-# :j/pi>L f^^ C E E [I II II C c E C C E C C E E K 3S^SBSS^^^SSS^^SE Marine Biological Laboratory Library Woods Hole, Mass. .^=>#<'=s. Presented by Dr. Philip Person Feb. 26, 1962 s 3ES^^^^^^^^^^^^^83E := i-n (_) WHO nj D i_) lAI r^ LJ m LJ HAEMATIN ENZYMES I.U.B. Symposium Series Volume 19 INTERNATIONAL UNION OF BIOCHEMISTRY SYMPOSIUM SERIES Vol. 1. The Origin of Life on the Earth — A. I. Oparin ?/ a/. (Editors) Vol. 2. Enzyme Chemistry: Proceedings of the International Symposium in Tokyo- Kyoto PROCEEDINGS OF THE FOURTH INTERNATIONAL CONGRESS ON BIOCHEMISTRY VIENNA, 1958 Vol. 3. (I) Carbohydrate Chemistry of Substances of Biological Interest Vol. 4. (II) Biochemistry of Wood Vol. 5. {l\\) Biochemistry of the Central Nervous System Vol. 6. (IV) Biochemistry of Steroids Vol. 7. (V) Biochemistry of Antibiotics Vol. 8. (VI) Biochemistry of Morphogenesis Vol. 9. (VII) Biochemistry of Viruses Vol. 10. (VIII) Proteins Vol. 11. (IX) Physical Chemistry of High Polymers of Biological Interest Vol. 12. (X) Blood Clotting Factors Vol. 13. (XI) Vitamin Metabolism Vol. 14. (XII) Biochemistry of Insects Vol. 15. (XIII) Colloquia Vol.16. (XIV) Transactions of the Plenary Sessions Vol. 17. (XV) Biochemistry Vol. 18. Biochemistry of Lipids — G. Popjak (Editor) Vol. 19. Haematin Enzymes (Parts 1 and 2) — J. E. Falk, R. Lemberg and R. K. Morton Vol. 20. Report of the Commission on Enzymes, 1961 (I.U.B.) PROCEEDINGS OF THE HFTH INTERNATIONAL CONGRESS ON BIOCHEMISTRY MOSCOW, 1961 {Provisional titles) Vol. 21. (I) Biological Structure and Function at the Molecular Level Vol. 22. (II) Functional Biochemistry of Cell Structures Vol. 23. (Ill) Evolutionary Biochemistry Vol. 24. (IV) Molecular Basis of Enzyme Action and Prohibition Vol. 25. (V) Intracellular Respiration: Phosphorylating and Non-Phosphorylating Systems Vol. 26. (VI) Mechanism of Photosynthesis Vol. 27. (VII) Biosynthesis of Lipids Vol.28. (VIII) Biochemical Principles of the Food Industry Vol. 29. (IX) Transactions of the Plenary Sessions Vol. 30. (X) Abstracts of Papers and Indexes to the Volumes of the Proceedings The building of the Australian Academy of Science, Canberra, where the Symposium was held. HAEMATIN ENZYMES A SYMPOSIUM OF THE INTERNATIONAL UNION OF BIOCHEMISTRY ORGANIZED BY THE AUSTRALIAN ACADEMY OF SCIENCE CANBERRA 1959 Edited by J. E. Falk, R. Lemberg and R. K. Morton PART 1 (Pages 1 to 362) SYMPOSIUM PUBLICATIONS DIVISION PERGAMON PRESS OXFORD • LONDON • NEW YORK • PARIS 1961 PERGAMON PRESS LTD. Headington Hill Hall, Oxford 4 & 5 Fitzroy Square, London W.I PERGAMON PRESS INC. 122 East 55th Street, New York 22, N. Y. 1404 New York Avenue N.W., Washington 5 D.C. Statler Center 640, 900 Wilshire Boulevard, Los Angeles 17, California PERGAMON PRESS S.A.R.L. 24 Rue des £coles, Paris V^ PERGAMON PRESS G.m.b.H. Kaiserstrasse 75, Frankfurt am Main Copyright © 1961 Pergamon Press Ltd. LIBRARY OF CONGRESS CARD NUMBER 60-53463 Set in Monotype Times 10ll2pt and Printed in Great Britain by the Pitman Press, Bath PREFACE This volume contains the papers and discussion material presented at the Symposium on Haemotin Enzymes. It was held at Canberra between 31st August and 4th September, 1959, and was organized by the Australian Academy of Science for the International Union of Biochemistry. The Symposium was arranged for the Academy by a Committee com- prising A. H. Ennor, J. E. Falk, R. Lemberg and R. K. Morton (Convener). The Committee is grateful to several organizations, cited in the address by Dr. R. Lemberg, President of the Symposium, for financial and other assistance. The titles and addresses of participants are given on pp. xvii-xx. For convenience, titles have been omitted from the scientific communications. It is with profound regret that we record the untimely death in October, 1960, of Professor Enzo Boeri, one of the distinguished participants in the Symposium. He made many notable contributions to our knowledge of haematin enzymes and he will be remembered with admiration, respect and affection by all who were privileged to know him. R. K. Morton ^ f L ! S .*? A -^ S '" CONTENTS ^ PAGE Participants ........... xvii Presidential Address ......... xxi The Electronic Structure and Electron Transport Properties of Metal Ions Particularly in Porphyrin Complexes by L. E. Orgel 1 Discussion Terminology in Ligancl-Field Theory . . . . . .13 Spin States of Haem Compounds . . . . . . .15 Electron Transport . . . . . . . . .16 Mechanism of Oxidative Phosphorylation . . . . .18 The Role of the Metal in Porphyrin Complexes by F. P. DwYER 19 Discussion Higher Oxidation States ........ 27 Effects of Metal on Reactivity at Periphery . . . . .28 The Physico-Chemical Behaviour of Porphyrins Solubilized in Aqueous Detergent Solutions by B. Dempsey, M. B. Lowe and J. N. Phillips . . .29 Discussion Cations of Porphyrins and Their Spectra . . . . .37 The Reactions Between Metal Ions and Porphyrins by J. H. Wang and E. B. Fleischer ...... 38 Some Physical Properties and Chemical Reactions of Iron Complexes by R. J. P. Williams 41 Discussion Oxidation-Reduction Potentials of Haem Compounds . . . .53 vii 79501 via CONTENTS PAGE Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins by J. E. Falk and D. D. Perrin 56 Discussion Correlations between Structure and Physical Properties. . . .71 Models for Haemoproteins ....... 74 Some New Compounds of Haems with Bases by J. E. Falk 74 Carbon Monoxide-Pyridine Complexes with Haems by J. H. Wang 76 Equilibrium Constants for Reactions of Haems with Ligands by J. N. Phillips 79 Modification of the Secondary Structure of Haemoprotein Molecules by K. Kaziro and K. Tsushima 80 Discussion The Haem-Binding Groups in Haemoproteins . . . . .94 The Nature of Haem-Binding, and the Bohr Effect by J. H. Wang and Y. N. Chiu 94 Models for Linked Ionizations in Haemoproteins by P. George, G. I. H. Hanania, D. H. Irvine and N. Wade . . .96 On the Stability of Oxyhaemoglobin by J. H. Wang 98 Discussion Oxygenation of Haemoglobin . . . . . . .102 Ferrihaemoprotein Hydroxides: A Correlation between Magnetic and Spectroscopic Properties by P. George, J. Beetlestone and J. S. Griffith . . . 105 Discussion Spin States and Spectra of Haemoproteins . . . . .139 The Electronic Origins of the Spectra by J. S. Griffith and P. George 139 Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins by D. L. Drabkin 142 Discussion Interpretation of Absorption Spectra of Haemoproteins . . .170 The Bands in the Region of 830 and 280 m/t . . . . .171 CONTENTS IX PAGE The Haem-Globin Linkage. 3. The Relationship between Molecular Structure and Physiological Activity of Haemoglobins by J. E. O'Hagan 173 Discussion Native globin . . . . . . . . .190 The Linkage of Iron and Pi otein in Haemoglobin . . . .190 Early Stages in the Metabolism of Iron by J. B. Neilands 194 The Enzymic Incorporation of Iron into Protoporphyrin by R. A Neve 207 Discussion The Formation of Metal-Porphyrin Complexes . . . . .211 Co-ordination of Divalent Metal Ions with Porphyrin Derivatives Related to Cytochrome c by J. B. Neilands . .211 Metal Incorporation in Model Systems . . . . , .214 On the Enzymic Incorporation of Iron . . . . . .215 Biosynthesis and Metabolism of Cytochrome c by D. L. Drabkin 216 The Location of Cytochromes in Escherichia coli by A. TissiERES . , 218 Discussion The Origin of the Respiratory Granules of Bacteria .... 223 On the Cytochromes of Anaerobically Cultured Yeast by Paulette Chaix 225 Discussion The Lactate Dehydrogenase of Yeast . . . . . .233 Components of the Respiratory Chain in Yeast Mitochondria . .233 On the '^ Haemoglobin' Afjsorption Bands of Yeast .... 233 X CONTENTS PAOB Irreversible Inhibition of Catalase by the 3-Amino-l: 2:4-Triazole Group of Inhibitors in the Presence of Catalase Donors by E. Margoliash and A. Schejter 236 Catalase Oxidation Mechanisms by M. E. WiNFiELD 245 Discussion Oxidation States of Haemoproteins ...... 252 Peroxide Compounds of Catalase and Peroxidase .... 254 The Nature of Catalasc-Peroxide Complex I by B. Chance 254 Studies on Problems of Cytochrome c Oxidase Assay by LuciLE Smith and Helen Conrad 260 Discussion Assay of Cytochrome c Oxidase ...... 275 Inhibition of Cytochrome c Oxidase by Cytochrome c ... 275 Interaction of Cytochrome c with Other Compounds .... 276 The EflFect of Cations on the Reactivity of Cytochrome c in Heart Muscle Preparations by R. W. Estabrook . 276 Composition of Cytochrome c Oxidase by W. W. Wainio 281 Discussion Function of Copper in Cytochrome Oxidase Preparations . . . 301 Cytochrome Oxidases of Pseudomonas aeruginosa and Ox-Heart Muscle and Their Related Respiratory Components by T. HoRio, I. Sekuzu, T. Higashi and K. Okunuki . . 302 Discussion Properties and Nomenclature of Cytochromes di and 3.^ . . .311 The Prosthetic Groups of Pseudomonas Cytochrome Oxidase by T. Horio and M. D. Kamen 314 The Reaction of Cytochrome c Oxidase with Oxygen . . . .316 The Oxygen-Reducing Equivalents of Cytochromes a and 03 by B. Chance 316 CONTENTS XI PAGE The Isolation, Purification and Properties of Haemin a by D. B. MoRELL, J. Barrett, P. Clezy and R. Lemberg . 320 Discussion Model Systems for Cytochrome Oxidase ..... 330 Absorption Spectra of Ferro- and Ferri-Compounds of Haem a by R. Lemberg ......... 330 Cryptohaem a ........ . 333 Mitochrome in Relation to Cryptohaem a .... . 334 Cytochrome Oxidase Components by M. Morrison and E. Stotz 335 The Structure of Porphyrin a, Cryptoporphyrin a and Chlorin «2 by R. Lemberg, P. Clezy and J. Barrett .... 344 Discussion The Structure of Haem ?i and Haem a.2 . ..... 358 The Structure of Porphyrin a by M. Morrison ......... 358 The Properties of Haem o^ and Cytochrome a.^ by J. Barrett and R. J. P. Williams . . . , . .360 Extractability of Ferro- and Ferricytochrome c . . . . .361 A Haemopeptide from a Tryptic Hydrolysate of Rhodospirillum rubrum Cytochrome c by S. Palpus and H. Tuppy 363 Electrometric and other Studies on Cytochromes of the C-Group by R. W. Henderson and W. A. Rawlinson .... 370 Discussion Protein Configuration and Linkage to the Prosthetic Group in Cytochrome c 382 Studies of the Haemochrome-forming Groups in Cytochrome c by E. Margoliash 382 The Amino Acid Sequence in Horse Heart Cytochrome c by E. Margoliash and R. Hill 383 Comments on the Structure of Cytochrome c . . . . . 384 Structure and General Properties of Cytochrome c . . . .385 XU CONTENTS Comparative Properties of Cytochrome c from Yeast and Heart Muscle by J. McD. Armstrong, J. H. Coates and R. K. Morton Properties of Native Cytochrome c . . . . Reactivity of Native Cytochrome c in Oxidative Phosphorylation Structure of Bacterial Cytochromes of c-Type . Structure and Redox Potentials of Cytochrome c 385 388 389 389 390 The Electron Transfer from Cytochromes to Terminal Electron Acceptors in Nitrate Respiration and Sulphate Respiration by F. Egami, M. Ishimoto and S. Taniguchi .... 392 Cytochrome Cg by J. PosTGATE 407 Discussion Nature and Properties of Cytochrome C3 . . . . . .414 Functional Aspects of Cytochrome c^ ■ • • • • .415 Evolutionary Aspects of the Sulphate-reducing Bacteria and of Cytochrome C3 416 The Atypical Haemoprotein of Purple Photosynthetic Bacteria by M. D. Kamen and R. G. Bartsch 419 Discussion The Functions of Cytochrome h^ and of Cytochrome c^** (Halotolerant Coccus) ....... 432 Nomenclature of CO-hinding Pigments ...... 432 Cytochrome o by B. Chance 433 On the Oxidase Function of RHP . . . . . . .435 Spectrophotometric Studies of Cytochromes Cooled in Liquid Nitrogen by R. W. EsTABROOK 436 Discussion 'Trapped' Steady-states by B. Chance ... 457 Low-temperature Absorption Spectra of Cytochromes in Relation to Structure . . ...... 458 CONTENTS Xlll PAGE Studies on Microsomal Cytochromes and Related Substances by C. F. SlRlTTMATTER 461 Discussion On the Nature of Cytoplasmic Pigments of Liver Cells by B. Chance 473 Possible Functions of the Cytochromes of the Endoplasmic Reticuhun of Animal Cells 476 The Significance of Eq Values of Cytochromes in Relation to Cellular Function ......... 477 The Cytochromes of Plant Tissues by W. D. Bonner, Jr . . .479 Discussion Cytochromes Ci and 63 of Particulate Components of Plants by R. K. Morton 498 The Cytochromes of Roots ....... 499 The Chemical and Enzymic Properties of Cytochrome Z^g of Bakers' Yeast by R. K. Morton, J. McD. Armstrong and C. A. Appleby . 501 Conditions for the Autoxidation of Flavocytochrome Z>2 by E. Boeri and M. Rippa 524 Kinetic Studies on the Action of Yeast Lactate Dehydrogenase by H. Hasegawa and Y. Ogura 534 Various Forms of Yeast Lactate Dehydrogenase by A. P. Nygaard 544 Studies on Bakers' Yeast Lactate Dehydrogenase by T. Horio, J. Yamashita, T. Yamanaka, M. Nozaki and K. Okunuki 552 Discussion The Problem of Cytochrome bg . ...... 558 Properties of Intact and Modified Cytochrome h.^ .... 560 XIV CONTENTS PAOB Nature of Bakers' Yeast Lactate Dehydrogenase by T. Horio 560 Nomenclature of Cytochrome h^ and Derived Proteins . . . 562 The Substrate Specificity of Cytochrome b2 . .... 563 On the Cytochrome b Components in the Respiratory Chain in Yeast . 564 The Kinetics of Reactions Catalysed by Cytochrome h^ ■ • . 565 The Kinetics of Reduction of Cytochrome b^ by B. Chance 565 The Absorption Spectrum in Relation to the Structure of Cytochrome hz 565 The Contribution of the Prosthetic Groups to the Absorption Spectrum of Cytochrome 62 by R. K. Morton 567 The Oxidation-reduction Changes in the Reaction of Lactate with Cytochrome 62 by E. Boeri and E. Cutolo ........ 568 The Function and Bonding of the Flavin Group of Cytochrome bg . . 569 The Bonding between the Flavin Group and Apoprotein of Cytochrome 62 by J. McD. Armstrong, J. H. Coates and R. K. Morton . . . 569 Possible Free Radical Formation in Flavoproteins byG.D. Ludwig 572 Autoxidation of Cytochrome h^ ...... . 573 The Role of Cytochrome b in the Respiratory Chain by E. C. Slater and J. P. Colpa-Boonstra .... 575 Discussion The Cross-over Theorem and Sites of Oxidative Phosphorylation . . 592 The Oxidation-reduction Potential of Cytochrome b . . . . 593 On the Redox Potential of Cytochrome b, the Kinetics of Reduction of Cytochrome h and the Existence of Slater' s Factor . . . 593 The Influence of Cyanide on the Reactivity of Cytochrome b . . . 596 Energy Transfer and Conservation in the Respiratory Chain by B. Chance 597 Discussion Mechanism of Oxidative Phosphorylation ..... 622 CONTENTS XV PAGE The Significance of Respiratory Chain Oxidations in Relation to Metabolic Pathways in the Cell by F. Dickens 625 Discussion On 'Additional DPN' of Incubated Mitochondria .... 636 Possible Structure of Complexes of DPN or of DPNH involved in Oxidative Phosphorylation ........ 637 Author Index .....,,... 641 Subject Index .......... 653 PARTICIPANTS Dr. C. a. Appleby Mr. J. McD. Armstrong Mr. J. Barrett Dr. N. K. Boardman Professor E. Boeri* Dr. W. D. Bonner Mme. Professor P. Chaix Professor B. Chance Dr. p. S. Clezy Dr. E. Cutolo Professor F. Dickens Professor D. L. Drabkin Professor F. P. Dwyer * Died, 1960. H.E. — VOL. I — B Biochemistry Section, Division of Plant Industry, C.S.I. R.O., Canberra, Australia. Department of Agricultural Chemistry, Waite Agricultural Research Institute, University of Adelaide, Adelaide, Australia. Institute of Medical Research, Royal North Shore Hospital, Sydney, AustraUa. Biochemistry Section, Division of Plant Industry, C.S.I.R.O., Canberra, Austraha. Institute of Human Physiology, University of Ferrara, Ferrara, Italy. Johnson Research Foundation, University of Pennsylvania, Philadelphia, U.S.A. Laboratory of Biological Chemistry, Univer- sity of Paris, Paris, France. Johnson Foundation for Medical Physics, University of Pennsylvania, Philadelphia, U.S.A. Institute of Medical Research, Royal North Shore Hospital, Sydney, Australia. Itahan Serum Research Institute, Naples, Italy. Courtauld Institute of Biochemistry, Middle- sex Hospital Medical School, University of London, London, England. Department of Biochemistry, Graduate School of Medicine, University of Pennsylvania, Philadelphia, U.S.A. The John Curtin School of Medical Research, Australian National University, Canberra, Australia. xvii PARTICIPANTS Professor F. Egami Dr. R. W. Estabrook Dr. J. E. Falk Professor P. George Dr. R, W. Henderson Dr. T. Horio Professor M. D. Kamen Professor K. Kaziro Mr. J. W. Legge Dr. R. Lemberg, Prof. a. D. (Heidelberg) Mr. W. H. Lockwood Dr. G. D. Ludwig Dr. E. Margoliash Dr. D. B. Morell Dr. M. Morrison Department of Biophysics and Biochemistry, Faculty of Science, University of Tokyo, Tokyo, Japan. Johnson Foundation for Medical Physics, University of Pennsylvania, Philadelphia, U.S.A. Biochemistry Section, Division of Plant Industry, C.S.I.R.O., Canberra, Austraha. John Harrison Laboratory of Chemistry, University of Pennsylvania, Philadelphia, U.S.A. Department of Biochemistry, University of Melbourne, Melbourne, Australia. Graduate Department of Biochemistry, Brandeis University, Waltham, U.S.A. Graduate Department of Biochemistry, Brandeis University, Waltham, U.S.A. Biochemical Laboratory, Nippon Medical School, Tokyo, Japan. Department of Biochemistry, University of Melbourne, Melbourne, Australia. Institute of Medical Research, Royal North Shore Hospital, Sydney, Australia. Institute of Medical Research, Royal North Shore Hospital, Sydney, Austraha. Hospital of the University of Pennsylvania, Philadelphia, U.S.A. Laboratory for the Study of Hereditary and Metabolic Disorders, College of Medicine, University of Utah, Salt Lake City, U.S.A. Institute of Medical Research, Royal North Shore Hospital, Sydney, Australia. Department of Biochemistry, School of Medicine and Dentistry, University of Rochester, Rochester, U.S.A. PARTICIPANTS Professor R. K. Morton Dr. F. J. Moss Professor J. B. Neilands Dr. R. a. Neve Dr. a. p. Nygaard Dr. Y. Ogltra Dr. J. E. O'Hagan Dr. L. E. Orgel Dr. S. Paleus Dr. D. D. Perrin Dr. J. N. Phillips Dr. J. Postgate Assoc. Professor W. A. Rawlinson Professor E. C. Slater Dr. L. Smith Dr. C. F. Strittmatter Department of Agricultural Chemistry, Waite Agricultural Research Institute, University of Adelaide, Adelaide, Australia. School of Biological Sciences, University of New South Wales, Sydney, Australia. Department of Biochemistry, University of California, Berkeley, U.S.A. Department of Biochemistry, University of California, Berkeley, U.S.A. Nutrition Institute, Blindern, Oslo, Norway. Department of Biophysics and Biochemistry, Faculty of Science, University of Tokyo, Tokyo, Japan. Red Cross Blood Transfusion Service, Brisbane, Austraha. University Chemical Laboratory, University of Cambridge, Cambridge, England. Biochemistry Department, Nobel Institute of Medicine, Stockholm, Sweden. Department of Medical Chemistry, John Curtin School of Medical Research, AustraHan National University, Canberra, Australia. Biochemistry Section, Division of Plant Industry, C.S.I.R.O., Canberra, Australia. Microbiological Research Establishment, Porton, Wiltshire, England. Department of Biochemistry, University of Melbourne, Melbourne, Australia. Laboratory of Physiological Chemistry, University of Amsterdam, Amsterdam, The Netherlands. Department of Biochemistry, Dartmouth Medical School, Hanover, New Hampshire, U.S.A. Department of Biological Chemistry, Harvard Medical School, Boston, U.S.A. PARTICIPAhfTS Dr. a. Tissieres Dr. p. Trudinger Professor S. F. Velick Professor W. W. Wainio Professor J. H. Wang Dr. R. J. P. Williams Dr. M. E. Winfield The Biological Laboratories, Harvard Univer- sity, Cambridge, U.S.A. Biochemistry Section, Division of Plant Industry, C.S.I.R.O., Canberra, Australia. School of Medicine, Washington University, St. Louis, Missouri, U.S.A. Bureau of Biological Research and Depart- ment of Physiology and Biochemistry, Rutgers, The State University, New Brunswick, New Jersey, U.S.A. Department of Chemistry, Yale University, New Haven, U.S.A. Inorganic Chemistry Laboratory, and Wad- ham College, Oxford, England. Division of Physical Chemistry, Indus- trial Chemical Laboratories, C.S.I.R.O., Melbourne, Australia. PRESIDENTIAL ADDRESS I DECLARE open the Symposium on Haematin Enzymes of the International Union of Biochemistry and welcome all who attend it. It is a great pleasure for me to welcome the many distinguished scientists from overseas who have come to Austraha to discuss with us the problems which interest us all. I hope that the stimulation which you may receive will repay you for your long and strenuous journeys to our distant shores, that you may carry back happy memories of this week spent in Australia — and that you wiU return! Your visit will certainly be a stimulus to Australian science. I am particularly happy to receive you in this home of the Australian Academy of Science.* When we discussed plans for its erection in the Council of the Academy only a few years ago, I little dreamt that I should have the honour of opening the first International Conference in it. Our thanks are due to the International Union of Biochemistry, not only for accepting our invitation to hold the Symposium, but also for financial support; to the AustraUan Commonwealth Government; to the Wellcome Trust ; and to the various Academies and national bodies, particularly to the National Science Foundation of the United States and to the Royal Society. The support of these various organizations made this meeting possible. We decided that we wanted an intimate symposium in which all participants could be rehed upon to make valuable contributions. We wanted to discuss our problems critically, but with some degree of leisure. This is a Symposium and even if we cannot do as the Greeks did, having a meal, wine and dancers in this hall, the comfortable lounge chairs are the nearest approach to it possible in these hurried times. There you may recline for a little snooze if the richness of the intellectual feast endangers your mental digestion. Our Symposium has a pecuhar note in that it calls together scientists of different branches, from quantum mechanics to microbiology, and asks them to direct the spotlights of their knowledge on to a comparatively narrow field, but a field of great biological importance and chemical interest. After reading the prepublished papers I am convinced that we were right in assuming that the difficulties of finding sufficient common ground for our various denomin- ations are no longer insuperable. Still, we shall have to exert some patience and forbearance. I ask the theorist not to be impatient with the experimenter, if he asks questions which reveal his lack of knowledge of theory, but to answer them in * See frontispiece {Editors). XXII PRESroENTIAL ADDRESS brotherly love ; I ask the experimenter not to be shy to ask such questions, for they may turn out to be quite searching and may indeed enforce modifications of theory. Again, I ask the theorist not to be shy to apply his theories to facts with which he may become familiar only at this meeting; and the experi- menter to try to enlighten the theorist about these facts, again in brotherly love. Finally, I am convinced that quite apart from the direct scientific results of the Symposium, our living together here for one week will cement bonds of friendship and comradeship which will remain a force long after the Symposium. R. Lemberg THE ELECTRONIC STRUCTURE AND ELECTRON TRANSPORT PROPERTIES OF METAL IONS PARTICULARLY IN PORPHYRIN COMPLEXES By L. E. Orgel Department of Theoretical Chemistry, University Chemical Laboratory, Cambridge INTRODUCTION The electronic structure of metal-porphyrins has often been discussed in terms of the valence-bond theory as developed by Pauling (1940). In recent years a related but more quantitative theory of metal complexes has been developed and is known as ligand-field theory (Griffith and Orgel, 1957; Moffitt and Ballhausen, 1956). The first part of this paper attempts to give an elementary account of this theory insofar as it is of interest to biochemists working with haem compounds. In particular I shall discuss the relevance of magnetic susceptibility and magnetic resonance data. In the later part of my paper I shall discuss some recent work on electron-transfer processes involving metal ions, and also show how the electronic structures of the haem enzymes may be relevant to the types of electron-transfer which can take place. LIGAND-FIELD THEORY OF REGULAR OCTAHEDRAL COMPLEXES The five 3- Distortion >- (a) (b) E^., ' Fig. 5. Energy level diagram showing the effect of tetragonal distortions; (a) electrostatic theory; (b) a possible consequence of double bonding. z axis are gradually removed. (This is formally equivalent to replacing two ligands by groups which produce smaller crystal fields.) In Fig. 5a we illus- trate the results of calculations based on the electrostatic theory, and in Fig. 5b the way in which these calculations might be modified by covalent bonding effects. The principal features, namely the splitting far apart of the d^i_y2 and d^i orbitals and the maintaining of the degeneracy of the d^.^ and dy^ orbitals are unaffected by covalent bonding, but the order of the d^^y orbital and the d^.^ and dy^ orbitals might be altered. If the d^i. orbital becomes much more stable than the dj,2_yi orbital all the electrons may crowd together in the bottom /ow orbitals. Then we would get 0, 1, 2 and 3 unpaired electrons in d^, d'', d^ and d^ ions, respectively. These extreme conditions seem to apply in nickel phthalocyanine (diamag- netic), cobalt phthalocyanine (one unpaired electron) and ferric phthalo- cyanine chloride (three unpaired electrons). Electronic Structure and Electron Transport Properties of Metal Ions 7 In the biologically important haem compounds the environment seems more nearly octahedral and no "intermediate-spin" derivatives are known. (Magnetic resonance shows conclusively that ferrihaemoglobin hydroxide exists in a mixture of high- and low-spin configurations.) Paramagnetic resonance absorption has not been detected in ferrous com- pounds and it is probable that even in the paramagnetic compounds it is outside the range of normal experiments. Paramagnetic resonance experi- ments, however, are perhaps the most sensitive tool at present available for the study of the detailed energy-level scheme both in high- and low-spin Fe+++ complexes. Once this is known inferences can be made about the detailed geometrical structure, and correlations established with other properties. PARAMAGNETIC RESONANCE EXPERIMENTS Here I can deal only with the energy-level diagrams deduced (mainly by Griffith, 1956, 1957, 1958) by an analysis of the experimental data of Ingram and co-workers (Gibson and Ingram, 1957; Gibson et al, 1958). The Table 2. Resonance characteristics of some haemoglobin derivatives Compound Spin-type ^-values Ferrihaemoglobi n High .-. = 2 ?, - 6 Ferrihaemoglobin fluoride High <5-l- " 0)5 t " t Weak bond Strong bond OMe O \v_C > ^ \.. transfer c \ O— Cr++^(HoO)5 O— Cr(H20)4(MeOH2)+ (ii) O— Cr+^+(H20)4(MeOH) + H+ In this system the reaction achieved is hydrolytic, that is, the redox reaction catalyses stoichiometrically the hydrolysis of the ester, but I shall now show that this is not an essential feature of the process. In order to bring the discussion closer to the subject of the cytochromes, I shall replace the reactants by ones which may be biologically important, but only for illustrative purposes : PO,H- / A Fe+++— O— (/ \)— O ^ \ Fe++ADP Weak bond Fe++— O— (/ \>— O PO.H- t Fe+++— ADP strong bond ^-^ > Fe++— O— M+++— Rj > M+++— Rx— Ra \ \ the "electron mediator" and the transferred group are part of the same molecule. The advantage of this in ensuring the efficient coupling between the redox and transfer processes is obvious. If mediated electron transfer is important in oxidative phosphorylation, then the details of the electronic configurations of the metal enzymes con- cerned will be critical to the understanding of the process. Here I can summarize only a few of the most important considerations : 1 . The effect of a metal ion in inducing the hydrolysis of a bond as in the reactions (ii) and (iii) will be large if the metal has a maximum number of spins paired after reaction, e.g. if the Fe^++ ion produced is low- spin rather than high-spin (of course it will also be greater for a trivalent than for a divalent ion, other things being equal). 2. The rate of transfer will be greatest if there is no need for a change of spin configuration during the process, i.e. if it occurs between two high-spin or two low-spin complexes. 3. The rate of transfer also depends on the degree of overlap between metal orbitals and mediator orbitals. This is greatest for low-spin complexes. 4. The most effective mediator molecules are likely to be ones which can act either as good electron donors or as good electron acceptors. In conclusion I should like to note that even if oxidative phosphorylation does not involve steps which have an obvious appeal to the theoretician, but rather a sequence of conventional coupled redox reactions, both the redox potentials and the rates of reaction will certainly depend critically on the spin-states of the metal ions and hence on the details of their interaction with their environments, REFERENCES George, P. & Griffith, J. S. (1959). The Enzymes, 2nd edition, Ed. Boyer, Lardy and Myrbiick, Chap. 8. Gibson, J. F. & Ingram, D. E. (1957). Nature, Lond. 180, 29. Electronic Structure and Electron Transport Properties of Metal Ions 13 Gibson, J. F., Ingram, D. J. E. & Schonland, D. (1958). Disc. Faraday Soc, No. 26, 72. Grifhth, J. S. & Orgel, L. E. (1957). Quart. Rev. XI, 381. Griffith, J. S. (1956). Proc. Boy. Soc. A. 235, 23. Griffith, J. S. (1956). J. Inorg. Nucl. Chem. 2, 229. Griffith, J. S. & Orgel, L. E., Unpublished calculations. Griffith, J. S. (1957). Nature, Lond. 180, 31. Griffith, J. S. (1958). Disc. Faraday Soc. 216,91. MoFFiTT, W. & Ballhausen, C. J. (1956). Ann. lier. phys. Chem. 7, 107. Orgel, L. E. (1956). Proc. lOtli Soliay Conference in Chemistry, Brussels. Paultng, L. (1940). The Nature of tiie Chemical Bond, Cornell University Press. Taube, H. (1959). Advances in Inorganic and Radio Chemistry 1, 1, Academic Press, London, New York. Eraser, R. T. M., Sebera, D. K. & Taube, H. (1959). /. Amcr. chem. Soc. 81, 2906. DISCUSSION Terminology in Ligand-Field Theory Williams: Dr Orgel's paper contains a discussion of the quantity. A, which is used by him both in thermodynamic and spectroscopic calculations. In one place it is sug- gested that there is a critical value of A such that if it is exceeded a change of paramagnetic mom.ent would be observed. A is here used in a thermodynamic argu- ment and is an indication of the field strength in the ground state of a complex. At another place it is stated that extensive studies of spectra have led to the relative values of A ; A is related to an excitation energy difference between two states, the ground and the excited states. What is A ? Is it a parameter of the field or is it only to be correlated with differences in character between excited and ground states, or does it represent a confusion of these two factors ? We shall now indicate why we consider the last statement to be true. The order of A given by Orgel for a series of ligands is part of the spectro-chemical series I- < Br- < CI- < F- < OH- < H2O < carboxylate- < SCN" < pyridine < NH3 < NOr < CN-. This series is often confused with a series of increasing field strength (Williams, J. chem. Soc. 8 (1956)). The effect of ligands in reducing the paramagnetic moment of ferric complexes (Scheler, Schoffa & Jung, Biochem. Z. (1957) 329, 232) follows the different order F- < CI- < H2O < carboxylate- < OCN" < SCN- < OH- < NOr < NH3 < pyridine < N3- < CN-; note particularly the underlined ligands. Again there is no doubt that the stability of metal complexes with different ligands does not follow a given order of ligands. With some cations an order somewhat like the spectrochemical series is observed, with others the order is almost completely reversed. Thus experiment shows that A, obtained from spectra, is not to be used as a guide in discussions of thermodynamic quantities (see Jorgenson, Orgel, Williams, Disc. Faraday Soc, 26, 1958, p. 123-130, 110-115, 180-187). However, theory (same references as above) also shows why this is so dangerous. The correlation of A with field strength is only apparent in a first order perturbation treatment of a simple electrostatic crystal field model. A second order perturbation treatment introduces polarization of the ligand dependent upon the cation or, if one wishes to put it this way, covalency, and changes in the radial as well as the angular dependences of the d wave-functions aftect the energies of states. This was pointed out by Owen {Disc. Faraday Soc, 19 (1955)). 14 Discussion This covalency is in part independent of field symmetry and is often called central field covalency. It is best explained by saying that not only is there a tendency for the d electrons of the cation to go into particular directions in space when a field is applied but that they spread out radially in space over the ligands also. The stabihty of a complex depends not only on the ability of a ligand to polarize the d electrons into given directions, where they get out of the way of the ligand electrons, but also on the ability of the ligands to allow the d electrons to spread out over them and the ability of the cation to allow the ligand's electrons to spread over it. Jorgensen has shown that the series of ligands which allow increasingly the d electrons to spread out, is F- < H2O < NH3 < L^^^^'g) < SCN- < Pyridine < Cl^ < CN- < Bi- < (I-?) ; Note the underlined Hgands. This series is different again from the spectrochemical series. Thermodynamic properties such as the stabilities of complex ions and the relative stability of two spin states are just as likely to follow either one of the two series, spectrochemical and nephelauxetic, which are both derived from spectra. The order of ligands which will bring about a change of spin state is even dependent upon the valency state of the cation. Orgel: 1. The procedures used to derive A, the spectroscopic ligand-field strength, and to determine the "nephelauxetic" series are well defined. To suppose that a single A suffices for ground and excited states of a complex is an approximation, but one which is justified by the success of the simple theory in interpreting spectra. If, as Williams suggests, A varies greatly from state to state, the ligand-field theory would never have been adopted, since it would have failed to accommodate even the simplest observa- tions on the spectra of high-field complexes. 2. The approximations implicit in the treatment of spin-pairing are more serious since (a) The internuclear distances decrease on spin-pairing and so the ligand-field increases (this is in contrast to the situation encountered in spectroscopy, where all energies are determined for the same internuclear distance). (See, for example, Orgel, 10th Solvay Conference Proceedings, Brussels, 1955.) (b) Increased delocalization in the low-spin state facilitates spin-pairing. (See, for example, the many papers on Co+++ in the spin-paired state.) It may well be that in special cases these effects can change slightly the ligand-field order; I would only question whether any experimental evidence for this is available in the regular octahedral complexes to which the theory applies. 3. The series of ligands of Scheler, Schoff"a and Jung may reveal a reversal of the ligand-field parameter. Before concluding that this is so Williams should establish: (a) That detailed calculations show that the conclusions of the theory of regular octahedral complexes can be taken over for non-regular complexes. (This is plausible but by no means obvious or easy to demonstrate.) (b) That the NOg" group is present as a nitro group (not a nitrito group) in both high- and low-spin forms of ferric haem complexes. (c) That the addition of ligands to haems (and changes of pH, etc.) do not affect the protein-metal interactions. 4. Williams' identification of electrostatic (as contrasted with covalent) theories with first order perturbation theories, and the latter with ligand-field theory, is incorrect. In conclusion, I should like to say that I usually find myself less in disagreement with Williams' view than with those which he attributes to others. In presenting material in reviev.'s or introductory articles it is normal to omit the many qualifications which appear in the detailed literature, and to indicate that this has been done. That, for example, is why I say that the theory of spin-pairing as presented is "much- oversimplified" and why I placed pyridine and ammonia, for the purposes of an elementary treatment, together in the ligand series as amines. (The ligand-fields of the compounds are very similar and the order probably does change from one com- pound to another, both for electronic and stereo-chemical reasons.) A useful purpose Electronic Structure and Electron Transport Properties of Metal Ions 1 5 may be sened by drawing attention from time to time to the well-recognized approxi- mations of a theory; an even more useful purpose is served by doing something about them (see, for example, Hush and Pryce, J. chein. Phys. (1958), whose work could well be extended to cover the transition from high-spin to low-spin states). Williams: It is not suggested that A, the field strength in any one state, varies greatly from every state to every other state. The variation of A will depend upon the character of the different excited and ground states. The spectrochemical series was observed to be a series roughly independent of whether one is dealing with low or high spin complexes or with metals from different transition metal series. Will Orgel state whether he believes this series to be also the series of the heats of interaction of ligands with a given cation, independent of cation ? The case of the octahedral complexes could be taken as an example. Orgel has made this assumption himself in his discussion here (3a) and elsewhere. .o- o- Under 3 (b); if the NOg- group changes from ^N^' to -<-0— N^ then this is but an indication of a change in ligand character with cation or spin state and/or valence state which I wish to demonstrate. This will be described shortly, not only for this case, but for the SCN~ complexes also. Under 3 (c), the hydroxide ion pro- duces spectroscopically the same effect in complexes of iron porphyrins where there are no proteins. (4) needs amplification before I can discuss it. I agree with Orgel's conclusion. My criticisms stand if the over-simplifications of theory lead to inconsistency with experi- ment. I say that they do. Spin-states of Haem Compounds Falk: I agree with Orgel that the relatively easy transition of many haemoproteins between the low- and high-spin states is very interesting. This phenomenon has fascinated me for some years, and Falk & Nyholm {Current Trends in Heterocyclic Chemistry (1958), p. 130) have discussed it briefly. But I think it is pertinent to remark that this phenomenon is established only for compounds of the haemoglobin, catalase and peroxidase types, and not for haemoproteins which are electron-transport agents in the classical cytochrome c fashion. I am not aware of any evidence, from magnetic susceptibility m.easurements, of high-spin low-spin changes in cytochromes of c, b or a types. If I may be allowed to guess, I would suggest that of these cyto- chromes, the properties of cytochromes a point to them as the most likely of the three types to shov/ this phenomenon. Orgel mentioned at one point the old observation of three unpaired electrons in ferric phthalocyanine chloride. In this context I think it is interesting to draw attention to the unpublished investigations of Nyholm and myself, mentioned in Falk and PeiTin (this volume, p. 56) on ferriprotoporphyrin chloride ("haemin chloride"). We found no conductivity in nitrobenzene solutions, indicating that the compound is not an electrolyte, and in view of the 5 unpaired spins, reported in the literature and confirmed by us, have suggested that it must be a square pyramidal complex with AsApHd'^ hybridization. Orgel: The observations of Falk and Nyholm are very relevant here. I v/onder whether the ferric protoporphyrin chloride has the same structure both in the solid and in nitrobenzene solution. Perhaps it would have the 3-spin ground state in nitrobenzene corresponding to a pyramidal structure, but have five unpaired electrons and an octahedral structure (with shared chloride ions) in the solid. George: I think there is some doubt whether the paramagnetic resonance absorption measurements at pH 7 and 8-5 carried out by Gibson, Ingram and Schonland {Disc. Faraday Sac, 26, 72 (1958)) prove that ferrihaemoglobin hydroxide is a mixture of high- and low-spin forms. First, the pK of the ionization is about 8 at room temperature so that at pH 7 only about 10% of the ferrihaemoglobin would be present as the hydroxide. Secondly, Keilin and Hartree {Nature Land., 164, 254 (1949)) have shown that on cooling the conjugate acid is favoured in the dissociation equilibrium, as 16 Discussion would be expected for an endothermic ionization process. Hence, if the measurements were carried out at low temperatures (liquid air or liquid hydrogen) as is normally the case, without suitable control experiments it is not certain how much hydroxide was present. Thirdly, unless ferrihaemoglobin is carefully freed from ammonium salts there is a marked tendency for the ammonia complex to be formed in alkaline solution : so, if the experiments were made with either single crystals or a microcrystalline paste from an ammonium sulphate mother liquor, in the absence of suitable controls a contribution to the observed signal from the ammonia complex cannot be disregarded. In the case of measurements at low temperatures this would be more serious since there is good reason to believe that the formation of the complex would be exothermic. The experiments reported in our present paper are not subject to these uncertainties, and they provide equally direct evidence for the existence of a thermal mixture. Electron Transport Lemberg: The interesting theory of oxidative phosphorylation involving a quinone- hydroquinone system (Todd and others) appears less likely in the phosphorylation step connected with the oxidation of cytochrome c, although Glahn and Nielsen {Nature, Lond., 183, 1578 (1959)) have recently suggested that this step involves binding of the phosphate to the formyl group of haem a. Orgel's explanation still leaves us with the difficulty that we do not know conjugated systems which might take up phosphate groups, except perhaps the histidine imidazoles bound to haem iron. The electron transport through a respiratory chain of several cytochromes makes it appear sterically unlikely that electron transport through the imidazoles as postulated by Theorell can be a sufficient explanation; similar difficulties exist with regard to haem-haem inter- actions in haemoglobin. I therefore ask whether the physicochemists consider it impossible that electron transfer may occur through "aliphatic" portions of a protein, or possibly through a chain of water molecules bound in the protein. Winfield: The example of electron transfer given by Orgel (terephthalic acid complex) is one which can readily be demonstrated experimentally. But may there not be some kinds of conduction which are important in biology and yet not readily demonstrable? If one were able to remove an electron from one end of a paraffin chain simultaneously with addition of an electron at the other end, would not the resulting electron move- ment along the chain take place with negligible activation energy? It seems possible that conducting chains of this kind could be interposed between conjugated con- ducting groups of the type described by Orgel. In other words, conduction of electrons between the prosthetic groups of adjacent enzymes {in vivo) may not require a path which is conjugated throughout its length. In the passage of electrons through a series of cytochromes in the living cell, I think that the individual enzymes are joined by metal bridges or hydrogen bonds. If the metal atom were calcium, one might expect that the electrons would pass across the bridge either not at all or with no pause. But with a metal atom such as iron or copper acting as bridge, I think that the electron would reside for a finite time in the metal ion and that there would be an activation energy required to move an electron across such a bridge. The pause might well be of biological significance. A small activation energy for the transfer of electrons between an interconnected series of cytochromes would restrict "hunting" in a system which would otherwise be uncon- trollably sensitive to transient fluctuations in the environment. In addition the metal bridge could provide for by-passing part of the electron flow along paths which branch from the main respiratory chain. Lockwood: In proposing models for electron transport two essentially diff"erent ones have been given. There is one in which electrons can be put in at one end of the chain and taken out at the other and that can be repeated an indefinite number of times. The comparison to a piece of copper wire is convenient. I take it that the model given in reaction (1) of Orgel's paper is an example of this type of conductor. In other models that have been proposed an electron can be put in at one end of the chain and taken out at the other but this produces an alteration of the configuration and the process cannot be repeated till the electron transport has been reversed. An Electronic Structure and Electron Transport Properties of Metal Ions 17 example of this is electron transport transverse to the polypeptide chains of protein where the transport occurs through the CO and NH group via a hydrogen bond. Orgel: This could be compared to a condenser. LocKWOOD : Yes. The picture of the cytochrome change in particular preparations where the cytochromes are situated spatially side by side is a legitimate one and the distinction between the two types of models becomes important. The transport of the electrons through the members of the cytochrome chain is a process which is repeated an indefinite number of times and the model, to be satisfactory, should belong to the copper wire type. It appears to me that the condenser type of model would be useless to explain the transport of electrons through the cytochrome chain. Orgel: We cannot be sure that electron transport will never take place through aliphatic side chains. However, I myself would be very surprised to find transport through more than at most three or four carbon atoms. We are currently investigating this problem by magnetic resonance methods. Transport through the a helix or similar protein structure via a long series of hydrogen-bonded C=0 and NH groups is more problematical; again I suspect that this process is not favoured except in systems which have been excited optically. Chance: Although we have been discussing in some detail the mechanisms by which electrons might be transferred through the peptide chain of the protein, an experi- mental test of this possibility suggests that an insufficient conduction rate would occur at least in the case of cytochrome c. Experiments carried out by my collaborator Patrick Taylor on dried cytochrome c in an atmosphere of nitrogen, show that less than 1 ,000th of the conductivity w ould be obtained when compared with the rate at which electrons are transferred in the cytochrome chain. While this experiment may not be conclusive it is certainly indicative of the difficulty of applying this approach. Our early experiments on the reaction of cytochrome c and the peroxidase intermediate have been reviewed and considerably extended by John Beetlestone. He finds that an active centre of the size of 5 A would be adequate to explain the observed kinetic data. This size is larger than that of the iron atom but would fit nicely with the idea that a histidine group is involved. Thus to within the accuracy that is possible with this determination, some group on the outside of cytochrome c may be responsible for the interaction. George: I would like to add a few comments to those of Chance on the subject of kinetic data for haemoprotein reactions. Even though some of the velocity constants are quite low, i.e. 10^ to 10^ M~^ sec^^ in comparison with high values of 10® to 10* M~^ sec~^, these low values are often found to originate in large (unfavourable) activation energies E, so that when the temperature independent factor A in the Arrhenius equation, k = A e-^l^^'^, is evaluated it is found to have remarkably high values of the order lO^^ to lO^'. Now in terms of the simple collision theory for bimolecular reactions A is equated to PZ, where Z is the collision frequency, 10^\ and P is the steric factor. Considering "target areas" for haemoprotein reactions one would expect P to be a fraction, yet it is apparent that P can in fact be several powers of ten. It would seem that other features are extremely important in these reactions of which we know very little at present. For example, the haem plate is hydrophobic in nature and undoubtedly alters the structure of the liquid water in its vicinity. In addition, around tlie haem plate, there is a constellation of ionic charges on the protein, which may be very important when a reaction between two haemoproteins occurs. Chance: I agree with George that temperature-independent factors in haemoprotein interactions are high and variable and thus the accuracy with which one can determine tlie size of the active centre is definitely limited. However, the results are useful indicators nevertheless. In this connexion, increasing knowledge of cytochrome c structure is of importance and the apparent inaccessibility of the haematin, due to the surrounding structures, provides independent support for the idea that the active centre of cytochrome c in the peroxidase reaction may have to exceed the size of the iron atom. 18 Discussion Margoliash: Orgel's idea of the importance of the native configuration of the protein of haemoproteins in determining the closeness of attachment of the haem-iron hgands and hence the nature of the complex formed, fits well with the results of our study of the denaturation of cytochrome c. With this haemoprotein it appears that denatur- ation probably does not change the haem iron bound groups but rather has a quanti- tative effect on the haem iron-ligand bonds resulting, as denaturation proceeds, in the gradual disappearance of the specific properties of cytochrome c and its trans- formation into a normal chemical haemochrome (Margoliash, Frohwirt & Wiener, Biochem. J., 71, 559, 1959). Mechanism of Oxidative Phosphorylation George : Lemberg has raised the question of the kind of mechanism by which the oxida- tion-reduction of cytochrome c can be coupled with phosphorylation, since, for structural reasons, it is difficult to see how an electron mediator can be involved like terephthalic acid or its ester in the oxidation of Cr" by Co"^ Arising from our studies of the opening of the crevice in ferricytoclirome c, which happens when the azide and cyanide complexes are formed, Glauser and I have suggested that a conformational change in the protein may be involved as a conse- quence of a switchover to a different bonding group during the oxidation-reduction cycle. For example, if the most stable crevice structures for the ferric and ferrous forms, at the pH at which oxidation-reduction occurs, differ in having a primary amino group and a histidine group respectively coordinated to the iron as in A and C. I I I red I II (A) Prot— Fe"'— NH2 imid > Prot— Fe"— NHg imid (B) I 1 I oxid I I I (D) Prot— Fe"i— imid NHj < " ' Prot— Fe"— imid NHg (C) then upon reduction of ferricytochrome c (A) a metastable, "energy-rich" form of ferrocytochrome c (B) would be produced, reverting to the stable form (C) with a release of energy. Likewise on oxidation of ferrocytochrome c (C) an "energy-rich" form of ferricytochrome c (D) would be produced, reverting to the stable form (A) with a release of energy. The switchover of the crevice group in the reactions (B) -* (C) and (D) -> (A) would entail a conformational change in protein structure which could conceivably be linked in some way to a phosphorylation step (George, P., & Glauser, S. C. Abstracts Third Meeting Biophysical Soc, Pittsburgh, April 1959, D4). Chance : I should like to ask Orgel for more information on equation (iii) of his paper. At first I thought that you wished to distinguish between electron transfer reactions and the coupling to phosphorylation. However toward the end of your paper you show them to be intimately associated, unless I have misunderstood you. Further, is the iron atom to which ADP is linked a haematin or a non-haematin iron? Would you be willing to indicate to me arguments in favour of one or the other alternative? Orgel : I should like to make clear that in this paper I have tried to describe a very general scheme for preserving the energy of oxidation-reduction reactions. I had no particular chemical system in mind. If the Fe+++ is part of a haem compound then the ADP or other acceptor could not be attached to the metal atom but would have to be held in position by attachment to the protein; if the Fe+++ atom is not in a porphyrin ring, then the ADP could be attached to the metal directly. I have no view on the relative likelihood, if any, of the possible alternatives. The main idea is that if an electron is extracted from a metal ion which is weakly associated with a ligand then the metal in its new valency may decompose the ligand in such a way as to preserve the redox energy. One illustration is given. THE ROLE OF THE METAL IN PORPHYRIN COMPLEXES By F. P. DwYER John Curt in School of Medical Research, Australian National University INTRODUCTION The more obvious implications of the co-ordination of organic molecules to metal ions : the effective charge reduction of the metal ion, and the polariza- tion of the organic moiety, have tended to become obscured by the wide interest in ligand-metal bond theories. In as much as these of their nature incline to emphasize the separate entities of metal and ligand, attention has been directed away from the properties of the complex as a whole. Sugges- tions that many reactions can occur at the periphery of the metal-porphyrin molecule rather than exclusively at vacant or labile sites on the metal (Williams, 1956a; Chance, 1951; King and Winfield, 1959), deserve more serious consideration. The purpose of this paper is to direct attention to properties which are those of the whole complex unit rather than the ligand and metal components. The metal-porphyrins are derived from a planar di-acid molecule which differs from the usual planar quadridentates such as 1 : 2-bis(a-pyridyl- methyleneaminoethane) (Fig. 1), by implication of the metal in a closed-ring system which probably contributes considerably to the stability of the complex structure, since the organic molecule cannot be detached point by point and hence unwrapped from the metal. Recent exchange work with the sexadentate molecule 1 : 2-propanediaminetctraacetic acid has shown that the six points of attachment to a metal can be broken progressively ^^ ^ in this way (Dwyer and Sargeson, 1960). Co-ordination proceeds with the extrusion of two protons and the metal complexes have zero or a small overall positive charge. As a result, since it is obviously easier to detach electrons from complexes with zero or a small positive charge, oxidation is facilitated. This is shown by the redox potential shift on passing from the simple hydrated ions to the iron and manganese complexes. Silver(I) acetate and protoporphyrin react, with the extrusion of a single proton, to yield a silver(I) complex, which spontaneously oxidizes with the extrusion of a second proton. The rather rare formal Ag(II) complex is favoured by the low charge and the planar arrangement of the bonds (Falk and Nyholm, 1958). 19 20 F. P. DWYER The two co-ordination positions at right angles to the plane of the quad- ridentate ring can be occupied by a variety of ligands: water, halide or cyanide ion, organic bases, the histidine anion, carbon monoxide. There is a good deal of still somewhat empirical evidence, e.g. labihty, that these out of plane bonds are rather long, and the geometry is therefore tetragonal. The bonds may be so long that the complex is essentially planar. Long bonds in the polar, (1:6), positions of many copper, nickel and palladium complexes are well known (Nyholm, 1953; Nyholm et al, 1956). In cytochrome c the interaction of the imidazole groups is sufficient to promote the maximum electron pairing in both oxidation states and the geometry must be octahedral. The higher oxidation states favour co-ordination of anions because of the greater polarizing power of the metal and also the greater electronegativity. In the oxidized forms of the metal-porphyrins there is thus a stronger tendency to co-ordinate OH' and CI' to available sites or by displacement of another ligand. The ionic structure ascribed usually to haemin chloride is unlikely. The 'Vra«5'-effect" may be of considerable significance in the 6-co-ordinate metal-porphyrins. The effect, which has been extensively studied in planar complexes (Chatt et al, 1955), refers to the labilizing effect of groups, e.g. CI, CN, CO, on other groups or ligands attached in the opposite {trans) position. In octahedral complexes, though the chemistry is more compli- cated, the "/ra/75-eflFect" has been fruitful in elucidating substitution reaction mechanisms (Quagliano and Schubert, 1952; Basolo and Pearson, 1958). Strongly trans influencing groups : CO, CN, or the thiol anion, should modify the strength of attachment or even the properties of other ligands in the polar position. The imposition of a fixed spatial arrangement on groups attached in the 1 : 6 co-ordination positions is an important function of the metal atom, especially when it is realized that donor atoms of the protein itself are often linked in this way. Part of the functional role of the cobalt atom in vitamin Bi2 is the rigid and unique conformation imposed on the large organic moiety. Another part is probably the lability of the sixth co-ordination position normally occupied by cyanide ion, water or hydroxyl, but which can be used to attach a donor atom from protein. CHARGE DISTRIBUTION IN COMPLEXES The fundamental principles involved in the formation of metal complexes, first enunciated by Pauling (1938) have been elaborated by numerous authors (Martell and Calvin, 1952; Basolo and Pearson, 1958). Co-ordination of a Hgand to a metal ion decreases the charge on the ion and makes the donor atom more positive. Since donor atoms are amongst the most electronegative of the elements, part or most of the positive charge spreads over the ligand molecule. In effect, this means that the ligand molecule is polarized, with the withdrawal of peripheral electrons, or electrons from electron donating The Role of the Metal in Porphyrin Complexes 21 groups. It is well known that the stability of metal complexes is usually related directly to the strength of the ligand as a base, and this is merely another way of expressing this concept. Electron withdrawing substituents in the ligand molecule promote polarization in the wrong sense : compete with the metal atom for electrons. These ideas have been expressed succinctly in the "principle of essential neutrality" (Pauling, 1948). The pronounced curariform activity of complex cations containing phenanthroline, and bipyridine [M phengj^+j [M bipy3]++ in which the characteristic biological response must be due to distributed charge, supports the principle (Dwyer et al, 1957). The extrusion of protons during the formation of porphyrin metal com- plexes reduces the charge by two units but, even in the reduced form, the zero charge does not necessarily imply electrical neutrality of either the H Oij — C- H.C- C C — CH, m-i .0 0, ,0 NHj ~-o cr ^0 1^1 II T '^'"'^*^'^^« HjC-C^^i— CH, ° H Fig. 2 Fig. 3 metal or the ligand. Apart from the electrical capacities of the substituent groups, the transition metals are moderately electronegative. This property may well be enhanced by the spin-paired electronic situation existing in strongly interacting complexes. Recently, it has been shown that the methylene groups in the neutral complexes Z?/.s(glycine)copper (Fig. 2) and rr/j'(glycine)cobalt are sufficiently activated in this environment to enable Knoevenagel type condensations to be performed with acetaldehyde (Sato, Okawa and Akabori, 1957; Ikutani, Okuda, Sato and Akabori, 1959). A mixture of threonine and allothreonine was obtained from the cobalt complex in the presence of sodium carbonate. Djordjevic, Lewis and Nyholm (1959), found that nitrite ion and nitrogen dioxide attacked the neutral complexes 6/Xacetylacetone)nickel (Fig. 3) and /)/5(acetylacetone)copper, with the formation of complex organic nitrogen compounds, as yet unidentified. It is probable that the sites of attack are the activated resonating — CH — groups, which may carry a small positive charge. In common with phthalocyanine, phenanthroline and bipyridine, metals are bound more firmly in the porphyrins than might be anticipated from the base strength of these ligands. The donor power of the ligand, concerned primarily with the primary co-ordination or a bond, is responsible for the dissipation of charge from the metal atom. It is believed that much of the bonding strength of these molecules derives from at least two tt bonds in which the d electronic orbitals of the metal overlap the vacant/? orbitals of the donor atoms. These bonds tend to make the metal more positive. In the ferrocyanide ion 22 F. P. DwYER [Fe(CN)6]*" the excess negative charge conferred by six negatively charged donors is supposed to be nearly off-set by three tt bonds from iron to carbon (Pauling, 1938). The charge interaction picture of donor atom and metal is thus quite complex. The molecules 4:7-dihydroxy-l :10-phenanthroline (Fig. 4a) and 4:4- dicarboxy-2 : 2'-bipyridine (Fig. 4b) exist normally in the zwitterion forms. In neutral solution very little reaction occurs with iron(II) salts, but in alkaline solution very strong co-ordination occurs not only because the nitrogen atoms are now free, but because at least two protons have been Fig. 4 detached and neutralized. The tris-chelatQ iron complexes have therefore zero charge or are anions depending on the pH. At the biological pH the carboxylic acid side-chains in many porphyrins, e.g. protoporphyrin, haemato- porphyrin, etc., make some contribution to the stability of the complex. Quite independently of other factors, such groups when sufficiently acidic promote oxidation by reducing the overall positive charge. OXIDATION STATE OF THE METAL IN COMPLEXES In the absence of obvious oxidizing or reducing conditions, the co-ordina- tion of a ligand to a metal ion is taken to involve no change in the oxidation state. The number of unpaired electrons, but not necessarily their location, can be obtained from magnetic moment measurements. The validity of Hund's Rule, which usually needs to be invoked to translate magnetic data into the oxidation state, has been frequently questioned when the magnetic evidence is at variance with the chemical properties. 5/5'(dimethylglyoxime)- copper has the moment characteristic of one unpaired electron, but from the absence of metal-metal interaction in the crystalline state, it has been deduced that the unpaired electron is mostly located on the ligands (Rundle, 1954). It would not be unreasonable to think of the Cu atom as in the -f3 diamag- netic state and the ligands as reduced. The observation that the 1 -electron oxidation of copper phthalocyanine removes an electron from the ligand and not the metal suggests a similar disposition of the unpaired electron (Cahill and Taube, 1951). The elaborate system of conjugated double-bonds in iron protoporphrin makes it feasible that oxidation could yield stable seini-quinone structures without affecting the oxidation state of the iron. Recently, Gibson and Ingram (1956), using the electron spin resonance method, showed that the The Role of the Metal in Porphyrin Complexes 23 oxidation of methaemoglobin removed an electron from peripheral carbon atoms and not from the metal, which was taken as formally remaining in the +3 state. Some sim.ple metal complexes containing nitric oxide provide examples of where chemical, magnetic and electronic structure considerations fail to establish the oxidation state of the metal. The ions [Fe(CN)5N0]-~ and [RuCljNO]^" are both obtained by boiling salts of the tervalent metal ions [Fe(CN)6]^~ and [RuCIgHaO]^^ with concentrated nitric acid. They are diamagnetic and hence the oxidation state is assumed to be +2. It is proposed that nitric oxide co-ordinates as N0+ following the loss of its odd electron to the metal which is thereby reduced. A tt bond also is formed between a d orbital of the metal and the vacant p orbital of the nitrogen. Exactly the same ultimate electronic structure would result had the nitric oxide formed the usual a bond, and the tt bond had come about by pairing the odd d electron of the metal with the odd p electron of the nitrogen, or had the metal lost an electron to nitrogen, which then utihzed four electrons to form a double bond. It is questionable whether the donation of four electrons by N0~ is more objectionable electronically than of two electrons by NO+. The metals should then be considered in the +4 state, which is certainly more consistent with the method of preparation and the resistance of the iron complex to oxidation. Metal complexes are generally regarded simply as Lewis acid-base entities but it is possibly more fruitful, especially in their oxidation-reduction reactions, to regard some of them as integral internal redox systems in which the metal alone is not the sole electron source or sink. Certain band spectra of the strongly interacting transitional metal complexes with the porphyrins, phenanthroline and bipyridine have been assigned to the transfer of negative charge from the metal to the ligands, or in highly oxidized states of the complex, in the opposite sense. In the similar activated states in which reac- tion occurs we are, in effect, dealing with an oxidized or reduced ligand. Rapid racemization of both species has been found to occur when aqueous solutions of d[Os bipy3]"'"+ and /[Os bipy3]+++ are mixed. This must proceed through a peripheral electron, located most likely on a carbon atom in the 4-position to the nitrogen atom, leaking across to the oxidized form. Because of the large organic molecules and the octahedral geometry, the metal atoms themselves are inaccessible for direct electron transfer, even through a water bridge (Dwyer and Gyarfas, 1952): J[Os bipy3]++ -^ ^[Os bipy3]+++ -f e /[Os bipy3]+++ + Q-> /[Os bipy3]++ If we think of the oxidized complex as having an electron deficiency, i.e. a positive charge, localized on a similar carbon atom, which is then solvated, a water bridge is provided for electron transport (Fig. 5). 24 F. P. DwYER This mechanism, which is similar to that proposed by Wilhams (1956b) for the haemin catalysed oxidation of cysteine by molecular oxygen, is applicable also to the remarkable reaction first discovered by Blau (1889). The oxidized forms of the tris complexes of Fe, Ru and Os with phenanthroline, bipyridine and terpyridine undergo spontaneous reduction when the pH of the aqueous solutions is raised. Hydroxyl radical has been detected (Uri, 1952). This may reoxidize the complexes if the pH is lowered soon enough, or decompose to ozone (Blau, 1898) or hydrogen peroxide (Brandt, Dwyer II2. II2* Fig. 5 and Gyarfas, 1954). The polarization of the carbon atom (Fig. 6) may be sufficient to lead to dissociation of a proton, as happens with simple hydrated cations, and the electron then is captured from the attached OH group. Reaction mechanisms of this kind could well be applied to oxidation- reduction reactions in the cytochrome systems, but may have applications to many synthetic processes involving activated — CH — and — CH2 groups, as discussed previously. Undue attention seems to have been paid in metal porphyrins to the formal oxidation state of the metal in relation to possible oxidation states as H^ /^OH H^ deduced from simple compounds or salts. As a result, jf 1 ► r "jj + OH there has been much hesitation in invoking otherwise II 1^ feasible reaction mechanisms involving, for instance, formal Fe(IV), Fe(V) or Mg(I). In such strongly interacting systems both the metal and the ligand are in unique electronic states because of their combination. The relevant fact is the number of electrons that can be added to or detached from the complex unit. The source or fate of the electrons is immaterial. Often this informa- tion can be obtained by electrolytic methods or simple chemical reagents. REDOX POTENTIALS OF RUTHENIUM COMPLEXES Simple model metal-complex systems offer much promise in elucidating problems in metal-porphyrin chemistry (Williams, 1956a, b). This is especially so when considering redox potentials. Much useful information on the effect of substituents has been obtained from the iron /m(phenanthroline) and bipyridine complexes. In general, the results parallel those obtained with various porphyrins (Martell and Calvin, 1952). Electron attracting substituents (NO2, Br, CI) render oxidation of the complexes more difficult (potentials are more positive than in the unsubstituted complexes), whilst The Role of the Metal in Porphyrin Complexes 25 electron donating groups cause the opposite effect (Brandt, Dwyer and Gyarfas, 1954). Because of the non-equivalence of the electronic states in the oxidized and reduced forms of many metal-porphyrins true equilibrium is not attained on an electrode. This raises the question of the applicability of redox potential results obtained from truly reversible model systems to the metal-porphyrins. The effect of substitution in the ligand molecule itself is but one aspect of the problem. There is little precise information available from models on the effect of the overall charge upon the redox potential, or of the effects that might be anticipated from various ligands when added to a basic planar complex. Recently, we (Dwyer and Goodwin, 1959) have prepared a large number of mono- and Z?/5(bipyridine) and phenanthroline ruthenium(II) and (III) complexes which serve as better models for the metal-porphyrin systems than the //•/5(chelate) iron complexes. The Z?/5(chelate) complexes which evidently are the more appropriate, however, always have the labile two groups in the cis{\ : 2) position instead of the desirable trans{\ : 6) position. Ruthenium is the heavier analogue of iron in Periodic Group 8, and unlike iron, the mono- and bisichdaie) complexes do not disproportionate. The complexes are spin-paired in both oxidation states and reversible redox potentials can be obtained. By suitable replacement of the labile positions, anions, cations and neutral complexes can be prepared. The effects of overall charge and of the nature of the ligand are shown in Table 1 . Oxidation is greatly facilitated by lowering the positive charge. The replace- ment of bipyridine, (pK^ = 4-33) by two molecules of the stronger base pyridine (pA:„ = 5-20), also facilitates oxidation, but slightly. Large poten- tial changes are associated with the co-ordination of ammonia, ethylene- diamine and water. These seem much too large with the basic ligands to be ascribed wholly to their greater strength as bases. Water, of course, is a much weaker base than pyridine. The implied enhanced stability of the oxidized state can be related in considerable part to the capacity of the ligands to dissipate positive charge to their hydrogen atoms. The latter are then more strongly solvated or can form hydrogen bonds to the solvent water. At the acid concentrations used, dissociation of a proton from the aquo groups is unlikely, though this would stabilize the oxidized form most effectively by reducing the overall charge. There are still insufficient data to make much of a comparison between the ruthenium systems and the metal-porphyrins containing various co-ordinated addenda. The potentials of the latter systems certainly cover a much narrower range, possibly because of the smaller overall charge. The replacement of the water molecules (or water and hydroxyl) in protoporphyrin, for instance, by pyridine only changes the potential from —0-14 V to -f 0-107 V. A much more positive potential might have been anticipated. Similarly, the potential of the dicyano-protoporphrin couple (—0-183 V) would be expected to be 26 F. P. DWYER more negative. In some of these couples, however, the electronic states are different. It is questionable whether comparisons between the spin-free protoporphyrin and the spin-paired Z)/5(pyridine) and dicyano complexes can be made on the same basis as the electronically equivalent ruthenium complexes. Table 1, Redox potentials of ruthenium complexes in SULPHURIC ACID (1 N) Couple £0 [Ru bipy3]++ -[Ru bipy3]+++ 1-257V [Ru b!py2py2]++ ^[Ru bipyjpyal''""''"'" 1-25 [Ru bipy py4]++ — [Ru bipy py4]"'"'"*' 1-246 [Ru bipy py3Cl]+ — [Ru bipy py3Cl]++ 0-894 [Ru bipy CI4]— — [Ru bipy CI4]- 0-35 [Rubipypy3-H201++ -[Rubipypy3-H20]+++ 1-041 [Ru bipy py2-(H20)2]++-[Ru bipy ^y^-{H^O)Y++ 0-782 [Ru bipy.,(NH3)2]++ -[Ru bipy2(NH3)2]+++ 0-875 [Ru bipy2-en]++ — [Ru bipy2-en]+++ 0-74 (py = pyridine, en = ethylenediamine). SUMMARY The metal-porphyrins have been discussed as typical strongly interacting metal complexes in respect to such properties as the vacant co-ordination positions about the metal, the peripheral charge distribution and the oxidation state. A series of mono- and Z)zXbipyridine) ruthenium complexes has been proposed as model systems. The importance of the whole complex unit is emphasized in opposition to the concept of metal with attached ligand. REFERENCES Basolo, F. & Pearson, R. G. (1958). Mechanisms of Inorganic Reactions, 34-90; 177-210 John Wiley, New York. Blau, F. (1889). Mh. Chem. 10, 367. Brandt, W. W., Dwyer, F. P. & Gyarfas, E. C. (1954). Chem. Rev. 54, 959. Cahill, a. E. & Taube, H. (1951). /. Amer. chem. Soc. 12,, 2847. Chance, B. (1951). The Enzymes, 2, 428, Sumner and Myrback, Academic Press, New York. Chatt, J., Duncanson, L. H. & Venanzi, L. M. (1955). J. chem. Soc, 4456. Dwyer, F. P. & Goodwin, H. (1959). Unpublished work. Dwyer, F. P., Gyarfas, E. C, Shulman, A. & Wright, R. D. (1957). Nature, Lond. 179, 452. Dwyer, F. P. & Gyarfas, E. C. (1950). Nature, Lond. 166, 1181. Dwyer, F, P. & Sargeson, A. M. (1960). Nature, Lond. 186, 966. Djordjevic, C, Lewis, J. & Nyholm, R. S. (1959). Chem. and Ind. 4, 122. Falk, J. E. & Nyholm, R. S. (1958). Current Trends in Heterocyclic Chemistry, 130-139, Butterworths, London. Gibson, J. F. & Ingram, D. J. (1956). Nature, Lond. 178, 871. Itukani, Y., Okuda, T., Sato, M. & Akabori, S. (1959). Bull. chem. Soc. Japan 32, 203. The Role of the Metal in Porphyrin Complexes Tl King, N. K. & Winfield, M. E. (1959). Aiist. J. Chem. 12, 47. Martell, a. E. & Calvin, M. (1952). Chemistry of tlw Metal Chelate Compounds, 101 237; 373-375, Prentice-Hall, New York. Nyholm, R. S. (1953). Chem. Rev. 53, 263. Nyholm, R. S., Harris, C. M. & Stephenson, N. C. (1956). Rec. trav. chim. 75, 687. Pauling, L. (1938). The Nature of the Chemical Bond, Cornell University Press, Ithaca, N.Y. Pauling, L. (1948). J. chem. Soc. 1461. QuAGLL\NO, J. V. & Schubert, L. (1952). Chem. Rev. 50, 201. RuNDLE, R. E. (1954). Conference on Coordination Compounds, Indiana University, Bloomington, 25. J. Amer. chem. Soc. 76, 3101. Sato, M., Okawa, K. & Akabori, S. (1957). Bull. chem. Soc. Japan 30, 937, Uri, N. (1952). Chem. Rev. 50, 375. Williams, R. J. P. (1956a). Nature, Lond. Ill, 304. Williams, R. J. P. (1956b). Chem. Rev. 56, 299. DISCUSSION Higher Oxidation States Lemberg : While it is certainly correct that we have strongly interacting systems in which both the metal and the ligand are in unique electronic states because of their combina- tion, I feel that Dwyer has somewhat overstated his case. In such instances as the RO2H or HjOa-complexes of peroxidase, catalase and ferrimyoglobin, one may well be in doubt about the exact valency state of the iron, but in most other haemoprotein compounds there is little doubt about the valency of the iron. George: Gibson and Ingram {Nature, Lond. 178, 871, 1956) demonstrated the presence of a free radical in the oxidation of ferrimyoglobin by H2O2 by paramagnetic resonance absorption measurements, and identified this with the higher oxidation state, Mb'^. However, in more recent experiments, Gibson, Ingram and NichoUs {Nature, Lond. 181, 1398, 1958) have shown the radical to be present in much lower concentration than the oxidation state IV of the prosthetic group, invalidating the previous conclusion. It is not unexpected that radicals can be detected in such a system because there is already ample chemical evidence for the production of a radical in the formation reaction, i.e. Mb"' -h H2O2 -> Mb^ -f- radical and furthermore, in the reduction of Mb"", which occurs spontaneously and more rapidly the higher the concentration, radical species must again be formed, since Mb'^ is a one-equivalent oxidation product of Mb"'. (George and Irvine, Biochem. /. 52, 511, 1952.) Some years ago it was shown that a simple radical structure of the type that was proposed by Gibson and Ingram would not account for the hydrogen ion participation in Mb'^ reactions, whereas the "ferryl ion" structure or an isomer of this structure is in accord with the experimental data (George and Irvine, Sympos. on Coordination Compounds, Danish chem. Soc, p. 135, 1954; Biochem. J. 60, 596, 1955). It should be emphasized that although paramagnetic resonance absorption provides an excellent technique for the detection of free radicals, other evidence must also be considered in discussing possible structures for intermediates in oxidation-reduction reactions. The same is true of the mechanism of oxidation-reduction reactions, since radical species could be formed in side reactions, and not necessarily be involved in the principal reaction path. George: Another example where higher oxidation states are formed, somewhat similar to that of the Cu", Co", Zn" and Al"' phthalocyanines, is that of a-/3-}'-5-tetraphenyl- porphin (TPP). Whereas with the metal-free phthalocyanine the one-equivalent H.E. — vol. 1 — D 28 Discussion higher oxidation state is very unstable, TPP yields both a one-equivalent and a two- equivalent higher oxidation state that are appreciably more stable, i.e., one one TPP ^ ^ TPP^ ^ ^ TPP" equiv. equiv. In phosphoric acid solution TPP is bright green, TPP' and TPP" are dull violet and orange-brown respectively. These oxidations are completely reversible Uke the one- equivalent oxidation of copper phthalocyanine, the addition of ascorbic acid, hydro- quinone or ferrous salts regenerating the TPP. The structure of TPP" probably corresponds to the removal of the two pyrrole hydrogen atoms with the introduction of a new double bond into the ring system, as in the oxidation of reduced flavin. The copper salt of TPP yields a one-equivalent higher oxidation state Uke the phthalocy- anine derivative, but on addition of more oxidant a whole series of highly coloured products are formed from which the original compound can no longer be recovered by the addition of reducing agents (George and Goldstein, Abstracts \19th Meeting Amer. chem. Soc, Dallas, K 16, p. 13, 1956; George, Ingram and Bennett, J. Amer. chem. Soc. 79, 1870, 1957). Effects of Metal on Reactivity at Periphery Barrett: Concerning Dwyer's remarks on the effect of the introduction of metals into porphyrins, and the consequent events occurring at the periphery of the molecule, I would like to make this comment. Fischer and Bock (Hoppe-Seyl. Z. 255, 1, 1931) exposed protoporphyrin in pyridine solution to light and obtained a substance with a chlorin-like spectrum. The substance is not a true chlorin, or dihydroporphyrin, but carries two or possibly three oxygen atoms. The addition of these oxygen atoms results in the formation of a hydroxy group and a carbonyl group (Barrett, Nature^ Loud. 183, 1185, 1959). A vinyl group is necessary for the formation of dioxy- protoporphyrin. Pertinent to Dwyer's remarks is the observation that photo-oxidation of the tetrapyrrole does not occur when complexed with a metal, e.g. Cu++, Fe+++, or if irradiated in 1-10% hydrochloric acid. Could Dwyer comment on these effects: the suppression of photo-oxidation by (1) the formation of a metal complex and (2) the formation of the di-hydrochloride ? DwYER : One can anticipate a common effect as far as the peripheral charge is concerned by either protonation or the formation of a metal complex. However, I feel that the altered charge distribution is not per se the reason for the inhibition of photo- oxidation, but rather the effect of the proton or the metal is on the fluorescence of the protoporphyrin, and hence its ability to form active oxygen (or hydroxyl) which is presumably the attacking agent. THE PHYSICO-CHEMICAL BEHAVIOUR OF PORPHYRINS SOLUBILIZED IN AQUEOUS DETERGENT SOLUTIONS By B. Dempsey,* M. B. LowEf and J. N. PHiLLiPsf Department of Chemistry, Royal Military College, Duntroon, and Division of Plant Industry, C.S.I.R.O., Canberra The Tetrapyrroles, and in particular the haem pigments, occur in a biologi- cal environment which is essentially aqueous. It is therefore desirable to determine their physico-chemical properties in an aqueous medium. Unfor- tunately, the information available, particularly as regards their ionization and co-ordination behaviour, is meagre (Phillips, 1960). The major experi- mental obstacle to obtaining such data has been the very low solubility of these compounds in water. Some measurements have been carried out in non-aqueous and mixed solvent media (Conant, Chow and Dietz, 1934; Aronoff and Weast, 1941; Aronoff, 1958; Barnes and Dorough, 1950; Caughey and Corwin, 1955; Corwin and Melville, 1955). However, such systems are unsuitable for electrochemical studies because of unknown ionic activity effects. This situation led us to explore the use of aqueous detergent solutions as solvent media (Pliillips, 1958). Studies have been carried out using fully esterified porphyrin derivatives to avoid electrostatic effects arising from the ionized carboxylic acid groups on the periphery of the nucleus. A wide variety of porphyrin esters has been shown to disperse molecularly in a number of detergent solutions, presumably by solubilization within the lipid micelle. This phenomenon is analogous to a phase distribution equilibrium in which one of the phases is of molecular dimensions. Macroscopically, such a system would behave as a single phase and equilibration between the phases would be expected to be extremely rapid. The distribution of the porphyrin molecules between the aqueous and micellar phases can be repre- sented by a simple equilibrium of the type : (PHg)^ ^ (PH2)(j the solubilization constant K^ being defined by [«(PHjJM * Royal Military College. t C.S.I.R.O. 29 30 B. Dempsey, M. B. Lowe and J. N. Phillips where the subscripts d and vi' refer to the micellar and water phases respec- tively, and Tij. is the number of molecules of species x\ Na is the number of detergent molecules in the micellar phase ; and A'^ is the number of water molecules in the system. When an ion, e.g. a hydrogen ion H+ or a metal ion M++, is introduced into such a system it will presumably prefer the aqueous environment exclu- sively to that of the lipid micelle. Accordingly, any reaction involving such an ion and the porphyrin molecule must take place within the aqueous phase, the detergent micelles acting as a readily available reservoir for the porphyrin molecules. Typical reactions which may be studied in this way are : PH2 + 2H+ PH, + M++ PH3+ + H+ MP -f- 2H+ PH. ionization chelation (1) (2) The products of such reactions may or may not be solubilized depending on their nature. For example, one might expect the nonionic metal complex // y—NH =N HC H NH N-< to) (b) (c) Fig. 1 (a) Dimethyl protoporphyrin ester (DMPP) (b) Dimethyl mesoporphyrin ester (DMMP) (c) Tetramethyl coproporphyria III ester (TMCP) M = — CH3 V = — CH=CH2 E = — CH2— CH3 P = — CH2— CH2— COOCH3 (MP) but not the ionic porphyrin species (PH3+ and PH4++) to be readily solubilized. The purpose of this paper is to indicate the type of data that may be obtained using the solubilization technique and to suggest how such data may be interpreted. The following discussion is primarily concerned with the ionization (as in equation (1)), co-ordination (as in equation (2)) and spectro- scopic behaviour of porphyrins. In particular, the discussion will refer to the behaviour of the fully esterified derivatives of mesoporphyrin IX (DMMP), protoporphyrin IX (DMPP) and coproporphyrin III (TMCP) (see Fig. 1). The detergent solutions used Physico-Chemical Behaviour of Porphyrins in Aqueous Detergent Solutions 31 were either 2-5% (w/v) sodium dodecyl sulphate (SDS) or 0-25% (w/v) cetyltrimethyl ammonium bromide (CTAB). Ionization Behaviour The ionization of porphyrins solubilized in aqueous detergent solutions can readily be studied spectroscopically. Two general types of behaviour occur, the one with cationic and non-ionic and the other with anionic detergent solutions. In the former case two species only, the neutral porphyrin (PHo) and the dication (PH4++) are observed upon spectroscopic titration within the pH range 0-12. The variation in optical density {E) with pH at a given wave- length is such as to indicate that the reaction involves two protons : PH2 + 2H+ ^ PH4++ (3) the overall dissociation constant (A'obs) of the conjugate acid PH4++ being given by _[PHJ[H+P ^^^'^ ~ [PH,++] • ^^^ This does not necessarily imply the simultaneous addition of two protons in the kinetic sense. A more likely explanation is that under these conditions the monocation behaves as a stronger base than the free porphyrin. The equihbria postulated to account for this behaviour are illustrated below. ^, K,= [(H+)J[(PH,)J (PH,).;.'(PH,), "" [(PH3+)J +°" ^^ ^^ ^ [(H+),][(PH3+)J (PH3+), A4 = -——_——, (PH4++), ^ _ [(H+)J^[(PH,), + (PH,),] [(PH,++)J ^obs = In terms of the scheme outUned the observed constant (ATobs) would corres- pond to K^K^{\ + K,). The pXobs (= - logio/sTobs) values for DMMP and TMCP in 0-25% CTAB were found to be 2-08 ± 0-05 and 2-24 ± 0-05 respectively. Unfor- tunately the corresponding value for DMPP was too low (< -f-O-S) to be determined with any certainty. It is clear, however, that the observed values will depend on the nature and concentration of the detergent, and that com- parisons of pAT values for different porphyrins in the same detergent solution will reflect differences both in their aqueous basicities and in their solubiliza- tion constants. In anionic detergent solutions three species, corresponding to PH,, PH3+ and PH4++, may be observed upon spectroscopic titration, each ionization 32 B. Dempsey, M. B. Lowe and J. N. Phillips step corresponding to a one-proton addition. It has not been possible fully to interpret this behaviour theoretically, although it is certain that both the ionic porphyrin species PH3+ and PH4++ are being stabilized by some type of solubilization process. This leads to the observed pK values being a function of the intrinsic aqueous basicity constant and of the ratio of the solubilization constants of the species concerned in the equilibrium. Such a ratio would tend to eliminate specific porphyrin solubilization effects and hence one would expect the relative p^ values for a series of porphyrins in anionic detergent solutions to parallel their values in water. In the case of DMMP, TMCP and DMPP, the ipK^ values observed in 2-5% SDS were 5-94, 5-58 and 4-88, and the piQ values 2-06, 1-80 and 1-84 respectively. The pATg values appear to reflect the relative electrophilic character of the side chains, and this effect has been confirmed with a number of other porphyrin derivatives. The pK^ values appear less sensitive to the nature of the side chain. It is of interest to note that the pK^ of 1-84 for DMPP in 2-5 % SDS is equal to the value for DMPP in water (pK^ = 1-8) as estimated indepen- dently from solubility measurements (Dempsey and Phillips, unpublished). This suggests that the solubilization constants for the species PH3+ and PH4++ not only parallel each other but are in fact very similar in magnitude. Co-ordination Behaviour The interaction between porphyrin molecules and divalent metal ions can be represented by an equilibrium of the type shown in equation (2), There are little quantitative kinetic or thermodynamic data available about such reactions, and it was therefore thought desirable to explore them using the solubilization technique. It has been found that in cationic detergent solutions the reactions between porphyrins and metal ions are markedly dependent on temperature and also on the nature of the metal ion. At 20°C in 0-25 % CTAB no reaction has been observed with Co++, Ni++, Cu++, Zn^, Cd++, Mg++, Mn++, Pb++, or Fe++, over a period of weeks. On the other hand, at 100°C very rapid reac- tions occurred with Cu++ and Zn++ under suitable pH conditions, though not with any of the other metal ions studied. Accordingly the reactions involving Zn++ and Cu++ were investigated in greater detail. These reactions were shown to conform to equation (2), the apparent equilibrium constant (K'g) being given by: [MP][H+]^ ' [M++][PH2]' ^ ^ Such constants were evaluated by determining spectroscopically the ratio MP/PH2 at equilibrium, for a range of hydrogen and metal ion concentrations. Typical formation curves are shown in Fig. 2 for the zinc-mesoporphyrin Physico-Chemical Behaviour of Porphyrins in Aqueous Detergent Sohttions 33 reaction at 60°C. The time required to attain equilibrium at this temperature in the presence of IQ-i m zinc sulphate and 5 X 10"' M DMMP in 0-25% CTAB was approximately 48 hr. The reversibility of the equilibrium was demonstrated by studying both the forward and backward reactions as expressed by equation (2). The KJ value determined for the zinc mesoporphyrin equilibrium at 80°C, extrapolated to zero ionic strength is 6-0 X 10~^; the corresponding value at IrO n 05 i / / i y / p / / 20 30 40 PH Fig. 2. Formation curves for the zinc mesoporphyrin complex at 60°C Forward reaction equilibrium points O Zn++ + PHg ^ ZnP + 2H+ Backward reaction equilibrium points • ZnP + 2H+ ^ Zn++ + PHj ZnP n = , where TpHo is the total concentration of mesoporphyrin present. TPH, 60°C is 2-5 X 10"^. The relationship between the observed equilibrium constants {K^) and their value in water {K^ is given by : f (l +^,^) \ (6) where K^ is the solubilization constant of species x. If both species PH2 and MP were equally solubilized then the observed equilibrium constant would correspond to the water value. It seems likely that the solubilization constants will be of a similar order of magnitude, and in any event K^ values are likely to parallel K^ values either when comparing the one porphyrin with different metals or different porphyrins with the same metal. 34 B. Dempsey, M. B. Lowe and J. N. Phillips Preliminary results suggest : (i) that Zn++ reacts faster and forms a more stable complex with DMMP and TMCP than DMPP, as might be expected from the greater electro- philic character of the unsaturated vinyl side-chain compared with its saturated analogues; (ii) that Cu++ reacts faster and forms more stable complexes than Zn++ with DMMP, in accord with the normal relative chelating ability of the two ions (see Bjerrum, Schwarzenbach and Sillen, 1956-57); and (iii) that Co++ and Ni++ react infinitely more slowly than Cu++ or Zn++, although there is evidence (Caughey and Corwin, 1955) to indicate that in general Co++ and Ni++ form the more stable porphyrin complexes. It is suggested that this reluctance on the part of Co++ and Ni++ may be associated with their tendency to form hexaco- ordinate compounds as compared with the tetraco-ordinating tendency of Cu++ and Zn++. The overall kinetics conform to a simple bimolecular reaction involving metal ions and the neutral porphyrin species (PHg). This, and the fact that Zn++ reacts more readily with DMMP and TMCP than DMPP suggests that the reaction mechanism is of the displacement rather than the dissociation type (Basolo and Pearson, 1958). For purposes of comparison it is convenient to express KJ values in terms of the more conventional stability or formation constants {Kf) defined by : _ [MP] ^^"[M++][P=]' ^^ In the absence of acidic ^K data for mesoporphyrin the value for aetio- porphyrin II at room temperature {"pK^ + pA'g '^ 32 (McEwen, 1936)) has been used. This leads to extrapolated logio Kf values for zinc mesoporphyrin at 20°C of '->-' +29. The high stability of porphyrin metal complexes is illustrated by comparing this value with the corresponding figures in water for the zinc complexes of, for example, 8-hydroxyquinoline-5-sulphonic acid ('^16-0), ethylene diamine ('-^ll-O), and glycine ('-^9-5) (Bjerrum et al, 1956-57). Spectroscopic Behaviour Aqueous detergent solutions form useful solvent systems for studying the spectroscopic properties of porphyrin molecules, their salts and metal com- plexes. Anionic detergents are of particular interest in that they permit a study of the spectral behaviour of the monocationic species (PH3+), a species which has proved virtually impossible to obtain in normal solvent media. Much theoretical argument (Piatt, 1956; Kuhn, 1959) concerning electron distribution in the porphyrin nucleus has been based on the absorption spectra of the symmetrical free porphyrin (PHg) and its dication (PH4++). It seems Physico-Chemical Behaviour of Porphyrins in Aqueous Detergent Solutions 35 - /~\ - 1 1 1 1 \ 4, \PH. \ \ \ PH*„ l^\ \ / ^/^ X\PH.\^ / .<^ \\ \ /"y^^. / ^\\ ^^^^^ ^ y^ 1 1 0300 200 0100 /'\ / \ V_/ \ 4000 4200 A, A 5500 6000 o' A, A phV 0400 0300 Fig. 3. Soret and visible absorption spectra for the various protoporphyrin species. Table 1. Spectroscopic properties of some PROTOPORPHYRrN DERIVATIVES IN 2*5% SDS Species Absorption maxima ±5A Fluorescence excitation maxima ±20 A Fluorescence emission maxima ± lOA PHj 4080 Soret 5050 \ 5405 ,,. .^, 5780 ^'^'^'^ 6330 j 4150 5080 5420 5820 6340 6340 PH3+ 3985 Soret 5350] 5680 Visible 6095 ' 4080 5370 5660 6100 6125 PH4++ 4120 Soret 5570] ^,. ... 6020/ ^'^'^^^ 4130 5580 6020 6060 ZnP 4120 Soret 5425] ,,. ... 57901 ^'^''''s 4150 5440 5780 5890 36 B. Dempsey, M. B. Lowe and J. N. Phillips likely that similar information on the asymmetric intermediate species (PH3+) would further facilitate the understanding of this problem. Figure 3 compares the absorption spectra of the species, PHg, PH3+ and PH4++ for DMPP solubilized in 2-5% SDS. Such curves are typical for porphyrins having actio type spectra. It will be observed that in the Soret region the absorption maximum for the monocation is displaced towards the violet relative to both the other species. In the visible region the four-banded spectrum associated with the neutral porphyrin (IV > III > II > I) changes upon ionization to a three-banded type (II > III > I) and finally to the typical two-banded dication spectrum (II > I). Measurements have been made also of the fluorescence excitation and emission spectra of the various protoporphyrin species and of the zinc protoporphyrin complex in 2-5% SDS. The fluorescence excitation and emission maxima are shown in Table 1 along with the absorption maxima. It will be observed that in each case the single fluorescent emission maximum lies at wavelengths 10 to 100 A longer than the a absorption band. In general the fluorescent excitation maxima correspond to the absorption bands. SUMMARY This paper is concerned with physico-chemical studies of porphyrin esters solubilized in aqueous detergent solutions. In particular, quantitative data have been reported for: (i) the relative basicity of proto-, meso- and copro-porphyrins at 20°C ; (ii) the chelation of zinc ions by mesoporphyrin at 60° and 80°C ; and (iii) the absorption and fluorescence spectra of the monocationic species of protoporphyrin at 20°C. Preliminary results have also been reported for the zinc-protoporphyrin and copper-mesoporphyrin reactions at 80°C. The technique could readily be adapted to other physico-chemical studies, e.g. the further co-ordination of metalloporphyrins with other ligands and oxidation-reduction equilibria in metalloporphyrin systems. Such studies, aimed at providing basic information on the physico- chemical behaviour of these pigment prosthetic groups, seem an essential prerequisite to a detailed understanding of the role of such molecules in biological processes. A cknowledgem en t The authors are indebted to Miss I. Verners for her skilled experimental assistance and to Dr. J. E. Falk for providing the purified porphyrin esters. REFERENCES Aronoff, S. (1958). J.phys. Chem. 62, 428. Aronoff, S. & Weast, C. A. (1941). J. org. Chem. 6, 550. Barnes, J. W. & Dorough, G. D. (1950). J. Amer. chem. Soc. 11, 4045. Physico-Chemical Behaviour of Porphyrins in Aqueous Detergent Solutions 37 Basolo, F. & Pearson, R. G. (1958). Mechanisms of Inorganic Reactions, p. 91, John Wiley, New York. Bjerrum, J., ScHWARZENBACH, G. & SiLLEN, L. (1956-7). Stability Constants, Pts I & II, The Chemical Society, London. Caughey, W. S. & CORWIN, A. H. (1955). J. Amer. chem. Soc. 77, 1509. CoNANT, J. B., Chow, B. F. & Dietz, E. M. (1934). /. Amer. chem. Soc. 56, 2185. CoRWiN, A. H. & Melville, M. H. (1955). /. Amer. chem. Soc. 77, 2755. KUHN, H. (1959). Helv. chim. Acta 42, 363. McEwEN, W. K. (1936). J. Amer. chem. Soc. 58, 1124. Phillips, J. N. (1958). Current Trends in Heterocyclic Chemistry, p. 30 (Ed. by A. Albert, G. M. Badger and C. W. Shoppe), Butterworths. London. Phillips, J. N. (1960). Rev. pure appl. Chem. 10, 35. Platt, J. R. (1956). Radiation Biology, Vol. Ill, p. 71 (Ed. by A. Hollaender). McGraw- Hill, New York. DISCUSSION Cations of Porphyrins and Their Spectra Orgel: I should like to ask Phillips why it is that in the detergent system it is possible to measure the addition of a single proton to a metal-free porphyrin, while in the past those measurements which have been done have suggested that two protons are added simultaneously. Phillips: We believe the reason that the monocation is so readily obtained in anionic detergent solution to be due to its stabilization at the negatively charged micelle-water interface. It is of interest to note that no monocationic species can be detected in non-ionic or cationic detergent solutions. Lemberg: It is reassuring that on the whole there seems to be a satisfactory agreement between the conclusion as to basicity of porphyrins derived from the Willstiitter eniM and R I/IV and R III/IV of porphyrins I III Porphyrin £mM i?I/IV fniM R III/IV Deutero 4-33 0-27 8-59 0-54 Aetio 5-18 0-38 9-50 0-70 Copro Meso 5-15 5-41 0-35 0-38 9-97 9-82 0-68 0-69 Proto 5-58 0-38 11-58 0-79 Diacetyldeutero Diformyldeutero 3-52 3-48 0-29 0-29 6-7 8-00 0-55 0-67 Monoacetyldeutero Rhodo 1-56 2-0 0-17 0-17 10-5 15-0 1-14 1-29 Crypto a 2-35 0-21 14-7 1-32 Chlorocruoro 2-25 0-21 15-1 1-40 Acrylic acid 3-01 0-27 16-13 1-46 Phaeo 05 1-88 0-20 16-06 1-71 Formylpyrro Formyldeutero Vinylrhodo Diacetylpyrro Acetylrhodo a 1-86 1-82 1-2 1-42 2-21 1-30 019 0-20 0-115 0-245 0-27 0-15 16-74 15-7 19-8 13-62 21-0 21-0 1-77 1-73 1-89 2-36 2-40 2-40 38 Discussion number, and their basicity now more correctly established by Phillips' interesting method. It is also interesting to note that evidence for a monocation had previously been obtained by Neuberger and Scott for deuteroporphyrin disulphonic ester, somewhat resembling anionic detergents, although Walter had not been able to confirm this. With regard to the spectra of porphyrin dications, it is of interest that on closer observation four, not two, absorption bands can be observed in the visible part of the spectrum. This leads me to doubt the correlations assumed by Piatt for the neutral and acid spectra of porphyrins. In this connexion I should like to point out that the two bands I and III of neutral porphyrin spectra are not at all influenced in a similar manner by the substitution of an electron-attracting group on the porphin nucleus. Thus all formyl and ketonyl-substituted porphyrins have their band III greatly increased, but their band I greatly diminished as the table on p. 37 shows. While it is true that band I is the most variable and band III the second in variability, there is, in fact, little difference between the variability of ^niM of bands II, III and IV in a variety of different porphyrins. Phillips: Walter's failure {J. Amer. chem. Soc. 75, 3860, 1953) to detect the monocationic species observed by Neuberger and Scott {Proc. Roy. Soc. A. 213, 307, 1952) was due to an unfortunate choice of experimental conditions. This aspect has been discussed by Scott (/. Amer. chem. Soc. 77, 325, 1955). The Reactions between Metal Ions and Porphyrins By J. H. Wang and E. B. Fleischer (Yale) Wang : Phillips and his co-workers suggested that the combination of porphyrin esters with Zn++ to form Zn++-porphyrin derivatives takes place through a displacement rather than a dissociation type of mechanism. I would like to report some work which not only confirms his suggestion but also gives a more detailed understanding of this displacement mechanism. We found that the rates of the successive steps in the combination of a metal ion and a porphyrin derivative can be markedly affected by varying the solvent composi- tion, presumably due to the change in solvation of the metal ion. In acetone solutions some metal ions, Cu++, Bi+++, Hg++, Cd++, etc., react readily with the dimethyl ester of protoporphyrin even at room temperature to form the corresponding metallo- porphyrin, whereas other metal ions, Fe++, Fe+++, Cr+++, Pt+++"'", Sn++, Zn++, etc., form a new type of complex with absorption spectra markedly different from that of the corresponding metallo-porphyrins; only upon heating do their spectra change to those of the metallo-protoporphyrins. The spectra of some protoporphyrin derivatives in chloroform solution are shown in Fig. 1 . The spectra are respectively (A) protoporphyrin dimethyl ester, (B) haemin dimethyl ester, (C) the new type of complex formed between ferric chloride and proto- porphyrin dimethyl ester, and (D) the dihydrochloride of protoporphyrin dimethyl ester. If alcohol or pyridine is added to the chloroform solution of this new complex formed between ferric chloride and protoporphyrin dimethyl ester, the spectrum changes immediately to that of protoporphyrin dimethyl ester, (A) in Fig. 1 . This shows that the binding between Fe"''++ and protoporphyrin in the new complex is quite weak, since the protoporphyrin can readily be displaced by other ligands. We suggest that this new complex has a sitting-atop type of structure as illustrated in Fig. 2. It is also of interest to note that the spectrum of the sitting-atop complex of Fe+++ bears striking resemblance to that of protoporphyrin dication, PH4++, as shown by diagram (D) in Fig. 1. This observation suggests that the observed absorp- tion in (C) is probably due to electronic transitions in the protoporphyrin rather than the metal part of the complex. Similarly, if acetone solutions of the sitting-atop complexes of Fe++, Fe+++, Cr+++, Pt++++, Sn++, Zn++ respectively are diluted with water the spectrum immediately changes Physico-chemical Behaviour of Porphyrins in Aqueous Detergent Solutions 39 o to that of the dimethyl ester of protopoq^hyrin itself, {A) in Fig. 1, as water molecules rapidly displace the organic ligand from the complex. We have isolated the ferric sitting-atop complex in pure crystalline form. The structure suggested in Fig. 2 was confirmed by the infra-red spectrum of this complex dissolved in deuterated chloroform, 99% CDCI3. In protoporphyrin dimethyl ester the N-H groups are responsible for three distinct infra-red absorption peaks at 3-05 fx (stretching), 6-15 /< (deformation), and 9-06 // (rocking) respectively. In the infra-red spectrum of haemin dimethyl ester all of the above three peaks disappear as expected. 20 15 10 05 20 15 10 OS ■t ° c -o 20 8 15 " aA yvTV^^ B ■^ 1 1 1 D 1 r 1 C,H COoCH, COXH, Fig. 2. Proposed sitting-atop type of structure for the reaction intermediate. 650 600 550 500 Wavelength, m/i Fig. 1 450 We found that in the infra-red spectrum of the ferric sitting-atop complex the 3-05 and 6-15 [i peaks shifted to 3-11 and 6-64 /t respectively, and that the 9-06// peak is no longer easily detectable. This observation confirms our proposed sitting-atop structure, since it shows that the corresponding N — H bonds still exist in the said complex. The observed frequency shift of the 3-05 and 6-15 /< peaks is presumably due to the polarizing influence of the ferric ion. Phillips: Wang's suggestion that the sitting-atop complex is involved in metal porphyrin formation is both novel and interesting. However, I find the suggestion that the dication structure involves two nitrogens each with two hydrogens and two nitrogens without hydrogens rather disturbing for a number of reasons, viz. (i) the spectra of the dication and dianion are identical, which would seem to support the symmetrical dication structure (i.e. 1 hydrogen per nitrogen atom); 40 Discussion (ii) energetically one would expect the symmetrical structure to be favoured because of its greater conjugation; and (iii) Wang's postulatedfstructure would imply that porphyrin basicity should be comparable to pyrrole itself whereas in fact it is many p^ units stronger. In our experience the reaction between anhydrous ferric chloride and dimethyl- protoporphyrin ester led either to the mono- or di-cationic species, probably due to the difficulty of removing the last trace of water from the system. It would be inter- esting to have the infra-red spectrum of the dihydrochloride for comparison with the other species. SOME PHYSICAL PROPERTIES AND CHEMICAL REACTIONS OF IRON COMPLEXES By R. J. P. Williams Inorganic Chemistry Laboratory and Wadham College, Oxford This article is intentionally speculative in that it will use a comparison between known properties and reactions of some simple iron complexes with chelating and conjugated ligands on the one hand, and of the wide range of haem-containing compounds of biology on the other, in order to make comments about the structure and mechanism of reactions of the latter molecules. First we restrict our attention to simple iron complexes. There are three classes of these. In class I all the properties and reactions are consistent with the iron (Fe++ or Fe+++) being in a high-spin state (ionic complexes). In class III the properties are consistent with the cation (Fe++ or Fe+++) being in a low-spin state (covalent complexes). In the intermediate group, class II, the properties and reactions are consistent with the cation being present in the complexes as an equilibrium mixture of high- and low- spin states. Table 1, which hsts the properties and reactions of the groups, is a summary of evidence presented in detail elsewhere (Williams, 1955, 1956, 1958, 1959). The change in the character of the iron complexes (from one group to another) is to be associated with the change in the strength of the effective ligand field increasing from class I through class II to class III. The porphyrin unit supphes a ligand field which, together with groups above and below the porphyrin plane, places its iron complexes in either class II or class III, although the extreme weak-field members in class II can have properties very like complexes in class I. The evidence for this statement is collected and discussed elsewhere (Williams, 1955-59). Here we take this position as estabUshed. However we will illustrate it by reference to some physical properties of porphyrins. SPECTRA The absorption bands in the spectra of iron porphyrins have been discussed (Williams, 1956) except for the Soret band. Here we will examine the shifts in the position of this band ignoring the others. Some band positions are given in Table 2. In Table 2 wavelengths are in m^. The description 'ionic' and 'covalent' is only applicable to the first three rows of data. 41 42 R. J. P, Williams Table 1. The three classes of iron complexes Class I Class II Class III Physical Properties Magnetic moment (+ +) 4-90 4-9O-000 0-00 (+ + +) 5-90 5 •90-2-00 2-00 Visible spectra (++) weak emax < 10^ £max ^ 10^ strong fmax = 10* (+ + +) strong £max ~ 10* intermediate weak fmax < 10^ Stability (++) weak intermediate very high (+ + +) quite high high high Chemical Properties Metal and ligand exchange (++) rapid rapid-slow slow (+ + +) rapid rapid-slow slow Autoxidation (++) rapid intermediate or rapid slow Reaction with H2O2 rapid usually rapid slow Electron transfer reactions rapid rapid rapid Examples of ligands in complexes (++) H2O, NH3, aT>MGUU^O\, pMG)2(NH3)2, and (+ + +) oxalate. oxine. (DMG)2(pyridine)j,, Enta. ^ I CN-. (DMG is dimethylglyoxime.) In the table £max is the molar extinction coefficient. Table 2 Ferrous Ferric Ionic - > Covalent Ionic - > Covalent H2O CO CN- Pyridine H2O OH- CN- Peroxidase 438 425 420 404 418 425 Myoglobin 435 424 420 405 414 425 Haemoglobin 430 418 420 404 409 419 Cyt. ^3 448 432 430(?) 420(?) Cyt. a 444(?) 430 440 430 420 428 Cyt. 6 425 no effect 416 no effect Cyt. c 415 no effect 410 no effect Cyt. ^ 423 415 420 415 390 407 404 Data taken from Lemberg and Legge (1949) and Morton (1958). Some Physical Properties and Chemical Reactions of Iron Complexes 43 Tbeorell (1942), Williams (1955, 1956) and Scheler, Schoffa and Jung (1957) have drawn attention to the band-shifts as the magnetic moment of the ferric complexes changes. The Soret band moves to longer v^avelengths in the lower moment complexes. Here we point out that in ferrous complexes the band moves in the opposite direction. The lower the moment of a Fe++PX2 complex (P is porphyrin and Xg the further co-ordinating ligands) the shorter the wavelength of the absorption band. Now if we plot the difference in Soret band position Fe++-Fe+++ (AA) against the sum of the magnetic ^ ^ 30 X" ^ y ^^"h.o x y y 20 ^^0H~ y"^ y y - y y y y y ■^ m 10 /^ ®co y y y ® y _ y* • _ ^ , , , 20 70 12-0 Fig. 1 . The relationship between the sum of the magnetic moments of the ferrous and ferric haemoglobin complexes in the presence of the additional ligands shown and the difference in position of the Soret band in the two complexes. moments (/fFe++ -f ^Fe+++) we can obtain a qualitative guide to the character of iron porphyrin complexes from their spectra (Fig. 1). A large value of AA implies ionic complexes. On this basis we have the series of complexes of decreasing ionic character: peroxidase = myoglobin = haemoglobin > cytochrome a^ > cytochrome d > cytochrome a > cytochrome h > cyto- chrome c. This series was devised by entirely different reasoning in earlier papers (Williams, 1958, 1959). The extreme members are ionic and covalent respectively (from magnetic observations), whereas the intermediate members are apparently mixtures of two spin forms in equilibrium (from spectroscopic evidence). A possible explanation of the opposed band shifts in the ferrous and ferric series is that the two cations are differently affected in their character on going from the high-spin to the low-spin state. If we take the Soret transition to be either an « -> tt or a tt' -> tt transition in which the electrons are more concentrated on the nitrogen in the ground state than in the excited state, H.E. — VOL. I — E 44 R. J. P. Williams this transition is naturally more difficult in ionic ferric than in ionic ferrous complexes on account of the charge difference. The band is at shorter Amax in the ferric complex. In the low-spin states both ferrous and ferric ions become better cr-acceptors and on these grounds the Soret transitions should both be more difficult in the promoted state (low-spin) complexes and a shift of Amax to shorter wavelengths would be expected. Now the ferrous ion also becomes a stronger tt donor in the low-spin state which again makes the transition of the electrons concentrated on the nitrogen more difficult (Note 1). Thus in the case of Fe""^ both a and tt interactions of the cations in the ligand change so as to shift Amax (porphyrin) to shorter wavelengths when the complex becomes of low spin. On the other hand, the ferric ion in the promoted state is a tt electron acceptor with one hole in the de shell. This property makes the Soret transition of the porphyrin easier. We conclude that the increased stabilization of tt electrons in the excited state overrides the increased stabilization of the electrons in the ground state on change from the high- to the low-spin ferric state. Thus we can see that the difference in TT-bonding characteristics of the cations can explain the opposed shifts. A similar explanation has been advanced in discussing the phenanthroline series of ferric and ferrous complexes where spectral changes in opposed directions are observed in a series of ferrous and ferric complexes which we have considered as models for the study of the physical properties of iron porphyrins (Williams, 1955, 1956). A good illustration of the opposed shifts of the Soret band in the Fe++ and Fe+++ states is that observed in the study of the effect of pH upon the spectra of Rhodospirillum cytochrome (Morton, 1958). At pH 7 the band positions are: Fe++, 424 m/t, and Fe+++, 390 m/<, with a band at 640 m/< in the ferric spectrum indicating an ionic complex. At pH 11-8 the bands are: Fe++, 413 m/i, and Fe+++, 407 m/.<, with no band at longer wavelengths than 565 m/i in the Fe+++ spectrum indicating a covalent complex. Thus at pH 7-0 this cytochrome is very largely ionic but at pH 11-8 it is largely covalent and the value of Am/t has changed from 34 m// to 6 m/i. The suggested change of magnetic moment is in keeping with the observed fall in the ratio of the y to a peak intensities of the reduced cytochrome on increasing pH. The observations are also very suggestive with regard to the group of the protein which is responsible for the pH dependence. As both the ferric and ferrous complexes show changes it can not be the reaction HoO -> 0H~. Imidazole groups are completely ionized at a pH just greater than 7-0 and we are left with the strong impression that the change involved is — NH3+ -^ NHg (see pp. 46, 49, 50). It will be observed that from the above analysis we consider that cytochrome ^3 is largely ionic. This tentative conclusion is supported by the following evidence. (1) The intensity ratio of the a to the y peak is 1:11. The usual ratio in covalent complexes is 1 : 6 as in the pyridine complex of haemoglobin Some Physical Properties and Chemical Reactions of Iron Complexes 45 while in ionic complexes it is approximately 1:12, as in haemoglobin. (2) The reaction of cytochrome a^ with carbon monoxide is rapid and only ionic forms of ferrous complexes can undergo rapid substitution reactions of this kind (Table 1). All ferrous complexes which react rapidly with carbon monoxide also react rapidly with oxygen. (3) The ferric form of cytochrome ^3 has a weak absorption maximum at '^ 650 m/^, which only appears when at least some of the ferric couple is in the ionic state. A similar discussion of the same spectroscopic features would lead us to suppose that cytochrome a is largely covalent with but a small percentage of the ionic form. The correlations between Soret band position and the magnetic moments of the ferrous complexes permit comment on the interaction between the ferrous ion and the protein. For example, Gibson (1959) observed that on intense illumination of carboxyhaemoglobin (HbCO) a short-lived species Hb* was produced. This species has its Soret band at slightly longer wavelengths than Hb itself. The discussion presented here would lead us to conclude that the protein groups interacting with the ferrous ion must be more weakly bound in Hb* than in Hb. In keeping, the reaction of Hb* with CO is forty times as fast and of lower activation energy. Again, the Soret band of myoglobin (438 m/<) is at a longer wavelength than that of haemoglobin (431m//) which are to be compared with MbCO 424 mju, Mb02 416 m/^ and HbCO 418 m//, HbOa 414 m/< (Lemberg and Legge, 1949). The values suggest that the protein groups are less strongly bound to myoglobin than to haemoglobin. This is in keeping with the more rapid reactions of myoglobin. Before leaving this point it is clear that the spin state of a ferrous or ferric complex is very sensitive to environment. We must expect that extraction of a cytochrome will sometimes alter its properties either through a minor denaturation of the protein or even through a change of the medium di-electric constant. OXIDATION-REDUCTION POTENTIALS We have made several comments upon the redox potentials of ferrous/ferric couples (Williams, 1959; Tomkinson and Williams, 1958). The general impression of the variations in the FePX, potential with change of X is that either by (1) a continuous adjustment in the nature of X from water to increasingly improved donors such as ammonia or (2) a gradual reduction in the Fe-X distance for a group X which is a good donor, the redox potential can be made to go through a continuous series of values which show a maximum, see Fig. 2. The maximum is reached because the ferrous as opposed to the ferric ion undergoes electron rearrangement at lower effective electronegativities of the group X. At low electronegativities, increase in electronegativity favours ferrous over ferric, while at higher effective electro- negativities increase in donor properties of the ligand favours ferric over ferrous. The maximum will be most accentuated for ligands which are 46 R. J. P. Williams TT-electron acceptors (Note 2). In order to develop further the discussion of iron-porphyrin redox potentials we present some new data, Table 3, on the redox potentials of Fe(NiOx)2X2 couples in water where NiOx is cyclo- hexanedionedioxime. From the table we see that the potential of the complex with pyridine is higher than that with imidazole which is higher than that with ammonia. All these complexes except the hydrates in both ferrous and ferric forms are covalent (of low spin). The values fall with the increasing donor character of the groups as shown by the p^ values, pyridine 5-2, imidazole 7-1 Table 3. Redox potentials of some iron nioxime complexes Ligand X in Fe(NiOx)2A'2 HaOCpH —3 0) pyridine imidazole ammonia Redox potential (mV) Ease of autoxidation + 180 slow very rapid at pH > 40 + 130 very slow + 30 very slow -370 slow rapid at pH < 80 The redox potential in water alone is pH-sensitive, falling to about —250 mV at pH approximately 100. NiOx is cyclohexanedionedioxime. and ammonia 9-1. Now the difference between the redox potential of Fe(NiOX)2, (NH3)2 and Fe(NiOx)2 (imidazole)2 is +400 mV. This difference is close to that between cytochrome c and cytochrome h, +200 mV, but cytochrome Z) is a protoporphyrin complex whereas cytochrome c is a meso- porphyrin complex. The redox potential difference produced by the different porphyrins is '-^ +80 mV in their pyridine forms (Lemberg and Legge, 1949). This factor should be added to the observed difference between cytochromes h and c giving +250 — 300 mV difference between the h and c type cyto- chromes had their porphyrins been identical. We conclude that if cytochrome c is an imidazole (histidine) complex, cytochrome h is likely to be an amine- imidazole complex. We will show later that the reactions of Z>-type cyto- chromes are in accord with this hypothesis as are their spectra (WiUiams, 1958). We now turn to cytochromes d (using the nomenclature of Morton, 1958) which appear to us to be related to h cytochromes structurally in much the way haemoglobin is related to cytochrome c. For although the cytochromes h and d have similar redox potentials they have very different chemical properties. The cytochromes d are very readily autoxidized for example. As we have noted in Fig. 1 the value of AA (Soret) between the Fe++ and Fe+++ forms of cytochromes d suggests that they are largely ionic complexes. This is supported by the easy autoxidation, by the uptake of carbon monoxide in the reduced state, and by the position of the Soret band of the Fe+++ form Some Physical Properties and Chemical Reactions of Iron Complexes 47 (>^max !^ 400 m-fi). Again there is evidence of bands at about 640 m/t in the Fe+++ complexes, the band expected for ionic Fe+++ haems. On the basis of this analysis of the properties of the cytochromes we need to modify the picture which we have given previously for the relationship between redox potential and the basicity of the co-ordinating groups amongst these compounds. Figure 2 gives our present impression. For all groups of haem-proteins there is a range of redox potentials and a range of otiier Ligand basicity (effective) Fig. 2. The suggested relationship between redox potentials and ligand basicity for the cytochromes. The top three curves are for haem a, protohaem and mesohaem complexes in descending order of potential. All these complexes are assumed to be imidazole-coordinated. The lowest curve is for either protohaem or mesohaem complexes where the further binding of the haem from the protein is assumed to be through amino-nitrogen. By effective ligand basicity we imply that basicity which obtains under the conditions of steric hindrance realized in the protein and we suggest that steric hindrance increases from right to left in all the cases. physical properties, such as magnetic moments and absorption spectra, as well as a range of chemical properties such as affinity for oxygen and carbon monoxide and rate of autoxidation. There is no reason to think that these groups of haem proteins can be rigidly differentiated, in fact. The variation in combination between porphyrins and the type of group X so far suggested would then be (see Williams, 1958), as shown in table on page 48. Inspection of these combinations leads to two immediate comments. Far from every possible intercombination between different porphyrins and different groups X or between different combinations of the two X groups for any one porphyrin has been postulated, let alone demonstrated. Second, where it is stated in the table that only one group is involved, say under 48 R. J. P. Williams myoglobin, and the other group is water we imply that there is no other co-ordinating group very near to the iron. Between this extreme and the case where there are two groups equally and strongly co-ordinated, as in cyto- chrome c, every possible intermediate may arise through the inability of the protein to satisfy simultaneously the stereochemical requirements of the iron and those of hydrogen-bonding in its own structure. Haem protein Porphyrin substituent Group X* Cyt. fl3 Aldehyde (hydroxyU?)) Imidazole HgO Cyt. fl Aldehyde (hydroxyU?)) Two imidazoles (one weakly held) Myoglobin Vinyl Imidazole HjO Haemoglobin Vinyl Two imidazoles (one weakly held) Cytochrome b Vinyl Two amino Peroxidase Vinyl Carboxylate HjO or — NH2 Catalase Vinyl Two carboxylates Cytochrome d no unsaturation One amino HgO Cytochrome c no unsaturation Two imidazoles * See also Williams, p. 72 of this volume. The 'Imidazole' Hypothesis Pauling and Coryell (1936) considered that the two dissociation constants of haemoglobin in the pH range 5-8 could be accounted for by assigning one ^K to a dissociation of type (1), of an imidazole group which was at a con- siderable distance from the iron atom and a second to the reaction (2). Basicity Fig. 3 This hypothesis is readily tested by examining the ionization of Fe(DMG)2 (imidazole)2. We find that the imidazole ionizes as in reaction (1) but that reaction (2) does not occur up to a pH of 11-0 (Croft and WiUiams, unpub- lished). We add the following evidence against any such ionization. Some Physical Properties and Chemical Reactions of Iron Complexes 49 (1) The absorption spectrum of Fe(DMG)2 (imldazole)2 does not change with pH from 6-11 except in intensity (Croft and WilUams). (2) The complex Fe++(DMG)2 (imidazole)2 is extractable into /j'o-amyl alcohol (Croft and WilUams). (3) The complexes Cu++(histidine) and Cu++(histidine)2 show no ioniza- tion of the type (2) (Leberman and Rabin, 1959; James and Wilhams, unpublished). (4) Amongst biological molecules, Fe++-cytochrome c has no ionization in the expected range (Lemberg and Legge, 1949). In the presence of oxygen it is observed that the lower ^K is raised. We accept Pauling and Coryell's (1936) explanation, that this implies that the oxygen inserts itself between the imidazole and the Fe++ ion. It would appear that the second p^is not due to the imidazole groups at all. The ^K shift could be due to the ionization of an — NH3+ group in the protein, the basicity of which was altered by the change in protein stereochemistry on deoxygenation of haemoglobin. Again, we have no evidence to show that Fe+++(NiOX)2 (imidazole)2 undergoes any ionization up to pH 10-0. The titration of the complex with alkali, its insolubility in organic solvents, and its absorption spectrum all indicate that the ligand is imidazole and not the imidazole anion. CHEMICAL REACTIONS OF IRON COMPLEXES It has always been our intention to proceed from a detailed study of the physical properties of ferrous and ferric complexes to a study of their chemical reactions. We have now made a start with the latter phase of this work. Reactions of Molecular Oxygen Oxygen can either combine with ferrous complexes (oxygenation) or oxidize them (autoxidation). In biological systems both reactions occur. We have observed both reactions also in the chemistry of the complexes Fe++(DMG)2X2 and Fe++(NiOX)2X2, where DMG is dimethylglyoxime. Our studies show that in the model systems the oxygenated complexes Fe(DMG)2X02 are not stable if X is readily exchanged for water or if X is a group containing labile hydrogen. The reactions can be illustrated by examples. When X is imidazole or pyridine the oxygenated complex is stable with certain qualifications. The replacement of the ligands X in the imidazole and pyridine complexes is much slower than in other complexes. When X is water, hydrazine, ammonia, aniline or other substituted amines, or sterically hindered pyridines or imidazoles (e.g. histidine) the oxygenated complex is not as stable but undergoes autoxidation. The replacement of ligands in these complexes is more rapid. Autoxidation is different in different cases giving oxidation of the group X in some cases (e.g. N2H4) and not in 50 R. J. P. Williams others. In both cases the same ferric complex is always obtained, Fe+++(DMG)2(H20)OH. We suggest the mechanisms: Fe+ ^(DMG)2X2 Fe++(DMG)2(H20)02 -> Fe+++(DMG)2(H20)OH Fe++(DMG)2X02 Fe++(DMG)2X+02- -> Fe+++(DMG)2(HoO)OH m^ (Reaction (2) is a catalysed autoxidation of X.) No complex of a saturated base X is known which carries oxygen (cf. cytochromes b and d) ; however, all the complexes of unsaturated bases X can carry oxygen (cf. myoglobin, haemoglobin, cytochrome a). It is of great importance here to note that Fe(DMG)2 (imidazole)2 can even pick up molecular oxygen in the presence of borohydride, sodium formaldehyde sulphoxylate, or sodium dithionite. This reaction is common to cytochrome a (Sekuzu, Takemori, Yonetani and Okunuki, 1959) but not to haemoglobin. It implies that the iron complex undergoes slow dissociation of its ligands as free oxygen reacts rapidly with borohydride or sulphoxylate. Haemoglobin undergoes very rapid de-oxygena- tion under these circumstances. We can now make some comments about the binding in haemoglobin. We note first that covalent ferrous complexes (Class III, Table 1) undergo slow ligand exchange. Magnetic data show oxyhaemoglobin to be covalent, yet it takes part in fast reactions involving ligand replacement. The iron must be in an energy state which is only slightly more stable than its high-spin states. This is in agreement with spectroscopic evidence as well as with the value of its redox potential. Now if the oxygen is labile in HbXOg the group X, the histidine, must be labile also. What is it then that prevents the autoxidation of haemoglobin in accord with equation (1) above? The answer which we suggest to this question is that it is the high activation energy of the rearrangement of the protein which prevents a water molecule replacing X and thus prevents autoxidation. On the other hand, the cytochrome a oxygen complex dissociates slowly to a covalent haemocliromogen (judged by the spectra) whence there is little danger of dissociation of groups X leading to autoxidation. In cytochrome a we suggest iron is more strongly bound to imidazole than in haemoglobin. Elsewhere (WiUiams, 1958) we have reached this conclusion from a very different argument. Amongst cytochromes some of the cytochromes b appear rapidly autoxidiz- able. We believe that this observation is irrelevant to biological function. If, as we suppose, cytochromes b (d) are amine ( — NH2 -^ Fe) complexes, then bringing them out of a cell environment to a pH of about 7-0 in free solution may well dissociate the NHg -^ Fe link. We can show this easily with Fe(DMG)2(NH3)2 which is fairly stable to autoxidation at pH 100 but Some Physical Properties and Chemical Reactions of Iron Complexes 51 rapidly autoxidized at pH 7-0. The behaviour is also very similar to that of Rhodospirilhim haemoprotein (cytochrome d) and could well arise in both cases through the high acid dissociation content of the — NH3+ group. It seems to us that some lower organisms may not have suitable histidine-con- taining proteins to give rise to Fe-histidine cytochromes but must be content with Fe-amine cytochromes. If this is the case and our discussion is valid then these organisms are unlikely to be able to store or transport molecular oxygen. Their cytoclirome oxidases and electron-transporting cytochromes are amine complexes whereas those of higher organisms are both amine and histidine complexes. One such conjecture about these compounds leads immediately to another. The development of histidine cytochromes a and c in a cell gives the organism the advantage over cells containing only amine cytochromes b and d that the energy of the oxygen molecule can be more efficiently used. Some 300 mV more energy (the difference in redox potentials) can be stored chemically for each electron transported. ELECTRON TRANSPORT There are two reasons for thinking that electron transport occurs across the porphyrin of the cytochromes. If catalytic activity resided in the imidazole-Fe-imidazole bonds then it should be demonstrable in Fe(DMG)2 (imidazole)2. The model complexes do not have the electron transporting properties of cytochromes as far as we can discover. Again using the models we have shown that although there is strong charge transfer interaction be- tween Fe++ and (DMG) there is no evidence for it in Fe++-imidazole. On the other hand, there is good evidence in Fe++(DMG)2 (pyridine)2. In this complex there is a band (absent in other Fe++(DMG)2X2 complexes) at '^ 400 m^a. For differently substituted pyridines it moves in the following manner (Jillot and Williams, 1958): Substituent None 4-bromo 3-cyano 4-cyano Maximum Absorption 385 385 460 475 (m/<) If it is assumed that this band is due to a partial charge transfer of an electron from the ferrous atom to the pyridine, the band positions are explicable in terms of the electron-acceptor properties of the substituents. The band position is solvent-dependent, again suggesting a charge transfer band. The absence of such a band in the imidazole complexes would suggest that charge transfer and therefore electron-transport across the imidazole is not facile. Finally, if we are correct in saying that cytochromes b and d are amine complexes then as amines are not unsaturated systems and presumably could not carry out electron transport we must assume in these cytochromes that the electron moves through the porphyrin. If this is so then it is very likely that electrons are mobile in the porphyrin of FeP (histidine)2, but of 52 R. J. P. Williams course this by itself does not eliminate the possibility that electron transport occurs across both the porphyrin and the imidazole in cytochrome c. NOTES 1. The position of the Soret band moves in the order of increasing wavelength with ligand ammonia, pyridine, carbon monoxide and oxygen, cyanide, water. This order, saturated bases < unsaturated bases < water is similar though not exactly the same as that found for the a and |3 bands and the explanation which we offer is that given earlier (Williams, 1956). 2. We imply here that the ferrous ion is more stabilized than the ferric ion by 7r-electron acceptors. The evidence is given elsewhere (Tomkinson and Williams, 1958). Thus in Fig. 2 we expect that the maximum will be more sharply defined in a series of pyridine complexes of increasing pyridine basicity than in a series of ammines. Imidazoles will occupy an intermediate position while for a series of oxygen anion-donors there need be no maximum as the ferric ion may well go over into the strong-field complex the more readily. SUMMARY An account is given of the properties of iron-porphyrin complexes of biological interest which is largely developed from a consideration of the properties of simpler iron complexes. Spectroscopic criteria for distinguishing between high- and low-spin complexes are suggested. New features of the inter-relationship of different cytochromes are proposed, based upon their redox potentials and their chemical reactions. Some comments are made upon the reactions of haemoglobin and their pH dependences. A discussion of the model iron complexes which are autoxidizable as opposed to those which can carry oxygen leads to a discussion of autoxidation and oxygenation of haem complexes. A cknowledgement I would like to acknowledge the help of the late B. A. Jillot, and of J. M. F. Drake and D. Croft, who have done all the experimental work connected with this paper. REFERENCES Coryell, C. D. & Pauling, L. (1936). Proc. nat. Acad. Sci. Wash. 22, 159. Croft, D. & Williams, R. J. P. Unpublished observations. Gibson, J. F. (1959). Disc. Faraday Soc. 29. James, B. R. & Williams, R. J. P. Unpublished observations. Jillot, B. A. & Williams, R. J. P. (1958). J. chem. Soc, A61. Leberman, R. &. Rabin, B. R. (1959). Nature, Lond. 183, 746. Lemberg, R. & Legge, J. W. (1949). Hematin Compounds and Bile Pigments, Chap. 5 & 6, Interscience, New York. Morton, R. K. (1958). Rev. pure appl. Chem. 8, 161. Scheler, W., Schoffa, G. & Jung, F. (1957). Biochem. Z. 329, 232. Sekuzu, I., Takemori, S., Yonetani, T. & Okunuki, K. (1959). J. Biochem. Tokyo 46, 43. Theorell, H. (1942). Ark. Kemi. Min. Geol. 16A, No. 3. Tomkinson, J. & Williams. R. J. P. (1958). J. chem. Soc, 2010, Some Physical Properties and Chemical Reactions of Iron Complexes 53 Williams, R. J. P. (1955). Special Lectures in Biochemistry, University College, London. H. K. Lewis & Co., London. Williams, R. J. P. (1956). Chem. Rev. 56, 299. Williams, R. J. P. (1958). Disc. Faraday Soc. 26, 123. Williams, R. J. P. (1959). TIw Enzymes (Ed. by P. D. Boyer, H. Lardy & K. Myrbiick), vol. I, p. 391, Academic Press, New York. DISCUSSION Oxidation-reduction Potentials of Haem Coinpounds Perrin : I should like to ask Williams what evidence he has for believing that with increasing ligand basicity the redox potentials of ferrous/ferric couples pass through a maximum. The only experimental evidence that I have so far found suggests, on the contrary, that in a related series of ligands there is a roughly linear dependence of redox potential on the pA^ of the ligand: the potential decreases continuously as the ligand pAT increases. This is true, for example, of iron complexes with a number of 5-substituted- o-phenanthrolines (Brandt and GuUstrom, 7. Amer. diem. Soc. 74, 3532, 1952). Other systems where linearity is found include the 1 : 1 iron-amino-acid complexes (Perrin, J. chem. Soc. 290, 1959) and the iron complexes of 8-hydroxyquinolines and polyaza- 1-naphthols (Albert and Hampton, /. chem. Soc. 505, 1954; Albert, Biocliem. J. 54, 646, 1953). The reported potential of 0-7 V for the iron complex of 4-hydroxy-3- carbethoxy-o-phenanthroline (Hale and Mellon, /. Amer. chem. Soc. 72, 3217, 1950) appears at first sight to be anomalously low but I think it can be readily explained. In heterocyclic compounds a hydroxyl group in a gamma position relative to a nitrogen makes two tautomers possible — an enol form, where the H is on the oxygen, and an amide form where the H is on the nitrogen. Contrary to the way the formulae are generally written, the amide form is greatly favoured relative to the enol form (for example, in 4-hydroxyquinoline the ratio is 24,000 to 1 (Albert and Phillips, J. chem. Soc. 1294, 1956) and it would be expected to be even higher for a 1-hydroxy- o-phenanthroline). I suggest that this effect, together with the sparing solubility of the substance, leads to insufficient complex formation to prevent extensive hydrolysis of ferric ion and this is what causes the potential to be so low. It should be pointed out (1) that this system did not behave reversibly and (2) that the corresponding substance without the carbethoxy group was more soluble and gave a higher and reversible potential that was closer to the expected value (Hale and Mellon, loc. cit.). In the metalloporphyrins and related substances also there seems to be, in all cases where both are known, a continuing decrease in redox potential with increasing pA' of the ligand. Some examples are given in Tables 5 and 6 of the paper of Falk and Perrin (this volume, p. 69). It seems reasonable to suppose that the change from high-spin to low-spin in a complex does not result in any great alteration in the nature of the metal-ligand bonds. The main differences would lie in the extent to which the metal's Id orbitals are made more or less available to take part in bond formation. As discussed more fully elsewhere (Perrin, Rev. pure andappJ. Chem., 9, 257, 1959) I believe these and ligand field stabilization energy changes for any series of iron complexes would make only a slight contribution to change in their overall stabilities and hence their redox potentials. Could Williams give any examples of iron complexes where the redox potentials in any series do, as he suggests, increase with the pA" of the ligand ? Williams: The arguments I use in discussing redox potentials are set out in full in my papers. There is as yet no direct evidence for the maximum Perrin discusses. On the other hand for the series of ligands H2O, pyridine, histidine, ammonia, there is good evidence that ^ohas little relationship to pA: of the base (see Dwyer, this volume, p. 25, and Falk and Perrin, this volume, p. 69). In both cases E^ goes through a maximum with pAT. I do not accept either the discussion of iron phenanthroline or iron 8-hydroxy- quinoline complexes given by Falk and Perrin {loc. cit.) and in the question, but prefer 54 Discussion our own interpretation of the data (Tomkinson and Williams, /. chem. Soc. 2010, 1958) for the reason given in that paper. I do not agree with Perrin's interpretation of ligand field theory as presented in the question and in his paper (see discussion of Orgel's paper, p. 13). I think that bonds, both in energy and in length, undergo considerable changes on change of spin type and that in biological systems these changes are of greater importance than almost any other factor. I was well aware of the data on the redox potentials of iron porphyrin complexes and have tried to use them properly and with due reservation. Perrin : The experimental values of E^ for metal porphyrin complexes with bases which Falk and I list represent probably the most complete series available in the literature. They show a general decrease with ^K of the ligands and so cannot be quoted as support for Williams' suggestion that they should pass through a maximum. I think it is dangerous to attempt to argue too finely from redox potential differences. For example, the porphyrin in cytochrome c differs from mesoporphyrin (with which it was compared) by having two CH3CHSR groups instead of ethyls. It would seem to be better to take haematoporphyrin, with two CH3CHOH groups, rather than meso- porphyrin, as an approximation to porphyrin c. The difference in redox potentials is then much smaller for the reactions : Fe++ (porph)-Py2 + OH- = Fe+++ (porph)-Py-OH + Py + e at pH 9-6, £'0 is + 15 mV for protoporphyrin, + 4 mV for haematoporphyrin, and — 63 mV for mesoporphyrin. (Lemberg and Legge, Haematin Compounds & Bile Pigments, 1949.) Potentials are even more different depending on whether the reaction is for the loss of a molecule of base from the ferric complex and its replacement by OH", as in the example just quoted, or simply for the loss of an electron from the Fe(porph)-B2 complex. The difference between the pyridine and the histidine complex of iron-pro toporphyrin is < 80 mV in the first case and about 210 mV in the second case (calculated from Barron, /. biol. Chem. Ill, 285, 1937, and Shack and Clark, /. biol. Chem. 171, 143, 1947). In the absence of other evidence as to the nature of the extra ligands in cytochromes and other metalloporphyrins, any suggestions from redox data must be almost entirely speculative. It should also be pointed out that there is not one, but many, members of each of the cytochrome families. There are, for example, many cytochromes c which, although similar in absorption spectra, do not have the same redox potential — compare cytochromes Cj and c^ where the Eo-values differ by 0-545 V (Morton, Rev. pure appl. Chem. 8, 161, 1958). Which of these should we assume from spectra or redox potentials to contain two imidazole groups bound to the metal, and how are mixed ligands to be ruled out? Williams : In all my discussions of redox potential data I have been fully aware of Perrin's points. I therefore restate that (i) All conclusions about structure are made taking into account spectra, redox, and magnetic properties (see Chem. Rev. 56, 299, 1959 for the way in which I do this). (ii) A maximum in redox potential is only expected on change of spin type. None is expected for the compounds in the series of Falk and Perrin (p. 69) except in the sequence water, pyridine, imidazole, NHg. (iii) Mixed complexes are treated later in this discussion (p. 55). The cytochromes c of different kinds are discussed by myself and Chance in Disc. Faraday Soc. 27, 269, 1959. Mixed complexes are included in that discussion. George: In answer to Williams' question about the £'0 for the Fe+++/Fe++ tetrapyridyl couple, I would like to report on the results of an investigation recently carried out in collaboration with G. Haight and A. Bergh, We thought originally, following Morgan and Burstall, that both ferrous and ferric derivatives were square planar complexes containing one tetrapyridyl molecule, somewhat analogous to haem and haemin. But while the ferric complex has the composition Fe(tetrapy)i+++, two ferrous complexes are formed with K^ >-^2» Fe(tetrapy)i++ and Fe(tetrapy)2++. K for the ferric complex is greater than K^ for the first ferrous complex, so that upon the addition of tetrapyridyl, E'q first falls below the value of 0-77 V for the Fe+++/Fe++ aquo-ion couple. But, in principle at least, as Some Physical Properties and Chemical Reactions of Iron Complexes 55 the tetrapyridyl concentration is increased Eq will eventually increase again when the oxidation-reduction reaction becomes Fe(tetrapy)i+^ + + tetrapy + t'~ -;-=^ Fe(tetrapy)2++ bright yellow reddish-violet Using Courtauld atomic models we found that co-ordination of all four TV-atoms to give square planar complexes is not possible. In all probability the structure of the Fe(tetrapy)2++ complex is that of a distorted octahedron with only three of the four nitrogens of each tetrapyridyl molecule co-ordinated to the iron. Its absorption spectrum resembles that of Fe(tripy)2++ very closely, which supports this hypothesis. The 1 : 1 complexes probably have only three bonded N-atoms, and we suppose that steric hindrance prevents the formation of the ferric complex corresponding to Fe(tetrapy)2++. Margoliash: For cytochrome c the evidence we have obtained from a study of the chemical and physico-chemical properties of the denaturation products, as well as of those of the pepsin digested 'core', indicates that cytochrome c is probably not a di-imidazole haemochrome, but probably a mixed haemochrome with a primary amino-group and an imidazole group bound to the haem-iron. Moreover, by denatur- ation it is possible to obtain products having Eq values ranging from that of native cytochrome c down to not far from V. In cytochrome c the E^ value seems to be an expression of the effect of the protein configuration on the haem iron-ligand bonds rather than an intrinsic property of the particular groups involved. I should therefore think it would be difficult to ascribe specific ranges of E'q values to specific haemo- chrome-forming ligands in haemoproteins. Williams : I consider that Margoliash has studied a series of complexes, often mixtures varying from di-imidazoles through mixed complexes, to di-amines. No simple explanation of his results is possible. SPECTRA AND REDOX POTENTIALS OF METALLOPORPHYRINS AND HAEMOPROTEINS By J. E. Falk* and D. D. PERRiNf Division of Plant Industry, C.S.I.R.O., Canberra and Department of Medical Chemistry, Australian National University, Canberra Why is it that there is no relationship between the oxidation-reduction potential of the cytochromes a, b and c and their absorption spectra (Table 1)? The porphyrin side-chains of these cytochromes increase in electron-attracting power in the order c, b, a. This sequence is reflected in the spectra of the Table 1 Cyto- chromes Side-chains in positions: Fe++- cytochrome Absorption maxima (m/ii) £•0'* PH 7 (V) Pyridine haemochromes Absorption maxima (m/j) c b a 2 4 8 — CHSRCH3 — CHSRCH3 — CH==CH2 — CH=CH2 — CHOHCH2R2 — CH=CHRi — CHO 550 564 603 -i-0-255 +0077t 4-0-29 551 557 587 Here and throughout this paper, the data, for the cytochromes of animal mitochondria, are taken from Morton (1958). Side-chains other than those shown are methyl and propionic acid groups, which have little effect upon the properties discussed. For side- chains in haem a, see Lemberg, Clezy and Barrett, this volume, p. 344. * The Eo' for a reaction is the electrode potential for 50% oxidation at a stated pH. t From Colpa-Boonstra and Holton (1959). cytochromes and of the pyridine haemochromes of their prosthetic groups. But while the pyridine haemochrome of haem c has a spectrum very like that of the cytochrome itself, the spectra of cytochromes b and a are displaced far to the red of their respective haemochrome spectra. The redox potentials of the three cytochromes follow no sequence whatever in relation to their spectra, or the chemistry of their haem prosthetic groups. In the absence of protein, however, the electron-attracting power (Falk and Nyholm, 1958) of porphyrin side-chains is correlated with changes of both spectrum (Table 2) and redox potential (Table 3). * C.S.I.R.O. t Australian National University. 56 Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 57 Table 2 Suljctitiipntc in Porphyrin absorption Pyridine haemochrome 0...^. ... ... maxima, band I absorption maxima a-band deuteroporphynn IX (dioxan) at positions: (m//)* (m/i) 2 4 — H — H 618 545t — C2H5 -QHs 620 547t — CH=CH2 — CH=CH2 630 558* — H — COCH3 634 571* — COCH3 — COCH3 639 575* — H —CHO 640 578* — CHO — CH=CH, 644 583* — CHO —CHO 651 584* * From Lemberg and Falk, 1951. t This study. Table 3 Haems Side chains in positions: £0 of the (CN-)2 derivatives* of the Pyridine2 derivatives! -CH2CH2COOH -C2H5 -CHOH— CH3 -CH=CH2 -CHO -CH2CH2COOH -C2H5 -CHOH— CH3 -CH=CH2 -CH=CH, -0-247 -0-229 -0-200 -0-183 -0-113 -0-04 +0-107 * From Martell and Calvin (1952). t Vestling, 1940. Among those cytochromes b for which it has been estabhshed that the prosthetic group is protohaem, there is again no correlation between spectrum and redox potential; the same is true for those cytochromes c for which it has been established that haem c is the prosthetic group (Morton, 1958). These anomalies can be due only to some particular properties of the proteins. If we regard these biological compounds as further co-ordination complexes of haems, to what extent can these differences from model com- plexes be explained in terms of the nature of the protein-haem iron bonds ? It is possible that, as ligands, the proteins may have properties difficult or impossible to reproduce in model systems. Thus, the stereochemical environ- ment of the protein ligand atom may influence the way in which this atom co-ordinates. The protein, and in intact tissue perhaps other macromolecules, 58 J. E. Falk and D. D. Perrin may well create a specific micro-environment about the haem molecule. The evidence v/hich Dr. Wang has found (Wang, Nakahara and Fleischer, 1958) that the dielectric properties of the medium affect some co-ordination properties of haem may be an example of such an effect. Nevertheless, we believe that it is important to see how much or how little can be learned from the study of model systems. The haemochromes which have been studied in any detail that bear any likely relationship to natural haemo- proteins are, practically without exception, complexes between haems and ligands containing unsaturated nitrogen atoms (=N — ). But complexes of protohaem with such ligands do not have the range of absorption maxima of the cytochromes b (which are protohaem complexes), nor do they have comparable redox potentials. We wish to outline here some theoretical aspects of porphyrins and metalloporphyrins and to draw some inferences from existing data. We discuss below (p. 74) some studies we are making of complexes formed by several different haems with ligands of a variety of chemical types. SOME THEORETICAL ASPECTS OF PORPHYRINS AND METALLOPORPHYRINS It is convenient to picture the porphyrin molecule as a framework of atoms held together by ordinary, two-electron, single {g) bonds, while the remainder of the valence electrons occupy molecular orbitals which extend over the whole of this framework. The strong delocalization of these mobile (tt) electrons confers considerable stability and 'aromatic' character on the porphyrins. Electron-withdrawing substituents on the peripheral carbon atoms reduce the 77-electron density on the pyrrole nitrogens, so that it becomes easier for the protons to dissociate from the two pyrrole N — H groups which make the porphyrin molecule a weak dibasic acid. Though little precise data exist, this clearly increases the acid strength of the porphyrin (lowers its pi^„) and, as discussed later, raises the oxidation-reduction potential of metalloporphyrins. The Absorption Spectra of Porphyrins Another consequence of the extensive 7r-electron delocalization is that the highest of the occupied molecular orbitals and the lowest of the vacant orbitals differ in energy by an amount small enough for transitions between them to give rise to absorption bands in the visible and near ultra-violet. In the porphyrins themselves in neutral solvents there are four bands in the visible in addition to the Soret band at about 400 m^. There appears to be good reason to believe (Piatt, 1956) that the four visible bands are really two pairs of bands which would be superimposed if the porphyrin nucleus were strictly square and uniformly substituted. X-ray analysis of the closely similar phthalocyanine molecule has shown that its structure is slightly Spectra ami Redox Potentials of Metalloporphyrins and Haemoproteins 59 distorted from square (Robertson, 1936), probably because the hydrogens bound on opposite nitrogen atoms each form hydrogen bonds with an adjacent nitrogen atom. This distortion probably occurs in the porphyrins also (Mason, 1958; Piatt, 1956), leading to the fairly constant difference of 6-7 kcal between the energy levels of corresponding absorption bands (I and III, II and IV). This difference is removed in the di-anion, the di-cation and the metal complexes, and in all these cases only two of these bands are found. By making some simplifying assumptions Longuet-Higgins, Rector and Piatt (1950) and Seely (1957) have carried out molecular orbital calculations to find the nature of the transitions giving rise to the observed porphyrin spectra. Essentially the same conclusions are reached using an electron-gas model (Kuhn, 1959). The visible bands all arise in a similar manner, as transitions between filled orbitals of /IgM'typ^ symmetry and vacant ^'^-type orbitals (Fig. 1). In all cases these bands are associated with an electronic displacement towards the periphery and this may be either along (bands III and IV) or perpendicular to (bands I and II) the axis tlirough the two H's which are on opposite nitrogens (Piatt, 1956; Mason, 1958). Because bands I and III are for a 0-0 vibrational transition which is classed as forbidden, their intensities will depend very much more on any loss of symmetry in the poi-phyrin molecule than will bands II and IV which are interpreted as 0-1 vibrational bands (Piatt, 1956). This symmetry, which is to be thought of in terms of possible pathways for the mobile electrons, will not be much affected by substituents such as alkyl groups, or carboxyl groups which are insulated by at least two CHg's as in propionic acid side-chains. Much greater distor- tion would be expected from substitution of a peripheral hydrogen by a group such as — CHO, — COCH3, — COOCH3 and COCgHj, and it is among such porphyrins that 'oxorhodo' (III > II > IV > I) (Lemberg and Falk, 1951) and 'rhodo' (III > IV > II > I) type spectra are found rather than the 'actio' (IV > III > II > I) type which is the usual one with alkyl and similar substituents (Stern and Wenderlein; for references see Lemberg and Falk, 1951). Extension of tliis generalization to porphyrins containing more of the symmetry-disturbing groups is difficult because of the necessity to allow for the vector directions of the moments of the substituents, but if this is done there is a reasonable correlation between observed and predicted spectra (Piatt, 1956). The a- (long wavelengths) and /S-bands of metal-porphyrin complexes seem to be related to bands I and III, II and IV, respectively. Thus bands II, IV and /5 are little affected by substituents, while bands I, III and a vary considerably. It has also been shown that the intensity of the a-band in some copper-porphyrin complexes varies with the intensity of band III of the corresponding porphyrins (Williams, 1956). The Soret band is attributed to the transition to an ^^-type orbital of an electron in an y4i„-orbital (Fig. 1) in which it was confined to the carbon atoms H.E. — VOL. I — F 60 J. E. Falk and D. D. Perrin of the pyrrole rings. The ^'^-orbital is one of a pair of equal energy strongly polarized along and perpendicular to the axis of the two NH's. Although the overall movement of electronic charge towards the periphery is negligible, the increase of electron density on the non-pyrrole carbons and a pair of Fig. 1. Nature of transitions giving rise to porphyrin spectra. The signs + and — indicate wave functions. The areas of the circles give electron densities in the vicinities of atoms ; the atomic distances are the same as in phthalocyanine (Robertson, 1936). opposite nitrogens (Fig. 1) affects the absorption spectrum of porphyrins and metalloporphyrins. Co-ordination of Porphyrins with Divalent Metals The property that distinguishes transition metals from other elements in the Periodic Table is that their 6?-orbitals are incompletely filled with electrons. These orbitals have the directions in space shown in Orgel's paper {loc. cit.). Depending on whether electrons occupy these orbitals singly or in pairs, complexes will be para- or dia-magnetic. This difference in magnetic properties is interpreted in the Valence Bond Theory as distinguishing two types of complexes. The main concept underlying Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 61 this theory, as applied to co-ordination complexes, is that suitable vacant orbitals of the metal are hybridized, and these hybrid orbitals are filled by electron pairs 'donated' by ligand atoms with the formation of cr-bonds. To be suitable for hybridization in this way an orbital must have an appreciable component in the directions finally occupied by some or all of the ligands. The diamagnetism of pyridine haemochromes is interpreted to mean that two of the 3i/-orbitals of Fe++ {d^i^yi and d^^ are used in this hybridization and are occupied by two pairs of ligand electrons ; the electrons already in these orbitals are forced to pair up in the remaining 3c/-orbitals. Such com- plexes have long been called 'covalent' and, more recently, 'inner-orbital', 'spin-paired', or 'low-spin'. Their formation is favoured by ligands of low electronegativity and in octahedral complexes such as pyridine haemochrome they are described as being ZdHsAp^, or d'^sp'^, types. On the other hand, haemin chloride, like the Fe+++ ion itself, has 5 unpaired electrons. Such paramagnetic complexes ('semi-ionic', 'outer-orbital', 'spin- free', 'high-spin') are generally formed by ligands of high electronegativity which, in addition, have little or no d acceptor capacity for double bonding (e.g. F" as against pyridine-N). It is now believed that high-spin complexes do have some degree of covalent bonding (cf. Craig et al., 1954) and it is convenient to regard haemin chloride, for example, as a hybrid of the type AsApHd"^. There is little doubt that in 'haemin chloride' (ferriprotohaem chloride), traditionally regarded as a square-planar complex with the Cl~ ionically associated, the Cl~ is 'co-ordinately' bound. Falk and Nyholm (unpublished) have found a 0-001 m solution in nitrobenzene to be a non- conductor of electricity. Under these conditions, univalent electrolytes have conductivities of 20-30 r.o. It appears likely that ferriprotohaem hydroxide ('haematin') is a similar complex. A more recent and more satisfying interpretation of the magnetic and other properties of complexes is provided by the Ligand Field Theory (Griffith and Orgel, 1957). In essence, this theory says that as co-ordinating groups, or ligands, approach a metal ion to form a complex, ^-orbitals pointing towards the ligands are raised in energy and electrons in them become less stable, while J-orbitals pointing away from the ligands become more stable. Bonding molecular orbitals are formed by suitable electron-filled orbitals on the ligands with the metal's vacant s- and /7-orbitals and the ^/-orbitals which point tov/ards the ligands; in octahedral complexes, the c^-orbitals involved are d^2_y2 and d^2. Any electrons already in these ^/-orbitals are removed by promoting them into antibonding orbitals, but their presence reduces the stability of the final complex. In any complex the magnitude of the differences in the energy levels of the various J-orbitals is a function both of the ligand and of the geometrical shape of the complex itself. The electrostatic effect of ligands in splitting these levels is enhanced by *back double bonding', which arises from the ability of suitably placed, 62 J. E. Falk and D. D. Perrin occupied ^-orbitals on the metal to form molecular orbitals with vacant 7T-orbitals on the ligand, so that these electrons gain in stability. (If the 77-orbitals are already filled the interaction is repulsive.) Similar interactions can take place between vacant orbitals on the metal and filled 7r-orbitals on the ligand. Effects such as back double bonding cannot be separated from True octahedral complex ZA ----B- I-9J 0=0 /^/ N — i-N Globin 10 Square planar complex HjO / Pe / N— |-N Globin 05 Ratio of ligand field in z direction to ligand field in x or y direction Fig. 2. Ligand field splitting of ^-orbitals in Metalloporphyrins (qualitative only). purely electrostatic contributions and they are probably important factors in unsaturated ligands, including porphyrins, oxygen, carbon monoxide, cyanide ion and unsaturated heterocyclic bases, all of which exert strong ligand fields. The porphyrin nucleus confers a planar configuration on its metal com- plexes, and any additional co-ordination sites on the metal are perpendicularly above and below this plane (i.e. along the z-axis). This leads with most transition metals to a more or less vertically distorted octahedral structure and, in the metalloporphyrins, the effect of ligands occupying these positions is to split the ^^-orbital energy levels as shown qualitatively in Fig. 2. If the energy separations are greater than the energy needed to pair electrons in the lower energy levels, diamagnetic or low-spin complexes will be formed; Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 63 otherwise there will be as little spin pairing as possible. For example, in haemoglobin the situation is probably something like that shown at A in Fig. 2. The diflference in ligand field stabilization energy (L.F.S.E.) between the low-spin and the high-spin alternatives is not great enough to prevent the ^/-electrons of the ferrous iron from spreading over all five of the 3fif-orbitals, four of which are occupied singly while the fifth, and lowest, holds a pair of electrons. Replacement of the water molecule in the sixth co-ordination position by oxygen, to give oxyhaemoglobin, displaces conditions towards B in Fig. 2, where the L.F.S.E. difference is sufficient to make the filling of the three lowest orbitals with pairs of electrons the more energetically-favoured process, giving a diamagnetic complex. For theoretical reasons complexes intermediate between high-spin and low-spin are unlikely. It is unnecessary to postulate any sudden changes in the nature of the ligand-metal bonds and, in fact, it is an important consequence of this theory that the magnetic behaviour of complexes does not provide a means of classifying complexes into 'outer-' and 'inner-orbital' or 'ionic' and 'covalent'. THE SPECTRA OF SOME PORPHYRINS AND THEIR METAL COMPLEXES Because the visible absorption spectra of porphyrins are associated with a displacement of electrons towards the periphery of the porphyrin nucleus, any effect which results in an extension of the distance the electrons can move in this direction reduces the energy required for these transitions, so that visible absorption maxima move to longer wavelengths. Any effect operating in the opposite direction moves these absorption maxima to shorter wave- lengths. A number of examples illustrate this. The Ejfect of Porphyrin Side-chains As has already been seen (Table 2) the spectra of a series of free porphyrins, and of the pyridine haemochromes of their Fe '^ "^ complexes, move to longer wavelengths stepwise as the electron-attracting power of the porphyrin side- chains increases. This porphyrin side-chain effect operates similarly in the simple (square) porphyrin complexes with a variety of divalent metals (cf. Table 2 of Falk and Nyholm, 1958), and indeed throughout the metallo- porphyrins of all types, including, in a broad sense, the haemoproteins (cf. Table 1). Among the latter, replacement of the protein with pyridine is a convenient way to obviate effects on spectrum peculiar to the protein; pyridine haemochrome spectra reflect accurately the effects of electron- attracting side-chains on the haem nucleus. The Effect of Co-ordinated Metal Ions Falk and Nyholm (1958) have compared the protoporphyrin complexes of a number of different divalent metal ions. It was found that the following 64 J. E. Falk and D. D. Perrin complexes fell into three classes, according to magnetic susceptibility, spectroscopic and other properties : A Co++, Ni++ B Cu++, Ag++ C Zn++, Cd++ From the point of view of valence-bond theory, the conclusion was reached that class A were spin-paired (3d4s4p^), class B spin-free (4s4pHd, 5s5p^5d respectively) ; both A and B have great (qualitative) stability. Class C is also spin-free {4s4pHd, 5s5p"5d) but of great (qualitative) instability. Within the three classes the spectra were very similar, but from ^4 to 5 to C the visible absorption moved to longer wavelengths. From Fig. 2, the essential difference between classes A, B and C is that they have 0, 1 and 2 electrons, respectively, in the d^i_y2 antibonding orbital. These electrons are favourably placed for strong repulsive electrostatic interaction with electrons on the pyrrole-N's, so that the more electrons in d^2_y2 the easier it is to displace 7T-electrons towards the periphery. Hence, in agreement with experiment, the visible absorption maxima move to longer wavelengths in passing from yl to jB to C. In the transition which gives rise to the Soret band there is a displacement of some of the electron density to the non-pyrrole (methene bridge) carbons, i.e. in a direction away from d^2_y2 (Fig. 1). This transition should be facilitated in the same way and the Soret maximum should shift to longer wavelengths in a similar sequence, as observed by Falk and Nyholm (1958) for the protoporphyrin complexes in benzene, which at the time appeared to be an inert solvent. It has recently been found (J. N. Phillips, unpublished) that in fact this solvent modifies the spectra of certain of the metalloporphyrins. The measurements have been repeated in CCI4, which shows no evidence of interaction, and the follov/ing maxima have been found : Co Ni Cu Ag Zn Cd a-band, m/< Soret, m/.i 561-5 403 561 403 573 409 570 417-5 579 411-5 5875 414 The a-bands have virtually identical maxima in benzene and in CCI4, as do the Soret bands of the Co, Ni, Cu and Ag complexes. The Soret bands of the Zn and Cd complexes in benzene, however, were at 415 and 423 m/< respectively. In the square planar, low-spin, cobalt (d"^) and nickel (d^) complexes, since there are the same number of electrons in d-^y (=2) and none in dj.2_y2, similar Soret and visible maxima would be predicted and have been found. In the same way the ferrous 6/>pyridine complexes which are diamagnetic Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 65 with an octahedral distribution of ^-orbital splittings, have no electrons in d^2_yi and two in d^y. From a spectroscopic point of view they therefore approach conditions for square planar nickel and cobalt complexes. Back double bonding should shift the visible bands to slightly shorter wavelengths and the Soret band to slightly longer wavelengths. Thus, for the proto- porpiiyrin metal complexes the predicted order of the visible wavelength maxima is: Co c;^ Ni > FePyg, and the observed values are 561, 561 and 558 m/< respectively. Although the stability of porphyrin metal complexes is due in large measure to the difficulty of providing enough energy to rupture four bonds, whether electrostatic or covalent, simultaneously, an additional effect in transition elements is the ligand field stabilization energy. Considerations of L.F.S.E. would suggest that the stability of these metalloporphyrins should lie in the series, A> B> C. This is because electrons occupying low lying ^-orbitals increase the stability of a complex, while those in higher (antibonding) orbitals will remove some of this stability. The additional stabilization becomes zero when all five of the J-orbitals are equally occupied. We estimate the L.F.S.E. for the bivalent Co, Ni, Cu and Ag complexes of protoporphyrin dimethyl ester to be at least 40 kcal, and this may be one factor contributing to the quahtative differences in the difficulty of dissociation of the metal from these complexes, as against the Zn, Cd and Pb complexes (which have no L.F.S.E.) (Falk and Nyholm, 1958). For ferro- and ferrihaemoglobin the estimated L.F.S.E.s are 20 kcal and zero, respectively. The great stability of ferric iron in complexes is probably mainly due to the electrostatic forces of attraction between opposite charges. The Effect ofLigands in the 5th and 6th Co-ordination Positions We restrict our discussion here to further complexes of iron-porphyrins, i.e. the haemochromes. The 5th and 6th positions on these complexes, above and below the plane of the haem molecule, correspond to positions 1 , 6 in octahedral complexes in general co-ordination chemistry. We discuss the spectra of the Fe++ complexes only, since, as is commonly recognized, their spectra are not complicated by the large component which has been attributed (Williams, 1956) to charge-transfer from the ligand to the metal in the Fe+++ complexes. Ligands to haem iron such as pyridine and chemically similar bases, and CN~ ion, allow back-double-bonding of the electrons in the dy^ and d^^ orbitals of the metal. The time these electrons spend in the plane of the porphyrin molecule, and hence the average electron density in this plane are reduced, so that it is harder for electrons to move towards the periphery, and the visible absorption maxima move to shorter wavelengths. That is, there is less electrostatic repulsion in the ground state if double bonding can occur. 66 J. E. Falk and D. D. Perrin Ligands such as ROH (including HOH), RS", RCOO-, HQ- and others, by their dipolar or electrostatic interactions with the metal ion, should facilitate the movement of the mobile porphyrin electrons away from the metal, and shift absorption to longer wavelengths. Among the haemoproteins, one clear example of these two effects is seen when Fe++ cytochrome c and Fe++ peroxidase are compared. The former is low-spin (diamagnetic) and the latter is high-spin (4 unpaired electrons found). The visible absorption maxima lie at 520, 550 m/^ and 558, 594 m/< respec- tively (Lemberg and Legge, 1949). In cytochrome c one, and possibly both, protein groups bound to the haem iron are histidine nitrogens, and in peroxidase the groups are a — COO on one side of the haem and probably a water molecule on the other (Chance, 1952; Theorell and Paul, 1944). Among Fe++ haemoglobin derivatives a similar change from high- to low-spin (para to diamagnetism) is found on substitution of water (Hb itself) by the double-bonding ligands O2 or CO. Similarly, when the 6th position on the haem of Fe+++ haemoglobin is occupied by HgO, F~, 0H~, EtOH, high-spin complexes, and by the double-bonding ligands CN~, — SH, Ng" or imidazole, low-spin complexes are found (for references see Falk and Nyholm, 1958). Similar examples are to be found among peroxidase derivatives (Lem- berg and Legge, 1949). Complexes of the haems with non-protein ligands have been studied extensively, but many of the data which may be very relevant to understanding of the haemoproteins are lacking. For comprehen- sive reviews of this subject, see Lemberg and Legge (1949) and Martell and Calvin (1953). As might be expected, with ligand atoms of high field strength and also capable of double bonding, such as =N — (in pyridine, nicotine, a-picoline, imidazole, etc.), low-spin complexes are formed. Thus all the complexes of protohaem listed in Table 5 are diamagnetic. Though there must be variations in complex-forming ability between these ligands, as indicated by their pA!' and Eq values (Table 5), the spectra of the complexes are very similar; this is probably because of comparable back-double bonding ability. In these hexaco-ordinate complexes, the wavelengths would be expected to be modified by the ligands on the 5th and 6th co-ordination positions only if these hgands alter the spatial distribution of the electrons round the central metal ion in such a way as to affect electronic transitions in the plane of the porphyrin molecule. Such effects would be expected if spin-free and spin-paired com- plexes were compared, or even in complexes with very different amounts of back-double-bonding, spin-paired and least back-bonded having maxima displaced towards longer wavelengths. However, especially with unrelated ligands, there is no reason why the factors that govern their complex-forming ability with two different valence states of a metal should bear any relation to the effective component of the metal's c?-electrons at right angles to the direction of bond formation, Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 67 as would be required for any general correlation of spectra with redox potential. OXIDATION-REDUCTION POTENTIALS Factors affecting the oxidation-reduction potentials of metal complexes (Perrin, 1959b) include: (i) Purely electrostatic effects of attractions between ions or dipoles of opposite charge. Thus if the ligand is an anion this will always favour the higher valent state of the metal and the potential will be less than for the corresponding free metal ions in water. This is because the standard oxidation-reduction potential, E^, of any pair of ligand- metal complexes is related directly to the stabiUty constants of the complexes by the identity, 2'3036RT^, ,, , .. ^ E1 = Em- 7 7^ (log ^M"^ - log Kj^m+) (n — m)r where E^ is the potential of the free metal ions. (ii) Back-double-bonding. Because this involves the removal of electrons from the vicinity of the positively charged metal ion, the effect is always greater for the lower valent state of the metal, so that it tends to raise the oxidation-reduction potential. (iii) Ligand field stabilization energies. The difference in L.F.S.E. varies with the particular pairs of cations and the ligands, as discussed in Dr. Orgel's paper (loc. cit.). For example, in weak ligand fields ferrous ion, but not ferric ion, is stabilized in this way : this raises the potential. On the other hand, in manganous, manganic systems manganic ion is stabilized but not manganous ion, so the potential is lowered. (iv) The acid dissociation constants (p/sTa's) of the ligands. In many series of closely related ligands the stabiUty constants of metal com- plexes vary with p^^ ^^ ^^ approximately linear fashion : log K ^^ apAr„ •\- c. Such a relation would be expected if the factors governing the binding of protons and cations by ligands were similar. From a simple electrostatic model it has been predicted that a should increase with increasing cationic charge (Jones et al., 1958). As a direct consequence, the oxidation-reduction potentials of metal complexes in which the ligands are sufiiciently similar should decrease linearly with increasing ipK^. For several series of iron complexes this has been found to be the case (Perrin, 1959b). In the porphyrins, electron-withdrawing substituents at the peripheral carbons increase the acid strength of the porphyrin (lower the pATJ and hence raise the oxidation-reduction potential of the metal complex. Martell and 68 J. E. Falk and D. D. Perrin Calvin (1953) have discussed this correlation between electron-withdrawing effect and Eq; potentials of iron complexes rise as the electron-attraction of side-chains increases (Table 3). The differences, although rather small, lie in the expected order. In suflEiciently alkaline solutions, complexes of nitrogenous bases with iron- porphyrins give oxidation-reduction potentials which vary Unearly with pH. The reaction can be written: Fe+++B2 + OH- ^ Fe++B . OH + B + e although the ferric complex may be present mainly as an easily split dimer (Shack and Clark, 1947). Such potentials should therefore be compared at approximately constant base concentration and pH. Around pH 9-6 reported potentials are as shown in Table 4, Table 4 Haems Side-chains in positions: £■0 at pH 9-6* (Pyr.)2 complex Eo at pH 9-6* (a-Picoline)2 complex 2 4 — CgHj — C2H5 — CH2CH2CO2H — CH2CH2CO2H 2 6 4 7 — C2H5 — C2H5 — C2H5 — C2H5 2 4 — CHOHCH3 — CHOHCH3 — CH=CH2 — CH^=Cri2 —CHO — CH=CH2 -0-063 -0-036 -0-029 +0-004 +0-015 -0-099 -0-033 -0010 Data from Martell and Calvin (1952). * For the reaction Fe+++B2 + OH- ^ Fe++BOH + B + e. A different kind of comparison can be made by keeping the porphyrin nucleus constant and varying the base co-ordinating with the iron complexes. Taking the data of Barron (1937) for protoporphyrin-iron we find that at pH 9-2 nicotine, pyridine and a-picoline give oxidation-reduction potentials showing the expected pH-dependence — A^"/ ApH = 0-06. On the other hand, the slope for histidine and pilocarpine is much less, indicating that the complexes Fe+++B2 are also present in significant amounts. To minimize this interference the Eq values listed in Table 5 have been calculated from the data using as low a concentration of bases as possible. The series shows a roughly linear dependence on pK^, with a slope of the order of — AEfApK <^ 0-04 V. This slope is similar in magnitude to that found for 1 : 1 iron Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 69 complexes with amino-acids, 8-hydroxyquinoline and o-phenanthroline (Perrin, 1959a). Table 5. Complexes of protohaem with bases Base ^Ka [B] M E',* pH 7 Haemochrome a-band, m/< Nicotine 3-15 0-04 0-200 5581 Pyridine 5-2 0-7 0-158 558t a-Picoline 6-2 0-5 0-115 558t Histidine 6-0 0-03 > 0-079 Pilocarpine 7-0 002 > 0-052 556-2t * Fe+++B2 + OH- t This study. + Barron (1940). Fe++BOH + B + e. Much less information is available for reactions of the type Fe++B2 ^ Fe+++B2 + e which become important in neutral solutions. For iron-protoporphyrin we find the values shown in Table 6. As would be expected the anion, CN"", Table 6 Complexes of protohaem with: ^Ka p * Pyridine Histidine Pilocarpine Water Cyanide ion 5-2 6-0 7-0 +0-107t -0-1051: -0-13t -0-14t -0-183t * For the reaction Fe^+Bg t Shack and Clark (1947). t Barron (1937). Fe+++B2 + e. stabilizes the ferric state to a greater extent than any of the neutral ligands : among the latter the ferric state is favoured the higher the p/f^ of the ligand. Compared with the cyanide complexes, potentials for these reactions appear to be much more sensitive to change in the porphyrin in the complex if the ligand is neutral (pyridine. Table 3). SUMMARY AND CONCLUSIONS We have outlined some theoretical aspects of the absorption spectra of porphyrins and metalloporphyrins, of the co-ordination chemistry of metallo- porphyrins, and of the redox potentials of haems and haemochromes. 70 J. E. Falk and D. D. Perrin It has been shown that the following firm correlations exist among the non-protein complexes : (a) The more electron-attracting the side-chains on the porphyrin nucleus, the greater the shift to longer wavelength of the visible absorption maxima of porphyrins, their square metal complexes, and their haemochromes with double-bonding ligands. (b) The more electron-attracting the porphyrin side-chains, the more positive the redox potential of the haemochromes formed by double- bonding ligands. (c) The greater the electron-donation (higher pK^) of double-bonding ligands, the more negative the redox potential of the haemochromes formed by them. (d) For ligands, in the above group, of ipK^ from about 3 to 7, although Eq varies by about 150 mV, the spectra of the haemochromes hardly differ. Reasons for this have been discussed. Among the haemoproteins : (a) The effects of porphyrin side-chains upon both spectrum and redox are obscured and overweighed by the effects of the proteins. (b) The haemochromes which have been studied in any detail are mainly those formed by double-bonding ligands, and their spectroscopic and redox properties may have little value as models for haemoproteins, even in cases hke cytochromes c, which resemble haemochromes in some ways. REFERENCES Barron, E. S. G. (1937). J. biol. Chem. Ill, 285. Barron, E. S. G. (1940). J. biol. Chem. 133, 51. Chance, B. C. (1952). Arch. Biochem. Biophys. 40, 153. CoLPA-BooNSTRA, J. & HoLTON, F. A. (1959). Biochem. J. 72, 4P. Craig, D. P., Maccoll, A., Nyholm, R. S., Orgel, L. E. & Sutton, L. E. (1954). J. chem. Soc. 332. Falk, J. E. & Nyholm, R. S. (1958). Current Trends in Heterocyclic Chemistry, p. 130. Butterworths, London. Griffith, J. S. & Orgel, L. E. (1957). Chem. Revs. 11, 381. Jones, J. G., Poole, J. B., Tomkinson, J. C. & Williams, R. J. P. (1958). J. chem. Soc. 2001. Kuhn, H. (1959). Helo. chim. Acta 42, 363. Lemberg, R. & Falk, J. E. (1951). Biochem. J. 49, 674. Lemberg, R. & Legge, J. (1949). Haematin Compounds and Bile Pigments, Interscience, New York. Longuet-Higgins, H. C, Rector, C. W. & Platt, J. R. (1950). J. chem. Phys. 18, 1174. Martell, a. E. & Calvin, M. (1952). Chemistry of the Metal Chelate Compounds, Prentice-Hall, New York. Mason, S. F. (1958). /. chem. Soc. 976. Morton, R. K. (1958). Rev. pure appl. Chem. 8, 161. Perrin, D. D. (1959a). J. chem. Soc. 290. Perrin, D. D. (1959b). Rev. pure appl. Chem. 9, 257. Platt, J. R. (1956). Radiation Biology, 3, 101. McGraw-Hill, New York. Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 71 Robertson, J. M. (1936). /. chem. Soc. 1195. Seely, G. R. (1957). J. chem. Pliyi. 27, 125. Shack, J. & Clark, W. M. (1947). J. biol. Chem. 171, 143. Theorell, H. & Paul, K.-G. (1944). Ark. Kemi. Min. Geol. 18A, No. 12. Vestling, C. S. (1940). J. biol. Chem. 135, 623. Wang, J. H., Nakahara, A. & Fleischer, E. B. (1958). J. Amer. cliem. Soc. 80, 1109. Williams, R. J. P. (1956). Chem. Rev. 56, 299. DISCUSSION Correlations between Structure and Physical Properties George : With regard to the correlations that are being sought between cheniical reactivity, physical properties and structural factors in co-ordination chemistry, and their exten- sion to haemoprotein compounds, I would call attention to the more detailed and revealing information that can often be obtained from A/f" and AS" data, which it is not possible to get if only AG" data (e.g. E'q and pAT values) are considered. In many cases, notably when ligand field effects are being investigated, A/f" is the significant thermodynamic quantity: and the successful correlations based on AG" probably result from the values of A^" remaining relatively constant throughout a series of compounds. But in some instances apparently valid conclusions based on AG" turn out to be rather misleading. For example there is the well-known correlation between the affinity of structurally similar ligands for a metal ion and for the hydrogen ion; linear plots are obtained for log (stability constant) against ^K. This suggests that the stronger the bond to hydrogen, the stronger is the bond to the metal. Yet an examination of the rather scanty thermodynamic data which are available indicates that the correlation is only determined by the A//" values, which contain the bond energy terms, for certain families of ligands, and that for others the AS" values dominate the relationship. Similarly oxidation-reduction potentials are often regarded as a relative msasure simply of the energy required to remove an electron from the reduced form of the couple. But this is not always so. For example, the Eq values for the Fe(dipy)3^+/^+, Feaci^+/^+ and Fe(CN)6=^-/*- couples are about 1-0, 0-77 and 0-36 V respectively. Yet this sequence is not determined by the electron-donating property of the ligands following the sequence CN" < HjO < dipyridyl. The values of Ai/" for the cell reaction in the presence of these ligands Fe'" + iH2^Fe" + H+ are about —30, —10 and —26 kcal/mole respectively, which show that the contri- bution from the ionization potential of the Fe++ compounds is more nearly the same for the dipyridyl and the cyanide complexes, and that entropy changes play a very dominant role in determining the magnitude of Eq. This is not unexpected in view of the entirely different charge changes in the cell reactions, -1-3 to -1-2 in contrast to —3 to —4 respectively. Somewhat more surprising is the unfavourable entropy contribution for the corresponding couples of haemoglobin and myoglobin as com- pared to the favourable entropy contribution for the aquo-ion couple. The apparent charge changes are -f-l to zero and -1-3 to +2 respectively, both of which should make a favourable contribution to AS'". A//" is nearly the same as that for the aquo- ion couple, and it is the entropy change which is responsible for the Eq values for the two haemoproteins being some 0-6 V lower, i.e. at about 0-2 V compared to 0-77 V. In setting up correlations, therefore, it can be extremely important to determine whether A//° or A5" is the dominant factor. Spin Type and Spectra of Haem Compounds Williams: The spectra of metal porphyrins have recently been analysed by Gouterman (/. chem. Phys. 30, 1 139, 1959). The suggestion that the band positions are solely due 72 Discussion to the CT-bonding power of the metal is, I believe, incorrect. The band positions are partly affected by metal acceptor power through a-bonds and partly through the 77-bond system (see Williams, Cliem. Rev. 56, 299, 1956). However the total analysis of the spectra of porphyrins given by Gouterman is the best available to date. The diagram in my paper. Fig. 1 (this volume, p. 43), is based on relatively little evidence for protoporphyrin and mesoporphyrin complexes. Lemberg has pointed out to me that it is inadequate for haem a complexes. While I deal with his complexes later I should like now to state the general case (see Fig. 1). In the region of high-spin complexes in both oxidation states Amax of both ferrous and ferric complexes moves to longer wavelengths with increasing a-donor power of the ligands perpendicular to the porphyrin plane. This is also true for the band positions of completely covalent complexes. Both these statements have theoretical and practical support. In the region of mixtures of high- and low-spin complexes the Soret band of Fe++ complexes moves to shorter wavelengths while that of Fe+++ does not. The reasons I give for these shifts remain unaltered but here the discussion is empirical first and theoretical afterwards. In this region all the spectra are composite, part being due to low-spin and part due to high-spin forms. Unlike Falk, I consider that some correlation between spectra and spin form must be established by us for otherwise there is little hope of knowing in what state the cytochrome is in the cell. It may be that the correlations I suggest are only partially true but I know of few exceptions to them. In order to avoid terminological difficulties could I point out the equivalent definitions: ionic = high-spin = weak field covalent = low-spin = strong field The discussion of redox potentials given in my paper needs modification. In an exchange with Chance (Faraday Society Discussion on Energy Transfer, 1959) I was led to the conclusion that £)-type cytochromes are mixed imidazole, amine porphyrin complexes. The diagram of redox potential against ligand basicity now should be as in Fig. 2 of my paper (this volume, p. 47). I have attempted to show the relationships to simple haem complexes at the risk of over-simplification. My table of conclusions on the basis of spectra and redox poten- tials (this volume, p. 48) is incorrect now and should read: Cytochrome b, one amino, one imidazole. From some work we have done recently (Brill and Williams, un- published) we have good reasons for supposing that peroxidase is an amine carboxylate complex rather than a simple carboxylate complex. The chemical reactions of model compounds with oxygen have recently revealed a possible cause of the differences between cytochrome 03, uptake of oxygen and rapid autoxidation, and haemoglobin, uptake of oxygen only. The reactions we have ob- served are that whereas Fe (dimethylglyoxime)2 (imidazole)2 picks up oxygen reversibly, Fe (cyclohexane dione-dioxime)2 (imidazole)2 first picks up oxygen and is then slowly oxidized to a conjugated hydrocarbon. The oxidation is also possible with ferri- cyanide, and iridichloride. Schematically we have: A B Fe++A'(imidazole)2 <^ Fe++A' (imidazole)02 ^ Fe+++A' (imidazole)2 The reversible reaction. A, goes if X has no hydrogens which can react with O2 of the complex, while B follows if X contains suitable hydrogen. If analogies are helpful, the oxidation of cytochrome a^ by O2 is due to the removal of a hydrogen from a sensitive group interacting with the porphyrin and the electron transport is initiated by hydrogen oxidation. Falk : In the paper with Perrin (this volume, p. 56), the haemochromes in our Table 5 are low-spin by magnetic susceptibility measurements; the ligands forming these compounds are =N — atoms of pAT values varying from 3 to 7. The visible spectra hardly differ from each other. I do not have Soret band measurements. In the com- pounds with primary and secondary aliphatic amines described a little later in this discussion, changes directly related to ^K occur in both X and e of the visible bands. Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 73 Soret band data are not available; magnetic susceptibility measurements have not yet been made on these. Because I suspect that there may be surprises in store in the magnetic properties of some of these new compounds, I suggest that we should be careful to refer to low-spin and high-spin compounds only when definitive magnetic susceptibility measurements have been made. If inferences are being made from spectra only, pending magnetic measurements, perhaps we could say 'low-spin spectral type', and so on. With this reservation, I do agree with Williams that we must continue to develop correlations between spectrum and spin-type. Perrin : I suggest that the electronegativity of the central metal is of less consequence in determining spectral shifts in porphyrin complexes than are the number of its d- electrons and the directions of the orbitals they occupy. Falk and I discussed this. In the case of ferrous ion the change from a high-spin to a low-spin configuration means that the unpaired electrons in the d^^_y''- and d^^ antibonding orbitals vacate these positions and all six of the rf-electrons fill, in pairs, the d^y, dy^ and d^.^ orbitals. From the spatial distribution of these orbitals it is easy to see that this leads to an increase in electron density along the xy axes (i.e. through opposite methene carbon atoms) and a reduction in density along the x and y axes; the net increase in electron density in the plane of the porphyrin ring is probably small. If the transitions giving rise to porphyrin spectra are of the types shown in Fig. 1 of our paper (based on Longuet- Higgins, Rector and Piatt, 1950; Seely, 1957; and Kuhn, 1959; see Falk and Perrin, p. 56, for references), what effects have the change from a high-spin to a low-spin ferrous complex on these transitions? The important thing to remember, and what distinguishes low-spin ferrous complexes from low-spin ferric complexes, is that in the former the dx^_y''- electron of the high-spin state has gone into the d^y orbital, whereas in the latter it is in the dy, or d^^ orbital. In ferrous complexes this makes the Soret transition more difficult because in the excited state there is increased electro- static repulsion with electrons on the methene carbon atoms. This is diminished by back-double-bonding whh ligands in the 5th and 6th co-ordination positions and this might be expected to displace Amax to longer wavelengths. On the other hand, for the visible bands in low-spin ferrous complexes there is decreased electrostatic repulsion in the excited state but this effect is reduced by back-double-bonding, so here Amax moves to shorter wavelengths the greater the back-double-bonding. If the electronic transition which gives rise to the Soret band does not involve a net movement of electronic charge away from the metal in the high-spin ferric and ferrous complexes (which have symmetrical d electron distributions in the plane of the porphyrin) Amax would be expected to be at shorter wavelengths for ferric than for ferrous, irrespective of whether electron density on the pyrrole nitrogens is decreased as suggested by Williams or increased as suggested by Falk and myself. In the ferric haemoproteins the Soret band shifts to longer wavelengths in the low- spin complexes. This is in line with expectation. The change from high-spin to low-spin includes taking an electron from d^.-^^y"- and putting it in dy,, d^^ (or possibly dxy): this makes it easier to put more 7r-electron density on the pyrrole-nitrogens in the excited state and hence lowers the energy needed for the transition if it is of the form Falk and I suggest in our Fig. 1 . The opposing effect of increasing the electrons in d^y, dy, or d,j. would be diminished where back-double-bonding is possible and, in fact, one might expect, other things being equal, that the longest wavelengths for Soret maxima will be given by the best back-double-bonding ligands. Williams : As far as I can see Falk and Perrin used the same theory as I do, that of Piatt, in the discussion of the spectra of porphyrins (see Chem. Rev. 56, 299, 1956). There is then the question of how metals affect the spectra. I appear to agree with the treat- ment given recently by Gouterman (J. chem. Phys. 30, 1139, 1959), and which is the best I know, while Falk and Perrin's discussion would say something different. I say that the Soret band shifts to longer wavelengths with decrease of effective electro- negativity of the central metal. From what Perrin says I feel he must conclude that the opposite is true. Lemberg : I am not convinced that there is a general relationship between the Soret band positions and the spin type. The Soret band of high-spin ferrohaemoglobin lies at 74 Discussion longer wavelengths than that of the low-spin pyridine protoferrohaemochrome, but that of high-spin protohaem lies at still shorter wavelengths. Nor does there seem a clear relationship between bond-type and autoxidizability. We find autoxidizable haem compounds both with high-spin (haem) and low-spin (pyridine and imidazole haemochromes), and compounds slowly or not at all autoxidizable also both with high-spin (haemoglobin) and low-spin (cytochrome c). Williams: I have discussed above the shift in Soret band with ligand (Fig. 1, p. 43). Only for haemoproteins will the Soret band shift to shorter wavelengths (with change of spin-type) as ligand basicity increases. For the change haem to haemoprotein (haemoglobin) there is no change of spin-type and the Soret band moves to longer wavelength with basicity. This is also likely to be true for all low-spin complexes. I believe bond-type and autoxidizability to be clearly related. I wonder whether the compounds (pyridine and imidazole haemochromes), which Lemberg speaks of as low-spin, are in equilibrium with amounts of dissociated complexes (high-spin). It does seem that imidazole and pyridine reduce the rate of autoxidation of haems; thus imidazole and pyridine complexes of haems in strong solution pick up oxygen reversibly without being autoxidized (Corwin and Bruck, /. Amer. chem. Soc. 80, 4736, 1958). I think oxidation occurs only on addition of water in this case. Models for Haemoproteins Some New Compounds of Haems with Bases By J. E. Falk (Canberra) Falk : The properties of some new complexes of meso-, proto- and 2 : 4-diformyldeutero- haems are indicated in Figs. 1 and 2 and Table 1. These complexes were made in 0-01 N NaOH. The ligands hydrazine (pA'a8-l), ethanolamine (9-5), n-propylamine (10-5) and dimethylamine (10-6) form an interesting series of compounds with these haems. As shown in Fig. 1 both Amax and cmax of the meso- and proto-haem com- plexes increase with pKa of the ligand, and the effects are very similar for both haems. With the diformyl haem, the e of the dimethylamine complex is sim.ilar to that found with meso- and proto-haems. It was not possible to measure A values with the other amines because of spectral shifts associated with Schiff's base formation between the ligands and the formyl side-chains. This reaction was slow enough, however, to permit measurement with the reversion spectroscope of the position of the absorption bands of the initial complexes, and as shown in Table 1, the a-bands of all these complexes, from hydrazine to dimethylamine, were at the same wavelength as the a-band of the pyridine haemochrome. This lack of influence of the ligands upon the spectrum, as com.pared with meso- and proto-haems, is apparently related to the much greater electron-attraction in the side-chains of diformyldeuterohaem. We were able to obtain interesting complexes of protohaem and the diformylhaem with 2-mercaptoethanol (Table 1 and Fig. 2) ; the a-band of the protohaem complex is displaced 17 m/« to longer wavelength than that of pyridine protohaemochrome — the greatest shift we have yet encountered. Ethanol did not react in the same way under these conditions, so that as a first presumption, it appears that the thiol-group of mercaptoethanol is complexing with the haem iron. The electrophilic side-chains of proto- and diformyl haems appear to have some influence, since mercaptoethanol reacted poorly with mesohaem (Fig. 2). The Srnu of the pyridine haemochrome of 2 : 4-diformyldeuterohaem at the a maxi- mum was assumed to be the same as that of protohaem. As shown in Table 1 , the a-bands of these compounds of meso- and proto-haems more than cover the ranges of a-band positions (cf. Morton, Rev. pure appl. Chem. 8, 161, 1958) of the known cytochromes of types c and b respectively. Absorption spectra were determined with a Beckman DU Spectrophotometer, except compounds marked *, which were measured with a Beck-Hartridge reversion spectro- scope (see text). The fmM=31 for the pyridine haemochrome of 2: 4-diformyl- deuterohaem is assumed, and the £niM for the other ligands calculated on this basis. 1 1 Protohaem 4 A / '. t; 1 ■'/ ^\ r\ 1 / \\ iA 2\1 I " 2 /, \ 1^^ T > 1 ^ J ^ ^ 1 1 1 ■■ 1 1 2 4-Diformyl deuterohaem '\ / p 1 1 4 i \, I P 510 530 550 520 540 560 A, m// 540 560 580 600 X, xw/x Fig. 1, Spectra of compounds of haems with bases. The haems (2 X 10~^m) were dissolved in 0-01 n NaOH; samples were reduced with dithionite imme- diately before adding the ligands and measuring the absorption spectra on the Beckman DU Spectrophotometer. Ligands were added in excess, higher concentrations causing no further change in absorption, in the following molarities : P. pyridine 1. hydrazine 2. ethanolamine 3. «-propylamine 4. dimethylamine Mesohaem 3-41 0-488 0-655 0-282 15-06 Proto- and 2 : 4-diformyl haems 3-41 0-195 0-301 0-159 15-06 HK) Mesohoem 1 P 30 A \ \ lac 10 1 1 \ \ \ 1 \ \ \ \ / / / / / / I \ \ ME ^^^ ^s S \ Pr otoha am P / \ / / \ \ ME A / J \ 1 \ 1 ^\ P \ \ \ ME / ^ K / \ \ ^ -^ E 510 530 550 520 540 560 580 A, m// A, m^ Fig. 2. Spectra of mercaptoethanol-haem compounds. The haem solutions and the procedure were as described in Fig. 1, The ligand concentrations were(M): pyridine (P), 3-41 ; ethanol (E), 8-5; mercaptoethanol (ME), 3-45. H.E. — VOL. I — G 76 Discussion Table 1. a-BANDS of some haem complexes 2:4-Di- Ligand Meso A (rcifi) fmM Proto A (m/<) ^mM formyldeutero A (m/i) f mil Pyridine 547 33-2 558 31 584 31 pKa 8-1 Hydrazine 545 17-2 556 17-3 584* 9-5 Ethanolamine 546 24-2 557 22-2 584* 10-5 rt-Propylamine 550 35-7 562 28-0 584* 10-6 Dimethylamine 550 38-8 563 37 584 36-4 2-Mercaptoethanol 560 9-8 575 25-7 595 21-2 Lemberg: There is considerable difficulty in finding suitable models for haemoproteins among simpler haem compounds. Thus amino acids at neutral pH are zwitterions and therefore unsuitable. Research with amino acids at a physiological pH should therefore be carried out with amino acid esters or poly-aminoacids, but such data are missing. Moreover, even then the affinity to haem may be far smaller than if the combining group is held in the protein in a restrained position close to the haem iron. In this regard Kaziro's approach offers perhaps more hope than model experiments. I was interested to note, in Falk's discussion (this volume, p. 74), the lack of vari- ability of the diformyldeuterohaem spectrum with different ligands. This is quite different from the great variability of band position of nitrogenous compounds of monoformyl haems such as haem a. Falk: Lemberg reinforces the point I have made — that relatively few model haem com- pounds have been studied; those which have been are largely of one class, in which the ligand atom is — N^. The reasons for this are partly the difficulties, as Lemberg points out — we are, in fact, studying amino acid esters and small peptides — and partly, perhaps, that we have all been over-impressed with the suspected role of histidine, and have modelled our models upon it. I agree that model studies can never entirely solve the problems of haemoprotein structure, but I believe that there still remains a great deal to discover about the chemistry of the combination of haem with new types of ligands, and that knowledge on this level will be a prerequisite for our eventual understanding of the haemoproteins. Not only do we need data on compounds of haems with primary and secondary nitrogen atoms, with thiols, etc., but as Perrin and I have pointed out, there is a par- ticular need for studies of mixed compounds, with one ligand of these types and the other of the — N== type. In this regard Wang's model (this volume, p. 98), in which the haem is held down on one side to a — N= bond, should be particularly valuable. Phillips: It is interesting to note that although free amino acids form haemochromes only with great difficulty, the corresponding amino acid esters react readily. The reluctance of the free amino acids to react is not simply due to electrostatic repulsion betv^een the amino acid anion and the carboxylate anions on the porphyrin side chains, since a similar behaviour is observed in detergent solutions using porphyrin esters. Carbon Monoxide-Pyridine Complexes with Haems By J. H. Wang (Yale) Wang: In connexion with Falk's remarks on the special significance of mixed haemo- chromes, I should like to discuss some interesting results obtained in our equilibrium studies on the combination of carbon monoxide with haem in aqueous solutions containing small amounts of pyridine. Spectra and Redox Potentials of Metalloporphyrins and Haemoproteins 77 Our results show that as the concentration of pyridine in the solution increases, the affinity of haem for carbon monoxide also rapidly increases (Nakahara and Wang, J. Anier. chem. Soc. 80, 6526, 1958). This observation shows that in dilute solutions the mixed complex pyridine-haem-CO has greater thermodynamic stability than both the complex HaO-haem-CO and the complex pyridine-haem-pyridine. Indeed the affinity of haem for carbon monoxide is so high that when the latter is bubbled through a solution of dipyridine haemochrome in pure pyridine, some mixed complex pyridine-haem-CO is formed. On the other hand no detectable amount of dicarbon- monoxyhaem, OC-haem-CO, was found when aqueous haem solutions, whether with or without added pyridine, were equilibrated with carbon monoxide at even 1 atm. pressure (Wang, Nakahara and Fleischer, /. Amer. chem. Soc. 80, 1109, 1958). O III 12 II 9 I c :Fe' 810 :Fe: -I,^Z *■ Fe r 49 671 P) ^ Identical J'^P^S^ ligands (c) i) Fig. 5. Visible spectra of ferrimyoglobin, ferrihaemoglobin and ferriperoxidase fluoride (Keilin and Hartree, 1951; Hanania, 1953). 12 10 mM 8 6 4 2 650 + + + Fe CN /'^X />" x\ / A J \ ^ yf \ A^ // y ^Per. - 1^/ Hb^ ..— y^ '^ 600 550 500 A(mjj) Fig. 6. Visible spectra of ferrimyoglobin, ferrihaemoglobin and ferriperoxidase cyanide (Keilin and Hartree, 1951 ; Hanania, 1953). 114 P. George, J. Beetlestone and J. S. Griffith band at about 500 m.ii ; and all the low-spin ferric complexes have spectra like the cyanide derivatives, with a very pronounced absorption band at about 540 m/j, and a shoulder, or second band, at about 580 m/« (Theorell, 1942). The high-spin complexes have additional minor bands of lower intensity, at about 580 and 540 m/<, but for the present it is the positions of 1.2 1.0 'mM 0.8 0.6 0.4 0.2 + + + ■ ^^f + + + Mb / \ / N / \ 900 800 ;\(m;j) 700 Fig. 7. Near infra-red spectra of ferrimyoglobin fluoride, hydroxide and cyanide, and ferrihaemoglobin hydroxide (Hanania, 1953). the major bands which differentiate the two types of complex that are important. Similar contrasting features appear in other regions of the absorption spectrum. In the near infra-red the fluoride complex has a v/ell-defined absorption band at about 850 m/( with a shoulder at about 750 m//, whereas the cyanide complex has remarkably low absorption throughout the whole range 700 to 950 m// as shown in Fig. 7 (George and Hanania, 1955). In the ultra-violet, from 280 to 450 m/<, there are three regions to consider. The very intense Soret band lies between 405 and 410 m/t for the acidic ferri- haemoproteins and the fluoride complexes, the latter having lower absorption in the case of myoglobin and haemoglobin but higher in the case of peroxidase. On the other hand, the low-spin derivatives have the band shifted towards the red in the neighbourhood of 418 to 425 m^a (see Fig. 8). Minor bands occur at about 350 m/u. These are unresolved in the case of the acidic ferri- haemoproteins and the fluoride complexes, but two distinct bands at about Fenihaemoprotein Hydroxides 115 345 and 360 mfi can be distinguished in the case of the cyanide complex. At shorter wavelengths, from 260 to 300 m/i, absorption due to both the ferri- porphyrin prosthetic group and tyrosine and tryptophane residues in the protein occurs, as evidenced by the greater absorption of the ferrihaemo- proteins as compared to their apo-proteins. As shown in Fig. 9 the low-spin 450 400 350 300 7\(m>i) Fig. 8. Ultra-violet spectra of ferrimyoglobin and ferrihaemoglobin cyanide and fluoride (Keilin and Hartree, 1951; Hanania, 1953). cyanide derivative has greater absorption than the high-spin fluoride deriva- tive throughout this region, although the band at 290 m/( is less well resolved. While the spectra of the high- and low-spin derivatives exhibit these characteristic distinguishing features, which as far as can be judged are common to myoglobin, haemoglobin and peroxidase, the spectra of the hydroxides vary a great deal as shown in Figs. 7, 10, 1 1 and 14. Moreover these variations are not haphazard, but appear to be related to the change in magnetic moment, i.e. FerriMb -^ FerriHb -> FerriPer (5) 5-11 B.M. 4-45 B.M. 2-66 B.M. To take but one example, in the region of 600 m/<, the extinction coefficients follow the order as shown in Fig. 10, 'Mb > £Hb > e Per (6) 116 P, George, J. Beetlestone and J. S. Griffith 50i — 10 300 250 A(nnjj) Fig. 9. Ultra-violet spectra of ferrimyoglobin fluoride, hydroxide and cyanide in the region of tyrosine and tryptophane absorption (Hanania, 1953). 'mM 650 600 550 500 7\imp) Fig. 10. Visible spectra of ferrimyoglobin, ferrihaemoglobin and ferriperoxidase hydroxide (Keilin and Hartree, 1951 ; Hanania, 1953). Fenihaemoprotein Hydroxides 117 Now the regular and systematic differences between the spectra of high- and low-spin complexes suggest very strongly that if the hydroxides are mixtures of high- and low-spin forms their spectra and magnetic moments should conform to a certain pattern. (a) For the same haemoprotein, the extinction coefficients for the hydroxide should be intermediate in value between those for typical high- and low-spin complexes in the regions where the major absorption bands occur. (b) For a series of hydroxides, there should be a regular trend in the extinc- tion coefficients in the region of the major absorption bands, such that the 450 400 350 300 A(mjj) Fig. 11. Ultra-violet spectra of ferrimyoglobin and ferrihaemoglobin hydroxide (Hanania, 1953). higher magnetic moment hydroxides resemble more closely the high-spin complexes, and the lower magnetic moment hydroxides resemble more closely the low-spin complexes. In the case of ferrimyoglobin, the only haemoprotein for which complete data are available at present, the first criterion is found to hold throughout the entire range of wavelength, 250 to 950 m/<. For the visible region the myoglobin curve in Fig. 10 is to be compared with those in Figs. 5 and 6; Fig. 7 covers the region 700 to 950 m/t ; Figs. 8 and 1 1 give the Soret bands, and the smaller bands in the region 330 to 370 m/i ; and Fig. 9 covers the region of composite absorption, 250 to 300 m//. The second criterion is borne out by a comparison of the spectra of ferrimyoglobin and ferrihaemoglobin hy- droxides in Figs. 7 and 10, where the extinction coefficients follow the sequence Fluoride Complex -» FerriMb Hydroxide -* FerriHb Hydroxide -> Cyanide Complex (7) high spin 5-11 B.M. 4-47 B.M. low spin 118 P. George, J. Beetlestone and J. S. Griffith in order either of increasing or decreasing magnitudes, depending on the particular wavelength. In the ultra-violet region, 330 to 450 m//, the trend is not so clear-cut, but, as will be shown in the next section, this can be attributed to the small but significant shift of all the ferrimyoglobin band maxima relative to those for ferrihaemoglobin, together with systematically 300 250 A(rTVj) Fig. 12. Ultra-violet spectra of ferrimyoglobin and ferrihaemoglobin hydroxide in the region of tyrosine and tryptophane absorption (Hanania, 1953). lower extinction coefficients (see Fig. 8). In the region 250 to 300 mft no strict evaluation is possible because myoglobin and haemoglobin are not alike in tyrosine and tryptophane content. Nevertheless it is interesting that the curve for ferrimyoglobin hydroxide, which has higher moment, in contrast to that for ferrihaemoglobin with the lower moment, has a well-defined shoulder at 290 m/u Hke the high-spin fluoride complex (see Figs. 9 and 12). The data for ferriperoxidase are not quite suflficient for it to be included in the sequence in equation (7), although there are ample indications that it would fit into the pattern and come between ferrihaemoglobin hydroxide and the low-spin cyanide complex. The magnetic moment has been determined for horseradish peroxidase, 2-66 B.M. (Theorell, 1942), but the absorption spectrum, recorded by Keilin and Hartree (1951) and reproduced in Fig. 10, refers to a pH of 11-4, which, judging from the pK of 10-9-1 1-3, would give Ferrihaemoprotein Hydroxides 1 19 only about 60-75 % hydroxide formation. It is already evident from Fig. 10, however, that the hydroxide has a pronounced peak at about 540 mn with a second peak at about 575 ra^i, and no peaks either in the region 600 to 640 m/i or at about 500 m,a. Spectroscopically, as well as magnetically, it can safely be classified as a low-spin complex. The spectroscopic type is fully substantiated by the corresponding spectrum for Japanese root peroxi- dase (Morita and Kameda, 1958), which has absorption bands at 548 and 578 ma, with £niM = 12-3 and lOT respectively, together with relatively lower absorption in the region 620 to 650 m//, compared to horseradish peroxidase in Fig. 10. But its magnetic moment has not yet been measured. The data for two other haemoglobins may be considered at this point. The first, Chironomus haemoglobin (Scheler and Fischbach, 1958), presents some anomalous features. The visible spectrum of the hydroxide is most like that of ferrimyoglobin, and the shape of the curve in the region of 600 m,a suggests that it should come between ferrimyoglobin and ferrihaeraoglobin in the sequence in equation (7), but on a quantitative scale nearer the former. However its magnetic moment is 4-45 B.M., a little less than that of ferri- haemoglobin hydroxide (Scheler, Schoffa and Jung, 1957). No explanation can be offered for this discrepancy, although it is to be noted that the magnetic moment of the acidic ferrihaemoglobin is appreciably lower than the values determined for ferrimyoglobin and erythrocyte ferrihaemoglobin, namely 5-68 and 5-80 B.M. respectively (Theorell and Ehrenberg, 1951; Coryell, Stitt and Pauling, 1937). The second, root nodule haemoglobin (leghaemoglobin) is particularly interesting. The visible spectrum of the hydroxide, reproduced in Fig. 16 (Sternberg and Virtanen, 1952), is almost the same as that of Japanese root peroxidase, which indicates that it is a low-spin complex. Furthermore, preliminary spectroscopic observations by George, Hanania and Thorogood (1959) in the near infra-red have shown it to have significantly lower absorp- tion in the region 700 to 900 m,a than the myoglobin and haemoglobin derivatives, which is in keeping with the trend in extinction coefficients from high- to low-spin complexes (see Fig. 7). The magnetic moment however still remains to be determined. Hence, provided it is appropriate to regard leghaemoglobin as a true haemoglobin*, the haemoglobins themselves, without recourse to peroxidase, furnish a series of hydroxides covering almost the whole range of spectroscopic characteristics. There is thus a substantial body of evidence to suggest that the hydroxides, especially of ferrimyoglobin and ferrihaemoglobin, are mixtures of high- and * This classification is based on the ability of ferroleghaemoglobin to form an oxygen complex, and it is further substantiated by the reaction of ferrileghaemoglobin with hydrogen peroxide. An intermediate compound is formed with absorption bands at 550 and 575 m//, resembling the ferrimyoglobin and ferrihaemoglobin derivatives, in contrast to ferriperoxidase and ferricatalase, which give two such compounds neither having bands at these wavelengths. 120 P. George, J. Beetlestone and J. S. Griffith low-spin forms, and in the next section this hypothesis will be put to a quantitative test. QUANTITATIVE CORRELATION BETWEEN THE MAGNETIC MOMENTS AND THE SPECTRA OF FERRIHAEMOPROTEIN HYDROXIDES Making the assumption that the hydroxides are mixtures of high- and low- spin forms, the magnetic moments and extinction coefficients at each wave- length should be interrelated in the following way. Denoting the moments of the high- and low-spin forms by /^^ and i^ii, the moments of, say, ferrimyoglobin and ferrihaemoglobin hydroxide, /^ji,, and / / / \ S. / / \ \ n > ^---''''^^^---O^ 450 400 350 300 7^(mjj) Fig. 17. Ultra-violet spectra of the high- and low-spin hydroxides calculated from the data for ferrimyoglobin and ferrihaemoglobin hydroxides. The absorption curve of ferrimyoglobin hydroxide has been corrected by a 5 m/i displacement toward the red, and all extinction coefficients multiplied by 1-08: /i) 700 Fig. 18. Near infra-red spectra of the high- and low-spin hydroxides calculated from the data for ferrimyoglobin and ferrihaemoglobin hydroxide. No A or fniJVl corrections: /tj = 2-24, n^ = 5-92. throughout this region like the cyanide derivative (see Fig. 7). The small negative extinction coefllicients actually obtained for the low-spin form can be attributed to the uncertainties in the calculation procedure when the correction factors are unknown, and also to experimental error. Precise extinction coefficients are very difficult to determine in this region because the magni- tudes are so small, and errors introduced by extraneous background absorp- tion, which is hard to remove completely, become significant. Summary Assuming that the ferrihaemoprotein hydroxides are thermal mixtures, and adopting /Uf^ = 5-92 and /tj = 2-24 as the magnetic moments of the 128 P. George, J. Beetlestone and J. S. Griffith high- and low-spin forms, calculations give the following percentages for the various haemoproteins : Myoglobin : High - 70 %, Low - 30 % Haemoglobin : High - 50 %, Low - 50 % Peroxidase: High- 7%, Low -93% From the extinction coefficients of the hydroxides the spectra of the high- and low-spin forms have been obtained over the range 250 to 950 m^. Major absorption bands, or shoulders, occur at about the following wave- lengths, with extinction coefficients having the approximate values given in brackets : High-spin form: 830 (1-2), 740 (0-9), 600 (11), 540 (8), 490 unresolved band (10), 405 (116), 350 shoulder (44). Low-spin form: 575 (10), 545 (12), 417 (104), 360 (20), 340 (14). The calculation procedure has certain inherent limitations, nevertheless these absorption bands correspond so closely to those which distinguish high- from low-spin derivatives that the assumption of a thermal mixture can be regarded as entirely consistent with the spectroscopic and magnetic data. THE EFFECT OF TEMPERATURE ON THE SPECTRUM AND ON THE MAGNETIC MOMENT OF FERRIMYOGLOBIN HYDROXIDE As suggested previously, if the ferrihaemoprotein hydroxides are thermal mixtures of high- and low-spin forms, then changing the temperature would be expected to influence the equilibrium. High-spin hydroxide ^^ Low-spin hydroxide (12) and a change should therefore be observable in the magnetic moment and in the absorption spectrum. The magnitude of the change would depend on the value of AT/ for reaction (12), since this determines the variation of equili- brium constant with temperature according to the van't Hoff Isochore. But, since a close balance between the energies of the two forms is to be anticipated, A// is likely to be small, and, as a consequence, K^ to have a small temperature dependence resulting in only slight changes in magnetic moment and absorp- tion spectrum. This is borne out by the observation that there were no noticeable variations in the optical densities of ferrimyoglobin or ferri- haemoglobin hydroxide solutions at 582 m// and 578 m/t respectively over the temperature range 7-5° to 37°C in experiments carried out to obtain thermo- dynamic data for the ionization reactions (George and Hanania, 1952, 1953). However, further experiments have now been made using a sensitive recording spectrophotometer, and a temperature effect has been detected. The absorption spectrum of a concentrated solution of ferrimyoglobin Ferrihaemoprotein Hydroxides 129 hydroxide at pH 11-0 and 5°C was recorded from 470 to 650 m,a, the optical density at 540 m/t being 0-7411. The reference cuvette, which previously contained buffer, was then filled with more of the hydroxide, and a baseline was recorded over the wavelength range with both solutions at 5°C. The solution in one cuvette was then rapidly warmed to 35°C, and maintained at +0.02 — Fig. 19. The difference spectrum in the visible region for 8-4 x 10~^ m ferrimyoglobin hydroxide at 5° and 35°, plotted as id^° - ^35°). this temperature, by the insertion of a specially constructed hollow metal heating unit through which water from a thermostat was circulated. A difference spectrum was recorded, from which the curve illustrated in Fig. 19 was obtained after correction for the baseline. As can be seen, the effect of a 30° alteration in temperature is rather small. The change in optical density is at the most 2-9% at 542 m/^, while at 582 m/i it has dropped to 1-6%, which accounts for the effect escaping notice in the earlier investigations. Control experiments using the cyanide and fluoride derivatives showed no similar effect. The negative regions from 480 to 520 m// and above 600 m/<, together with the positive region in between which shows two well-defined bands, are qualitatively consistent with an increase in the fraction of the high-spin form as the temperature is increased. Some indication of its magnitude can be obtained from the difference between the extinction coefficients of the 130 P. George, J. Beetlestone and J. S. Griffith high- and low-spin forms calculated in Section IV. At 540 m/< the difference in e^^ is about 4, which gives an increase of between 0-05 and 0-07, i.e., between 5 and 7 %. However, in order to obtain the individual spectra of the high- and low-spin forms from the difference spectrum, an independent determination of the fractions present at the two temperatures is required. This can be seen from the equations, fg = agSj -}- (1 - a5)£^ (13) £35 = aaaCj + (1 - a35)e;, (14) where £5 and £35 are the extinction coefficients of the hydroxide at 5° and 35°, a5 and cc^^ are the fractions of the low-spin form at the two temperatures, and £f^ and £j are the extinction coefficients of the high- and low-spin forms. Since £5 is known and £35 can be obtained from the difference spectrum, provided cc^ and cc^^ can be determined, jUj^ and jUi can be evaluated from the equations rearranged in the form, ^3b'^5 ~ ^5*^35 "■35 (15) €5(1 - ^35) - ^35(1 - 0^5) .... El = (16) 0^5 - ^35 The variation of a with temperature has been obtained in the following way. Using a sensitive Gouy balance, constructed from a Varian electro- magnet V4004 and a Sartorius Microbalance MPR 5 II, and equipped with a coaxial glass thermostat surrounding the sample tube and suspension fibre, the change in Aw was measured as a function of temperature over the range 1° to 30°C for the fluoride, cyanide and hydroxide derivatives of ferrimyo- globin. Calibration with nickel chloride solution enabled these changes in Aw to be converted into changes in molar susceptibility, Xu- The value of Xu obtained by Theorell and Ehrenberg (1951) for the three derivatives at 20°C were adopted, namely, 14,790, 2,340 and 11,040 x 10-^ c.g.s. units respectively, and hence values of^Xu o^^^ the temperature range were obtained. The variation of Xu for the fluoride and cyanide derivatives was found to follow very closely the simple Curie law, x = constant/r. The magnitude of the change is illustrated by the following data: from 20° to 1°, Xu for the fluoride and cyanide derivatives increases by 1,046 x 10~^ and 82 x 10~^ c.g.s. units respectively. On the other hand, Xu for the hydroxide, although it has a high value approaching that of the fluoride, only increases by 145 X 10~^ c.g.s. units for the same decrease in temperature. As a conse- quence, the values of Xm do not follow the Curie Law, and the type of devia- tion is just what would be expected if, on lowering the temperature, the fraction of the high-spin form decreases. Ferrihaemoprotein Hydroxides 131 The simplest method by which the fractions of the high- and low-spin forms can be calculated, throughout the temperature range, is to use the experi- mental values of x^si for ^^'^ fluoride and cyanide derivatives at various temperatures as the values appropriate to the high- and low-spin forms, and substitute in the equation, yfM(hydroxide) "" °^XM(cyanide) + (^ ^)ZM(fluoride) (17) In practice, this is equivalent to taking // j = 2-34 B.M., i.e., the value for the cyanide derivative, instead of 2-24 B.M., as in the majority of calculations in 1 1 1 p V-^ low 10 fo ^^^ \ y iJyQ~K mM high-W^ / TJ— cr^ ^"""^ \^ J 5 Jo ■ 1 ( / 1 Ferrimvoqlobin Hydroxide 650 600 550 A(mjj) 500 Fig. 20. The visible spectra of the high- and low-spin hydroxides calculated from the difference spectrum in Fig. 19, and the corresponding change in the fraction of the low-spin form, 0055, as obtained from the variation of magnetic suscepti- bility with temperature. Section IV. Such a slight change in Hi only affects the value of a to a negli- gible extent, i.e., from 0-300 to 0-304. Values of a and (1 — a), obtained in this way, are listed in Table 5 for temperatures from 0° to 30°, together with values for K^, the equilibrium constant for the conversion reaction (12). Interpolation and extrapolation for the temperature interval 5° to 35° gives 0-055 for the corresponding change in a. The spectra of the high- and low-spin forms of ferrimyoglobin hydroxide were then obtained by calculating Ey^ and £j throughout the wavelength range according to equations (15) and (16). The similarity between these spectra, shown in Fig. 20, and those in Figs. 13, 14 and 16 is very gratifying. But it must be remembered that the previous spectra, calculated from data for pairs of haemoproteins, are 132 P. George, J. Beetlestone and J. S. Griffith approximations, consisting of average values of the extinction coefficient appropriate to the two individual high-spin forms and the two individual low-spin forms. The new spectra in Fig. 20, based entirely on data for one haemoprotein, are therefore more valid. THE SPECTRA OF FERRIMYOGLOBIN DERIVATIVES IN HEAVY WATER The interplay of structural and electronic factors necessary for a ferri- myoglobin derivative to exist as a mixture of high- and low-spin forms is evidently so critical that the ligands most closely related to the hydroxyl group in chemical type give predominantly, or entirely, high-spin or low-spin complexes. On the basis of spectroscopic data, or magnetic data, or both, it is clear that the complexes with phenol, and presumably ethanol, i.e., Fejib+++ — OCgHg and Fe]^+++ — OC2H5 come in the former category; whereas the complexes with the sulphur analogues, hydrogen sulphide, ethyl mercaptan and thiophenol, i.e., Fejib'^++ — SH, Fe]yii,+++ — SC2H5 and Fejnj+++ — SCgHg, come in the latter (George, Lyster and Beetlestone, 1958; Coryell and Stitt, 1940; Keilin, 1933; Heussenstam and Coryell, 1954). The least drastic of all substitutions that can be achieved, with the exception of employing HaO^^, is the replacement of hydrogen by deuterium, and the spectrum of the alkaline form in heavy water, which should accordingly have the structure Feniij+++ — OD, has therefore been studied. In the prelimi- nary experiments, reported below, the highest mole ratio of DgO to H2O that could be attained was 134: 1. Hence, although the affinities of the iron atom for OH~ and 0D~ also have to be taken into consideration because they determine the relative amounts of Fe]ynj+++ — OH and FejyQj+++ — OD formed, it is unlikely that in pure D2O the effect observed would be very much enhanced. A very concentrated solution of acidic ferrimyoglobin in ordinary water was used, so that only 0-02 ml in a total of 3 ml was required to give optical density values of about 0-7 at the band maxima in the visible region. Solutions of the alkaline form were prepared in the following way. Tiny quantities of caustic soda solution were added to acidic ferrimyoglobin (0-02 ml stock solution -f 2-98 ml ordinary water) from a micro-syringe until the pH was 11-0. The same volume of caustic soda was added to a corresponding solution of acidic ferrimyoglobin made up in heavy water. Difference spectra were then recorded with the heavy water solution in the reference cuvette for the alkaline form, and, as controls, for the acidic form and the cyanide derivative, which was prepared by adding a little solid KCN. With the cyanide derivative no difference could be detected and with the acidic form there was scarcely any change. But with the alkaline form a well- defined difference spectrum was obtained, very similar to that in Fig. 20, and the optical density differences were about the same in magnitude. The Ferrihaemoprotein Hydroxides 133 simplest interpretation of this result is that for the Fej]^+++ — OD formed in heavy water the fraction of the low-spin form is about 6 % higher than the fraction for Fejj^+++ — OH under similar conditions. Using the data given Table 5. The fractions of the low-spin and high-spin forms of ferrimyoglobin hydroxide, a and (1 — a) respectively, at different temperatures, calculated from the temper- ATURE VARIATION OF Xm FOR FERRIMYOGLOBIN HYDROXIDE, FLUORIDE, AND CYANIDE Kg is the equilibrium constant for the conversion high-spin form ^^ low-spin form and is given by a/(l — a). TCO a (1 -a) Ke 0-34 0-66 0-52 10 0-32 0-68 0-47 20 0-30 0-70 0-43 30 0-285 0-715 0-40 in Table 5 for the hydroxide, an approximate value of the equilibrium con- stant for the reaction high-spin deuteroxide ^^ low-spin deuteroxide (18) is found to be 0-55 at 25°C, compared to 0-41 for the hydroxide. Substitution of hydrogen by deuterium thus favours the conversion by about 0-2 kcal/mole in units of free energy. It is probable that the bulk of this difference arises via the water of solvation and not from any effect on the ligand field. The change of mass of the hydrogen nucleus affects the free energy of 'crystalliza- tion' around the iron ion directly but the ligand field only very indirectly through the effect of the change of vibrational amplitudes for the OH group on the mean ligand field. GENERAL REMARKS The experiments with heavy water offer particularly direct evidence for the existence of a thermal mixture in ferrimyoglobin hydroxide. Further, because the fundamental difference between the high- and low-spin form lies in the electronic structure of the iron ion, they show that water molecules (or those protons which are exchangeable with those of water) play an essential part in determining the free energy change. We would naturally guess that the water molecules which are 'crystallized' around the iron ion (and also the hydrogen of the OH~ group) are the ones concerned here. 134 P. George, J. Beetlestone and J. S. Griffith Although this is far from direct evidence for the existence of the hydroxide structure it is at least thoroughly consistent with it. The hypothesis of a thermal mixture is also fully borne out in the case of ferrimyoglobin hydroxide by the temperature variation of its spectrum and magnetic moment as described in Section V; and, in view of the self-consistent results of the calculations using magnetic and spectroscopic data in Section IV, it can be concluded that ferrihaemoglobin hydroxide is also a mixture of high- and low-spin forms. This accounts equally well for its apparently anomalous magnetic moment as the explanation in terms of the electronic configuration with three unpaired electrons, which was shown to be unlikely on theoretical grounds (see Section II). Thermodynamic data for the conversion of the high-spin to the low-spin form can be obtained from the values of K^ for ferrimyoglobin in Table 5. A plot of log Kg against \jT gives A// = — 1-5 ± 0-2 kcal/mole, and from the equation AG° = A// - T^S°, with ^G° equal to 0-5 kcal/mole at 25°C, AS" is found to be — 6-7 ±0-7 e.u. The conversion is thus favoured by the enthalpy change, but is appreciably hindered by the entropy change to such an extent that the resulting free energy change has a small positive value. In other words, with respect to their heats of formation the low-spin form is the more stable, whereas in terms of their entropies the high-spin form is the more stable. The favourable value of A// may be regarded as purely fortuitous, because, although the conversion to the low-spin form implies an increase in the value of A, and hence extra stabilization, pairing energies have also to be taken into consideration and in addition solvent interaction effects may be important (see below). In order to discuss the entropy change accompanying the conversion, it is convenient to distinguish the contribution arising from the degeneracy of the electronic state of the iron in the two forms from the remainder. The following estimate shows that this contribution is unlikely to be more than about — 2 e.u. In the high-spin form the ferric ion has a ground term which is spatially non-degenerate but has a sixfold degeneracy due to the spin 5* = 5/2. The ligand field combined with the spin-orbit coupling lifts the degeneracy into three Kramers doublets. If this splitting is large compared to kT, only one Kramers doublet is occupied and the effective degeneracy of the ferric ion is 2. If it is small then the degeneracy is 6. This means that the entropy associ- ated with the degeneracy of the electronic state of the iron in the high-spin form lies between the two limits of i? log^ 2 = 1-38 e.u. and R log^ 6 = 3.56 e.u. The actual magnitude of the splitting is unknown. If we assume that it may be represented in a spin-Hamiltonian for the ground term with S = 5/2 by the quadratic expression Ferrihaemoprotein Hydroxides 135 then electron resonance measurements show that D can hardly be less than 4°K (Bennett and Ingram, 1956; Griffith, 1956c). With D in these units, the partition function, Z, is given by the equation, IQD 2D 8D Z = 2e sr +2e3r + 2e3r (19) from which the entropy follows from the formula S = 5(/?nogg Z)ldT. For D/r small we deduce S = 3-56 — (28D-i?/9r^). The second term is inappre- ciable (< 0-01) at room temperature for D < 12°K, i.e., an overall splitting of 48 cm~i. Therefore it seems likely, although not certain, that at room temperature this contribution to S is close to 3-56 e.u. In the low-spin form we have a spatial degeneracy of three and a spin degeneracy of two. Here, however, it is probable that the three Kramers doublets have a separation large compared with kT at room temperature (Griffith, 1957) so that the contribution to S from the degeneracy is close to 1-38 e.u. This means a contribution to A^ for the conversion of the high-spin to the low-spin form of 1-38 — 3-56 = —2-18 e.u. If our assumptions are incorrect the numerical value of this contribution will almost certainly be lower. It is much more difficult to obtain any a priori numerical estimate for the remainder of the entropy change, which, using the value obtained in the last paragraph for the degeneracy contribution, is seen to amount to about —5 e.u.* We should expect it to be negative, however, for the following reason. In the high-spin form the overall distribution of the five ^-electrons about the iron has nearly spherical symmetry, thus producing no orientating effect on the environment. On the other hand, in the low-spin form the five electrons are in the three orbitals away from the bond directions, thus im- posing an extra rigidity on the environment of the iron. This would result partly in a more rigid ferrimyoglobin molecule, and partly in a more rigid arrangement of water molecules around the Fe — OH group. Just as A// for the conversion is determined by other energy terms besides the electronic stabilization energy arising from the splitting of the ^-orbitals, so the values of A// for the formation of complexes with different ligands cannot be taken as an accurate indication of the variation in A. From one extreme to the other, however, a rough correlation would be expected, with the high-spin complexes having the less favourable values of AH. This trend, which has also been discussed by Havemann and Haberditzl (1958), is illustrated by the data in Table 6. The values of AS° become progressively more negative from fluoride to cyanide, but they are not amenable to any straightforward correlation because the entropies of the ligands themselves vary so much, with S"" for F~, 0H~ and CN~ having the values —2-5, —2-3 and -|-28 e.u. respectively. Some allowance for this can nevertheless be made * The assumed additivity of entropies is equivalent to a factorization of the partition function, which is probably a good approximation here at room temperature or below. 136 P. George, J. Beetlestone and J. S. Griffith by comparing the differences in partial molal entropies of the complexes and the parent haemoprotein (George, 1956). Table 6. IS.H and A^" values for the formation of ferrimyoglobin FLUORIDE, HYDROXIDE AND CYANIDE: AND THEIR MAGNETIC MOMENTS (GEORGE AND HANANIA, 1952, 1956: THEORELL AND EHRENBERG, 1951) Ligand Aiykcal/mole ^S° e.u. /f B.M. Type of complex F- OH-* CN- -1-5 -7-65 -18-6 + 1-8 -2-6 -24 5-75-5-92 5-11 2-35 high-spin 70% high, 30% low-spin low-spin * See footnote to Table 7. A more rigorous correlation can be sought, if, for the same ligand, data are available for closely related haemoproteins. But if the whole range from high- to low-spin complexes is to be covered, this would clearly be restricted to those ligands capable of giving theimal mixtures in some cases. For example, the data in Table 7 for the various haemoglobin hydroxides show that the increase in the fraction of the low-spin form is accompanied by more negative (i.e., favourable) values of Ai/, which was to be anticipated from the overall trend illustrated in Table 6. Furthennore, with a series of derivatives of this type, where the ligands are identical and the structure of the complex in the immediate neighbourhood of the iron is presumably very similar, the values of AS* can be taken as a true indication of a general trend paralleling the trend in Ai/. As the fraction of the low-spin form increases, IS.S assumes more negative (unfavourable) values, while A// assumes more negative (favourable) values. This trend in AS* is entirely in accord with the entropy change obtained above for the conversion of the high-spin to the low-spin hydroxide in the case of ferrimyoglobin, and it can likewise be associated with a greater structural rigidity in the vicinity of the iron atom of the low-spin form. The T^K values for the ionization of ferriperoxidase and ferricytochrome c are so much higher than those for the haemoglobins (see Table 1) that inevitably either the A// values, or the A^* values, and very probably both, would show marked deviations from the correlation set out in Table 7. This is not unexpected because the acidic forms of these haemoproteins have different structures, and as a consequence the formation of the hydroxide is a different type of chemical reaction. In the case of the haemoglobins, the reactions of the acidic form can be very adequately expressed by the hydrate structure, e.g., Prof. — Fegb"'^++(H20), and the ionization is accordingly the simple dissociation of a proton. Prot.— FeHb+++(H20) ^ Prot.— Fc Hb -OH + H+ (20) Ferrihaemoprotein Hydroxides 137 Table 7. A^ and A5° values for hydroxide formation* Haemoprotein \H kcal/mole ^S° (e.u.) Type of hydroxide Ferrimyoglobin Ferrihaemoglobin Ferrileghaemoglobin -7-65 -9-5 -11-0 -2-6 -7-9 -13-0 70% high, 30% low-spin 50% high, 50% low-spin approaches 100% low-spin * These values have been calculated from the corresponding data for the ionization reaction, and for the ionization of water (A// = -1-1 3-4 kcal/mole and AS^ = —19-2 e.u.). The references for ferrimyoglobin and ferrihaemoglobin are given in Table 1 ; for ferrileghaemoglobin, see George, Hanania and Thorogood (1959). On the other hand, in the acidic form of ferricytochrome c the iron is bonded in an intricate crevice structure to nitrogenous base groups of tlie protein at both the fifth and sixth co-ordination positions. One of the crevice bonds must be broken if 0H~ is to replace one of the groups, and the ionization reaction therefore takes the form, Prot. - Fe+++cyt.c - N(base) + H2O v^ Prot. - Fe++cyt.c - OH N(base) + H+ (21) With ferriperoxidase the nature of the reaction is rather more obscure, because in less alkaline solution the pH variations of the equilibrium con- stants for complex formation with cyanide, fluoride, azide, etc., differ systema- tically from the corresponding variations for ferrimyoglobin, ferrihaemo- globin and ferricytochrome c. The difference lies in the consumption of a proton accompanying the formation of the complex. George and Lyster (1958) have discussed various explanations that have been advanced, and suggested as a further possibility that acidic ferriperoxidase also has a crevice structure but with the labile bond, which is broken in complex formation and upon ionization, to some group other than a nitrogenous base, with a ^K of about 10 in horseradish peroxidase. Whatever the true explanation may be, it is clear that no strict comparison can be made between thermodynamic data for the ionization of ferriperoxidase and ferricytochrome c and the corresponding data for the haemoglobins. Finally, the question naturally arises as to whether any other haemo- protein derivatives are mixtures of high- and low-spin forms, Scheler, Schoffa and Jung (1957) and Havemann and Haberditzl (1958) have suggested that this may be the case for several other derivatives where the magnetic moments differ appreciably from the usual high or low values. However, no quantitative correlation of magnetic and spectroscopic properties, of the 138 P. George, J. Beetlestone and J. S. Griffith kind used in Section IV to calculate the individual spectra of the high- and low-spin hydroxides, was considered. Since azide gives a high-spin complex with ferricatalase and a low-spin complex with ferrihaemoglobin, it is quite likely that the ferrimyoglobin derivative would contain a significant fraction of the high-spin form. Preliminary calculations tend to confirm this. Never- theless, until temperature variations of the magnetic moment and absorption spectra have furnished direct experimental evidence for the existence of a thermal mixture, as in the case of ferrimyoglobin hydroxide, it is perhaps better to leave it as an open question with regard to the other derivatives. Acknowledgements The work reported above forms part of a research programme on haemo- proteins supported by grants from the National Science Foundation (G2309 and G7657). We wish to thank Dr. Britton Chance for making the facilities of the Department of Biophysics and Physical Biochemistry available to carry out the spectrophotometric study described in Section V. REFERENCES Basolo, F. & Pearson, R. G. (1958). Mechanisms of Inorganic Reactions — A Study of Metal Complexes in Solution, John Wiley, New York. Bennett, J. E. & Ingram, D. J. E. (1956). Nature, Lond. Ill, 275. Coryell, C. D. & Stitt, F. (1940). J. Amer. chem. Soc. 62, 2942. Coryell, C. D., Stitt, F. & Pauling, L. (1937). /. Amer. chem. Soc. 59, 633. Deutsch, H. F. & Ehrenberg, A. (1952). Acta chem. Scand. 6, 1522. Gamgee, a. (1868). Phil. Trans. {London) 158, 589. George, P. (1956). Currents in Biochemical Research, p. 338 (Ed. by D. E. Green), Interscience, New York. George, P. & Hanania, G. I. H. (1952). Biochem. J. 52, 517. George, P. & Hanania, G. I. H. (1953). Biochem. J. 55, 236. George, P. & Hanania, G. I. H. (1955). Disc. Faraday Soc. 20, 293. George, P. & Hanania, G. I. H. (1956). Currents in Biochemical Research, especially Table 5, p. 353. George, P., Hanania, G. I. H. & Thorogood, E. (1959). Unpublished results. George, P. & Lyster, R. L. J. (1958). Proc. nat. Acad. Sci. Wash. 44, 1013. George, P., Lyster, R. L. J. &. Beetlestone, J. (1958). Nature, Lond. 181, 1534. Gibson, J. F. & Ingram, D. J. E. (1957). Nature, Lond. 80, 29. Grifhth, J. S. (1956a). J. inorg. nuclear Chem. 2, 229. Griffith, J. S. (1956b). J. inorg. nuclear Chem. 2, 1. Griffith, J. S. (1956c). Proc. Roy. Soc. A 235, 23. Griffith, J. S. (1957). Nature, Lond. 180, 30. Griffith, J. S. (1958). Biochim. biophys. Acta, 28, 439. Griffith, J. S. & Orgel, L. E. (1957). Quart. Rev. 11, 381. Hanania, G. I. H. (1953). Ph.D. Thesis, The University of Cambridge, England. Havemann, R. & Haberdftzl, W. (1958). Z.phys. Chem. 209, 135. Heussenstam, p. &. Coryell, C. D. (1954). See Coryell, C. D., Chemical Specificity in Biological Interactions (Chap. 8, p. 108 and 113), Academic Press Inc., New York, 1954. Howard, J. B. (1935). J. chem. Phys. 3, 813. Keilin, D. (1933). Proc. Roy. Soc. B. 113, 393. Keilin, D. & Hartree, E. F. (1951). Biochem. J. 49, 88. KoTANi, M. (1949). J. phys. Soc. Japan 4, 293. Fenihaemoprotein Hydroxides 1 39 MoRiTA, Y. & Kameda, K. (1958). Mem. res. Ins fit. Food Science, Kyoto, Japan, No. 14,61. Orgel, L. E. (1955). J. chem. Phys. 23, 1819. ScHELER, W. & FiscHBACH, I. (1958). Acta Biol. Med. Germ. 1, 194, ScHELER, W., SCHOFFA, G. & JuNG, F. (1957). Biocliem. Z. 329, 232. Sternberg, H. & Virtanen, A. I. (1952). Acta. chem. Scand. 6, 1342. Taube, H. (1952). Cliem. Rev. 50, 69. Theorell, H. (1941). J. Amer. chem. Soc. 63, 1820. Theorell, H. (1942). Ark. Kemi, Min. Geol. 16A, No. 3, 1. Theorell, H. & Akesson, A. (1941). /. Amer. ciiem. Soc. 63, 1812. Theorell, H. & Ehrenberg, A. (1951). Acta chem. Scand. 5, 823. DISCUSSION Spin States and Spectra of Haeryioproteins The Electronic Origins of the Spectra By P. George and J. S. Griffith (Philadelphia) George : It is perhaps desirable to say something about the electronic origins of the spectra of the pure high-spin and low-spin compounds (ferrous and ferric), although we have not yet made a detailed analysis of them. Considering first the iron-porphyrin group we may divide the possible electronic transitions into three categories: porphyrin transitions, metal transitions and charge- transfer transitions. Free porphyrin has strong absorption in the visible and also a Soret peak and the intensity associated with these cannot be lost in the metal com- pound. Therefore one naturally supposes the Soret band of the latter and some at least of its visible absorption to be porphyrin transitions. These porphyrin transitions have the characteristic that the electric vector of the light lies in the porphyrin plane so that when it is at right-angles to it the light does not get absorbed. This is not necessarily true of the other transitions discussed later. In thinking about the part of the spectrum which arises from the porphyrin ring one should also remember that the singlet-triplet transitions may enhance their intensity considerably through coupling with the metal ion when the latter has non-zero spin. The metal transitions in the visible and infra-red are d-d transitions which would be of low intensity and probably completely masked by the porphyrin bands. At least they can hardly be responsible for the gross visible structure. The metal 3d-4p transi- tions would probably be in the ultra-violet although it is possible that the 4pz orbital might have its energy lowered sufficiently by interaction with a porphyrin tt orbital to invalidate this view. If this were so, however, the transition would also involve charge-transfer to the ring and so be partly included in our third category. Charge-transfer transitions are of two kinds — to and from the metal ion. Naturally we expect the low energy ones to be to the metal ion for ferric compounds, and from the metal ion for ferrous compounds. In each case, then, they would involve the iron atom commuting between the ferrous and the ferric state. It is natural to suppose that the infra-red bands characteristic of high-spin ferric compounds, oxy-haemoglobin and myoglobin, and the single-equivalent higher oxidation states, are indeed charge- transfer bands. Some components of such transitions are of course fully allowed for electric dipole radiation, and can therefore account for the relatively high intensities. The ligands in the fifth and sixth positions can also have transitions and give charge transfer to and from the iron, and so in a particular compound one or more bands may arise which have no counterpart in other compounds. It would be quite feasible, for example, that the infra-red band of oxyhaemoglobin might be of this type: it could be a transition from a weakly bonding to a weakly antibonding orbital embracing the ferrous ion and the oxygen molecule. We have for simplicity deliberately treated the system as if it can be broken up in a unique and well-defined manner into a number of pieces. This is not strictly H.E. — VOL. I — L 140 Discussion allowable and it is important to remember that interaction may occur among the various types of excited state, resulting in shifts of their positions and donations of intensity from one transition to another. Williams: In his introduction George has stated that the work he described on the equilibrium between two spin states was initiated in 1956. I do not wish to claim any priority in the discussion of the equilibrium between two spin states as it has been mentioned by a large number of authors since 1944 and extensive reference has been made to the factors controlling this equilibrium both by myself and others. In par- ticular, however, I would refer to a three cornered discussion between myself, George and Griffith, in discussions of the Faraday Society, 1955, where I was the only one to maintain this point of view. I have held this point of view consistently since 1953 in discussing the very compounds now examined by George and co-workers. The general idea that equilibrium exists between spin states has had excellent experimental foundation in the study of model complexes for some years past. I should add that I do not wish to detract from the importance of George's contribution which sets my discussion, and that of other earlier authors, on a firm basis. Falk's contribution (see Orgel's paper and discussion, this volume, p. 15) should be read in the light of a summary of my views in Lardy and Myrback, The Enzymes, 1959. There, as previously, I elaborate on spin state equilibria in the cytochromes. Boardman: In their paper, George, Beetlestone and Griflith (this volume, p. 126) correct their calculations for differences in the milliinolar extinction coefficients of the corres- ponding ferrihaemoglobin and ferrimyoglobin derivatives. I feel that these corrections are unnecessary as the extinction coefficients for the myoglobin derivatives appear to be too low by a factor of approximately 1-08. The extinction coefficients for the myoglobin derivatives are taken from the work of Hanania, who assumed a molecular weight of 17,000 for myoglobin whereas now there is evidence to suggest that the molecular weight of myoglobin is above 18,000. A few years ago, Adair and I at Cambridge succeeded in isolating two CO-myo- globins from horse-heart extracts by means of ammonium sulphate fractionation and chromatography on columns of Amberlite IRC-50. The main myoglobin fraction accounted for 90% of the total myoglobin. The molecular weight of the main com- ponent was determined from measurements of osmotic pressure and a figure of 18,400 was obtained. The extinction coefficient of the ferrimyoglobin cyanide derivative at 542 vnn was 0-613 for a 0-1 % solution and this corresponds to a millimolar extinction coefficient of 11-3, if we assume a molecular weight of 18,400. Fig. 6 of the paper by George et al. (this volume, p. 113) shows a millimolar extinction maximum of 10-6 for ferrimyoglobin cyanide. A value of 11-3 agrees closely with the corresponding value for ferrihaem.oglobin cyanide as determined by Drabkin. Theorell and Akeson have concluded also that the molecular weight of horse myoglobin is above 18,000. Their preparation was purified electrophoretically. Three myoglobin components were obtained and the iron content of the main component was 0-297 %. This figure gives a molecular weight of 18,800. George : I would Uke to thank Williams for drawing attention to the early suggestion of Willis and Mellor (/. Amer. chem. Soc. 69, 1237, 1947) that some co-ordination compounds may be thermal mixtures of high- and low-spin forms, and to his own remarks on the subject in the 1955 Faraday Discussion. I agree with Boardman's comments on extinction coefficients. The spectrophoto- metric data employed in the calculations were obtained before the electrophoretic separation of myoglobin into a major and two minor components had been demon- strated. The sample used had been subjected to repeated recrystallizations from ammonium sulphate and treated with strong phosphate buffer, pH 5-7, to remove haemoglobin. For a more exact analysis of the type described in Section IV of our paper it would not only be necessary to have extinction coefficients but also magnetic susceptibility measurements for single components. But an analysis of this kind only gives average spectra of the high-spin form and of the low-spin form for the pair of haemoproteins upon which the calculations are based. Undoubtedly there will be minor variations Ferrihaetnoprotein Hydroxides 141 from one haemoprotein to another as there are for other complexes, e.g. the cyanides and the fluorides. We regard these analyses rather as semiquantitative evidence for the existence of a thermal mixture, in that the calculated band maxima for the two forms occur at appropriate wavelengths with extinction coefficients of the right magni- tude. The spectra of the two forms calculated from the temperature dependence of the spectrum and magnetic susceptibility of a single haemoprotein derivative, e.g. ferrimyoglobin hydroxide in Section V, are however free from this uncertainty. Lemberg: Applying similar spectrophotometric methods to those used by George, Beetlestone and Griffith, we find that in contrast to horseradish peroxidase itself, the compound of haematin a with the apoprotein of horseradish peroxidase has the most 'high-spin type' of spectrum of the haemoproteins a which we studied. There is thus an interesting difference between two compounds of the same protein with the two diff"erent haematins, haematin a and protohaematin. O'Hagan : Preliminary results suggest that aetiohaematin does not attach to apomyoglobin at pH values higher than about pH 8. Holden and Hicks {Aiist. J. exp. Biol. med. Sci. 10, 219, 1932) considered alkaline ferrihaemoglobin to be a mixture of a compound in which linkage was not through the iron atom, and globin ferrihaemochrome. My results would appear to confirm this view and would seem to have some bearing on these results of George. ANALYSIS AND INTERPRETATION OF ABSORPTION SPECTRA OF HAEMIN CHROMOPROTEINS By David L. Drabkin Department of Biochemistry, Graduate School of Medicine, University of Pennsylvania, Pennsylvania The molecular spectra of various common haemin chromoproteins, their derivatives, and such related compounds as the haemochromes (nitrogenous ferro- and ferriporphyrins) exhibit selective absorption over the broad spectral range of 1000 to 200 m/t. The spectrum may be conveniently subdivided into four regions, in which individual maxima have a 500-fold difference in density (see Fig. 1). The most frequently examined region is the visible (2 in Fig. 1), the location of the a and /5 bands, as they have been designated historically. The selective absorption in this region is very different for the different haemin chromoproteins and some of their derivatives, whereas the absorption in regions 3 and 4 (Fig. 1), respectively the location of the y band (Soret, 1878, 1883a and b; Grabe, 1892) and the ultra-violet, is more generally similar for the different chromoproteins. Although the differences in absorption, as in the visible and near infra-red regions, form a very convenient and accurate basis for the quantitative determination of the various pigments (Drabkin, 1950; Gordy and Drabkin, 1957), they yield little obvious information concerning the relationship of the absorption to structures in the complex molecules. Dhere's finding that haemoglobin, like other proteins, had an absorption maximum in the vicinity of 275 m/< led him to conclude that the haemin nucleus v/as responsible for the absorption in the visible region, while the globin caused the absorption in the y band and ultra-violet regions (Dhere, 1906). In an early attempt to analyse the spectra of haemoglobin derivatives. Vies (1914) presumably accepted Dhere's earlier generalization, which has persisted to the present day. An examination of the spectrum curves of several haemoglobin derivatives led me to deduce about a quarter of a century ago that all the maxima (which represent bands) in the ultra-violet and some in the visible region were spaced at approximately equal frequency distances from each other (Drabkin, 1934). This was a potentially important discovery, since it materially simplified the interpretation of the complex spectrum from the viewpoint of its origin in the molecular structure. Before this work such a distribution of integrally related absorption bands, belonging to a spectral 142 Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 143 series, had been demonstrated only in relatively simple molecules, such as KMn04 (Hagenbach and Percy, 1922) and CoCU in concentrated HCl (Erode, 1928). It was deduced that the absorption spectra of haemoglobin derivatives were largely an expression of iron in a co-ordination complex, and attention was called to certain similarities in the spectra of K3Fe(CN)g '- ^fh 1 1 / : P ^ 'J ; ^^' 1 H r : i f%\ r : 1 I 1 : HbC 32 r.^' -^^^^ /" ^^i;i>« •"^ / '- J / HbC / "i"iiVli 1 1 1 1 II ll !| M ilLllllll ll 11 lllIM 'Ml III II 1 1 1 1 1 1 1 J 1 iiiiiiiii mil 11 II II 1 ll 1 1 1 1 tOOO 900 800 700 600 500 400 300 200 A, tn/^ Infrared 4" Visible 4*— Ultraviolet " d ore =10-50 d ore =1 d ore =01 Fig. 1. Absorption spectrum curves of oxyhaemoglobin, HbOj, deoxygenated haemoglobin, Hb, and carbonyl haemoglobin, HbCO (Drabkin, 1950; Gordy and Drabkin, 1957). The values of cuvette depth, d, or concentration, c, suggest the relative thicknesses of layer or concentrations required for optimal spectro- photometry in the different spectral regions, due to a 500-fold difference in the densities of the maxima over the spectral range of 1100 to 200 m/<. The use of log £ (the log of the molar or 1-Fe-atom equivalent extinction) permits the portrayal of the maxima in region 1 together with the rest of the spectrum. Many derivatives of the chromoproteins have weak bands in this region (the red and near infra-red). As examples, proto- and mesohaemin hydroxides have maxima in the region 810-820 m/<. Met- or ferrihaemoglobin hydroxide (alkaline methae- moglobin, pH 9), the spectrum of which in the near infra-red was originally studied by Horecker (1943), also has an absorption maximum at 820 m/< {v x 10~^ = 122), with £ (1 mM/1.) = 0-544 and log E = 2-735 (Gordy and Drabkin, 1957). and cyanmethaemoglobin (Drabkin, 1936). The same deduction was later made from the similar paramagnetic susceptibilities of these iron complexes by Coryell, Stitt and Pauling (1937). The absorption spectrum curves were resolved into component bands by means of a novel graphic-mathematical analysis (Drabkin, 1937, 1938, 1940, 1950). Interestingly enough, the analysis indicated that the a and ^ bands did not belong to the main spectral series and that the band at 275 m^< was not primarily owing to globin (Drabkin, 1937, 1938). Furthermore, the analysis predicted the potential occurrence of bands in the neighbourhood of wavelengths 833, 313 and 250 m/<. This was 144 David L. Drabkin verified by the finding of a definite maximum at 820 m/t in the spectrum of methaemoglobin hydroxide and at 314 mjLi in the spectrum of ferrocytochrome c (Drabkin, 1941a). The maximum at 314 m/< was later designated as the 6 band (cf. Theorell and Nygaard, 1954; Tsou and Li, 1956; Morton, 1958). The very brief past reports embracing this phase of our studies have presumably escaped notice by those more recently concerned with the analysis of the spectra of chromoproteins (Williams, 1956; Morton, 1958). In this communication a more complete description of our graphic analytical method will be supplied, together with an assembly of absorption data relevant to an interpretation of the spectra of haemin-protein complexes and possibly to the development of the theory of molecular spectra. MATERIALS AND METHODS Materials The haemoglobins were crystallized preparations (Drabkin, 1946, 1949a), the solutions of which were rendered 'salt-free' either by dialysis or passage through 'Deeminac 16-4' resin (Drabkin, 1954). The cytochrome c was prepared from horse heart by Tint and Reiss (1950), and by our analyses (see below) was 96 % pure, with reference to 0-43 % iron content. Methods Spectrophotometry. Most of the more recent measurements were carried out with the Beckman DU instrument, but in the earlier work the Hilger Spekker photometer and medium quartz spectrograph were used. When other types of instrumentation such as recording spectrophotometers were em- ployed, this will be indicated in the legends to the figures. Table 1 . f max of cyanide derivatives in the direct determination of HAEMIN IRON Compound Wavelength (in m/<) £max Fe-contentt (%) 1-Fe-atom equivalent weight Cyanmethaemoglobin Cyanmetmyoglobin Ferricytochrome c cyanide Ferriprotohaemin dicyanide 540 545 537 545 11-5 11-5 11 •o§ 11-3 0-338 0-339 0-41211 8-5711 16,552t 16,503t 652 * £max, the millimolar extinction referred to 1 milliatom of iron, t Iron determined independently as ferrous 1 : 10-phenanthroline (Drabkin, 1941b). % May be rounded out to 16,500. § Provisional. II Sample was 96% pure on basis of 0-43% of iron, or 91-6% pure on basis of 0-45; of iron. H Iron content of a-chlorohaemin. Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 145 Concentration of Reference. The spectrum curves are plotted for a conven- tional depth of 1 cm. The concentration of reference, unless otherwise indi- cated, is 1 mM/L, where 1 niM represents 1 milliatom of iron. The iron content was determined independently (Drabkin, 1941b), but the spectrophotometric determination of the extinction at the maximum in the region of 545 to 170 180 190 200 210 220 - /" ol I - / 1 P\ — \ y / r^^ - // // / / / / X -^~ la - / 1 1 1 1 ! y t 1 1 1 / / !2- '" 1 ^ 1 1 1 1 1 1 K 510 Fig. 2. Absorption spectrum curves of complexes of iron. Curve 1 , ferrous 1 : 10- phenanthroline (ferrous o-phenanthroline). Curve la, ferrous dipyridyl complex. Curve 2, cyanmethaemoglobin (ferrihaemoglobin monocyanide). Curve 2a, haemin dicyanide (ferriprotoporphyrin dicyanide). The e (c = 1 mM/1.) for ferrous 1 : 1 O-phenanthroline is 11 -05 at the maximum of 500 m//, read against an appropriate blank. The open circle corresponds with a value of 11-25 for this complex read against water. For details see Drabkin (1941b). (Absorption spectra closely similar with those of cyanmethaemoglobin and haemin dicyanide are yielded by the cyanide derivatives of the ferrihaemochromes, such as monocyanide monopyridine ferriprotoporphyrin, with e = 11-7 for the maximum at 545 m/t (Drabkin, 1942a) and by mesohaemin dicyanide and coprohaemin dicyanide with maxima respectively at 537 and 535 m/<, displaced about 10m/< toward the shorter wavelengths in comparison with protohaemin derivatives (Drabkin, 1942b).) 535 m/f of corresponding cyanide derivatives of the ferri-complexes served in the direct, unequivocal evaluation of haemin iron (Drabkin, 1942a and b, 1949, 1954). Table 1 and Fig. 2 supply pertinent information. It may be noted that the writer's value of 0-338% for iron in haemoglobin (Drabkin, 1949b) has been tentatively accepted by the Protein Commission of the International Union of Pure and Applied Chemistry (cf. Drabkin, 1957). This iron content appears to be valid on a dry weight basis, and corresponds with a 1-Fe-atom equivalent weight of 16,500 for haemoglobin and 15,850 for globin, the total molecular weight of which may be taken as 15,850 X 4 = 63,400. 146 David L. Drabkin Notation. In the molecular interpretation of spectra, the frequency v (the number of waves passing a fixed point in a unit time, as 1 sec) is of more fundamental interest than the wavelength I. A term closely related to the frequency is the wavenumber v (the number of waves in a unit of length, as 1 cm). The relationship between )' and v is given by v = v x c, where c = 3 X 10^° cm/sec, the speed of light, and the relationship between v in cm~^ and X in vnfx is given by r = (l/A) x 10^. Thus A500 vo-fx corresponds Xov = 20,000 cm~^. It is convenient to use v X 10~^, and most of the graphs are so plotted. Anotherterm, the fresnel,/, has been employed. /= v X 10~^^. Hence, v x 10~^ =fl^- The absorption curves are plotted logically against v in ascending order from left to right, which is in the descending order with reference to A (cf. Drabkin, 1950). The Graphic Analysis of the Absorption Spectrum Curves. In their analysis of the visible absorption spectrum of permanganate Hagenbach and Percy (1922) made the assumption that the component bands (represented by maxima in the spectrum) could be resolved into individual simple curves, all of the same shape, but of different height. The shape of the curves was determined by the curvature of the slope at the lowest frequency end of the absorption curve. The summation of the resolved curves yielded the absorp- tion spectrum curve. Erode (1928) used effectively a similar method of analysis for the spectrum of CoClg. The spectra of these relatively simple molecules could be regarded as possessing essentially one broad band in which the multiple maxima represented a finer structure (Harrison, Lord and Loofbourow, 1948). The general character of the absorption spectra of the haemin chromoproteins is very different from that of KMn04 or CoCL, in HCl (see Figs. 1 and 4), and the spectrum curves could not be shown to be reproduced by a summation of component curves of the same shape, varying only in height. A highly satisfactory resolution of the complex haemin-protein spectrum curves has, however, been attained (Drabkin, 1937, 1938, 1940 and 1941a) by assuming that the component elements (bands) did not have the shape of simple curves but could be described by the 'bell-shaped' normal frequency curves of the form ;; = A: e ^ 2a== ; (1) or y = -tke:^ ^^^ q\ 2o^ ) The summation of unit curves of this type, each with different values for k and a (Fig. 3), reproduced very accurately the original (determined) spectrum curves. The families of curves in Fig. 3 are drawn for orientation to equation (1). In such symmetrical curves the values of j and x on each side of the I Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 147 centre of distribution at are the same, y is the height from the base of a point on the curve at a distance .v from the centre, k is the height at the centre of distribution a (usually designated the 'mean') for a particular case (in the diagrams of this figure a = 0). a, the standard deviation, is a measure 1 / \ ^ 3 k. i/ariable 3 , / \ cr, constant y = ke" 20-2 C, varioble / f \ \ 2 1 \ 1/ /- ^ \ \ J \ 1 / ^ / V -s \/ ^ \ 1 ^ y ^ n -\ S >-, \ ^ ■—/ ^ I ^ A /, " ^ ^ y J, 1 I \ ^1 \ -L_±x-0. -i- ± X -Q^ I 2 ^ / VI K \, k, constant ^ Vh ^ .^ FiG. 3. Families of normal frequency curves of the form: _ fi x - a)- \ y = kc ^ 2a^ ^ The eflfect of variability in either ^ or c is shown in 1 and 2. In 3, most pertinent to the present analytical assumptions, both k and a are variable. of the spread or variation about the centre of the individual points, over two- thirds of which in such curves lie within the interval Icr and 95% within la. e is the base of natural logarithms, 2-7183. In our application to spectra, y and k are in a units, x and a (the locations of the centres of the curves or bands, assumed or deduced from the locations of the maxima in the absorption data) in units of/ or v. In the construction ot the curves, 2cr^ is conveniently evaluated by a rearrangement of terms in equation (1). (.V- a)^ (3^ 2a2 = loge kly k and y in equation (3) are derived either directly from the absorption data or by adjustment for overlapping of neighbouring component curves. In the case of prominent bands, as the so-called y or Soret band in the spectra of the chromoproteins, or bands at the lower frequency end of spectrum as with 148 David L. Drabkin C0CI2 (Fig. 4), a mean value of la"- is readily obtained from several points along the left contour of the determined absorption curve. In other cases simultaneous equations are useful in deriving appropriate values, aided by suitable processes of curve fitting, details of which cannot be supplied here. Having settled on values for 2a^ and /c, values for y are calculated with Fig. 4. The graphic-mathematical analysis of the absorption spectrum curve of C0CI2 in concentrated HCl (Drabkin, 1940). The continuous solid line with multiple inflections is the absorption spectrum obtained by Erode (1928), and the open circles represent summational points obtained by his method of analysis (see text). The individual curves, numbers 34 to 41, were obtained by the writer's method of analysis with curves of the normal frequency form. The black dots show the summation of these curves, expressed by the equation inserted in the figure. For 734 to J41 the values of k are 0-19, 1-20, 0-75, 1-08, 0-55, 0-80, 0-57, and 0-19. The corresponding values for Id^ are 102-0, 102-0, 102-0, 97-5, 93-8, 92-3, 27-2, and 66-2. equation (2) to yield the curves. Experience will suggest labour- and time- saving devices in this type of graphical analysis. It should be clear that the choice of spectral interval in terms of i^ X 10"" determines the locations (at equal frequency distances from each other) of the values of a. It is of interest that the analysis of the absorption spectra of complex molecules into component bands of the shape of normal frequency curves makes it possible to express the spectra in relatively exact mathematical terms. This cannot be done with the earlier successful analytical procedure used for KMn04 and CoCU. Hence it was desirable to test the applicability of the new method by the analysis of the spectra of the simpler molecules. Figure 4 and its legend give the analysis of the spectrum of CoClg, and the Analysis and Interpretation of Absorption Spectra of Hacmin Chromoproteins 149 conclusion appears justified that practically as good a result is yielded by the new method as by the one used by Brode (1928). The latter's frequency spacing of/= 12-28 was purposely retained in the analysis. It should be pointed out that the writer's method yielded one extra band (number 34) at the low-frequency end of the spectrum. This is explained by the fact that a values, calculated from the left slope of the absorption curve were suflftciently discrepant from each other to denote skewness and suggest the presence of an additional component. Tentative Corollaries or Rules of the Analysis and Notation of Bands. (a) The presence of bands in the absorption curves may be indicated either by well-defined maxima, by inflections ('bumps') in the curve, or by regions of relatively flat absorption. On the other hand, several component bands may be merged together into an apparently single band or may be hidden in the final 'summational' spectrum curve. These propositions can be demonstrated by the summational results obtained with two or more neighbouring (over- lapping) curves of the normal frequency type, (b) The positions of the maxima in the determined absorption spectrum may be displaced from their theoretically correct locations in the resolved components. Indeed, this is an expected consequence of the overlapping of the component elements in the proposed analytical method, (c) The analysis itself best discloses the multiple components, and, as has alieady been stated, 'predicts' the possible presence of hidden bands. Hence, certain component bands are represented by prominent maxima in the spectra of all haemin chromoproteins, others only in the spectra of some of the complexes or their derivatives (see Table 2). The descriptive notation used for a particular band in the complex spectra may prove controversial. This matter need not be debated here. The historical designations a and (i have been retained for the bands in the visible green spectral region. Even this may be illogical, since bands are present in the red and near infra-red spectral regions (Fig. 1). The a and /5 bands are deduced by our analysis to have a structural significance differing from the main frequency distributed series. For the latter, which includes the Soret and ultra-violet bands the designations y and 6 appear inappropriate, and they will be assigned a number, n, which is an integer (3, 4, 5, 6, 7, etc.) based on the frequency spacing v x 10"^ = 40. Thus 6 x 40 = 240, the wavenumber location or a of the y or Soret band; 8 x 40 = 320, the postulated location of the d band (Table 2, Figs. 5 to 10). EXPERIMENTAL Contribution of Haemin to Over-all Spectroscopic Character of the Haemin Chromoproteins. In Tables 1 and 2 and in Figs. 1, 2 and 5 to 11, which with their legends are largely self-explanatory, the basis for the analysis of the spectra and deductions drawn therefrom is furnished. Attention may be directed to several points. 150 David L. Drabkin 150 250 30 — - K3Fe(CN)s,IOOmlVIA - Hematoporphyrin, ImM/l. - Cyanmethemoglobin, ImM/L /" ]■ ■ ■■ III Ferriprotohemin dicyanide, ImM/L \ i V \ ( N k u - f 1 J 1 \' i U v ^ f 1 J 1 * / /^ =^-^_../" Multiples of 40 = 4 5 6 7 9 10 II 600 500 tT\/i Fig. 5. Comparison of absorption spectra of K3Fe(CN)6, haematoporphyrin (in alkaline solution), cyanmethaemoglobin or ferrihaemoglobin monocyanide (from dog haemoglobin), and ferriprotohaemin dicyanide or haemin dicyanide at pH 13. Hogness and colleagues (1937) reported a maximum at 230 m/t (corresponding with V X 10~^ = 435), with e = 32-1, for haemin dicyanide. The abscissal scales indicate the postulated locations of bands in the equally spaced, frequency distributed series (Drabkin, 1936, 1938). ■xlO" 1 ■ - — Cyanmethemoglobin 1 1 1/ ooo Cyonmetmyoglobin ; V / 1 \ / 1 V a^^ J r ' y^ / \ y ^^ Multiples of 40=4 \ n\// Fig. 6. Comparison of spectra of cyanmethaemoglobin (from the haemoglobin of man), ferricytochrome c cyanide at pH 10 to 11 (for preparation see legend to Fig. 7), and cyanmetmyoglobin (from horse heart myoglobin). The remarkable similarity of these spectra is evident. The abscissal scales indicate the postulated locations of bands in the equally spaced, frequency distributed series. Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 151 (1) The determination of haemin iron is most unequivocally and quantita- tively accomplished through the characteristic absorption in the visible region of the cyanide derivatives of haemin and the haemin proteins (Drabkin, 1942a, 1949b), and the spectra are remarkably similar in both the visible and ultra-violet regions (Figs. 2 and 6). These findings indicate that a single major molecular characteristic is responsible for the general over-all spectrum. In the writer's analysis for iron by the 1 : 10-phenanthroline method the e value for the maximum of ferrous 1 : 10-phenanthroline was found to be close to identical with that of the cyanide derivatives of the haemin complexes (Table 1 and Fig. 2). Accordingly it was deduced that in the analytical procedure iron was liberated from one complex (hexaco-ordinated haemin iron; Drabkin, 1936, 1938) and bound up in another, diimine iron (Drabkin, 1941b). The similarity of the spectra favoured the idea of a spectroscopically operative structural similarity in these different classes of compounds. The spectroscopic similarity of the cyanide derivatives of the ferrichromoproteins has its counterpart in their low paramagnetic susceptibilities (Coryell, Stitt and Pauling, 1937; Theorell, 1941). It was deduced from their magnetic behaviour that they are essentially (though not fully) octahaedral d'^sp^ co- valent bonded stabilized structures, in essence of the Werner hexaco-ordina- tion type (cf. Pauling, 1940, 1948, 1949). (2) In Table 2 it may be seen that in some haemin complexes only a limited number of the bands, postulated by the analysis, are represented by definite maxima in the absorption spectra. However, considering the data on the different haemin complexes as an interrelated whole, maxima representative of at least eight, possibly nine bands (numbers « = 3 to 1 1) of an equally spaced frequency distributed series are found. As has been stated, the spectrum of ferrocytochrome c (Drabkin, 1941a) proved to be particularly rich in maxima (Fig. 7) and disclosed the presence of bands missing from the earlier examined spectra of haemoglobin derivatives, but 'predicted' by the analysis. The basis for these differences in the spectra of reduced and oxidized cytochrome c and haemoglobin derivatives is not clear, unless it can be attributed to the difference in the bonding of the haemins with the protein (Theorell, 1938, 1941). With the exception of small 'shifts' in the location of some of the maxima, such spectral differences are erased in the spectra of the respective cyanide derivatives of these chromoproteins (Fig. 6). (3) An examination of Table 2 will disclose that bands number 3, 6, 9 and 11 in the series Vq x 10^^ = 40 can also be distributed at regular fre- quency intervals on the basis of Vq x 10"^ = 60. In the latter case the /9 band would be included as number 3 and the bands would be represented by n = 2, 3, 4, 5, 6 and 7, with number 5 in an intermediate position between 7 and 8 of the spacing Vq x 10"^ = 40. The 60 spacing was originally assumed (Drabkin, 1934), and led to the analysis, illustrated for the spectrum of cyanmethaemoglobin, in Fig. 8. In this figure it may be seen that, utilizing 152 David L. Drabkin the frequency interval of 60, the graphic-mathematical analysis into com- ponent bands fails to reproduce by their summation the observed absorption in the regions v X 10^- = 160, 200, 280 and 320. These spectral regions do have representative maxima in the absorption spectra of some of the chromo- proteins (Table 2), and either two or more separate series would have to be 150 250 1 1 1 1 ff] Ferrocytochrome c i Ferrici tochrorr e c, p H8 45 -/ 1 1 -f - 1 1 1 I \ V h ji > J V; >L V Multiples of 40 = 4 » /? 5 400 350 Fig. 7. The spectra of reduced and oxidized cytochrome c from horse heart, pH 8-45. The molecular weight of reference was taken as 13,000 (0-43 % of iron). The preparation had 0-412% of iron by Drabkin's o-phenanthroline method (1941b). Reduction to ferrocytochrome c was by means of NaoSgOi for the data to 380 m/ii and with palladium asbestos and hydrogen for the data in the ultra- violet region beyond 380 m/<. To insure complete oxidation to ferricytochrome c 0-4 of an equivalent of ferricyanide was added to the 60 % oxidized preparation. In the measurements this was balanced out by an equivalent amount ferrocyanide. The abscissal scales indicate the postulated locations of bands in the equally spaced, frequency distributed series (Drabkin, 1941a). assumed or a more appropriate single frequency spacing adopted. The latter alternative was taken, and accordingly the 40 spacing was tested. This spacing was preferred not only because of the locations of the observed maxima, but also intuitively since it excluded both the a and (j bands. The supplemental Figs. 9 and 10 illustrate the analysis of the absorption spectrum of cyanmethaemoglobin, taking Vq x 10~^ = 40. It may be seen that the summation of the analytically resolved ten absorbing units or bands agrees excellently with the observed spectrum, which can be expressed mathematically (Drabkin, 1937) by sj, = ( kQ {x - [n X 40])- \ + S(>'«,J^) (4) Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 153 The closeness of fit of the analytically derived spectrum with the experi- mentally determined one is reflected by values of 3-29 for the mean of devia- tions between them, 0-24 for the mean about which the deviations fall, 6-06 for S.D., and 0-47 for S.D. from 151 to 205 i5 x IQ-^ (Fig. 10). All the absorption spectra of the haemin chromoproteins and their derivatives thus far studied by the writer (a partial list of which is given in Table 2) can be similarly analysed into their component bands, and equation (4) may be regarded as generally applicable. (4) The dissection of the broad band of the spectrum of cyanmethaemo- globin at wavelength 540 m// or v X 10-^ = 185 (Drabkin, 1937, 1938 and Fig. 10) merits particular attention. The accurate establishment of the left contour of this band reveals a very slight 'bump' in the vicinity of 570 m/< (see curve 2 in Fig. 2 and the solid line in Fig. 10). There is also the appreciable absorption in the regions of 630-600 and 500 m^w, the locations assigned in the analysis to bands 4 and 5, which are represented by definite maxima in some of the chromoprotein spectra (Table 2). These findings in themselves Table 2. Location of maxima (bands) in chromoprotein spectra, postulated to belong to series « = (jt x 10"^)/(vo x 10~^) Vq X 10~^ is assumed = 40 (where n = 1). The values in brackets give the number n of the band in the series. The values in parentheses give locations of inflections (or bumps) in the absorption curve, as distinguished from obvious maxima. Compound a P f X 10 - observed 173 184 f, Oxyhaemoglobin 109 241 290 362 417* [3?J [6] [71 [9] [11?] Carbonylhaemoglobin 176 186 111-125 [3?1 238 [6] 290 [7] 366 [91 426 [11?] Cyanmethaemoglobin (from 185 239 285 368 437 dog haemoglobin) [6] [7] [91 [11?] Cyanmethaemoglobin (from 185 237 282 366 haemoglobin of man)t [61 [7] [9] Methaemoglobin, pH 5-9 (173) (184) 159 198 246 285 359 435 [4] [5] [61 [7] (91 [11?] Methaemoglobin hydroxide. 173 184 122 167 (206) 241 282 365 431 pH9-2 [3] [4] [5] [6] [71 [91 [11?] Ferrocytochrome c, pH 8-45 182 192 (135) 241 (284) 319 (365) (403) (446) [3?] [61 [71 [81 [91 [101 [111 Ferricytochrome c, pH 8-45 182 191 (143) 244 280 (315) 359 (430) [3?] [6] [71 [81 [91 [11?] Ferriprotohaemin dicyanide. 183 235 278 (360) t pH13 [6] [71 [9] * With the Beckman DU spectrophotometer, measurements in this spectral location are unreliable; with ferrocytochrome c a definite inflection in the curve at v x 10~^ = 446 was obtained with Hilger's Spekker photometer and medium quartz spectograph. t For the spectra of ferricytochrome c cyanide and cyanmetmyoglobin see Fig. 6. X For a maximum reported to be present in this location see the legend to Fig. 5. suggest the composite nature of this band. In the analysis, the order of solution was band 4 first, then 5, then a, and finally /5. The need for a /5 band and the locations of the centroids of a and /5 were consequences of the analytical procedure. The centroids of the analytical a and /? bands are respectively at i^ x 10"- = 179 (559 m//) and 187 (535 m/0 and k is larger 154 David L. Drabkin for a. The locations and relative densities of the bands are very suggestive of those found in the spectra of ferrohaemochromes, like pyridine and globin ferroprotoporphyrins (see Fig. 17 and Drabkin, 1937, 1938). It is difficult to regard this analytical result as pure coincidence. It is at the least highly provocative, revealing as it does fundamental similarities in the spectrally ^ 80 MHbCN (■-;4o v y^= Il5e" 243 I. -3001 yj,= 3IOe -lozo (x-360) yg=3l'3e 360 (x-120) ' y^= 100 e 7go~ 500 Vu X 10" Fig. 8. The graphic-mathematical analysis of the absorption spectrum curve of cyanmethaemoglobin (from dog haemoglobin). The continuous solid line is the absoqDtion spectrum obtained experimentally. The broken line represents the summation of the individual component elements or bands (solid dotted lines) of the normal frequency form, derived by the writer's method of analysis. The bands are numbers 3 to 7 in an equally spaced, frequency distributed series, with Vq X 10~^ = 60. The summation of the resolved component bands is given by the equation inserted in the figure. The equations y^ to J7 (insert in figure) are for the respective components, with the applicable values for a and 2a- sub- stituted in equation 1 (see Methods). divergent visible region of haemin-protein spectra. The single band of deoxygenated haemoglobin at 555 m/< can be similarly resolved into bands 4, a, /? and 5. To summarize, it may be concluded from the graphic-mathematical analysis that (1) the absorption spectra of all haemin chromoproteins and their derivatives (both oxidized and reduced) are fundamentally similar, (2) the a and /5 bands have a different origin from bands 3 to 11 of the equally spaced frequency distributed series, and (3) the differences in the spectrum curves (largely evident in the visible spectral region) are an expression of the relative densities or intensities of the absorption bands in the different compounds (Drabkin, 1937, 1938, 1941a; Table 2 and Figs. 10 and 12). W 60 MHbCN / " 2'y„=ke x-tn40l)' r - y4 = 80e" 108 "\ 1 - 7^=630 6" 155 1 1 y^=4 32e- <-l87)^ 121 1 y^=5 00e" .-200)^ 320 1 1 y, = ll8e- «-240)^ 236 1 y^ = 270e- >-280)^ 1613 1 !l yg = 14 8e" .-320)^ 1332 1 i ,;/^ \ 1 . y9 = 333e- .-360)2 564 Un ^ I - y,o=2l7e- «- 400)2 1345 i K'^^ /y * \L \ - y,| = 190e- 1 1 «-440)2 B90 /"^^ J^l ^ 1/ ''"^^SK :>< .1. n*-.t. . 250 300 VuXlO" Fig. 9. The graphic-mathematical analysis of the absorption spectrum curve of cyanmethaemoglobin. (This figure and Fig. 10 are supplementary.) The continu- ous solid line is the absorption spectrum obtained experimentally. The broken line represents the summation of the individual components or bands (solid dotted lines) of the form of normal frequency curves, obtained by the writer's method of analysis. In contrast vi'ith Fig. 8, the bands in the equally spaced frequency distributed series are resolved on the basis of Vq X 10"^ = 40. Bands with numbers « = 6 to 1 1 are shown in this figure; numbers 4 and 5 and resolved bands a and /S in Fig. 10. The summation of the bands (4 to 11) in the single series and the mathematical formulation of each unit is given by the equations inserted in the figure. See Fig. 10 and the text. 140 150 160 170 180 190 Fig. 10. The graphic-mathematical dissection of the band in the green spectral region of cyanmethaemoglobin. This figure is supplemental to Fig. 9. See legend to Fig. 9 and the text. H.E. — VOL. I — M 156 David L. Drabkin Contribution of Protein to the Over-all Spectroscopic Character of the Haemin Chromoproteins. From what has already been said, the affect of the metalloporphyrin complex is dominant throughout the spectral range, and the absorption of the haemoglobins and of cytochrome c in the ultra-violet region, even at the location of the so-called protein band (280-270 m/t ; band number 9 in the analysis), cannot be attributed exclusively to the protein moieties. Fig, 11. The contribution of the protein moiety, globin, to absorption in the region 275 to 280 m/t (v x 10"^ = 360; band number 9). Curve 1, heavy solid line, oxyhaemoglobin, 1 him/I. (1 milliatom of Fe); curve 2, heavy broken line, carbonyl haemoglobin, 1 mM/1. (1 milliatom of Fe); curve 3, light solid line with open circles, met- or ferrimyoglobin, 1 mM/1. (1 milliatom of Fe); curve 4, light solid line, globin (from haemoglobin), 0-25 mM/1. (reference mol. wt. = 15,850); curve 5, light solid line with black dots, bovine plasma albumin, 1 roM/l. (reference mol. wt. = 68,000); curve 6, broken line with dots, bovine albumin, 0-25 mM/1. (reference mol. wt. = 17,000); curve 7, broken light line, denatured globin, 0-25 mM/1.; curve 8, light line with crosses, 1 mM/1. with reference to haemin (protohaemin added to amount of globin used for curve 7). Curves 4 and 6 aflFord a comparison of globin and albumin at approximately similar concentrations by weight. Curves 1 and 2 indicate that, if expressed on a molar basis (mol. wt. = 66,000), the absorption in this spectral region would be four times greater for the haemin proteins than for albumin. See the text. The spectroscopic data plotted in Fig. 11 substantiates this conclusion. It may be seen that the bands of haemoglobin derivatives and metmyoglobin, though generally similar yet differ slightly from each other (curves 1 to 3). On the other hand, they are broader and smoother than the band of bovine albumin (curve 5). Misleading deductions as to the molecular origin of this band in chromoproteins may have been drawn because of the convention of expressing information on the chromoproteins on a 1-Fe-atom equivalent basis, whereas for proteins like albumin the molecular weight base has been used for reference. If the absorption for the haemoglobin derivatives were to be referred to 1 mM/l., instead of 1 milliatom of Fe/1., its maximum in this Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 157 spectral region would be four times higher than that of albumin (see curves 1 to 6 and the legend, Fig. 1 1). However, denaturation does have an influence on chromoprotein spectra in this region, as it does on the albumin spectrum (compare curves 4 and 7). The influence of haemin in this spectral region is suggested by curves 4, 7 and 8. Contrariwise, if the protein moieties do exert spectroscopic influence, they do so more evidently in the visible spectral region, the spectral changes in which may be deduced to be a function of the nature of co-ordinating groups or ligands bonded to the iron, which is also bonded to the four pyrrolic nitrogens. Thus, the alkaline denaturation of haemoglobin yields globin haemochrome, and the spectrum of globin ferro- protoporphyrin in the visible spectral region is practically indistinguishable from that of pyridine ferroprotoporphyrin (Drabkin, 1942a and Fig. 17). The situation with respect to spectroscopically operative protein structures is presumably different in the case of cytochrome b^. The pronounced maximum at 265 m/i {v x 10^^ = 377) appears to be clearly ascribable to the riboflavin phosphate and a polydeoxyribonucleotide, which are structural components of this complex molecule (Appleby and Morton, 1954; Morton, 1958). Whether the influence of the nucleotide is confined to this spectral region or may be reflected in other spectral regions remains to be ascertained. Thus far no convincing evidence has appeared that the number 9 band in the chromoproteins may be a composite with finer structure attributable to the aromatic amino acids, as has been shown for non-haemin proteins (Holiday, 1936,1937; Lavin, Northrop and Taylor, 1933; Lavin and Northrop, 1935). It may be concluded that from the qualitative viewpoint the protein moiety of the haemin proteins contributes only negligibly to their over-all spectra (see Discussion). The a and (j Bands and the Neighbouring Visible Spectral Regions. Despite the fundamental similarities disclosed by the analysis of the spectra of the haemin proteins, in the past major attention has been devoted to characteristic and pronounced diff'erences in these spectra in the visible spectral region. Such diff'erences have been valuable in the accurate determination by means of spectrophotometry of biologically important derivatives as well as in the study of certain equilibria (Austin and Drabkin, 1935; Drabkin and Singer, 1939; Drabkin and Schmidt, 1945; Drabkin, 1950; Gordy and Drabkin, 1957), but may have directed attention away from the basic similarities. It may be said that the combination of haemoglobin with gases (Og, CO, NO), the oxidation of haemoglobin to ferrihaemoglobin, the alkaline denaturation of haemoglobin to haemochrome, and the pH dependency of ferrihaemoglobin are all spectroscopically operative, and that many of these reactions have parallels in the spectroscopic behaviour of the cytochromes. Figures 12 to 17 and Table 3 are relevant in the interpretation of the diff'erences in the visible spectra of chromoproteins, consonant with the analytical viewpoint that all the spectra have a and /3 components straddled by bands 4 and 5 of the dominant 158 David L. Drabkin Table 3. Correlation of spectroscopic pattern with electronic structure S = strong band; W = weak band; M = moderately strong band; N = negligible band Compound Figure number Spect 4 rum bands a B Number of unpaired Typet r electrons* Ferrihaemoglobin 12 S W w 5 Ionic (1) Ferricytochrome c I and II 7 St W w 5 Ionic (2) Ferrohaemoglobin 1 St MJ Mt 4 Ionic (3) Ferrihaemoglobin hydroxide 12 Mt S s 3 Ionic (1) Ferrihaemoglobin cyanide 10 wt Mt Mt 1 Covalent (1) Ferricytochrome c cyanide 6 wt Mt Mt 1 Covalent (2) Oxyhaemoglobin 1 N S S Covalent (3) Carbonylhaemoglobin 1 N s s Covalent (3) Globin ferroprotoporphyrin 17 N s s Covalent (4) Pyridine ferroprotoporphyrin 17 N s s Covalent (4) * The permanent magnetic dipole moment in these complexes is due almost entirely to Hsy the spin moment of the unpaired electrons, /is is derived from the measured molal paramagnetic susceptibility. Theory for //s for 1 to 5 unpaired electrons is 1-73, 2-83, 3-83, 4-90 and 5-92 Bohr magnetons. t The terms ionic and covalent should be prefaced by 'essentially' to indicate partial ionic and covalent character. Thus, ferrihaemoglobin is more ionic than ferrihaemoglobin hydroxide (Pauling, 1940, 1948, 1949). References: 1, Coryell, Stitt and Pauling, 1937; 2, Theorell, 1941 ; 3, PauUng and Coryell, 1936b; 4, Pauling and Coryell, 1936a. t Deduced from the graphic analysis of the spectrum. series. The relative prominence of these four components resuUs in the spectral differences. This is well illustrated in supplementary Figures 12 and 13, which show the transition with change in pH of the spectra of human met- er ferrihaemoglobin to ferrihaemoglobin hydroxide. The pH dependency of the spectrum of ferrihaemoglobin was originally studied by Hartridge, 1920 and Haurowitz, 1924. A detailed and very careful spectrophotometric study by Austin and Drabkin (1935) of dog ferrihaemoglobin, MHb, with reference to the equilibrium MHb ^ MHbOH revealed a reaction of the first order with OH ion, and pi^g (as it is now usually designated) was accurately established as 8- 12 ± 0-01 at /< (ionic strength) = 0-1 and a = 0-6. This value for p^3 has been confiimed by the independent techniques of magnetometric titration (Coryell, Stitt and Pauling, 1937) and differential acid-base titration (Wyman and Ingalls, 1941). Furthermore, the demonstration of the effect of ionic strength, regarded in that day as unusual for such complex com- pounds, was also apparent in their magnetic behaviour. Thus, the haemin- linked group, responsible for ^K^ (i.e. the release of a proton from a molecule of water co-ordinated with the iron in acid ferrihaemoglobin, MHb • HOH+, to form MHb -OH -H H+) was spectroscopically, magnetometrically and titrimetrically operative. It appeared that at least in this case the electronic structural change involved in MHb ^ MHbOH (MHb, with a magnetic Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 159 susceptibility corresponding with 5 unpaired electrons, to MHbOH with 3 unpaired electrons; Coryell, Still and Pauling, 1937) could be correlated with the spectra of the transition of one spectroscopic species to the other. (In ferricytochrome c, appreciably more pH-stable than ferrihaemoglobin, four such spectroscopically operative acid groups have been uncovered (Theorell and Akesson, 1941).) The pA'3 for human MHbOH is 8-15 at /i = 0-1, nearly the same as that of dog MHbOH. This value is derived from the automatically recorded spec- trum curves in Fig. 12 (Drabkin and E. Thorogood, unpublished). That only two species participate in the spectroscopic transition was evident both in the visible and near infra-red spectral regions and was reflected by the presence of 5 isosbestic points (see legend to Fig. 12). However, it may be seen in Fig. 13 that there is a departure from this behaviour in the near ultra- violet region, since band number 6 is appreciably higher and located at a shorter wavelength for MHb than the corresponding band for MHbOH (cf. also Hicks and Holden, 1929). In the present connection, the main point which may be stressed is that bands 4, a, /5 and 5 are evident in the spectra of both MHb and MHbOH, but in the latter a and ^ dominate, whereas in the former a and ^ are only weakly expressed, while 4, particularly, and 5 are relatively dominant (see Fig. 12). Using the spectra of MHb and MHbOH as models, the relation between the quantum mechanical deductions drawn from measurements in Linus Pauling's laboratory of the molal paramagnetic susceptibilities of haemin and its derivatives and the spectra of these compounds may be placed upon a somewhat broader base by a suggestive correlation of the spectroscopic pattern with the corresponding electronic character. In general, all essentially covalent structures have prominent a and ^5 bands and weak number 4 bands, whereas essentially ionic structures may have relatively weak a and /5 bands, but are mainly characterized by strong number 4 bands or marked absorption in the spectral region of the 4 band. This is brought out in Table 3. The spectral patterns in the visible region are not as diverse as may have been supposed (Drabkin, 1942a and b, and Figs. 15 to 17). However, oxy-, carbonyl, cyanide and pyridine derivatives of ferrohaem are spectro- scopically distinguishable from each other, whereas their electronic configura- tion is the same (Table 3). It seems reasonable to infer that both the electronic structure of the haemin iron and the nature of the co-ordinating ligand contri- bute to the spectrum (cf. also Williams, 1956). The co-ordination of haemin iron with OH ion is a general reaction, exhibited also by the ferrihaemochromes (cf. Haurowitz, 1927; Davies, 1940). In Fig. 14, the writer's spectrophotometric measurements of the equilibrium pyridine ferriprotoporphyrin ^ pyridine ferriprotoporphyrin hydroxide are supplied. From these data a pAT value of 9-64 may be derived (see Insert to Fig. 14). However, there is a most interesting difference between these 160 David L. Drabkin D I Fig. 12. Absorption spectrum curves of solutions of crystalline human met- or ferrihaemoglobin, buffered to various pH values. All the buffers had an ionic strength, /<, of 0-1. D is the optical density. The spectra were obtained with the General Electric recording spectrophotometer on solutions with a haemoglobin iron concentration of 0-0584 mw/l. (Figs. 12 and 13 are supplementary.) Curve 1, methaemoglobin at pH 6-20 (taken as species 1). Curve 2, methaemoglobin hydroxide at pH 9-35 (taken as species 2). The nine intermediate curves represent mixtures of the two species at respective pH of 6-61, 7-05, 7-34, 7-7, 8-17, 8-50, 8-60, 8-80 and 9-20. Methaemoglobin hydroxide has an additional maximum at 820 m/< and there is an additional isosbestic point for the two species at 845 m/t (Gordy and Drabkin, 1957). For the earlier work on the equilibrium of methae- moglobin-methaemoglobin hydroxide and the treatment of spectrophotometric data in a two component system see Austin and Drabkin, 1935. Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 161 spectra and the corresponding ones for MHb ^ MHbOH. The colours of the spectroscopic species are reversed with reference to pH. MHb is brown, MHbOH red, whereas in the pyridine complexes the colour is olive brown at D I A ^ - n - V - 1 - 1 - J '^Tl 1 1 2 1 1 1 r 1 1 1 1 r -r Fig. 13. Absorption spectrum curves of solutions of crystalline human met- or ferrihaemoglobin, buffered to various pH values. The solutions correspond to those shown in Fig. 12, but the concentration of haemoglobin is 0-01 17 mM/1. Tsee legend to Fig. 12). The e (lmM/1) values for the maximum of methaemoglobin (pH 6-2) at 407 m/< and for the maximum of methaemoglobin hydroxide (pH 9-35) at 415 m/i are 169 and 113 respectively. pH 11-38, red at pH 7-26. Uncertainties still exist as to the structure of pyridine ferriprotoporphyrin (Davies, 1940) and magnetometric measure- ments are available only for the hydroxy form (Rawlinson, 1940), which was deduced to be essentially covalent in contrast with MHbOH (see Table 3). Further information is required for the clarification of this situation, but at present the spectra cannot be easily reconciled with the proposed correlation. 162 David L. Drabkin Contribution of Porphyrin to the Over-all Spectroscopic Character of Haemin Chromoproteins. In the early days of the investigation of the struc- ture of cytochrome c, Theorell (1939) questioned whether the thio-ether linkage he had proposed (Theorell, 1938) for the union of the protein with the haemin at positions 2 and 4 was present in the native compound or was 190 200 Fig. 14. Absorption spectra showing transition, with change in pH, from pyridine ferriprotoporphyrin to its hydroxide. In all solutions the final concentrations of haemin Fe and pyridine were 0-1 mM and 5700 mM/1. respectively. The pH was modified by the inclusion of HCl in all solutions except that represented by curve 9. Curve 1, absorption spectrum of species 1, probably dipyridine ferriproto- porphyrin, pH = 7-26. Curve 9, absorption spectrum of species 2, pH = 11 -38. Curves 2 to 8, absorption spectra of mixtures of species 1 and 2. The pH values of the solutions represented by curves 2 to 8 were 7-62, 8-32, 8-95, 9-45, 9-86, 9-98, and 10-36. The insert in the figure shows the partition of species 1 and 2 against pH (see legend to Fig. 12). The solid circles are from data presented in the figure. The open circles are based upon absorption data not shown. See the text. artifactually obtained by the vigorous hydrolysis conditions employed in the isolation of porphyrin c. This led the writer (Drabkin, 1942b) to investigate the spectra of cyanide, pyridine and carbonyl derivatives of proto-, meso-, and coproferrohaem, and corresponding derivatives of haemoglobin and ferrocytochrome c. In protohaemin positions 2 and 4 are occupied by the unsaturated vinyl group, whereas in meso- and coprohaemin the substituents in this position are respectively ethyl and propionic groups. The spectra of the meso- and copro- derivatives were virtually indistinguishable from each other (Figs. 15 to 17 and see Drabkin, 1942b, for the individual spectra of Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 163 carbonyl derivatives), indicating tliat the CoHg and CH3CH2COOH side chains had individually no distinguishable spectroscopic affects, but the maxima in the spectra of all protohaemin complexes were shifted some 10 m^« toward the longer wavelengths. Three distinctive spectroscopic patterns were found for all the ferrohaems, characteristic for the combination with 16 17 18 19 20 : 1 1 1 1 1 1 ' ' T 1 1 1 III' — 1 — '-. 1 2A - r r 2. 12 ': r^ ^ r J \/n \ \ r 1, V ►3 rir «(=!»-: ^. 1 1 1 \ 1 1 1 1 1 1 1 I 1 620 580 540 500 A, m/^ Fig. 15. The three distinctive spectral patterns exhibited by ferrohaems co- ordinated with three types of ligands, exemplified by derivatives of ferromeso- porphyrin. Curve 1, cyanide ferromesoporphyrin, representative of Pattern Type 1. Curve 2, pyridine ferromesoporphyrin, representative of Pattern Type 2. Curve 3, carbonyl ferromesoporphyrin, representative of Pattern Type 3. In all cases the NaOH concentration was 0-2 m/1. and the NagSjOi concentration 5 mM/1. The pyridine concentration was 6-19 m/1. and that of cyanide 400 mM/1. The horizontal arrows and appended numbers represent the magnitude in m/i of the shift of maxima towards longer wavelengths in corresponding derivatives of ferroprotoporphyrin (Drabkin, 1942b). cyanide, pyridine and carbon monoxide (Fig. 15). These conclusions were reached: (1) The shape and intensity of absorption in the visible region was a function of the nature of the co-ordinating ligand. (2) The wavelength location of the maxima (of the a and /5 bands) was a function of the haemins (or porphyrins) themselves, and most probably of the groups substituted in positions 2 and 4. The maxima in the spectra of haemoglobin derivatives were in the locations expected for protohaemin derivatives. On the other hand, the 164 David L. Drabkin spectra of derivatives of cytochrome c (Figs. 16 and 17) were in the meso- or copro- locations. Hence, this was offered as evidence that the spectrum itself of cytochrome c revealed a structural difference of its haemin in positions 2 and 4, namely a vitiation of the unsaturated vinyl bond structure, such as would occur in Theorell's thio-ether linkage (Drabkin, 1942b) Fig. 16. Pattern Type 1; absorption spectra of cyanide derivatives of ferro- haems, haemoglobin, and ferrocytochrome c. Curve 1, cyanide ferroprotopor- phyrin. Curve 2, cyanide ferromesopoi"phyrin. Curve 3, cyanide ferrocopropor- phyrin. Curve 4, the reduced cyanide derivative prepared from dog haemoglobin in alkaline solution, probably cyanide ferroprotoporphyrin. Curve 5, the reduced cyanide derivative prepared from cytochrome c in alkaline solution, ferrocyto- chrome c cyanide. See legend to Fig. 15 (Drabkin, 1942b). The spectral displacement of the protohaemin complexes toward the longer wavelengths was ascribed (Drabkin, 1942b) to the increase in the conjugated double bonds in the porphyrin system by the presence of the unsaturated vinyl radicals, analogously to the situation disclosed by the systematic studies of Hausser and Kuhn and their collaborators on polyene dyes of the type R— (CH=CH)„— R' (Hausser, 1934; Hausser et al, 1935a to e). Incidentally, on the basis of the above analysis, it could be prophesied that the haemin of cytochrome b.^ must be ordinary protohaemin with the vinyl residues intact (Morton, 1958). I believe that in the haemin proteins the influence on spectral location of the maxima in the visible region, evident also in ferrihaemin complexes such as the cyanide derivatives Analysis and Interpretation of Absorption Spectra of Haeniin Chromoproteins 165 (Drabkin, 1942a), represents the major contribution of the structure of the porphyrin nucleus to the spectra, although it is tempting to draw analogies between the band distribution and location in the spectra of porphyrins and /"xlO^^ Fig. 17. Pattern Type 2. Absorption spectra of denatured globin (globan) derivatives of ferrohaemins (reduced haemochromogens) and the spectra of haemoglobin and ferrocytochrome c in alkaline solutions. Curve 1, globan ferroprotoporphyrin, prepared from human globin and protohaemin. Curve 2, globan ferromesoporphyrin, prepared from human globin and mesohaemin. Curve 3, globan ferrocoproporphyrin, prepared from human globin and copro- haemin. Curve 4, globan ferroprotoporphyrin, prepared from haemoglobin of man. Curve 5, ferrocytochrome c in alkaline solution. See legend to Fig. 15 and for details see Drabkin, 1942b. their two-banded hydrochlorides and in haemin derivatives (Williams, 1956). Nevertheless, this is a matter of predilection, and Williams' interpretation that in porphyrins (see Fig. 5) the near ultra-violet band ('Soret band') is accounted for by a tt -> 77' electron transition, and the visible bands, I, II, III and IV by a second -n electron transition may be valid for porphyrin spectra (Williams, 1956). DISCUSSION Theory of Absorption Spectra. In general the absorption of light energy by atoms and molecules may be considered to be the reverse of emission. 166 David L. Drabkin Enlargement of this assumption involves the inference that the internal energy of the atom or molecule is increased by the absorption of light, and transfers from lower to higher quantized energy states occur, which give rise to absorption bands. In the ultra-violet and visible spectral regions the bands may represent either transitions of optical or valence electrons (as distin- guished from core electrons) through different quantized energy levels, or internal vibrational phenomena set up in the molecule by the absorption of hght energy. Henri (1919, 1923a and b) and Lifschitz (1920) were among the first to recognize that orderliness exists in the distribution of bands in simple molecules, which is expressed by the spacing of the bands at constant fre- quency distances from each other. Such integrally related bands, forming a spectral series, were early demonstrated in the spectra of KMn04 and C0CI2 (see Fig. 4), and have been interpreted as vibrational fine structure in a broad band of electronic origin (Harrison et al, 1948). The important implication lies in the inference that all bands which are members of a single series probably originate from a common configuration in the molecule, or from the same fundamental molecular disturbance incident to the absorption of light energy. The Spectra of the Haemin Chromoproteins. The finding — extended and supported by a graphic-mathematical analysis — that most of the bands (exclusive of a and (i) in the spectra of haemin chromoproteins and their derivatives are spaced at equal frequency distances allows the interpretation that they originate (as in the simple molecules above) in a common molecular structure and from fundamentally the same dynamic source. This enormously simplifies the interpretation of the complex spectra of these complex molecules. In effect, spectrally they behave like much simpler molecules. The location of the bands in the ultra-violet and visible regions permits the deduction that they represent electron transitions. Furthermore, the total iron-porphyrin structure as a unit is held responsible for the bands in the spectral series. In the resonating conjugated double bond system of the porphyrins, each atom contributes an optical electron to the molecule's collection, but the electrons belong collectively to all the atoms in the complex, not to a particular atom. Such electrons, belonging to several atoms, are characterized by relatively low energy (hence the spectral bands in the near infra-red, visible and ultra- violet regions), and the energy levels associated with them are regarded as closely and regularly spaced (cf. Harrison et al, 1948; Braude, 1945). The iron may be thought of as either facilitating or modifying the movement of the electrons over the atoms of the porphyrin ring. At any rate, the over-all spectrum is viewed as an expression of the spectrum of iron in a hexa- co-ordinated Werner type structure. Supporting evidence for the common origin of the bands in the spectral series may be drawn from Warburg's classical deduction of the photochemical spectrum of cytochrome c oxidase (Warburg and Negelein, 1928; Warburg, 1929). The photochemical spectrum, Analysis and Interpretation of Absorption Spectra of Haemln Chromoproteins 167 which measured the release of the enzyme from its carbonyl derivative (poisoned state) had several maxima in the ultra-violet, including the y band. This would not have been the case unless the bands had a similar molecular origin. The a and (i bands, as disclosed by the analysis, do not belong to the main spectral series, and have been shown to reflect more intimately the effect of different co-ordinating ligands. A broad correlation has been found between the paramagnetic susceptibilities and the corresponding spectral patterns in the area covered by bands 4, a, (i and 5 (Table 3). This correlation would appear to justify the deduction that the visible spectra reflect electron transitions involved in the hybridization of the atomic s, p and d orbitals of the iron cation (Pauling, 1949 ; Williams, 1956). Four major spectral patterns have been uncovered, namely those of ferrihaem and its derivatives with cyanide, and those of ferrohaem complexes with cyanide, pyridine or globin, and carbon monoxide. Identical patterns (see the text and Figs. 15 to 17) are obtained for corresponding derivatives of proto-, meso- and coprohaemin, except for the displacement of the maxima of the protohaemin complexes toward the longer wavelengths. This displacement is also apparent in the spectra of proto- and mesoporphyrin, and is regarded as the main evident contribution of porphyrin />e/-5e to the over-all haemin protein spectra. Since the disclosure of the close similarity of the spectra of ferrous 1 : 10- phenanthroline and the cyanide complexes of ferrihaemin derivatives (Drabkin, 1941b, and Fig. 2), the structure of the ferrous diimine has been shown to be Fe^Pheug (Gould and Vosburgh, 1942; Harvey and Manning, 1952) and the anomalous colours of the metallic diimines have been explained as due to the resonance of 77-electrons (and their increased mobility in the metal complex) over the non-metallic atoms composing the entire chelate ring, assuming that ^/-electrons of the metal are involved in the formation of the co-ordinate bond (Sone, 1952; Krumholz, 1953; see also Pauling, 1940; Calvinand Wilson, 1945; Chatt, 1949). Williams (1956) has used the metallic diimines and the postulated electronic basis for their spectra as models in his interpretation of the visible absorption spectra of the haemin chromoproteins. The contribution of the protein moieties to the over-all spectra of the haemin chromoproteins is not identifiable in any alteration in spectral pattern, nor does it appear to be confined to a particular spectral region such as 280-275 m^w. The role of the protein is postulated to be that of a 'resonator' or 'enhancer'. The relative intensity of the bands, not their pattern, may be influenced, analogously with the effect of the alkaline earth metals on the emission spectrum of copper. The Graphic-mathematical Analysis. It must be frankly stated that this is an empirical approach, and it is recognized that the solutions yielded by the adopted analytical procedure are influenced by underlying assumptions. 168 David L. Drabkin The novel use of curves of the normal frequency form — admittedly agree- able to the writer — need not imply that absorption bands actually possess this shape. Yet, made up as they are (in different spectral regions) of an electronic band, with blurred out vibrational or rotational elements, their shapes may really be similar to those employed in the graphical resolution. It is hoped that methods will be forthcoming which may permit an experimental resolu- tion of the spectra of haemin derivatives at least, if not those of the haemin proteins. Among the possibilities are the spectra at very low temperatures in which interest has been renewed, and which were originally explored in porphyrins by Conant and his colleagues (Conant and Kamerling, 1931). SUMMARY A graphic-mathematical analysis, using curves of the normal frequency form, has been made of the absorption spectra of haemin chromoproteins. The spectra in the near infra-red, visible and ultra-violet regions of all the derivatives examined are fundamentally similar, and represent the summa- tional effect, which can be expressed mathematically, of the a and /5 bands and bands numbers 3 to 11 of an equally spaced frequency distributed series. The bands represent a series of electronic transitions and those in the demonstrated spectral series are inferred to originate from the same molecular structure, the resonating conjugated double bond system of the iron- porphyrin unit. The special significance of the a and /5 bands, the contributions of porphyrin and protein, and the influence of co-ordinating ligands have been discussed. A ckfio 1 vledgemen t The writer's investigations of the chromoproteins have been supported by grants from the Office of Naval Research and the Bureau of Medicine and Surgery of the Navy, and, more recently, by a grant from the National Science Foundation, U.S. REFERENCES Appleby, C. A. & Morton, R. K. (1954). Nature, Lond. 173, 749. Austin, J. H. & Drabkin, D. L. (1935). /. biol. Chem. Ill, 67, Braude, E. a. (1945). Am. Rep. chem. Soc. 42, 105. Erode, W. R. (1928). Proc. Roy. Soc. A, 118, 286. Calvin, M. & Wilson, K. W. (1945). /. Amei: chem. Soc. 67, 2003. Chatt, J. (1949). /. chem. Soc. 3340. Conant, J. B. & Kamerling, S. E. (1931). /. Amer. chem. Soc. 53, 3522. Coryell, C. D., Stitt, F. & Pauling, L. (1937). /. Amer. chem. Soc. 59, 633. Davies, T. H. (1940). /. biol. Chem. 135, 597. Dhere, C. (1906). Compt. rend. Soc. biol. 61, 718. Drabkin, D. L. (1934). Proc. Soc. exp. Biol. Med. 32, 456. Drabkin, D. L. (1936). J. biol. Chem. 114, xxvii. Drabkin, D. L. (1937). J. biol. Chem. 119, xxvi. Drabkin, D. L. (1938). Proceedings of the Fifth Summer Conference on Spectroscopy and Its Applications, p. 94. John Wiley, New York/Chapman & Hall, London. Analysis ami Interpretation of Absorption Spectra of Haemin Chromoproteins 169 Drabkin, D. L. & Singer, R. B. (1939). /. biol. Chem. 129, 739. Drabkin, D. L. (1940). Proceedings of tlie Seventh Summer Conference on Spectroscopy and Its Applications, p. 116. John Wiley, New York/Chapman & Hall, London. Drabkjn, D. L. (1941a). J. opt. Soc. Amer. 31, 70. Drabkin, D. L. (1941b). J. biol. Chem. 140, 387. Drabkin, D. L. (1942a). J. biol. Chem. 142, 855. Drabkin, D. L. (1942b). J. biol. Chem. 146, 605. Drabicin, D. L. & Schmidt, C. F. (1945). /. biol. Chem. 157, 69. Drabkin, D. L. (1946). /. biol. Chem. 164, 703. Drabkin, D. L. (1949a). Arch. Biochem. 21, 224. Drabkin, D. L. (1949b). Haemoglobin, p. 35 (Ed. by F, J. W. Roughton & J. C. Kendrew), Butterworths, London. Drabkin, D. L. (1950). Medical Physics, 2, p. 1039 (Ed. by O. Glasser), Year Book Publ., Inc., Chicago. Drabkin, D. L. (1954). Report to Ad Hoc Panel of the National Research Council (U.S.) on a Standard for Haemoglobin Measurement. Unpublished. Drabkin, D. L. (1957). Fed. Proc. 16, 740. Gordy, E. & Drabkin, D. L. (1957). /. biol. Chem. Ill, 285. Gould, R. K. & Vosburgh, W. C. (1942). J. Amer. chem. Soc. 64, 1630. Grabe, H. (1892). Untersuchungen des Blutfarbstoffes auf sein Absorptionsvermogen fiir violette und ultraviolette Strahlen, Dorpat. Hagenbach, a. & Percy, R. (1922). Helv. chim. Acta, 5, 454. Harrison, G. R., Lord, R. C. & Loofbourow, J. R. (1948). Practical Spectroscopy, Chaps. 10 and 1 1 (pp. 228-299), Prentice-Hall, New York. Hartridge, H. (1920). /. Physiol. 54, 253. Harvey, A. E. Jr. & Manning, D. L. (1952). /. Amer. chem. Soc. 74, 4744. Haurowitz, F. (1924). Hoppe-Seyl. Z. 138, 68. Haurowitz, F. (1927). Hoppe-Seyl. Z. 169, 235. Hausser, K. W. (1934). Z. tech. Physik. 15, 10. Hausser, K. W., Kuhn, R. & Seitz, G. (1935a). Z.phys. Chem. B. 29, 391. Hausser, K. W., Kuhn, R. & Smakula, A. (1935b). Z.phys. Chem. B. 29, 384. Hausser, K. W., Kuhn, R., Smakula, A. & Deutsch, A. (1935c). Z.phys. Chem. B. 29, 378. Hausser, K. W., Kuhn, R., Smakula, A. & Hoffer, M. (1935d). Z.phys. Chem. B. 29, 371. Hausser, K. W., Kuhn, R., Smakula, A. & Kreuchen, K. H. (1935e). Z.phys. Chem. B. 29, 363. Henri, V. (1919). Etudes de Photochimie, Paris. Henri, V. (1923a). C.R. Acad. Sci., Paris, 176, 1142. Henri, V. (1923b). C.R. Acad. Sci., Paris, 111, 1037. Hicks, C. S. & Holden, H. F. (1929). Aust. J. exptl. Biol. med. Sci. 6, 175. Hogness, T. R., Zscheile, F. R. Jr., Sidwell, A. E. Jr. & Barron, E. S. G. (1937). /. biol. Chem. 118, 1. Holiday, E. R. (1936). Biochem. J. 30, 1795. Holiday, E. R. (1937). /. sci. lustrum. 14, 166. Horecker, B. L. (1943). J. biol. Chem. 148, 173. Krumholz, p. (1953). J. Amer. chem. Soc. 75, 2163. Lavin, G. L & Northrop, J. H. (1935). /. Amer. chem. Soc. 57, 874. Lavin, G. I., Northrop, J. H. & Taylor, H. S. (1933). /. Amer. chem. Soc. 55, 3497. Lifschitz, J. (1920). Z.phys. Chem. 95, 1. Morton, R. K. (1958). Rev. pure appl. Chem. 8, 161. Pauling, L. (1940). The Nature of the Chemical Bond, and the Structure of Molecules and Crystals, 2nd ed., Cornell University Press, Ithaca/Oxford University Press, London. Pauling, L. (1948). The Valences of the Transition Elements, Victor Henri Memorial Volume. Lidge: Desoer. Pauling, L. (1949). Haemoglobin, p. 57 (Ed. by F. J. W. Roughton and J. C. Kendrew), Butterworths, London. 1 70 Discussion Pauling, L. & Coryell, C. D. (1936a). Proc. nat. Acad. Sci. Wash. 22, 159. Pauling, L. &. Coryell, C. D. (1936b). Proc. nat. Acad. Sci. Wash. 22, 210. Rawlinson, W. a. (1940). Aust. J. exp. Biol. med. Sci. 18, 185. SoNE, K. (1952). Bull. chem. Sac. Japan, 25, 1. SoRET, J. L. (1878). Arch, sc.phys. et nat. 61, 322. SoRET, J. L. (1883a). Arch, sc.phys. et nat. ser. 3, 9, 513. SoRET, J. L. (1883b). Arch. sc. phys. et nat. ser. 3, 10, 429. Theorell, H. (1938). Biochem. Z. 298, 242. Theorell, H. (1939). Biochem. Z. 301, 201. Theorell, H. (1941). /. Amer. chem. Soc. 63, 1820. Theorell, H. & Akesson, A. (1941). J. Amer. chem. Soc. 63, 1812, 1818. Theorell, H. & Nygaard, A. P. (1954). Acta chem. Scand. 8, 1649. Tint, H. & Reiss, W. (1950). J. biol. Chem. 182, 385, 397. Tsou, C. L. & Li, W. C. (1956). Scientia Sinica. 5, 253. Vles, F. (1914). Compt. rend. Soc. biol. 158, 1206. Warburg, O. (1929), Naturwissenschaften 16, 245. Warburg, O. & Negelein, E. (1928). Biochem. Z. 200, 414. Williams, R. J. P. (1956). Chem. Rev. 56, 299. Wyman, J. Jr. & Ingalls, E. N. (1941). /. biol. Chem. 139, 877. DISCUSSION Interpretations of Absorption Spectra of Haemoproteins Perrin: Would Drabkin please indicate the theoretical significance he attaches to his serial band analysis? In, say. Fig. 9 of his paper (p. 155), how much latitude does his assigning of seven bands, with the resulting sixteen adjustable constants, allow in analysing the spectrum? Drabkin: I have frankly admitted in my paper that my analytical approach was an empirical one. My proposal was originally made a good many years before high- and low-spin complexes were recognized and before it was fashionable to speak of d- and 7T-electrons. It seemed to me very worthwhile, as a first step, to make the most simplifying assumptions. I used Gaussian curves because I hked their shape, and perhaps naively believed that the components (bands) in the complex absorption curves might indeed have such a form. I may say that I anticipated Perrin's searching questions, and posed them to myself, without a wholly satisfactory answer. I recognize that from a rigid physical viewpoint the bands are spaced too far apart to be regarded as electronic in the usual sense, and probably represent a special case. As to the latitude of the analytical procedure, it is probably rather broad. For the pronounced maxima in the absorption spectra they would appear to be unequivocal. For regions of masked absorption the solution may not be unique. However, reference to my Table 2 (p. 153) will disclose that assembling the various derivatives, actual representatives with definite maxima in the postulated locations have been found. It should be stressed that this is the important experimental finding, independent of and, indeed, guiding the analysis. Yet, the proposal of the equally spaced, frequency- distributed series was made before all the data were available. The possible existence of certain maxima was prophesied, a prophecy fulfilled by later finding them in the spectra of oxidized and reduced cytochrome c. Williams: My views and those of Drabkin as to the nature of iron porphyrin spectra are rather different. Drabkin has given us a detailed survey, empirically based, of a large number of absorption bands. This will be most valuable. I have attempted to use Piatt's theory {Radiation Biology, 3, 1956) of the porphyrin spectra to interpret the spectra of metal porphyrin complexes. The analysis led to the conclusion that some bonds in Fe+++ porphyrins were due to the iron and had little to do with the porphyrin, notably at 650 m/f and about 500 m/f. In Fe++ porphyrins there is a band at about 500 m/< due to the iron alone. There is no requirement for a frequency series in Piatt's theory. Analysis and Interpretation of Absorption Spectra of Haemin Chromoproteins 171 Drabkin: Williams has suggested that the band at 500 m/i, which is No. 5 of my frequency distributed series and ascribed by me to originate in the dynamic haemin structure, is rather owing to iron itself. Having earlier discussed this question with him, I believe his deduction originates from finding a maximum at about 510 m/t in such complexes as ferrous orthophenanthroline (see my Fig. 2). This is a very broad maximum and doubtless includes several component bands. Of course my proposal also involves a co-ordination complex. I must respect his theoretical knowledge, but naturally I prefer my own proposal. At any rate, I believe we agree on many other aspects of our individual analytical approaches. Thus, we have brought out somewhat similar correlations between the absorption spectra and corresponding magnetometric data (see my Table 3 and text, p. 158). It may be pointed out that, in Table 3, in the case of ferrihaemoglobin and ferri- haemoglobin hydroxide, one may now substitute the terms 'high-spin' and 'mixture of high-spin and low-spin' in place of the older 'essentially' or 'partially' ionic (see Orgel's, Williams', and George's papers, this symposium). I wonder whether it is possible that the unexpected behaviour in the vicinity of the Soret band (my band No. 6) in the transition spectra of MHb -> MHbOH (i.e. the absence of an isosbestic point) may be related to a change from a high-spin complex to a mixture of high- and low-spin ? Wainio: I wish to ask Drabkin, if the a- and )5-bands of the haem of any one of the haem- proteins is replaced by the bands of the corresponding porphyrin, will the latter fall into the frequency distribution series? Drabkin: In the ultra-violet region they do not fit well. Since I was making the most simplifying assumption, namely that the spectra are an expression of iron in a co- ordinated complex, I was content to start with, and confine myself to iron-porphyrin complexes. I am not at all convinced that there is real validity in drawing parallels between porphyrin bands and correspondingly located bands of haemin derivatives. The bands in the region of 830 and 280 m/x George: I should like to ask Drabkin whether the near infra-red bands of the ferri- myoglobin fluoride complex at 740 and 830 m/t can be fitted into the frequency series with successive values of 'rt'. If the wave-number separation is too small then one or the other would have to be regarded as belonging to another category like the a- and /S-bands. Drabkin: The band at 830 m/i is present in other haemin protein derivatives (as evident in Table 2 of my paper). This band corresponds to v x 10""^ = '-^120, and is No. 3 of my proposed equally-spaced frequency series. The band at 740 m/< may correspond with the at present anomalous band at 760 m/z of deoxygenated haemoglobin (see my Fig. 1), and may in fact belong to a different series, or may be 'odd-man-out' as you have expressed it. Incidentally, in some derivatives (see Fig. 1, p. 143) there are maxima in the vicinity of wavelength 920 m/n which do not fit in the main series. This is brought out in my paper. George: For myoglobin derivatives the millimolar extinction coefficients in the protein absorption region, 260-290 m/<, are about 30 compared to the value of about 13 for apomyoglobin. Similar values have been reported for haemoglobin derivatives and apohaemoglobin, hence there can be no doubt that the haem absorbs significantly in this region. The same is true of vitamin B12 and free benzimidazole, so the eff'ect is not restricted to haemoproteins. Margoliash: In the case of cytochrome c, for which the amino acid composition is reasonably well known, it is easy to calculate the contribution of tryptophane, phenyl- alanine and tyrosine to the 280 m/i band. The results are roughly the same as those quoted by George for metmyoglobin. There is, moreover, a striking difference between the spectra of the ferro- and ferri- forms of cytochrome c in the 'protein' band region, indicating a distinct contribution of haem absorption to this band (Margoliash and Frohwirt, Biocliem. J. 71, 570, 1959). H.E. — ^VOL. I — N 172 Discussion PosTGATE : I think there exists some evidence against Drabkin's view that the 280 m/n peak in haematins is mainly due to some frequency in the iron-porphyrin system and not to residues of aromatic amino-acids. Cytochrome C3, which has a molecular weight closely similar to that of cytochrome c, has two haemins/molecule. Yet in spite of this the absorption of this material at 280 m/i is negligibly small; the 280 m/f peak of cytochrome c is absent. On the other hand, we find from qualitative observations that the aromatic residue content of c^ is very small, which would be consistent with the traditional view of the significance of the 280 m/t band. Morton: As will be seen from the tables in my review (Morton, Rev. pure appl. Chem. 8, 161, 1958), there is a paucity of information concerning the influence of the state of oxidation on the absorption of cytochromes in the ultra-violet region. Our studies with cytochrome b.^ suggest that the position and height of the band in the 260-280 m/i region does change between the oxidized and the reduced compound. Could Postgate comment on the diff"erence between ferricytochrome c^ and ferrocytochrome C3 in the 280 m/< region. Postgate: We have never observed the 280 m/t range of ferrocytochrome c^; because of its low Eq we have found no reducing agent which will maintain the reduced form without absorbing strongly in this range. Drabkin: The situation Postgate describes in cytochrome C3 is certainly most unusual, perhaps unique for haemin chromoproteins. I would not be astonished by the absence of a well-defined maximum, but am puzzled by negligible absorption, particularly since co-ordination complexes such as haemin dicyanide (my Fig. 5) absorb rather strongly in this region. Perhaps the presence of a weak masked band could be brought out by the type of analysis I have employed. What is the nature of the ultra-violet absorption of reduced cytochrome Cg? That approximately 30% only of the total absorption at 280 m/t can be ascribed to the specific absorption of the aromatic amino acids in most haemoproteins is in essential agreement with my own deduction, based upon the content of tyrosine, phenylalanine and tryptophane, as discussed in my paper. I believe that these aromatic amino acids contribute to the absorption, but the main contribution is owing to the haemin structure. It is misleading in the case of the haemin derivatives to speak of the protein band (at 280 m/t). Williams: I wonder should not one re-examine, in respect to the absorption at 280 m/t, the question of energy transfer from aromatic amino-acids to haemoproteins, e.g. in the photochemical decomposition of CO-complexed haemoproteins (Weber, Disc. Faraday Soc. 11, 1959). As I understand Drabkin's remarks, there is a co-operative enhancement of the absorption of the aromatic amino-acids at 280 m/t by the haem unit. Drabkin: I am sorry that I am not acquainted with the actual experimental work of Weber to which you refer. In any event the energy would have to be quantized. In my paper I do refer to Warburg's classical study, and believe that his photochemical spectrum which includes a broad spectral coverage, with several maxima besides the 'protein' band at 280 m/t, must indicate that the same haemin structure is involved and photochemically eff'ective in spectral regions which cannot be ascribed to protein. This appears to support my proposal of a similar origin for the various bands. On the other hand, the total co-ordination complex includes residues from the protein. Hence, the eff"ect of protein cannot be wholly separated from the haemin, and some protein contribution may be present over the whole spectral range (see discussion in paper). Lemberg: Most porphyrins have, indeed, a weak absorption in the region 260-280 m/t. For protoporphyrin in dioxane we have found an £„,« of 14 at 280 m/t. The absorp- tion is thus far weaker than that of the Soret band, whereas many haemoproteins absorb at 280 m/t as strongly or more strongly than in the Soret region. THE HAEM-GLOBIN LINKAGE 3, The Relationship between Molecular Structure and Physiological Activity of Haemoglobins* By J. E. O'Hagan Red Cross Blood Transfusion Service, Brisbane, Queensland, Australia The reactions of haemoglobins have been interpreted in terms of the imidazole, steric hindrance, haem-haem interaction and intermediate com- pound hypotheses, which have had wide acceptance. However, the opinion that the haem iron was linked by groups other than imidazoles has been expressed by Haurowitz (1954, 1959) and by O'Hagan (1959a). Theorell and Ehrenberg (1951) considered that a more acid group was responsible for this linkage in horse myoglobin, while Wyman (1948) in expounding the imidazole hypothesis, made a reservation that the evidence for the identification as imidazole of the more acid of the groups co-ordinating the iron in horse haemoglobin was not completely certain. Recent X-ray studies of ferri- myoglobin by Kendrew, Bodo, Dintzis, Parrish, Wyckoff and Phillips (1958), neither appear to support a hypothesis of co-ordination of the haem iron between two amino acid side-chains of the protein, nor confirm steric hindrance relationships. Roughton (1944), while paying tribute to the value of Wyman's work, pointed out that it did not account for the important carbamino reaction. Strong arguments against the steric hindrance (or embedded haem) concept were presented by Keilin (1953), and these were supported to a certain extent by the preparation of artificial haemoglobins from haems of dimensions larger than protohaem (namely coprohaem III and tetramethyl-coprohaem III (O'Hagan, 1955, I960)). George and Lyster (1958), after a careful analysis of the evidence, considered steric hindrance effects unlikely, at least for small ligands, which, after all, are the physiologically important ones. The haem-haem interaction hypothesis as proposed by Pauling (1935) to interpret the sigmoid dissociation curve of oxyhaemoglobins has not been able to account for a number of apparent exceptions. The addition of oxygen to ferrohaemoglobin is linear when measured by either spectro- photometric (Nahas, 1951) or magnetometric methods (Coryell, Pauling and * Part I, O'Hagan; Part 2, O'Hagan and George; Biochem. J., 74, 417, 424 (1960). 173 174 J. E. O'Hagan Dodson, 1939). Spectrophotometric titrations of the combination of imida- zole (Russell and Pauling, 1939) or hydroxyl (George and Hanania, 1953) with ferrihaemoglobin also show linear relationships. None of the equations proposed, on the basis of haem-haem interaction, to explain the sigmoid oxygen dissociation curves, has been found to be satisfactory, except in a special case at pH 9-1, outside the range of the Bohr effect (Roughton, Otis and Lyster, 1955). There appears to be reliable evidence that specimens of mammalian haemo- globins have at times exhibited hyperbolic dissociation curves (Barcroft, 1928; Hartridge and Roughton, 1925). Takashima (1955) found that at ionic strength 0-03-0-3 the «-value for the Hill equation was about 3, which would not be in accordance with Pauhng's equation. At lower ionic strength, pro- nounced deviation from the Hill equation occurred, especially at the lower portion of the curves. Rossi-Fanelli, Antonini and Caputo (1959) found that human haemoglobin in 2 M sodium or potassium chloride (under which conditions it is dissociated into half molecules), gave an oxygen dissociation curve with slightly increased 'haem-haem interaction'. They did not show that the four haems were divided between the two fragments but if they were, as is most probably the case, their results could mean that the sigmoid curves of haemoglobins were not due to interaction between the haems. Gastrophilus haemoglobin with two haems per molecule has a hyperbolic oxygen dissociation curve, i.e. no 'haem-haem interaction' (KeiUn and Wang, 1946). The sigmoid dissociation curve of diluted chlorocruorin (Fox, 1932) of molecular weight of about 3,000,000, with the possibility of inter- action between about 200 haems (Lemberg and Legge, 1949), and the 'atypical' curves of haemoglobins of species such as the duck and carp (Red- field, 1933), are difficult to reconcile with this hypothesis and have generally been conveniently ignored. Some experimental support for the intermediate compound hypothesis would appear to have been found from the work of Itano and Robinson (1956) who detected intermediates when normal human adult carboxy- haemoglobin was partly oxidized with ferricyanide and the mixture submitted to electrophoretic separation. It does not necessarily follow, however, that their findings are appUcable to the ferrohaemoglobin-oxyhaemoglobin system. The finding by Hill and Holden (1926), Holden (1941) and Granick (1949) of attachment of porphyrins to apohaemoglobins, the instability of aetio- haemoglobin (O'Hagan, 1950, 1955, 1960) and the X-ray studies of ferri- myoglobin (Kendrew et al., 1958) suggested the likehhood of linkages between the haematin propionate side-chains and basic side-chains of the apohaemo- globins and apomyoglobins. While studying the nature of these linkages it appeared that it might be possible to explain more satisfactorily the oxygen dissociation curves, the Bohr effect, the alkali-stabihty, and the differences The Haem-Globin Linkage 1 75 in crystal structure in terms of such linkages. They have therefore been examined with this in view and, as a result, a new interpretation of the structural and functional relationships of haemoglobins and myoglobins is presented. MATERIALS AND METHODS Aetiohaemin III The preparation was as described by O'Hagan (1960). Nickel Mesoporphyrin IX Mesoporphyrin IX was prepared by the method of Grinstein and Watson (1943) from dimethylprotoporphyrin IX (Grinstein, 1947). In 5% w/v HCl the absorption spectrum had bands at I, 590-7; la, 570-5; II, 547-5 m// (order of intensity II > I > la, with the Hartridge Reversion Spectroscope). Bands I and la were 1 m/i lower than those reported by Fischer and Orth (1937). Two methods of conversion to the nickel complex were employed. The first followed the technique used by Fischer and Piitzer (1926) to prepare nickel protoporphyrin, but the yield of crystals, even after standing several days in the refrigerator, was poor. The second, a very simple method of preparation, was as follows. About 0-1 g mesoporphyrin was dissolved in about 5 ml acetic acid, the solution was quickly heated to boiling and nickel acetate in acetic acid (prepared by leaving a piece of pure nickel, just covered with acetic acid, in a beaker for a few days) added drop by drop, with continued boiling, until the fluorescence of the porphyrin under ultra-violet light had disappeared. After cooling and adding \ vol. distilled water, the precipitated nickel mesoporphyrin was centrifuged off, washed with alcohol and ether and dried in a vacuum desiccator over NaOH. It was then dissolved in 0-04 N NaOH, the solution centrifuged to remove a trace of undissolved material, the pigment in the supernatant precipitated with 0-2 N HCl, the precipitate centrifuged off, washed several times with distilled water and dried in a vacuum desiccator over NaOH. The nickel mesoporphyrin prepared by both methods had absorption peaks (Hartridge Reversion Spectroscope) as follows: in dioxane, I, 550-5; II, 513 m/< (order of intensity I> II, cf. Lemberg and Legge, 1949); in pyridine, I, 552-0; II, 514 m/* (order I > II). Proteins The horse apohaemoglobin was prepared by the method previously described for the human material (O'Hagan, 1960) and the apomyoglobin as reported by O'Hagan and George (1959), and both were estimated spectro- photometrically at 280 m// using e^^ = 13 as calculated by Hanania (1953). Before use, the solutions were left for several days at 1°C after adjustment to pH 7-8 to remove as much denatured material as possible. The 25 % human serum albumin was a special batch of salt-poor albumin for transfusion 176 J. E. O'Hagan purposes which had not been subjected to the usual heat treatment to destroy hepatitis virus. Imidazoles Caffeine B.P. and theophyUine B.P. were suppHed by Drug Houses of AustraHa Ltd. They were recrystalhzed from water and used as saturated aqueous solutions. Bujfer Solutions These were prepared with British Drug Houses A.R. grade chemicals from tables calculated by George and Hanania (unpubhshed), and kindly supplied by Professor P. George. The buffers were of constant ionic strength (I = 0.05) and of the following composition: pH 2-0-3-8, HCl + KH phthalate; pH 4-0-6-2, NaOH + KH phthalate + NaCl; pH 5-6-8-0, NaOH + NaH2P04 + NaCl; pH 7- 5-9- 5 HCl + Na4Po04 + NaCl; pH 9-9-1 M, NaOH + glycine + NaCl; pH 11-0-12-0, Na'gHPO^ + NaOH + NaCl; pH 130, NaOH. Dithionite 1 % w/v solutions were prepared immediately before use from sodium hydrosulphite B.D.H., which was taken from a freshly opened ampoule (repacked from a 500 g bottle). Standardization of Instruments The Hilger Uvispek Spectrophotometer, Beck Hartridge Reversion Spectroscope and Jones Electronic pH Meter were standardized as previously described (O'Hagan, 1960). All pH measurements were made with the standard glass electrode, and, although corrections were applied, all readings at high pH should not be regarded as exact. RESULTS Decrease in the Acid Strength of Haematin Propionate Groups on Reduction to Haem It appeared possible that since substituents of R in RCH2CH2COOH could alter the pAT value of the carboxyl by as much as 1-3 units (Edsall and Wyman, 1958), a change in the electronic structure of the haem iron atom producing alterations in the high resonance of the porphyrin ring system might have the same effect as a substitution of R in simple com- pounds, with subsequent change in the acid strength of the propionate groups. To test this hypothesis two reactions known to involve the propionate groups were investigated, those with human serum albumin, and with caffeine. A new reaction with theophylline was found, and prehminary results obtained supported those observed with caffeine. The Haem-Globin Linkage 1 77 Attachment of Haematin and Haem to Human Serum Albumin Haematin combines with human serum albumin to form ferrihaemalbumin (see discussion by Lemberg and Legge, 1949). J. Keilin (1944) considered that the attachment of both haematin and haem to the albumin was through the porphyrin and Lemberg and Legge indicated the haematin carboxyls as being most hkely involved. O'Hagan (1955, 1960) in showing by both spectrophotometric and paper electrophoretic studies that mesohaematin (but not aetiohaematin) combined, demonstrated that the carboxyls were responsible. Keilin (1944) believed it most likely that haem is attached in similar fashion, on account of the spectral differences of ferrohaemalbumin from free haem and from the haemochromes. The extent of the combination of human serum albumin with haematin and haem was investigated by measuring the increment in absorbance in the Soret region on addition of the pigments to excess of the albumin in a series of buffer solutions of constant ionic strength. Tubes of 7 ml capacity were filled with 5 ml buffer (phthalate or phosphate), 1 ml 2-5% human serum albumin (the 25% solution diluted 1:10 with distilled water) and 0-1 ml 2x 10~^M freshly prepared protohaematin solutions (1-30 mg haemin dissolved in 1 ml 005 n NaOH, then 9 ml distilled water added). Other sets of tubes, one containing water in place of albumin and another with albumin but no haematin were set up at the same time. The solutions were stood at 21° for 3 hr, then portions transferred to a 10 mm cuvette and the absorbances read at 404 m/f in the spectrophotometer. Another series of three sets of tubes had 0- 1 ml freshly prepared 1 % w/v dithionite added to each tube and the tubes closed by long stoppers which excluded air almost completely, except for a small bubble which assisted mixing on inversion. After the 3 hr standing, portions of these solutions were carefully transferred with a Pasteur pipette and as little agitation as possible, to the cuvette and the absorbances read at 414 m/i, the Soret peak of ferrohaemalbumin. The curves obtained in Fig. 1 show the increment in absorbance due to the addition of the albumin, i.e. they represent Aj^^ — A^ — A^, where ha = ferri- or ferrohaemalbumin, h = haematin or haem and a = albumin. The difference in the attachment of the haematin and haem, as indicated by the increment in absorbance, is at once apparent. Similar results were obtained for the ferrohaemalbumin when the dithionite was added to the ferrihaemalbumin after it had stood for 3 hr and the mixture stood a further 3 hr. No trace of verdohaem compounds was detected, but these were formed, as expected, when the reduced solutions were reoxidized, so that the reverse reaction (ferro- to ferrihaemalbumin) could not be investigated under these conditions. To check whether the dithionite itself interfered in any way, nickel mesoporphyrin (on which dithionite has no effect) was added to the albumin and the absorbance increment measured before and after addition 178 J. E. O'Hagan of the dithionite. The first curve was obtained after standing 3 hr, dithionite added at the same concentration as for the iron porphyrins, and the solutions stood another 3 hr to give the second curve. The shght difference is due to the Fig. 1 . Attachment of metalloporphyrins to human serum albumin. (a) Soret absorbance increment (A) curves prepared, as described in the text, for haematin + human serum albumin at 404 m/n • — • — •, for haem + human serum albumin at 414 m/n O— O — O. (b) Soret peak absorbance increment curves for nickel mesoporphyrin + human serum albumin at 394 m/t before A — ▲ — ▲ and after ■ — ■ — b adding di- thionite. (Buffers, pH 4-0-6-2 phthalate, pH 6-2-7-2 phosphate, I = 0-05, T = 2rc.) decreased value for the metalloporphyrin (without albumin) which is not in true solution and whose absorbance is decreasing with time. It was concluded that reduction significantly decreased the acid strength of at least one of the haematin propionate groups. Attachment of Haematin and Haem to Caffeine Caffeine was found by J. Keilin (1943) to combine with copper uroporphyrin III and with manganese mesoporphyrin, but she detected no reaction between haematin and caffeine. This seemed unusual and O'Hagan and George (unpubhshed, quoted by O'Hagan and Barnett, 1958) found attachment at pH 11-3 (1-33 mol of caffeine/mol of haematin) and also at pH 7-0 (stoichio- metric relationship not determined). This suggested stronger attachment of haematin than haem to caffeine since J. Keilin had found at least 20 mol of caffeine/mol of haem to be required for caffeine-haem formation at high pH. A saturated solution of caffeine (about 10~i m) was substituted for the albumin in the experiments reported above and the absorption increments plotted as shown in Fig. 2. The peaks for the ferrihaemcaff'eine and ferro- haemcafifeine were 402 and 420 mju respectively. Nickel mesoporphyrin also The Haem-Globin Linkage 179 showed absorption increments on adding caffeine, with Soret peak at 392-5 mn and attachment occurring from about pH 6-0, rising to a maximum at pH 7-0, and being unaffected by addition of dithionite. Fig. 2. Attachment of haematin and haem to caffeine. Soret absorbance incre- ment curves prepared as described in the text, for haematin + caffeine at 402 m/z • — • — •, and haem + caffeine at 420 m/f O — o — o. (Buffers, pH 4-0-6-2 phthalate, pH 6-2-7-5 phosphate, I = 0-05, T = 2rc.) Preliminary Experiments with Theophylline Theophylline has an unsubstituted imino group and therefore resembles more closely the imidazole ring as it would be expected to occur as the side- chain of proteins. In a single series of experiments with theophylHne, attach- ment of haematin occurred from about pH 5, with a pronounced maximum at pH 6-5 and a minimum at about pH 7-3, but no attachment of haem was detected in this range. At lower hydrogen ion concentration, J. Keilin (1943) also had not found combination of haem with theophylline, though she did with caffeine. These experiments with the imidazoles are prehminary; more points should be obtained to plot the increment curves exactly, but if the curves are nearly correct, a mean increase of propionate pA^ value of about 0-9 unit could be indicated on reduction, of the same order as that suggested from the experiments with the albumin compounds. Further studies on the reactions of haematins with imidazoles will be reported later. Haem Propionate - Protein Linkages in Adult Horse Haemoglobin and Horse Myoglobin Nickel porphyrins would appear to be very useful compounds for the study of the attachment of haem propionate groups to the side-chains of proteins. They would be expected to be of almost identical size and shape to the iron porphyrins. Haurowitz and Klemm (1935) found nickel dimethyl- mesoporphyrin and Pauling and Coryell (1936) found nickel protoporphyrin to be diamagnetic and concluded that these porphyrins contained no unpaired 180 J. E. O'Hagan electrons. This means that neither the pyrrole nitrogens nor the metal atom are capable of further combination, only reactive side-chains can effect attachment to other compounds. The covalent linkage of the metal resembles, too, that of the iron in oxy- and carboxyhaemoglobins, and the resonance state of these metalloporphyrins might be expected to be akin to that of the haem in those proteins. In the studies reported here, nickel mesoporphyrin was employed, both because it was found to be more readily prepared in pure form than the corresponding protoporphyrin complex and because its use eliminated con- fusion in interpretation of results, through possible (though unhkely) linkage to protein through vinyl groups. Hill and Holden (1926) had shown that it combined with ox apohaemoglobin, as also did nickel protoporphyrin (Holden, 1941). To investigate attachment, 10 ml of buffer was placed into each of a series of tubes, then 0-1 ml 5x 10"*M apohaemoglobin solution (assumed M.W. = 16,500) and 0-1 ml 1 x 10^^ m nickel mesoporphyrin (1-34 mg pigment + 2 ml 0-05 n NaOH + 8 ml water) added. Other series replacing the apohaemoglobin solution or nickel mesoporphyrin solution with water were prepared at the same time and the tubes containing the three series stood at 1° for 16 hr and then at 21° for 2 hr before reading absorbances at 389 mfi, using a cuvette of 40 mm path length. For the apomyoglobin experiments the undiluted protein solution used was 2 x 10~^ m and the absorbances read at 410 mfi. Excess apoprotein was used because of the likehhood of a small variable quantity of denatured protein being present (O'Hagan, 1960), not removable by any treatment yet described, and likely to precipitate on bringing solutions to room temperature. In order to rule out 'protective colloid' effects or non-specific attachment, carboxyhaemoglobin and ferrimyoglobin were substituted for the apoproteins in the experiments. A small peak centred about pH 9 was found with the haemoglobin, while no attachment was detected with the myoglobin. The results, shown in Fig. 3, indicate that attachment to the apohaemoglobin occurs over the pH range 5-12 with two maxima at about pH 7-4 and 10-0. At pH 9-8 nickel mesoporphyrin had an absorption peak at 380 m/;, intensi- fying and moving to 389 m/t (pH 5-7, 8-0, 9-95) on addition of apohaemoglobin (cf. with caffeine, 392-5 m/u). With apomyoglobin the stability range was wider and the peak of the absorption curve shifted much further to 410 m/<. A difference in the position of the peaks with apohaemoglobin and apomyo- globin is in accord with the finding of differences by Hill (1939) when proto- porphyrin was added to these proteins. The section of the curve with maximum centred at about pH 7-4 for the apohaemoglobin complex is strongly suggestive of imidazolium combination with one or both of the propionates. The other section with maximum at about pH 10-0 varied in height and width with the preparation and is most The Haem-Globin Linkage 181 probably due to a combination with a group in some denatured protein present. Perhaps this group results from unmasking of the one giving the maximum attachment at pH 9 in the unsplit native protein. It might also be the group detected by Theorell (1942) in apohaemoglobin, with pA: of 10 at 0°C, not present in ferrihaemoglobin. The increment curves presented here should not be considered to be exactly reproducible since comparison between solutions and colloidal sus- pensions is being made. The increments are, however, of such magnitude 0-5 - . . - / ^ 0-3 - ^ \ i /"\ l\ \ /\ Y f V > 0-1 - 1; \ \ / ^ ^--\ ^^ V -'r'T:>i L,.X 1 1 1 1 1 pH 10 12 Fig. 3. Attachment of nickel mesoporphyrin to apohaemoglobin, apomyoglobin and carboxyhaemoglobin. Soret absorbance increment curves prepared, as described in the text, for nickel mesoporphyrin + apo Hb at 389 m/i 9 — • — •, + apo Mb at 410 m/i O— O— O, and + CO Hb at 390 m/ii A— A— A. (Buffers, pH 2-6-2 phthalate, 6-2-8-0, phosphate; 8-5-9-5, pyrophosphate; 9-9-11, glycine; 11-5, phosphate; 12-5, NaOH; I = 0-05, T = 2rC.) (e.g. apomyoglobin increased the absorbance of nickel mesoporphyrin at 410 m/t from 0-22 to 0-67 at pH 10-0) that it seems legitimate to make quantitative comparison. The technique should prove useful in the detection and identification of linking groups in other haemoproteins. The Nature of the Acid Groups Linking the Haem Iron The likelihood of combination of haematin propionate side-chains with imidazolium side-chains of horse haemoglobin called for a re-examination of the evidence for the mode of attachment of the iron atom to the protein. Stability curves of ferrihaemoglobin and ferrimyoglobin reported by O'Hagan (1959a) suggested the possibility that groups of ^K value lower than 5-3 were involved. Theorell and Ehrenberg (1951), after their exhaustive study of myoglobin, concluded that a group of more negative character than an imidazole was responsible for iron Unkage, but gave the group a pA^ value of 5-3. Since Coryell, Stitt and Pauling (1937) had shown that in acid ferri- haemoglobin the atom of iron was joined to other atoms surrounding it by 182 J. E. O'Hagan 'essentially ionic' bonds, it could be assumed that as the haematin was being removed at increasing acidity the reaction could, for practical purposes, be regarded as ionic. Since the haematin would be likely to have a tendency to polymerize, with drop in absorbance, it might well act as an 'indicator' of the suppression of the ionization of the group ligating it. To rule out the effects of linkages through the haematin propionate groups, aetiomyoglobin, the properties of which have been described by O'Hagan and 0-5- 0-3- 01 3 4 5 6 7 8 pH Fig. 4. Soret absorbance increment curve for ferriaetiomyoglobin at 393-5 m/n prepared as described in text, o — O — O phosphate buffer, • — • — • phthalate buffer, I ppte., theoretical curve for acid with pK = 5-3, ■ — ■ — H curve for nickel mesoporphyrin + same cone, of apomyoglobin for comparison. (Buffers, I = 0-05, T = 21°C.) .- ; i i / / / / / 7 George (1959) was utilized. It was prepared by adding 0-5 ml of 3 X 10~^ m aetiohaemin in methanol to 5-5 ml of 2-9 X 10~*m apomyoglobin at pH 6-5 (no added buffer) and standing at 1°C overnight. A solution prepared by adding 0-5 ml aetiohaemin solution to 5-5 ml distilled water was treated throughout in the same manner as the aetiohaemin-apomyoglobin solution, to act as a control. Next day the solutions were spun at 15,000^ for 15 min (to remove free aetiohaematin) and the supernatant added in 0-2 ml ahquots to 5 ml portions of buffer solution. After standing 6 hr at 1°C and 3 hr at 21°C the absorbances were read (10 mm cuvette). The curve showing the absorbance increment due to the apomyoglobin is shown in Fig. 4. The nature of the buffer, phthalate or phosphate, made little difference to the shape of the curve in the range pH 5-5-6-8. At about pH 4-95 a discontinuity in the curve appeared, probably due to the taking up of the aetiohaematin by carboxyl groups liberated in the protein. The curve drawn through the points represents a theoretical curve for a group ionizing with pK = 4-95, and for comparison curves for a group with pK = 5-3 and for the attachment of nickel mesoporphyrin to apomyoglobin are given. While by no means The Haem-Globin Linkage 183 conclusive, these studies could indicate the existence of a group of pA!" less than 5-3, if the aetiohaemin is behaving as a weak base (it does not attach to apomyoglobin above pH 8-0) and is acting as an indicator of the ionization of the haem-linked group of the protein. These studies are being extended to apohaemoglobin combination. DISCUSSION That a change occurs in the acid strength of the haematin propionate groups on reduction of the haematin iron is not unexpected, since it has been well established that variation of the side-chains influences the oxidation- reduction potentials of ferro-ferrihaemochrome systems (Lemberg and Legge, 1949). If alterations to the side-chains can affect the reactions of the iron atom, it is not unreasonable to expect that modifications to the electronic structure of the iron could alter the acid strength of ionizable side-chains. The parallelism between changes of the oxidation-reduction potential at 50% oxidation and log/?, where /j is the oxygenation pressure at 50% satura- tion of haemoglobin, with alteration in hydrogen ion concentration, has been clearly demonstrated (Wyman and Ingalls, 1941; see Lemberg and Legge, 1949). We may therefore infer that oxygenation affects the acid strength of the propionate groups in much the same way as does oxidation of the iron. It was first suggested by Altschul and Hogness (1939) that, on oxygenation, a change occurred in the acid strength of groups of the haem rather than of those of the apoprotein. They considered it very probable that the two carboxyl groups of each haem were influenced by oxygenation. They calcu- lated the pAT values of these carboxyl groups as shown in Table 1. The Table 1. Calculated pA" values of haem propionate groups IN FERRO- and OXYHAEMOGLOBIN (AlTSCHUL AND HoGNESS, 1939) Assumed no. of acid groups Calculated pAT values* affected oxygenated reduced ^^K 1 2 5-5 5-8 6-9 6-5 1-4 0-7 * Temperature unspecified, presumably 25°C. actual values may lie somewhere between the extreme values of 5- 5-6-9 since the two acid groups may not be altered in an exactly uniform manner, since the dissociation constants may differ as do those of dibasic acids. Examination of the haem structure shows that the vinyl groups at positions 2 and 4 give asymmetry, which may mean influence of the propionate at 184 J. E. O'Hagan position 6 to a greater or lesser extent than the one at position 7. Considera- tion of the grouping together of rings 1 and 4 and 2 and 3 seems justified, and there may be shared resonance between them as pairs, since on rupture the ring sphts first at the a position to form verdohaems and the tetrapyrrohc ring system of bihrubin sphts at the central methylene group on diazotization (Lemberg and Legge, 1949). This could suggest that one propionate group would be more affected by oxygenation than the other, so that if they were joined by electrostatic hnkages to the apoprotein, one might be a labile link under physiological conditions, while the other only split by such conditions as high ionic strength, high urea or high hydrogen ion concentration, with subsequent parting of the protein molecule into halves. Whatever the actual ipK values, those calculated by Altschul and Hogness show that carbonic acid of pK = 6-352 at 25°C in water (Edsall and Wyman, 1958) could be displaced by a group or groups changing between minimum p^ values of 6-5 and 5-8. The curves of Figs. 1 and 2 could suggest a change of 0-9 unit, compared with a calculated change of 0-7 for two groups and of 1-4 for one group. The asymmetric haem could conceivably be attached to the apoproteins directly or inverted so that structural isomers would be possible unless some orientation by the side-chains occurred. Differences in the acid strength of the two propionates might decide the orientation and also which group could detach from the apoprotein on reduction. The curves obtained for the increment in the absorbance of nickel meso- porphyrin on addition to apohaemoglobin show that a molecule of the size and shape of haem can link by its propionate groups in the pH range 5-9 to a residue in the apohaemoglobin not present in the carboxyhaemoglobin. A specific structure in the proteins binding one or both of the propionates is clearly indicated. That it is not the same as the one binding the iron atom can be deduced from the work with aetiomyoglobin (Fig. 4). The measurements given by Wyman (1948) of the apparent heat of dissociation of horse oxy- haemoglobin and the work reported here, very strongly suggest that in this species the groups binding the propionates are imidazolium side-chains. The shape and range of the curves, and their similarity to the ones obtained for combination with caffeine (Fig. 2), support this. Studies have not yet been extended to haemoglobins of other species; it may be that other amino acid residues are responsible for bonding in these, perhaps explaining the higher heat of dissociation given by Roughton (1944) for ox haemoglobin at pH 6-8, and accounting for his findings in respect to the carbamino reaction. The relationship between the new data, representing the attachment of the propionate groups to the residue in horse apohaemoglobin, and the curve of German and Wyman (1937) is shown in Fig. 5. The top curve of Fig. 5 was obtained by subtracting the increments found for the attachment of nickel The Haem-Globin Linkage 185 mesoporphyrin to apohaemoglobin and to carboxyhaemoglobin. These curves are most probably also related to a differential carbamate equilibrium curve which can be prepared by subtracting the points of the curves of Figs. 5a and 5b of Stadie and O'Brien (1937) for ferro- and oxyhaemoglobin (species unspecified). A new interpretation of the differential titration curve is suggested — (1) that the alkaline loop represents the attachment of propionate groups to imida- zolium side-chains; (2) that the acid loop mirrors the difference between Fig. 5. Comparison of specific increment curve (upper) for attacliment of nickel mesoporphyrin to apohaemoglobin with the differential titration curve (lower) of German and Wyman (1937) for ferroHb-oxyHb (upper curve) prepared by deducting increment with COHb from increment with apoHb). the detachment of the haem iron, in the 'essentially covalent' (and stronger) linked oxyhaemoglobin and the 'essentially ionic' ferrohaemoglobin. As the hydrogen ion concentration increases, the group bonding the iron in ferro- haemoglobin will be liberated, that in oxyhaemoglobin will still be held by the stronger bond. The maximum difference occurs at about pH 5-4 after which the 'covalent' bond begins to break and the extent of ionization of the two groups becomes the same at pH 4-3. Preliminary studies (O'Hagan, unpub- lished) show very marked differences in the stability of carboxyhaemoglobin and ferrihaemoglobin in the pH range 4-6. If the above interpretation is correct, a more pronounced acid loop should be found for the differential titration of carboxyhaemoglobin-ferrohaemoglobin. The experiment with apomyoglobin shows that a much stronger basic group or groups is bonding the propionate(s) as evidenced by the greater height of the curve in the pH range 5-9 and the major shift (30 m/t) in the position 186 J. E. O'Hagan of the Soret peak. This finding of a stronger basic group could explain the considerably lower heat of oxygenation (A//q = — 17-5kcal) found for myoglobin by Theorell (1934), compared with the values of haemoglobins (A/Zq '->-' — 12-5 kcal, Wyman, 1948; see also George, 1956). The greater stabihty of myoglobin towards alkali (Haurowitz and Hardin, 1954) may also be explainable in terms of this basic group. Change in acid strength of the haem propionates would be expected to have less effect on bonds to such stronger basic groups, accounting for the small Bohr effect and minor effect on the shape of the oxygen dissociation curve with a change in the hydrogen ion concentration. Examination of the attachment in the pH range 11-13, using a hydrogen electrode, may give further information on the nature of the linking group. German and Wyman (1937) and Wyman (1948) in discussion of the nature of the group binding the haem iron, pointed out the uncertainty that the group ionizing in the more acid range was imidazole, and that the second carboxyl group of dicarboxylic acids should also be considered. This seems to have been generally overlooked, and the matter calls for more attention in the fight of the observations made here. Even a pK value of 5-3 appears low for an imidazole group by comparison with its value in compounds resembfing those occurring naturally. The strongest evidence against the hypothesis that the haem iron is linked to a group other than imidazole is the ingenious experiment of Wang (1958), who found that diethylprotohaem linked to l-(2-phenyl-ethyl)-imidazole in a film of polystyrene, bound carbon monoxide which could be replaced by oxygen. However, substitution at position 1 on the imidazole ring would appear to change its character, making it unlike that presumed to occur in haemoglobins. It could be that such a substitution makes the free nitrogen acidic and that the acid strength of the linking groups rather than their structure is the important factor. This may apply also to the imidazole haem compound found by Corwin and Reyes (1956) to bind oxygen, though apparently very poorly. Whatever type of linkage exists between the iron and the apoprotein, whether it be imidazole, ^ or y carboxyl or an unusual type not yet detected in proteins, should not materially affect the discussion which follows. The oxygen dissociation curves of the haemoglobins can now be examined in the light of the new findings. It is interesting that Barcroft (1938) likened the curves to oxygen titration curves and we could perhaps visualize them as representing the 'titration' by haem propionates (of changing acid strength) of the weakly basic groups (haemoglobins) or more strongly basic groups (myoglobins). The oxygen dissociation curve of haemoglobin could be considered as representing the titration, from no combination to full combina- tion, of the propionate group with the imidazolium group. The curve for myoglobin would be only the upper section of the curve for the titration of The Haem-Globin Linkage 1 87 a propionate group with a much more basic group, the strength of the com- bination being only shghtly affected on reduction. A weak acid-weak base Hnkage would be expected to be considerably affected by neutral salts and specific ions, as is the oxygenation of haemoglobin, while a linkage to a stronger base would be less affected. This concept could explain the effect known as 'haem-haem interaction', and at the same time account for the Bohr effect. That these effects are apparently interrelated is indicated by the loss of both under certain conditions of reconstitution of haemoglobins (Wyman, 1948). Rossi-Fanelli and Antonini (1959) from a study of human deuterohaemoglobin considered the vinyl groups to be involved in 'haem-haem interaction' and illustrated, but did not explain, the influence of removal of these groups in decreasing the Bohr effect at a given pH. The results seem comparable to the findings of Riggs and Wolbach (1956) on addition of mersalyl to horse haemoglobin. In both cases these effects would appear to be secondary, in the first from a decrease in acid strength of the haem propionates, brought about by removal of the vinyl groups; in the second on account of the propionate binding group of the protein becoming more basic, through combination of a neighbouring sulphydryl group with the mersalyl. A labile electrostatic linkage between at least one of the two haem pro- pionates of each of the four haems and imidazolium side-chains of the apoprotein might explain, better than 'haem-haem interaction', the sigmoid oxygen dissociation curve and the Bohr effect in horse haemoglobin. Linkage to more or less basic groups than the imidazoliums of horse haemoglobin could conceivably explain the curves for other species, including those with 'atypical' curves. The detachment of the propionates in ferrohaemoglobins could explain the change in molecular shape on deoxygenation and also partly account for dry oxyhaemoglobin not releasing oxygen at low oxygen pressures (Haurowitz and Hardin, 1954), the electrostatic bond breaking only in solution. While the exact state of the haemoglobin in the erythrocyte is unknown (see Wintrobe, 1956), the most likely condition — attachment to a framework of stromatin at an equivalent concentration of 34 % haemoglobin — could be conceived as decreasing the velocity of diffusion of oxygen into the interior of the cell. A change in shape of the haemoglobin molecules on breaking the electrostatic link might counteract this to some extent by 'agitation' of the cell contents. SUMMARY 1. Reduction of haematin to haem markedly decreases the acid strength of one or (more probably) both of the propionate side-chains. 2. Nickel mesoporphyrin which can only combine through its propionate side-chains, links with a group or groups in horse apohaemoglobin v/ith imidazolium characteristics. H.E. — VOL. I — o 188 J. E. O'Hagan 3. With horse apomyoglobin nickel mesoporphyrin combines with a more basic group or groups. 4. Fresh evidence is presented that the group in apomyoglobin binding the haem iron is more likely to be of the nature of a /J or y carboxyl than an imidazole group. 5. An interpretation of the sigmoid shape of the oxygen dissociation curve, the Bohr effect, the alkaline stability, and the change of molecular shape on oxygenation, in terms of labile electrostatic linkages between the haem propionate groups and imidazolium side-chains of horse apohaemoglobin is presented. 6. It is suggested that the change in shape of the haemoglobin molecule on oxygenation or reduction may 'agitate' the contents of the erythrocyte and thus assists the exchange of oxygen and carbon dioxide between the cell interior and the plasma surrounding its envelope. Acknowledgement Without the continued encouragement of Dr. A. E. Shaw and generous provision of facilities by the Queensland Division of the Australian Red Cross Society this work could not have been accomplished. ADDENDUM (Note added in proof) The subsequent evidence of Kendrew, Dickerson, Strandberg, Hart, Davies, Phillips and Shore (1960) indicates that the haem iron in myoglobin is attached to imidazole nitrogen, not to a carboxyl group. Their work was carried out with crystalline ferrimyoglobin, mine with this compound in solution, but it is unlikely that the mode of attachment would be essentially different in the crystal and in solution. REFERENCES Altschul, a. M. & HoGNESS, T. R. (1939). /. biol. Chem. 129, 315. Barcroft, J. (1928). The Respiratory Function of the Blood, Pt. 2, p. 102, Cambridge Univ. Press. Barcroft, J. (1938). Features in the Architecture of Physiological Function, p. 71, Cam- bridge Univ. Press. Corwin, a. H. & Reyes, Z. (1956). /. Amer. chem. Soc. 78, 2347. Coryell, C. D., Pauling, L. & Dodson, R. W. (1939). J.phys. Chem. 43, 825. Coryell, C. D., Stitt, F. & Pauling, L. (1937). /. Amer. chem. Soc. 59, 633. Edsall, J. T. & Wyman, J. (1958). Biophysical Chemistry, 1, 558. Academic Press, New York. Fischer, H. & Orth, H, (1937). Die Chemie des Pyrroles, 2, 442. Akademische Verlags- gesellschaft, Leipzig. Fischer, H. & Putzer, B. (1926). Hoppe-Seyl. Z. 154, 39. Fox, H. M. (1932). Proc. Roy. Soc. B. Ill, 356. German, B. & Wyman, J. (1937). J. biol. Chem. 117, 533. George, P. (1956). Currents in Biochemical Research, p. 338 (Ed. by D. E. Green), Inter- science, New York. The Haem-Globin Linkage 1 89 George, P. & Hanania, G. I. H. (1953). Biochem. J. 55, 236. George, P. & Lyster, R. L. J. (1958). Conference on Haemoglobin, May 2-3, 1957, p. 33. Publication No. 557, National Academy of Sciences — National Research Council, Washington. Granick, S. (1949). Harvey Lectures, 44, 228. Grinstein, M. (1947). J. biol. Cliem. 167, 515. Grinstein, M. & Watson, C. J. (1943). /. biol. Cliem. 147, 671. Hanania, G. I. H. (1953). Ph.D. Thesis, University of Cambridge. Hartridge, H. & RouGHTON, F. J. W. (1925). Proc. Roy. Soc. A. 107, 654. Haurowitz, F. (1959). Private communication. Haurowitz, F. & Hardin, R. L. (1954). The Proteins, (Ed. by H. Neurath and K. Bailey), vol. 2, p. 332. Academic Press, New York. Haurowitz, F. & Klemm, W. (1935). Ber. dtscli. cliem. Ges. 68, 2312. Hill, R. (1939). Perspectives in Biochemistry, p. 131 (Ed. by J. Needham and D. E. Green), Cambridge Univ. Press. Hill, R. & Holden, H. F. (1926). Biochem. J. 20, 1326. Holden, H. F. (1941). Aiist. J. exp. Biol. med. Sci. 19, 1, Itano, H. a. & Robinson, E. (1956). J. Amer. chem. Soc. 78, 6415. Keilin, D. (1953). Nature, Lond. Ill, 922. Keilin, D. & Wang, Y. L. (1946). Biochem. J. 40, 855. Keilin, J. (1943). Biochem. J. 2,1, 281. Keilin, J. (1944). Nature, Lond. 154, 120. Kendrevv, J. C, BoDO, G., Dintzis, H. M., Parrish, R. G., Wyckoff, H. & Phillips, D. C. (1958). Nature, Lond. 181, 662. Kendrew, J. C, Dickerson, R. E., Strandberg, B. E., Hart, R. G., Davies, D. R., Phillips, D. C. & Shore, V. C. (1960). Nature, Lond. 185, 422. Lemberg, R. & Legge, J. W. (1949). Hematin Compounds and Bile Pigments, pp. 94, 121, 194, 214, 243, 295 and 458, Interscience, New York. Nahas, G. G. (1951). Science, 113, 723. O'Hagan, J. E. (1950). M.Sc. Thesis, University of Queensland. O'Hagan, J. E. (1955). Abstr. 3rd int. Congr. Biochem., Brussels, p. 10, O'Hagan, J. E. (1959a). Nature, Lond. 183, 393. O'Hagan, J. E. (1959b). Ph.D. Thesis, University of Queensland. O'Hagan, J. E. (1960). Biochem. J. 74, 417. O'Hagan, J. E. & Barnett, C. (1958). Abstr. 3rd Gen. Meeting, Aust. Biochem. Soc. Adelaide, 1958. O'Hagan, J. E. & George, P. (1960). Biochem. J. 74, 424. Pauling, L. (1935). Proc. nat. Acad. Sci. Wash. 21, 186. Pauling, L. & Coryell, C. D. (1936). Proc. nat. Acad. Sci. Wash. 22, 159. Redfield, a. C. (1933). Quart. Rev. Biol. 8, 31. RiGGS, A. F. & Walbach, R. A. (1956). J. gen. Physiol. 39, 585. Rossi-Fanelli, a., Antonini, E. (1959). Arch. Biochem. Biophys. 80, 308. Rossi-Fanelli, a., Antonini, E. & Caputo, A. (1959). Nature, Lond. 183, 827. RouGHTON, F. J. W. (1944). Harvey Lectures, 39, 96. Roughton, F. J. W., Otis, A. B. & Lyster, R. L. J. (1955). Proc. Roy. Soc. B. 144, 29. Russell, C. D. & Pauling, L. (1939). Proc. nat. Acad. Sci. Wash. 25, 517. Stadie, W. C. & O'Brien, H. (1937). /. biol. Chem. Ill, 439. Takashima, S. (1955). J. Amer. chem. Soc. 11, 6173. Theorell, H. (1934). Biochem. Z. 268, 73. Theorell, H. (1942). Ark. Kemi Min. Geol. 16A, No. 14. Theorell, H. & Ehrenberg, A. (1951). Acta chem. Scand. 5, 823. Wang, J. H. (1958). J. Amer. chem. Soc. 80, 3168. Wintrobe, M. M. (1956). Clinical Haemotology, 4th ed., p. 89, Lea & Febiger, Phila- delphia. Wyman, J. (1948). Advanc. Protein Chem. 4, 410. Wyman, J. & Ingalls, E. N. (1941). J. biol. Chem. 139, 877. 190 Discussion DISCUSSION Native Globin Drabkin: I would be glad if O'Hagan could tell us a little more about the state of his apohaemoglobin as against that of his apomyoglobin. I am thinking in terms of the possibility that, since myoglobin is far, far more stable toward alkali than is haemo- globin, some question may be raised as to the strict validity of comparing the two apoproteins. Lemberg: Rossi-Fanelli has found that by his method any irreversible denaturation of globin from haemoglobin can be prevented, and O'Hagan followed his method rather closely. Denaturation will, however, occur if the recombined haemoglobin solutions contain an excess of free globin and are measured in the spectrophotometer at room temperature. O'Hagan: Initially trouble was experienced with these apoproteins due to precipitation of the denatured material at room temperature and neutral pH. The trick is to bring the pH to 7-8, leave at 2rc for 1 hr, leave at O'^C for three days to remove coagulated material, and centifuge for 10 min at 20,000 X g. With one preparation, which I considered to be not as good as the others, on other evidence, the peak at pH 10 (see Fig. 3 of my paper) was higher and broader. George: In our work on reconstituted ferrimyoglobin, O'Hagan and I obtained data similar to that of Rossi-Fanelli on reconstituted ferrohaemoglobin. In a comparison with native ferrimyoglobin we found that the affinity for fluoride is scarcely altered, and that the pK values of the haem-linked ionizing group associated with its Bohr effect are identical to within experimental error. The Linkage of Iron and Protein in Haemoglobin Perrin : In systems such as haem, or haematin plus albumin, we have equilibria such as : Hm COOH ^ Hm COO- + H+ pK^ Alb H+ + Hm COO- ^ Hm COOH Alb K Alb -H H+ ^ Alb H+ pJ^T/ This is certainly a gross over-simplification but will serve to illustrate a difficulty in using O'Hagan's spectral absorption difference approach. At constant albumin and total Hm concentration there are still three light-absorbing species in such a system and their concentrations are governed by three unknown constants, K, K^ and AT/. In addition, each of the absorbing species has its own molecular extinction coefficient, so that the absorbance increment at any given pH is not a simple function of the species. There is no reason to assume the A^'s are the same for haem and haematin, and in fact they are most unlikely to be if any binding through the iron is involved. I should like to ask O'Hagan how he arrives at the conclusion that 'reduction (of haematin to haem) significantly decreases the acid strength of at least one of the haematin propionate groups.' Concerning Fig. 4, 1 should like to point out that stability curves of metal complexes such as ferrihaemoglobin cannot be used to obtain the pATs of the ligands. Falk, Phillips and I have discussed this elsewhere (^Nature, Lond. 184, 1651, 1959). The 'apparent p/Ts' in such cases are, in fact, functions not only of the pAT of the ligand but also of the stability constant of the metal complex and the concentration of the ligand. It is quite erroneous to identify such 'apparent pATs' with the pA:'s of groups such as carboxyl or imidazole. The nature of the metal-to-ligand bond does not affect this conclusion: if a complex is present, there is, of course, a AG = RTXnK which leads to significant changes in the thermodynamics of the system relative to the proton-ligand system. One cannot use the A/f for haem-protein dissociation where the iron-to-protein link is involved as evidence for COO- or imidazole linkage if one takes the data for The Haem-Globin Linkage 191 proton addition to the latter groups. There is no reason why A// for metal-ligand bonding should be the same as for proton addition to the ligand. Phillips: The conclusion reached in part 4 of O'Hagan's summary seems not only un- justified but also contrary to the facts. One cannot deduce the ^K of a ligand from the pH at which the co-ordination complex is half dissociated without knowing the stability constant of the complex and the concentration of the various species: One can be certain, however, that the pA" will be greater than the pH of half-dissociation and all one can deduce from O'Hagan's results is that the pAT of the co-ordinating group is > 5 which could well be histidine (pAT — 7) or even lysine (pAT '~ 9) but is unlikely to be a carboxyl group (pAT — 5). O'Hagan: It is true that the pAT values I have suggested are not exact; they were not intended to be. I was not attempting to obtain an absolute value under these con- ditions, but one which would at least indicate the most likely group ionizing. In regard to Phillips' point, curve (a) shows some preliminary results on the dissociation of horse oxy haemoglobin in acid buffers. There is evidence that at pH 50 groups are dissociating. These cannot be the haematin side-chain carboxyls, the evidence points to the groups linked to the iron. At pH 4-9, Ferry and Green (/. biol. Chem. 81, 175, 1929) found horse haemoglobin stable enough to obtain an oxygen dissociation curve. If we examine the curve for the apparent heat of dissociation of horse oxy- haemoglobin of German and Wyman {J. biol. Chem. 117, 533, 1937), we can see that at pH 5-0 the apparent heat of dissociation is about — I kcal. If there were imidazole groups ionizing we would expect a value of about 6 kcal. The value found would appear to indicate carboxyl groups are linked to the iron atoms. 4- 3- . The Haem-Globin Linkage (a) Change in the Soret peak absorbance at 410 m/< of horse oxyhaemoglobin in phthalate buffers (pH 4-0-6-2, I = 005). (b) Curve for apparent heat of dissociation of horse oxyhaemoglobin from German and Wyman {J. biol. Chem. Ill, 533, 1937). 192 Discussion Lemberg : Are we really sure that the combination between haematin and serum albumin is only through the haem carboxylic acid groups? This is certainly not so for the combination of haematin a with serum albumin, in which the spectrum clearly indi- cates combination with protein nitrogen. I was inclined to accept the suggestion of J, Keilin for the protohaematin compound, but I feel no longer sure about it now. George: In considering various mechanisms that might account for haem-linked ionization effects, I recently calculated the pH range over which a salt bridge would remain intact, and I think this will clarify the points just raised by Orgel, Perrin, Phillips and Williams. If the equilibrium constant for the formation of a salt bridge between, say, a carboxylate group and a substituted ammonium group is K^, —COO- + NH3+— -COO-NH,+— and the ionization constants for the two separate groups are ATcooh and K-^n^^ respectively, then in acidic solution the pH at which 50% formation occurs is given by /'(A's^cooh), and in alkaline solution the corresponding pH is given by p{K^Yii^lK^. As illustrated in the diagram the formation of the salt bridge results in the 'titration' of the carboxyl group in a lower pH range and the amino group in a higher pH range than usual — the apparent shift in the pA^ values being determined by the magnitude of K^. The shift in the pX^ values is in an opposite sense, because in acidic solution the bridge is broken by combination with H+, I.e. — COO-NH3+— + W — COOH + NH3+- (1) whereas in alkaline solution by the dissociation of H+, i.e. — COO-NH3+— ^ COO- + NHj— + H+ (2) In reaction (1) NH3+ — can be thought of as competing with H+ for the COO- group, thus lowering the 'p^'; in reaction (2) the combination of H+ with NHg — giving NH3+ — is favoured by the formation of the salt bridge; this has the effect of making NH3+ — a weaker acid, and hence raises its pA'. The Haem-Glohin Linkage 193 Falk: It is clear that the basis of the difference which Perrin, Phillips and I have had with O'Hagan is that he seems to consider the protein-haemoglobin bond as something like a simple electrostatic bond, with virtually zero stability constant, where we consider that co-ordination occurs to give a haem-giobin complex with a finite stability constant. Regarding COO~ as the protein-iron bond in haemoglobin, my objection to this is that of all the possible protein ligand groups, — COO" would tend to stabilize the ferric state most. The stabilization of its ferrous state, compared to haem itself, is perhaps the most important and the most outstanding special property of haemoglobin. EARLY STAGES IN THE METABOLISM OF IRON By J. B. Neilands Department of Biochemistry, University of California, Berkeley, California There are certain chemical and biochemical characteristics of iron which place it in a unique category in relation to the other common biocatalytic elements. Both ferric and ferrous ions are quite insoluble in aqueous solution at physio- logical pH; this property is of special significance in the case of ferric ion (solubility product 10^^^) since most of the iron available to living organisms will be encountered, at least initially, in the higher oxidation state. If one considers the very large proportion of iron that takes part in the transport and storage of oxygen as catalytic iron, then the latter element is seen to be quantitatively the single most important biocatalytic element in the entire realm of animal enzymology. Finally, since iron is bound relatively weakly to the usual type of naturally-occurring ligand, it might be expected that living cells, especially those with a high requirement for this metal, would have found it necessary to evolve special complexing agents which have the capacity to overcome such problems that are inherent in the transport and metabolism of this particular element. The complete synthesis of an iron-enzyme involves the ultimate conver- gence of at least two, possibly three, biosynthetic pathways. If the broad definition of an iron-enzyme given above is adopted, it is clear that in many instances the major portion of enzyme iron will occur as the porphyrin chelate, i.e. haem. The early stages in the biosyntheis of the organic part of this pros- thetic group have been thoroughly elucidated at least to the level of porpho- bilinogen (Shemin, 1955; Laver, Neuberger and Udenfriend, 1958) and partially clarified from thence to coproporphyrinogen (Granick and Mauzerall, 1958). Little exact information is available as yet concerning the immediate precursor of protoporphyrin although from the vitamin require- ments for haem synthesis (Lascelles, 1957) it might be speculated that an acrylic acid side chain should occur as an intermediate between the carboxy- ethyl and vinyl side chains. In recent years many investigators have examined the mechanism of uptake of iron by protoporphyrin. The concensus of opinion appears to be that the reaction is enzyme-catalyzed although the specific protein responsible for the observed effect has not been isolated. Until the latter has been achieved, the precise nature of the iron donor in the reaction will remain obscure. 194 Early Stages in the Metabolism of Iron 195 Essentially nothing is known of the structure and origin of the active centres of the non-haem iron enzymes. The special problems which arise in iron metabolism, as contrasted, for example with copper metabolism, can be illustrated by comparison of the characteristics and behaviour of the two elements in question. In the case of copper it is certain that smaller quantities are required by living tissues, the somewhat greater solubility of the hydroxide (saturated water solution of cupric hydroxide is > 10~^ m at 25°, Seidell, 1940) and, finally, the ubiquitous a-amino carboxylic acid structure provides an effective ligand capable of holding the cupric ion in solution at physiological pH. In the present paper, results will be presented for certain experiments dealing with the early stages of iron metabolism in micro-organisms. The latter form of life has been chosen for investigation on account of the well-known metabolic flexibility characteristic of unicellular organisms; however, in spite of this advantage, it must be recognized that micro-organisms as experi- mental subjects suffer from the fact that each species may exhibit certain metabolic variations. This will effectively preclude the formulation of sweeping generalities about the detailed mechanism of iron metabolism in all forms of life. The technique employed in the present instance has been that of cultivation of the aerobic micro-organisms Bacillus subtilis and Ustilago sphaerogena in the presence of diminished levels of iron. Such very aerobic species can be expected to have a reasonably high requirement for iron and a correspondingly well-developed system for the intermediary metabolism of this element. This statement is particularly true for Ustilago sphaerogena since this organism is known to form large quantities of cytochrome c (Grimm and Allen, 1954). The growth of such cells under conditions of iron deprivation provides valuable information about the early stages of iron metabolism. At very low levels of iron there is sparse growth, a feeble metabolism and essentially nothing can be learned about the intimate processes of iron utilization. Similarly, at abnormally high levels of iron, the substances usually involved as intermediates in iron metabolism may be produced in greatly diminished quantities in spite of the excellent cell yields. On the other hand, at inter- mediate levels of iron, three possible metabolic adjustments come into play which lead to the accumulation and excretion of iron-complexing agents : (i) The biosynthesis of specific ferric complexing agents, normally com- petitively inhibited and maintained at a low level by the presence of a variable amount of the ferric chelate, becomes a major metabolic activity of the cell. (ii) The deficiency of iron creates a metabolic block, the latter being manifested by the appearance of iron-complexing products which normally require iron for their further metabolism. 196 J. B. Neilands (iii) The new substances produced in iron deficiency are intended to serve, either as such or as the ferric complex, as a by-pass for electron trans- port around the normal cytochrome system. The isolation, characterization and chemical synthesis of itoic acid (iron- transferring-orthophenol ; 2 : 3-dihydr