in, 150-162 hypothesis of, 368-369 stoichiometry of, 150 thermodynamics of, 145-175 thyroid regulation of, 367-388 Spherical liposomes, 73 Spherocytosis, erythrocyte membrane phosphorylation in, 269 Staphylococcus aureus, nitrate respiration in, 188 Stem equation, 96-99 experimental tests of, 99-108 Streptococcus faecalis fumarate reduction in, 194, 195 ATP synthesis coupled to, 198 Subcellular compartmentalization, definition of, 2-3 Subcellular organelles, method for obtaining, 3-4
3 94
SUBJECT INDEX
Surface potential carriers as “probes” of, 82-87 photochemical reaction studies by, 127 Synaptic membranes, phosphorylation of, 286-291 Synaptosomes, phosphoprotein phosphatases in, 302
T 3T3 cells, surface potential studies on, 130 Tetraphenylborate, adsorption to bilayer membranes, 113-114 Tetrodotoxin, permeation of, through membranes, 124-125 Thermodynamics, of active sodium transport, 145-175 Thiobacillus a!enitrl&ans, nitrate respiration in, 188 Thyroid gland in sodium transport regulation, 367-388 potassium in, 375-379 Thyroid hormone, in sodium-transport respiration, 372-374 TNS, adsorption of, to bilayers, 126-127 Toad bladder, phosphorylation in, 297298 Transport, composite system for, model of, 148
Troponin, phosphorylation of, 249 Tubulin, phosphorylation of, physiological significance of, 297 Tumor cells, calcium accumulation in, 351-352 Tyrosine hydroxylase, phosphorylation of, 250-25 1
V Veillonella alcalescens active transport in, 208 fumarate reduction in, 195 nitrate respiration in, 189 Vesicles, small, osmolarity of solutions in, 128-129 Volmer isotherm, 95
W Water in biological systems, NMR spectroscopy of, 19-41 ordered model systems, 21-27 protein solutions, 27-34
Z Zeta potential, measurements of, 87-88
A 8 7
c a D 9 E O
F 1 6 2
H 3 1 4 1 5