|
1. |
Contents pages |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 008-009
Preview
|
PDF (68KB)
|
|
摘要:
Quarterly Reviews No 3 Vol 24 1970 Page TILDEN LECTURE The Biochemistry of Sodium Potassium Magnesium and Calcium By R. J. P. Williams 331 CENTENARY LECTURE Roads to Corrins By A. Eschenmoser The van der Waals Fluid A Renaissance By M. Rigby 366 41 6 The Separation of Polar Steric and Resonance Effects in Organic Reactions by the Use of Linear Free Energy Relationships By J. Shorter 433 Application of Computers in Chemical Analysis Amino-acid Analysis and Sequence Determination By B. Sheldrick 454 The Chemical Society London Quarterly Reviews contains articles by recognised authorities on selected topics from general physical inorganic and organic chemistry. The Journal and Annual Reports interest primarily the research worker Quarterly Reviews is designed for a wider range of readers.It is intended that each review article shall be of interest to chemists generally and not only to workers in the particular field being reviewed. The submission of reviews for publication is welcomed but intending authors are advised to write in the first place to the Editor The Chemical Society Burlington House Piccadilly London W 1V OBN. Such pre- liminary communications should be accompanied by an outline of the ground to be covered (about two quarto pages) rather than by the completed manuscript. Price to non-fellows f 4 10s. Od. per annurn. The Chemical Society-Endowed Lectures. The Council of The Chemical Society has decided that the Endowed Lectures of the Society shall in future be published in full in Quarterly Reviews to ensure that there will be a permanent record. These will be in addition to the normal Review articles. 0 Copyright reserved by The Chemical Society 1970 Published by The Chemical Society Burlington House London. Printed in England by Eyre & Spottiswoode Ltd Thanet Press Margate
ISSN:0009-2681
DOI:10.1039/QR97024FP008
出版商:RSC
年代:1970
数据来源: RSC
|
2. |
Front cover |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 009-010
Preview
|
PDF (94KB)
|
|
摘要:
CONTENTS PAGE PRESIDElVTIAL ADDRESS. TRANSITION-METAL COMPLESES OF SOME MOLECULAR COMPLEXES OF WATER I N ORGANIC SOLVENTS AKD I N THE VAPOUR PHASE. By Sherril D. Christian Ahrned A. Taha and Bruce W. Gash . . 20 PHOTOCHEMICAL REACTIONS I N NATURAL PRODUCT SYNTHESIS. By P. G. Sarnrnes 37 THE CO-ORDINATION OF AMBIDEXJATE LIGANDS. By A. H. Norbury and A. I. P. Sinha . . 69 WAVE FUNCTIONS FOR SMALL MOLECULES BASED ON LINEAR COMBINATIONS OF ATOMIC ORBITALS. By R. G. Clark and E. Theal Stewart . . 95 DEVELOPMENTS IN THE CHEMISTRY OF DIAZO-ALKANES. By G. W. Cowell and A. Ledwith. . 119 STRUCTURE AND PROPERTIES OF AQUEOUS SALT SOLUTIONS. By M. J. Blandamer 169 THE THERMAL DECOMPOSITION OF HYDROUS LAYER SILICATES AND T H ~ R RELATED HYDROXIDES. By N. H. Brett K. J. D. MacKenzie and J. H. Sharp .. 185 THE STEREOCHEMISTRY OF POLYSULPHIDES. By R. Rahman S. Safe and A. Taylor 208 THE STUDY OF SIMPLE LIQUIDS BY COMPUTER SIMULATION. By I. R. McDonald and K. Singer . . . 238 VOLATILE COMPOUNDS OF THE HYDRIDES OF SILICON AND GERMANIUM WITH ELEMENTS OF GROUPS V AND VI. . 263 TRIMETHYLENEMETHANE AND RELATED CY,CY’-DISUBSTITUTED ISOBUTENES. By Francis Weiss . . I . 278 PERFLUORO-LIGANDS. By Sir Ronald Nyholrn F.R.S. . . . I By John E. Drake and Chris Riddle . ORGANOTHALLIUM CHEMISTRY. By A. G. Lee . . . . 310 AND CALCIUM. By R. J. P. Williams . . . . 331 TILDEN LECTURE. THE BIOCHEMISTRY OF SODIUM POTASSIUM MAGNESIUM CENTENARY LECTURE. ROADS TO CORRINS. By A. Eschenmoser . . 366 THE VAN DER WAALS FLUID A RENAISSANCE. By M. Rigby . . 416 THE SEPARATION OF POLAR STERIC AND RESONANCE EFFECTS I N ORGANIC RE- ACTIONS BY THE USE OF LINEAR FREE ENERGY RELATIONSHIPS.By J. Shorter . 433 APPLICATION OF COMPUTERS I N CHEMICAL ANALYSIS AMINO-ACID ANALYSIS AND SEQUENCE DETERMINATION. By B. Sheldrick . . . . . . 4 5 4 Bartlett . . 473 By E. W. Abel and F. G. A. Stone . . 498 ACIDIC AND BASIC AMIDE HYDROLYSIS. By Charmian O’Connor . . 553 THE PHOTOLYSIS OF SIMPLE INORGANIC ANIONS IN SOLUTION. By Malcolm F. Fox 565 BASE-CATALYSED ISOMERISATION OF ACETYLENES. By R. J. Bushby . . 585 DIFFUSION IN TONIC SOLIDS. By J. M. Pollock . . 601 THE REACTIONS OF HYDRAZINE WITH TRANSITION-METAL COMPLEXES. By F. Bottornley . . . . 617 CENTENARY LECTURE. MECHANISMS OF CYCLOADDITION. By Paul D. THE CHEMISTRY OF TRANSITION-METAL CARBONYLS SYNTHESIS AND REACTIVITY. CONTENTS PAGE PRESIDElVTIAL ADDRESS.TRANSITION-METAL COMPLESES OF SOME MOLECULAR COMPLEXES OF WATER I N ORGANIC SOLVENTS AKD I N THE VAPOUR PHASE. By Sherril D. Christian Ahrned A. Taha and Bruce W. Gash . . 20 PHOTOCHEMICAL REACTIONS I N NATURAL PRODUCT SYNTHESIS. By P. G. Sarnrnes 37 THE CO-ORDINATION OF AMBIDEXJATE LIGANDS. By A. H. Norbury and A. I. P. Sinha . . 69 WAVE FUNCTIONS FOR SMALL MOLECULES BASED ON LINEAR COMBINATIONS OF ATOMIC ORBITALS. By R. G. Clark and E. Theal Stewart . . 95 DEVELOPMENTS IN THE CHEMISTRY OF DIAZO-ALKANES. By G. W. Cowell and A. Ledwith. . 119 STRUCTURE AND PROPERTIES OF AQUEOUS SALT SOLUTIONS. By M. J. Blandamer 169 THE THERMAL DECOMPOSITION OF HYDROUS LAYER SILICATES AND T H ~ R RELATED HYDROXIDES. By N. H. Brett K. J. D. MacKenzie and J. H. Sharp . . 185 THE STEREOCHEMISTRY OF POLYSULPHIDES.By R. Rahman S. Safe and A. Taylor 208 THE STUDY OF SIMPLE LIQUIDS BY COMPUTER SIMULATION. By I. R. McDonald and K. Singer . . . 238 VOLATILE COMPOUNDS OF THE HYDRIDES OF SILICON AND GERMANIUM WITH ELEMENTS OF GROUPS V AND VI. . 263 TRIMETHYLENEMETHANE AND RELATED CY,CY’-DISUBSTITUTED ISOBUTENES. By Francis Weiss . . I . 278 PERFLUORO-LIGANDS. By Sir Ronald Nyholrn F.R.S. . . . I By John E. Drake and Chris Riddle . ORGANOTHALLIUM CHEMISTRY. By A. G. Lee . . . . 310 AND CALCIUM. By R. J. P. Williams . . . . 331 TILDEN LECTURE. THE BIOCHEMISTRY OF SODIUM POTASSIUM MAGNESIUM CENTENARY LECTURE. ROADS TO CORRINS. By A. Eschenmoser . . 366 THE VAN DER WAALS FLUID A RENAISSANCE. By M. Rigby . . 416 THE SEPARATION OF POLAR STERIC AND RESONANCE EFFECTS I N ORGANIC RE- ACTIONS BY THE USE OF LINEAR FREE ENERGY RELATIONSHIPS. By J. Shorter . 433 APPLICATION OF COMPUTERS I N CHEMICAL ANALYSIS AMINO-ACID ANALYSIS AND SEQUENCE DETERMINATION. By B. Sheldrick . . . . . . 4 5 4 Bartlett . . 473 By E. W. Abel and F. G. A. Stone . . 498 ACIDIC AND BASIC AMIDE HYDROLYSIS. By Charmian O’Connor . . 553 THE PHOTOLYSIS OF SIMPLE INORGANIC ANIONS IN SOLUTION. By Malcolm F. Fox 565 BASE-CATALYSED ISOMERISATION OF ACETYLENES. By R. J. Bushby . . 585 DIFFUSION IN TONIC SOLIDS. By J. M. Pollock . . 601 THE REACTIONS OF HYDRAZINE WITH TRANSITION-METAL COMPLEXES. By F. Bottornley . . . . 617 CENTENARY LECTURE. MECHANISMS OF CYCLOADDITION. By Paul D. THE CHEMISTRY OF TRANSITION-METAL CARBONYLS SYNTHESIS AND REACTIVITY.
ISSN:0009-2681
DOI:10.1039/QR97024FX009
出版商:RSC
年代:1970
数据来源: RSC
|
3. |
Back cover |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 011-012
Preview
|
PDF (283KB)
|
|
摘要:
Ami no-acids Peptides and Proteins Volume I The Chemical Society announces the publication of Volume 1 in this the third title in their series of Specialist Periodical Reports. This volume reviews and evaluates progress reported during 1968 to a depth and comprehensiveness not available elsewhere. The scope is indicated by the principal chapter headings Amino-acids ; Structural investigation of peptides and proteins ; Peptide synthesis ; Peptides of abnormal structure; The relationship between structure and biological activity of peptides and proteins ; Metal derivatives of amino-acids peptides and proteins. The coverage includes naturally-occurring synthetic and chemically modified materials and chemical physical stereochemical analytical structural and synthetical studies. Publication will be annual authorship being undertaken by a team of eight scientists led by Dr.G. T. Young of Oxford as Senior Reporter. Specialist Periodical Reports are designed to assist the research worker or specialist in his own field t o give the non-specialist a concentrated but complete view of the topic being reported and to provide libraries with a useful source book. Size 83" x 53" Pages xii + 308 Cloth Bound SBN 85186 004 4 Price per volume Fellows of The Chemical Society f 3. 0.0 (US $7.20) Non- Fellows f4.10.0 (US $1 0.80) This publication may be ordered from the Publications Sales Officer The Chemical Society Blackhorse Road Letchworth Herts England. Ami no-acids Peptides and Proteins Volume I The Chemical Society announces the publication of Volume 1 in this the third title in their series of Specialist Periodical Reports.This volume reviews and evaluates progress reported during 1968 to a depth and comprehensiveness not available elsewhere. The scope is indicated by the principal chapter headings Amino-acids ; Structural investigation of peptides and proteins ; Peptide synthesis ; Peptides of abnormal structure; The relationship between structure and biological activity of peptides and proteins ; Metal derivatives of amino-acids peptides and proteins. The coverage includes naturally-occurring synthetic and chemically modified materials and chemical physical stereochemical analytical structural and synthetical studies. Publication will be annual authorship being undertaken by a team of eight scientists led by Dr. G. T. Young of Oxford as Senior Reporter. Specialist Periodical Reports are designed to assist the research worker or specialist in his own field t o give the non-specialist a concentrated but complete view of the topic being reported and to provide libraries with a useful source book. Size 83" x 53" Pages xii + 308 Cloth Bound SBN 85186 004 4 Price per volume Fellows of The Chemical Society f 3. 0.0 (US $7.20) Non- Fellows f4.10.0 (US $1 0.80) This publication may be ordered from the Publications Sales Officer The Chemical Society Blackhorse Road Letchworth Herts England.
ISSN:0009-2681
DOI:10.1039/QR97024BX011
出版商:RSC
年代:1970
数据来源: RSC
|
4. |
Tilden Lecture. The biochemistry of sodium, potassium, magnesium, and calcium |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 331-365
R. J. P. Williams,
Preview
|
PDF (2258KB)
|
|
摘要:
TILDEN LECTURE The Biochemistry of Sodium Potassium Magnesium and Calcium By R. J. P. Williams INORGANIC CHEMISTRY LABORATORY SOUTH PARKS ROAD OXFORD 1 Introduction In this article I wish to illustrate three features of inorganic biochemistry (a) the controlled level of metal ion concentrations in biological systems (b) the high degree of the selectivity of their functional activity and (c) the manner in which we can set about the task of understanding the cation activities. In previous work on the functional significance of metal ions our major concern has been with the transition metals1s2 and with zinc3 but here I shall be considering largely the biochemistry of sodium potassium magnesium and calcium. The biological significance of these four cations is very different from that of the transition metals and zinc as is shown in Table 1.Whereas the transition metals and zinc Table 1 Classification of cations in biological systems Na+ K+ Mg2+ Ca2+ Charge-carriers Structure formers and triggers Mobile Semi-mobile Oxygen-anion Oxygen-anion binding binding Weak complexes Moderately strong complexes Very fast Moderately fast exchange exchange Zn2+ Super-acid catalysts Static Ni trogen/sulp hur ligands Strong complexes No exchange Fe Cu Co Mo Redox catalysts Static Nitrogen/sulphur ligands Strong complexes No exchange Bioenergetics’ vol. 3 ed. A. Sanadi Academic l R. J. P. Williams in ‘Current Topics in Press New York 1969 p. 80 et seq * R. J. P. Williams Roy. Inst. Chem. Rev. 1968 1 13. 8 B. L. Vallee and R. J. P. Williams Chem. in Britain 1968 4 397; R. J. P. Williams in ‘The Enzymes’ vol.1 ed. P. D. Boyer H. Landy and K. Myrback. Academic Press New York 1959 p 391. The Biochemistry of Sodium Potassium Magnesium and Calcium are strongly bound and immobile the Group IA and IIA metals are weakly bound and mobile. These differences greatly affect the functions of the metals for whereas the cations of copper iron cobalt and molybdenum are redox catalysts and that of zinc is a super-acid catalyst the IA and IIA cations are concerned more in structural and transport (of ionic charge) rbles than directly in catalysis. In addition and because of their mobility their concentration in a given cell or in a part of the cell space can be controlled by metabolism so that cell activity is proportional to the free cation concentration. Our first task then is to examine the individual cation concentrations in different cells and cell compartments.Subsequently we can attempt to understand their functional significance at the molecular level and finally we shall turn to the gross implications of the concentrations and activities in the biochemistry of large organisms. A major chemical procedure which we shall use in order to follow the chemistry of these elements in biological systems and which should be kept in mind from the outset is that of strict isomorphous repla~ement.~ It is known from mineralogy that cations substitute for one another on the basis of similarities in radii and in stereochemical demands. Ionic charge is important but replacement of Ma+ by Mf and M3+ by M2+ is often possible. This idea has been extended to zinc and transition metals in biochemical systems with considerable success.3 By choosing cations for substitution in a given metal enzyme on the basis of size and geometric demand and also on the basis of the physical probe properties which they display it has proved possible to uncover features of binding sites and reaction paths.In particular cobalt(@ is an excellent substitute for zinc,3 and gallium(m) is equally useful as a replacement for iron(rn~).~ Here we shall be looking for probe substitutions for the Group IA and IIA cations. Quite independently of the ability of cations to act as a probe of metal sites they can be used generally as probes of proteins and membranes (see later). The procedure of isomorphous replacement not only permits an examination of a binding site by physical techniques but it also allows us to generate an extended series of cation complexe~.~ For example the study of magnesium and zinc enzymes has been helped by examining the thermodynamic and catalytic properties of long series of metal complexes prepared by substituting the follow- ing bivalent cations for the naturally occurring one Ca Mg Mn Fe Co Ni Cu Zn Cd Hg.As we know the chemistry of each of these cations in detail changes in biological activity along the series can be traced back to fundamental properties of the cations such as ion size electron affinity and geometric demand. Already it has been shown that the sites of metal action in biological systems are extremely selective with regard to cation diameter to within say 0-2 A are sensitive to changes in electron affinity of the cation and are critically dependent upon stereochemistry.There is the strong indication that activity owes itself to a R. C. Woodworth K. G. Moralle and R. J. P. Williams Biochemistry 1970 9 839. 332 Williams co-operative interaction between metal ion and protein. It is one of the purposes of this article to search for similar features in the biological complexes of the simplest cations Na+ K+ Mg2+ and Ca2+. 2 AceumuIation of Group IA and IIA Metals It has been known for a long time that the concentration of free potassium inside is much higher than that outside cells whereas the level of sodium is generally much lower. Such a separation requires constant expenditure of energy and a selective pump which recognises the difference between the two cations.6 The minimum energy used per unit time is proportional to the ion gradient and the flux required to maintain the gradient.Ignoring the flux problem the free energy requirement is proportional to [Na+lout 1- RTln - dG = RTln - [K+ ]in [K+]out tNa+lin For a series of cells which have the same external cellular environment then the relative free energy required is [K+]in dG = RTln- INa+]in It has also been known for some time that total intracellular magnesium concentrations were often higher than extracellular concentrations but that in-cell calcium was maintained at a rather low level. In 1967 we suggested that cells operated a general pumping of these bivalent cations much as they pumped univalent cations.6 For a series of cells which have the same cellular environment the energy required to generate the concentration gradient is proportional to A G3 [ Mgz+]in [ Ca2+ ]in dG3 = RTln - (3) In all three equations (1)-(3) the concentrations should be those of the free unbound cations.We shall find it useful to define the concentrations more loosely for it is generally the case that free concentrations of the ions are not available. In place of free concentrations therefore we shall use the measured analytical concentrations. This implies that bound and free cations are proportional to one another. Clearly this is a gross simplification but as the discussion of cell activity develops it will be shown that it has considerable justification. Table 2 lists some observed gross concentrations of cations in some biological systems. Now the ‘pumping’ of the cations is brought about by the hydrolysis of adenosine triphosphate (ATP) through enzymes called Na+/K+ ATP-ases for Na+ and K+ pumping and Ca2+ ATP-ases for Ca2+ pumping which are in the See for example P.R. Kernan ‘Cell K‘ Butterworths London 1965; and C. P. Bianchi R. J. P. Williams and W. E. C. Wacker J. Amer. Med. ASSOC. 1967 201 18. ‘Cell Calcium’ Butterworths London 1968. 333 The Biochemistry of Sodium Potassium Magnesium and Calcium Table 2 Ionic content of some living systems (mmoles/lOOOg)s Kf Na+ Mg2+ Ca2+ Human red cell (wet) 92.0 11.0 2-5 0.1 Squid nerve extract (relative to Na+ = 1) 5.0 1.0 0.5 0.1 Yeast cells (dry) 110.0 10.0 13.0 1.0 Euglena cells (wet) 103.0 5.0 4.8 0-3 Escherichia coli cells (wet) 250 80 20 5 Skeletal muscle 92 27 22 3 Figure 1 A diagrammatic representation of a cell showing B the cytoplasmic membrane; A the cytoplasm C a membrane of an organelle of the cell; D the internal fluid of the organelle.On the right the region of the cytoplasmic membrane between the two lines is expanded to show the way in which a cation X could be transported by a carrier C across the membrane. Energised transfer requires a modification of C at the interfaces as shown outer cell membrane (Figure 1). Thus both processes are regulated by the energy supply the concentration of ATP.6 Initially then we looked for a relationship between log [K+]in/[Na+]in and log [Mg2+]in/(Ca2+]in in very similar cell systems-the blood cells of different animals where the concentration of the external solution (to the cells) is the same.6 Figure 2 shows the result. For a wide range of animal cells a linear relationship holds so that we may presume that the concentration gradients of K+ and Na+ and Mg2+ and Ca2+ are limited by the same primary source of energy ATP remembering that this energy is being used against a constant flux opposing the concentration gradient.It follows that the relationship in Figure 2 could be a consequence of a membrane permeability decrease and/or of increase in metabolic activity from cattle to duck erythrocyte cells. In fact it is metabolism which increases most markedly from 334 Williams +1.6 +1.0 r PI 6 Y 1 +0.4 r t-4 Y ba c1 -0.2 -0.8 -1.0 0 +1.0 +2.0 log IK+l / "a+] Figure 2 A plot showing the relative accumulation of cations by diflerent erythrocytes blood cells and the composition of the surrounding fluid serum for comparison.The points are explained in the text left to right.' Interestingly the accumulation of the cations Mg2+ and K+ is paralleled by an anion concentration gradient.+* [phosphatelin [chloride ]in dG = RTln (4) A very crude measure of the activity of a cell is provided by bound internal phosphate for it requires energy to condense phosphate with other inorganic and organic residues. It follows that we should inspect [PI [Cl-1 "a+] [K+] [Mg2+] and [Ca2+] changes together. There are marked exceptions to the general picture of Figure 2. The erythro- cytes of the dog the cat and certain sheep give a low ratio [K+]in/[Na+]in but quite a high ratio fMg"]in/[Ca2+]in. Thus it cannot be generally true that the univalent and bivalent metal concentration gradients are limited in the same way by energy.Amongst particular sheep the univalent ionic concentration gradient [K+]in/ma+]in of the red blood cell is genetically transmitted from 7 S. Rapoport in 'Essays in Biochemistry' vol. 4 ed. P. N. Campbell and G. D. Greville Academic Press New York 1968 p. 69. 8 W. E. C. Wacker and R. J. P. Williams J . Theor. Biol. 1958,20 65. 335 The Biochemistry of Sodium Potassium Magnesium and Calcium generation to generation and we may conclude that the limiting factor must be one based on a process specific to Na-' and Kf transport. This is now known to be due to a specific inhibition of the Na+/K+ ATP-ase protein. Low potassium blood-cells of cats and dogs must also arise from a lowered activity of the speci- fic K+/Naf ATP-ase and not from a liinitation of energy or an excessive mem- brane permeability for these factors would also affect the [Mg"]in/ [Ca2+]in and anion gradients.The above treatment of erythrocytes led us to a parallel consideration of other cell types which have the same environment e.g. the different cells of man.8p9 In general there is a close relationship between the following quantities [K+]in/ [Na+]in [Mg2+]in/ [Ca2+]in and [P]in/ [Cl-]in. Bonting and coworkerslO showed furthermore that the [K+]in/[Na+]in the [Mg2+]in and the [Plin in different types of cell of one animal the cat are connected closely to thecon- centration of Na+/K+ ATP-ase of its different cells. Even the different muscles of the body can be classified on the basis of their inorganic contents which in turn may be related to total ATP-ase acti~ities.~ Moreover the onset of many diseased conditions e.g.dystrophy and eye cataracts is accompanied by changes in all three concentration ratios towards high ma+] [Ca2+] and [C1-].s All this evidence points to the general conclusion that there is a strictly controlled separation of K+ from Na+ and of Mg2+ from Ca2+ in all living cells as we now see them. We stress at this point that these ion accumulations do not need six different pumps for the six different ionic components. The extrusion of Ca2+ and Na+ (and Cl-) and the uptake of K+ and Mg2+ (and P) could occur through separate transport processes for all four cations (and two anions) or the extrusion of one ion could be coupled to the uptake of another by hetero-exchange diffusion. Both cases are now known. In Escherichia cuZi,11~12 erythrocytes,13 and in liver slices the extrusion of Ca2+ and Na+ and the uptake of Mg2+ and (K+?) are independent processes.In cardiac muscle1* and in squid axons16 the extrusion of Ca2+ is linked to the inward movement of Na+. (Curiously liver mitochondria transport Ca2+ independently of Na+ movement so that free Ca2+ in liver cells is not closely linked to free Naf.) The exchange of Na+ for Ca2+ in cardiac muscle and squid axons employing a common carrier mechanism is of great interest as these ions have identical sizes and show isomorphous replacement in many minerals. The external environments of all these cells are far from identical. The extreme conditions of living in fresh water mean that a biological system has to maintain all four cations against concentration gradients. It is still the case that K+ and R.J. P. Williams Bioenergetics 1970 1 215. J. W. Dicks and D. W. Tempest J. gen. Bact. 1966 45 347. H. J. Schatzman Experimentia 1966,22,364; E. J. Olson and R. J. Cazort J . Gen. Pliysiol. l o S. L. Bonting K. A. Simon and N. M. Hawkins Arch. Biochem. Biophys. 1961 95 416. l2 J. E. Lusk R. J. P. Williams and E. P. Kennedy J . Biol. Chem. 1968,243 2618. 1969,53 31 1 ; G. D. V. Rossum J . Gen. Physiol. 1970,55 18. l6 H. Reuter and N. Seitz J. Physiol. (London) 1968,195,457. lb M. P. Blanstein and A. L. Hodgkin J. Physiol. (London) 1968,198,468. 336 Williams Mg2+ are accumulated preferentially as compared with Na+ and Ca2+ respec- tively. At the other extreme life in sea water which is 490 Na+ 9-8 K+ 54 Mg2+ and 10 mmole 1-1 Ca2+ the cations Na+ and Ca2+ are strongly rejected Mg2+ is somewhat rejected or held closely in balance (its internal concentration is usually lower than in the sea) while K+ is accumulated against the concentration gradient.It remains true that the preferential pumping causes the gross rejection of Na+ relative to K+ and of Ca2+ relative to Mg2+. These statements are not affected by the binding of cations in cells. Firstly Na+ and K+ are only weakly bound so that analytical concentrations and free concentrations are closely related. Secondly although the binding of Mg2+ in cells is very considerable closely following phosphate accumulated the binding constants to enzymes for which magnesium is required and the known in-cell activity of these enzymes puts the free [Mg2+] at 1-10 mmole 1-I while the free concentration of Ca2+ is probably as low as from to lo-' mole 1-l.3 The General Use of Ion Distributions The ability to reject Na+ and Caa+ and to accumulate or maintain K+ and Mg2+ have led to a striking functional differentiation of these cations in biological systems. It is probable that the rejection of sodium was required early in the development of cells in order to maintain osmotic balance. Sodium ions still serve this function. It was also necessary that some other ion potassium should therefore neutralise the anionic groups of the biological macromolecules built up in the cell. In bacteria the separation of the elements could then be evolved into control systems associated with replication. In more complex living systems which have protected environments the cell membranes became more permeable.Thereby the Na+ and K+ gradients set up in the higher organisms established membrane potentials and it is these potentials which have permitted the develop- ment of nerve muscle and brain. The rejection of calcium has led to its partial use in membrane potentials associated with some muscles (cf. sodium) but largely it is used as an external structure-forming element. Shells bones and cell walls of crustacea animals bacteria and plants are frequently calcium compounds but the organisation of single cells into multicellular systems also depends on this cation. As calcium has a high affinity for many proteins it can be used as an initiator of structural change provided that its concentration can be dynamically controlled (see later) and it can also be used as a cofactor of enzymes external to cells.Magnesium is not normally involved in this way for it is maintained at a high concentration in the cell. There it stabilises many structures and acts as a general but weak acid catalyst. The utilisation of the four elements can be seen in more detail by considering first single-cells e.g. bacteria and then more organised cell systems. A. Cation Gradients and Bacteria.-Bacteria have relatively impermeable membranes and they use their metabolic energy very largely in replication i.e. synthesis. We12 and othersl1tl6 have followed the inorganic composition through l 8 D. W. Tempest J. W. Dicks and D. C. Ellwood J . gen. Illicrobiol. 1966 45 135; H. Y . Neujahr Biochirn. Biophys. Acra 1970 203 261. 337 The Biochemistry of Sodium Potassium Magnesium and Calcium the life cycle of bacteria in our case Escherichia culi and in different growth conditions.A colony of bacteria at rest has a rather low in-cell content of K+ Mg2+ and P and has associated with it relatively high Na+ Ca2+ and C1-. This resting state can be maintained by limiting organic or inorganic materials. When the limitation is removed no matter what it is the bacteria pick up the required materials for duplication. Analysis shows that the increased rate of growth is paralleled by increases in [K+]in and [Mg2+]in and in-cell phosphorus and that these increases are generated through cation-linked ATP-ases similar to those found in erythrocytes.17 Moreover the initial reaction step of the bac- teria as they move out of stationary conditions is to raise the inorganic gradients.It has been shown in vitro that the r81e of the magnesium and potassium is the stabilisation of RNA and the protein- RNA- and DNA-synthesising machinery. In vitro experiments also show that RNA molecules require ca. mole 1-1 free Mg2+ if they are to bind to other RNA molecules. Bacteria can pump Mg2+ to this level from an external solution of mole 1-l. The exact geometry of magnesium binding in RNA is important. For example the substitution of Mg2+ by Mn2+ in the protein-synthesising system leads to a mis-translation of the genetic code.18 In a general way the genetic code is a magnesium code and not just a code of base triplets. The levels of sodium potassium and magnesium are also important for the structure of bacterial DNA and in several en- zymes which will be described below.The general relationship between ion gradients and the bacterial life cycle is shown in Figure 3. Na+,Ca2+ HARD ( o u n - STRUCTURES ORIGINAL CELL ION P STRUCTURE - ENERGY GRADIENTS PLUS NUTRIENT K+ Mg2 + NEW RNA DNA AND PROTEINS \ NEW CELLS Figure 3 A scheme of the relationship between ion gradients and the growth cycle in bacteria. In many other types of cells the steps leading to synthesis are replaced by a loss of energy through ion di’usion l7 J. C. M. Hafkenscheid and S. L. Bonting Biochim. Biophys. Acta 1969 178 128. la H. R. Mahler and E. H. Cordes ‘Biological Chemistry’ Harper and Row New York 1966 p. 737. 338 Williams B. Cation Gradients and Higher Cells.-Whereas bacterial cells are stable in a great variety of ionic media advanced animal cells have developed in a con- trolled ionic medium.The medium blood serum is controlled by pumping mechanisms akin to those in the cell membranes of bacteria but which are very different in different living creatures. In some forms of life it is the outer skin (the frog) while in others it is the kidney (mammals) the gills (fish) or even a nasal salt gland (seabirds) which assist this control of the ionic medium. The blood stream in most of these creatures is a high sodium low potassium and a dilute magnesium and calcium solution. (There are numerous exceptions to this statement and for example the blood stream of insects is of totally different ionic composition.) Many of the cells of the more advanced living systems do not multiply rapidly although they accumulate cations Table 2.As the membranes of these cells are permeable to cations a major utilisation of their energy is that of maintaining ionic gradients. The mobilities of the ions coupled with their gradients which closely parallel those in bacteria8 can then be used in the following new ways. (1) As stated above the gradient can produce a junction potential at the membrane of say K+ and Na+.19 If the mobility of one of two ions through the membrane is much less than the other then a simple potential V is set up [see equations (1)-(3)] The sign of the potential depends on which cation is mobile for the concentration gradients are opposed. As the membrane is open to perturbation the potential can be reversed by external changes-such is the basis for our senses. The best- known example of the utilisation of the sodium and potassium gradients is in the nerve.(A nerve message is an electrolytic depolarisation wave.) Similar gradients exist across muscle membranes and the membranes of other excitable cells. In muscles of molluscs earthworms and perhaps in many slow muscles the depolarisation or action wave is propagated by calcium instead of sodium inward fluxes.20 The levels of magnesium inside and outside the cells are critical to the processes too but the lack of good radioactive magnesium isotopes has hindered a knowledge of Mg2+ fluxes. All in all then the use of the analytical con- centration gradients described earlier is controlled by changes in the membranes. The development of the dynamic interaction between membrane state and ionic gradients was a major evolutionary change following cell organisation.(2) The gradient represents a store of energy and it can be dissipated either in exchange reactions so as to reject or accumulate other chemicals e.g. amino- l 9 B. Katz 'Nerve Muscle and Synapse' McGraw-Hill New York 1966. 2 o S. Hagiwara and K. Takahashi J . gen. Physiol. 1967 50 583; Y. Ito H. Kuriyama and N. Tashiro J. Expt. Biol. 1970,52 59; P. F. Baker J. Gen. Physiol. 1968,51 172. 339 The Biochemistry of Sodium Potassium Magnesiiirn arid Calciiim acids,21 or in coupling with other chemical reactions e.g. the high energy inter- mediates of oxidative and pho to-phosphoryla t ion.22 (3) The chemical distribution of ions can be used as a basis of reaction control for the different ions K+ Na+ Mg2+ and Ca+ are associated with catalysts selectively located inside and outside the cell.By varying their concentrations the level of enzyme activities is effectively varied (see later). C. Internal Cell Membranes and their Gradients.-Advanced cells are composed of an outer cell wall and membrane and several well-separated inner-cell compart- ments (Figure 1). The inner-compartments are separated from the bulk of the inner fluid the cytoplasm by membranes. Thus the nucleus and mitochondria and chloroplasts (energy-producing units) and many vesicles have independent means of controlling the concentrations of cations. In fact some of the compart- ments act as stores for cations e.g. the sarcoplasmic reticulum of We return to these stores in the last section. In other cases the inner working of the compartment is very sensitive to ion concentrations.The structures of certain nuclei are known to depend on the amount of sodium and magnesium to which they are exposed.24 This has been observed directly in the ‘puffing’ of chromo- somes and is thought to be connected with differentiation. Again the mito- chondria can pick up calcium to such a degree that they fill their inner space with a type of calcium hydroxy-phosphate-they make ‘bones’.25 As yet little is known of even the analytical details of these processes. D. Summary of Biological Systems.-The diverse functions of the four cations in all these systems of outer and inner membranes and outer and inner solutions pose questions as to the chemical selectivity of interaction between them and the biological molecules. The selectivity in vivu arises in two ways.The membranes generate gradients of the ions and must interact with them in a highly selective manner. Once inside the cell selectivity of action can be based on intrinsic selection for one ion rather than another or upon the different permitted con- centration levels of the ions. Summarising four different types of selective binding have been devised. (1) Na+ > K+ (2) K+ > Na+ (3) Mg2+ > Ca2+ (4) Ca2+ > Mg2+. In the following an attempt will be made to solve the molecular problems involved in the generation of these series. Starting from our knowledge and understanding of the simple complex-ion chemistry of the cations we shall proceed to protein interactions with the metal ions and finally to the consider- ation of highly organised systems. A satisfactory attack on the problem can be made only with the help of studies of series of cations and of probe methods which we shall describe in all the sections.The account starts with a description of the chemistry of the cations. E. Riklis and J. H. Quartel Canad. J . Biochem. Physiol. 1958 36 347. ** P. J. Garraham and I. M. Glynn Nature 1966 211 1414. 2a S. Ebashi and M. Endo Progr. Biophys. Mol. Biol. 1968 18 123; W. Hasselbach and M. Makinose Biochem. Z. 1963 339 94. a‘ M. Lezzi and H. Kroeger 2. Naturforsch. 1966 21b 274. Is A. L. Lehninger Biochem. J. 1970 (Jubilee lecture). 340 4 The Chemistry of the A-Subgroup Cations26 A. Structural Features of Group IA Compounds.-Some crystal structure data for the Group IA cations are given in a previous article.27 The obvious fall in co-ordination number is from Li+(4) through Na+(6) to K+(8) Rb+(8) and Cs+(lO?).The change is an example of the well-known radius ratio effect which was introduced into discussions of ion-packing in crystals by Pauling.28 The binding of macrocyclic l i g a n d ~ ~ ~ by the Group IA cations has recently been shown to involve high co-ordination numbers20. The difference in co-ordination number is connected with the degree of hydration of the cations in their salts. When bound by complicated or large anions potassium has a lower hydration than Naf e.g. in the salts of PtC&’-. B. Stability Constants and Solubilities of Group IA Metals.-There are two general stability sequences for Group IA cation complexes which are also reflected in the solubility of their salts.2s For the most part the anions of the simple weak acids e.g.hydroxides give the stability and insolubility order Lif > Na+ > K+ > Rb+ > Cs+. Such anions are small. The anions of large strong acids give the reverse order. A more selective order for cations in the middle of the series can be obtained by changing the nature of the co-ordinating atom to some intermediate type or by generating a stereochemical relationship Table 3 Some stability constants and other data for Group IA cations Li Na K Rb Cs Ref. edta (log K) 2.8 1.7 0 - - a P20,4- (log K) 3.1 2.3 2.3 - 2.3 a Dibenzoylmethane (log K) 5.9 4.2 3.7 3.5 3.4 a NO,- (log K) -1.0 -0.4 0 - 0.1 a (log K) 0.6 0.7 0.9 - - a Ring Chelate XXXI (log K) 0-0 0.0 2.0 1.5 1.1 b Football ligand (log K) - 3.6 5.1 3.7 - C Substituted picrylamine anion (extraction const ant) 1.0 1.8 3.7 4-2 5.2 d U‘Stability Constants’ Special Publication No.17 ed. A. Martell and L. G. Sillen Chemical Society London 1964. bR. M. Izatt J. H. Rytting D. P. Nelson B. L. Haymore and J. J. Christensen Science 1968 168 443. CB. Dietrich J. M. Lehn and J. P. Sauvage Tetrahedron Letters 1969 2889. dJ. Rais and M. Krys J . Inorg. Nuclear Chem. 1969 31 2903. C. S. G. Phillips and R. J. P. Williams ‘Inorganic Chemistry’ vol. 2 Oxford University Press Oxford 1966 p. 48 et seq. R. J. P. Williams in ‘The Protides of The Biological Fluids’ vol. 14 ed. H. Peeters Elsevier Amsterdam 1967 p. 25. L. Pauling ‘The Nature of The Chemical Bond‘ Cornell University Press Ithaca 1948 p. 335 et seq. y y C . J. Pedersen Chem. Eng. News 1970 26; J. Amer. Chem. Soc. 1970 92 391; B. T. Kilbourn J.D. Danitz L. A. R. Pioda and W. Simon J. Mol. Biol. 1967 30 559; M. Truter unpublished observations. 341 The Biochemistry of Sodium Potassium Magnesium and Calcirrm I 2,5,8,15,18,21-Hexaoxatricyclo (20.4.0.09714 )- hexacosane (Crown X X X I ) (Dibenzo- 18-Crown 6) I I Val ino m y cin I l l Nonactin Figure 4 Some biological and synthetic ring chelates. Note that all the donor groups are oxygens of ethers or carbonyl and that the exterior of the molecules are such as to make their complexes soluble in hydrophobic solvents Williams between ligand geometry and cation size. Both situations arise through the influence of radius-ratio effects as shown in the appendix to this article. The same influence effects the binding of macrocyclic ligands such as those designed by Pedersen 29 (Figure 4).These ligands can give rise to almost any order of binding constants30 (see for example Table 3) for they can be so designed that the ‘hole’ size they generate best matches the cation size of any one of the Group I A metals. There are biologically important ligands which have similar ring structures and bind potassium in preference to sodium valinomycin or sodium in preference to potassium actinomycin (Figure 4).31 It is not correct to conclude that macrocyclic systems only will generate the order K+ > Na+ found in biology though they may show the greatest selectivity Figure 5. We must keep in mind the following other observations. The ex- traction and precipitation of Group IA cations by perchlorate tetraphenylbor- ate and picrylamine anions (Table 3) can give the inverted order Cs+ > Rbf > K+ > Na+ > Lif.In many such cases potassium enters a ‘hydrophobic’ medium easily losing its hydration whereas relatively speaking sodium does not. It is important in biological systems that the reagents which give the selectivities Kf > Na+ and Na+ > K+ should also exclude calcium and magnesium. Such selectivity can be produced for potassium by ring ligands as in size K+ > Naf = Ca2+ > Mg2+ but such rings should be heavily blocked by barium. Barium is not extracted by picrylamine anions nor is it precipitated by tetraphenylborate. Thus the two types of system can be distinguished-hydrophobic systems may well produce the greatest selectivity for potassium. It is more difficult to devise ligands which accept sodium but exclude calcium though some weakly polar ring ligands may be capable of this selectivity.Before describing the reasons for the changing order in more detail we turn to the parallel case of the bivalent cations. C. Structural Features of Group IIA Cations.2s-Some details of the structures of Group IIA cations are given in previous article^^^^^^ which illustrate the increasing co-ordination number from Be2+(4) and Mg2+(6) to Ca2+(8) Sr2+(8) and Ba2+(8). The difference is established in the binding of complex anions as well as in that of simple anions. Just as with Group IA the radius ratio effect which is reflected in these co-ordination numbers also differentiates between magnesium and calcium in that magnesium salts often remain hydrated when calcium forms salts of low hydration.There are two types of example. Probably the strong acid anions are the best known examples e.g. sulphates and per- chlorates with which magnesium remains hydrated but calcium is dehydrated. The critical dependence on radius is shown in the series of increasing radii MgS0,,7H20; MnS0,,7H20; CaSO,,H,O; SrSO,; BaSO,. DNA and RNA are strong acid diester-phosphate anions and Mg remains hydrated when * O R. M. Izatt J. H. Rytting D. P. Nelson B. L. Haymore and J. J. Christensen Science 1968,168,443. I1 C . Moore and B. C. Pressman Biochem. Biophys. Res. Comm. 1964,15 562. 343 The Biochemistry of Sodium Potassium Magnesium and Calcium 4.0 3.5 3.0 2.5 % g 2.0 - 1.5 1 .o 0.5 0 Isomer A Isomer B -- ~ I I I Ag+ A. KS B A \Rb+ I,’ 0.3 0.4 ‘ 0.5 0.6 0.7 0.8 0.9 Ratio diameter of cationldiameter of hoie Figure 5 The logarithm of the stability constants of the contplexes of ligand XXXI (see Figure 4) plotted against the cation radii; A and B are two diferent isomers of ligand XXXI.(With permission of Prof. J. J. Christensen and Prof. R. M. Izatt) bound to them as shown by the work of Peacocke Sheard and Richards.32 Presumably calcium would be much more dehydrated and would be structure- forming on binding these phosphate-esters. It is a somewhat general feature of caIcium chemistry as opposed to that of the other three cations that it readily acts as a bridge a ‘cement’ between anions inducing precipitation. 32 A. R. Peacocke B. Sheard R. E. Richards J. Mol. Pliys. 1969 16 177. 344 Williams D. Stability Constants and Solubilities of Group IIA Cations.-As in Group IA the sequence of stability constants2s with many strong acid anions e.g.SO,2- and NO3- is Ba2+ > Sr2+ > Ca2+ > Mg2+. In chemical systems magnesium does not bind strongly to organic sulphonic acid residues and it is calcium which is expected to be associated with these and other strong acid anions in biological systems. By way of contrast weakly acidic and neutral groups such as amines bind to magnesium much more strongly than to calcium and magnesium occurs in some nitrogen-complexes in biology e.g. chlorophylls. Selectivity of complex ion formation of Group IIA cations with weak acid anions follows very similar patterns to that in Group IA Table 4.28933 The effect TabIe 4 Some stability constants* for Group IIA cations (log K) Acetate Oxalate GI ycine Imidodiacetate Nitrilotriacetate edta Sulphate Phosphate (S.P.) ATP Carbonate (S.P.) Football ligand egtat (S.P.) Mg2+ 0-82 3.4 3.4 2-9 5.3 8.9 5 4 2.0 0.0 24.0 4.2 7.5 2.0 Ca2+ 0.77 2.0 1.4 2-6 6.4 10.7 10-7 2.3 5.0 27.5 4.0 8.5 4.1 Sr2+ 0.44 2.5 0.9 5.0 8-8 8.1 6-5 27.4 3.5 9.0 13-0 - Ba2+ 0.41 2.3 0.8 1.7 4.8 7.9 8.0 10.0 22.5 3.3 8.5 15.0 - *Data from ‘Stability Constants’ Special Publ.No. 17 ed. A. Martell and L. G. Silltn Chemical Society London 1964. fegta is 2,2’-Ethylenedioxybis(ethyliminodiacetic acid); S.P. is solubility product data; ATP is adenosine triphosphate. $B. Bietrich J. M. Lehn and J. P. Sauvage Tetrahedron Letters 1969 2889. of steric hindrance the radius ratio effect is now much more marked however for the magnesium cation is very small. The data in Table 4 show that even with carboxylate groups as the complexity of the ligand increases so the stability order changes e.g.Acetate edta Mg2+ > Ca2+ > Sr2+ > Ba2+ Ca2+ > Sr2+ 3 Mg2+ > BaZ+ Hydroxy-acids also give the second order. Again although all acetates of Group IIA are soluble and magnesium forms a stronger acetate complex than calcium the solubility of oxalates follows the pattern Mg2+ > Mn2+ > Ca2+ < Sr2+ < 35 R. J. P. Williams J . Chem. Suc. 1952 3770. 345 The Biochemistry of Sodium Potassium Magnesium and Calcium Ba2+. Such a pattern is due to the difficulty of packing large carboxylate groups as opposed to several water molecules around a small cation-a radius ratio effect. It is not a field effect for the field of a single weak acid anion even in the presence of water molecules is such that the binding order is invariably Mn2+ > Mg2+ > Ca2+ > Sr2+ > Ba2+.The solubilities of phosphates and carbonates also follow the order Ca2+ > Mg2+ through a radius ratio effect. Undoubtedly this has led to the utilisation of calcium carbonates and phosphates in the external hard structures of living things partly explaining why the path of evolution has led to the increasing rejection of calcium. (There are close chemical parallels here with man-made cement-forming chemicals.) Thus a site of a protein which binds magnesium need not bind calcium so strongly or alternatively a calcium binding site may be a very poor site for magnesium and this distinction will be critically dependent upon the number of co-ordinating centres as well as on their type.As was pointed out many years ago ring chelates can be made the basis of Group IIA cation ~electivity.~~ The best examples are provided by the work of Christensen and Izatt and their coworkers31 (Figure 5) and by the ‘football’ ligands shown in Table 4. Such stability sequences also arise from radius ratio effects (see Appendix). Exactly the same stereochemical and chemical features affect the stability constants of the lanthanide series of complexes and almost any order of stability or solubility can be generated by a controlled use of binding groups and their geornetrie~.~~ In all cases it is the radius ratio effect which is being utilised as we show in the Appendix to this article. Whereas the sites for the two bivalent ca- tions Mg2+ and Ca2+ can be designed so as to bind monovalent cations weakly e.g.edta the calcium sites can not be designed to exclude lanthanides. Perhaps it is fortunate that biological systems have not had to face this problem. Table 5 Probable biological ligands of do cations Example Order of stability -OS03- -0P(OR)02- Ligand type Strong acid anions K+ > Na+ Ca2+ > Mg2+ Weak acid anions -OP0,2- P043- -coz- co32- Neutral oxygen donors Alcohols and ethers All orders are possible depending upon radius ratio considerations and ’especially upon the number of ligands co-ordinated Neutral nitrogen donors -NH2 imidazole Mg2+ > all others Li+ > Na+ > K+ 3 4 R. J. P. Williams Analyst 1953 78 586. Press Oxford 1966 p. 106 et seq. C. S. G. Phillips and R. J. P. Williams ‘Inorganic Chemistry’ vol. 2 Oxford University 346 Williams We see from this summary of the chemistry of the four elements with which we are concerned that the reason for the selectivity orders found in biology may well be readily recognisable if we know something about the chemical nature of the binding groups whether they are weak or strong acid anions and about the steric crowding involved (see Table 5).Such chemical groupings can often be recognised by a study of their complexes with an extended series of cations. The final definition of the binding groups and their stereochemistry is a task for crystallography and for certain forms of spectroscopy. In what follows we indicate how spectroscopy can be used within the limitations imposed by isomorphous replacement. Table 6 lists possible ion substitutions. A summary of the use of this procedure has been given by Vallee and Williams with special reference to zinc enzymes which are outside the scope of this review.Table 6 Possible probe ions for substitution Native cation Substitution* Na+ (0.95) Li+ (0.60) (poor) Mg2+ (0.65) Ca2+ (0.99) K+ (1.33) T1+ (1.40) Rb+ (1.48) CS+ (1.69) NH4+ (1.45) Mn2+ (0.80) to Zn2+ (0.65) Eu2+ (1-12) Mn2+ (0-80) La3+ (1.15) to Lu3+ (0.93) U022+ ( M 1.1) etc. *Ionic radii are in parentheses. 5 Probes for Group IA and IIA Metals A. Probes for Sodium.-Sodium complexes can be studied directly by n.m.r. using 23Na. Relatively little work has been done in this field as yet.36 Unfortu- nately a search of the Periodic Table shows that no other cation is likely to mimic the properties of sodium closely so that a detailed understanding of its biochemistry must depend on an extension of the n.m.r.method and not on isomorphous substitution. Lithium does not replace sodium in a biological system. The differences between the two are exemplified by the use of lithium salts in the treatment of nervous disorders. Unfortunately too very few enzymes depend on sodium so that its properties can not be studied even in relatively small molecules. B. Probes for Potassium.-In this section we shall show how series of cations and their physical properties can be used to study potassium in biochemical sys tems. Potassium is the main Group IA cation which activates enzymes. Table 7 lists some of the enzymes which have been shown to be potassium dependent and includes the order of effectiveness in them of different univalent cations.36 F. W. Cope Proc. Nat. Acad. Sci. U.S.A. 1966 54 225. 347 The Biochemistry of Sodiiim Potassium Magnesium and Calciirm Table 7 The eflect of thallium on several enzymes and other biological systems Biological system Diol-dehydratase (BIZ) Pyruvate kinase Phosp ha tases Na/K ATP-ases (K-func tion) Erythrocyte transferases Muscle excitation Order of cation efficiency Ref. T1+ > NH4+ > K+ > Rbf > Cs+ > Na+ > Lif a Tlf > K+ > Rbf > Csf > Na+ > Lif b Tl+ > Kf > Rb+ > Cs+ > NH4+ > Naf Li+ C Tlf > K+ > Rbf > Csf > Naf > Lif T1+ moves with Kf f Tlf > Kf > Na+ Li+ 4 e g aM. E. Foster and R. J. P. Williams to be published. bG. K. Radda and R. J. P. Williams to be published. CC. E. Inturrusi Biochem.Biophys. Acta 1969 173 567; ibid. 1969 174 630. d J . S. Britten and M. Blank Biochem. Biophys. Acta 1968,159 160. eP. J. Gehring and P. B. Hammond J. Pharmacol. 1967 155 187. fP. J. Gehring and P. B. Hammond J . Pharmacol. 1964 145 215. QL. J. Mullins and R. D. Moore J . Gen. Physiol. 1960 43 759. There are probably many additional enzymes which are affected by potassium but which have not been properly studied as potassium is usually added to the buffers used in the study of enzymes. It is immediately apparent that cations of the same size as potassium i.e. TI+ and NH4+ are equally or even more effective than K+ itself whereas smaller or larger cations are relatively ineffective. The maximum in the plot of activity against radius (Figure 6) is due to the strength of binding Km of the cations of a particular size and is not necessarily reflected in the maximum velocity V,,,.at which the enzyme can operate when the cation is present in saturating concentrations (Figure 6). This suggests that the binding site is of a size which ‘matches’ the potassium radius. The nature of the binding groups are not known but we can examine various possibilities through the use of probes. As shown in Table 6 the two best probes for potassium are thallium(1) and caesium (Figure 6). The physical and chemical properties of caesium systems are under study by Professor R. E. Richards and we have concentrated upon a study of thallium@. In Table 8 is a series of stability constants for its complexes and a comparison with potassium where possible. Thallium(1) binds considerably more strongly than potassium or sodium T1+ > Na+ > K+ to all the weak acid anion ligands we have studied and to the cyclic ethers XXXI.The differences increase with increasing ligand charge. Strength of binding however Tlf > Na+ > K+ is known not to be the only source of the selectivity in the order of enzyme activation TI+ > K+ > Na+ (see Table 7 or Figure 6). Presuni- ably thallium is so effective as its radius is almost identical with that of potassium and because it has a greater binding power to the enzyme ligands. There is the indication here that the potassium site has one weak acid anion group associated 348 10 5 0 Li+ Na+ K+ T1+ NH4+ Rb+ Cs+ Figure 6 The eflect of diferent metal ions upon the maximum activity of propylenegIycol dehydratase arid the logarithm of the binding constant of the metals to the enzyme.The cations ( X ) are plotted in the order of increasing radius Table 8 The stability constants and absorption spectral data of some thallium(1) comptexes at ionic strength 0 . 1 5 ~ Ligand HPOd2- PO4'- P,o,~- HP,o,~- Ribose-5-phosphate2- Adenosine diphosphate3- Adenosine triphosphate4- E t h y lenediamine te tr a-ace t a te4 - Nitrilotriacetate3- 1OgloKTl 2-25 0.75 3.05 2-35 0.90 1-20 2.00 5.8 4.4 hmax(nm) logdK 230 225 227 1.5 21 9 219 1.0 246 1.0 243 1.0 with it. The site of binding can be found in principle by using the physical properties of thallium(1) complexes as follows. Firstly the TP 7s - 7p (triplet) excitation which gives an intense absorption band at 215 nm in the aquated cation moves considerably and differentially on 349 The Biochemistry a f Sodiiim Potassiim Magnesirrm and Calcium ligand binding (see Table 8).[Unfortunately most ligands quench the fluor- escence of thallium(1) so that this property is of restricted use.] Thallium(r) has a significant temperature independent paramagnetism and it is probably this property that causes the large observed shifts in proton and phosphorus reson- ances of its bound l i g a n d ~ . ~ ~ We have studied the proton resonances of for example thallium(1) ethylenediaminetetra-acetate. The shifts on binding TI1 ‘CH2 of acetate -0.21 p.p.m. ‘CH of ethylene -0.17 p.p.m. The shifts of phosphorus resonances on binding T1I are pyrophosphate - 1.4 p.p.m. ; adenosine diyhosphate a - P = -2.0 p - P = - 1.3 p.p.m.; adenosine tri- phosphate 01 - P = -0.5 p - P = -2.2 y - P = - 1.0 p.p.m.A full pH study has been carried out on these series of compounds. The shifts make TI1 a very effective probe for the binding groups of potassium but as yet we have not tested the procedure with the enzymes of Table 7. nucleus (thallium-215) so that a direct study of thallium nuclear resonances is feasible. Model studies have been made by Richards and and an initial series of measurements on a protein pyruvate kinase has been made by Kayne and Reuben.38 In this system it has proved possible to use a double probe TI1 and MnII. When more is known of the model chemistry of TF it seems likely that it will make an excellent probe. An extension of T1I probe studies is possible not only in the field of enzymes but also in the study of DNA and RNA structures and in membrane transport.The cyclic ethers which bind potassium also bind thallium very effectively so that the thallium probe can be introduced into an organic membrane easily. It is already known that erythrocytes will accumulate thallium(1) mistaking it for potassium and that thallium(1) will activate the muscle spike potential in place of potassium (Table 7). In order to see if these sites are very like those found in the enzymes Dr. M. E. Foster and myself have tested the activity of the diol dehydratase enzyme with molecules which are known to block potassium action in nerves e.g. NMe,+. The organic cation has no effect and presumably the site of K+ absorption in nerves is more hydrophobic than that in enzymes. are / / Thallium(1) has a spin = C. Probes for Magnesium-A large number of enzymes are activated by magnesium and calcium and a considerable number of proteins bind them differentially.The role played by the metal can be merely structural or catalytic at the active site. It is generally true that magnesium is the activator of intra- cellular enzymes where its concentration is greater while calcium activates enzymes external to the cell. The relative effectiveness of series of cations in the intracellular enzymes is shown in Figure 7 where it is clearly seen that a magnesium enzyme is usually poorly activated by calcium (and strontium or barium). s7 J. P. Manners K. G. Morallee and R. J. P. Williams Chem. Comm. in the press; R. P. Gasser and R. E. Richards Mol. Phys. 1959 2 357. a E F. J. Kayne and J. Reuben J . Amer. Chem. SOC. 1970,92,214.3 50 Williams Figure 7 A comparison of the relative effectiveness of different cations in non-enzymic (a) and three diferent enzymic reactions. (a) is drawn schematically to show the increasing effectiveness of cations as the electron afinity of the cation increases and is generalisation of a large number of examples. (b) isphosphoglycerate kinase (c) ispyruvate kinase and (a) isphosphoglucomutase C ~ h n ~ ~ has divided the kinases in particular into two classes on the basis of their activation by different metals and from her studies of their physical properties. In the first class e.g. phosphoglycerate kinase there is less metal selectivity than in the second e.g. pyruvate kinase. Cohn concludes that the first group contains cases of metal-substrate-enzyme complexes whereas the J' M.Cohn Qirart. Rev. Biophys. 1970 3 61. 351 The Biochemistry of Sodium Potassium Magnesium and Calcium second group contains substrate-metal-enzyme complexes. Now it is a curious feature of biological systems that in general transition-metal cations are in- effective in magnesium enzyme^.^ By way of contrast in model reactions of small molecules and in the binding of either small molecules or proteins there is an invariant order of catalytic and binding power of bivalent cations (see Figure 7) Cu2+ > Ni2+ 3 Zn2+ > Co2+ > Fez+ > Mn2+ > Mgz+ > Ca2+. Presumably the transition metals bind the wrong co-ordination centres of the enzymes and are therefore ineffective as catalysts. I n vivo the concentrations of the free cations Cu2+ Ni2+ Co2+ Fez+ and Zn2+ are also so low that they do not compete for many protein sites and they are therefore not effective inhibitors.On the other hand Mn2+ is present in reasonable concentrations and its chemistry is there- fore very relevant to that of magnesium and calcium. Reference to Figure 7 for example shows that manganese will replace magnesium in many biological systems. This replacement is probably successful because the chemistries of magnesium and manganese as opposed to those of most other bivalent cations of the first transition series are rather similar. Shulman'* and Cohngg have made great use of this exchange in the study of magnesium binding to nucleic acids and enzymes for manganese(@ is an excellent probeeither through its e.p.r. signal or its paramagnetic perturbation of proton fluorine or phosphorus nuclear resonances.In principle and to some degree in practice a metal site can be mapped using such techniques. This work has been summarised recently3@ and we shall not describe it further here. Manganese is not a very good match for magnesium as judged by its radius and nickel would clearly be a much closer fit. Moreover nickel(1r) and magnesium have very similar geometric demands for they both have a strong tendency to octahedral geometry. Unfortunately as stated above the binding of nickel to ligands is much stronger and nickel tends to bind nitrogen rather than oxygen centres. Furthermore the exchange of nickel from a given site is slow. It is rarely the case then that nickel is a satisfactory probe for magnesium. However in the case of phosphoglucomutase Ray4' has shown that nickel is an excellent substi- ture for magnesium (Figure 7d).Nickel [like cobalt(11)1 offers many probe possibilities and Ray has looked at the absorption spectra of the nickel(@ and cobalt(11) phosphoglucomutase enzymes. The spectra show that the nickel geometry is strictly octahedral but that when cobalt(I1) occupies the site the geometry is irregular possibly 5-co-ordinate. The nickel but not the cobalt enzyme is active and we may suppose that nickel (magnesium) generates a special protein geometry because of its stereochemistry whereas another protein geometry obtains with the cobalt(@. The enzyme is not very active with mangan- ese (too big?) nor with zinc (wrong geometry?) so that isomorphous substitution has very exact demands with this flexible enzyme.It seems to be a general finding in the study of isomorphous replacement that substitution needs to be very exact if it is to be successful. The correspondence depends on size to the limit 4 0 J. Eisinger R. G. Shulman and B. M. Szymanski J. Chem. Pliys. 1962 36 1721. 4 1 E. J. Peck and W. J. Ray J . Biol. Cliem. 1969 244 3748. 352 Williams of 0.2 8 in radius and to bond angles probably to an equal degree of atom positioning say 10". An understanding of enzyme catalysis may well depend on appreciating structures to this degree of exactness. This is one of the reasons why Williams and Vallee42 have drawn attention to the special geometric factors operating in enzymes under the heading of the entatic state. Apart from thermodynamic considerations the rates at which magnesium and calcium react with ligands are very different and Eigen and ham me^^^ have proposed a possible mechanism by which one of these cations magnesium rather than calcium could be activating.The mechanism depends upon the slow rates of the forward and back reactions of magnesium permitting a con- formational change of a protein whereas this would not be possible with calcium. The thermodynamic explanation which we favour for the above phenomenon depends upon differences in geometry at equilibrium. There may well be examples of both types of behaviour. D. Probes for Cal~ium.~~-Both in binding to calcium-activated (usually extra- cellular) enzymes (Table 9) and in the binding of proteins such as t r ~ p o n i n ~ ~ Table 9 Extracellular calcium enzymes and enzyme precursors Protein Function Pro t hrombin Trypsinogen Staphylococcus aureus nuclease Amylase B.subtilis enzyme Bacterial protease Aryl sulphatase Sulphate ester hydrolysis Precursor of thrombin blood clotting enzyme Precursor of trypsin digestive protease Bacterial enzyme for DNA and RNA hydrolysis Hydrolysis of starches in digestion (Figure 8) the order of binding is Ca2+ > Sr2+ > Ba2+ > Mg2+. This order can be very closely matched by the ligand 2,2'-ethylenedioxybis [ethylimino- di(acetate)] (Table 4) or by the solubility products of the corresponding oxalates (Figure 8). Here radius ratio effects have been pushed close to the limit observed in the difference in solubility product between magnesium and calcium sulphates. It is a reasonable tentative hypothesis that the sites of binding of calcium in biological systems are multi-carboxylate or -phosphate centres.Already it is known that the staphylococcus nuclease in the presence of the ligand-a strong acid diester of phosphate and the trypsin precursor trypsinogen provide such ~e~ltres.*~ 4 y B. L. Vallee and R. J. P. Williams Proc. Nat. Acad. Sci. U.S.A. 1968 59 498. 43 M. Eigen and G. G. Hammes Adv. Enzymol. 1963,25 1. 4 4 R. J. P. Williams Proceedings of the International Congress of Pharmacology 1969 to be published. 4 5 A. Amone C. J. Bier F. A. Cotton E. E. Hazen D. C. Richardson and J. S. Richardson Proc. Nat. Acad. Sci. U.S.A. 1969 64,420. 353 The Biochemistry of Sodium Potassium Magnesium and Calcium I I I I 1 - Mg 2+ Mn2+ Ca2+ Sr2+ Ba2+ Figure 8 A comparison between the binding constant of metal ions to the protein troponin (a) and the solubility product (S.P.) for the corresponding metal oxalates (b) There are two potential probe cations for calcium other than the use of 43Ca n.m.r.A search of the Periodic Table (see Table 6) shows that europium(I1) and manganese(@ are bivalent cations with approximately the correct ionic radius. We have started a detailed study of the chemistry of Ed1 (Table lo) after observing in preliminary experiments that it was partially effective in stimulating muscle. Europium(r1) clearly lies between calcium(I1) and strontium@) in its proper tie^.^^ The use of europium(I1) has noteworthy advantages over that of manganese(@ as a calcium substitute for manganese(@ has a considerable affinity for nitrogen ligands and in this and in other respects e.g.the solubility of the sulphate it is like magnesium not calcium. Europium(n) has several possibilities as a probe. 4O E. Nieboer R. J. P. Williams and A. Xavier to be published. 354 Williams Table 10 Properties of europium(II) calcium and magnesium Property Ion size (A) Main co-ord. number log K (edta)* log K (egta) log K (Pa) Solubility of sulphate Europ ium(11) Calcium Magnesium 1.12 0.99 0.65 8 8 6 9-6 10.6 8.7 9.6 11.0 5.2 2.8 2.5 2.5 insoluble insoluble soluble *edta is ethylenediamine tetra-acetate pa is picolinate egta is 2,2’-ethylenedioxybis[ethyl- iminodi(acetate)]. The cation has an absorption band around 300 nm which is sensitive to the chelating groups and the europium nucleus can be used in Mossbauer studies. Table 11 shows the isomer shifts for some EuII and Eu1I1 compounds.There is Table 11 Isomer shifts for europium compounds (rnm/sec) EuF - 15.0 E u F ~ 0.0 Eu(HCO2)z - 13.0 Eu(HC02) 3 $0.20 EuO - 11.2 Eu203(cubic) +O-8 EuCO - 13.1 Eu2( oxalate) +0.35 Eu(OW2 - 13.0 Eu(OH)3 +Om55 EuSO~ - 13.8 Eu2(S0 4) 3 $0.35 Note. The sensitivity of the europium(I1) Mossbauer spectra is well illustrated by 0. Berkooz J. Phys. Chem. Solids 1969 30 1763. The data in the Table are unpublished results of C. E. Johnson E. Niebor and R. J. P. Williams. EuS - 11.6 no confusion as to which oxidation state is being examined in marked contrast with the situation in iron chemistry. Again the isomer shifts are extremely sensitive to the nature of the compound and the work of Berkooz on inorganic systems would suggest that changes in EuII ligand distances can be readily studied.Thus EuII Mossbauer could help in the examination of Ed1 proteins and enzymes i.e. calcium complexes in vivo. Again EuII is a 4f7 cation so that it can be used as a paramagnetic perturbation of the resonances of other nuclei. A study of the information available from this attack has been initiated. A further series of probes for calcium may well be provided by the fifteen Ianthanide~.~~ As we stressed at the beginning of this article isomorphous replacement is less sensitive to charge than to size. In biology Na+ interacts with Ca2+ (certain muscle membranes) Ba2+ interacts with K+ (nerve and muscle membranes). The lanthanides have about the same radius as Ca2-+ the same co-ordination number and the same sensitivity to steric effects. Already several 355 The Biochemistry of Sodium Potassium Magnesium and Calcium Table 12 Lanthanides in biological systems System Action Lobster axon Bone proteins Bacterial nuclease Mitochondria Ln3+ blocks calcium transport Ln3+ behaves as ‘Super-calcium’ Ln3+ competes for calcium sites Ln3+ inhibits at calcium site Ref.a b d C aM. Takata W. F. Pickard J. Y . Lettvin and J. W. Moore J . Gen. Physiol. 1966 50 461; M. P. Blanstein and D. E. Goldman J. Gen. Physiol. 1968 51 279. bA. R. Peacocke and P. A. Williams Nature 1966 211 1140. CP. Cuatrecasas S. Fuchs and C. B. Anfinsen J. B i d . Chem. 1966 242 1541. dA. L. Lehninger see ref. 25. observations have been made as to their effect in a biochemical system Table 12. If further work proves their value then a chemical and physical series of fifteen elements is available for probe studies.The physical methods made available by these probes includes every spectroscopic method and as they are heavy elements they are also useful in X-ray crystal structure and electron-microscope studies. We have made a start with a very detailed study of the proton n.m.r. of EuIII edta egta nta complexes and of the water proton relaxation rates of the hydrates of these complexes.47 We have also shown that gadolinium(I1r) sits exactly between the two carboxylate residues of ly~ozyme*~ and we have gone on to a study of the perturbation of the n.m.r. spectrum of lysozyme and its substrates by gadolinium(nI) europium(rrr) and holmiurn(~rr).~~ The chemical advantages of these probes can not be over-stressed. Many studies are available of changes in complex stability with atomic number in the lanthanide series.The effect of the radius ratio changes is seen in the structures degree of hydration and the stability constant sequences. We can imagine that the lanthanide complexes can be developed for many different probe purposes. For example they can be used to tackle the hydration state of membrane phases. In a series of complexes Ln(acac),(H,O)n the value of n and the binding strength of water varies systematically. Thus water activity can be followed and should this change on energising the membrane this change can be followed. Such deductions will be possible not only through the use of n.m.r. but also through fluorescence studies. The next few years should show how valuable such approaches will be. 6 Summary of Simple Binding Sites In the preceding sections I have demonstrated three points.The first is the effect of ion size upon the chemistry of the four cations as seen in structures hydration and in thermodynamic quantities. The co-ordination number the hydration and the stability of complexes are very sensitive to radius ratio effects but the influence of size need not be seen simultaneously in changes of all three 47 R. A. Dwek K. 0. Morallee E. Nieboer F. J. C. Rossotti R. J. P. Williams and A. Xavier to be published. 48 D. C. Phillips and co-workers unpublished results. 356 Williams properties. The second is the use of closely matched cations for substitution into Group IA and IIA sites as probes. The third is the study of long series of closely related cations as a guide to the chemistry of a given site.The study by these methods of some enzymes and proteins which bind the IA and IIA cations has been described but much work remains to be done. Even so there is good reason to suppose that we understand the basic reasons for the different selec- tivities of sites for the cations. The selectivities are Magnesium is bound preferentially by nitrogen bases Mg2+ > Ca2+ > Na+ > K+ and sites showing strong preference for this cation probably contain at least one such base and at least one phosphate or carboxylate group. Calcium is bound preferentially by multidentate anions and strong acid anions Ca2+ > Mg2+ > Na+ > K+. Sites showing strong preference for this cation could include phosphate carboxylate or sulphonate groups and no nitrogen bases. Magnesium and calcium bind equally and much more strongly than sodium and potassium to multi-anion sites.Potassium is taken up into a ‘Iarge’ (hydrophobic) site of neutral oxygen- donors or singly charged oxygen donors K+ > Ba2+ > Na+ > Ca2+ > Mg2+. Sodium combines with a rather ‘smaller’ (hydrophobic) centre of neutral or singly charged donors. Na+ b Ca2+ > K+ > Mg2+. Hydrophobic anionic sites exclude bivalent and small univalent cations. These general rules allow us to inspect more complicated biological systems in an effort to understand how cation selectivity arises. We consider firstly how the cation concentration gradients could have developed through membrane re- actions. The membranes to be considered are organic rather than aqueous phases. 7 Complex Biological Systems A.Ion Transport through Membranes.-A biological membrane is a thin organic phase into which cations can partition. In the case of sodium and potassium it is clear from the study of the above ligands e.g. ring ligand XXXI and picryl- amines that potassium can partition into such a highly hydrophobic environment but that this is much less readily achieved by sodium. As seen from the above a good transporting agent for potassium which will not accept sodium would then be one which was large and apolar thcugh it could carry a single negative charge. It is not required to be a ring ligand. Such a centre will not accept magnesium or calcium for they are at least as difficult to dehydrate as sodium. However weak competition by barium is possible and strong competition from thallium(I) rubidium caesium and ammonium is to be expected.Competition by all such ions is seen in the simple enzymes of Table 7 and in membrane transport but additionally membrane transport is blocked especially by tetra-alkyl ammonium salts which do not affect the enzymes. These cations act as effective drugs restricting the access of K+ to the channels through which 357 The Biochemistry of Sodium Potassium Magnesium and Calcium it moves in membranes. The membrane sites we presume are the more hydro- phobic and may not be of a restricted size. The binding of thallium to the potassium sites has not yet been studied quantitatively but it may reveal the nature of these sites. By way of contrast the sodium centre must be smaller and more highly polar than the potassium site and may have a group such as phos- phate ROPOZ2-.This would explain the competition by calcium at some sodium sites and could also account for the blocking effect of lithium which would bind more strongly than sodium and which is used in treating mental disorders. The site could be the phosphorylated protein which is associated with the Na/K ATP-ase of the sodium pump in all outer membranes (see Figure 1). The transport of calcium would also seem to involve phosphorylated proteins; for example a phosphorylated protein carries calcium in the blood stream and probably in the mitochondria1 membrane. The binding constant of these centres for Ca2+ (ca. lo6) and lanthanides (ca. log) indicates that there are probably two (or three) additional anionic (carboxylate?) groups as well as the phosphate at the binding centre.The calcium-binding protein discovered by Wasserman and his colleagues has binding constants for Ca2+ Sr2+ and Ba2+ which are very similar to those of troponin (Figure 8). This could be the transport protein of cell membranes in the kidney. The carriers for magnesium in bacterial membranes are more likely to have one nitrogen base probably imidazole and one or two carboxylate or phosphate groups (see earlier). Steric restrictions could be built in as in chlorophyll. The presence of the nitrogen base would imply that the carrier would bind transition metals such as cobalt nickel and manganese@) which could make excellent probes. Certain bacteria can be loaded with transition metals and perhaps the mechanisms of loading utilises the magnesium carrier. An important feature of transport is that it is often linked to metabolism.The function of phosphorylated carrier proteins for sodium and calcium may lie in the ease with which their formation can be linked to energy Protein + ATP --f Protein-P + ADP Hydrolysis at the opposite side of a membrane from phosphorylation (Figure 1) then yields an energy coupled system for rejecting Na+ and Ca2+ and perhaps also for accumulating K+ and Mg2+. The binding of a metal in a ring chelate as in many of the postulated membrane complexes may be such that the co-ordination of further ligands is restricted. In the ring chelates of porphyrin corrin and chlorin only certain metals in certain valence and spin states can sit in-plane. Others sit above the plane-for example magnesium sits out of plane in chlorophyll.Such a cation can be expected to bind but one additional ligand perpendicular to the ring plane and hence magnesium becomes 5-co-0rdinate.~~ Very small factors can now influence the exact binding of the ligand and the stereochemistry of the chlorin. In biological systems different types of magnesium chlorophyll arise with different 4 $ R. Timkovich and A. Talinsky J. Amer. Chem. SOC. 1969,91,4430. 358 Williams reactivities and both different iron-porphyrin and different cobalt corrin geometries have been studied. It is possible that the geometry of a ring chelate is controlled by the membrane state or even the site of the membrane at which it resides. Transport could then be coupled to energy by a conformational rather than a chemical change. Let us presume that we understand the transport problem.How does the cell utilise the unequal concentrations of cations which it has generated? We shall now elaborate somewhat on the introductory statements regarding cation function. We start with the outside of the cell. B. Crystallisation of Salts in Biology.-The solubility product of calcium salts is generally less than that of magnesium salts owing to the radius ratio effect. The precipitation of calcium carbonate oxalate phosphate and even fluoride commonly occur in biological systems. This precipitation is assisted by the rejection of calcium from the interior of cells and in many living systems the blood stream is super-saturated with calcium salts. It would appear that there are fibrous protein structures on the outside of some cells and these proteins act as initiators of crystallisation.Given such a fine kinetic control of pre- cipitation and solution bone and shell material can be transferred in the blood stream to be deposited in a new region. The growth of the skeleton of animals the deposition of shells of eggs and the building of many other structures de- mand this type of activity. As the system is in such close balance very small changes can bring about catastrophic faults. Let us assume that proteins and polysaccharides slowly become more oxidised to more anionic polymers with age which is thermo- dynamically speaking reasonable. They will then bind calcium slightly better as they age. These binding sites could lead to the initiation of crystal growth and thus the deposition of calcium salts. Is this why ageing is associated with calcium deposition in cataracts stones hardening of soft tissues and arteries? Table 13 Calcium binding compounds of cell walls Living system Binding chemicals (orgarzic) Inorganic" deposit Bones Chondroitin sulphate Calcium hydroxy- Shells and plant Pectic acids (Galacturonic acid) Calcium carbonate glycoproteins apatite celluloses Algae Alginic acid (mururonic acid) Fucoidin (polyfucose sulphate) Carragenin (polygalactose sulphate) Spores Picolinic acid Calcium picolinate *Calcium fluoride oxalate silicates and various organic carboxylates have also been reported.359 The Biochemistry of Sodinnr Potassium Magnesium and Calcium We can now see that the laying down of calcium salts as hard structures is a consequence of the radius ratio effect as is the binding of calcium to the outer saccharides and proteins of cells Table 13.Slowly the walls of bacteria and spores have been extended by evolution to the celluloses of plants and the bones and shells of animals. External calcium is also essential for the binding of cells to one another the ability of cell material to bind to surfaces (pseudo-pod form- ation) and repair of cell membrane. Additionally as we have seen above there is the dependence on calcium of digestive and other extracellular enzymic processes Table 9. C. Intracellular Ions.-Inside cells where [Ca2+] is ca. 10-7~ and [Mg2+] is ca. 1 0 - 3 ~ there are many substrates and proteins which will form complexes with either cation of stability constants ca. lo3-lo4. Thus only magnesium is bound e.g. to ATP ADP pyrophosphate RNA and enzymes like enolase and phosphoglucomutase.However in many cells e.g. muscle there are also sites with binding constants of los for calcium (Figure 8) and less than or about lo3 for magnesium. On exposing the inside of the cell to calcium at 10-5~ generated by an influx of calcium on exciting the membrane these sites become occupied and action is triggered for complex formation alters the protein geometry. The distinction between the first group and the second group of sites is probably no greater than that between NH(CH2C02-)2 and (CHzC02-)2N-CH2CHz.N (CHzC02-),. Thus while magnesium is an intracellular cofactor which is present in large concentration (> 10-3~) everywhere and adjustments in its concentra- tion can be used as a fine control on the level of enzyme activities calcium is a trigger which can be called into use by suddenly raising its in-cell concentration from 2 1 0 - 7 ~ to 10-5~ or by suddenly exposing proteins to the 10-3~-Ca2+ outside cells.Calcium can therefore affect major changes in constituents of low concentration. Inside cells K+ is > 100m while Na+ is l o r n so that a binding site which binds sodium ten times more strongly than potassium will be equally occupied by the two cations. The evidence of binding strengths of potassium-activated enzymes is that potassium is bound ten times more strongly than sodium so that sodium does not compete at potassium sites to more than the 1 % level. The use of energy to manage the level of potassium therefore controls many in-cell reactions being used as a fine control rather like magnesium.D. Calcium and Vesicular Membranes.-Calcium has a gross action on the vesicles which store hormones transmitters digestive and other proteins and even calcium.6 The effect of the calcium (Figure 9) is to cause the contents of the vesicle to be ejected often from the cell by breaking the vesicle membrane. Such membranes are composed of phospholipid and long-chain alkyl carboxylates which are typical chemicals of emulsion-forming soaps. The effect of cations on such emulsions is well known.60 Univalent ions Naf and K+ allow one particular 5 0 J. H. Schulman and E. G. Cockburn Trans. Faraday SOC. 1940,36 651 and 661. 360 Williams Na+ Ca*+ Figure 9 A schematic drawing of a synapse showing the movement of the diferent cations with respect to their concentration gradients. In different systems magnesium could move with calcium or against it.The small circles are the vesicles which contain acetylcholine Ach and a message runs from left to right structure and stabilise an oil in water emulsion probably because they do not bind to the anions. Another structure is generated by small highly-charged cations which stabilise a water in oil emulsion by binding to the anions and yet remaining partially hydrated. They are not able to co-ordinate large numbers of the anions through the radius ratio effect. Large cations of high charge such as calcium break the emulsion through precipitation of calcium salts of the emulsifier. Calcium binds to several such large anions after loss of its water of hydration. It is a structure-forming cation. In other words calcium induces a temporary solidification of a membrane film.These observations provide a possible explanation of many biological phenomena involving vesicles. In the resting cells vesicles are stabilised by the magnesium which is not structure forming and are not much affected by sodium and potassium which do not bind. The binding of magnesium to the strong acid anions di-ester phosphates is weak and that to the assembly of carboxylate anions in the membrane is weakened by steric restrictions-the binding constants are probably ca. or less. Injection of calcium (the triggered state) leads to stronger binding to either of these types of anion site say for the binding constant and causes the membranes of the vesicle and the outer membrane to come together and collapse into a single structure. This causes ejection of the vesicle contents.Figure 9 illustrates the several cation effects. We can now look again at the nature of the nerve message. We do not know how it is triggered initially whether it is by touch temperature change light or chemical action. However the immediately subsequent observation is that an 361 2 The Biochemistry of Sodium Potassium Magnesium and Calcium electrolytic depolarisation wave runs along the membrane. This is often pictured as a physical eventelectrostatic field changes altering the membrane so that it changes from a potassium to a sodium permeable condition. Could it not rather be that inward diffusion and binding of calcium causes a running wave of structure change along the membrane? This could arise through the interaction of the calcium with the anions on the inside of the outer membrane (Figure 10).IN OUT Figure 10 A proposed mechanism for the transmission of an electrolytic pulse along a membrane. At the extreme right and left the membrane is shown at rest and the down-pointing vertical arrows indicate the strong pumping of sodium and calcium from the cell. The thinner central region of the membrane is shown to be contracted by the inward calcium flow indicated by upward vertical arrows Permeability changes may then be associated with the structure change. Recovery is activated by the self diffusion of calcium away from the region of structure change and its subsequent rejection by the calcium pump. Such a problem as this and the many problems associated with vesicle (emulsion) stability can be ideally tackled by the probes of hydration and structure which we have already described.The problems presented in this section are of the greatest possible importance. No matter to which biological problem we turn-evolution the working of the code differentiation movement or the working of the brain the four cations Naf Kf Mg2+ Ca2+ have a role. The role can be chemically defined only by intensive studies by inorganic chemists who are familiar with the biological problems. The time when these problems could be left to inspection by gross tools has gone and it is necessary now to use the sophistication of modern spectral 3 62 Williams methods to analyse the molecular events underlying the changes in organised systems. It is hoped that this article has pointed to some of the problems and the methods which might be used in their study.8 Appendix The Ionic Model and the Radius Ratio E f f e ~ t ~ ~ g ~ ~ In the gas phase in solution or in crystals the smaller the cation the greater is the interaction between the cation and any given ion or dipole if the energy is measured from the free gaseous cation state. Thus a stability sequence following the order of the inverse of the cation ionic radius is the most obvious order of the free energy change of association. In exchange reactions ML1 + L2 -+ ML2 -1 L1 the total free energy change is the difference between two such simpler free energies. M+(gas) + L1 -+ ML1 M+(gas) + L2 - ML2 In such circumstances we can show that the order of the free energy change of the exchange reaction for a series of cations can be varied at will by changing L1 and L2.The simplest way in which this can be demonstrated is to compare hydration ML1 energies and energies of lattices or complexes ML2. Empirically and with some theoretical justification it has been found that the hydration free energy of a gas cation is given by -A Y+ + 0.85 reff - - -A AG = where r+ is the Pauling ionic radius A is a constant dependent upon the water dipole and the cation charge mainly. The second part of the equation defines the effective radius Yeff. (The implied very small water radius 0.85 A is possibly a reflection of the short bond distances and consequent polarisation of the water -see later). The interaction between another ligand or anion L2 and the cation is given by - B r+ + r- AG2 = - where B again includes a product of the charges on the ligand and those on the metal and Y- is the radius of the anion.Now there are no ligand atoms smaller than 0.85 A i.e. than water and so Y- > 0.85 but the value of B can be greater or smaller than A . (The hydration of the anion is deliberately omitted in what follows.) When we plot AG and dG2 against l/reff a straight line is obtained for the hydration energy but several possible curves for the interaction free energy of ML2 (Figure 11). Inspection of the case [Figure ll(i)] when B < A shows that the smaller reff the more the hydration energy exceeds dG, and al- though no systems ML2 are stable their stability relative to water falls as ionic size decreases. Stability (dG2 - dG,) is directly related to ion size. The second situation where B > A shows that there are three regions of any 3 63 The Biochemistry of Sodium Potassium Magnesium and Calcium Figure 11 The free energy of formation from the gas cation of the hydrate (AG,) and three different complexes with ligands of different types cr;lll lines) plotted against the reciprocal of the eflective cation radius as defined in the text.(ii) is for an anion of high charge or small size a weak acid anion (iii] is for an anion of lower charge or larger size one curve; (a) where dG > dGl and no complexes are stable which happens at very high l/refr for all curves; stability is again directly related to ion size but all complexes are unstable; (b) where dG > dGl and dG is increasingly greater than AG as the size of the cation decreases i.e. where l/reff increases (this is true for all curves near the origin of the plot); (c) where dG > dG, all complexes are stable and dG is decreasing relative to dG as the cation size decreases l/reff increases (this is true for all curves at some higher values of l/reff).Series in which (dG - dGl) passes through a maximum (smoothly) for an ion of intermediate size arise in the region between (b) and (c). Now let l/refr vary between extreme limits X and Y of Figure 11 for a given cation sequence e.g. Group IA or Group IIA independent of the anion. In this region the curve of dG will be the higher the smaller the anion and it follows that the smaller the anion the more it will give rise to the sequence (b) rather than (c). Thus we have (b) small anions (dG - dG,) for a small cation > (dG - dG,) for a but (c) large anions (dG - dG,) for a small cation .c (dGz - dGJ for a which are the two basic observations of this article.large cation large cation 364 Williams We need to enquire further into the factors affecting (dG - dGl). We assume that dGl is well understood and that r1 refers to Pauling crystal radii. The size of B depends upon the charge on the anion and leaving aside any polarisability terms we see that smaller charge will lead to lower curves for dG and sequence (iii) rather than (ii). This is true for example in the sequences of anions of the same size P04,- ROP03,- (RO),PO,-; Po43- SO4,- C104-. Secondly we need to consider (r+ + r-) which has been treated in too simple a manner in the above. We have fixed r+ as the Pauling cation crystal radius but then as r+ + r- is an equilibrium distance r- is really a variable and not the Pauling crystal anion radius.For small anions comparable in size to water the equilibrium distance is dependent only on cation-anion contact and so r- may well be close to the anion crystal radius for all cations. For larger anions the distance r+ + r- is dependent upon anion-anion contact which leads either to larger measured anion sizes for the smaller cations or to an enforced change to smaller co-ordination numbers smaller B as the cation decreases in size. Both type of effect depend only on the radius ratio r+/r- using Pauling crystal radii. Both have the result that for smaller cations dG values are reduced relative to dGl and consequently as anion size increases no matter what the size of B for the largest cation the order of (dG - dG,) is pushed toward order (c).Thus radius ratio r+/r- has two influences one of which can be described by Figure 11 but the other of which is a consequence of stereochemical limitations upon bond distances or co-ordination numbers or both. As we have observed above orders not following (b) can be generated with or without co-ordination changes though falling co-ordination number with ion size is one result of radius ratio effect and with or without hydration number changes. However increased hydration with decreased ion size is another result of radius ratio effect for water is a small ligand and with a weak acid or a strong acid donor depending upon the proper control of the steric factors. In conclusion the simplest result of all is the change from (b) to (c) on going from a unidentate small anion F- or OH- to a unidentate large anion I- or NO,-. The next easiest system to visualise arises from anion-anion contacts when (b) changes to ( c ) on going from a 1 1 equilibrium in solution to the precipitation of a salt for example. Thus (b) changes towards (c) for oxalates and phosphates on going from their solution equilibria to their solubility pro- ducts. Finally @) changes towards (c) if the ligand is multidentate edta (compare acetate) but the change can be augmented by still greater restrictions on the ligand geometry e.g. egta and cyclic ligands. The discussion applies without modification to Groups IA IIA and IIIA of the Periodic Table and we consider that it is sufficient to explain all the effects described in this article in both model and biological chemistry. 365
ISSN:0009-2681
DOI:10.1039/QR9702400331
出版商:RSC
年代:1970
数据来源: RSC
|
5. |
Centenary Lecture. (Delivered november 1969). Roads to corrins |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 366-415
A. Eschenmoser,
Preview
|
PDF (2359KB)
|
|
摘要:
CENTENARY LECTURE (Delivered November 1969) Roads to Corrins By A. Eschenmoser OF TECHNOLOGY (ETM) ZURICH LABORATORY OF ORGANIC CHEMISTRY SWISS FEDERAL INSTITUTE The beautiful and intriguing structure of vitamin B12 stands among the finest contributions of British science to the chemistry of low molecular weight natural pr0ducts.l This structure happened to have a rather fateful influence on our own research activity; it induced us a number of years ago to embark on a major endeavour in organic synthesis the synthesis of corrinoids.2 To expose and summarise results of these studies in a Chemical Society Centenary Lecture seems therefore not inappropriate. At the time the development of synthetic roads to corrins appeared as one of the prerequisites to a systematic study and exploration of corrin chemistry.Moreover there always was and there still is in the background of this work the alluring and exacting problem of a chemical total synthesis of vitamin B12 itself. In fact the permanent confrontation with the numerous complex problems of a vitamin B, synthesis did in more than one instance exert a most fruitful feed-back effect on the progress to be attained in the broader field of corrinoid and porphinoid synthesis. The deeply interwoven stories of corrin and vitamin B12 syntheses can actually be taken to illustrate a type of research mechanism by which involvement in a synthetic project towards a structurally novel and sufficiently complex natural product is bound to fertilise organic synthesis beyond the immediate structural boundaries of the specific synthetic problem.In an account of the present state of our work on corrinoid synthesis special reference must be made to the important accomplishments in this field of A. W. Johnson and his collaborators.s Furthermore it seems appropriate to mention here the truly pioneering efforts and contributions made by the Cambridge school4 to the corrin problem and by J. W. Cornforth’s6 group to D. Crowfoot-Hodgkin A. W. Johnson and A. R. Todd Chem. SOC. Special Publ. No. 3 1955 109; D. Crowfoot-Hodgkin J. Kamper J. Lindsey M. Mackay J. Pickworth J. H. Robertson C. B. Shoemaker J. G. White R. J. Prosen and K. N. Trueblood Proc. Roy. SOC. 1957 A 242 288; for a review see R. Bonnett The Chemistry of the Vitamin B, Group Chem. Revs. 1963 63,573. A. Eschenmoser Pure and Appl. Chem. 1963,7,297. A. W. Johnson Chem.in Brit. 1967,253 ; cf. also H. H. Inhoffen J. Ullrich H. A. Hoffmann and G. Klinzmann Tetrahedron Letters 1969 613. *R. Bonnett V. M. Clark A. Giddey and A. R. Todd J. Chem. Soc. 1959 2087 and subsequent papers. ‘ J. W. Cornforth reported by P. B. D. de la Mare Nature 1962 195 441 ; J. W. Cornforth Discussions on recent experiments on the chemistry of corrins The Royal Society London June 4 (1964). 366 Eschenmoser the one of vitamin BIZ synthesis. The epochal impact on chemiciil theory of R. B. Woodward’s involvement in the BI2 problem is well recorded.6 The specific synthetic achievements of the Harvard group have been the subject of a recently published lecture‘ and some aspects of this work are bound to be dealt with also in this report. Formula (1) in Figure 1 introduces the structure of the simplest known corrinoid natural product cobyric acid.* Since this compound has already I co + I CH- \ coo- Figure 1 CONH2 ’ONEI R. B. Woodward “The Conservation of Orbital Symmetry” Chem. SOC. Special Publ. No. 21 1967 217. 7 R. B. Woodward Pure and Appl. Chem. 1968,17 519. * K. Bernhauer H. Dellweg W. Friedrich G. Gross F. Wagner and P. Zeller Helv. chim. Acta. 1960 43 693; K. Bernhauer F. Wagner and P. Zeller ibid. p. 696; D. Dale D. Crowfoot-Hodgkin and K. Venkatesan ‘Crystallography and Crystal Perfection’ 1963 p. 237 Academic Press London; D. Crowfoot-Hodgkin Proc. Royal SOC. 1965 A 288,294. 367 Roads to Corrins served as the starting material for a partial synthesis of vitamin Blzs and since B,,-coenzymes can nowadays be made easily from the vitamin,lO cobyric acid represents the immediate goal of all the work aiming at a total synthesis of vitamin B12 and the B,,-coenzymes.Its structure lacks the characteristic nucleotide parts of the vitamin and the coenzymes but contains all other essential elements of the vitamin’s corrinoid nucleus. In particular it contains the peripheral carboxy functions in their primary amide form except the one of the propionic acid chain attached to ring D; this free carboxy group has been crucial in the partial synthesis of the vitamin and its synthetic differentiation from all other carboxy functions presents a particular hurdle in a total synthesis. At the heart of any project for a synthesis of vitamin BI2 must be a concept for the construction of the central corrin chromophore.In keeping with this- and with the historical development-the first section of this account deals with the earlier work on this problem. 1 ‘The Old Road’ Synthesis of Corrh Complexes via A/B-Iminoester Cyclhtionsll The corrin ligand system (2) (see Figure 2) is at the same oxidation level as the Figure 2 W. Friedrich G. Gross K. Bernhauer and P. Zeller Helv. chim. Acta 1960,43 704. l o E. L. Smith L. Merwyn A. W. Johnson and N. Shaw Nature 1962,194 1175; K. Bern- hauer 0. Muller and 0. Miiller Biochem. Z. 1962 336 102; 0. Miiller and G. Muller ibid. p. 229; A. W. Johnson L. Merwyn N. Shaw and E. L. Smith J. Chem. SOC. 1963 4146. 11 (a) E. Bertele H. Boos J. D. Dunitz F. Elsinger A. Eschenmoser I. Felner H. P. Gribi H. Gschwend E. F. Meyer M. Pesaro and R. Scheffold Angew Chem. 1964 76 393; Angew.Chem. Internat. Edn. 1964 3 490; (b) A. Eschenmoser R. Scheffold E. Bertele M. Pesaro and H. Gschwend Proc. Royal SOC. 1965 A 288,306; (c) M. Pesaro I. Felner and A. Eschenmoser Chimia 1965 19 566; ( d ) I. Felner A. Fischli A. Wick M. Pesaro D. Bormann E. L. Winnacker and A. Eschenmoer Angew. Chem. 1967 79 863; Angew. Chem. Internat. Edn. 1967 6 864. 368 Eschenmoser Figure 2-continued assembly of structures (3); in other words the latter can be considered to represent the products of a formal hydrolytic ‘retrosynthesis’ of the corrin chromophore. As simple and transparent as this formalism may appear it had in fact served as an heuristic basis on which the strategy of the ‘old‘ corrin synthesis was conceptually erected. From this way of looking at things the 369 Roads to Corrins problem of constructing the double-bond system of the corrin chromophore presented itself in principle as a reversal of formal hydrolytic processes that is to say as a series of stepwise carbon-carbon condensations between imide or lactam carbonyl groups and suitably activated carbon bridge components.The saturated structural district of rings A and D with their direct ring junction could of course not be covered by this type of formalism and as a consequence bicyclic dilactam derivatives of type (3a) were the obvious choice as starting materials containing these two rings. It was only recently that the formalistic concept of ‘equality of oxidation level’ could be boldly extended to include also the A/D-ring junction [see assembly of structures (4)] in order to serve once more as a retrosynthetic stimulant for the concept of still another (the ‘new’) type of corrin synthesis (see Section 4).The concept of involving lactam and imide carbonyl groups in successive carbon-carbon condensations clearly required a method that would allow a smooth and dependable activation of the electrophilic reactivity of these systems. H N \ / N OR I I -C -H Figure 3 370 Eschenmoser Alkylation at carbonyl oxygen with Meerwein’s trialkyloxonium ions12 proved to be such a method. Not in all but in a great many instances iminoesters served satisfactorily as intermediates in the route which had to be developed for constructing derivatives of the vinylogous amidine system which is [see formula (2)] the characteristic structural element of the corrin chromophore.Figure 3 gives an abstract of this condensation principle. Figure 4 summarises the typifying final steps in the ‘old’ synthesis of corrin Et Et 0 R 0 I CN Lt M = Ni+/Pd+/Co(CN)2 M =# Zn+/Li/Na Et CN CN ( 7 ) CN (8) (9) Figure 4 lP H. Meerwein H. Hinz P. Hoffmann E. Kroning and E. Pfeil J. Prakt. Chem. 1937 147 17; H. Meerwein E. Battenberg H. Gold E. Pfeil and G. Wilfang J. Prukt. Chem. 1939 154 83. 371 Roads to Corrins complexes. The approach demands the preparation of two components (5) and (6) containing rings A/D and rings B/C respectively; the various ways which were developed for their preparation have been discussed e1sewhere.l' The two components are combined in a structurally specific way by two consecutive iminoester-enamine condensations. The first of them is induced by sodium ethoxide by NH-deprotonation of the enamine system of the AID-component (5) leading to a specific intermolecular attack of the iminoester carbon of the B/c-component (6).Self-condensation of the A/D-component ( 5 ) is neither expected nor observed to occur since the conjugated iminoester system of the B/c-component is for obvious reasons more reactive towards C-nucleophiles than the isolated iminoester system in ring A. The labile precorrinoid ligand system is complexed with a transition metal ion [e.g. nickel(n) palladium(r1) or cobalt(~~-tm)] in order to stabilise the system and at the same time to arrange the four nitrogens of the ligand in a common plane bringing thereby the two remaining condensation centres of rings A and B into close proximity.The crystalline diamagnetic metal complexes (7) do in fact cyclise smoothly under the influence of a strong base e.g. potassium t-butoxide. The function of the base is to increase the nucleophilic reactivity of the exocyclic enaminoid methylidene carbon at ring B by deprotonation of the peripheral methylene group either at ring B [see intermediate (9)] or ring c. Two corrin complexes prepared in this way namely nickel(11)-7,7,12,12,19- pentamethyl-1 5-cyano-corrin chloride and dicyanocobalt(m)-l,2,2,7,7,12,12- heptamethyl-15-cyano-corrin [(8) R = CH3] have been subjected to X-ray crystallographic structure analysis in the laboratories of Professor J. D. Duni tz13 and Professor P. Galen-Lenhert14 respectively so that full knowledge of the detailed structural properties of the ligand system in these complexes is available today.Thanks to a twofold contribution of Professor Dunitz's g r ~ ~ p ~ we are even in the rare possession of detailed insight into the structural reactant- product relationship in one of the precorrin-+corrin transformations. Figure 5 presents side views of the structure of nickel(@ complexes in the above- mentioned pentamethyl series before [(10)l6 ] and after [(l 1)13] cyclisation.ls In the diamagnetic precorrinoid complex (10) the four nitrogen atoms and the nickel ion do in fact lie approximately in a common plane and the two cyclisation centres in rings A and B are almost optimally juxtaposed for cyclisation being separated by a distance of about 3.4 A which amounts to about twice the value of the van der Waals radius of a trigonal carbon.What are the limitations of this type of corrin synthesis? First of all the method seems to be strictly confined to the preparation of corrin complexes of those transition metal ions which form robust complexes l3 J. D. Dunitz and E. F. Meyer jun. Proc. Royal SOC. 1965 A 288 324. l4 P. Galen-Lenhert and T. J. Shaffner (unpublished); cf. T. J. Shaffner Thesis Vanderbilt 1969. l5 M. Dobler and J. D. Dunitz Acta Cryst. 1966 21 A110. l6 I thank Professor Dunitz ETH for kindly permitting the reproduction of Figure 5 before publication. 3 72 Eschenmoser with the corresponding precorrinoid ligands. Experiments aiming at a cyclisation of the labile precorrinoid complexes of sodium lithium or zinc have failed. This limitation precludes in particular the preparation of metal-free corrins since neither under acidolytic nor reductive conditions (cyanidation included) was it possible to remove metal ions such as nickel palladium or cobalt from their very stable corrin complexes without concomitant destruction of the ligand.The great stability of these complexes is by the way not due to an 373 Roads to Corrins inherently strong electronic metal-chromophore interaction but is clearly a kinetic consequence of macrocyclic chelation (corresponding A/D-seco-corrinoid nickel complexes lose the nickel ion to cyanide ion with great ease; see later Figure 24). A further limitation of the method concerns the influence of the substitution pattern in the precorrinoid ligand on the ease of cyclisation of dicyano cobalt(m)- 374 Eschenmoser complexes (12)-+(13) (see Figure 6).Whereas in the least substituted series R=R1 =H (and likewise in the case of the corresponding 19-methyl derivativellb) the cyclisation proceeds smoothly at room temperature in over 90% yield a n M U v 375 Roads to Corrins vitamin B1,-like substitution by methyl groups in ring A [(12) R=CH3 R1=H] necessitates a somewhat higher reaction temperature and tends to have an adverse effect on the cyclisation yield. Not surprisingly an additional methyl group at the critical methylidene reaction centre [(12) R=CH3 R1=CH3]17a hampers the cyclisation step to such an extent that the outcome may rather be called a 'mode of formation' of 5-methyl-corrin complexes [(13) R1=CH,] than a synthesis of them. Ethyl di-isopropylamine chosen as an example of a less aggressive base hardly induces cyclisation of precorrinoid complexes of type (1 2) (R = CH3 R1 = H X=OC,H,) at all even at high reaction temperatures.However with a view to the problem of vitamin B12 synthesis it was of interest! to learn that the cor- responding thioiminomethylester derivative [(12) R=CH3 R1=H X=SCH,] does cyclise at 150 "C in the presence of this base.17* Irrespective of such limitations this 'old' approach to corrins served its original purpose of bringing corrin complexes within the reach of organic syn- thesis thereby making possible the initiation of studies contributing to an understanding of the chemical and physico-chemical properties inherent in this biologically important type of structure. Such studies have been done in our18 and otherla laboratories and some are still under way.Their results will not be discussed here with the exception of one which has an immediate bearing on a synthetic problem. It concerns the relative reactivity of the meso position of the corrin chromophore towards electrophiiic substitutions.18 Figure 7 shows a reaction sequence by which methyl groups can be introduced with some degree of selectivity into the meso positions 5 and 15 of the corrin chromophore.18b Chloromethylphenylsulphide in the presence of silver tetra- fluoroborate reacts at room temperature with the cobalt complex (14) to give the mono-substitution product (1 5). A subsequent treatment with Raney nickel produces the 15-methyl derivative (1 6). The observed specificity in this methyl- ation is believed to be predominantly the result of steric control; however a repetition of the process at a somewhat elevated reaction temperature (ca.50 "C) starting from (16) produces the 5,15-dimethylated complex (17) and the isomeric 10,15-dimethyl derivative in a ratio of about 5:l. It seems therefore-and this is corroborated by deuteriation as well as cyanidation experimentslsa~ * and coincides by the way with calculated HMO-localisation energies18C-that the two 'natural' meso positions at the corrin chromophore are the somewhat more reactive ones in electrophilic substitution reactions. This fact merits attention with respect to vitamin B12 synthesis (and biosynthesis?) in the light of what has been said above about the A/B cyclisation to 5-methyl-corrin complexes. l7 (a) D. Miljkovic and M. Roth; M. Roth Thesis ETH (to be published); (b) K.J. Schossig and M. Roth; M. Roth Thesis ETH. (a) D. Bormann A. Fischli R. Keese and A. Eschenmoser Angew. Chem. 1967,79 867; Angew. Chem. Internat. Edn. 1967 6 868. (b) E. L. Winnacker Thesis ETH 1968; (c) R. Keese Tetrahedron Letters 1969 149. H. Kuhn K. H. Drexhage and H. Martin Proc. Royal SOC. 1965 A 288 384; P. Day Theor. chim. Acta 1967 7 328; J. Seibl Org. Mass. Spectr. 1968 1 216; R. Briat and C. Djerassi Nature 1968 217 918 Bull. SOC. chim. France 1969 135. 376 Eschenmoser 2 The Methods of Sulphide Contraction via Oxidative and Alkylative Coupling A fresh and lasting impetus to the field of corrin synthesis came from the work towards a synthesis of vitamin Blz. The relevant part of this development together with some of its consequences will now be discussed.The translation of the concept of corrin synthesis A B A B A -B I f 1 - I I -+ I I D C D-C D-C into a synthesis of vitamin B12 itself requires the construction of two bicyclic optically active components of known chirality such as (18) and (19) (see Figure 377 Roads to Corrirts 8). In view of the complementary directions of the work that had at the time independently been started towards these goals in the Harvard and ETH laboratories it was agreed with Professor R. B. Woodward to economise and channel the respective efforts towards a Harvard (A/D)- and an ETH (B/c)- MARVARD H3 COOC 0 ETH / Figure 8 COOCHS component and to merge eventually in the final problem of bringing the two components together. The present discussion remains confined to that crucial situation which built itself up in the central phase of the synthesis of the B/C- component (19) that is in the coupling of the precursors of rings B and c (see Figure 9).20 s o A discussion of the synthesis of the ring B precursor (20) and of its conversion to the ring c precursor (21) is given in A.Eschenmoser Accademia Nazionale dei Lincei Conferenze X. Corso Estivo di Chimica Roma 1968,183; cf. theses ETH of J. Wild 1964 U. Locher 1964 A. Wick 1964 R. Wiederkehr 1968 and P. Dubs 1969; exploratory work by J. Muchowski J. Sims D. Coffen and T. Bogard; see also ref. 7. The ring c material (21) used for the coupling experiments B+C+B-c has been prepared in the Harvard Laboratories’ from (+)-camphor by a synthesis paralleling earlier investiga- tions of J. W. Cornforth et aL6 378 Eschenmoser 0 0 0 COOCH (21) Figure 9 The problem of combining the ring B precursor in its lactone-lactam form (20) with the precursor of ring c in its enamide form (21) appears-remember the discussion on corrin retrosynthesis in Section l-simply to be one of an inter- molecular elimination of the elements of water to form the tricyclic structure (22); it implies however that central problem in corrin synthesis namely the con- struction of a ring-bridging vinylogous amidine system.Whereas the principle of iminoester-enamine condensation (see Figure 3) had served analogous purposes in a number of instancesl1s2l in the synthesis of simpler corrins it failed completely-and in a multitude of attempted structural versions and reaction conditions-to bring about a coupling of the ring precursors (20) and (21) in the desired direction.Combined experience out of these experiments and of investigations on related systems point at two main aspects of this failure. Firstly the niethylidene carbon of enamides of type (21) although quite reactive towards strong electrophiles,22 appears not sufficiently nucleophilic for reacting with non-activated iminoesters in neutral or basic medium (attempts to induce condensation through acid catalysis were hampered by the instability of such enamides towards acids2pz2) and secondly iminoester derivatives of the bicyclic lactone-lactam (20) turned out to be much less prone to undergo condensations with carbanionoid partners than structurally analogous but less substituted iminoesters of the a-pyrrolidone family. While the first-mentioned property of enamides is clearly related to electrophilic deactivation of enamine reactivity *l W.Hausermann Thesis ETH 1966. 22 (a) R. Scheffold Thesis ETH 1963; (b) W. Huber Thesis ETH 1969. 379 Roads to Corrins by the acyl group the latter reveals the iminoester condensation as a ‘soft’ process with a high susceptibility to steric hindrance. Whenever in the synthesis of complex organic molecules one is confronted with a situation where the success of an intermolecular synthetic process is thwarted by any type of kinetically controlled lack of reactivity one should look out for opportunities of altering the structural stage in such a way that the critical synthetic step can proceed intramolecularly rather than intermolecularly. Adherence to this pragmatic principle in the situation described above and setting priority for the development of a general instead of-in the structural sense-a ‘local’ solution of the problem led to what turned out to be a conceptual and eventually also a preparative breakthrough.The basic concept is abstracted in Figure 10. / “-( H\NqcH2 I 0 / P - 0 4 (24) Figure 10 The lactam group of the one condensation partner is first converted to the corresponding thiolactam system. Its nucleophilic sulphur atom sterically unhindered by being removed from any bulky substituents in the carbonyl environment by the long carbon-sulphur bond is then linked to the methylidene carbon of the enamide partner to form a sulphur-bridged intermediate of type (23). Formally such a system now fulfils the structural requirements for an intramolecular version of a (thio)iminoester-enamide condensation.The process [see ‘arrowism’ depicted in formula (23)] would result in an episulphide deriva- 380 Eschenmoser tive of type (24) which of course could not be expected to appear as an actual product; it could however offer the chance of acting as a labile reversibly formed intermediate in a reaction sequence leading either to a vinylogous N-acyl-amidine system (25) by loss of sulphur or to a corresponding mercapto derivative by rearrangement. It is well known that the sulphur of episulphide systems departs quite often with great ease especially to thiophiles like phos- phines or phosphites leaving behind a carbon-carbon double bond.z3 The overall result envisaged in this concept is of course reminiscent of a mechanistically loosely defined group of processes known in the chemical literature as 'sulphur extrusion reaction^'.^^ Extensive experimentation was required to clear the way from the concept to a preparatively acceptable synthetic process.z5 The most satisfactory coupling method turned out to be the oxidation of the thiolactam partner (20a) (Figure 11) with one equivalent of benzoylperoxidez6 in the presence of the enamide (21) and a trace of HCI.The identified intermediate in this process is the readily formed bis-imidoyldisulphide (26) which in turn induces an acid-catalysed electrophilic substitution at the methylidene carbon of the enamide; the liberated equivalent of thiolactam (20a) is recycled by further oxidation. The coupling product is subjected without isolationz7 to a thermal treatment with triethyl- phosphite to give a 1 2 mixture of the two epimeric B/c-components (22) and (22a) in 70% overall yield whereby the p-epimer (22a) is easily obtained in crystalline form.The configuration at the (CH)-position of ring B in vinylogous amidine derivatives of type (20)/(20a) has been found to be extremely labile traces of HCl equilibrate the epimers (22)/(22a) in CDCl at room temperature within a short time. The configuration of the crystalline main epimer (22a) is actually the 'unnatural' one; fortunately this fact can be considered harmless because it has recently become known that in authentic vitamin BIZ derivatives the corresponding position is likewise configurationally labile and that luckily enough the natural configuration is the more stable one. According to the original concept the structure of the final B/c-component had been envisaged to contain an enaminoid exocyclic methylidene group at ring B [see (19) in Figure 81.In practice however this structure turned out to be unstable relative to its endocyclic tautomer (27) which was formed as the exclusive product when the tricyclic lactone derivative (22a) was subjected to the ring-opening conditions of methanolic sodium methoxide in the presence of an excess of diazomethane2* (see Figure 12). On the other hand extended N. P. Neureiter and F. G. Bordwell J. Amer. Chem. SOC. 1959,81,578; D. D. Denney and M. J. Boskin ibid. 1960 82,4736; M. Sander Chem. Revs. 1966 66 326. 24 J. D. Loudon in 'Chemistry of Organic Sulphur Compounds' ed. N. Kharash Pergamon Oxford 1961 vol. 1 p. 299. 25 B. Golding P.Loliger and H. Gschwend; cf. P. Loliger Thesis ETH 1968. 36 cf. F. Hodosan Bull. SOC. chim. France 1957 633. 27 The structure of the sulphur-bridged coupling product has been fully characterised in a closely related model system derived from thiolactam (20a) and the enamide (39).'6 The corresponding equilibrium ratio between the tautomers with exocyclic and endocyclic double bonds is about 1 :1 (CHCI,; co. 30") in the case of the less substituted analogue (44).'lb 38 1 Roads to Corrins n 0 382 Eschenmoser model studies on the problem of the D/c-coupling of the A/D- and B/c-components had eventually made it clear that very probably not an iminoester condensation but a coupling via sulphide contraction would have to bring about this major synthetic step. Therefore the thiolactam structure (29) came to be considered as the final form of the ETH component; the reaction scheme (22a)+(28)-.(29) in Figure 12 summarises its preparation from the corresponding lactam derivative (22a).The intermediate conversion of the free lactam to the methyl- mercury complex (28) with methylmercury i~opropoxide~~ served the purpose COOCH3 ' b C 0 O C H 0 cH3&coocH3 \ I CH3 N- CH,N,/CH3 OH C/OOCH3 (22a) CH3Hg-O< Q 0. - 0,2 CH 0'- COOCH3 (28) Figure 12 0 .;3 A H 3 (27) 1 cH3 COOCH '' R. Scheffold Helv. chim. Actu 1969 52 56. 383 Roads to Corrins of achieving a smooth and specific 0-alkylation with trimethyloxonium-tetra- fluoroborate. Vinylogous amidines are diaza-derivatives of p-dicarbonyl systems. It should therefore be possible to adapt the condensation principle of sulphide contraction to the synthesis of other potential p-dicarbonyl systems and also of p-dicarbonyl compounds themselves.This indeed proved to be the case by an alternative preparative version of the condensation principle namely by the method of sulphide contraction via alkylative (compared to oxidative) coupling (see Figure 13). Sulphide contraction via alkylative coupling Figure 13 Thioamides and salts of thioacids are known to be S-alkylated by a-bromo- carbonyl compounds with great ease to give the corresponding thioiminoesters and thioesters respectively. As indicated in Figure 13 these alkylation products in their adequately enolised form possess the complementarily arranged reactivity centres required for intramolecular 1,3-~ondensations and subsequent sulphur transfer that is for processes completely analogous to the one discussed above in connection with the formation of vinylogous amidines.The results of experi- ments carried out on simple systems do in fact amply illustrate the preparative feasibility and apparent generality of this method for the synthesis of vinylogous amides and p-dicarbonyl compo~nds.~* 20(u) P. Dubs Thesis ETH 1969; (b) M. Roth Thesis ETH (to be published) see also P. Dubs E. Gotschi M. Roth and A. Eschenmoser Chimiu 1970 24 34. 384 Eschenmoser A Synthesis of Vinylogous Amides I 0 1 H R = 0 Br (C,H,O),P /70° 94% over all CH3 g3P/0,2t.-BuOK/A benzene 69% over all O-t-C4H p3P/O. 1 t.-BuOK/A benzene 56% over all Figure 14 A p -Diketone synthesis 0 0 I (C,Hs),N / LiC104. n- - cH2T R' in benzene R[IcH2TR' 0 0 RT/ > 20h 0 0 R R' CH3-CH2-CH2 CH3 -CH2 72% over all CH,-CHz-CH 80% over all Figure 15 385 Roads to Corrins The examples chosen to illustrate the formation of vinylogous amides31 (see Figure 14) reveal some plausible reactivity differences.In contrast to the case of the aromatic ketone derivative (R =p-BrC,H,) the examples with bromo-acetone and a-bromo-t-butylacetate require the presence of a catalytic amount of base in order to induce the enolisation considered to be necessary for the contraction process. The presence of a base is also compulsory for the contraction of the thioester derivatives of Figure 15 to the corresponding p-diketones. Not surprisingly these conversions can easily be achieved in high yields e.g. by potassium t-pentoxide and triphenylphosphine (or tributyl- phosphine) in benzene at somewhat elevated temperatures; however of greater preparative interest is the discovery that tertiary organic amines cleanly induce these contractions at room temperature in benzene provided that anhydrous lithium perchlorate is present.30b This salt has been found to be the most effective catalyst among a number of metal salts tested in order to find an S- and/or 0-complexing agent that would speed up either the enolisation step or the contraction process or even both.Admittedly there is hardly any dearth of methods available for the preparation of /3-dicarbonyl compounds; however mildness of reaction conditions potential ver~atility,~~ and structural control- lability are properties which may well hold out some prospect for the method as a new tool in organic synthesis.(Note added in proof. The potentially unsafe LiC104 can be replacedby LiBr.) Things in the field of corrin synthesis started moving again after the B/C- problem in vitamin B12 synthesis had so to say given birth to the sulphide contraction method. The next two sections deal with such developments. 3 Synthesis of Metal-free Corrins by AIB-Cyclisation via Suplhide C~ntraction~~ As previously discussed an important limitation of the synthesis via A/B- iminoester cyclisation is its restriction to the synthesis of corrin complexes with those metal ions which form robust square planar or octahedral complexes with the precorrinoid ligand but which cannot so far at least be removed again after cyclisation. It was the method of A/B-cyclisation by sulphide con- traction via oxidative coupling that was found to cure the situation.At the time a major impetus to get metal-free corrin ligands accessible by synthesis had come from J. I. too hey'^^^ surprising and important discovery of cobalt-free corrinoid natural products occurring in certain photosynthetic bacteria. Peculiar chemical and spectroscopic observations on these products clearly made a study of structurally well-defined synthetic derivatives desirable. Furthermore *I This type of formation of vinylogous amide systems has been adumbrated by observations described by E. B. Knott J . Chem. Soc. 1955 916. It has been found more recently that the method is also applicable to the synthesis of enolizable j?-diketones 8-formylketones and 8-keto-esters which are alkylated in the a-position.* ob 'a A.Fischli and A. Eschenmoser Angew. Chem. 1967 79 865; Angew. Chem. Internat. Edn. 1967 6 866; cf. A. Fischli Thesis ETH 1968; H. U. Blaser Thesis ETH (to be published). a t J. I. Toohey Proc. Nat. Acad. Sci. USA 1965,54 934; Fed. Proc. 1966,25 1628. 386 Eschenmoser an availability of synthetic metal-free corrin ligands could be expected to open the door to a colourful (and theoretically interesting) palette of new corrin complexes. An A/B-cyclisation via sulphide contraction was originally planned to start from a precorrinoid ligand containing a thiolactam group in ring A and being loosely complexed with a metal ion like zinc(@ which could be removed again after cyclisation by acidolysis. However the treatment of a precorrinoid sodium salt of type (7) (M = Na; R = CH3 see Figure 4) with hydrogen sulphide in the presence of trifluoroacetic acid followed by complexation with zinc@) per- chlorate did not produce a thiolactam derivative but the cyclic isomer (30) (see Figure 16).Notwithstanding reaction conditions could be elaborated which make use of this product as a starting material for the sulphide contraction process. Reaction with benzoylperoxide in methylenechloride in the presence of trifluoroacetic acid brings about the desired oxidative coupling of the sulphur to the exocyclic methylidene carbon of ring B; the crystalline complex (31) can be isolated in as much as 72% yield after contact of the reaction mixture with methanol. Subsequent treatment with trifluoroacetic acid in dimethylformamide at elevated temperature leads to contraction and produces the corrin complexes (32) and (33) the former as the main product in spectroscopic yields up to 80 %.The formulae (30a)-(31c) depict a tentative interpretation of the rather intricate series of processes involved. We assume that in the first step trifluoro- acetic acid can establish the conversion of (30) to the thiolactam derivative (30a) [(30) does not react in the absence of acid] and that the latter is attacked by benzoylperoxide to form the 0-benzoate of the thiolactam-S-oxide (30b) which then reacts-very probably assisted by the template effect of the zinc ionss- with the enaminoid methylidene carbon of ring B to give the sulphur-bridged intermediate (3 1 a). Consideration of molecular models reveals the interesting feature that the ring B double bond of this intermediate is not expected to return easily to the exocyclic position as long as the nitrogen atoms of rings A and B remain co-ordinated to the central zinc ion; the geometrical situation created by the sulphur bridge in a complex is such that the double bond in the exocyclic position is expected to be heavily strained.Yet restoration of that double bond in the exocyclic position is clearly a prerequisite for the reaction to proceed further in the desired direction. The fact that sulphide contraction (31)-+(32) (33) does occur with trifluoroacetic acid in dimethylformamide suggests that an acid induced decomplexation of the ligand system precedes the formation of an intermediate of type (31b) in which the system can now better accommodate the exocyclic double bond by virtue of the higher flexibility of the free ligand system compared to its zinc complex.Contraction to the hypothetical episulphide intermediate (31c) and a subsequent not unexpected rearrangement leads then to the 5-mercapto-corrin ligand which is isolated as its crystalline zinc complex (32). The following experimental fact strongly 3 5 The same treatment on a decomplexed derivative of (30) produces only small amounts of corrinoid products. 387 Roads to Corrins 388 Eschenmoser supports the decomplexation hypothesis treatment of the sulphur-bridged complex (31) with trifluoroacetic acid in dimethylformamide in the presence of 2 additional equivalents of zinc ions no longer produces appreciable amounts of the corrinoid products (32) and (33). It has been checked that these complexes would in fact survive these reaction conditions if they were formed.SH CN (34) Figure 17 The 5-mercapto-corrin zinc complex (32) can be cleanly desulphurised by triphenylphosphine in the presence of trifluoroacetic acid in chloroform to give the corrin zinc complex (33) which in sharp contrast to the behaviow of the corresponding robust complexes of nickel cobalt and palladium loses the metal ion with delightful ease under the influence of trifluoroacetic acid in acetonitrile. Various corrinium salts tentativelySg formulated as (34) have been isolated in crystalline form. The tentativeness in this formulation refers to the position of the two NH-hydrogens only. The gross constitution has been chemically proved by conversion of (34) to the known dicyanocobalt(m) complex [8 R = CH,; M = Co(CN),].389 Roads to Corrins 390 Eschenmoser Figure 18 reports a very recent contribution from Oxford that is the result of an X-ray structure analysis of the 1,2,2,7,7,12,12-heptamethyl-l5-~yano- corrinium bromide (34; X = Br)37. While the analysis fully confirms the constitution and the configuration of the synthetic ligand system it reveals surprising structural details about the substance in its solid state. Rings B c and D are almost perfectly placed in a common plane whereas ring A dramatically sticks out of that plane and its nitrogen-admittedly to our astonishment- appears bonded to one of the two immonium hydrogens which is also engaged in a hydrogen bridge to a molecule of ethanol the solvent of crystallisation. The introduction of various metal ions into the synthetic corrin ligand (34) has not as yet presented major difficulties; specific reaction conditions for different metal ions had to and could be found in order to produce the respective (crystalline and diamagnetic) metal complexes (35) in good yields [e.g.over 90 % for M = Co(CN),]. Figure 19 summarises the experience hitherto available. M # Co(CN)Z N i+ Pd+ M =Zn+ Li a CH3 CN CH3 CH 33 1 CH 3 CN (34) (35) M = Co(CN) COII(C104)2 /CH3CN /RT-tO,/CN- Ni+ NiIr (CH3C00)2 /CH3CN /70 O Pd+ PdII (CH3COO)2 /CZH60H /70° Rh(CN)2 [Rh1(CO),C1] /CH,COOH(Na) /100°-+02/CN- Znf Zn1~(C10,) /CH3CN/RBN /RT Li LiOH /CHC13 /RT Figure 19 37 E. Edmond and D. Crowfoot-Hodgkin unpublished results; I thank these authors for kind permission to report these results before publication.391 Roads to Corrins The situation with respect to the reversal of these complexations has been discussed already (see above). There is one feature among the chemical properties of synthetic metal-free corrinium salts of type (34) deserving here a special comment ironically the neutral metal-free species in this series prefers to exist as a non-corrin. Mono- deprotonation of the corrinium ion (34) by tertiary amines in nonpolar solvents or by sodium hydroxide in ethanolic solution leads to an ‘enaminised’ form of the neutral ligand (~K*Mcs 8-6 in titration by 0.1 N-HCI in dimethyl-cellosolve- water 1 1). The structure (36) is based on the distinct presence of three different vinyl protons in the n.m.r. spectrum the position of the enaminoid double bond being inferred from the experimentally established factlaa that in synthetic nickel(@ and cobalt (m)complexes the methylene group in ring B is the pre- ferred site of peripheral CH-deprotonation.Figure 20 reflects the course of deprotonation by the electronic spectrum which cleanly (and reversibly) decays to a non-corrinoid spectrum which is highly reminiscent of the one observed by J. I. too he^^^ for natural metal-free corrinoids in strongly alkaline solution. 4 ‘The New Road’ Synthesis of Corrins via Photochemical A/D-cyClOiSO- rneri~ation~~ A characteristic feature of the corrin syntheses discussed so far is a final metal- template-assisted cyclisation between rings A and B. Inevitably this type of approach requires the construction of bicyclic precursors containing rings A and D and implies the solution of the major configurational problem of con- necting these two rings together stereospecifically in a trans fashion.To impose stereospecificity on such a ring-connecting process would truly be a worthy task for a metal template and this in fact is a central feature of the following alternative concept of corrin synthesis To construct fist the corrin chromophore by joining rings A B c and D together to introduce then a metal ion and to achieve finally a cyclisation between rings A and D under both the constitutional and configurational control of the metal template. Such an approach beside being reminiscent of A. W. Johnson’s3 syntheses of corrole and tetradehydro- corrin systems has been inspired and forcibly promoted by the specific structural regularities present in the natural corrinoids.These regularities reflect themselves in a remarkable network of synthetic opportunities emerging from a retro- synthetic analysis of the cobyric acid structure (see Figure 21). Take the chiral dilactone-monocarboxylic acid (37) as the starting material An elongation of the free carboxylic acid chain by one methylene unit followed by a structurally specific replacement of one of the endocyclic lactone oxygens by NH and finally conversion of the potential methyl ketone system into its enamide form leads to a compound which can serve as the precursor not only of ring B but potentially also of ring A. The precursor of ring c in its enamide form differs from the ring A/B precursor by nothing more than the 8 8 Yasuji Yamada D. Miljkovic P. Wehrli €3. Golding P. Loliger R.Keese K. Muller and A. Eschenmoser Angew. Chem. 1969,81,301; Angew. Chem. Znternat. Edn. 1969,8,343. Eschenmoser 5 0 0 VI 0 0 ct a E k 0 0 m 0 0 N 393 3 Roads to Corrhs x 8 0 0 h P m W 394 Eschenmoser carbomethoxy group of the acetic acid side chain being replaced by hydrogen. A chemical realisation of this relationship makes the ring c precursor available from the A/B-intermediate. A reaction sequence essentially analogous to the one which leads from the dilactone-monocarboxylic acid (37) to the AIB-precursor can transform the enantiomeric form of the same starting material into a potential precursor of ring D provided that not the free acetic acid side chain but rather the lactonised (CH,-CO-0)-chain is lengthened by a methylene unit. In doing that an important requirement must be taken care of namely that the propionic carboxy function of ring D must eventually be chemically differentiable from all other carboxy functions.Finally the specific ring D concept illustrated in the Figure implies a reductive replacement of the lactam carbonyl group by a methylene This strategy ignores one special feature of the cobyric acid structure namely the two extra methyl groups bound to the corrin chromophore at the meso carbons between rings A/B and C/D. Placing a corresponding methyl group in the starting material (37) seems easily possible but would destroy at once the synthetic relationship between the four ring precursors. These methyl groups are to be introduced post festum (compare Figure 7). To sum up at this point clearly as a reflection of underlying regularities in the biosynthesis of the natural corrinoids we find ourselves confronted with the striking opportunity that the two enantiomers of one single starting material appear convertible to three optically active chirally correct intermediates which could serve as precursors of all four rings in a potential synthesis of cobyric acid- provided that methods are or could be made available for putting them together.A tailor-made method for the construction of the chromophore part had in the meantime come to hand the method of sulphide contraction via oxidative coupling. The extraordinary challenge to provide a potential solution for the other and most crucial problem the final A/D-cyclisation gave the impetus to recent investigations which led to a new and 'as we now know rather broad road to corrins.Figure 22 formulates the concept at the structural level of simple enamide enamine ring precursors. As delineated earlier the assembly (4) of the four simple ring precursors is on the same oxidation level as the ligand system of a corrin complex (35). Furthermore an A/D-seco-corrinoid metal complex of type (38) is isomeric with the corrin complex ( 3 3 the two systems differing only in the position of one hydrogen atom and of one carbon-carbon bond. Models of such seco-corrinoid metal complexes display the ligand system coiling around the metal ion and having a ring D methylene hydrogen atom lying directly underneath the exocyclic methylidene double bond of ring A; this is true the more one tries to hold all four ligand nitrogen atoms in a common plane with the co-ordination centre.A jump of this hydrogen atom to the methylidene carbon would formally create a new conjugated 15-centre-1 6-electrons rr-system of admittedly higher energy 3 9 The synthesis of the ring D precursor formulated in Figure 21 from the enantiomer of (37) has actually been accomplished (R. Wiederkehr P. Dubs and W. Fuhrer). 395 Roads to Corrins CN (4) 0 0 6 CN (38) O? CN (35) Figure 22 396 Eschenmoser (it cannot be presented by a classical formula if one neglects 1,3 a-bonds) but which in turn could gain the stabilisation of a carbon-carbon a-bond by simple collapse to the corrin complex (35). Stereochemically such a collapse within a helically deformed planoid metal complex could hardly avoid leading to the trans configuration of the A/D ring junction.At an earlier time such consideration might have remained untested by being relegated into the realm of wishful formalism but not so nowadays after the advent of R. B. Woodward and R. Hoffman’s40 generalisations on orbital symmetry control of concerted organic reactions. The two above mentioned R. B. Woodward and R. Hoffmann,J. Amer. Chem. SOC. 1965,87,395,2511; R. Hoffmann and R. B. Woodward ibid. p. 2046,4388,4389; R. B. Woodward and R. Hoffmann Angew. Chem. 1969,81 797. 397 Roads to Corr€ns 5 nn CdP - 398 Eschenmoser processes formally classify as an antarafacial sigmatropic 1,16-hydrogen transfer and an antarafacial electrocyclic 1,15-rr+o-isomerisation. In a simple frontier orbital analysis the hydrogen transfer emerges as a 'thermally forbidden' process whereas the antarafacial 1,15-cycloisomerisation appears symmetry allowed in the electronic ground state (symmetries of frontier orbitals rs and 7r8 respectively at reaction centres; see Figure 22).Figures 23 and 24 illustrate the reaction sequence by which A/D-seco-corrinoid complexes of type (38) can be prepared in a straightforward way by connecting three molecules of the enamide ring precursor (39) by the method of sulphide contraction via oxidative coupling and by subsequent addition of a fourth ring by an enamine-iminoester condensation. The first enamide coupling requires protection of the strongly nucleophilic enamide double bond of the potential thioamide partner. This protection proved to be best provided by the cyanide group which is introduced with KCN in aqueous It survives the conditions of the subsequent steps and can eventually be cleanly expelled from vinylogous amidine derivatives by strong base.In contrast to the enamide (39) itself the corresponding cyanolactam can easily be converted to the thiolactam (40) by reaction with PzS5. In a sequence of events completely analogous to those of the B/c-coupling in the &,-series (see Section 2 Figures 10 and ll) oxidation of the cyanothiolactam (40) with benzoylperoxide in the presence of the enamide (39) leads to the bicyclic thio- bridged intermediate (42) in high yield. Heating in triphenylphosphine brings about sulphide contraction to the vinylogous amidine derivative (43) and subsequent treatment with potassium t-butoxide eliminates the protecting group to form the bicyclic lactam (44).This compound had already served as a central intermediate in the earlier corrin approachll (see Figure 4) but had to be prepared at the time by less straightforward methods. Repetition of the ring-connecting process adds another vinylogous amidine unit to the bicyclic intermediate (44) producing the bicyclic analogue (45). In this more complex case a preparative version of the oxidative enamide- thiolactam coupling procedure has to be used involving base induced NH- deprotonation of the enamide partner (44) followed by reaction with the isolated disulphide intermediate (41). A further deviation refers to the subsequent contraction step which responds strongly to catalysis by boron trifluoride. The incorporation of the fourth ring by an enamine-iminoester condensation requires the transformation of the lactam group of the tricyclic intermediate (45) into its iminoester.In contrast to the experience with the tricyclic analogue (44),11 the direct O-alkylation of (45) with triethyloxonium-tetrafluoroborate proved preparatively useless because rather indiscriminate O- and N-alkylations occur. This difficulty is effectively overcome in the corresponding silver complex (46) (see Figure 24) in which the (presumably digonal) co-ordination with the OIThe addition of CN- is expected to proceed onIy after tautomerisation to the N-acyl- ketimine isomer [compare with the addition of nitromethane to (39)"dI. The method of masking an enamide double bond by cyanide was first used in another connection by R. B. Woodward and A. Wick (unpublished). 399 Roads to Corrins silver ion protects the sp2-electron pairs of at least two nitrogen atoms against N-alkylation.Reaction of this silver complex with triethyloxonium-tetrafluoro- borate directly followed by condensation with the ring D component in its free enamine form produces the desired tetracyclic system which is isolated in high yield as the beautifully crystalline NiII-complex (47 M = Ni+). Fortunately cyanide ions remove the nickel(@ with great ease thus allowing the preparation of complexes with other metal ions e.g. PdII PtII or CoIII. The cyanide protect- ing group in complexes (47) is expelled under the influence of potassium t-but- oxide to yield AID-seco-corrinoid systems of type [38 M = Nif Pdf Pt+ Co(CN),]. The very labile complexes with ZnII and MgII can be prepared by carefully controlled cyanide induced metal exchange on the corresponding nickel@) complexes (38 M = Ni+).42 All these seco-corrinoid complexes have been obtained in pure and crystalline form and the assignment of their structures is fully supported by analytical and spectral data.A somewhat problematic situation has been encountered in the cobalt series. The removal of the cyanide protecting group with t-butoxide from the dicyanocobalt(m) complex [47 M = Co(CN),] resulted in the isolation of not one but two different crystalline dicyanocobalt(1rr) complexes in about equal amounts both having the molecular weight corresponding to structural formula [38 M = Co(CN),] and both showing n.m.r. signals compatible with this constitution. On the basis of differences in their U.V. and i.r.spectra diastereo- meric structures of type (48) and (48a) (see Figure 25) are tentatively assigned to these two compounds. The one believed to represent the trans-dicyano complex (48) appears thermodynamically less stable; it can be quite cleanly converted into the isomeric complex e.g. by heating in t-butanol. A summary of the expanding and continuingly exciting experience of the photochemical cycloisomerisation of AID-seco-corrinoid metal complexes to corresponding corrin complexes is given in Figure 26. The reaction revealed itself as an all-or-none process its success most remarkably depending on the nature of the central metal ion. Whereas the seco-corrinoid complexes of palladium(II) platinium(II) zinc@) and magnesium(I1) [38 M = Pdf Pt+ Zn(C1) and Mg(CI)] in degassed solutions cyclise at ambient temperature in essentially quantitative yield to the corresponding trans-corrin complexes (35) on irradiation with light in the range of ca.300-530 nm38s4* (Pyrex-filtered U.V. light sunlight or just artificial visible light43) no cyclisation whatsoever has been detected under similar and other conditions in the cases of the nickel(rr) and both dicyanocobalt(Ir1) complexes [38 M = Nif and Co(CN), compare (48) and (48a)l. The successful cycloisomerisations are by far the cleanest and most delightful steps we have ever encountered in synthetic corrin chemistry. On the other hand preliminary experiments have given no u.v.-spectroscopic in- 4 2 Work by H. Wild (Pt Zn Mg) and L. Ellis (Co). 43 The electronic absorption spectra of these seco-corrinoid metal complexes have intense absorption bands in the region 300-350 nm and 450-530 nm; e.g.M = Zn(C1) Amsx = 268 nm (log E = 4-41) 293 (4*18) 328 (4~58)~ 377 (3.85) 405 (3~66)~ 51 1 (4.15) nm (in C2H,0H). k Eschenmoser v) (Y 40 1 Roads to Corrins h v 3 1 5 0 0 rcl A 2 I n t ? E $. A 0 m 5 W e N h I V 2 W M 402 Eschenmoser dication of the occurrence of a thermally induced cycloisomerisation of either the palladium(n) or the nickel(@ and dicyanocobalt(m) complexes (38). Both constitution and configuration of the photochemical cycloisomerisation products [35 M = Pdf Pt+ Zn(CI) and Mg(CI)] are beyond any doubt u.v. ix. n.m.r. and mass spectra of the Pd- and Zn-complexes are identical with the spectra of the corresponding 1,2,2,7,7,12,12-heptamethyl-15-cyano-trans- corrin complexes prepared by the classical routes described in Sections 1 and 3 (see Figures 4 and 19) whereas the structure of the magnesium complex has been proved by conversion into the metal-free derivative.The structural assignments in this synthetic series rest solidly on two X-ray analyses contributed by P. Galen- Lenhert and T. J. Shaffner [35 M = Co(CN),]14 and E. Edmond and D. Crow- foot-Hodgkin (34) (see Figure 18).37 The plain and in a way still surprising fact that certain A/D-seco-corrinoid complexes are found to cycloisomerise so smoothly to corrin complexes does in a sense lend substance to the reaction formalism envisaged in the planning stage of the work; but much more importantly it confronts us now with a number of incisive and provocative mechanistic questions. An important lead to these problems is the apparent yes-or-no dependence of the cycloisomerisation on the nature of the metal ion.It had originally been suspected that the first observed remarkable difference in the behaviour of the two (diamagnetic) square-planoid seco-corrinoid com- plexes of nickel(@ and palfadium(n) might be simply caused by corresponding differences in their respective molecular geometry re~nembering~~ that the palladium ion can be expected to enforce square-planarity of a ligand system more tightly than the more tolerant nickel ion.45 However the smooth cyclo- isomerisation of the supposedly stereochemically non-ideal chloro-zinc(rr) and chloro-magnesium(@ complexes provide support for the idea that factors others than purely geometrical ones play a (or the) decisive role in the process.The lifetime of the reacting photoexcited state (or states?) of the chromophore might depend on the specific electronic structure of the metal ion in such a way that this dependence is critical for the occurrence or non-occurrence of the cyclisation process. It has been found in G. Q~inkert’s~~ laboratory that the quantum yield of the cyclisation of the palladium complex at room temperature is very low and that it vanishes at low temperature. Another important piece of information comes from A. J. Thompson’sQ7 recent work on the fluorescence of 44 See for example the synthesis of the corphin system A. P. Johnson P. Wehrli R. Fletcher and A. Eschenmoser Angew. Chem. 1968 80 622; Angew. Chem. Internat. Edn. 1968 7 623. ‘’ Since the time that this lecture was given the results of X-ray structure determinations of the seco-corrinoid complexes of nickel(I1) and palladium(@ (38 M - Ni+ and Pd+ as perchlorates) have become available.The crystals of the two complexes are isomorphous and the square-planoid tetrahedrally-distorted conformation of the ligand system in borh com- plexes appears ideally suited for the hydrogen transferprocess. The critical (> CH . . . .CH,=!- distances are 3-46A (nickel) and 3.33 A (palladium); private communication by Dr. M. Cunie and Professor J. D. Dunitz ETH. 46 Private communication from Professor G. Quinkert and G. Prescher. 4 7 A. 1. Thompson J. Amer. Chem. Soc. 1969,91,2780. 403 Roads to Corrins naturally occurring corrinoids according to which cobalt(rr1)-corrinoids do not fluoresce whereas corresponding metal-free derivatives do the reason for this striking difference being that the open-shell transition metal ion would quench the fluorescing excited state of the corrin chrom~phore.~~ At present these facts together with the behaviour of the seco-corrinoid closed-shell metal ion complexes of zinc@) and magnesium(:r) hint at the possibility that occurrence of the photocycloisomerisation might be the result of its successful competition with intersystem crossing and/or internal conversion processes mediated by the metal ion in the lowest singlet excited state of the seco-corrinoid chromophore.Emission studies on a whole series of synthetic corrin complexes currently under way in Dr. A. J. Thompson’s laboratory extended quantum yield measure- ments and spectroscopic studies by Professor G.Quinkert’s group X-ray structure determinations by Professor J. D. Dunitz and his collaborators and finally systematic chemical investigations in our laboratory are expected to provide the basis for an understanding of what we suspect to be a pregnant problem in the field of porphinoid metal complex photochemistry. It has been mentioned earlier (see Figure 19) that zinc(~:)-corrins can serve as starting materials for a whole series of other corrin complexes including those of cobalt. Therefore the photochemical cycloisomerisation of AID-seco-corrinoid zinc(@ complexes quite apart from its inherent mechanistic interest promises to be a general synthetic approach to corrinoids containing a vitamin B,,-like substitution pattern in ring A. Quite specifically the earlier discussion referring to cobyric acid (see Figure 21) now appears to describe a realistic synthetic opportunity.In the laboratory however another problem has recently been more acute namely to join the A/D with the B/c-component in the Harvard-ETH approach to cobyric acid. 5 Recent49 Steps in the Harvard-ETH Approach to Cobyric Acid It was during 1967 that both the Harvard and ETH groups had finally reached their respective goals by having accomplished syntheses of compounds which seemed structurally apt to serve as A/D- and B/c-components in a construction of the cobyric acid molecule. It soon became clear in both laboratories that earIier expectations were too optimistic in assuming that this condensation problem could be solved by merely following the conceptual and experimental paths paved in the earlier work on the synthesis of corrins.For instance one of the candidates originally envisaged to serve as the A/D-component for an A/D-C/B coupling was the tricyclic enamino-ketone (49) (see Figure 27). This assignment soon proved illusory because the compound surprisingly emerged from the work in the Harvard laboratories as an extremely labile one 48 According to preliminary experiments of Dr. A. J. Thompson the synthetic chloro-zinc@) complex (33) also shows fluorescence (private communication); compare also the luminescence properties of porphyrin complexes R. S. Becker Theory and Interpretation of Fluorescence and Phosphorescence Wiley-Interscience N.Y. 1969 p. 190. 49 Refers to the time of delivering the lecture November 1969; for later developments see footnote 58.For previous progress reports see R. B. Woodward;’ A. Eschenmoser.20 Eschenmoser m 405 Roads to Corrins the reason for its lability being a ring closure between the enamine NH-group and one of the methoxy-carbonyl groups at ring A (see arrow) that most disturb- ingly takes place under almost any conditions. This cyclisation suppresses the system's enaminoid character required for all types of coupling processes under consideration. A derivative stable towards this sort of cyclisation was found in the enolether (50) but unfortunately its nucleophilic reactivity at the vinylic carbon appeared insufficient as judged from the completed failures of various coupling attempts. There is no need to describe here any of the other numerous unsuccessful experiments; it is equally informative to enjoy a glimpse of the cartoon60 of Figure 28 which describes the situation as it prevailed in the two laboratories for quite a while.Fortunately the cartoon grossly exaggerates in one important respect things started moving again after the Harvard group had tried and succeeded to ozonolyse with high selectivity the carbon-carbon double bond of the enolether system in their compound (50) (see Figure 29) to form the corresponding formyl-ketimine derivative which in turn could be reduced to the hydroxy compound and converted via the mesylate to the crystalline bromomethyl derivative (51). This type of structure represents an ideal solution to the problem of finding a suitable form of the A/D-component for the coupling with the B/c-component (29) the condensation method to be used being the sulphide contraction via alkylative coupling (see Section 2; Figure 13 and 14).The potassium salt of the B/c-component (29) is alkylated specifically and smoothly at sulphur by the A/D-bromide (51) to form the labile pentacyclic thioiminoester derivative (52) (see Figure 30). The system boron trifluoride- triphenylphosphine-methylmercury-isopropoxide in benzene51 converts this labile condensation product at about 70 "C within less than one hour into a desulphurised material in estimated overall yields of up to 50-60%. Although this material can be purified by chromatography it has as yet never been obtained in crystalline form (it presumably contains components with epimerised centres at rings B and c); however the assignment of structure (53) to it is well documented spectroscopically and fully supported by further transformations.Formulae (52a) and (52b) describe tentatively assigned structures of two isomeric condensation products which are formed with great ease from the thioiminoester (52) by chromatography on silicagel or in contact with traces of acids. The novel structures and reactions in this ADcB-series turned out to be full of intricacies and a major research effort invested mainly by the Harvard group was needed for defining experimental conditions under which these systems behave inter- pretably and reproducibly and above all under which the condensation product isomers (52) and (52a b) contract reliably in synthetically acceptable yields.62 Cartoon by Sattler published in 'Nebelspalter' Nebelspalter-Verlag Rorschach (Schweiz) and spotted by Dr.L. Werthemann ETH. '' It is possible that methylmercury-isopropoxide prevents the formation of the isomer (52b) which is known not to contract under these conditions. 52 A very recent and important result of these studies by the Harvard group is a reliable high yield procedure for the acid-catalysed contraction of the thermodynamically most stable condensation product isomer (52b). 406 Eschenmoser 407 Roads to Corrins A 8 0 X A u 0 0 5( A z- z 8 0 z 91 N i 408 Eschenmoser 0 cr) 409 Roads to Corrins It did not come as too much of a surprise when the next important step the A/B-cyclisation to the corrinoid system revealed itself again as a major synthetic obstacle. Initial exploratory experiments on an introduction of cobalt into a ring A iminoester derivative of the ADcB-condensation product (53) to be followed by base catalysed or thermal iminoester cyclisation-a sequence of processes well under control in the case of simpler precorrinoid systems-showed little promise of success.Therefore a major attack was launched towards the goal of achieving this cyclisation by making use of for the third time in the whole project the principle of sulphide contraction. In fact it was in such experiments that the corrinoid chromophore finally appeared on the scene. The present state of these still incomplete cyclisation studies is summarised in Figure 31.63 53 Work by P. Schneider F. Karrer D. Becker and W. Huber. 410 Eschenrnoser An A/B-cyclisation via sulphide contraction requires first of all the conversion of the ring A lactam group into the corresponding thiolactam system.Phosphorus pentasulphide under carefully controlled conditions brings about this trans- formation in high yield but not without concomitant replacement of the ring B lactone oxygen by sulphur to give the ADcB-thiolactam-thiolactone derivative (54). A sequence of operations so far carried out without characterisation of the intermediates but performed very closely to thereaction conditions extensively studied earlier in the synthesis of metal-free corrins (see Section 3; Figure 16) leads to a mixture of corrinoid cyclisation products clearly recognised as such by the very characteristic electronic spectrum. This series of operations includes (a) the methoxide-induced eliminative opening of the thiolactone system in the presence of diazomethane in methanol (see the analogous process in Figure 12) (6) complexation in the same solvent with zinc@) perchlorate (c) oxidative coupling with benzoylperoxide in dichloromethane in the presence of trifluoro- acetic acid (d) acid-catalysed contraction in dimethylformamide and finally (e) recomplexation with zinc@) chloride in methanolic solution.So far only one of the corrinoid components produced in this process has been isolated in chromatographically homogeneous form ; we believe that the structure of this component is that of the corrinoid zinc@) complex (55). This assignment rests for the time being on the following three pieces of evidence First the highly characteristic electronic spectrum (reproduced in full line in Figure 31) cor- responds convincingly to what one has to expect for the spectrum of a zinc(@ corrin complex;K* second the highest peak of significant intensity in the mass spectrum appears at the mass number 1013 which corresponds to the molecular weight of the complex-ion (55) and third although the exploratory experiments on the acidolytic decomplexation and subsequent introduction of cobalt with cobalt(I1) perchlorate cannot be said to have been successful already in a pre- parative sense such experiments have resulted in the isolation of a chromato- graphically homogeneous but still not quite pure66 material which clearly is a dicyanocobalt(m)-corrin complex the structure and position of all the absorption bands in its electronic spectrum are highly characteristic and virtually identical to those observed in the spectrum of authentic dicyano-5,15-bis-desmethyl- cobyrinic acid heptamethylester (56) a crystalline compound available by degradation of vitamin Neither the cyclisation nor the metal exchange procedure in the state of development indicated above represent what we can call a solution to the synthetic problem involved yields are too low and products and intermediates have not been crystallised and not fully characterised so far; however these '' This expectation relies on the known spectrum of the synthetic zinc(1r) corrin complex (33) (see Figure 17).55 The presence of impurities is indicated by the intensity of the absorption bands in the U.V. relative to those in the visible region. 56 Prepareds7 by acid-catalysed methanolysis of material obtained from vitamin B, by oxidation with KMnO (unpublished work of K.Bernhauer and F. Wagner). I thank Pro- fessor Bernhauer and Dr. Wagner Technische Hochschule Stuttgart for generously supplying us with this material. '' Work by L. Werthemann and H. Maag. 41 1 Roads to Corrins results clearly mark the direction in which a thorough study must and will uncover further insights and eventual success. Since the time that this lecture was given (November 1969) significant progress has been made by deviating from the above mentioned cyclisation procedure in the following way:58 The ADCB-thiolactam-thiolactone derivative (54) reacts very smoothly with dimethylamine in methanol at room temperature to form a labile intermediate assigned the structure (57). Complexation with zinc per- chlorate followed by internal A/B-coupling through oxidation with iodine in methanol containing potassium iodide [see hypothetical intermediate (57a)] acidolytic decomplexation and concomitant sulphide contraction with tri- fluoroacetic acid in dimethylformamide COOCH3 5 in the presence of triphenylphosphine COOCH H COOC COOCH3 \ H,C \ \ / "30CH COOCH (56) Figure 32 58 Work by P.Schneider N. Hashimoto and H. Maag. 41 2 Eschenmoser and finally recomplexation with zinc perchlorate leads to the corrinoid zinc complex (58) (characterised by u.v.-vis. i.r. n.m.r. and m.s.) in a spectro- scopically estimated overall yield [from (54)] of 60-70 %. After acidolytic removal of zinc by treatment with trifluoroacetic acid in acetonitrile cobalt can be introduced very smoothly in high yield by reaction of the metal-free corrinium salt with cobalt(@ in tetrahydrofuran in the absence of base followed by short treatment with aqueous potassium cyanide in air.The electronic spectrum of the chromotagraphically homogeneous (so far not crystallised) synthetic dicyanocobalt(m) complex (59) is reproduced in Figure 34 together with the spectrum of authentic dicyano-5,l S-bis-desmethyl- cobyrinic acid heptamethyl ester (56)*. There are three features in formula (56) which should not escape the readers’ attention the non-differentiable form of the propionic acid carboxy function at ring D the absence of methyl groups in positions 5 and 15 of the corrin chromo- phore and finally the configurational uncertainty with respect to the (CH)- position in ring c. These are problems still to be solved before the journey to synthetic vitamin BIZ can come to an end.The nature of the work described in this lecture is such that very little would have been accomplished without the excellence skill and enthusiasm of my doctoral students and postdoctoral collaborators who have devoted themselves to this endeavour. Their names appear in the list of references; those who have produced the results discussed here are R. Scheffold E. Bertele M. Pesaro E. L. Winnacker and K. J. Schossig (Section 1); B. Golding P. Loliger W. Huber P. Dubs and M. Roth (Section 2); A. Fischli and H. U. Blaser (Section 3); Yasuji Yamada P. Wehrli D. Miljkovic L. Ellis and H. Wild (Section 4); P. Schneider F. Karrer D. Becker N. Hashimoto L. Werthemann and H. Maag (Section 5). I express here my deep appreciation to all of them.I acknowledge Dr. R. Huff’s and Dr. B. Place’s efforts to correct the English manuscript; it is the author who is to be blamed for the remaining linguistic Germanisms. The investigation has been supported by the Swiss National Foundation for the Promotion of Scientific Research. * Note added in proof. In the meantime it has been possible to convert the dimethylamide group of the synthetic dicyano-cobalt(m) corrin complex (59) into the methoxycarbonyl function by alkylation of (59) with trimethyloxonium tetrafluoroborate followed by treatment with aqueous potassium bicarbonate. The product has been obtained in beautifully crystalline form and is believed to be synthetic dicyano-5,15-bis-desmethyl-cobyrinic acid heptamethyl- ester (56); the chromatographic behaviour the u.v.-vis. and i.r.-spectra as well as the 0.r.d.- curve of the synthetic material are identical with the corresponding data of the authentic compound (56).6e The complex (59) has now also been obtained by the Harvard group using the method of base catalysed thioiminoester cyclisation of the corresponding precorrinoid dicyanocobalt(in) complex. ’’ Cobalt(I1) chloride is according to recent findings of the Harvard group in the precorrinoid series much superior to cobalt(1r) perchlorate as complexation agent. 41 3 Roads to Corrins Is- h rn vl v h k 414 Eschenmoser / a( I I I / / 41 5
ISSN:0009-2681
DOI:10.1039/QR9702400366
出版商:RSC
年代:1970
数据来源: RSC
|
6. |
The van der Waals fluid: a renaissance |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 416-432
M. Rigby,
Preview
|
PDF (1147KB)
|
|
摘要:
The van der Waals Fluid A Renaissance By M. Rigby DEPARTMENT OF CHEMISTRY QUEEN ELIZABETH COLLEGE CAMPDEN HILL ROAD LONDON W. 8 The search for equations of state capable of describing accurately the properties of liquids and dense gases has been long and extensive. The literature abounds with examples of equations of widely varying complexity and having from two to eight disposable parameters. That this should be so is an inevitable conse- quence of the need of chemists and chemical engineers for a reliable basis for the calculation of thermodynamic data of great economic importance. It is however unfortunate that in some cases the desire to improve agreement with experi- mental data has rendered the physical basis of the equations less clear. Recent theoretical developments in the theory of dense fluids have shown the importance of a firm physical basis and have perhaps surprisingly led to a revival of interest in one of the oldest and surely the best known of such models that of van der Waals.The appearance in 1873 of the dissertation ‘On the continuity of the gas and liquid phases’ by J. D. van der Waalsl in which the equation of state [P + a (N/Y)’] [V - Nb] = NkT was introduced provoked considerable discussion amongst such distinguished scientists as Maxwell and Lord Rayleigh.’ The columns of ‘Nature’ in the following two decades carried several letters in which the merits and basis of the equation were discussed. This was perhaps not surprising for it has been remarked ‘the results of van der Waals do not appear to be due to any exact mathematical development of his physical ideas but rather as some German writers have commented to in~piration.’~ The resort to inspiration was not universally acceptable as may be seen from Maxwell’s review of the dis~ertation.~ It was however a necessary procedure in view of the complexity of the problem and in the light of contemporary knowledge.Before considering the nature of van der Waals’ approach it will be useful to consider in general terms the basic difficulties involved in any attempt to describe the properties of matter in terms of molecular behaviour. 1 J. D. van der Waals Dissertation Leiden 1873; English translation Threlfall and Adair Physical Memoirs 1890 1 333. An extensive list of references may be found in J. R. Partington ‘An Advanced Treatise on Physical Chemistry’ Longmans Green and Co.London 1949 vol. 1 660. F. G. Keyes and W. A. Felsing J. Amer. Chem. SOC. 1919 41 589. * J. C. Maxwell Nature 1874 10 477. 416 Rigby 1 Intermolecular Forces and Structure There are two essentially different aspects of this problem which must be con- sidered. In the first place the nature of intermolecular interactions must be understood. For an isolated pair of spherically symmetric molecules the inter- molecular potential energy is a function of the separation of the molecular centres which may be represented as u(r) and has the approximate form illus- trated in Figure 1. For non-spherical molecules the potential energy depends also on the relative orientation of the molecules but this discussion will not be concerned with such cases. Theforces between the molecules may be obtained from the gradient of the potential energy curve and we see from Figure 1 that for small separations the forces are repulsive and for larger separations they are attractive.2 1 0 -1 1.5 V f - 2.5 Figure 1 An intermolecular pair potential energy function u(r) and a typical liquid radial distribution function g(r) Information about the pair potential curve may be derived from studies of the properties of dilute gases although it is not in general possible to obtain the potential curve directly in this way. When studying the properties of solids liquids or dense gases it is often necessary to calculate the potential energy associated with several molecules which are simultaneously close together. A fairly accurate estimate may be obtained by assuming that the total energy of such an assembly is equal to the sum of the energies of each pair of molecules.One consequence of using this approximation is that a pair potential energy 41 7 The van der Waals Fluid A Renaissance function which gives satisfactory results for dense matter may not yield good results when applied to dilute gas properties. In particular the Lennard-Jones 12 -6 potential function u(r) = 4 E [(;)I2 - (;)*I where u(o) = 0 and E is the depth of the potential well appears to give rise to a satisfactory description of the liquid inert gases but is not appropriate to the dilute gas properties. The second problem which must be overcome is that of relating the bulk properties of a system to the properties of the constituent molecules. For this we use the methods of statistical mechanics.In general we may calculate the thermodynamic properties of a system provided that we know both the appro- priate potential energy function and the structure of the system.6 It is a fairly easy matter to visualise the nature of a solid or dilute gas but the liquid and dense gas states differ from these cases in that no simple model is available to describe their structures. Although the density of a liquid may approach that of a solid the molecules are not localised and there is no regular lattice. The structure of a liquid or dense gas is best described in terms of molecular dis- tribution functions the most important of which the pair distribution function g(r) defines the number density of molecular centres at a distance r from the centre of a given molecule. When N molecules are present in a total volume V the number of molecular centres in a volume element of size 4nr2dr at a distance r from the reference molecule is defined as 47rr2(N/ V)g(r)dr and the total number of molecular pairs separated by a distance r is then given by 4nr2N2g(r)dr/2V.The mean number density of the fluid is N/V and g(r) which is in general a function of both temperature and density thus defines the microscopic devia- tions from the mean density due to the presence of neighbouring molecules. Values of g(r) may be obtained experimentally from X-ray diffraction studies. A typical pair distribution function for a liquid is shown in Figure 1. It is seen that at separations less than the collision diameter 0 g(r) is very small. This is due to the effect of repulsive intermolecular forces which effectively prevent molecules from approaching to very short distances.At slightly greater values of r the distribution function rises to a maximum which indicates that the likeli- hood of finding a molecuk at these separations is considerably greater (two or three times) than would be expected on the basis of a completely uniform dis- tribution [for which g(r) = 11. At still greater separations the value of g(v) stays fairly close to the value of 1 appropriate to a uniform distribution although small oscillations (which decrease with increasing r ) occur as a result chiefly of packing requirements in the second and subsequent shells of neighbours. This overall situation is commonly summarised in the statement that liquids possess short-range order combined with long-range disorder and it is this combination which causes considerable difficulty in the description of the liquid state.J. S. Rowlinson ‘The Structure of Liquids’ in ‘Essays in Chemistry’ ed. J. N. Bradley Academic Press London 1970 1 1 . 41 8 Rigby If the pair distribution function is known a rather obvious route to some of the bulk thermodynamic properties is available. Since the distribution function enables us to calculate the number of molecular pairs having a given separation r then knowing the potential energy function U(Y) the calculation of that part of the internal energy which is derived from the intermolecular forces follows directly. U’ = jm4nr2g(r). u(r)dr 2v 0 (3) For monatomic systems the total internal energy is the sum of U’ (the con- figurational internal energy) and the ideal gas translational energy.U = U’ f 3 NkTI2 The calculation of the pressure is less readily performed. It may be carried out by the use of a rigorous theorem the virial theorem developed in 1870 by Clausius by means of which the average kinetic energy of a system may be related to its average potential energy. Using this theorem the following equation may be derived. (4) du co PV = NkT - 5 4nr2g(r)rdr . dr 6v 0 ( 5 ) It is evident that if the intermolecular force - du/dr is zero for all values of Y the ideal gas equation of state is recovered. The last term in equation (5) thus represents the corrections to the ideal gas equation resulting from intermolecular forces. It is possible to establish the overall effect of these forces by considering the deviations from ideal gas behaviour.For gases at low to moderate pressures and at temperatures below the Boyle temperature the observed values of PV are less than NkT and the predominant interaction between the molecules is thus an attractive force. In liquids PV exceeds NkT and the more important factor in this case is seen to be the repulsive forces. 2 The van der Waals Model At the time of van der Waals’ original work little was known of the details of either the structure of liquids or of intermolecular forces. The formal relation- ship between the pressure and intermolecular forces had just been derived but the information needed to make even a semi-quantitative evaluation of the terms involved was not available. The essence of van der Waals’ derivation was thus the introduction of judicious approximations based on a physically reason- able model.The most important approximation introduced was the separation of the effects of intermolecular repulsion and attraction. van der Waals believed that there were two essentially distinct corrections to the ideal gas laws one resulting from the tendency of molecules to attract one another at moderate separations and the other a consequence of the finite size of the molecules associated with the short-range repulsive forces. He made the additional important assumption 419 The van der Waals Fluid A Renaissance that the net force on a molecule in the body of a dense fluid resulting from the summed attractions of all its neighbours was zero. Thus although the potential energy of the molecule was lowered as a result of the intermolecular energy the overalI potential field in which the molecule moved was regarded as essentially uniform.Since there were no potential gradients the resultant force was zero. However when a molecule was close to the walls of the containing vessel its neighbours could no longer be symmetrically disposed about it and the molecule was subjected to a resultant force directed away from the wall. The velocity of the molecule as it approached the wall was therefore reduced to a value some- what less than its simple kinetic theory value and the observed pressure was therefore less than the ideal gas value. van der Waals estimated the correction in a manner described by Maxwell4 as ‘ingenious and on the whole satisfactory’ and suggested that the term P + a(N/ V)2 should replace the pressure in the ideal gas equation of state.The empirical constant a was a measure of the strength of the attractive forces and was assumed to be independent of temperature. van der Waals treated the repulsive forces by considering each molecule to have a hard core of diameter 0 so that it was not possible for the centres of a pair of mole- cules to approach to a distance less than 0. Thus each molecule was surrounded by a sphere of radius (T from which the centres of all other molecules were excluded. van der Waals attempted to calculate the extent to which the total volume V available to the molecular centres was reduced in this way and con- cluded that the available volume was V - Nb where b = 2ro3/3 four times the volume of a hard core. He admitted that this was an approximation valid only at low densities.Replacing the total volume by this estimate of the available volume we obtain van der Waals’ equation of state in its familiar form [P + a(N/V)2] [ V - Nb] = NkT which may be written alternatively p= - NkT - a ( ; ) 2 V - Nb (7) in which the effect of each correction term on the pressure is more clearly shown. A. The Generalised van der Waals Equations.-It is important to recognise that the van der Waals equation is just one of a general class of equations of state3 which may be represented where /3 and a are functions of the density N/V but are temperature indepen- dent. We shall describe equations of this type as generalised van der Waals equations. The function 18 represents the effects of molecular size and a relates to the effects of intermolecular attraction.The specific van der Waals form is obtained by the use of the approximate relations 420 Rigby V V - Nb p = - and a = a (9) We must distinguish between results following from the general form of equation (8) and those dependent on particular choices for the functions a and 18. An interesting comparison may be made between equation (8) and the thermo- dynamic equation of state p=T(;)v - (yT It is immediately seen that the generalised van der Waals equation requires that the thermal pressure coefficient ( $)v be a function only of the molar volume and consequently that at constant volume the pressure should increase linearly with temperature. For a wide range of liquids the observed behaviour closely approximates to this.s Perhaps the most important consequence of equation (8) is that the entropy is determined solely by the term p(N/V) and is thus governed only by considera- tions of molecular size.The structure and hence the disorder of the fluid is therefore independent of the attractive forces and is identical with that of a system of molecules having the same repulsive forces but no intermolecular attraction. This result is fundamental to the generalised van der Waals equations and greatly simplifies the treatment of attractive forces. If the structure due to repulsive interactions can be determined the thermodynamic properties of any fluid with an arbitrary attractive potential u(r) may be readily calculated. Defining properties of the fluid without attractive forces by the superscript 0 we may write u=uo+- N 2 5" 4nr2gO(r). u(r)dr 2v 0 and since S = SO the Helmholtz free energy is given by 03 A = A0 + 5 4m2go(r)u(r)dr 2v 0 If following van der Waals we associate the fluid without attractive forces with the hard sphere fluid these equations permit us to calculate the properties of any fluid provided the distribution function for hard sphere systems is known.B. Comparison with Experimental Data.-As is well known the original van der Waals equation gives a qualitatively correct description of the critical properties of fluids. The critical properties may be expressed in terms of the constants a * J. S. Rowlinson 'Liquids and Liquid Mixtures' Butterworths Scientific Publications London 2nd edn. 1969. 421 The van der Waals Fluid A Renaissance and b of equations (6) and (7) and experimental values of two of the critical temperature pressure and volume may be used to derive characteristic values of these constants for specific substances.Below the critical temperature the van der Waals isotherms do not have the experimentally observed horizontal portion in the two-phase region but are sinuous curves. These may be re- interpreted using the equal areas rule of Maxwell. A horizontal line is drawn intersecting the isotherm at three points and defining regions of equal area above and below the line. The extreme intersections correspond to liquid and vapour states of equal free energy and the modified isotherms then give a de- scription of a liquid in equilibrium with its vapour. To this extent it may be said that van der Waals achieved his objective. However a closer study of the pre- dicted critical behaviour shows considerable disagreement with the experimental results.In particular the form of the critical isotherm and of the coexistence curve near to the critical point are badly at variance with the data. This dis- agreement appears to be of a fundamental nature and recent theoretical studies’ have shown that correlations resulting from short-range attractive forces are of major importance in determining properties in the critical region. The van der Waals model with its assumption of a uniform attractive potential field cannot reproduce this behaviour correctly. The use of van der Waals equation to calculate PVT data for simple systems leads to results which are qualitatively in agreement with experimental values but which are inadequate for accurate work.One important factor is that the pressure calculated from equation (7) is the difference of two terms of similar size and a relatively small error in either term may have a large effect on the final result. This cancellation does not always occur in the estimation of other properties such as the internal energy for which the van der Waals theory is generally much more accurate. One apparently gross inadequacy of van der Waals equation is its failure to predict the existence of a solid phase. It is true that van der Waals was not attempting to account for the solid but it does not seem an unreasonable expectation if the physical basis of the equation is sound. The origins of these discrepancies may best be investigated by considering separately the approximations used for the functions a and /3 of equation (9).The van der Waals approximation for a implies that the configurational energy of the fluid is a linear function of the density and is independent of temperature. Experimental values of the configurational energy of liquid argon may be obtained from known thermodynamic data and reveal that this approximation is remarkably accurate. The only significant deviations occur near to the critical point. For other substances the detailed data needed for the calculation are not available. However values of the van der Waals ‘constant’ a may be derived for a number of simple liquids. From equations (8) (9) and (10) we see that a = (5)’ (g) T B. Widom Science 1967 157 375. 422 Rigby For several simple liquids the values of a calculated using equation (13) and known experimental data are found to vary only slightly over the whole liquid range.6 These results suggest that the approximation for a is quite accurate and that the use of refined versions cannot be expected to improve significantly the overall quality of the equation of state.The function p describes the behaviour of molecules which have finite size but no attractive forces. They are thus regarded as hard spheres and the equation PV V NkT- V - Nb -- should reproduce the hard-sphere equation of state. Although the hard-sphere model is not a very realistic physical model for real molecules it has been ex- tensively studied in the development of theories of fluids and the equation of state is well established. Equation (14) is not a good approximation to the true equation of state at other than very low densities.The explanation of this failure is that the excluded volume is not simply proportional to the number of molecules in a given volume. The true excluded volume is less than the van der Waals value because the ‘excluded volume per molecule’ relates to the exclusion of a molecular centre and it is possible for the excluded volumes to overlap to a certain extent in such a way that the volume excluded by a pair of molecules which are close together is less than the sum of the two separate excluded volumes per molecule. The van der Waals approximation thus overestimates the size of the correction for the molecular size and the pressures calculated from equation (14) are greater than the true values. The exact hard-sphere equation of state is compared with the van der Waals form in Figure 2 and it is seen that although the approxi- mate form is correct at low densities it is badly in error at higher densities and indeed gives rise to an infinite pressure at about one-third of the close packed density.It is largely in the treatment of the effects of molecular size that the van der Waals equation is inadequate and we shall now consider in more detail the properties of the hard-sphere fluid. 3 The Hard-sphere Fluid A great deal of work has been done in the past thirty years on the development of theories of fluids expressed in terms of distribution functions. These theories provide a complete formal basis for the description of the liquid state but it is not possible to calculate the values of the distribution functions from first principles.In order to obtain values of g(r) it is necessary to introduce approxi- mations into the theory. Much effort has been expended in attempts to devise satisfactory approximations but in no case have completely satisfactory solu- tions been obtained. In order to test the various approximations it is desirable to compare the theoretical results with those found experimentally from X-ray diffraction studies. However a direct comparison is not simple since a know- ledge of the pair potential energy function is a prerequisite for the calculation of distribution functions and our present knowledge of these functions is in- complete. To some extent these problems may be overcome by the use of 423 The van der Wmls Fluid A Renaissance 3c Pressurt - Pb kT 2c I I I I I I I I V.W.I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 I I P.Y. 1 I I I I I Density Nb/V Figure 2 The hard-sphere equation of state. The solid curve shows the results of computer studies and includes the solid-fluid phase transition. The dashed curves show the approximate equations of state based on the van der Waals (equation 14) and Percus-Yevick (equation 15) approximations 424 Rigby simulation studies. By use of the Monte Carlo and Molecular Dynamics methods with fast electronic computers,8 it is possible to study the behaviour of an imaginary model fluid in which the potential energy function is specified by the investigator. The results obtained are essentially exact and are commonly regarded as pseudo-experimental data for the purpose of comparison with theory.Because of the considerable simplifications afforded by its use the hard sphere model has been very extensively studied in this way and the equation of state has been firmly established. The most striking result of these simulation studies was the discovery0 that at certain densities the system may exist in two different states one of which has the long-range order characteristic of a solid and the other which is associated with a higher pressure having no long-range order. The latter is believed to correspond to a metastable fluid state and the phenomenon is thought to be associated with a solid-fluid phase transition. Considerations of molecular size without any attractive forces thus lead to the existence of two states of matter. Some pair distribution functions have been derived from simulation studies and may be compared with the predictions of the various approximate theories.In general however comparisons have been made with the calculated equation of state which is determined (for hard-sphere systems only) by the value of the pair distribution function at the point of contact g(o). One theory in particular has been found to yield good solutions for the hard-sphere systems over the whole liquid range. This theory due to Percus and Yevick,lo leads to an equation of state which may be written in the form Nb where 6 = - PV 1 + 6 + k " N k T - (1 - b3 4v - However it does not predict the observed phase transition and its application to systems having more realistic pair potentials is often difficult. It is now possible to replace the approximate hard-sphere term in the van der Waals equation by the exact result.Since the hard-sphere system can exist in both solid and fluid forms and the simple van der Waals theory shows both the liquid and gaseous phases this corrected equation may reasonably be expected to show the existence of all three possible states of matter and this expectation is realised. One major failure of the van der Waals equation is therefore removed and is shown to be a consequence of the inaccurate descrip- tion of the hard-sphere equation of state. 4 Perturbation Theory We may now consider also the re-derivation of the term describing the effects of the attractive forces. van der Waals made the assumption that the distribu- tion of the molecules was uniform i.e. that g(r) = 1 for all Y 3 CT. It is evident from Figure 1 that this is not a good approximation for values of r between * M.A. D. Fluendy and E. B. Smith Quart. Rev. 1962,16 241. lo J. K . Percus and G. J. Yevick Phq's. Rev. 1958,110 1. B. J. Alder and T. E. Wainwright J . Chem. Phys. 1960 33 1439. 425 4 The van der Wads Fluid A Renaissance about CJ and 20 although it becomes better for larger separations. The van der Waals arguments will lead to a good approximation if the potential function is long ranged. In this case the bulk of the configurational energy will arise from interactions between molecules which are far apart and for which g(r) N 1. Indeed it has been shown by rigorous methods" that the van der Waals attractive term will arise as an exact result in the case of molecules which consist of a hard core together with an attractive energy term of infinite range and vanishingly small magnitude.However we know that these conditions are not satisfied for real molecules and are led to consider an alternative approach to the problem. Zwanzig12 showed that the hard-sphere fluid may be used as the basis of a perturbation expansion the physical basis of which is that of the generalised van der Waals equation namely that the geometrical distribution of real moleculm is deter- mined by the repulsive intermolecular forces and may be regarded as identical to that of a hard-sphere fluid of a suitably chosen density. The attractive forces act as a source of internal energy which maintains the high density of the liquid but do not otherwise significantly affect the distribution of the molecules. The first-order perturbation theory leads to equation (12) when gO(r) is the pair distribution function of the hard sphere system and u(r) is the difference between the potential energy function of the perturbed system and the hard- sphere potential.It is possible to write formal expressions for the second-order perturbation contributions but these involve higher order distribution func- tions about which very little is known. If only the first-order term is obtained the perturbation theory has the characteristics of the generalised van der Waals equation mentioned earlier. The neglect of higher order perturbation terms may be shown to be formally acceptable at high temperatures and early applica- tions of the theory were concerned with the development of a high-temperature equation of state for dense gases.We have not so far considered the basis for the selection of the hard-sphere diameter. Since this value determines both the hard-sphere contribution to the pressure and also the size of the correction terms it is evident that a reliable basis for its estimation is essential. It is possible to make a qualitative assessment of the hard-sphere diameter on the basis of simple physical arguments. The true pair potential rises rapidly as the separation of a pair of molecules is decreased from the collision diameter 0. The hard-sphere potential becomes infinite for pair separations less than its collision diameter and such a situation has zero prob- ability. In real systems the molecules may approach to separations less than o and their closest distance of approach will be reached when the potential energy is equal to the kinetic energy which was possessed at large separations.At higher temperatures the average kinetic energy is higher and the average value of the distance of closest approach would be expected to be smaller. The effective hard-sphere diameter would thus be expected to decrease with rise in temperature. l1 M. Kac G. E. Uhlenbeck and P. C. Hemmer J . Math. Phys. 1963,4 216. R. W. Zwanzig J . Chem. Phys. 1954,22 1420. 426 In the earlier calculation^^^ made using the first-order perturbation theory a rather arbitrary choice of hard-sphere diameter was made. Although the em- phasis in this work was on the development of a high-temperature equation of state for gases the use of a Lennard-Jones 12-6 potential as the perturbing potential led to a reasonably good estimation of the critical temperature of the inert gases.The hard-sphere distribution functions used in this work were calculated from exact analytic expressions appropriate only to fairly low densities. With the development of the Percus-Yevick approximation reliable estimates of the hard-sphere distribution functions over a wide density range became avail- able and further calculation^^^ were made again using the 1 2 - 6 potential. These calculations revealed the great sensitivity of the theory to the choice of the hard-sphere diameter but after some slight adjustment of parameters an equation of state was developed which was in good agreement with the data for gaseous argon at high densities. The applicability of the perturbation theory to liquids was demonstrated by Smith,15 who pointed out that the experimental data for many liquids were consistent with an equation of state of the form of equation (8) and showed that the function fi could be derived from experimental thermal pressure coefficients and agreed closely with the hard-sphere equation of state.The most successful extension of the perturbation theories has been due to Barker and Henderson,16 who have considered both the choice of hard-sphere diameters and the development of approximate methods for the estimation of the second-order perturbation term. By a suitable definition of the perturbing intermolecular potential these authors were able to develop a theory which dealt separately with the attractive forces and the slightly soft nature of the repulsive energy.The treatment of the repulsive forces was based on a method due to Rowlins~n,~~ and provided a means of defining an effective hard-sphere diameter which was independent of density and decreased with increasing tem- perature. The first-order correction for the attractive forces was essentially that of Zwanzig and in addition Barker and Henderson were able to approximate the second-order attractive term using accessible functions. By use of this theory the equation of state was calculated for the Lennard-Jones potential and the results compared with those obtained from Monte Carlo and Molecular Dynamics studies. These are shown in Figure 3. The agreement is seen to be excellent over a wide range of temperature and density and this success may be seen as a strong justification of the basic validity of the approach.It seems that the effect of neglecting the higher order terms is negligible except perhaps at the lowest temperature shown which corresponds roughly to the triple point tem- perature of the inert gases. The critical constants obtained from this model are in excellent agreement with those obtained from computer studies. In addition l3 E. B. Smith and B. J. Alder J. Chem. Phys. 1959 30 1190. l4 H. L. Frisch J. L. Katz E. Praestgaard and J. L. Lebowitz J. Phys. Chem. 1966,70,2016. l5 E. B. Smith J . Chem. Phys. 1962 36 1404. l6 J. A. Barker and D. Henderson J . Chem. Educ. 1968,45,2; J. A. Barker and D. Henderson J . Chem. Phys. 1967 47 4714. I 7 J . S. Rowlinson Mol. Ph,vs. 1964 8 107. 427 The van der Waals Ffuid A Renaissance the densities and pressures of the coexisting gas and liquid phases were calculated using the Maxwell equal area rule and the derived coexistence and vapour pressure curves were found to be in good agreement with the experimental data P L NRT 4 2 0 I 1 I I I .5 1 .o 1.5 2.0 Density Nb/V Figure 3 The equation of state of aj?uid of Lennard-Jones molecules.The compressibility factor PV/NkT is shown as a function of the density at four values of the reduced temperature kTJE. The curves are based on the perturbation theory and are labelled with the value of the reduced temperature. The critical isotherm corresponds to a reduced temperature of ca. 1.35. The points show the results of computer calculations. (W. W . Wood and F. R. Parker J. Chem. Phys. 1957 27 720; 1. R. Macdonald and K. Singer Discuss. Faraday SOC.1967 43,40; L. Verlet Phys. Rev. 1967 159 98.) 428 Rigby for argon. We should note that this appears to be the first theory of liquids capable of describing the vapour pressure curve correctly and may infer from this that the approach leads to an accurate description of the entropy of a liquid. 5 Applications of the van der Waals Concepts It would be inappropriate to describe in detail the numerous applications of the van der Waals model. Instead three examples have been selected which emphasise the basic simplicity of this approach and the insight which it can provide. A. Phase Changes.-The identification of the entropy of a fluid with that of an appropriately chosen hard-sphere system provides a simple basis for calculating entropies of vaporisation and fusion. For example the entropy of vaporisation of a fluid at the normal boiling point is equal to the difference between the entropy of the hard-sphere gas at one atmosphere pressure and that of the hard- sphere fluid at the liquid molar volume.This difference may be readily calculated if the hard-sphere equation of state is known. Yosim and Owens1* used the hard-sphere equation of state of equation (15) and took values of the collision diameters derived from studies of dilute gas properties. They were then able to calculate entropies of vaporisation for a large number of non-ionic liquids and obtained results which were in very good agreement with the experimental values in almost all cases. These authors also applied a similar approach to the calculation of entropies of fusion of the inert gases and obtained results in fair agreement with experiment.An alternative approach to melting phenomena was investigated by Widom and Longuet-Higgins,l8 who used a generalised van der Waals equation consist- ing of the exact hard-sphere results including the solid-fluid phase transition and the original form of the function a. The use of this equation leads to isotherms which have a zig-zag portion in the density region near the hard-sphere melting transition. These may be re- interpreted using the Maxwell equal area rule and the equation then describes the temperature dependence of the melting density for a system of hard spheres immersed in a uniform potential field. Widom and Longuet-Higgins were able to establish the value of a/kT corresponding to the triple point by using the criterion that the activities of the solid liquid and gas phases should be equal.They then evaluated such properties as the ratio of the liquid and solid molar volumes and the entropy of fusion at the triple point and obtained values in good agreement with the experimental data for argon. These calculations were made without the introduction of any arbitrary parameters and the success of this approach gives a striking confirmation of the validity of the physical model. S. J. Yosim and B. B. Owens J . Chem. Phys. 1963 39 2222. la H. C. Longuet-Higgins and B. Widom Mol. Phys. 1964 8 549. 429 The van der Waals Fluid A Renaissance B. Thermodynamic Properties of Mixtures.-Until recently the study of liquid mixtures has been dominated by a number of related theories which were based on the assumption that the distribution of different types of molecule within the liquid was random.The possibility of preferential packing arrangements result- ing from differences in the intermolecular forces associated with the various types of pair interactions were neglected in the simplest forms of these theories. Although they have been successfully applied to simple mixtures of components similar in molecular size the application of these theories to mixtures in which there is a large size difference has not proved satisfactory. In such cases large positive values of the excess free energy GE are predicted and these are not observed experimentally. Mixture theories based on the van der Waals concepts have proved much better able to take account of the effects of size differences and seem likely to provide the basis for future work in this field.The van der Waals picture of a liquid mixture is again based on the assump- tion that the structure is determined largely by the repulsive intermolecular forces and may be closely approximated by the use of a hard-sphere model. However the different species in a mixture will in general have different repulsive forces and will be represented by hard spheres of different sizes. The overall picture is thus of a hard-sphere mixture immersed in a uniform field due to the attractive intermolecular forces. There have been several computer studies20 of the properties of hard-sphere mixtures and the Percus-Yevick theory has also been successfully applied to these systems.21 Calculations of the thermodynamic mixing functions22 have shown that hard spheres mix at constant pressure with a small decrease in volume and a small negative excess free energy GE.The decrease in volume occurs because it is possible to pack spheres of different sizes more efficiently than those of a single size and the structure of the mixture is therefore rather different from that of a pure hard-sphere fluid. The van der Waals theory suggests that this structural change will also occur when real molecules are mixed and it now appears that it was the failure of earlier theories to take account of this effect which was responsible for their inability to describe mixtures whose components differed significantly in size. Two different methods of applying the van der WaaIs model to mixtures have been described. Snider and Herringt~n~~ used a generalised form of equa- tion (16) to describe the equation of state of a mixture.Leland Rowlinson and Sather24 avoided the choice of an explicit equation of state by using an approach based on the principle of corresponding states which gave accurate results when applied to hard sphere mixtures. In both cases the calculated values of the thermodynamic properties of simple binary mixtures were in considerably better agreement with the experimental data than those based on earlier theories. 2o E. B. Smith and K. R. Lee Trans. Faraday SOC. 1963 59 1535; B. J. Alder J. Chem. Phys. 1964,40,2724. 21 J. L. Lebowitz Phys. Rev. 1964 133 A895. 22 J. L. Lebowitz and J. S. Rowlinson J. Chem. Phys. 1964,41 133. 23 N. S. Snider and T. M. Herrington J . Chem. Phys. 1967,47 2248.24 T. W. Leland J. S. Rowlinson and G. A. Sather Trans. Faraday SOC. 1968 64 1447. 430 R&by C. Transport Properties of Dense Gases and Liquids.-Although the dilute gas transport properties of simple substances may generally be calculated the extension to higher densities is very complex owing to the possibility of many- body collisions. For hard spheres this problem does not arise since the potential is not long ranged and the theoretical treatment is less difficult. E n ~ k o g ~ ~ deve- loped an approximate theory in which it was assumed that a dense hard-sphere system behaves like a dilute hard-sphere system with the modification that the collision rate is higher in the dense system. If the ratio of the collision rate at high density to that at low density is Y the Enskog theory gives values of the transport properties of the dense hard-sphere fluids in terms of their low density values and the factor Y.Values of Y may be obtained from the hard-sphere equation of state and the transport properties of dense hard-sphere systems may then be calculated using the known results for the low density coefficients. The extension of these results to real molecules is based on the van der Waals concept of a uniform potential field in dense fluids. If this were exact molecules would travel in straight lines between collisions. Although this is undoubtedly an approximation the successful application of the van der Waals model to equilibrium properties has suggested that the true situation is probably close to this. In order to use the Enskog equations for real systems effective hard-sphere diameters for the molecules are needed.Dymond and Alderz6 obtained these from experimental compressibility data and calculated the coefficients of viscosity and thermal conductivity for the heavier inert gases at several temperatures and densities above the critical obtaining results in good agreement with the avail- able experimental data. This approach has also been successfully applied to the calculation of the viscosity of liquid mixtures of argon and kryptonz7 and to the study of gaseous diffusion in liquids.28 6 Conclusion Almost a century has passed since the van der Waals equation of state was first proposed. After a period in which it was extensively used in early studies of phase equilibria and the properties of liquid mixtures it passed into disuse as its inadequacies were revealed.Only in recent years has it become apparent that these were due to an inadequate treatment of the physical model on which the equation was based rather than to weaknesses in the model itself. There is now much evidence that the structure of a liquid or dense gas is determined almost entirely by repulsive intermolecular forces and may be accurately reproduced by the use of a hard-sphere model. The attractive intermolecular forces give rise to an essentially uniform potential field and maintain the high density. Equations of state which accurately embody this picture of a liquid have proved remarkably 26 See S. Chapman and T. G. Cowling ‘The Mathematical Theory of Non-Uniform Gases’ Cambridge University Press Cambridge 2nd edn. 26 J. H. Dymond and B. J. Alder J . Chem. Phys. 1966 45 2061. 27 N. Jhunjhunwala J. P. Boon H. L. Frisch and J. L. Lebowitz Physicu 1969,41 536. 28 E. McLaughlin J. Chem. Phys. 1969,50 1254. 43 1 The van der Waals Fluid A Renaissance accurate and the use of this model has led to important advances in a number of fields. The author wishes to thank Professor J. S. Rowlinson and Dr. E. B. Smith for their encouragement and helpful advice. 432
ISSN:0009-2681
DOI:10.1039/QR9702400416
出版商:RSC
年代:1970
数据来源: RSC
|
7. |
The separation of polar, steric, and resonance effects in organic reactions by the use of linear free energy relationships |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 433-453
J. Shorter,
Preview
|
PDF (1511KB)
|
|
摘要:
The Separation of Polar Steric and Resonance Effects in Organic Reactions by the Use of Linear Free Energy Re1 at ionships By J. Shorter DEPARTMENT OF CHEMISTRY THE UNIVERSITY HULL 1 Introduction. Structural modification of a reactant molecule may influence the rate or equi- librium constant of a reaction through polar steric or resonance effects. When a substituent is introduced at a point remote from the reaction centre only the operation of polar effects need usually be considered and quite detailed under- standing of the influence of structure upon reactivity is possible. Linear free energy relationships notably the Hammett equation are very important in this. When the structural modification is close to the reaction centre our under- standing is less advanced. Thus for many years it was impossible to assess the relative importance of different factors contributing for example to the ‘ortho- effect’ in aromatic systems.About 1952 R. W. Taft suggested a basis for separat- ing polar steric and resonance effects by the use of linear free energy relation- ships and developed his ideas in detail. Taft’s work involved an analysis of the rate coefficients of alkaline and acid-catalysed hydrolysis of esters. Considerable use has been made of this ‘Taft analysis’ and of the various substituent para- meters developed from it. The present Review seeks to survey the basic assumptions and development of the Taft analysis and the use which has been made of the various parameters during the last eighteen years. It is necessary to define the terms used in the title. We consider a reactant molecule RY and an appropriate standard molecule ROY.Initially we suppose that Y is not conjugated with either R or R,,. For RY the polar efect of the group R comprises all the processes whereby a substituent may modify the electrostatic forces operating at the reaction centre Y relative to the standard ROY. These forces may be governed by charge separations arising from differences in the electronegativity of atoms (leading to the presence of dipoles) the presence of unipoles or electron delocalisation. Field inductive (through-bond polarisa- tion) and mesomeric effects may in principle be distinguished. Because of the difficulty of distinguishing between field and through-bond effects in practice the term ‘inductive effect’ is often used to cover both and is so used in this Review.The term ‘resonance’ is sometimes used in connection with the meso- meric effect (zk. the ‘resonance polar effect’; see Section 2) but this is not its ordinary usage in this Review. When R and Ro may be conjugated with Y the above discussion holds but The Scope of the Review 433 The Separation of Polar Steric and Resonance Eflects in Organic Reactions additional influences may arise from the more extensive electron-delocalisation. The mesomeric part of the polar effect will be modified. There will also be the resonance efect of R (as in the title) which is concerned with the extent of conjugation of R with Y relative to the standard ROY and is not part of the polar effect. This distinction is sometimes not clearly made ‘resonance effects’ in a wide sense are lumped together and treated as if they are polar in nature (see Section 2).In this Review ‘resonance effect’ will usually have a narrow connotation as above. Steric eflects are caused by the intense repulsive forces operating when two non-bonded atoms approach each other so closely that such approach involves non-bonded compression energy. The primary steric efect of R is the direct result of compression which arises because R differs in structure from R in the vicinity of the reaction centre. A secondary steric efect involves the moderation of a polar effect or resonance effect by non-bonded compressions. In discussing the influence of polar steric and resonance effects it is necessary to consider differentially interactions in initial and transition states in the case of rate processes and in initial and final states in the case of equi1ibria.l 2 Introduction to Linear Free Energy Relationships.The Hammett Equatien Empirical correlations of the reactivities of organic compounds are usually linear relationships involving logarithms of rate or equilibrium constants. At constant temperature the logarithm of a rate constant (k) is proportional to the standard free energy of activation (dG:) and that of an equilibrium constant (K) to the standard free energy change of reaction (AGO). The term linear free energy relationship is thus appropriate.2 In the early 193Os Hammetta at Columbia and Burkhardt4 at Manchester discovered linear relationships involving log k or log K for a number of systems. This work led to the formulation of the Hammett equation (1937) which describes the influence of polar meta- or para-substituents on the sidechain reactions of benzene derivatives.s The Hammett equation takes the forms where k or K is the rate or equilibrium constant respectively for a sidechain reaction of a meta- or para-substituted benzene derivative and ko or KO is the constant for the ‘parent’ compound.The substituent constant 0 measures the polar effect of the substituent (relative to hydrogen) and is in principle inde- pendent of the nature of the reaction. The reaction constant p depends on the 1 General account of substituent effects C. K. Ingold ‘Structure and Mechanism in Organic Chemistry’ Bell London 2nd edn. 1969. 2 J. Shorter Chem. in Britain 1969 5 269 and references therein. 3 L. P. Hammett Chem. Rev. 1935,17 125. 4 G. N. Burkhardt W.G. K. Ford and E. Singleton J. Chem. SOC. 1936 17. 6 L. P. Hammett ‘Physical Organic Chemistry’ McGraw-Hill New York 1940 chap. 7. 434 Shorter nature of the reaction and measures the susceptibility of the reaction to polar effects. Hammett chose the ionisation of benzoic acids in water at 25 “C as the standard process for which p was defined as 1.0oO. Values of u for many substituents viz log (K/Ko) were then calculated. The equation was found to apply satisfactorily to all but a few members of fifty-two reaction series. In 1953 Jaffb found data for 371 reaction series to which the equation could be applied with considerable succes~.~ By this time however a number of exceptions to the uniqueness of the substituent constant had been noted. Indeed one had been apparent to Hammett the ionisation of para-nitrobenzoic acid gave a = 0.778 for p-NO, but this value proved inapplicable to the reactions of phenol or anilineq5 In these a value of 1.27 was required.This exalted value was attributed to cross-conjugation of p-NO (- Meffect)’ with OH or NH (f Meffect) so thatp-NO was effectively more electron-attracting than it was in para-nitrobenzoic acid. Jaff6 found further examples of this ‘duality of substituent constants’.6 The application of the Hammett equation with increasing refinement has involved the elaboration of this concept. Brown’ extended the Hammett equation to highly electron-demanding reactions notably electrophilic substitution in the aromatic ring which had previously appeared to be outside the scope of the equation. This involved a new set of substituent constants urn+ and up+.(Special substituent constants for nucleophilic reactions cf. p-NO above are designated a-.) In the late 1950s the Hammett equation was re-examined by Wepster and by Taft. WepsteP strongly criticised the ‘duality of substituent constants’. In his view ‘mesomeric para interaction’ inevitably depends on both the mesomeric effect of the para-substituent and of the reaction centre. Thus a ‘multiplicity’ or ‘sliding scale’ of a values would be expected rather than a single exalted constant. Wepster introduced un to designate the normal unexalted value. Certain o values considered to be free from the effects of the above interaction were taken as primary values e.g. those for m-CI m-Me and m-NO,. Only these were to be used in evaluating p . For other substituents u values relevant to particular reaction series were calculated and un values for such substituents were suggested.The ‘sliding scale’ was convincingly demonstrated. Taft’s9 approach was through a quantitative separation of substituent effects into ‘inductive’ and ‘resonance’ i.e. mesomeric contributions. The inductive contribution was considered to be given by a parameter based on aliphatic and alicyclic reactivities (see below). The resonance contribution OR was shown to be reaction-dependent but an unenhanced resonance parameter ORO was defined by reference to ‘insulated reaction series’ i.e. when the reaction centre is not conjugated with the ring. The values of ORO and 01 were appropriately combined to give a normal substituent constant oo. 6 H . H. Jaffb Chem.Rev. 1953,53 191. 7 H. C. Brown and Y . Okamoto J . Amer. Chem. Soc. 1958 80 4979 and earlier papers. 5 H. van Bekkum P. E. Verkade and B. M. Wepster Rec. Truv. chim. 1959,78 815. R. W. Taft J. Phys. Chem. 1960 64 1805 and earlier papers. 435 The Separation of Polar Steric and Resonance Efects in Organic Reactions 3 The Separation of Polar Steric and Resonance Effects A. Introduction to R. W. Taft’s Work.-The occurrence of steric as well as polar effects in aliphatic systems and ortho-substituted aromatic systems compli- cates the devising of linear free energy relationships. Little progress was made in this direction until the early 1950s when Taft proposed a procedure (based on a suggestion by Ingoldlo) for evaluating polar and steric parameters in such systems. His analysis essentially involves a quantitative separation of polar steric and resonance effects in the total effect of structure upon reactivity in ester hydrolysis.l1 Taft suggested the following equation for evaluating the polar effect of a substituent R as manifested in the ester RC02R1 (T* = [log ( k / k o ) ~ - log (k/ko)~]/2’48 (3) o* is a polar substituent constant for R. The rate constants k refer to reactions of RCO,RE and k to reactions of MeCO,R1 as a standard. B and A refer to basic and acidic hydrolysis carried out for the same R1 solvent and temperature. The factor 2-48 puts the o* values on about the same scale as u. The equation may also be applied to ortho-substituted benzoic esters o-XCsH4-CO2R1 with o-MeC,H,CO,R1 as standard. The terms have the following significance log ( k / k ) ~ gives the sum of the polar steric and resonance effects of R (or X); log ( k l k ) ~ gives the sum of the steric and resonance effects of R (or X); the difference gives the polar effect of R (or X).A further equation Es = log (k/ko)A (4) gives a steric substituent constant Es although for systems which are conjugated with -C02R1 E contains a resonance contribution. This procedure is based on three assumptions (1) the relative free energy of activation may be treated as the sum of inde- (2) in corresponding acidic and basic reactions the steric and resonance (3) the polar effects of substituents are markedly greater in the basic than in Assumption (1) is by no means self-evident if the various effects interact linear analysis is inappropriate. No progress is possible however unless such a simplifying assumption is made as a first approximation.Its validity is tested by the usefulness of the results obtained. Assumption (2) lies at the heart of the analysis. Taft justified it by the following argument. The transition states (having structures closely resembling those of thc carbonyl addition intermediates) for the acidic and the alkaline reactions (1) and (21 differ by two protons. pendent contributions from polar steric and resonance effects; effects are the same; the acidic reaction. l o C. K. Ingold J . Chem. Suc. 1930 1032. l1 R. W. Taft ‘Steric Effects in Organic Chemistry’ ed. M. S. Newman Wiley New York 1956 chap. 13. This article summarises Taft’s original papers. 43 6 Shorter 0 I R-?-- OH I The small size of these protons makes it reasonable that the steric interaction produced in attaining the transition state should be the same in the two reactions.Further any resonance effect between R and CO,R1 should be the same since both transition states are effectively saturated at the carboxylate carbon. Thus the steric and resonance effects of R relative to Me should be eliminated in equation (3). This assumption will be examined in detail later in this article. Assumption (3) is supported by the Hammett p constants for the reactions of rn- or p-substituted benzoates.6 For basic hydrolysis p values are commonly in the range +2.2 to f2.8. For acid-catalysed hydrolysis or the related esterifica- tion of benzoic acids p values lie close to zero in the range - 0.2 to +0.5. Thus o* may be approximately scaled to CT through a factor 2-48 the average of the p values recorded by Jaffe for basic hydrolysis of benzoates,6 and the polar contribution to Es may be neglected.Taft regards the ultimate justification of the assumptions to be the nature of the results to which they lead. He carried out a careful analysis of a vast amount of data to give o* and Es values and showed that these could be used to interpret reactivity in other reactions. For any wide range of substituents the conditions regarding constancy of solvent and R1 could not be completely met. Further Taft used data from acid-catalysed esterification to supplement those from acid- catalysed hydrolysis. He justified this by pointing out the similarity of the transition states for the two processes. For some substituents R or X for which there were plenty of data it appeared that log (k/k,)A or log ( k / k ) ~ did not depend much on solvent or R1 and that log ( k l k ) ~ was approximately the same for hydrolysis and esterification.Supposedly equivalent log (klk,) values where available were averaged Taft holding that ‘the use of average values appears desirable on the basis that small specific effects and experimental errors will be reduced’.l2 Selected values of u* and Es are shown in Tables 1 and 2. For a full discussion of the substituent constants for numerous groups the reader should consult Taft’s articles.ll These include a limited quantity of data for variation of the group R1 in the ester RCO,R1. While the Es values in Taft’s Tables are always based on the original defining reaction many of the o* values are not directly based on the alkaline hydrolysis of esters but are based on the application of the original o* values to other reactions.Table 1 contains two examples of this. l 2 R. W. Taft J . Amer. Chem. SOC. 1952 14 2729. 437 The Separation of Polar Steric and Resonance Efiects in Organic Reactions Table 1 Steric and polar parameters for aliphatic systemsa R H Me Et Pri But Prn Bun Bui neopen t y I ICH C12CH CI,C MeO-CH PhCH2 Ph(CHd2 ClCH2 MeCH=CH Ph ES $1.24 0.00 - 0.07 - 0.47 -1.54 -0.36 -0.39 -0.93 -1.74 -0.24 -0.37 -1.54 - 2.06 -0.19 -0.38 -0.38 ( - 1.63)‘ ( - 2.55)d o* + 0.49 0.00 -0.10 -0.19 -0.30 -0.115 -0.13 -0.125 -0.165 +1.05 +0.85c + 1 -94 +2-65 $0.52 +0*215 +0*08 +0-36 $0.60 aR. W. Taft ‘Steric Effects in Organic Chemistry’ ed. M. S. Newman Wiley New York 1956 chap.13; 6From sulphation of alcohols; CFrom ionisation of carboxylic acids; dFor significance see text. Table 2 Steric and polar parameters for ortho-substituents in benzoatesa X H OMe Me F Cl Br I NO2 Es + 0.99 0.00 +0.49 +0-18 0.00 -0.20 -0.75 - UO* -0.22 0.00 +Om41 +0*37 +0*38 3-0.38 +0*97 - 0 0 0.00 -0.39 -0.17 + 0.24 +0.20 +0.21 +0.21 +0-80 Ue 0.00 -0.27 -0.17 +O*M + 0.23 +0*23 +0-27 $0.78 aR. W. Taft ‘Steric ERects in Organic Chemistry’ ed. M. S. Newman Wiley New York 1956 chap. 13; bD. H. McDaniel and H. C. Brown J. Org. Chem. 1958 23,420. Taft regards the values of (T* as agreeing with ‘the qualitative English school theory’ of the polar effect.ll He points out that U* and Es are quite different functions of structure e.g. the series Me Et Pri But shows an additive effect for the polar influence of successive methyl groups whereas the values of Es ‘telescope’.Es values in general conform well with qualitative assessment of ‘steric hindrance’ by means of molecular models. For @%unsaturated substitu- ents R however the Es value is often governed mainly by the conjugation of R 438 Shorter with C02R1 in the initial state and is thus not a measure of the steric effect (see last two entries in Table 1). When Taft’s u0* values are changed to a o0 scale with hydrogen as the standard by assuming a. =ap for Me it is seen that in various other cases u0 N u p . This apparently means that the polar effects of substituents operate equally from the ortho- and para-position but as a general rule this seems unlikely. Values of o* are of importance in the analysis of a values in terms of 01 and (TR (see above).@ For a few substituents an inductive parameter (T’ was based on the reactions of 4-substituted-bicyclo[2,2,2]octane-l-carboxylic acids and The bicyclo-octane system was considered to provide a good model for the transmission of the non-mesomeric part of the polar effect through the benzene ring.It was found that the (T’ values for X were proportional to u* for CH2X.11 A new inductive parameter for substituents in general was therefore defined as 01 = 0*45(~*. B. Applications of the Taft Parameters to the More General Understanding of the Reactivities of Organic Compounds.-At this stage we discuss mainly the contributions of Taft and his colleagues Later we shall consider a variety of examples of the uses made of these parameters by other authors.(i) The linear free energy-polar energy relationsh@.ll Taft found that the rate or equilibrium constants for a wide variety of reactions of RY conformed respectively to equation ( 5 ) log (klk,) = o* p* or the corresponding equation for equilibrium constants where (T* is the polar substituent constant for R andp* is a reaction constant analogous to the Hammett p constant. A few reactions of ortho-substituted aromatic systems o-XC6H4.Y also obey these equations. Examples are given in the Figure. Conformity to these equations clearly implies that all effects other than polar remain nearly constant throughout each reaction series. Notably this means either the complete absence of steric effects of substituents or that they are approximately the same as the steric effect of a methyl group within the range considered.The predictive power of these equations is thus limited deviations may occur if substituents markedly different from those involved in the initial correlation are considered. For many of the reactions the minor r81e of steric effects is readily understood some involve no change in bond angles at the reaction centre; in others the reaction centre is somewhat remote from the substituent or the reagent involved is very small. Deviations from equation (5) were used by Taft to assess other effects quanti- tatively. A good example of the assessment of steric effects is provided by the application of the modified equation A A H ~ = (C(T*)P* J. D. Roberts and W. T. Moreland J. Amer. Chem. Soc. 1953 75 2167. 439 The Separation of Polar Steric and Resonance Eflects in Organic Reactions -0.6 0 + I .O + 2.0 4- 3.0 0" Figure The divisions on the ordinate are 1.00 units of pK or log k apart. The relative positions of the lines A to D with respect to the ordinate are arbitrary. A. p K aliphatic carboxylic acids water 25 "C. (J. F. J. Dippy Chem. Rev. 1939 25 151.) B. p K ortho-substituted benzoic acids water 25 "C. Abscissa Tuft's u, values. Benzoic acid deviates markedly. ( J . F. J . Dippy Chem. Rev. 1939 25 151.) C. log k alkaline hydrolysis of [Co(NH,),.O,C*R]*' water 25 "C. ( J . Basolo J. G. Berg- mann and R. G. Pearson J . Phys. Chem. 1952,56,22.) D. log k catalysis of dehydration of acetaldehyde hydrate by R.CO,H aqueous acetone 25 "C. [R. P. Bell and W. C. E. Higginson Proc. Roy. SOC. ( A ) 1949,197 141.1 to the enthalpies of dissociationdHd of the addition compounds formed between boron trimethyl and aliphatic amines R1R2R3N.14 The term Ca* is the sum of the a* values for the groups R1 R2 and R3 so that effectively the parent system is that involving NM%.Ammoniaand straight-chain primary amines conformed to equation (6) but branched-chain compounds and secondary or tertiary amines showed marked deviations. Thus ButNH2 deviated by - 6.6 kcal mol-l (from calculated AH^ = 19-6) which was attributed to steric strain in the complex. The acid-catalysed hydrolysis of diethyl acetals of general formula R1R2C(OEt)2 (R1 = H or Me; R2 variable) indicated the importance of resonance effects.15 Application of equation (5) with Ca* for R1 and R2 to the hydrolysis of compounds derived from non-conjugated aldehydes and ketones gave well- separated parallel straight lines for the two classes of compound with a number of deviant points.A single straight-line however was given by equation (7) H. C. Brown and G. K. Barbaras J. Amer. Chem. SOC. 1953,75 6 and earlier papers. l 5 M M. Kreevoy and R. W. Taft J. Amer. Chem. SOC. 1955,77 5590. 440 Shorter log (k/ko) = (co*)p* + (n - 6)h (7) where n is the number of a-hydrogen atoms in R1 and R2 i.e. (n - 6 ) is the decrease in the number of such atoms compared to the six in acetonal (R1 = R2 = Me) and h is a proportionality constant. The term (n - 6)h is interpreted as a contribution from hyperconjugation in stabilising the transition state [resembling an 0x0-carbonium ion R1R2C(OEt)+] relative to the initial state (which is saturated).Some compounds did not conform to equation (7) e.g. the acetal with R1 = Me and R2 = neopentyl shows clear signs of steric ac- celeration due to the bulky substituent. A number of other reaction series are correlated by equation (7).16 (ii) The Iinear free energy-steric energy relationship.ll Taft found that a number of reactions conformed to the equation log (klk,) = SES (8) where 6 is a steric susceptibility constant. In such reactions the polar effects of substituents must be very small. Several of the reactions are closely related to acid-catalysed ester hydrolysis e.g. the acid-catalysed hydrolysis of ortho- substituted benzamides but others are very different e.g. methyl iodide reacting with 2-monoalkyl pyridines. For examples of the latter type equation (8) tends to be of rather limited applicability; deviations occur when the range of sub- stituents is extended.(iii) The combined Iinear free energy-polar energy and steric energy relationship. Pavelich and Taft17 suggested that the equation log (klk,) = a*p* + SES (9) would be found to be of wider applicability than equations ( 5 ) or (8). It implies that the relative free energy of activation may be regarded as a sum of independent contributions from polar and steric effects. Equation (9) was found by Pavelich and Taft to correlate the results for acid-ls or ba~e-catalysedl~ methanolysis of ( -) -menthy1 esters RC02CloHls in methanol at 30 “C. Biechler and Taft lU found it also applied in a modified form to the basic hydrolysis of anilides. C. Further Analysis of Steric Effects in Ester Reactions.-By applying tran- sition state theory Taftll accomplished a separation of the overall steric effect of a substituent as measured by Es into ‘steric strain’ and ‘steric hin- drance of motions’.For an account of the procedures Taft’s articles should be consulted. From a series of more than twenty alkyl groups it appears that small groups give only small steric strain but the hindrance of motions is considerable. The l6 R. W. Taft and M. M. Kreevoy J. Amer. Chem. SOC. 1957,79,4011. W. A. Pavelich and R. W. Taft J . Amer. Chem. SOC. 1957,79,4935. W. A. Pavelich Thesis Pennsylvania State University 1955. l 9 S. S. Biechler and R. W. Taft J. Amer. Chem. SOC. 1957 79 4927. 441 5 The Separation of Polar Steric and Resonance Efects in Organic Reactions latter rapidly comes to a limit with increasing size and branching of the alkyl group while steric strain continues to increase.Steric hindrance of motions shows little solvent dependence suggesting that steric inhibition of solvation is not a factor contributing to this. Taft showed that the estimates of steric strain in M%B.NR1R2R3 addition compounds (see above) were related to the estimates of steric strain caused by R1R2R3C in ester reactions. The addition compound and the ester transition state show structural similarities so the relationship is reasonable. There is no relationship to Es. 4 Application of Taft's Polar and Steric Parameters The parameters have been used to correlate a wide range of phenomena in nature often far removed from the systems from which the parameters were derived.A selection of material is presented to show the great variety of applica- tions. A. Ionisation of Acids.-Taft correlated the strengths of aliphatic carboxylic acids through the appropriate form of equation (5). Chartoi120 has suggested that this system should be a basis for defining (TI values (01 = 0.450"); this is doubtless satisfactory for fairly small substituents. However for a series of arylaliphatic carboxylic acids with bulky substituents Bowden Chapman and Shorter,21 and Bowden and Young22 correlated the pK values by means of the Pavelich-Taft17 equation; a significant term in Es indicated steric inhibition of solvation of the carboxylate ion. The application ofa* to many series of non-carboxylic acids has been examined. These are of interest because the substituent is often attached to an element other than carbon.The largest amount of work has been done on the ionisation of the conjugate acids of nitrogen bases. For non-aromatic amines Hall23 applied equation ( 5 ) in the form (10) where p* = -3.14 -3.23 and -3.30 and pKo = 13-23 12.13 and 9.61 for primary secondary and tertiary amines respectively (in water). Tertiary amines conformed well ; certain secondary and primary amines showed deviations which were attributed to steric effects. The results indicated the importance of solvation of the alkylammonium ion through the N-H bonds and of steric inhibition of this. Folkers and Runquist2* reported that Hall's data conformed to one line given by equation (1 1) M. Charton J . Org. Chem. 1964 29 1222. K. Bowden N. B. Chapman and J. Shorter J. Chem.SOC. 1963 5239; 1964 3370. ** K. Bowden and R. C. Young Canad. J. Chem. 1969,47,2775. 23 H. K. Hall J. Amer. Chem. So.. 1957 79 5441. 24 E. Folkers and 0. Runquist J. Org. Chem. 1964 29 830. 442 Shorter where n is the number of hydrated NH groups in the alkylammonium ion. The 1.12 (n) term gives an acid-weakening effect of hydration of the NH groups. The above ideas have been developed further by C o n d ~ n ~ ~ with regard to statistical effects and to hydration both of free base and of ammonium ion. Henderson and Streuli2* have correlated the basicities of substituted phos- phines with Co*. Hallz7 found equation (5) of little relevance to the nucleophilic reactivity of amines but Henderson and Buckler2* were more successful with that of phosphines. B. Heterogeneous Catalysis.-Kra~s~~ has reviewed the recent development of linear free energy relationships for heterogeneous reactions.Values of (T or o* based on reactions in solution at room temperature or a little above have proved applicable to the rates of reaction of substrates on catalysts up to 500" C. This is very surprising since complex interactions between catalyst and substrate might well modify reactivity in a way quite unrelated to polar effects in sol- ution. The correlations involving o* are often good. For example results on the elimination of water from five alkanols over AlzO at 380 "C are well corre- lated by o*. However it is often necessary to exclude certain substituents from the correlation. But is prominent among these which suggests that steric effects must not be ignored. This has been emphasised by Mochida and Yoneda,,O and by Ruzicka and his colleague^,^^ who have used the Pavelich-Taft17 equa- tion (9).Work on linear free energy relationships in this field is hampered by the difficulty of using a wide range of substituents. C. Polymerisation.-Taft's polar and steric parameters have been found rel- evant to polymerisation. Otsu and his c011eagues~~ have copolymerised various alkyl methacrylates or acrylates M2 with styrene or #?-chloroethyl methacrylate M1. The relative reactivities l/rl of Ma towards attack by the polymer radicals were examined in terms of a Pavelich-Taft equati0n.l' The correlation with o* was good but there was no significant relationship to Es. On the other hand Chikanishi and T s ~ r a t a ~ ~ reported that for the attack of polystyryl radicals on methyl a-alkylacrylates there was significant correlation with Es but not o*.No doubt the steric effect here is due to the alkyl groups being directly attached to the reacting vinyl bond. For the same system however Cameron and claim that correlation with o* as well as Es is significant. 2s F. E. Condon J . Amer. Chem. SOC. 1965,87,4481 et seq. as W. A. Henderson and C. A. Streuli J. Amer. Chem. SOC. 1960 82 5791. 28 W. A. Henderson and S. A. Buckler J . Amer. Chem. SOC. 1960,82 5794. 2 9 M. Kraus Adv. Catalysis 1967 17 75. 3 0 I. Mochida and Y. Yoneda J . Catalysis 1968,11 183. 31 V. Ruzicka L. Cerveny and J. Pachta Coll. Czech. Chem. Comm. 1969,34 2074. 3 p T. Otsu T. Ito and M. Imoto J. Polymer Sci. Part C Polymer Symposia 1967 2121 and references therein. 33 K.Chikanishi and T. Tsurata Makromol. Chem. 1964,73 231. a4 G. G. Cameron and G. P. Ken European Polymer J. 1967 3 1. H. K. Hall J. Org. Chem. 1964 29 3539. 443 The Separation of Polar Steric and Resonance Efects in Organic Reactions D. Optical Spectra.-a* Values have been much used in the correlation of i.r. data and such correlations have been used to estimate new -a* values. 1.r. frequencies are related to force constants of bonds whereas intensities are governed by the rate of change of dipole moment with respect to bond length. Polar effects of substituents may influence both of these although sometimes one is influenced much more than the other. Frequencies v have been correlated with -a* while integrated intensities A have also been variously correlated in the forms log A A and All2.T. L. has argued that the use of AIP is theoretically sound. Earlier however,3s he had successfully correlated A for the OH stretching frequency of aliphatic alcohols R1R2R3C.0H with &*. Deviations were observed where internal hydrogen bonding appeared possible. Rao3’ showed that the OH stretching frequency in eleven carboxylic acids gave excellent correlation with u*. The CO stretching band has been much examined. For example O’Sullivan and Sadler38 found good correlation of v with -a* for R-CO-Me with R as a variety of heteroatom groups while Morcillo et aZ.39 have correlated A with u* for RCO-Ph. Various studies of aromatic compounds involve using Taft’s u0 values to place ortho-substituted compounds on the same plot as meta- and para-deriva- tives. Weigmann and Malewski40 do this for various bands in aromatic sulphonyl chlorides with mixed success.Correlation is quite good for the antisymmetrical stretching frequency of -SO2-. Correlations involving -a* values and U.V. spectra are much more limited. Polar effects operating in the formation of electronically-excited molecules will not necessarily be closely related to those influencing rate or equilibrium con- stants. However Gosavi and Rao41 found a fairly good linear relationship to ED* for vmax of the n -+T* CS band in thioureas R1R2N(CS)NR3R4. E. Nuclear Magnetic Resonance.-Various aspects of n.m.r. have been cor- related with -a* but the situations are rarely simple. Kan42 correlated the chemical shifts of the CH or CH protons of acetates or succinates respec- tively with u* values for the alkyl group of the alkoxy-substituent of the ester.The correlation was not very good deviations being particularly marked for bulky substituents. This behaviour was considered due to steric inhibition of resonance involving the canonical form X-CO =0-R. + - 35 T. L. Brown J. Phys. Chem. 1960,64 1798. a6T. L. Brown J. Amer. Chem. SOC. 1958,80,6489. 37 C. N. R. Rao and R. Venkataraghavan Canad. J. Chem. 1961,39 1757. 38 D. G. O’Sullivan and P. W. Sadler J. Chem. SOC. 1957 4144. Ser. B 1964,60 199. 4 0 H.-J. Weigmann and G. Malewski Spectrochim. Acra 1966 22 1045. 41 R. K. Gosavi and C. N. R. Rao Canad J. Chem. 1967,45 1897. 4 2 R. 0. Kan J. Amer. Chem. SOC.~ 1964,86 5180. J. Morcillo E. Gallego R. Madronero and A. R. Trabazo Anales de Quim. (Madrid) 444 Shorter Proton chemical shifts in aromatic systems have been related to cr values for meta- and para-substituents and to Taft’s uo values for ortho-substituents.This has been done for the -NH2 shifts in aniline.43 Alternatively such cor- relations have been made the basis of cro values e.g. Tribble and Traynham’s work on the OH shifts in phenols leading to cro- values said to be useful in correlating ca. thirty reaction series.44 Chemical shifts for 14N I9F and 31P coupling constants and nuclear quad- rupole resonance frequencies have also been related to cr*. F. Biological A~tivity.~~-Linear free energy relationships are increasingly used in the interpretation of biological activity. The problem is complex and certain special parameters have been developed e.g. Hansch’s ‘partition co- efficient’ factor n which measures the ‘hydrophobic bonding’ character of a drug a property which has received little attention in mechanistic chemistry.Polar and steric effects of substituents often play only a subsidiary r81e but some- times their part is important. In various systems different types of substituent constant have proved relevant including cr* and Es. Thus results for the esterase activity of human serum using a series of six p-nitrophenyl esters R-Co2C6H4NO2 are well correlated by n and Es with the addition of a term in cr* making a slight improvement. It is of interest that steric parameters are sometimes needed even when the substituent is remote from the functional group responsible for the drug activity. This is presumably connected with a steric effect on the interaction of the drug molecule as a whole with the bio- logical site.5 Further Consideration of the Steric Parameter Taftll recognised that even for substituents incapable of normal conjugation with a carboxylic function there might be a contribution to Es from a resonance interaction i.e. a hyperconjugative effect of a-hydrogen atoms. In Hanc~ck‘s~~ view this should be allowed for in the derivation of a ‘corrected steric substituent constant’ EsC in the equation ESc = Es - h(n - 3) (12) where h is a reaction constant for hyperconjugation and n is the number of a-hydrogen atoms. Quantum mechanical calculations by Kreevoy and Eyring were used as a basis for taking h as -0.306. Selected EsC values are shown in Table 3. H a n ~ o c k ~ ~ showed that the use of Esc values could lead to significant improvement in the correlation of certain reactions.The following relation- ships were found to hold for the saponification of nine esters RC02Me in 40 % aqueous dioxan at 35 “C 4 5 B. M. Lynch B. C. Macdonald and J. G. K. Webb Tetrahedron 1968,24 3595. 4 4 M. T. Tribble and J. G. Traynham J. Amer. Chem. SOC. 1969,91 379. 4 5 C. Hansch Accounts Chem. Res. 1969 2 232 and many references therein. 4 6 C. K. Hancock E. A. Meyers and B. J. Yager J. Amer. Chem. Soc. 1961,83 421 1. 445 The Sepuration of Polar Steric and Resonance Efects in Organic Reactions log k = 1.31 + 1*540* -t 0.709 Es log k = 1.36 f- 1.480* + 0.471 Esc log k = 1.25 + 1.750* + 0.848 Esc - 0.383 (n - 3) (13) (14) (15) For equation (1 3) the multiple correlation coefficient R and standard deviation s are 0-992 and 0.076 respectively i.e.the correlation is fairly good because hyperconjugation is involved in Es and in log k. Equation (14) is poor ( R = 0.970 and s = 0-149) because in Esc hyperconjugation has been eliminated. Equation (15) is excellent ( R = 0.998 s = 0.043) because it incorporates proper consideration of both steric and hyperconjugative effects. Table 3 7aft"s" Es and Hancock'sb Esc for R in R-CO2R1 R Es Esc Me 0.00 0.00 Et - 0.07 -0.38 Pri - 0.47 - 1 '08 But -1.54 - 2.46 Prn -0.36 -0.67 Bun -0.39 -0.70 Bui -0.93 -1.24 UTabIe 1; W. K. Hancock E. A. Meyers and B. J. Yager J. Amer. Chem. SOC. 1961 83 4211. Hancock4' has also shown the importance of 'change in the six-number'. N e ~ m a n ~ ~ showed that the number of atoms in position 6 from the carbonyl- oxygen as 1 makes a large contribution to the steric effect.When a given group is considered both as R and R1 in R-C02R1 there may be a change in the six- number 8 6 as between R and R1 e.g. for Me Et Pri and But 4 6 = 0 -3 -6 and -9 respectively. Hancock4' uses A6 as another structural parameter. For the saponification of nine acetates Me-C02R1 in'40% dioxan at 35 "C log k = 1.40 -+ 1.340* + 0.730 Esc (1 6) with R = 0.980 and s = 0.161 the Esc values being for the substituents a5 R. When 4 6 is included log k = 1-35 + 0,6880" + 0.664 E S C + 0.0477LI6 (17) with R = 0.997 and s = 0.070 which is a much improved correlation. Hancock's views do not appear to have achieved the notice they deserve. Bowden Chapman and Shorter2I regarded the improved correlations with some scepticism. P. D. B ~ l t o n ~ ~ however regards EsC and (n - 3) as relevant to 4 7 C.K. Hancock B. J. Yager C . P. Falls and J. 0. Schreck J . Amer. Chem. SOC. 1963 85 1297. 48 M. S. Newman J. Amer. Chem. SOC. 1950 72 4783. 4 9 P. D. Bolton Austral. J . Chem. 1966,19 1013. 446 Shorter the acid- or base-catalysed hydrolysis of aliphatic amides and has obtained significantly better correlations than with Es. Modified steric parameters have also been developed by Palm,50 who considers the contribution of both C-H and C-C hyperconjugation to Es. Equation (18) is proposed where ~ E I is the number of a-C-H bonds and nc is the number of a-C-C bonds. Eso is described as a ‘purely steric constant’ and linked with proper consideration of hyperconjugation has been much used by Palm and by other authors in the U.S.S.R.50 Eso = Es + 0-33 ( n ~ - 3) + 0.13 nc (18) 6 The Significance of (T* particularly for Alkyl Groups There seems little doubt that o* values measure the polar effects of substituents when these are substantial but the significance of small (T* values has been questioned.This applies to the values for all alkyl groups which lie mainly between 0 and -0.3. Such small values might arise from an imperfect cancelling of steric effects in the Taft analysis (see below). While Taftll asserted that (T* and Es are completely different functions of structure this is not strictly true for alkyl groups Koppe151 has shown that for twenty primary secondary or tertiary alkyl groups (o* = 0 to -0.4; Es - 0 to -4) with R = 0.956 and s = 0.292 Es = 0.88 + 27.78~* - 1.90 (n - 3) (19) where n is the number of a-C-H bonds.For electronegative substituents no such relationship was applicable. Ritchies2 holds that o* values for alkyl groups are not consistent with those for other substituents since different damping factors for interposing methylene groups are required for the two classes. By symmetry considerations developed for the Hammett equation by Hine Ritchie shows that the damping factor should be the same for all substituents. He argues that (T* values of alkyl groups do not really measure the polar effects of the groups and that o* is properly zero for all alkyl groups. Ritchie claims that data for various systems conforming to equation (5) are as well correlated by taking o* = 0 for all alkyl groups as by using Taft’s values although this is not properly substantiated by statistical procedures.He also concludes that the hyperconjugative interpretation of acetal hydrolysis (see earlier)ls is fallacious. The Taft (T* values for alkyl groups are considered to arise from interaction between the various types of substituent effects in ester hydrolysis i.e. Taft’s assumption (1) is erroneous cross-terms making an appreciable contribution. Most physical organic chemists however continue to believe that the electron- releasing properties of alkyl groups in aliphatic systems increase with chain- s O Numerous papers are in the issues of Reaktsionnaya sposobnost organicheskikh Soedinenii (with English summaries) ed. V. A. Palm commencing 1964 and available in English translation as Organic Reactivity (Consultants Bureau) from 1966. s1 I. A. Koppel Reakts.spos. org. Soedinenii 1965,2 (2) 26. 6 2 C. D. Ritchie and W. F . Sager Progr. Phys. Org. Chem. 1964,2,323 and references therein. 447 The Separation of Polar Steric and Resonance Eflects in Organic Reactions length and -branching and to use o* values as a measure of this. It may be statistically satisfactory to submerge the alkyl groups as a cluster of points at o* = 0 in a sea of highly polar substituents but this ignores small but weil- established structural effects which agree with the o* values. It is not clear how these effects are to be explained if o* = 0 for all alkyl groups. Palm53 was also aware of the problem posed by the damping factors but concludes that this indicates the operation of different electrical effects hydrogen and alkyl groups exert their influence by the through-bonds inductive effect while the field effect is more important for highly polar substituents.Thus the o* values for the two classes of substituent may really be on different scales. MaiS4 has shown that u* values may be correlated with ionisation potentials and electron affinities of substituents but alkyl groups and highly polar sub- stituents require different forms of relationship. T. L. BrownS5 has discussed the possible r6le of polarisability effects in the o* values of alkyl groups. In the transition state for alkaline ester hydrolysis dispersion forces between R and the remainder of the system will stabilise the transition state to a greater extent than in that for acidic ester hydrolysis. o* is thus a resultant of polarisability and inductive effects acting in opposite directions.The values are often applied however to systems in which these effects act in the same direction. This may be the case in the hydrogenation of aldehydes and ketones for which Taft and Kreevoy’slS separation of hyper- conjugative effects through applying equation ( 5 ) is therefore probably fallacious. Much of the literature merits re-examination in the light of the ideas summarised above. 7 The Ortho-Effect. The Work of M. Chart~n~~ Steric phenomena have long been recognised as playing a major part in the peculiar effects of ortho-substituents.l Primary steric effects of various kinds including steric hindrance to solvation or to the approach of the reagent and secondary steric effects have been invoked. In certain systems hydrogen-bonding and other intramolecular interactions have been postulated.The main approach to understanding the ortho-effect has been the attempt to separate steric effects from polar and other effects; the application of Taft’s procedure to aromatic ester hydrolysis and the use of the steric and polar parameters thus derived is the best known of these attempts. Some authors have approached the problem in a different way. Whereas Taft’s analysis estimates the steric effect (Es) and then eliminates it from a reaction in which both steric and polar effects are known to occur other authors have selected a reaction believed to be free from the steric effects of ortho- ssV. A. Palm Rum. Chem. Rev. 1961,30 471. w L. A. Mai Organic Reactivity 1967 4 220. 5 5 T. L. Brown J . Amer. Cliem. Soc. 1959 81 3229.5 6 M. Charton J. Amer. Chem. Soc. 1969 91 624 6649 and papers referred to therein. Professor Charton has also written a summary article to appear in Prog. Phys. Org. Chem. 448 Shorter substituents and have derived a scale of uo values by assuming po = p for this reaction. The u0 values have then been used to interpret other reactions. For instance McDaniel and Brown5’ believed that the pKa values of 2- substituted-pyridinium ions provided a basis for uo values. Usually however systems in which the reaction centre is somewhat removed from the ring have been chosen e.g. the dissociation of phenylpropiolic acids. uo Values have also been based on spectroscopic data. Charton has compiled available data.5s There is often poor agreement between the various values determined for a given substituent.This has prompted a search for the ‘true’ ortho-substituent constant. Several authors have claimed peculiar virtues for their favourite reactions e.g. G. G. Smith5* argues that interactions with solvent are responsible for the difficulty of determining satisfactory a. values. Hence ‘true’ uo values should be determined in the absence of solvent. Thus gas-phase ester pyrolysis much studied by Smith and his The value to be taken of po for the standard reaction presents a problem. Each author takes po = p for his favourite reaction but when the derived uo values are applied to other reactions po is not in general equal to p. The Reviewer has recently stated that ‘the complexity of the influence of ortho-substituents on reactivity may make the search for a single generally applicable scale of a.values quite fruitless’.a Charton has expressed the same view and has marshalled much evidence in support of it.56 Charton has compiled and analysed a vast amount of data on the ortho-effect. This work merits much closer examination than can be given here; some of his findings and conclusions are very remarkable. Charton’s work is based on the separation of the electronic effect of a substituent X into inductive (non-meso- meric) and resonance (mesomeric) contributions through the equation is the ideal reaction. The factors contributing to Es values are examined through the equation E s x = am,x + / ~ R x + $ w x + h (21) A and 8 or a and p are constants defining the relative importance of inductive and resonance effects. r v x is related to the size of the substituent.Depending on the shape of the substituent YV,X values are assessed from van der Waals radii in various ways. $ gives the contribution of steric effects thus specified; h is a constant. For substituents in aliphatic systems the contribution of the a r ~ and UR,X terms was insignificant and ES,x could be regarded largely as a function of substituent size i.e. Es is indeed a steric parameter. For the Es,x values of ortho-substituents however the UI,X and UR,X terms proved significant while the rv,x term did not i.e. the Es values of ortho- substituents have little to do with steric effects. 5i D. H. McDaniel and H. C. Brown J. Amer. Chem. SOC. 1955 77 3756. 5 8 D. A. K. Jones and G. G. Smith J. Org. Chem. 1964 29 3531 and earlier papers. 449 The Separation of Polar Steric and Resonance Efects in Organic Reactions The approach is extended to the analysis of a vast amount of data.In (21) ES,X is replaced by Qx the value of an observed property under the influence of ortho-X. Q may be log k log K v etc. The rv,x term generally makes little contribution. Charton concludes that steric effects play a minor r6le in the ortho-effect. Different contributions of resonance and inductive effects i.e. /$a account mainly for the variety of phenomena. Charton’s work is undoubtedly a valuable contribution but some of his conclusions are so much at variance with long held ideas that they merit searching examination. His data sets frequently lack substituents whose steric effect is expected to be substantial. Often the necessary items are not available but some- times bulky groups e.g.But or I are admitted not to conform. It is possible that exclusion of data relating to bulky groups may distort the picture for the less bulky substituents appearing in the correlation. Another awkward matter is the frequent need to exclude the parent compound as showing a marked deviation. The correlation of data made possible by computers is valuable but one must not lose sight of the chemistry in a welter of statistics. Real effects can be obscured by good correlation coefficients and high confidence levels. 8 Critique of the Taft Analysis The earlier parts of this Review have shown that the Taft analysis achieved considerable empirical success and led to a deeper understanding of the influence of structure upon reactivity in a wide variety of processes.Nevertheless as Sections 5 6 and 7 have shown there are questions about the significance of the polar and steric parameters. The present Section continues this examination with particular reference to the fundamental assumptions of Taft’s procedure. In several papers Chapman Shorter and their colleagues have made extensive use of the Taft analysis and its parameters while remaining alert to possible weaknesses. They have drawn attention to the paucity of data relating to acidic and basic ester hydrolysis in the same solvent. Attempts to remedy this met with only limited success at temperatures necessary for studying acidic hydrolysis aqueous organic solvents often ‘crack’. It is much easier to study acid-catalysed esterification of carboxylic acids. In studying conformation and reactivity Chapman Shorter and Toynesg obtained results which cast doubt on Taft’s assumption (2) regarding the equality of steric effects in acid- and base-catalysed reactions.The use of a But group or a (CH,) bridge (trans-decalin system) to lock the conformation of cyclohexane compounds enables the reactivity of a functional group to be studied in both the equatorial and the axial disposition. The steric interactions in the two configurations are different and the procedure provides a means of changing steric effects without changing polar effects. For methyl trans- and cis-4-t-butylcyclohexane carboxylates in 1 :3 dioxan-water at 90 “C keq/kaz N 4.8 in acidic hydrolysis and N 8.3 in alkaline hydrolysis indicating a difference in N. B. Chapman J. Shorter and K. J. Toyne J .Chenr. SOC. 1961 2543; N. B. Chapman A. Ehsan J. Shorter and K. J. Toyne J. Chein. SOC. ( B ) 1967 570; 1968 178. 450 Shorter steric effects in the two reactions. In acid-catalysed esterification in methanol at 3 0 O C trans-decaliii-2fl- and 2a-carboxyIic acids gave kea/kas N 25. Alkaline hydrolysis of the corresponding methyl esters gave ratios between 18 and 22 in various mixtures of water with methanol or dioxan. It appears that Taft’s assumption (2) may be more closely fulfilled as between alkaline hydrolysis in aqueous organic solvents and acid-catalysed esterification in methanol. Chapman Shorter and UtleyGo studied acidic and basic hydrolysis of methyl ortho-substituted benzoates in the same aqueous organic solvents. Various features of this and related work led them to criticise Taft’s assumption (2) for neglecting the rble of the solvent since the transition states of the acidic and basic reactions carry opposite charges it is unlikely that the solvation patterns will be so similar that the steric interactions in the two system will be the same.By considering the thermodynamics of ions in solution R. M. NoyesG1 concluded that there are extreme differences in solvent structure around cations and anions and warned that ‘Taft’s treatment may not have separated steric effects as cleanly as was hoped.’ Chapman Rodgers and ShorterG2 studied the acid-catalysed esterification of ortho-substituted benzoic or phenylacetic acids in methanol. Their results in conjunction with those on related systems suggested a dependence of steric effects on solvent and the occasional importance of a polar contribution to log ( k / l i O ) ~ .Bowden Chapman and ShorteP examined the Taft analysis and the application of polar and steric parameters in the reactions of arylaliphatic carboxylic acids and esters. A Taft analysis of alkaline hydrolysis led to o* values with inconsistencies and not agreeing well with a set based on the reaction of carboxylic acids with diazodiphenylmethane. Criticism of Taft’s assumption (2) was repeated. A rather different approach to examining the validity of assumption (2) is to study the application of polar and steric parameters to reactions bearing a close resemblance to acid- and base-catalysed ester hydrolysis. Ester exchange provides a very suitable system,e.g. the methanolysis of( -) -menthy1 esters RC02C10H19 in methanol at 30 “C studied by Pavelich andTaft.l7n1* Equation (9) was applied to givepa* = 0.626,s~ = 1-549 for the acid-catalysed reaction andpB* = 2.702 6g = 1.301 for the base-catalysed process i.e.the steric susceptibility constants differ by 18%. Pavelichl* attributes this to the steric effects ceasing to be equi- valent when they are considerably greater than those operating in ester hydro- lysis (6 = 1). Howsoever interpreted the results are disturbing to assumption (2). Note also the appreciable polar effect in acid-catalysed exchange. P. D. B01ton~~ has studied the acidic and basic hydrolysis of aliphatic amides whose mechanisms closely resemble those of the corresponding ester reactions. The results are best interpreted in terms of a Pavelich-Taft equation (9) modified to use EpC values46 and to include a hyperconjugative term (see above).The 6 o N. B. Chapman J. Shorter and J. H. P. Utley J . Chern. SOC. 1963 1291. 61 R. M. Noyes J. Amer. Chem. SOC. 1964,86 971. 62 N. B. Chapman M. G. Rodgers and J. Shorter J Chem. SOC. (B) 1968 157 164. 45 1 The Separntion of Polar Steric and Resonance Efects in Organic Reaciions steric susceptibility constant 8~ = 0458 and 8~ = 1.08 i.e. the values differ but in the opposite sense from that found in ester exchange. On the other hand support for assumption (2) has come from a variant of this approach. I. V. T a l ~ i k ~ ~ interpreted acidic and basic ester hydrolysis in water in terms of u* and Eso values (based on the usual measurements in aqueous organic solvents) with appropriate hyperconjugative terms.She finds 8~ N 8~ - 0-80. The situation regarding the supposed equality of steric effects in acid- and base-catalysed reactions is thus obscure and deserves attention. There is room for well-designed experimental work. Taft’s other assumptions likewise require investigation particularly assumption (3) which is the basis for ignoring polar effects in the acid-catalysed reactions. A widespread disbelief in assumption (1) regarding the approximate separation of polar steric and resonance effects in a linear way would stultify the field. However it would be appropriate to investi- gate possible contributions from interaction terms in certain systems. The inclusion of cross-terms in correlations frequently improves them but often their physical significance is obscure. Taft’s papers in 1952 and 1953 represented a great achievement in dealing with a vast body of data.Our continued dependence on the consequences of the averaging procedures which he found necessary is unsatisfactory particularly since some of the data are disparate and of doubtful reliability. Further new values of U* and Es do not have this character they are based on the behaviour of individual systems e.g. a new Es value is derived from acid-catalysed esterifica- tion in methanol or a new a* value by interpolation in a spectroscopic cor- relation. There is too little attention paid by many organic chemists to the considerations in Sections 5 and 6 and there is much rather woolly application of Es and u* with insufficient attention to an adequately wide range of sub- stituents (a point emphasised by Taftll) and to proper statistical assessment of correlations.In general the Taft analysis its equations and parameters are in need of the kind of refinement to which the Hammett equation was subjected in the late 1 9 5 0 ~ . ~ # ~ APPENDIX Table 4 Substituenf constants Symbol a+ a- @ UO a 01 ORo Nature References Original Hammett constant 5 6 Constant for electron-demanding reactions (Brown) 7 Constant for electron-releasing reactions 5 6 64 Unexalted Hammet t constant (Wepster) 8 Unenhanced Hammett constant (Taft) 9 Inductive and resonance parameters used for analysis of U0 9 Resonance parameter (reaction-dependent) 9 63 I. V. Talvik Reakts. spos. org. Soedinenii 1964,l (2) 241. 64 R. W. Taft J . Amer. Chem. SOC. 1957 79 1045. 452 Shorrer Symbol u* uo* 0 0 u’ Es Esc Es O n A6 7T Nature Aliphatic polar substituent constant Polar constant for ortho-substituents (Taft) relative to methyl Polar constant for or tho-subst i tuent s (general) Polar constant for alicyclic reactivity (Roberts) Steric parameter (Taft) Steric parameter corrected for C-H hyperconjugation (Hancock) Steric parameter corrected for C-H and C-C Number of relevant bonds in correlations allowing for Change in the six-number (additional steric parameter Hydrophobic bonding constant in biological activity hyperconj uga ti on (Palm) hyperconjugation (subscripts may be attached) Hancock) (Hansch) References 11 11 56 13 11 46 50 15 46 50 47 45 I thank Professor R.W. Taft for the loan of a thesis Professor M. Charton for the opportunity of reading an article prior to publication and Professor N. B. Chapman for valuable comment on the manuscript of this Review. 453
ISSN:0009-2681
DOI:10.1039/QR9702400433
出版商:RSC
年代:1970
数据来源: RSC
|
8. |
Application of computers in chemical analysis: amino-acid analysis and sequence determination |
|
Quarterly Reviews, Chemical Society,
Volume 24,
Issue 3,
1970,
Page 454-471
B. Sheldrick,
Preview
|
PDF (1266KB)
|
|
摘要:
Application of Computers in Chemical Analysis Amino-acid Analysis and Sequence Determination By B. Sheldrick ASTBURY DEPARTMENT OF BIOPHYSICS THE UNIVERSITY OF LEEDS LEEDS 2 The widespread introduction of electronic digital computers in recent years has enabled chemists to perform two types of calculation which were previously too complicated i.e. those which involve complicated mathematical treatment of data and those which involve large numbers of comparisons or sorting move- ments as in literature searches. The second type of facility is useful when dealing with sequence analysis of chemical compounds which includes the sequence analysis of proteins or polypeptides when these are not analysed step-by-step but are investigated by examination of the fragments produced by random hydrolysis and the establishment of the sequence in a compound by examination of its mass spectrum.Both these types of sequence analysis have much in com- mon as they involve a considerable amount of sorting and comparison but without any complicated mathematical requirements. The advantage of the electronic computer in this field lies in its ability to sort and compare in a thorough and systematic manner the large number of combinations that can occur and which increases very rapidly with increase of the molecular weight of the compound under examination. For convenience we shall also consider the mass spectrometry of mixtures. 1 Automatic Amino-acid Analysis The introduction of automation to the anaiysis of proteins and polypeptides has progressed in a number of discrete steps. First was the introduction of the chromatography column with automatic sampling of the eluent.The sampling was then automated further so that the addition of ninhydrin reagent and the spectrophotometry of the samples resulted in the output of the results as a graphical display on a recorder chart the chart showing a plot of absorbance against tinie (or consecutive samples). From this point on the digital computer entered the picture in a variety of ways and some of the possibilities are shown in Figure 1. In the left-hand column is shown the method adopted without digital com- putation; the peaks corresponding to the various amino-acids on the chart are measured usually by measuring the height of the peak and the number of recorder points marked above a line drawn at the half height of the peak (the accuracy of these measurements is the limiting factor of the accuracy of the final results).The quantities of the amino-acids (c) are then calculated from the relation 454 Sheldrick Analog integrator Recorder Chroma t ogra pli y column Digital voltmeter 11 1 1 Hand measurement c = no. of dots x peak height x calibration constant where the calibration constant is given by const = The value of the calibration constant is found from a run which contains an internal standard (norleucine). The simplest method of applying a computer to this procedure is merely to carry out the calculations on the data measured by hand. An ALGOL program norleucine equivalent x norleucine concentration area of norleucine peak L Punched tape 455 I 1 f I 'I + Computer Hand calculatioii Application of Computers in Chemical Analysis put forward by Graham and Sheldrickl (plus correction) requires the measured data to be typed by hand into a remote terminal,* which is then used by the program permanently stored in the computer to produce values for each amino-acid of pmoles residues/1000 residues ,ug nitrogen and percentage total nitrogen plus a value for the total nitrogen which are output at the terminal.Not all the final sets of values are required but since results are sometimes quoted in the literature as pg nitrogen sometimes as residues/1000 etc. a small amount of additional computation produces the complete set of data which is suitable for direct comparison with other published results no matter what system is used. This procedure is reasonable as most computers will carry out calculations much faster than data are input i.e.once the data are read into the computer it is advisable to carry out all the simple calculations which may be required rather than have another program which handles the same set of data. The program just mentioned suffers the disadvantage that human error is not eliminated from the data handling as errors can and do occur in measuring the peaks and typing the data. Alternative methods have been developed to overcome this which involve direct production of a paper tape (or magnetic tape) by the chromatography equipment. Two methods are available one in which the peaks produced are integrated by an analog integrator to produce a peak area which is then punched automatically on to paper tape; a second method uses a digital voltmeter to punch each reading on to a paper tape.Obviously different programs are required to compute the results as the second tape will not have the individual peak areas but will require procedures which recognise a peak and integrate it before carrying out the final calculations. An example of the fist type of program has been published by Starbuck et aL2 This program written in FORTRAN was designed to accept information provided either by the hand calculation method or by an integrator either method giving the required information of the peak area. Additional refinements in this program allow previous standard runs to be averaged in with the run being calculated if required correction of the results for specified loss or destruc- tion of particular amino-acids allowance for ammonia production and calcula- tion of the number of amino-acid residues in one molecule of the protein (the molecular weight of the protein is supplied as part of the input data).An addi- tional program by the same authors calculates the empirical composition of peptides obtained by the proteolytic digestion of proteins. The completely digitised system may be considered in two parts the automa- tion of the analyser to produce a digital output and program techniques required to handle the data in this form. Discussions of various forms of digitiser have been given by Yonda et aZ,3 Porter and Talley,* Jones and S~ence,~ Krichevsky * Remoteterminal adeviceforinput andoutput ofdataconnected toacomputer bya direct link. G. N. Graham and B. Sheldrick Biochem. J . 1965,96 517.W. C. Starbuck C. M. Mauritzen C. McClimans and H. Busch Anafyt. Biochem. 1967 20 439. 3A. Yonda D. L. Filmer H. Pate N. Alonzo and C. H. W. Hirs Analyt. Biochem. 1965 10 53. W. L. Porter and E. A. Talley Anafyt. Chem. 1964 36 1692. H. J. Jones and D. W. Spence Infotronics Application Notes No. 1 1964. 456 Sheldr ick Schwartz and Mage,6 and Cavins and Friedman.' These vary according to the type of computer with which they are to be used but basically consist of a coding system which converts the voltage produced by the spectrophotometer into a set of digital values on a paper tape or magnetic tape with suitable arrangements to indicate the termination of a value. The consecutive values are assumed to be produced at equal intervals of time and hence need not be indexed as a simple count from the first value is sufficient identification.Since every measurement is recorded it is unnecessary to assume as is done in the height-width calculation that the peaks approximate to a Gaussian distribution and a peak area is obtained by summation of all the observed values within it. Usually a recorder trace is produced simultaneously and filed as this is much more meaningful to observe than a series of numbers if the output is queried. Using a string of digital numbers in place of a string of peak areas (or equivalent data) introduces difficul- ties which the computer program must be designed to overcome. These are discussed by Jones and Spence5 for an on-line? integrator and some of the items which have to be dealt with are (i) recognition or detection of the start and end of a peak; (ii) location of peak maximum; (iii) integration of peak area; (iv) variation of base-line value ; (v) resolution of overlapping peaks.Each item can be dealt with separately by an appropriate sub-routine but in a complete program there must necessarily be considerable overlapping of these sub-routines and care must be taken to ensure that e.g. two peaks which over- lap are not mistaken by the program for two peaks with a peculiar base-line shift. Consider first of all detection of the three required points of a single peak the beginning themaximum and the end. If we store say six consecutive numbers of the output string we can readily test these for characteristics which we define for the specific points e.g. for the beginning of a peak we may specify that i.e.that each successive value is greater than the previous value. Alternatively we may specify that Xt+2 - Xt+l i.e. that the differences increase. Such a test is answered only by yes or no. If yes then we have the beginning of a peak if no then each of the six values in store is shifted one to the left and another value read in from the data string to fill the empty sixth place; and the test repeated. This is equivalent to sliding a window along the string of numbers and checking at each position for the beginning of a peak. Variations can be introduced for the window size and the tests applied while a similar system with different tests may be used to find the end of the peak. During the search for the end of the peak the intermediate values are stored and when the process is complete these are then scanned for the maximum value i.e.the centre of the peak which can be used for identification of the position of the peak in time. t On-line directly connected. ' J. F. Cavins and M. Friedman Cereal Chem. 1968,45 172. X t f l > Xt X t + l - Xt M. I. Krichevsky J. Schwartz and M. Mage Analyr. Biochem. 1965 12 94. 457 Application of Con lputers in Chemical Analysis Before the peak area is calculated however the two other tests for base-line drift and for presence of overlapping peaks have to be carried out. A simple method of distinguishing between these two cases is to examine the next few values after the end of the peak; if the values are higher than at the beginning of the peak then they are tested to see whether or not they fit the criteria used to test for the beginning of a peak.If they do not then there is a base-line drift which must be allowed for in the calculation of the peak area and if they do then another sub-routine must be entered to decide how the areas of the two peaks are to be established. Another simple test is to check if the peak maximum lies about half-way between the beginning and end of the peak; if not then either the peak is of peculiar shape or overlapping is taking place. In amino-acid analysis it is unlikely that overlapping peaks will occur as the conditions of the analysis are usually adjusted to ensure that full resolution of all peaks over the range is obtained. The calculation of the buffer pH values in the varigrad* has been described by Burns Curtis and Kacser.8 Exceptions can however occur and various methods of resolution are available depending on how the overlapping occurs.The simplest form is when two symmetrical peaks of equal height overlap. Areas of the peaks can be calculated by drawing a vertical line from the lowest point between the maxima to the horizontal axis. The area defined by the horizontal axis the curve from the beginning of the peak to the central minimum and the vertical is the area of the first peak since just as much of the second peak is enclosed in this area as is cut off the first peak by the vertical line. Obviously if the peaks differ in height or are not symmetrical but skewed then this method becomes less accurate. A discussion of the resolution of unequal peaks has been given by Fraser and Suzukis where the peaks are analysed into components by using either a Fourier method or by Cauchy Functions.7 Though in practice the peaks obtained are not accurately sym- metrical the accuracy lost by assuming that they are will be quite small.An iterative procedure for decomposing the two peaks consists of calculating the theoretical constants for one peak subtracting the calculated values from the observed values and then calculating the theoretical constants for the second peak. The values obtained for the second peak are then subtracted and the first peak corrected. This process is then terminated when the corrections involved fall below defined limits. Finally the block diagram for a computer program which carries out the above tests checks and calculations is shown in Figure 2. The only portion not already discussed is the test for the end of the data.This requireseither a special number at the end of the data which is recognised by the program or a count of the number of data entered at the beginning of the data. * Varigrad a multichamber device used to produce a controlled variation of pH and/or salt concentration. t Cauchy Function a function of the type y = a/(l + [2(x - b>/cJ2}. J. A. Burns C. F. Curtis and H. Kacser J . Chromatog. 1965 20 310. R. D. B. Fraser and E. Suzuki Analyt. Chem. 1966,38 1770. 458 Sheldr ick L - Input T 1 Shift Check for peak beginning of 1 window Check for end A 1 of peak Q- begins? Shift window and move Subroutine for handling over- lapping peaks J 1 Check peak end I for 2nd peak 459 Application of Computers in Chemical Analysis The former is more reliable as it will always be the same number and if omitted by an operator will not invalidate the calculations.In the paper by Krichevsky Schwartz and Mage6 details of a program similar to the above are discussed. An additional refinement introduced is a check and correction for ‘noise’ (Le. spurious values caused by random fluctuations in the electronic section of the apparatus) with special precautions taken when testing points near the maximum of a peak. One difficulty arises if the values are pro- duced on paper tape and if this has then to be transferred to card format before the data can be read by the computer. It would perhaps help to avoid difficulties of this nature if all digital information could be recorded on magnetic tape to avoid the necessity of format changes.One factor which will have to be decided if this is ever to become a standard system is whether the data are recorded on the magnetic tape as an analog signal which is later converted to a digital signal by the computer or whether the data are initially recorded in a digital form on the magnetic tape. The latter has the advantage that an additional translation at the computer is avoided but requires additional equipment in the form of an A-D convertor to be on-line with the chemical equipment. Chemical methods of establishing the amino-acid sequence of a protein have been carried out in many cases on small quantities of material with consider- able success. Due to the difficulties of the chemical processes and the time involved attention has been directed towards alternative methods.One technique consists of hydrolysing the protein into fragments analysing the fragments and then calculating the unique chain which could produce the observed fragments. The process is not as fundamental as stepwise analysis and various factors have to be considered. 1. The fragments produced by the hydrolysis must be of reasonable length to allow information about the chain to be established e.g. total hydrolysis into single peptides would provide no sequential information at all. 2. The points on the chain at which breaks occur should be random and not systematic e.g. if the chain A.B.C. always breaks into A B and C even if the separate parts undergo random hydrolysis it would be impossible to distinguish between the chains A.B.C. and A.C.B. or B.A.C. Bernhard Bradley and Dudalo suggested a program which applies a set of rules in stages to the input data to produce the maximum amount of sequence information which can be established.They assumed that for each fragment produced by random splitting of the chain the total composition is known and also the identity of the N-terminal amino-acid e.g. a typical fragment could be represented as -Thr (Ala Pro Lys) where the tetrapeptide contains the specified four amino-acids; in this example it is known that threonine is the N-terminal group but the sequence of the other three is unknown. An alternative method of input codes the amino-acids as prime numbers (Sheldrick’l) with a separation between the items of known sequence e.g. in lo S. A. Bernhard D. F. Bradley and W. L. Duda IBM Journal 1963 246. l1 B.Sheldrick Biochem. J . 1966,100 llc. 460 Sheldrick the foliowing example if the coding is Pro 7 ; Gly 2 ; His 53; Ala 3; Val 11 then the information Pro (Gly Ala Val His) would be represented by 7;3498 where the prime number 7 represents the N-terminal amino-acid proline and the integer 3498 is the product of the prime numbers 2 x 3 x 11 x 53. This system has two advantages; all the necessary information is stored in one loca- tion limited only by the largest integer which can be stored in the computer and a test of the data for the presence of a specific amino-acid can be carried out by a single operation i.e. the product is divided by the appropriate prime number and the answer tested for the presence of any remainder; no remainder indicates that the amino-acid is present.Incomplete sequence data are repre- sented by the two systems as Pro (Gly Val His) Ala and 7 ; 1166; 3. Similarly Pro Val Gly (His Ala) = 7 ; 11 ; 2; 159. The process of elucidating the sequence of the amino-acid residues in the original protein from such data has been discussed in various papers. Bernhard Bradley and DudalO gave some preliminary results using artificially simulated fragments from a known sequence of insulin and showed how the process could be carried out in four stages 1. The fragments are sorted into groups each of which has a common known terminal acid. 2. Each group is broken down into sub-groups the number of such sub-groups being limited to the number of times the N-terminal acid occurs in the original chain. Each sub-group contains fragments which have at least one amino-acid in common in addition to the N-terminal group and within each sub-group comparisons are used to contrast the information to produce a single sequence which includes all the information available in that sub-group e.g.two frag- ments 3;35 and 3;70 would be combined to give 3;35;2. 3. All the contracted fragments are re-listed and replace the original fragments. 4. The sorting process used in 2 is now repeated to merge the contracted frag- ments into one sequence. The example given in this paper provided the sequence of a chain of twelve units from a collection of eighteen fragments some of the fragments providing superfluous information. The problem of calculating the minimum number of fragments which is necessary and sufficient to provide an unambiguous sequence for the original chain has not been solved for the general case.It is affected by the following factors (i) hydrolysis may not be truly random; (ii) the number of fragments required may vary according to the arrange- ment of the amino-acid residues in the original chain-an example given in the above paper deals with the difficulty of determining the sequence 5 ;7;5;7;5 ;7 ; (iii) variation of the size of the fragments will have an effect-e.g. if all the 461 Application of Coriiputers iii CheinicaI AnnIysis fragments are only two residues long each occurrence of a duplicate of one of these pairs in the original chain will produce an ambiguity; (iv) errors in the amino-acid analysis or data production may add to the difficulty of producing a sequence. Further papers dealing with this topic have been published by Dayhoff12 and Bradley Merril and Shapiro.13 Dayhoff allows for the presence of erroneous data by specifying that removal of a suspected error will remove two or more inconsistencies and not introduce others.Some differences from the previous program of Bernhard Bradley and Duda occur because Dayhoff uses data for which the C-terminal residues are identified in addition to the N-terminal residues. The first process is to collect together groups which have a specified amino-acid residue in common; these are then sub-divided into sub-groups each of which can be condensed to produce a merged sequence. Fragments which cannot be assigned unambiguously to one sub-group are set on one side and this collection of fragments may contain some erroneous pieces in addition to any ambiguous sections.Merging of the sub-group sequences is then used to produce the final sequence (or sequences). It is pointed out that one amino-acid may be difficult to detect and/or subject to error. Such an acid may be omitted from the sequence determination and added at suitable points specified by the fragments which contain it in the final chain. The process of determining the sequence without using a specified amino- acid is also useful to show how much importance can be assigned to that acid i.e. omission of one acid may have no effect on the final result whereas omission of another acid may produce several alternative possibilities. The paper by Bradley Merril and Shapiro being the first part of a series gives a thorough coverage of the problems involved and the rules put forward ten in number are designed to give accurate indexing of the various fragments (six rules) with a separate rule to stop the process if internal inconsistencies occur followed by merging of the fragments and reduction to the final sequence with a final rule to eliminate alternative sequences.The program is iterative as shown in the block diagram Figure 3 and cycles until either no further overlaps are found or some discrepancy is found in which case a diagnostic message is output. Using this program the authors pointed out a discrepancy between the published comp~sition'~ and the published sequence16 of the B-chain of insulin and were also able to show that when the presence of three valine residues in the molecules is assumed and not two only as in the composition data the dis- crepancy is removed and the h a 1 number of possible sequences produced was six.When fragments produced by acid hydrolysis were added to the input data the sequence identical to the published sequence was obtained immediately. Several textbooks and reviews are extant which deal with the mass spectro- meter and its application to the problems arising in organic chemistry (refs. l2 M. 0. Dayhoff J . Theor. Biol. 1964 8 97. l3 D. F. Bradley C. R. Merril and M. B. Shapiro Biopolymers 1964 2 415. l4 F. Sanger and H. Tuppy Biochem. J. 1951,49,463. l5 F. Sanger and H. Tuppy Biochem. J. 195 1 49,48 1. 462 Sheldrick Index all possible elements in sequence FI fragment If max. index results -+- exceededprint r L - t t Add to list sequence- d e t d . fragments I t Eliminate fragments incorporated Establish overlap indices final output found according to overlapping Print results of this cycle Figure 3 After Bradley Merrill and Sliapiro (ref.13) 463 Application of Computers in Chemical Analysis 16-25). In a low-resolution system each peak has an integral m/e ratio but in a high-resolution system very slight differences can be recorded e.g. 27.0109 (CHN) and 27.02347 (C,H,). With the scanning technique which can record a spectrum in a few seconds a popular method is to connect the inlet of the mass spectrometer to the outlet of a gas chromatograph thus identifying the com- ponents as they appear even though the quantities involved may be minute. This combination allows the separation and identification of complex mixtures c f volatile organic compounds.The second major use is in structural work where a pure compound is decomposed and the structure elucidated from a study of the resultant spectrum. Both methods may utilise either low- or high-resolution spectrometers but as the cost of a high-resolution spectrometer is higher than that of a low-resolution type there is a tendency to use low resolution with the gas-chromatograph which gives satisfactory results and a high-resolution type for structural work where the greater precision makes the results unequivocal but requires more calculation. A. Low Resolution-The sample usually a mixture of fairly low molecular weight organic compounds may be analysed directly or may be separated by a gas chromatograph and the components analysed as they appear in succession.The latter method gives a complete spectrum for successive samples and thus some may be of a pure material while others may contain ions from more than one compound. Identification of each spectrum is simpler than that of the total mixture and is facilitated by a library of the mass spectra produced by specific organic compounds e.g. the A.S.T.M. index of mass spectral data in which the cards are sorted in order of the six most intense lines. This method has also been used with a high-resolution system and will be dealt with in more detail later. When dealing with a mixture of compounds the mass spectrum is obviously more complicated and various additional factors have to be taken into account. For example each compound may be ionised to a different extent and may be present in different quantities in the initial mixture thus affecting the height of the peaks produced.Each peak on the low-resolution spectrum can be assigned an integer i (= m/e) and the height of this peak Pz produced by the substance j is propor- l6 H. C. Hill ‘Introduction to Mass Spectrometry’ Heydon and Sons Ltd. 1966. l7 ‘Mass Spectrometry of Organic Ions’ ed. F. W. McLafferty Academic Press New York 1963. l8 R. I. Reed Quart. Rev. 1966 20 527. l9 J. H. Beynon ‘Mass Spectrometry and its Application to Organic Chemistry’ Elsevier Amsterdam 1960. 2o R. I. Reed ‘Ion Production by Electron Impact’ Academic Press London 1962. 21 R. I. Reed ‘Application of Mass Spectrometry to Organic Chemistry’ Academic Press London 1966. 22 K. Biemann ‘Mass Spectrometry’ McGraw-Hill New York 1962. 23 H.Budzikiewicz C . Djerassi and D. H. Williams ‘Interpretation of Mass Spectra of Organic Compounds’ Holden-day Inc. San Francisco 1964. 24 ‘Structure Elucidation of Natural Products by Mass Spectrometry’ Holden-Day Inc. San Francisco 1964. 25 ‘Advances in Mass Spectrometry’ ed. J. D. Waldron Pergamon Press London 1959. 464 Sheldrick tional to the amount xj of that compound. The ratio of proportionality is specified as aij (see Hopp and Wertzler26). Hence for one peak from one compound we have Pi = aij x xj It has been shown that the height of such peaks are additive i.e. if two com- pounds 1 and 2 give peaks with the same value of i then so that if the sample contains n components then Pi = (ail x XI) + (az2 x x2) n Pi (i = 1 2 . . . m) = atj x xi 2 j = 1 The values of the proportionality ratios azj are known from calibration runs of pure compounds and for a complete analysis of the observed mass spectrum it is necessary to have some idea of the compounds present i.e.the solution of the equations gives information about the relative amounts of compounds in the original mixture rather than identification of these though unsuspected materials may be discovered by a study of the residual spectrum when analysis of the spectrum is thought to be complete in terms of the expected compounds. Once the instrumental corrections have been made the problem is reduced to solving the rn (number of peaks measured) equations for the n values of xj which requires that rn > n and the equations are independent of each other. The equations still however contain errors of measurement plus instrumental variations which can introduce varying amounts of error.Methods of solving the array of equations by inverting the matrix of co- efficients to produce equations of the type n q ( j = 1 2 . . . n) = 2 bjl x Pt i= 1 have been published by Hopp and Wertzler,26 McAdam~,~' Gillette,2s and Tunnicliff and Wads~orth.~~ Hopp and Wertzler use a method of triangular inversion (Gauss pivotal) which involves successive transformations of co- efficients of the matrix. They include a limitation which replaces any coefficient which is calculated as a negative quantity (an impossible value) by zero and show by calculation of the variance that the triangular inverse method produces a more accurate set of coefficients than square inversion. Gillette uses the Gauss-Seidel iteration to solve the equations and specifies a value epsilon to indicate that the required accuracy has been achieved by setting the sum of the squares of the differences between the calculated and observed peak heights to be less than or equal to epsilon.When this condition 26 H. F. Hopp and R. Wertzler Analyt. Chem. 1958 30 877. 27 D. R. McAdams Analyt. Chem. 1958,30 881. 28 J. M. Gillette Analyt. Chem. 1959 31 1518. 29 D. D. Tunnicliff and P. A. Wadsworth Analyt. Chem. 1965,37 1082. 465 Application of Conpiitem in Chemical Analysis is fulfilled the calculated values of the peaks are printed together with the delta peaks which are the differences between the observed and calculated peaks so that the user can check the accuracy of the analysis. It is not clear what happens if the initial postulated analysis was in error by an amount such that the value of epsilon could not be attained by the sum of the squares of the differences but if the program is taken off by the operator after a reasonable number of iteration cycles the delta values should indicate the nature of the component omitted.Tunnicliff and Wadsworth include the selection of the spectra to be used in the calculation as part of the program which can select from up to 150 reference spectra each of which can contain details of 110 mass peaks. The method based on a stepwise regression programme by Efroyrn~on,~~ chooses a group of spectra from a reference list which when multiplied by a concentration factor (xj) gives the best fit (by a least-squares criterion) to the observed spectrum.This has the advantage that the spectra to be used in the calculation are not pre- selected except in so far that the library is pre-selected but can as the authors show lose accuracy in some hypothetical cases though these may be regarded as unsolvable by hand calculation. A factor not specifically mentioned above is that the library of spectra should be recorded under the same conditions as the sample as with variation of temperature various effects may OCCUT including chemical breakdown or reaction of the components of the mixture and the relative distribution of the fragment ions may change. The first will result in identification of the products rather than the actual components while the second may produce wrong proportions of compounds or even failure to solve the problem.The combination of a gas chromatograph to separate the components of a mixture followed by mass-spectrographic analysis of the output is a powerful analytical method and is improved by the on-line use of a computer to analyse the details as discussed by Hites and Biemann3l and Abrahamsson.32 In the first system the output from the gas chromatograph is sampled at 4-sec. intervals over the 30 min. run and with each spectrum is recorded the sum of the data points which it contains. This sum is proportional to the unresolved ion beam current of the mass spectrometer as measured by the beam monitor and a plot of the value against the serial number of the scan is similar to the gas chromato- gram. Use of this plot enables the user to select only those mass spectra of special interest to be processed further and/or retained as permanent records.The mass spectra themselves are identified by comparison with standard spectra as previously discussed. Abrahamsson gives details of the system used for scanning the library of mass spectra (7500) which are recorded on magnetic tape and points out various methods of simplification to reduce the time required to scan the whole tape. 30 M. A. Efroymson ‘Mathematical Methods for Digital Computers’ eds. A. Ralston and H. S. Wilf Wiley New York 1960. 32 S. Abrahamsson Science Tools LKB Instrument Journal 1967,14,29. R. A. Hites and K. Biemann Analyr. Chern. 1965 40 1217. 466 Sheldrick As the number of recorded spectra increases some form of simple check test will become important or the computer time required will be wasteful.B. Metastable Peaks.-These peaks are produced when fragments traveIling through the mass spectrograph undergo further fragmentation producing characteristic peaks which have a considerable spread of mfe when compared with a normal peak. Analysis of the components producing such peaks can be of value in the determination of alternative fragmentation paths. A program by Mandelba~m~~ calculates the possible metastable ion peaks from the equation where m and m2 are the m/e values for the normal peaks of the fragments involved. The information required by the program is (a) total number of normal peaks; (b) list of m/e values for these peaks; (c) number of metastable peaks; (d) list of m/e values for these peaks. Values of (n~,)~/m are calculated for all permutations of the normal peaks and if the calculated value differs from a metastable peak by less than a specified amount (k 0403m~ in the exampIe quoted) then this combination of peaks is output together with the value of the metastable peak.The number of combina- tions which can produce a particular metastable peak may vary and if the number is large it may be impossible to assign a specific combination as their origin. m2 = (m2)2/m C. High Resolution.-Introduction of the high-resolution mass spectrometer by means of which the m/e ratio may be quoted to five decimal places has increased the resolution of the mass spectrum so that the elemental composition of a peak and of the difference between peaks may be calculated with a high degree of certainty. The amount of arithmetical manipulation has also increased considerably and computer handling of the data processing is almost obligatory to avoid errors and overlooked relationships.The differences which can be detected are perhaps best shown by the following examples where the m/e ratios are expresed in terms of I2C = 12~00000 Low resolution High resolution Elements present 89.02659 C303H5 89-062 5 5 C402Hg 88.01 680 C303H4 8 8.05456 C402H8 i 89 88 Biemann and McM~rray~~ give this example in details of a program which applies certain conditions to the mass spectrum in order to establish the molecular ion peak. This peak is usually the peak corresponding to the highest m/e value but this may not be the case if (a) the compound breaks down so 33 A. Mandelbaum Israel J. Chem. 1966,4 161. 34 K. Biemann and W. McMurray Tetrahedron Letters 1965 647.467 Application of Computers in Chemical Analysis readily that no molecular ion peak appears or (b) if an impurity is present which produces a high m/e peak. Five criteria are listed all of which can be checked very quickly and success or failure used to steer the search for the molecular ion. The conditions specified for the peak are (i) the species may not contain any heavy isotopes; (ii) the number of hydrogen atoms must be of the same parity as the number of nitrogen atoms (Le. even if even odd if odd); (iii) the mass differences between the peak and peaks of lower m/e must be equivalent to the loss of a reasonable chemical group; (iv) if the highest m/e peak does not fit these criteria then the true molecular ion should be related to these peaks by combinations of atoms which can be lost in simple fragmentation processes; (v) the elemental composition of the lower mfe fragments must not require more atoms of one type than the molecular ion processes.These criteria can be applied to postulated systems and the authors show that the correct molecular ion could still be identified when the two peaks corre- sponding to Mf and (M - 1)+ were removed from the mass spectrum of androsterone. For establishing the chemical fragments which correspond to specific differ- ences between peaks it is useful to have a list of the possible combinations of the elements involved and the m/e value for each possible fragment. A description of a program which can produce such a list has been published by Tunnicliff Wadsworth and S~hissler~~ in which they specify the requirements necessary to keep the number of fragments listed within reasonable bounds.These involve limitations on the number of types of atom allowed or the relative abundances of particular types of atom. The limitations serve to keep the size of the final tables to a manageable size and also restrict the time required for sorting the results into a list of increasing mass. A similar system to that already described for low-resolution mass spectro- metry by Hites and Biemann and by Abrahamsson has been described by Bowen Chenevix-Trench Drackley Faust and Saunder~~~ for a high-resolution system. The postulated system involves an Argus 500 computer directly linked to a high- resolution mass spectrometer to handle both on-line recording and data process- ing. A 10-sec. scan is sampled and if every measurement were recorded would soon overload the data storage.To avoid this only values which exceed a specified minimum are recorded thus excluding general background and random noise only using storage for necessary information. In addition time is thus provided between peak recording for calculations to be carried out. Calibration involves the presence of a reference compound and the m/e peaks of the sample are calibrated by interpolation from the known peaks. A perfluoro-compound is recommended for a reference compound if the sample does not contain fluorine as there is less tendency for overlapping to occur. 35 D. D. Tunnicliff P. A. Wadsworth and D. 0. Schissler Analyt. Chem. 1965,37 543. 3eH. C. Bowen T. Chenevix-Trench S. D. Drackley R. C. Faust and R.A. Saunders J . Sci. Instr. 1967 44 343. 468 Sheldrick 2 Peptide Sequence Determination Determination of the amino-acid sequence of a peptide or protein by mass spectrometry is an attractive proposition. By an analysis of the m/e peaks obtained it should be possible to establish the exact sequence of a very small amount of material. Two main difficulties arise however the first being due to the stability of the material i.e. proteins in general are difficult to volatilise and this stability increases with the size of the molecule. The usual way to over- come this is to attach a known group to the N-terminal group to form an ester which will be more volatile than the free peptide. A variety of groups has been used for this and the group may serve a second purpose by acting as a starting point for the analysis of the data.The second difficulty arises when breakdown of side-chains occurs in addition to the breaking of the primary chain. This occurs when residues with sizeable side-chains e.g. asparagine proline serine and may also be affected by interactions with neighbouring side-groups. Thus the three main areas of interest are (a) choice of end-group substituent; (6) investigation of possible modes of break-down ; (c) programming of computer to reconstruct the primary chain. Biemann Gapp and Sieble3' reduced small peptides with LiAlH4 to produce polyamino-alcohols and recorded the low-resolution mass spectra of these products. They found preferential rupture of the primary chain and the bond connecting the side-group to the chain. They point out that a peak correspond- ing to M + 1 produced by an ion-molecule collision occurs but this is recog- nised by its variation of relative intensity with change of pressure and focusing conditions.Discussions of the terminal substituents and the types of side-chain rupture have been published by Shemyakin et aZ.38 and by Ovchinnikov et al.39 These papers specify acylation of the free peptide followed by methylation to produce compounds which are sufficiently volatile and break down in a reasonably simple manner. In addition to the breaks at the amide bonds the resultant fragments can lose the elements of CO producing related peaks and metastable peaks which can be used to confirm the mechanism. Additional factors which may produce partial fragmentation are the nature of the side-groups and the relationships between neighbouring side-groups.These factors are listed below in terms of the nature of the side-group glycine alanine-little fragmentation ; valine leucine iso-leucine-some fragmentation of a simple nature; methionine-may lose all the side-chain or if some interaction with neighbouring groups occurs may lose only part of the side-chain ; proline-may condense the ring; serine threonine and cysteine-may lose the functional substituent ; cystine-S-S bond is easily 37 K. Biemann F. Gapp and J. Sieble J. Amer. Chem. SOC. 1959 81 2274. 38 M. M. Shemyakin Yu. A. Ovchinnikov A. A. Kiryushkin E. I. Vinogradova A. 1. Miroshnikov Yu. B. Alakhov V. M. Lipkin Yu. B. Shvetsov N. S. Wulfson B. V. Rosinov V. N. Bocharev and V. M. Burikon Nature 1966,211 361. 3nYu. A. Ovchinnikov A.A. Kryushkin E. I. Vinogradova B. V. Rosinov and M. M. Shemyakin Biokhimiya 1967 32 427 (Biochemistry U.S.S.R. 1967 351). 469 Application of Computers in Chemical Analysis broken; asparagine and aspartic acid-elimination of the p-substituent by losing first the elements of ammonia (or alcohol) followed by the elements of CO ; glutamine and glutamic acid esters-similar steps to asparagine plus some breaks in the C-C bonds of the side-chain; y-methyl esters of a- glutamic acid-this can lose the elements of water with the formation of a ring to the neighbouring amide group ; phenylalanine tyrosine histidine and tryptophan-side-chain elimination as RCH2 or RCH2+ or cleavage of the N-Ca bond. It is suggested that a method of overcoming the difficulty of vaporising the peptide (for long-chain samples) would be to apply Edman’s method to split off several residues followed by mass spectrometry of the remainder.A computer program to elucidate the peptide sequence of a chain even before the mass spectrum is considered has to be organised to start the calculation either from the molecular ion and work downwards to fragments of lower m/e or to start with a recognisable end-group and build up the sequence by addition of peptide fragments. Barber et aL40 say that the second method was not found to be satisfactory and describe a program which starts from the molecular ion. Additional chemical data are provided if available and checks are carried out to make all the data compatible; e.g. an example quoted shows that the amino-acid composition data can be changed to fit the molecular formula and molecular weight.During the calculation three types of fragmentation are considered (a) linear peptide break; (b) (cyclic peptides)-fragmentation with the loss of an amino-acid residue plus the elements of ammonia; (c) as (b) but with the loss of an amino- acid residue less the elements of CO. In addition the program is designed to allow for cyclodepsipeptides which contain hydroxy-acids in addition to amino-acids residues. Other programs by Gavrilov et aZ.,4l Senn et and Biemann et aZ.,43 start from the other end of the mass spectrum. Gavrilov et al. discuss an al- gorithm* for establishing the chain sequence from low-resolution data of the fragments produced by hydrolysis i.e. a combination of the first stage of the amino-acid sequence analysis by chromatographic methods and identification of the fragments produced by their mass spectra.The sorting process described is similar to those previously described and the authors claim success for chains of up to twenty-five residues even in cases where some peaks do not occur. In the latter case an alternative system is proposed in which partial chains are synthesised and finally combined to produce a single chain though this appears to produce certain ambiguities. * Algorithm a mathematically unambiguous set of instructions. *O M. Barber P. Powers M. J. Wallington and W. A. Wolstenholme Nature 1966,212 784. 41 V. Yu. Gavrilov A. D. Frank-Kamenetskii and M. D. Frank-Kamenetskii Biokhirniya 1966 81 799 (Biochemistry U.S.S.R. 1966 689). 43 M. Senn R. Venkataraghavan and F.W. McLafferty J. Amer. Chem. SOC. 1966,88,5593. 43 K. Biemann C. Cone B. R. Webster and G. P. Arsenault J. Amer. Chern. SOC. 1966,88 5598. 470 She fdrick The programs by Senn et af and by Biemann et al. identify the N-terminal substituent group. This process is simplified if the substituent group is markedly different from the normal peptide side-chains e.g. trifluoroacetyl and volatility is improved if the terminal carboxy-group is also esterified. Once the N-terminal group is found each possible amino-acid residue is added in turn and a search made of the mass spectrum for the presence of a corresponding peak. This search is obviously shortened if a previous analysis can limit the number of possibilities to be considered. To allow for different fragmentation mechanisms along the chain the search is also repeated for each residue less the elements of CO.A successful result of either search establishes the N-terminal residue and the process can be repeated to find the next residue and so on. If an ambiguity arises where two possible residues are found then the next stage is carried out for both combinations and if necessary repeated until one can be rejected. At each stage in addition to the various amino-acid residues the ester group OR is tested for the molecular ion. The whole process may also start using the C-terminal ester portion. These processes can also be used to check for frag- ments which are produced by side-chain rupture or re-arrangement if the appropriate values are calculated from chemical considerations. In all the programs discussed in this article it is essential that the maximum amount of information obtained by the program is output for consideration. This is necessary even if all appears to have fitted perfectly since a critical examination of the results may reveal possible variations which are not con- sidered in the program. Programs can be quite complicated and yet still fail to produce a correct answer merely because of some unforeseen item. 47 1
ISSN:0009-2681
DOI:10.1039/QR9702400454
出版商:RSC
年代:1970
数据来源: RSC
|
|