摘要:

Chemical Society Reviews Vol 2 No 4 1973 Page THE MELDOLA MEDAL LECTURE Chemical Aspects of Glycoproteins, Proteoglycans, and Carbohydrate -Protein Complexes of Human Tissues By John F.Kennedy 355 Chirality in Carbonium Ions, Carbaniom, and Radicals By J. W.Henderson 397 Chemistry of Azidoquinones and Related Compounds By H. W.Moore 415 Chemistry-a Topological Subject By R. W.Jotham 457 The Brmsted Relation- Recent Developments By A. J. Kresge 475 The Chemical Society London Chemical Society Reviews Chemical Society Reviews appears quarterly and comprises approximately 25 articles (ca. 600 pp) per annum. It is intended that each review article shall be of interest to chemists in general, and not merely to those with a specialist interest in the subject under review.The articles range over the whole of chemistry and its interfaces with other disciplines. Although the majority of articles are specially commissioned, the Society is always prepared to consider offers of articles for publication. In such cases a short synopsis, rather than the completed article, should be submitted to The Editor, Reports and Reviews Section, The Chemical Society, Burlington House, Piccadilly, London, WlV OBN. Members of The Chemical Society may subscribe to Chemical Society Reviews at E2.00per annum (beginning 1974, f3.00per annum); they should place their orders on their Annual Subscription renewal forms in the usual way. Non-members may order Chemical Society Reviews for €8.00 per annun (beginning 1974, f10.00 per annum) (remittance with order) from: The Publications Sales Officer, The Chemical Society, Blackhorse Road, Letchworth, Herts., SG6 lHN, England. 0 Copyright reserved by The Chemical Society 1973 Published by The Chemical Society, Burlington House, London, W1V OBN Printed in England by Eyre & Spottiswoode Ltd, Thanet Press, Margate

ISSN:0306-0012    

DOI:10.1039/CS97302FP009

出版商:RSC    

年代:1973

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS97302FX013

出版商:RSC    

年代:1973

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS97302BX015

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

THE MELDOLA MEDAL LECTURE" Chemical Aspects of GIycoproteins, Proteoglycans, and Carbohydrate-Protein Complexes of Human Tissues By John F. Kennedy DEPARTMENT OF CHEMISTRY, UNIVERSITY OF BIRMINGHAM, EDGBASTON, BIRMINGHAM B1S 2TT 1 Introductlon Most people, at an early age, learn either from elementary instruction in general science, or from their parents, or from the cornflake packet on the breakfast table, that carbohydrates and proteins are very important commodities in the processes and maintenance of life. Although for the majority the descriptions of these commodities go no further, few can be left entirely unmoved by some degree of fascination, even if only fleetingly and to a shallow depth, for the complexity of the automatic chemical reactions which go on day and night in our bodies.Some molecules, including carbohydrates and amino-acids, that are involved in these reactions which provide energy, form tissue and bone, and convey messages, are very simple in structure. For example, the small molecule D-glucose is an important constituent of blood and is involved in the energy cycle. However, most of the material which goes to make up our frame is composed of polymeric structures. A high proportion of these structures has proved to contain carbohydrate and protein and therefore comes under the general classifications of glycoproteins, proteoglycans, or carbohydrateprotein complexes. 2 General Considerations A. Definitions of Glycoproteins, Proteoglycans, and Carbohydrate-Protein Complexes.-In more detail, glycoproteins contain a protein [poly(amino-acid)] chain which may consist of, for example, some three hundred amino-acid units which can be any of the twenty or so naturallyoccurringL-a-amino-acids(Table1). This protein chain is essentially the backbone of the molecule and the carbo- hydrate part of the molecule takes the form of oligosaccharide chains which are pendant and covalently bound to the protein chain (Figure 1).The oligosac- charide chains are usually hetero-oligosacchdride chains which are frequently branched and consist of neutral monosaccharides (D-galactose, D-glucose, D-mannose, or L-fucose), basic monosaccharides (2-amino-2-deoxy-~-ga~actose or 2-amino-2-deoxy-~-gh.1cose),and acidic monosaccharide (neuraminic acid) (Table 2).The basic carbohydrate units are N-acetylated and the neuraminic acid * First dslivered on 27 September 1972 at the Autumn Meeting of the Chemical Society held at the University of Nottingham. 355 me Meldola Medal Lecture Table 1 Structures of a-amino-acids1 Name Formula Neutral amino-acids (one amino-group and one carboxy-group) 1. Glycine (g) CH,(NH,)*CO,H 2. Alanine (g) CH3-CH(NH2)CQ2H 3. Valine (g, e) (CEI3),CH 'CH(NH2) COZH 4. Leucine (g, e) (CH3),CHCH2 CH(NH2)COzH 5. Isoleucine (9,e) (C,H,)(CH,)CH *CH(NH2) CO,H 6. Norleucine (I) CH3.(CH2)3*CH(NHZ).C02H 7. PhenylaIanine (g, e) 8. Tyrosine (g) 9. Serine(g) HOCH, CH(NH2) CO2H 10. Cysteine (8) HS *CH2CHfNH2)-C02H 11.Cystine (g) (-S*CH,*CH(NH2)CO2H)2 12. Threonine (g, e) CH3 'CHOH 'CH(NH2) *CO,H 13, Methionine (g, e) CH,.S .CH,*CH, *CH(NH2)COpH I 14. Di-iodotyrosine or iodogorgic acid (2) HOQCHyCH(NH2) aC02H I 15. Thyroxine (I) 16. Dibromotyrosine (1) 17. Tryptophan (g, e) 356 Kennedy Table 1 (continued) Name Formula 18. Proline (g) Acidic amino-acids (one amino-group and two carboxy-groups) *CH(NH,) C02H 20. Aspartic acid (g) H02C*CH2 21. Glutamic acid (g) HOzC -CH2 CH, CH(NH2) C02H 22. 7-Hydroxyglutamic acid (I)* H02C*CHOH.CH2CH(NH2)CO2H Basic amino-acids (two amino-groups and one carboxy-group) 23. Ornithinet NHaCH2 .CHaCH2CH(NH2)COsH NH2I 24. Arginine (g) HN=C-NHCH2*CH2CH2*CH(NH2)COaH 25.Lysine (g,e) NH2*CHa'CH2 *CH2CH**CH(NH2) .COaH 26. 8-Hydroxylysine NH2*CH,*CHOH.CH2.CH2.CH(NH&CO,H 27. Histidine (g) y2 3General stereochemistry for L-amino-acids Rb$I dH I 6Q2 H * Occurrence in proteins uncertain. t Ornithine is probably not present in proteins, but is formed by the hydrolysis of arginine. g general occurrence; I less common occurrence; e essential in man. 357 The Meldola Medal Lecture branched hetero-oligosaccharide chain consisting of galactose, mannose, fucose 2-amino-2 -deoxyhexose, and #-substituted neuraminic acid residues PRNEdGLYCAIS P linear regular polysaccharide chain possessing alternating sequence .of hexuronic acid (or hexose) and 2-amino-2-deoxyhexose residues Figure 1 General representation of glycoproteins and proteoglycans units are N-acetylated or N-glycolylated and in some cases are also 0-acetylated.The oligosaccharide chains are therefore overall moderately acidic. Whilst proteoglycans are also based on a protein backbone, the carbohydrate takes the form of polysaccharide chains which are pendant and covalently bound to the protein chain (Figure 1). These polysaccharide chains are linear and regular, possessing a1ternati ng monosaccharide sequences which generally involve acidic monosaccharide (D-glucuronic acid or L-iduronic acid) and basic monosaccharide (2-amino-2-deoxy-~-ga~actoseor 2-amino-2-deoxy-~-g~ucose). The basic units are N-acetylated or N-sulphated and are frequently 0-sulphated.These polysaccharide chains are of course strongly acidic, and were therefore Kennedy Table 2 Structures of monosaccharides OHao,,CH20H HOjQ"0-H HO OH HO OH HO OH /3-D-galactopyranose j?-D-glucopyranose a-D-mannopyraxlose OH ,0H2WOH a-L-fucopyranose 2-amino-2-deoxy- P-D-galacto p yrano se 2-amino-2-deoxy -p-D-gluco p yranose OH 5-~no-3,5-d~deoxy-~-g~ycero-a-~-ga~octo-non1dose(neuraminic acid) HO p-D-glUCOpytanuroniC acid a-r-idopyranuronic acid called 'acidic mucopolysaccharides' but are now known as'glycosaminoglycans' in the interests of systematization. The number of polysaccharide chains attached toa unit length or unit molecular weight of protein backbone in a proteoglycan is much greater than the number of oligosaccharide chains attached per unit of protein backbone in a glycoprotein.fie Metdola Medal Lecture Thus carbohydrate predominates in a proteoglycan whereas protein predominates in a glycoprotein. The term carbohydrate-protein complex is applied to situations where glyco- proteins, proteoglycans, polysaccharides, oligosaccharides, proteins, peptides, or lipid-type molecules are linked with one another by non-covalent bonds to give complexes which contain both carbohydrate and protein. The nature of the intramolecular linkages is usually ionic. B. Historical Development of the Concepts of Glycoproteins, Proteoglycans, and CarbohydrateProtein Complexes.-Historically, proteins such as albumin and gelatin were known before glycoproteins.Probably the first paper on a glyco- protein appeared in 1805 when it was observedl that mucus differed in its physicochemical properties from proteins and therefore appeared to be a new type of macromolecular proteinaceous compound. It was not until 1865 that the first chemical evidence for the presence of carbohydrate in mucins was reported, it being observeda that elemental analysis of purified mucin gave values for carbon, and more so for nitrogen, that were significantly lower than the cor- responding values for proteins, and that acidic hydrolysis of mucin yielded a compound which appeared to be glucose. The presence of basic carbohydrate in glycoproteinaceous materials was particularly recognized in the period 1894- 1905, the compound isolated being found to be identical with the hydrolysis product (chi tosamine) of chitin, a linear poly(2-acetamido-2-deoxy-~-glucose) which occurs in crab shells.Nearly coincident with the work on mucinous secretions etc., another type of protein-linked carbohydrate was discovered in the ground substance of con-nective tissue, it being reported that a carbohydrate termed chondroitin sulphuric acid could be produced by the hydrolysis of hydine ~artilage.~ This carbohydrate was ultimately purified4 and was conceived5 in 1891 to be a polymer which contained glucuronic acid, sulphuric acid, and chi tosamine -a polysaccharide which is now known to be a glycosaminoglycan. Gradually, the picture of carbohydrate firmly bonded to protein emerged.It appeared that all glycoproteins associated with mucous secretions contained 2-amino-2-deoxyhexose and thus the term mucopolysaccharide was introduceds in 1938 to describe 2-amino-2-deoxyhexose-containingpolymeric materials of animal origin. As more and more information on the occurrence of macro- molecular carbohydrate-containing proteinaceous materials was obtained, so it became apparent that not only mucins but all these materials contained 2-ami no-Zdeoxyhexose. The occurrence of neuraminic acid in glycoproteins was not recognized until J. Bostock, J. Nat. Phil. Chem. Arts, 2nd Ser., 1805, 11,244.* E.Eichwald, Ann. Chem. Pharm., 1965,134, 177. a G. Fischer and C. Boedeker, Ann. Chem. Pharm., 1861,117, 111. C.T. Morner, Skand. Arch. Physiol., 1889, 1, 210. 6 0.Schmiedeberg, Arch. Exp. Pathol. Pharrnakol., 1891,28,354. K. Meyer, Cold Spring Harbor Symposium on Quantitative Biology, 1938, 6,91. Kennedy a much later date (1949’) on account of the complexity of the monosaccharide structure and the fact that it had not been previously detected in simpler form, as had neutral and basic monosaccharides before their recognition in glyco- proteins. C. Occurrence and General Function of Glycoproteins, Proteoglycans, and Carbohydrate-Protein Complexes.-The course of events has been such that these types of macromolecule have been found extensively in mammalian tissues and fluids, and the literature reporting their occurrence, isolation, purification, and structure is vast (see refs.8 and 9 for reviews). Thus there is now a general picture of the occurrence in humans of glycoproteins in connective tissues (including bone) as matrix formers ;in specialized organs as hormones; in blood cells (erythrocytes and leukocytes) as blood group active substances and in these and other cells as protective coatings; in serum as immunoglobulins and antibodies which give the subject immunity against infection; in serum, milk, urine, saliva, and other secretions and in body fluids as enzymes involved in biosynthesis and metabolism, etc. Proteoglycans occur in connective tissues as matrix builders, in joint fluids as lubricants and shock resistors, and in the eye as humour. Carbohydrate-protein complexes occur as cross-linking agents for/of matrix-forming molecules and possibly as glycoprotein-lipid complexes in tissues.Whilst less is known of such non-covalent complexes, there is no doubt that there is much to be discovered in terms of complexes formed by the known glycoproteins and proteoglycans as they perform their various functions. Overall, whilst many glycoproteins and proteoglycans have been documented as contained in the human body, it may be predicted tliat more will be found and that molecules presently regarded as proteins will be found to contain carbo- hydrate as more attention is given to systematic carbohydrate analysis. Clearly, much less is known of the precise function of the macromolecules and of the chemical processes which they perform and undergo.D. Isolation, Purification, and Compositional and Structural Analysis of Glyco-proteins, Proteoglycans, and Carbohydrate-Protein Complexes.-In spite of the fact that the general picture of the occurrence of such macromolecules represents a large number of publications, it is important to realize that a number of problems remain unsolved. Thus many glycoproteins etc. have yet to be isolated in a pure form. Even a definition of purity is no longer straightforward since microheterogeneity is now well recognized, i.e. a particular glycoprotein may occur in forms which differ from one another by one carbohydrate or amino- acid unit, and it is also recognized that a particular proteoglycan may occur in forms which differ slightly in their glycosaminoglycan composition.Ion-exchange fractionation and gel filtration, which distinguish molecules according to their charge and molecular size, respectively, are used most extensively to effect A. Gottschalk and P. E. Lind, Nature, 1949,164,232.* A. Gottschalk, ‘Glycoproteins’, Elsevier, Amsterdam, 2nd edn., 1972. J. S. Brimacombe and J. Webber, ‘Mucopolysaccharides’, Elsevier, Amsterdam, 1964. The Meldola Medal Lecture purification. Other techniques include dtracentrifugation, electrophoresis, iso- electric focusing, and affinity chromatography/immunoadsorption, which separate molecules according to their shapelsize, charge, pKa value, and ability to complex with a specific compound, respectively.Following purification, the next chemical step is a component analysis per- formed after hydrolysing the macromolecules into their component carbo- hydrate and amino-acid units. Complete and accurate amino-acid analysis is now an established technique but it is unfortunate that many workers tend to assume that glycoproteins contain L-fucose, ~-galactose, D-glucose, D-mannose, 2-acetamido-2-deoxy-~-glucose,2-acetamido-2-deoxy-~-galactose, and neura-minic acid as the monosaccharide units, determining neutral sugars and basic sugars only collectively by the classical colorimetric techniques and assuming that the basic sugars are in fact N-acetylated. However, with the problems of microheterogeneity it is essential to determine qualitatively and quantitatively each specific type of carbohydrate unit individually and ultimately to determine its D or L configuration.This necessarily involves the use of more advanced techniques such as quantitative gas-phase chromatography, mass spectrometry, ionexclusion chromatography, and ion-exchange chromatography. Both auto- mated amino-acid and carbohydrate analysers based on ion-exchange chroma- tography are now available and this should alleviate some of the difficulties of component analysis. Determination of the primary structure follows component analysis and requires even more skill and attention. Whilst it is true that amino-acid sequences can be determined by automated and semi-automated techniques and peptide mapping, it is not a foregone conclusion that a sequence of a few hundred amino- acid units can be determined without difficulty.Structurally, the carbohydrate portion of the macromolecule is even more elusive since in addition to determin-ing,as in the case of amino-acids, the sequence of the various units, it is necessary to determine which of the hydroxy-groups of the carbohydrate are involved in the linkage to the adjacent units and also the stereochemistry of carbon-1 of the carbohydrate unit. Such parameters give rise to at least eight possibilities for the way in which a neutral hexose unit such as D-galactose is linked within a carbo-hydrate chain. In the structural determination of both carbohydrate and protein moieties a number of techniques are now employed, and these depend essentially on selective cleavage unit by unit along the chain or on breakdown of the macromolecule into small peptides.oligosaccharides, or glycopeptides which present simpler cases for sequential analysis. As will be seen later, the linkage points and types in the carbohydrate moieties are determined by further specialized techniques. Finally, from the chemical viewpoint, there remain to be determined the secondary and tertiary structures. The secondary structure involves the com-plexity that many glycoproteins consist of subunits, i.e. they are oligomers of polymeric species. These subunits are frequently different from one another and each has a specific part to play in the overall properties and structure of the molecule.The tertiary structure involves the spatial three-dimensional structure Kennedy of the complete molecule. It must not be imagined that the backbone of the macromolecules under consideration is straight -generally it is not and may involve random orientation, coil/helix, ball, and rod type shapes, and indeed may be partially random and partially ordered. Information regarding the particular structures involved in a macromolecule is derived from X-ray diffrac- tion analysis, sedimentation analysis, circular dichroism and optical rotatory dispersion spectroscopy, and determination of solution properties. Complete primary structures have been determined for very few carbohydrate- containing macromolecules and complete information on any tertiary structure has yet to be achieved.This situation holds at present not only on account of the complexityof the structures to be identified but also in the case of human materials on the fact that many of the macromolecules are available only in milligram or even microgram amounts. Thus, as will be seen later, in any such work, the chemist is frequently forced to devise new sensitive analytical methods or to scale down drastically existing methods. Coupled with structural analysis are investigations of structure-biological activity relationships whether the activity manifested by the macromolecule be enzymic, hormonal, or immunological. Some idea of the involvement of a certain unit type in a structure may be obtained by applying to the macromolecule a reaction which is specific for that unit in bound form and testing for the effect of the derivatization on the activity. Further confirmation of the involvement of the unit may be obtained by reversion of the activity effect on de-derivatization.In other words, monitoring of the activity of the molecule during the reaction sequences is used as a test for the occurrence of a chemical reaction and this technique has a great advantage in that it can be carried out on a microgram or nanogram scale since many of the active macromolecules are active at such levels. E. Present Work.-Since macromolecules containing carbohydrate and protein are so prolific in the human, it is impossible to research all types and therefore specializationis necessary.Thus our work has involved the investigation of human hormonal glycoproteins in the fertility field and of proteoglycans as con- stituents of dermal tissue and as excess products of the body in disease. These two classes of molecule will be described in more detail. However, in the case of humans, the chemical and structural investigation is not only academically orien- tated and is not an end in itself, but is directed at giving a better understanding of health and disease and so to lead the way to chemically based diagnoses and monitoring of treatment. To achieve such aims emphasis must be laid on chemi- cally based analyses as well as structural determination of the macromolecules. In this respect our work has extended to the development of microscale techniques and to the preparation and application of new and improved water-insoluble agentslreactors.These agents are frequently water-insoluble derivatives of enzymes and glycoproteins which wn be used for molecular separa- tions, speciiic purifications, structural studies, and assays of biologically active macromolecules.Chief advantages of these water-insoluble reactors are that they me Meldola Medal Lecture have greater stability than their soluble counterparts, are easily separated from soluble material, can be re-used, and are amenable to packing into cartridges etc. for simplified automated analyses, including clinical chemical analyses. These types of reagent will be described in more detail later. 3 Hormonal Glycoproteins A.Nomenclature of Hormonal G1ycoproteins.-The group of glycoproteins which possess hormonal activity comprises follicle-stimulating hormone, luteiniz- ing hormone, human chorionic gonadotrophin, human menopausal gonado- trophin, pregnant mare serum gonadotrophin, and thyroid-stimulating hormone. With the exception of thyroid-stimulating hormone, these hormonal glycoproteins are also known as gonadotrophins on account of their gonad-stimulating ability and their involvement in the fertility cycle. Some aspects of earlier work on the subject of hormonal glycoproteins have been reviewed.1° B. Composition of Hormonal G1ycoproteins.-These hormones, as already implied, are all macromolecular and are composed of the naturally occurring amino-acids together with the usual carbohydrate components of glycoproteins. Methods and problems in the determination of carbohydrate compositions and structures of glycoprotein hormones have been reviewed.” Typical compositions are illustrated by those12 of human pituitary follicle-stimulating hormone13 and human chorionic gonadotrophin (Table 3).Generally, these hormones all contain these components in moderately different proportions. It should be noted that the precise composition of each hormone from a number of species has in many cases yet to be determined since complete purity in isolation has yet to be achieved. Furthermore, it now emerges that some of the carbohydrate and amino-acid units, particularly in certain terminal positions, may be ‘optional extras’, thus giving rise to the problem of microheterogeneity within any hormone preparation.Microheterogeneity may, of course, be the result of incomplete biosynthesis or of metabolism. It will be realized that many other, totally unrelated glycoproteins have sfmilar compositions, but that it is the precise quantitative composition and more especially the sequence of units along the macromolecular chain (primary structure), the combination of any subunits (secondary structure), the spatial arrangement of the chain (tertiary structure), and the size of the molecule which determine the hormonal character of the molecule. C.Occurrence, Function, and Use of Hormonal G1ycoproteins.-Follicle-stimulating and luteinizing hormones occur in the pituitary gland, and their IQ J.F. Kennedy, Endocrinologica Experimentalis, 1973, 7, 5. l1 J. F. Kennedy, in ‘Structure-Activity Relationships of Protein and Polypeptide Hormones,’ ed. M. Margoulies and F. C. Greenwood, International Congress Series 241, Part 2, Excerpta Medica, Amsterdam, 1972, p. 360. J. F. Kennedy and M. F. Chaplin, unpublished results. la W. R. Butt, S. S. Lynch, and J. F. Kennedy, in ‘Structure-Activity Relationships of Protein and Polypeptide Hormones’, ed. M. Margoulies and F. C. Greenwood, International Congress Series 241, Part 2, Excerpta Medica, Amsterdam, 1972, p. 355. Table 3 Compositions of human pituitary follicle-stimulating hormone and human chorionic gonadotrophin Amino-acid or curbohydrute unit Number of residues Percentage by weight Follicle-Chorionic Follicle-Chorionic stimulating gonadotrophin stimulating gonadotrophin hormone (per (per mol. wt.of hormone mol. wt. of 41 OOO)35 OOO)L-Asparticacid* 15.2 17.5 5.00 5.16 L-Threunine 17.3 14.2 4.98 3.68 L-Serine 13.8 15.3 3.43 3.40 L-Glutamic acid 25.0 17.2 9.20 5.65 L-Proline 12.2 29.2 3.37 7.24 Glycine 11.4 15.6 1.85 2.27 L-Alanine 11.4 14.3 2.31 2.60 L-Valine 12.5 13.8 3.54 3.50 r.-Methionine 3.2 3.1 1.18 1.02 Half L-cystine 15.2 11.1 4.45 2.86 L-Isoleucine 7.3 5.4 2.24 1.58 L-Leucine 10.5 14.3 3.39 4.13 L-Tyrosine 12.4 5.3 5.81 2.19 L-Phen ylalanine 7.6 6.5 3.22 2.43 L-Lysine 12.4 10.1 4.56 3.32 L-Histidine 7.0 4.6 2.73 1.60 L-Arginine 8.2 13.3 3.64 5.33 D-Galactose 13.4 16.6 6.20 6.81 D-Mannose 19.4 19.4 9.01 11.99 L-Fucose 2.0 1.5 0.84 0.56 ucose 14.5 15.5 8.45 7.992-Acet amid 0-2-deoxy-~-gl2-Acetamido-2-deoxy-~-galactose 1.6 5.0 0.93 2.61 N-Acetylneuraminicacid (acid hydrolysis) 10.6 14.8 8.81 11-10 N-Acetylneuraminicacid (enzymic hydrolysis) 11.1 16.2 9.25 12.10 * With the exceptions of D-galactose and L-fucose, the D and L configuration are assumed by analogy with other glycoproteins. ch l%e Meldola Meiial Lecture release from this gland is dictated by a releasing hormone/factor produced by the hypothalamus, this hormone being a decapeptide (1).l4 Thyroid-stimulating hormone is similarly released from the pituitary by a tripeptide (2)ls9le releasing- hormone of hypothalamic origin.Chemico-biological relationships of the two releasing hormones have been inve~tigated.~~,~~ Follicle-stimulating and luteiniz- ing hormones act upon the ovary to stimulate follicle growth/rupture according to the stage of the cyclelo and are ultimately excreted in the urine in macromolecu- Iar active forms. The combined follicle-stimulating and luteinizing hormone content of post-menopausal urine is termed human menopausal gonadotrophin. Thyroid-stimulating hormone acts upon the thyroid gland to stimulate produc- tion and release of thyroid hormones. Human chorionic gonadotrophin is pro- duced by placental trophoblasts, particularly in the first trimester. No releasing factor for this gonadotrophin has as yet been identified.Human gonadotrophins are used in clinical treatment of infertility where cases have proved resistant to the synthetic low molecular weight drug clomiphene. Great success in such treatment has been achieved by Dr. A. C. Crooke and his group in Birmingham, administration of a combination of follicle-stimulatingand luteinizing hormones being followed by human chorionic gonadotrophin. In this respect, apart from general interest in the structure of these glycoprotein hormones, determination of the complete structure is important with a view to establishing structure-activity relationships and identifying the active site($. Hopefully, this will permit an understanding of the way in which these molecules function and provide information for the design of simulators which can be synthesized.D. Activities of Hormonal G1ycoproteins.-Each hormone not only has a biological activity, i.e. its biological function in the species to which it is endo- genous, but also an immunological activity, i.e. its ability to act as an antigen in an antigen-antibody system. The biological activity is usually tested in vivo, the effect of the injected hormone upon certain organs of the test animal being determined. More recently an in vitro test has been developed for follicle- stimulating hormone whereby the biological activity can be determined by direct application of the test material to the ovary.20 Immunological activity is usually determined in a competitive binding tech- nique where unknown antigen (hormone) competes with a standard amount of M H.Matsuo, Y. Baba, R. M. G. Nair, A. Arimura,and A. V. Schally, Biochem. Biophys. Res. Comm., 1971, 43, 1334. C. Y. Bowers, A. V. Schally, F. Enzmann, J. Boler, and K. Folkers, Endocrinology, 1970, 86, 1 143. l6 R. M. G. Nair, J. F. Barrett, C. Y. Bowers, and A. V. Schally, Biochemistry, 1970, 9, 1103. l7 J. F. Kennedy, C. J. Gray, S.Ramanvongse, L. Albrighton, and W. F. White, LifeSciences, 1973, 12, Pt. I, p. 533. J. F. Kennedy, C. J. Gray, S. A. Barker, L. Albrighton, C. Y. Bowers, A. V. Schally, and W. F. White, L$e Sciences, 1971, 10, Pt. 11, p. 569. lo V. Petrow, Chem. in Britain, 1970, 6, 167. M. Ryle, M. F. Chaplin, C. J. Gray, and J.F. Kennedy, in ‘Gonadotrophins and Ovarian Development’, ed. W. R. Butt, A. C. Crooke, and M. Ryle, Livingstone, Edinburgh, 1970, p. 98. Kennedy z I 0 0 I I 00L I II I Z 2 I I I 00 II r--0 I I z I 00 i 3 0 0 0 0 r" z sc!I n Y r 3z I 00 Ic?0 s The Meldola Medal Lecture radioactively labelled but otherwise identical hormone for binding to a fixed and substoicheiometric amount of antibody which has been raised specifically to the hormone in question. In order to facilitate separation of bound antigen from unbound antigen, the antibody is frequently immobilized by attachment to a water-insoluble matrix, e.g. bentonite,*l cellulose,22 or Sephade~.~~ Bioassays require a number of days to elapse between administration and measurement and are rather imprecise, whereas the radioimmunoassay is easily completed in a few hours and is much more accurate.The methods for determination of the specific activity of hormonal glyco- proteins are not without criticism; the greater problem is not the determination of activity on a small scale, but the measurement of mass on a small scale. In view of the instability of the hormones and the small quantities being prepared, dry weight determinations are inaccurate. Significant variations exist24 between results based upon alternative methods such as U.V. absorption and colorimetric determination of the protein, and such variations must account for some of the differences in potencies reported by various laboratories for the most highly purified materials.Whilst spectrophotometric methods give tolerable approxima- tions, determination of mass by a full carbohydrate and amino-acid compo- sitional analysis is the only absolute method. E. Isolation and Purification of Hormonal G1ycoproteins.-The closely similar chemical and physical natures of the hormones to one another and to other macromolecules, including unrelated glycoproteins endogenous to the media of origin, give rise to difficulties in the isolation and purification of the particular hormone under investigation. A complex series of purification steps is often necessary in which it is important to take advantage of properties peculiar to the hormone molecule, e.g.acidic and basic groups provided by certain amino-acid units and acidic groups provided by N-acetylneuraminic acid units. Thus the purification methods used by many workers include ion-exchange chromato- graphy and electrophoresis in addition to molecular weight discrimination by gel filtration. It is essential to monitor the purification process by determination of activity as well as protein profiles since no purely chemical technique is hormone-specific. Owing to the sensitive natures of the hormones, where loss of a few residues can abolish all activity, care must be taken to avoid microbial contamination and conditions which deviate markedly from the physiological. Clearly, the final stages of purification prove more and more difficult as the impurities become predominantly more similar to the hormone.A typical purification is one which has been applied to human follicle-stimulating horm~ne~~~~~~~~and which involves a number of steps of chromatography on calcium phosphate, DEAE-cellulose, and Bio-Gel P-150. By such techniques a purification of 5000 fold has been achieved.13 21 W. R. Butt and S. S. Lynch, Clinica Chim. Acta, 1968, 22, 79. J. F. Kennedy and H. Cho Tun, Carbohydrate Res., 1973, 30, 11. a3 L. Wide, S. J. Nillius, C. Gemzell, and P. Roos, Acta Endocrinol., 1973, 73, suppl. 174. BQ S. A. Barker, C. J. Gray, J. F. Kennedy, and W. R. Butt, J. Endocrinol., 1969, 45, 275. 16 A. S. Hartree, Biochem. J., 1966, 100, 754.Kennedy Two other approaches to the purification of hormonal glycoproteins have now been pursued. Isoelectric focusing has been applied to the purification of follicle- stimulating and luteinizing hormone from the human and other species.26-28 However, this technique may suffer from the fact that other materials in the mixture may have PI values analogous to that of the hormone. More important, the technique does not overcome the problem of microheterogeneity since differences of, for example, one or two N-acetylneuraminic acid units between molecules will cause them to migrate to slightly different PI values under the electrostatic field in the static pH gradient column. This theory is supported by the fact that human luteinizing hormone was resolved into multiple components in the range pH 7-10 whereas each component showed comparable high biological and immunological activity.26 Treatment of the components with neuraminidase decreased their number, Immunoadsorption has been applied to the purification of ovine luteinizing Since antibodies highly specific for the glycoprotein hormones can be produced, this technique is theoretically the most important and advanced, permitting almost complete purification in one step. However, problems preventing the complete success of this type of purification include that of raising the specific antibody.This really requires the administration of high purity material to the animal. A further problem is that of achieving a successful elution of the hormonal glycoprotein specifically bound to the immobilized antibody.Many of the reagents used for such elutions in other antibody-antigen systems are known to cause disruption of the hormonal glycoproteins into subunits. The stabilities of the most highly purified preparations of certain hormonal glycoproteins have presented some problems, e.g. follicle-stimulating hormone preparations are often found to be stable in frozen solution but lose activity dramatically on concentration and freeze drying. Indeed some of the highest potencies described have been for preparations which were obtained in solu-ti~n.~~JlThe easy loss of activity is undoubtedly due to alterations to the macro- molecular structure and again emphasizes the sensitivity of these hormones and the precision with which hormonal activity is related to the molecular structure.Some studies have been carried out on the stability of human pituitary follicle- stimulating hormone to various parameter^,^^.^^ and a preparation of highest activity which is both biologically and immunologically stable to freeze drying and concentration has now been obtained.13 F. Primary and Secondary Structures of the Hormonal G1ycoproteins.-The determination of the primary structures of the hormonal glycoproteins is very much in its infancy on account of the difficulties in obtaining quantities of pure 18 L. E. Rekhert, Endocrinology, 1971, 88, 1029. 27 D. Graesslin, H. C. Weise, and G. Bettendorf, in 'Gonadotropins', ed.B. B. Saxena, C. C.Beling, and H. M. Candy, Wiley-Interscience, New York, 1971, p. 159. P. Rachnam and B. B. Saxena, J. Biol. Chem., 1970,245, 3725. D. Gospodarowicz, J. Biol. Chem., 1972, 247, 6491. 30 P. ROOS,Acra Endocrinof., 1968, 59, suppl. 131. 81 W. D. Peckam and A. F. Parlow, Endocrinology, 1969, 84,933. 369 me Meldola Medal Lecture material. The determination has been approached in three ways -determinations carried out on the intact molecules, determinations carried out on the subunits of the molecules, and via structure-activity relationships (see later). Human pituitary follicle-stimulating hormone has been studied by the direct approach. Treatment with the enzyme neuraminidase released all the N-acetyl- neuraminic acid units from the intact molecule and hence these units must occupy terminal non-reducing positions in the oligosaccharide The desialylized but not the intact molecule was susceptible to galactose oxidase, indicating that the galactose units have the D-configuration and do not occupy terminal non-reducing posit ions but are adjacent to the N-acetylneuraminic acid units. Some fucose was released by L-fucosidase, but generally the hormone is disappointingly resistant to the action of glycosidases.However, the number of possibilities of linkage positions of the various monosaccharide units in the hormone has been narrowed by periodate oxidation studies in which adjacent carbohydrate hydroxy-groups not involved in glycosidic linkages are oxidized and reduced and the products, liberated by acid hydrolysis, are identified q~antitatively.~~The identification of linkage type in the carbohydrate moieties of follicle-stimulating hormone was finalized by methylation analysis in which unoccupied hydroxy-groups are converted into methyl ether groups whereas occupied hydroxy-groups are exposed in the subsequent acid hydrolysi~.~~ Identification of the hydrolysis products by gas-liquid chromatography and mass spectrometry demonstrated that the L-fucose units occupy terminal non- reducing positions, the D-galactose units are linked in the 1- and 2-positions, the mannose units exist in three forms (some as terminal non-reducing residues, some as 1,6-linked residues, and some as 1,3,4-linked branch points), and the 2-acetamido-2-deoxyglucoseunits are 1,6-linked.The methylation analysis also showed that these four types of unit exist in the pyranose form and that the 2-amino-2-deoxyglucose units are N-acetylated. The application of glycosidases to human chorionic gonadotrophin has proved successful, revealing the sequences N-acetylneuraminosyl-p-D-galactosyl-2-acetamido-2-deoxy-~-~-g~ucosy~~-~-mannoseand a-L-fucosyl-@galac tosyl-~-acetam~do-2-deoxy-~-~-g~ucosy~-a-~-mannose.~~~~~From the data, tentative structures involving branch points at wmannose units were proposed. Prelim- inary studies also indicated that 2-acetamido-1-[(N-/3-~-aspartyl)amino]-2-deoxy-/?-D-glucopyranosylamineand 2-acetamido-2-deoxy-~-~-gluc~pyranosy1-serine type linkages are involved at the carbohydrate-protein junctions.Turning to the structural approach via the subunits, the phenomenon of subunit formation by the gonadotrophins and thyroid-stimulating hormone s* M. F. Chaplin, C. J. Gray, and J. F. Kennedy, in ‘Gonadotropins and Ovarian Develop- ment’, ed. W. R. Butt, A. C. Crooke, and M. Ryle, Livingstone, Edinburgh, 1970, p. 77. ** J. F. Kennedy and W. R. Butt, Biochem. J., 1969, 115,225. s4 J. F. Kennedy and M. F. Chaplin, Biochem. J., 1972,130,417. 0.P. Bahl, J. Biol. Chem., 1969, 244, 575. 0. P. Bahl, in ‘Structure-Activity Relationships of Protein and Polypeptide Hormones’, ed. M. Margoulies and F. C. Greenwood, International Congress Series 241, Part 1, Excerpta Medica, Amsterdam, 1971, p.99. Kennedy has been reviewed.a7 First evidence that the pituitary glycoprotein hormones could exist in the form of subunits came from work on ovine luteinizing hormone, which was dissociated by hydrochloric acid at pH 1.3.s8*39Many reports on the use of various dissociative reagents such as dichloroacetic acid, urea, and guanidine have followed. Human luteinizing hormone has been shown to be dissociated at a pH of 1.3,40 and by dichloroacetic acid,41 urea, or g~anidine.~~~~~ Evidence for the dissociation of human pituitary follicle-stimulating hormone was originally presented in terms of quantized molecular weights according to the ionic strength of the solvent used for gel filtration.44 Treatment of the hormone with sodium dodecyl sulphate gave two fragments of different molecular and treatment with acid at pH 1.0 at 20 “Calso gave rise to fragments.4s The hormone has also been segregated into its subunits by the action of urea47 and propionic Human chorionic gonadotrophin has been separated into non- identical subunits by preparative gel electrophore~is~~ and urea treatment.sO~sl Work on thyroid-stimulating hormone has been confined to that of bovine origin, but again two subunits have been separated and identified.From these studies and those on such hormones from other specie^^^^^^ it has emerged that all the glycoprotein hormones can be dissociated into dissimilar subunits (aand B)*Recent studies of the subunits of the hormonal glycoproteins have been revieweds2 and the complete amino-acid sequences have been determined in a number of instances for subunits from non-human sources.Of the human hormones, most work has been done on chorionic gonadotrophin, the a-and p-subunits (HCG-a,HCG-R possessing 92 and 139 amino-acid residues, respectively, [(3), (4)] the points of attachment of carbohydrate being indicated.63 The sequence of the N-terminal35 residues of the a-subunit of human luteinizing 37 W.R.Butt and J. F. Kennedy, in ‘Structure-Activity Relationships of Protein and Poly-peptide Hormones’, ed. M. Margoulies and F. C.Greenwood, International CongressSeries 241, Part I, Excerpta Medica, Amsterdam, 1971, p. 115. 38 C. H. Li and B. Starman, Nature, 1964,202, 291. 39 D. N. Ward and M. S. Amott, Anafyt.Biorhem., 1965, 12, 296. 40 A. S. Hartree, in ‘Protein and Polypeptide Hormones’, ed. M.Margoulies, Intefnational Congress Series 161, Part 3, Excerpta Medica, Amsterdam, 1969, p. 799. I1L.E. Reichert, A. R.Midgley, G. D. Niswender, and D. N. Ward, Endocrinology, 1970, 87, 534. 4s R. J. Ryan, N. Jiang, and S. Hanlon, Recent Prog. Hormone Res., 1970, 26, 105. p3 L. E. Reichert and D. N. Ward, Fed. Proc., 1969, 28, 505. 44 C. J. Gray, Nature, 1967, 216, 112. 46 J. F. Kennedy, W. R. Butt, W. Robinson, and M. Ryle, in ‘Structure-Activity Relation- ships of Protein and Polypeptide Hormones’, ed. M.Margoulies and F. C. Greenwood, International Congress Series 241, Part 2, Excerpta Medica, Amsterdam, 1972, p. 348. 46 J. F. Kennedy and M.F.Chaplin, J. Endocrinof., 1973, 57, 501. O7 B. B. Saxena and P. Rathnam, J. Biof. Chem., 1971,246, 3549. 48 L. E. Reichert, Endocrinology, 1972, 90, I I 19. 4B F.J. Morgan and R. E. Canfield, Endocrinufogy, 1971, 88, 1045. N. Swaminathan and 0.P. Bahl, Biochem. Biophys. Res. Cumm., 1970,40,422. 61 S. Donini, V. Olivieri, G.Ricci, and P. Donini, Acra Endocrinol., 1973, 73, 133. 61 S. M.Amir, Acta Endocrinof., 1972, 70, 21. 63 0.P. Bahl. R.B. Carlsen, R. Bellisario, and N. Swaminathan, Blochem. Biophys. Res. Comm., 1972,48,416. 371 3m H-Ala-Pro-Asx-Val-G1x-Asx-Cys-Pro-G~~-~s-~-Leu-Glx-Glx-~x-Pro-Phe-Phe-~r-Glx-Pro-Gly-Ala-Pro-Ile-~u-Glx-~s-CHO s I s Met-Gly-Cys-Cys-Phe-Ser-~g-rg-Ala-Tyr-Pro-Thr-Pro-~u-~g-Ser-Lys-Lys-~-Met-Leu-Val-Glx-Lys-Asn-Val-~-~r-Glx-5 CHO k bI Ser-Thr-~s-Cys-Val-Ala-Lys-Ser-Tyr-Asx-Arg-Vat-Thr-Val-Met-Gly-Gly-Phe-Lys-Val-Glx-Asn-His-~-~a-~s-His-Cys-82Ser-Thr-Cys-Tyr-Tyr-His-Lys-Ser-OH (3) CHO CHO I I H-Ser-Lys-Glx-Pro-Leu-Arg-Pro-Arg-Cys-Arg-Pro-Ile-Asn-Ala-~-Leu-Ala-Va~-Glx-Lys-Glx-Gly-~s-Pro-Val-Cys-Ile-Asn-Val-Thr-Thr-I le-Cys- Ala-Gly-Tyr-Cys-Pro-Thr-Met-Thr-Arg-Val-Leu-Glx-Gly-Val-~u-Pro-Ala-~u-~o-GIx-Leu-Val-Cys-Asx-Tyr-Arg-Asx-Val-Arg-Phe-GIx-Ser-Ile- Arg-Leu-Pro-Gly-Cys-Pro-Arg-Gly-Val-Asx-Pro-Val-Val-Ser-Tyr- Ala-Val- Ala-Leu-Cys-Arg-Ser-Thr-Thr-Asx-Cys-Gly-GIy-Pro-Lys-Asx-His-Pro-Leu-Thr-Cys-Asx-Asx-Pro-~g-Phe-G~-Asx-Ser-Ser-Ser-CHO CHO CHO I I I Lys-Ala-Pro-Pro-Pro-Ser-Leu-Pro-Ser-Pro-Ser-Arg-~u-Pro-Gly-Pro-Pro-Asx-~-Pro-~e-Leu-Pro-G~-Ser-~u-Pro-OH (4) H-Val-GIn-Asp-Cys-Pro-Glu-Cys-Thr-Leu-Gln-Glu-Asn-Pro-Phe-Phe-Ser-Gln-~o-Gly-~a-Pro-Ile-~u-Gln-~s-Met-Gl~-Cys-Cys-Phe-Ser-Arg-Ala-Tyr-Pro (5) Kennedy hormone (HLH-a) has been reporteds4 (5) and is identical with the N-terminal residues of HCG-a except that the latter has three additional residues at the N-terminus. The amino-acid sequence of HCG-a also shows extensive analogies with the a-subunits of ovine luteinizing hormone (OLH-a) and of bovine thyroid-stimulating hormone (BTSH-a) although the carbohydrate compositions are quite different.The amino-acid sequences of the /%subunits of these two hormones from non- human species are, however, quite distinct from one another and from that of HCG-p.No sequence data are as yet available for the subunits of human follicle-stimulating hormone but the a-subunit of human follicle-stimulating hormone (HFSH-a) has been shown to possess an amino-acid composition analogous to HLH-a and HCG-a.s5Although the picture is far from complete, it appears at this stage that the p-subunits are individually unique, whereas the a-subunits have some common amino-acid but not carbohydrate sequences. Tertiary structures have yet to be studied seriously but physicochemical studies of human chorionic gonadotrophin indicate a spherical structure for the hormone.s6 G. Primary Structure-Activity Relationships of Hormonal G1ycoproteins.-The involvement of specific units within the primary structures of the molecules may be determined by combined chemical and biological techniques as already described.Earlier results for such studies on the gonadotrophins have been ~ummarized.~~~~~In general it has been found that removal of the N-acetyl- neuraminic acid units from follicle-stimulating hormone and human chorionic gonadotrophin destroy the biological (in vivo) but not the immunological activities. In contrast, removal of the same units from luteinizing hormone has no effect upon either activity of the hormone. Further investigation of the desialylization of human pituitary follicle-stimulating hormone showed that whereas the modified molecule is inactive in vivo it is biologically active in vitro, i.e. when presented direct to the target organ of the hormone, the ovary, the molecule still exhibits follicle-stimulating abilityeZ0 This was interpreted as indicating the necessity of the N-acetylneuraminic acid units only for transport of the hormone from the site of production to the site of action.Such a phenom-enon has subsequently been shown to be the general case for a number of glyco-proteins. Reduced, periodate-oxidized human pituitary follicle-stimulating hormone was found to be immunologically active but biologically (in vivo) inactive, suggesting that although the carbohydrate is essential for biological activity, it is not a requirement for immunological activity, which appears to be a function of the protein portion of the Modification of human follicle-stimulating hormone by its reaction with chloramine-T, diazosulphanilic b4 T.Inagami, K. Murakami, D. Puett, A. S. Hartree, and A. Nureddin, Biochem. J., 1972, 126, 441. s6 P. Rathnam and B. B. Saxena, in ‘Gonadotropins’, ed. B. B. Saxena, C. G. Beling, and H. M. Candy, Wiley-Interscience, New York, 1971, p. 120. .m K. F. Mori and T. R. Hollands, J. Biol. Chem., 1971,246,7223. 67 W. R. Butt, Acfa Endocrinol., 1969, 64, suppl. 142, p. 13. 37re Meldola Medal Lecture acid, maleic anhydride, citraconic anhydride, N-acetylimidazole, t-butyl azido- formate, and proteases, and by photo-oxidation has shown that lysyl, histidyl, methionyl, cysteinyl, and some tyrosyl residues are essential for manifestation of biological (in vivo) activities whereas arginyl, tryptophanyl, and phenylalanyl residues are not so in~o1~ed.~~~~~~~~Biological properties of human chorionic gonadotrophin after removal of the sialic acid and D-galactose residues have been investigated in detail.6o H.Secondary Structure-Activity Relationships of Hormonal G1ycoproteins.-The subunits already described for the hormonal glycoproteins have been extensively examined for their biological and immunological activities with respect to those of the parent intact molecules and such phenomena have been reviewed and summari~ed.~~~~~~~~The a-and p-subunits of human luteinizing hormone were found to possess some residual immunological activity and small amounts of biological acti~ity.~lv~l Recombination of the subunits was achieved by incubation together on a 1:1 basis whereupon both immunological and biological activities were restored. Analogous results have been obtained for follicle-stimulating horrn~ne~~,~~ From these studies and for human chorionic gonadotr~phin.~~,~~ and those on gonadotrophins and thyroid-stimulating hormone from non- human species it is now clear that the loss of immunological and biological activity on subunit formation, which is reversible, is a general property of the hormones. However, the small amount of residual biological activity in the separate subunits may be an artefact, owing to the presence of traces of the complementary subunit; this view is supported by the fact that the residual biological activity of HCG-p has been largely eliminated by further purification.68 It has also been found that the subunits of a particular hormone from various species are interchangeable so far as activity is concerned.More important, however, is the fact that hybrid gonadotrophins have been produced, e.g. the combination of HCG-awith BTSH-/3 gives a molecule with thyroid-stimulating hormone activity.gs Also, the a-subunits of human luteinizing and follicle- stimulating hormones and chorionic gonadotrophin can be substituted for one another in combination with the &subunit of each A pattern is now emerging in which the type of hormonal activity of the hybrid gonadotrophin is designated by the activity in which the p-subunit used was originally involved. m W. R. Butt, S. S. Lynch, M.F. Chaplin, C. J. Gray, and J. F. Kennedy, in ‘Gonadotrophins and Ovarian Development’, ed. W. R. Butt, A. C. Crooke, and M. Ryle, Livingstone, Edinburgh, 1970, p. 171. 69 J. F. Kennedy, S. Ramanvongse, W. R. Butt, W. Robinson, M. Ryle, and A. Shirley, in ‘Structure-Activity Relationships of Protein and Polypeptide Hormones’, ed. M. Margou- lies and F. C. Greenwood, International Congress Series 241, Part 2, Excerpta Medica, Amsterdam, 1972, p. 351. T. Tsuruhara, M. L. Dufau, J. Hickman, and K. J. Catt, Endocrinology, 1972,91,296. O1 A. Nureddin, A. S. Hartree, and P. Johnson, in ‘Gonadotropins’, ed. B. B. Saxena, C. G. Beling, and H. M. Candy, Wiley-Interscience, New York, 1971, p. 167. Oa P. Donini and S. Donini, personal communication, 1973.69 J. G. Pierce, 0. P. Bahl, J. S. Cornell, and N. Swaminathan, J. Biol. Chem., 1971, 246, 2321. O4 P. Rathnam and B. B. Saxena, J. Biol. Chem., 1971,246,7087. Kennedy This phenomenon has yet to be tested for all the human hormonal glycoproteins but nevertheless it seems quite certain that the /&subunits are hormone-specific whereas the a-subunits are interchangeable. Such a finding is in keeping with the similarities of the amino-acid sequences of the a-subunits but the unique characters of the /3-subunits. Apart from such work, it is also important to try and produce fragments of the hormones, which retain activity, thus permitting a localization of the active site(s) of the molecules. Some success has been achieved in this respect, fragments produced from human follicle-stimulating hormone retaining immunological activity and biological activity (in~itro).~~,~~ Conclusions and future objectives are included in the final section of this article, page 392.4 ProteoglycansA. Nomenclature of Proteoglycans.--On account of the apparent regularity of the polysaccharide chains in proteoglycans and the earlier belief that the protein present in preparations of the polysaccharide parts represented impurity, greater attention has been given to the polysaccharides themselves. Thus the poly- saccharides, glycosaminoglycans (earlier named acidic mucopolysaccharides), have been classified and named according to their monosaccharide contents and apparent repeating structures.In comparison with glycoproteins, the proteo- glycans are apparently few in type and present only eight glycosaminoglycan components:hyaluronic acid, chondroitin, chondroitin 4-sulphate (chondroitin sulphate A), chondroitin 6-sulphate (chondroitin sulphate C), dermatan sulphate (chondroitin sulphate B, /&heparin), heparin, heparan sulphate (heparitin sulphate), and keratan sulphate (keratosulphate). Earliere and more recenP aspects of the subject of the glycosaminoglycans have been reviewed. B. Composition of Proteoglycans.-The proteoglycans, as already implied, are all macromolecular and are composed of the naturally occurring amino-acids, hexuronic acid (or in one case hexose), 2-amino-2-deoxyhexose, acetate, and sulphate (in some cases) together with the following types of unit which may be present in small amounts : hexose, pentose, 6-deoxyhexose, sialic acid.The determination of amino-acid and 2-amino-2-deoxyhexose presents no problem, methodology being well established and a sensitive fluorimetric methode6ss7 being available for identification of the latter in microgram amounts. Similarly, determination of the hexoses, although present in small amounts, is easily performed and the gas-phase chromatographic behaviour of the pentoses as volatile derivatives has been studied in detail.68 Although determination of the uronic acid content by the modified carbazole procedure is a well-established O6 J. F. Kennedy, Biochem. SOC.Trans., 1973, 1, 807. 46 H. Cho Tun,J. F. Kennedy, M.Stacey, and R. R. Woodbury, Carbohydrate Res., 1969, 11,225. J. F. Kennedy, in ‘Automation in Analytical Chemistry’, Technicon, Basingstoke, 1973, p. 528. J. F. Kennedy, Chromatograph&, 1970,3,316. 375 The Meldola Medal Lecture technique, identification of the uronic acid is not so easy. These units may be partially identified either enzymically (see later) or by conversion into the corresponding hexose and identification as such.sQ Thegas-phasechromatographic and mass spectrometric behaviour of the uronic acids as volatile derivatives has also been studied as an aid to identifi~ation.~~ The determination of sulphate has also presented a problem for microscale work where the suIphate content is below the lower limit of the techniques, which depend on the formation of insoluble sulphates.A flame-photometric method based on determination of barium sulphate in the range in which it is soluble has been reported71 and it is to be hoped that an ion-selective electrode which is being developed for ~ulphate‘~could also be used in this field. Determination of the amino-acid composition is similar to that for any proteinaceous molecule, but as will be evident from the trend throughout this section, much less attention has been given to the amino-acid compositions of the proteoglycans. C. Occurrence, Function, and Use of Proteog1ycans.-The glycosaminoglycans of the proteoglycans occur generally in mammalian species. Of the glycosamino- glycans, hyaluronic acid was reported first, although ‘unwittingly’, since it was extracted, albeit in a crude form, from human umbilical cords as Wharton’s jelly in 1656.73In terms of the modern approach it was discovered in bovine vitreous humour in 1934,74and in the human the glycosaminoglycan is now known to be an important component of bone, fibroblasts, skin, aorta, and connective tissues, and also of two body fluids -synovial fluid and vitreous humour; it is also produced by a few Streptococcal bacteria.Chondroitin 4-sulphate was first isolated in 1861 from cartilageS although the workers were somewhat uncertain of themselves; chondroitin 6-sulphate was reported in 1936.76These two isomers usually occur together, e.g. in bone, cornea, aorta, umbilical cord, and connective tissue from humans and other species.Polysulphated variants of these two glycosaminoglycans occur predominantly in squid and shark cartilage. Chon- droitin itself has a more limited occurrence, being found in cornea.76 Dermatan ~ulphate~~is widely spread in cornea, sclera, skin, lung, aorta, and connective tissues, and has also been detected in a ba~terium.~~ Heparin,7Q heparan sulphate,*O and keratan s~lphate~~ are found extensively in connective tissues etc. and the latter at least is also a component of cartilage. Polysulphated varieties of keratan sulphate have been found in shark cartilage. The reports of the occurrence of R. L. Taylor and H. E. Conrad, Biochemistry, 1972, 11, 1383. 70 J. F. Kennedy and S. M. Robertson, unpublished results.71 S. A. Barker, J. F. Kennedy, P. J. Somers, and M. Stacey, Carbohydrate Res., 1968,7,361. 72 G. G.Guilbault, personal communication, 1973. 73 T. Wharton, ‘Adenographia : Sive Clandularum Totius Corporis Descriptio’, Londini, 1656. 74 K. Meyer and J. W. Palmer, J. Biol. Chem., 1934,107, 629. 75 K.Meyer and J. W. Palmer, J. Biol. Chem., 1936, 114, 689. 76 K.Meyer, A. Linker, E. A. Davidson, and B. Weissman, J. Biol. Chem., 1953,205, 611. 77 K.Meyer and E. Chaffee, J. Biol. Chem., 1941, 138,491. 78 G.K.Darby, A. S. Jones, J. F. Kennedy, and R. T. Walker, J. Bacteriol., 1970, 103, 159. 79 J. McLean, Amer. J. Physiol., 1916, 41, 250. J. E. Jorpes and S. Gardell, J. Biol. Chem., 1948, 176, 267. Kennedy glycosaminoglycans now run into hundreds, and, as already indicated, these molecules are essentially parts of the proteoglycans. The corresponding proteo- glycans have more recently been isolated intact from the same sources.The general function of the proteoglycans as essential components in the block building of the macromolecular framework of connective and other tissues from these molecules and collagen has already been cited. Hyaluronic acid, on account of the viscosity it imparts to its solutions, appears to act as a lubricant for joints and also possesses physicochemical characteristics which permit the solution to be fluid except under compression (shock) when it acts more as a shock-absorbing gel. On account of their macromolecular and charged natures, the proteoglycans may operate to perform filtration and exclusion of diffusable molecules.Beyond these, the true role of the proteoglycans in greater detail is uncertain since they do not appear to manifest biological, hormonal, or enzymic properties. Clinically, the glycosaminoglycans and proteoglycans are of little use in terms of therapy of non-production since most diseases known to involve them specific- ally produce them in excess. However, heparin alone has potent blood anti- coagulent properties and is therefore used extensively for the prevention of clotting. D.Isolation and Purification of Proteog1ycans.-The quantitative isolation of the glycosaminoglycans frequently presents a problem since the parent proteoglycans are often strongly associated with insoluble collagen fibres etc.Various techniques have been used for such cases including enzymic degradation with proteases and collagenases, and alkaline extraction. Some tissues are relatively easily solubilized but studies on human skin, chemically one of the most inaccessible tissues, have shown that extensive disruption of the fibres is essential to permit complete attack of de-bonding etc. reagentss1 A typical process to obtain a mixed glycosamino- glycan fraction involves the following: drying of tissue, solvent extraction of lipid, extensive homogenization, proteolytic digestion, brief alkaline treatment, denaturation of residual protein, and dialysis, the product being purified as described subsequently. Although isolation of the proteoglycans intact may be considered an easier task, de-bonding agents are still somewhat necessary.However, initial emphasis was again laid on disruption, and the extraction of tissue with water alone, with high speed homogenization, yielded a proteoglycan fraction.82 It was subsequently found that the use of salts greatly enhanced the yield of proteoglycan, up to 90 % of the total being extracted when guanidinium chloride was This is attributable to the fact that such an extractant causes dissociation of proteoglycan aggregates, and the process is therefore a dissociative extraction. Dissociative extraction is to be preferred since the macromolecules are more likely to remain intact. Ethylenediaminetetra-acetic acid-sodium chloride solution has also proved useful for the extraction of proteoglycans, and many other agents have *I S.A. Barker, J. F. Kennedy, and P. J. Somers, Carbohydrate Res., 1969, 10, 57. 8% J. Shatton and M. Schubert, J. Biol. Chem., 1954, 211, 565. a* S. W. S@deraand V. C. Hascall, J. Biol. Chem., 1969,244,77. The Meldola Medal Lecture been tested. The use of cathepsins holds potential. Generally, it has beenproposed that the extractability of a proteoglycan is inversely parallel to the protein content and is also dependent upon the age of the Having obtained an initial crude extract, it is necessary to achieve a fraction-ation into the various molecular types present. In the case of glycosaminoglycans, one might expect this to be relatively easy on account of their individual charge characteristics. However, such simple theory is overshadowed by molecular weight heterogeneity within a fraction and the presence of small amounts of residual peptide.Numerous techniques, including selective precipitation, thin-layer chromatography, and isoelectric focusing, have been employed. No one met hod separates completely all glycosaminoglycans but ionexchange chromato- graphy (see ref. 81), electrophoresis (see ref. 85), and the cetylpyridinium chloride-cellulose column techniques6 are probably the best, the latter method being based on the fact that glycosaminoglycans can be partially separated by precipitation as their cetylpyridinium Gel filtration may prove useful at various stages in a purification procedure, but recent studies have shown that the glycosaminoglycans can become associated with some gel-filtration matrices.88 However, none of these techniques should be regarded as doing more than providing a means of separating the glycosaminoglycans, structural analysis being necessary for complete identification (see later).In the case of proteoglycans, a combination of selective precipitation and high speed centrifugation has been used extensively and separates the proteoglycans into heavy (PPH)and light (PPL) fractions, e.g. ref. 89. Fractionations of proteoglycans and derived subfractions may also be achieved by equilibrium densi ty-gradient ultra- centrifugation, gel filtration, and ion-exchange chromatography. E.Primary Structures of the Glycosaminoglycan Components of Proteoglycans.-The glycosaminoglycans may be distinguished by their compositions and primary structures, since repeating disaccharide structures have been discovered as general features of all of them. Thus the repeating unit of hyaluronic acid is (6), the molecule containing no sulphate. Chondroitin, the only other non-sulphated glycosaminoglycan, is an isomer of hyaluronic acid, the 2-amino-2-deoxyhexose component of the disaccharide repeating unit (7)having theD-gdactu- rather than the D-gluco-configuration. As implied, chondroitin 4-sulphate and chondroitin 6-sulphate are sulphated variants of chondroitin, the sulphate ester groups being situated on the 2-amino-2-deoxy-~-galactoseunits [(8) and (9), respectively].Dermatam sulphate (10) is an isomer of chondroitin dsulphate in which the uronic acid has the L-idu-rather than the D-gluco-configuration. As with other, perhaps better known, polysaccharides, irregularities have been observed in 84 A. A. Hallbn, in ‘Chemistry and Molecular Biology of the Intercellular Matrix’, ed. E. A. Balazs, Academic Press, London. 1970, vol. 2, p. 903. P. W. Lewis, J. F. Kennedy, and N. D. Raine, Biochem. SOC.Trans., 1973,1,844. J. Svejcar and W. Van B. Robertson, Analyt. Biochem., 1967, 18, 333. J. E. Scott, Methods Biochem. Analysis, 1960, 8, 145. 88 J. F. Kennedy, J. Chromatog., 1972, 69, 325. 80 R. M. Mason, Biochem. J., 1970,119, 599. Kennedy CO2H CH20H OW!SO’’HO OH NHAc *o&o, OH ’0 HO OH NHAc OSO3H ’o*o&o&dHO OH NHAc NHAc (10) 379 The Meldola Medal Lecture these primary structures.For example, using hyaluronidase, an enzyme which specifically cleaves 2-acetamido-2-deoxy-~-~-glucopyranosyl-(1+-4)-D-g~UCwOniC acid bonds in glycosaminoglycan structures, it has been found that the wonic acid units of dermatan sulphate may occasionally have the D-gluco-rather than the L-ido-conf?guration.Qo~Q1Degradation into disaccharide units with other specific enzymes has shown that the repeating disaccharide structures of the chondroitin sulphates may occasionally be di- or non-sulphatedQ2 whereas the L-iduronic acid units of dermatan sulphate may bear a sulphate group in addition to the normal sulphate group on the 2-amino-2-deoxy-~-galactoseunit.O3 Chondroitin 4- and 6-sulphate-type and dermatan sulphate-type structures also exist as copolymers, the heterogeneous molecules containing equimolar pro- portions of the standard disaccharide repeating units for these three glycosamino- glycan~.@~Dermatan and ‘dermatan 6-sulphate’ have not yet been reported as occurring naturally, although dermatan sulphate may be desulphated synthetic- ally.The structures of the repeating units of heparin, heparan sulphate, and keratan sulphate are, thus far, less clearly defined. It has, however, been established that the repeating structures for heparin and heparan sulphate possess a general disaccharide repeating sequence of -4)-hexopyranuronosyl-( 1-4)- 1(2-amino-2-deoxy-a-~-g~ucopyranosyl)-( and keratan sulphate of 43)-hexopyranosyl-(1-+4)-(2-amino-2-deoxy-~-glucopyranosy~)-(1-+.Structures (1 1) /02y+o*o OSO~H NH C02H /‘SO3H (1 1) OH /S03H (12) O0 L.-A.Ftansson and L. Rodh, J. Biol. Chem., 1367,242,4161. m L.-A. Fransson and L. Rodtn, J. Biol. Chem., 1967,242,4170. Ba K. Murata, T. Harada, T. Fujiwara, and T. Furuhashi, Biochim. Biophys. Acfa,1971,230, 583. O3 S. Suzuki, H. Saito, T. Yamagata, K. Anno, N. Seno, Y.Kawai, and T. Furuhashi, J. Biol. Chem., 1968, 243, 1543. O4 L.-A. Fransson and B. Havsmark, J. Biol. Chem., 1970,245,4770. 0, NH Kennedy and (12) have been found to contributeas~s6 to the overall structure of heparin but the structure appears to be much more complex, if not random, in terms of the location and abundance of the sulphate groups and also the configuration of the uronic acid units.Until quite recently, heparin was considered to contain D-glucuronic acid exclusively as the uronic acid. However, L-iduronic acid has been isolated from pure heparin via mild acid hydr~lysis,@~ additional experiments establishing that it was neither derived from contaminant dermatan sulphate nor through epimerization at C-5 of D-glucuronic acid. Sulphation of the 2-amino-2- deoxy-D-glucose residues of heparin occurs mainly at C-6,#*whereas the uronic acid residues are predominantly non-sulphated. Non-sulphated and 3,6-di-sulphated 2-amino-2-deoxy-~-g~ucoseresidues also occur in small proport ions.a @ The 2-amino-2-deoxy-~-glucoseunits of heparan sulphate are N-sulphated and less frequently O-sulphated,fOO~lol and nitrous acid degradationloa and enzymic hydrolysis103 studies have led to the suggestion of some repeating monosaccharide sequences. Structure (13) contributes to the overall structure of keratan sulphate, the 2-amino-2-deoxy-~-g~ucoseresidues sometimes bearing 6-0-sulphate groups in addition to the hexose nit^.^^^^^^^ Occasionally the hexose units may be non-sulphated.106 Although this glycosaminoglycan is generally assumed to contain D-galactose as its constituent hexose, it is frequently reported to contain mannose, fucose, and sialic acid.lo7 Apart from the glycosaminoglycans, the primary structures of the proteo- glycans have been given little attention, apart from determining the amino-acids involved in the glycopeptide linkage regions (see later).-0 HO OH NHAC IMA. S. Perlin, D. M. Mackie. and C. P. Dietrich, Carbohydrate Res., 1971, 18, 185. T. Helting and V. Lindahl, J. Bid. Chem., 1971, 246, 5442. s7 M. L. Wolfrom, S. Honda, and P. Y. Wang. Carbohydrate Res., 1969, 10, 259. M. L. Wolfrom, P. Y. Wang, and S. Honda, Curbohydrure Res., 1969, 11, 179. I. Danishefsky, H. Steiner, A. Bella, and A. Friedlander, J. Biol. Chem., 1969, 244, 1741. looJ. A. Cifonelli and J. King, Biochim. Biophys. Acru, 1970, 215, 273. lol C. P. Dietrich, H. B. Nader, L. R. G. Britto, and M. E. Silva, Biochim. Biophys.Actu, 197 1,237,430. lol J. A. Cifonelli, Carbohydrate Res., 1968, 8, 233. lo*A. Linker and P. Hovingh, Biochim. Biophys. Acra, 1968, 165, 89. lo4 V. P. Bhavanandan and K. Meyer, J. Biol. Chem., 1967,242,4352. lo6 V. P. Bhavanandan and K. Meyer, J. Biol. Chem., 1968,243, 1052. lo' J. Hirano and K. Meyer, Biochem. Biophys. Res. Cornrn., 1971,44, 1371, lo' N. Toda and N. Seno, Biochim. Biophys. Acta, 1970,208,227. lXe Meldola Medal Lecture F. Identification of the G1ycosaminoglycans.-The basic disaccharide repeating structures of the glycosaminoglycans were originally determined by classical carbohydrate chemistry. The relatively large scales employed were possible since the tissue sources were non-human. Subsequently, in investigating new sources of glycosaminoglycans, workers have relied either on compositional data obtained via acid hydrolysis or on comparison of chromatographic data for the unknowns with that of ‘standard preparations’. With the possible excep- tion of the simplest glycosaminoglycans, such methods are imprecise for at least two reasons.Compositional data may be misleading owing to the difficulty in determining individual uronic acids108 and owing to the variation in the degree of sulphation which may occur. Comparative chromatography may lead to a false identification since the ‘standards’ employed may have special characteristics related to the source from which they were prepared, and may in fact differ markedly in composition and molecular weight, and therefore physicochemical properties, from the unknown molecule under investigation.The only completely satisfactory method of individual measurement and identification is one based on structural recognition, and one which is applicable on a microgram scale to cope with the fact that often only milligram amounts of human tissue are available. New methods for the microscale chemical identifications of hyaluronic acid, chondroitin, chondroitin 4sulphate, chondroitin 6-sulphate, and dermatan sulphate have therefore been developed.109 Hyaluronic acid is degraded with hyaluronidase, the products being further degraded with p-D-glucuronidase and p-D-acetamidodeoxyglucosidase to a characteristic disaccharide (14). Chon-droitin, chondroitin 4-and 6-sulphates, and dermatan sulphate are distinguished with hyaluronidase and treated with chondroitin sulphate lyases and chondro- sulphatases from Proteus vulgaris to give a sulphate-free unsaturated disaccharide (15), which can be identified via specific periodate oxidation to p-formylpyruvic acid.The positions of linkages in the disaccharides (14) and (1 5) may be estab- lished by specific photometric analysis. Component analyses are combined with these specific methods, and new more-sensitive techniques were developed for the determination of sulphate71 (by flame photometry) and quantitative identifica- tion of the 2-arnin0-2-deoxyhexoses~~~~~(by fluorimetry). These combined techniques, which are specific for the identification and characterization of the C02H HO OH AcNH LOa J.F. Kennedy, Biochem. SOC.Trans., 1974, 2, in the press. loBS.A. Barker, J. F. Kennedy, and P. J. Somers, Carbohydrate Res., 1968, 8,482. Kennedy OH OH NHAC individual gIycosaminoglycans, require only 150 micrograms total of each glyco- saminoglycan. The chondroitin sulphate lyases and the chondrosulphatases have also been applied to chondroitin sulphates and dermatan sulphate by otherslloJ1l to differentiate between them, the products being identified by chromatographic comparison with standards. Application of the microscale techniques permitted the identification of three types of hyaluronic acid, chondroitin 4-and 6-sulphates, and dermatan sulphate in human skin.*' However, it occurred to us that such tissues might contain only nanogram amounts of proteoglycans, which would escape identification by the microgram techniques.Methods were therefore developed to label the glycos- aminoglycans by culture of the living tissue in the presence of radioactive pre- cursors,lla-lla the glycosaminoglycans being subsequently separated and being monitored for radi~activity.~~' Furthermore, by careful selection of the labelled precursor used in the culture medium (e.g. D-glucuronic acid, D-galactose, sulphate), it has been found that some specificity of labelling of certain glycos- aminoglycans can be achieved. Such studies permitted the additional identifica- tion of heparin and kerdtan sulphate in human skin and differences in the glycos- aminoglycan distribution between the dermal and epidermal layers.118 G.Primary Structures of the Glycopeptide Linkages of Proteog1ycans.-A common glycopeptide linkage sequence (16) has been found for chondroitin C02H 1NH OH HO (16) OH OH INH I T. Yamagata, H. Saito, 0. Habuchi. and S. Suzuki.J. Biol. Chem., 1968, 243, 1523. ln H. Saito, T. Yamagata, and S. Suzuki, J. Biol. Chem., 1968, 243, 1536. 11* S. A. Barker, J. F. Kennedy. and C. N. D. Cruickshank, Carbohydrate Res., 1969, 10, 65. llS J. F. Kennedy, J. Labelled Compounds, 1970, VI, 201. 11* J. F. Kennedy, Experienria, 1969, 25, 1120. me Meldola Medal Lecture dsulphate and heparin,llsJl6 chondroitin 6-sulphate,l17 and dermatan sul- phate118J1D and, as will be seen, this involves neutral sugar residues which are quite different from the normal monosaccharide units of the glycosaminoglycans themselves.The linkage between the D-xylose residue and serine is of the 0-glycosyl type to the hydroxy-group of the amino-acid. In the case of keratan sulphate the linkage is quite different (17),I2O does not involve atypical carbo- hydrate units, and is of the glycosylamine type to the side-chain of asparagine. The overall picture for the involvement of such glycopeptide linkages in proteo- glycans is gradually emerging and appears complex since the proteoglycan protein backbone may bear more than one type of glycosaminoglycan. Chon- droitin 4-sulphate and keratan sulphate frequently occur together in this way.121v122 Although hyaluronic acid isolated by non-degradative methods is known to contain protein, it is not certain that the overall protein-containing molecule is analogous to the other proteoglycans.Also in contrast, the glycopeptide linkage of hyaluronic acid has not been elucidated, although it has been suggested that this involves neutral carbohydrate residues. I NH I NHAc c=o I I(17) H. Secondary and Tertiary Structures of Glycosaminoglycans and Proteoglycans.- The proteoglycans are generally considered to lie in the tissue matrix with their backbones parallel to collagen fibres, with the pendant polysaccharide chains forming non-covalent bonds with the collagen, so forming a three-dimensional network. Many model systems have been devised to simulate and elucidate in greater detail the tertiary structures of the proteoglycans in the natural state, but at the moment no clear solution to the problem is available.However, evidence has been for the existence of specialised crosslinks between the proteoglycans themselves and of subunit secondary structures, whilst some information is forthcoming from ultracentrifuge data. A'-Ray crystallographic studies of the simpler glycosaminoglycuronans have, however, reached a more 116 U.Lindahl, Biochim. Biophys. Ada, 1966, 130, 268. 116 U.Lindahl and L.Rodkn, J. Biol. Chem., 1966. 241, 21 13. n7T.Helting and L.Rodh, Biochim. Biophys. Acru, 1968, 170,301. 11* A.Bella and 1. Danishefsky, J. Biol. Chem., 1968. 243, 2660.119 E.L.Stern, J. A. Cifonelli, L.-A. Fransson, B. Lindahl, L. Roden, S. Schiller, and M. L. Spach, Arkiv Kemi, 1969, 30, 583. la"H. W. Stuhlsatz, R. Kisters, A. Wollmer, and H. Greiling,Z.phySiol. Chem., 1971,352,289. 121 K.D.Brandt and H. Muir, Biochem. J., 1971, 123, 747. Ira H. Lyons and J. A. Singer, J. Biol. Chem., 1971,246,277. 123 M. B. Matthews, Biochem. J., 1971, 125, 37. Kennedy advanced ~tate~~~~~~~ and it is now possible to attempt rationalizations of the tertiary structures of the molecules in the crystalline state in terms of some of the functions of the molecules. I. The Involvement of Proteoglycans in Diseases.4n account of their funda- mental function in tissue structure, it is not surprising that the proteoglycans are involved chemically in conditions which camelgive rise to changes in tissue.The glycosaminoglycans have been examined in disease, but on account of the problem of availability of tissue from living subjects many investigations have relied only on histological methods. It has not been until the advent of the micro-techniques that differences could be examined in detail and on a quantita- tive and structural basis. Also, using the previously cited radioactive incorpora- tion techniques,f1z it has been possible to demonstrate glycosaminoglycan disorders in various tissue conditions and to monitor correction of the disorders during treatment.f2s-f2s A further approach has been the application of the radioactive incorporation method to the testing of drugs, etc, for their action and side-effects on the structure, biosynthesis, and metabolism of tissue as judged by glycosaminoglycan biosynthesis and metabolism.12B In view of the versatility of the radioactive incorporation technique, which can be carried out on a biopsy specimen of five milligrams, it holds potential for further development in the understanding of tissue in health and disease.A group of diseasesf3* (mucopolysaccharidoses) in which various glycosamino- glycans with peptide attached are excreted in excess (glycosaminoglycanuria) has been given more attention on account of the large and easily collected amounts of glycosaminoglycans produced. However, clinical distinction between the various mucopolysaccharidoses is somewhat difficult, and therefore a chemically based diagnosis is required.Many qualitative and quantitative glyco- saminoglycan tests have been reported, but research in this area has more recently been aimed at reproducible qualitative tests and methods of analysis for screening infantsss and identification on the basis of chemical A further approach to an understanding of the involvement of proteoglycans in disease and their faulty metabolism lies in simulation of the abnormal processes involved. In association with certain arthritic conditions in which hyaluronic acid chains become degraded, the free-radical degradation of the glycosamino- glycuronan with ferrous ions has been investigated.132 (Conclusions and future objectives are included in the final section of this article, page 392.) 124 E.D. T. Atkins, C. F. Phelps, and J. K. Sheehan, Biochem. J., 1972,128, 1255. 126 E. D. T. Atkins and J. K. Sheehan, Nature New Biol., 1972, 235, 253. 126 S. A. Barker and J. F. Kennedy, Life Sciences, 1969, 8, Pt. JI, p. 989. lZ7 E. J. Moynahan and J. F. Kennedy, Proc. Roy. SOC. Med., 1966,59, 1125. 12* E. J. Moynahan and J. F. Kennedy, ‘XI11Congressus Internationalis Dermatologiae -Miinchen, 1967’, Springer-Verlag, Berlin, 1968, p. 1543. lZg S. A. Barker and J. F. Kennedy, Carbohydrate Res., 1969, 11, 27. lS0V.A. McKusick, D. Kaplan, D. Wise, W. B. Hanley, S. B. Suddarth, M. E. Sevick, and A. E. Maumanee, Medicine, 1965,44,445. lS1 J. F. Kennedy, C. H. Sinette, and J. B. Familusi, Clinica Chim.Acra, 1970, 29, 37. lSp J. F. Kennedy and H. Cho Tun,Carbohydrare Res., 1972,22,43. 385 2 me Meldola Meclbl Lecture 5 Water-insoluble Biologically Active Reactors The chemical attachment of molecules, particularly biologically active macro- molecules, to water-insoluble matrices has received considerable attention in the past decade, and the products of such attachments have been put to a number of uses, the most common of which is as a new phase-type of biologically active Gompound. However, one must not be led by all the extensive literature which has now been published on the chemical production of such compounds to think that the principle of insolubilization is something new. The overall principle of attachment of a biologically active molecule to an insoluble matrix is simple and simulates the natural mode of action and environment of enzymes, antibodies, antigens, etc.which are carried on the surfaces or in the interiors of cells, or which are embedded in biological membranes and tissues. Indeed, as is often discovered, ‘Creation was there first’. In fact, it may be said that in the human the greater proportion of the biologically active molecules of the body exist at some time in insolubilized form. Apart from all the molecdes present in the body known to have biological activity, many if not all of the others can be regarded as insoluble biological reactors of some description, e.g. the proteoglycans in their tissue matrix-forming role, However, perhaps less attention has been given to the insoluble forms rather than the soluble forms of such molecules since most chemical techniques, analyses, and manipulations are designed to be carried out in solution, and the chemistry of activity in the solid phase is less well developed.In natural systems, the insolubilization of biologically active macromolecules such as enzymes and glycoprotein hormones may well be a reversible process, according to whether the macromolecule is originally synthesized in the solid or liquid phase. However, it is quite certain that insolubilized forms of such active macromolecules easily become converted into soluble forms to be transported to a new site at which they perform their function. In this respect the natural insolubilized molecules differ markedly from those produced in the laboratory.This is because synthetically insolubilized biologically active molecules are usually required to perform their biological function without being released into the surrounding solution and thereby contaminating it. As already indicated, there are many applications of insolubilized biologically active molecules. Insolubilized enzymes are principally used to effect the reaction catalysed by the free enzyme, but in a simplified form since the enzyme (insoluble) can be very easily and simply removed from the substrate and products (soluble) by filtration or centrifugation, whereas use of the soluble enzyme in the con- ventional fashion requires subsequent laborious separation of the enzyme from the products by, for example, gel filtration and ion-exchange chromatography.Further advantages of insolubilized enzymes are that an enzyme is often stabilized to decomposition in storage and to heat on insolubilization, and that such enzymes can be re-used. On these accounts, insolubilized enzymes also lend themselves to application as reactors in continuous analytical and industrial processes. Insolubilized antibodies are principally useful for the purification of homo- Kennedy logous antigens, usually by a type of column chromatography (immunoadsorp- tion) in which the solution of impure antigen is passed through a bed of insolubil- ized antibody: the specific antigen is adsorbed by the antibody whilst impurities are washed through the column.Subsequently, the antigen may be desorbed from the column in pure form. Thus the lengthy conventional techniques of various types of column etc. chromatography are short-circuited. Immunoadsorption can of course also be applied in the reverse sense, using insolubilized antigen to purify an antibody. Immunoadsorption is a very versatile technique since many macromolecules are antigenic and therefore antibodies can be raised to them; but an important prerequisite is of course that the antigen that is to be insolubil-ized be obtained in pure form or that the antibody to be insolubilized be obtained in pure and/or highly specific form. Insolubilized antigens and antibodies are also of use in radioimmunoassay techniques, as described earlier for the radio- immunoassay of follicle-stimulating hormone (see page 366).Analogous to immunoadsorption is the technique of affinity chromatography, a technique which can be applied to the purification of enzymes, etc. Here the insolubilized molecule is usually one of low molecular weight but one for which the macromolecule to be purified has a specific affinity. Nucleic acids may also be used in an analogous way for the purification of complementary nucleic acids. Insolubilized molecules may be useful as enzyme substrates in the simplification of enzyme assays. The most recent innovation in the field of insolubilization has been the preparation of insolubilized antibiotics. When attaching a biologically active molecule to an insoluble support, it is important to avoid a mode of attachment that reacts with or disturbs the active site(s) of the molecule, as otherwise a loss of activity will result on binding.It is also important to avoid overloading the matrix when binding molecules, since overloading leads to overcrowding and hence reduced activity by reason of steric hindrance of approach of the substrate etc. molecules to the active sites of the bound molecules. Attention to the way in which the macromolecule can be attached to the insoluble matrix and the choice of matrix is also a matter of importance. A number of matrix types have been used in the field of insolubiliza- tion, and polysaccharide derivatives have been used extensively.Methods of insolubilization of enzymes, antigens, antibodies, nucleic acids, antibiotics, and affinants and descriptions of the insolubilized molecules have been reviewed in detail in the sister article to to which the reader is referred for such information and for greater details of the principles of insolubilization. Presently, our own approach to the field of insolubilization will be described. We have been concerned with the preparation of a matrix suitable for insolubilization of a wide range of types of molecule with a view to providing a basis for (i) enzyme insolubilization for the use of enzymes in studies of carbo-hydrate unit sequences in glycoprotein structures and for simplification of industrial processes, (ii) glycoprotein purification by immunoadsorption, (iii) immunoglobulin purification by immunoadsorption, (iv) antigen and ''' J.F.Kennedy, Adv. Carbohydrate Chem. Biochem., 1973,29,in the press. The Meldola Medal Lecture antibody insolubilization for rapid, specific, and accurate radioimmunoassay techniques for clinical assays, (v) affinity chromatography in relation to carbo- hydrate structures, and (vi) antibiotic insolubilization, with a view to paving the way to a number of applications. Of the general modes available -covalent attachment, non-covalent attach- ment and physical entrapment/inclusion in a fibre or lattice, or cross-linking of the molecule itself to form a matrix -covalent attachment is to be preferred since it is more permanent and the attached molecule is unlikely to be inadvertently released into solution during use.We have therefore followed the lines of a covalent-type insolubilization and have given attention to the derivatization of the insoluble polysaccharide cellulose with ethyl chloroformate to give the reactive cellulose truns-2,3-carbonate (1 8).lS4Using microcrystalline cellulose, the conditions for achieving a maximum degree of substitution with the cyclic group and a minimum degree of substitution with the acyclic carbonate (ethyloxy- carbonyl) group which occurs as a side-reaction were defined.13rJ35 The strain and electronic arrangement in the ground state of a trans-1,2-carbonate ring fused to a six-membered ring in the chair conformation are such that the carbonyl carbon atom of the carbonate ring is susceptible to nucleophilic attack.Thus cellulose truns-2,3-carbonate was shown to exhibit suitable reactivity for the covalent coupling of enzymic protein under mild aqueous conditions to give an active insoluble derivative of ~-~-glucosidase.~~~~~~~From a study of the reaction of simple nucleophilic compounds with cellulose trun~-2,3-carbonate~~~ it is envisaged that the cbupling reaction involves the nucleophilic attack of a free amino-group in the enzyme etc. protein on the strained trans-cyclic groups (Scheme 1). As can be seen, the carbonate ring may open in two ways to give products in which the D-glucopyranose ring is substituted at the 2- or 3-position. The initial successes in forming a stable covalent bond between cellulose tr~ns-2~3-carbonateand an enzyme were magnified by the fact that, contrary to expectation, the optimum pH for the covalent coupling reaction is approximately pH 7.8, and not considerably higher, as might be expected for a nucleophilic reaction.This was of course very important in avoiding, during the coupling, damage to biologically active molecules by subjecting them to other than physio- logical (near neutral) pH values. Thus further studies were carried out to optimize the coupling conditions, and highly active derivatives of several enzymes have now been prepared.lSg Cellulose truns-2,3-carbonate has also proved very useful in the insolubilization of human immunoglobulins for purification of antibodies to them,loo and its properties for this immunoadsorption were found to surpass la' S.A. Barker, H. Cho Tun, S. H. Doss,C. J. Gray, and J. F. Kennedy, Carbohydrate Res., 1971, 17,471. lS6S. A. Barker, J. F. Kennedy, and C. J. Gray, B.P. 1289548. lSo S. A. Barker, S. H. Doss, C. J. Gray, J. F. Kennedy, M. Stacey, and T. H. Yeo, Carfm hydrare Res., 197 I, 20, 1. S. A. Barker, J. F. Kennedy, and C. J. Gray, B.P. 1289549. 13* J. F. Kennedy and H. Cho Tun,Carbohydrate Res., 1973, 29,246. 139 J. F. Kennedy and A. Zamir, Carbohydrate Res., 1973, 29,497. ld0D. Catty, J. F. Kennedy, R. Drew, and H. Cho Tun, J. Immunological Methods, 1973, 2,353. Kennedy (18) I”+ p”+ R = rest of molecule being insolubilized Scheme 1 those of other matrices which could be used for insolubilization.Purification of these antibodies is important for structural investigation and for investigation of antibody specificity and reactivity with respect to health and disease. Both human pituitary follicle-stimulating hormone and its homologous antibody have been covalently attached to cellulose ~rans-2,3-carbonate.~~ The anti-follicle-stimulating hormone retains its immunological reactivity on insolubilization, and is therefore suitable for use in the solid-phase radio- immunoassay of unknown amounts of the hormone by competitive binding of The Meldola Medal Lecture radioactively labelled and unlabelled hormone. Acceptable inhibition curvescan be obtained, and the low, non-specific adsorption characteristics have advantages over other systems.Follicle-stimulating hormone itself also retains immuno- logical reactivity on insolubilization, and this derivative holds potential for the radioimmunoassay of the hormone as it can be layered immunologically with anti-follicle-stimulating hormone and then the hormone itself. The comparatively recent discoveries of myeloma forms of human immunoglobulin IgE have made possible the development of a sensitive radioimmunoassay for estimating antibody IgE in normal and pathological sera and other body fluids with a view to diagnosis of myeloma conditions. Antibodies, to immunoglobulin IgE, attached to cellulose trans-2,3-carbonate have been applied in this respect.141 Work is presently in hand for the utilization of antigens and antibodies attached to cellulose trans-2,3-carbonate for other, automated clinical chemical analyses and diagnoses.The coupling of a number of antibiotics to cellulose trans-2,3-carbonate under a series of coupling conditions has been investigated, and it has been shown that by such couplings active insoluble derivatives of antibiotics can be pro- du~ed.'~~Such pioneer experiments are particularly exciting since they open the way to the production of antimicrobial surfaces, coatings, dressings, etc. for use in clinics and hospitals and for the protection of industrial membranes etc. It was also found that the antibiotics became firmly bound to cellulose itself, whereas use of the cellulose trans-2,3-carbonate extended the range of antibacterial activity retained.In certain cases slow release of the antibiotic from the matrix occurred when cellulose was used, and this phenomenon holds potential in the development of slow-release anti biotic and drug formulations. On the industrial side, since cellulose is the chemical basis of many packaging materials, it would appear that long-lasting antimicrobial protection may be afforded by a single treatment with antibiotic solution. Finally, the use of cellulose trans-2,3-carbonate derived from microcrystalline cellulose in the field of insolubilization of affinants typified by the phytohaemag- glutinin concanavalin A has been in~estigated.'~~ Thus cellulose trans-2,3-carbonate has proved to be a very versatile material for insolubilization. However, one disadvantage to be expected from the use of a standard powder for insolubilization of macromolecules is that substrate macromolecules cannot rapidly diffuse into the particles.For this reason we have extended the deri- vatization of cellulose to its trans-2,3-carbonate to macroporous cellulose in which there is a gel-type network that is permeable to macromolecules. Deri-vatization of such macroporous cellulose with retention of macroporosity has been achieved,ldq the degree of substitution with the cyclic carbonate groups being controlled by the addition of water to the reaction medium. Thus it has 141 P. McLaughlan, D. R. Stanworth, J. F. Kennedy, and H. Cho Tun, Nature New Biol., 1971, 232, 245.x's J. F. Kennedy and H. Cho Tun,Antimicrobial Agents and Chemotherapy, 1973,3, 575. wa N. Ling, J. Bray, and J. F. Kennedy, unpublished observations. 144 J. F. Kennedy, S.A. Barker, and A. Rosevear, J.C.S. Perkin I, 1973,2293. Kennedy been possible to achieve an insolubilized enzyme with a high retention of the natural activity against a substrate of high molecular Cyclic carbonate derivatives of other polysaccharides and cycloamyloses, including inulin cyclic carbonate, which involves some novel carbonate rings, have also been formed146 and some of these may have application in the attachment of enzymes to charged matrices for displacement of the enzyme pH optimum by microenvironmental effects. Mixed enzyme derivatives of other cellulose deriva- tives have also been prepared.l*' Whereas the use of a polysaccharide-type material for insolubilization is to be considered advantageous since in the solid state residual hydroxy-groups provide a hydrophilic environment for the attached macromolecule, and insolubilization within a macroporous structure gives added protection from exposure, poly- saccharides cannot be expected to be universally applicable since they are bio- degradable.We have therefore paralleled our approach to insolubilization by devising some non-biodegradable matrices. One of these, poly(ally1 cyclic carbonate)148(19), is the product of subjecting poly(ally1 alcohol) to the ethyl Jn (19) r CHz-CH-CH2-CHI CH2I 0 i c=o II NHRScheme 2 146 J.F. Kennedy and A. Rosevear, J.C.S.Perkin I, in the press. 146 J. F. Kennedy and H. Cho Tun, Carbohydrate Res., 1973,26,401. lP7 J. F. Kennedy, Z. klin. Chcm. klin. Biochem., 1971,9,71. lo8S. A. Barker,J. F. Kennedy, and A. Rosevear,J. Chem. Soc. (0,1971,2726. 391 The Melcibla Medal Lecture chloroformate reaction and it reacts with enzymes148J49 in an analogous way to cellulose truns-2,3-carbonate (Scheme 2). Poly(4- and 5-acrylamidosalicylic (20) have proved versatile matrices, being suitable for antibiotic ins~lubilization,’~~and enzyme insolubilization via metal ion hel la ti on.^^^ They are also applicable to selective extraction of certain metal ions from ~olution.~6~ Enzyme insolubilization, using alginic acid, chitin, and Celite as matrices,16* and antibiotic insolubilization, using cellulose as matrix,lSS have also been achieved via metal ion chelation, whilst studies with glass have shown that care must be exercised in handling of enzyme solutions in glass since the enzyme may become adsorbed with retention of activity.lSa (Conclusions and future objectives are included in the following section.) 6 Conclusions and Future Objectives Whilst it is clear that the present state of chemical knowledge of glycoprotein hormones and proteoglycans is well advanced, it is also certain that there is a considerable amount of work to be done, involving further development of microscale analytical techniques suitable for dealing with minute samples from humans, improved and more direct purifications, primary, secondary, and tertiary structural identification, and determination of chemico-biological relationships and biosynthetic pathways.In terms of improved and automated component analyses, our interests have required us to devise a number of techniques, principally for carbohydrate components. Carbohydrates are frequently separated and identified by ion-14@ J. F. Kennedy, s.A. Barker, and A. Rosevear, J.C.S. Perkin I, 1972,2568. lb0J. F. Kennedy, S. A. Barker, J. Epton, and G. R. Kennedy, J.C.S. Perkin I, 1973, 488. 161 J. F. Kennedy, J. Epton, and G. R. Kennedy, Antimicrobial Agents and Chemotherapy, 1973, 3, 29. J. F. Kennedy and J. Epton, Carbohydrate Res., 1973,27, 11.lS3 J. F. Kennedy, S. A. Barker, A. W. Nicol, and A. Hawkins, J.C.S. Dalton, 1973, 1129. 164 J. F. Kennedy and C. E. Doyle, Carbohydrate Res., 1973,28,89. lS6 J. F. Kennedy and A. Zamir, unpublished observations. u6J. F. Kennedy and P. M. Watts, Carbohydrure Res., 1974,32,155. Kennedy exchange chromatography (see ref. 108 for review) and by gas-phase chromato- graph~.~~~In connection with ion-exchange chromatography we have been interested in separations that may be carried out in water alone, and have extended the use of an ion-exchange resin as an ion-exclusion matrix to the separation of carbohydrates according to molecular weight,168 the pores of the resin being utilized as in gel filtration. Other applications of ion-exchange chromatography have been the sensitive identification of 2-amino-2-deoxy- hexose~~~~~~ Improved separations and sensitivities and erythritol and threit01.l~~ demand improved and sensitive automated analyses and such spectrophotometric methods have been reviewed.lo8 An even more sensitive general fluorimetric method for carbohydrates, amino-acids, peptides, proteins, glycoproteins, and proteoglycans has been devel~ped.~~,~~ Spectrophotometric determinations for phosphatela0 in monitoring columns in glycoprotein etc.fractionations and for formic acidlB1 in structural analysis by periodate oxidation have also been reported. In connection with gas chromatographic analysis we have been concerned with the separation of pentoses,B* of methylated monosaccharide^^^ in structural analysis by methylation, of hexoses and periodate oxidation in structural analysis, and of uronic acids,7o each compound being pre-derivatized to render it volatile.Mass spectrometry of the compounds coupled with gas-phase chromatography has provided additional identification data, and breakdown patterns have been in~estigated.~~~~~ In view of the increasing need for component analyses of glycoproteins, but the high cost of automated equipment, we have established and set up in our laboratory the University of Birmingham Macromolecular Analysis Centre. The purpose of this Centre is to provide a service of carbohydrate and amino- acid analysis to all members of the University. The Centre uses a Locarte analyser for amino-acid analysis and a Jeol JLC-6AH analyser for carbohydrate analysis : both machines operate on ion-exchange chromatography and are fully auto- matic.Programming work is in hand to process data and results from the two analysers to customer requirements, the two machines being on line to a Nova 1220 Mini-Computer in the Centre. It is hoped shortly to commission a peptide synthesizer and ultimately to extend the service to peptide sequencing. So far as better purifications are concerned, it appears at the moment that immunoadsorption and affinity chromatography will be of use here and will reduce the complexity of many purification procedures, including those for hormonal glycoproteins and proteoglycans. In the case of carbohydrate-contain- ing molecules, the lectins or phytohaemagglutinins are also of use since they complex specifically with certain carbohydrate structures.Ourassessmentls2 of a water-insoluble, active form of one phytohaemagglutinin, concanavalin A, has established that a number of straight-chain and branchedchain polysaccharides lS7 J. R. Clamp, Biochem. SOC. Trans., 1974, 2, in the press. lS8 S. A. Barker, B. W. Hatt, J. F. Kennedy, and P. J. Somers, Carbohydrate Res., 1969,9,327. log D. B. Lowrie and J. F. Kennedy, F.E.B.S. Letters, 1972, 23, 69. la*J. F. Kennedy and D. A. Weetman, Analyt. Chim. Acta, 1971,55,448. lal J. F. Kennedy, Methods Carbohydrate Chem., 1971, 6,93. 16a J. F. Kennedy and A. Rosevear, J.C.S. Perkin I, 1973,2041. Tle Meldola Medal Lecture and several monosaccharides may be separated by elution from a column of insoluble lectin, monosaccharides and weakly interacting polysaccharide fractions being eluted with phosphate buffer and more tightly bound fractions being eluted with borate buffer.The use of borate buffer for this purpose over- comes the problems arising in the common use of competitive carbohydrates for the elution of bound material. This work has also demonstrated that mixtures of carbohydrates which are not separable by complex formation with concanavalin A in solution may be separated by use of the immobilizedform; immobilized lectins are now finding use in glycoprotein fractionation. The importances of all types of structural identification have been mentioned, and for the hormonal glycoproteins it is to be hoped that further information of this type along with chemico-biological studies will further aid identification of biosynthetic, metabolic, and message-transport defects and assist in improved treatment of clinical conditions.For the proteoglycans, more attention needs to be given to the absolute primary structures of the glycosaminoglycans and to the structures of the protein backbone, and indeed might lead to an identification of a spectrum of sequences, thus revealing that the proteoglycans are not such a discrete set of compounds as is imagined at present. Furthermore, the question arises as to whether the distribution of carbohydrate units and sulphate residues along the glycosaminoglycan chains and the sequence of glycosaminoglycan chains along the proteoglycan backbone have any coding function, e.g. for deposition of various tissue types which go to make up the various regions and organs of the body.Other useful extensions of work on the proteoglycans would include: exploration of their interactions with other macromolecules and ions to provide connective-tissue models, determination of relationships between primary and tertiary structures and susceptibility to endogeneous enzymes in relation to living processes, and development of automated analyses for chemical diagnosis of clinical conditions. In the field of insolubilization, there is now sufficient known of how to insolubilizea biologically active molecule with retention of activity, and emphasis must now be laid on application.Thus, for example, insolubilized enzymes must now be developed more earnestly for application in clinical analyses and in clinical machines such as blood purifiers etc., and for the simplification of industrial processes. Insolubilized immunologically active agents must also be applied in clinical chemistry for the monitoring of patient conditions, for treat- ment, and immunological reaction of rejection of transplants. Insolubilized antibiotics and drugs must be developed for in situ sustained activity. Insolubiliza- tion of hormonal glycoprotein subunits may aid purification of complementary subunits and provide means of assessing the interactions of the hormones with insoluble target sites.Insolubilization of proteoglycans and glycosaminoglycans could also provide models for studies of tissue processes, and on account of the ionic natures of the glycosaminoglycans such derivatives could well prove useful as new types of matrices for fractionations. In this respect, it is worthwhile remembering that in the body many specific interactions OCCUT between molecules in solution and molecules (receptors) in the solid phase, and thus may ultimately be extrapolated to laboratory separations. Also on the synthetic side is the possibility of testing modified hormones, and the synthesis of pseudo-hormones and pseudo-proteoglycans by covalent attachment to non-biodegradable soluble molecular chains with a view to increasing their half-life in the human system.One specific application of such would be a joint lubricant for treatment where the natural lubricant, hyaluronic acid, has failed, as in arthritis, a common and painful complaint which demands attention since so little is known of the underlying chemistry. Automated systems must also be developed for testing the effects of drugs upon hormonal glycoprotein and proteoglycan biosynthesis and metabolism and for the discovery of better treatments of clinical conditions. Since enormous effort is currently being put into research on glycoproteins and proteoglycans, the number of papers published annually on the chemical, chemico- biological, chemko-medical, and biosynthetic aspects runs into thousands. This of course presents a problem for any worker to keep up with the literature.Computerized output of titles of computer-selected relevant papers is now being offered by some organisations, including the Chemical Society, whilst current awareness with detailed reports on each paper is facilitated by an annual review of such papers.le3 In conclusion, it is hoped that the reader will have been fascinated by the chemical architecture on which the chemical and biological aspects of glyco- proteins, proteoglycans, and carbohydrate-protein complexes of human tissues etc. are based. Indeed, with such complexity and yet harmony within ourselves, it is true that, as was acknowledged long before chemistry as we know it was established, we are ‘fearfully and wonderfully made’.164 It is also anticipated that the reader will recognize the tremendous scope for future work in this important area and will acknowledge that such chemistry will be supreme in deriving further benefits for the maintenance of health and treatment of disease on a physical plane.It is indeed a very great honour for me to receive the Meldola Medal and I am very grateful to the Society of the Maccabteans and the Royal Institute of Chemistry for making this award. I am also grateful to the numerous colleagues, relatives, friends and research students who have encouraged me to this success. 163 J. F. Kennedy, in ‘Carbohydrate Chemistry’, ed. J. S. Brimacombe, (Specialist Periodical Reports) The Chemical Society, London, 1971, 1972, 1973, vols. 4, 5, and 6, Part 11. 184 King David, The Psalms, BC 1040, Ps. 139, v. 14.

ISSN:0306-0012    

DOI:10.1039/CS9730200355

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Chirality in Carbonium Ions, Carbanions, and Radicals By J. W.Henderson SCHOOL OF CHEMICAL SCIENCES, SCIENCE UNIVERSITY OF MALAYSIA, PENANO, MALAYSIA 1 Introduction In the absence of another chiral centre in the system, a carbonium ion, carbanion, or radical is chiral if either of two conditions is fulfilled. First, a system (la) with a non-planar trisubstituted centre bonded to three different groups is non-superimposable on its mirror image (1 b) and thus is chiral. Second, a system (4a) with a planar centre bonded to three different groups, one of which is not conically symmetrical, is non-superimposable on its mirror image (4b) and thus is chiral. The barrier to racemization in system (1) is the barrier to inversion of the central atom X through the a-b-c plane; e.g.(Ib) inverts to structure (3) which is superimposable upon (1a), the enantiomer of (1b). The barrier to race- mization in system (4) is the barrier to rotation about the a-X bond; e.g. (4b) rotates to structure (5), which is superimposable upon (4a), the enantiomer of (4b). (31 d d Chirality in Carbonium Ions, Carbanions, and Radicals Ab initio and semi-empirical calculations' of the geometries of methyl cation, anion, and radical all predict a planar carbonium ion and all but one1* predict a planar radical. All but onele of the studies predict a non-planar carbanion. Although various geometries and inversion barriers have been predicted, the most acceptable result lh suggests an H-C-H angle of 106.8O and an inversion barrier of 5.2 kcal mol-l.Carbonium ions2a and radicalsZb were found in spectroscopic studies to be planar. Studies3 of the reactivities of bridgehead carbonium ions, radicals, and carbanions suggest that carbonium ions have a strong preference for planarity, that radicals have a weak preference for planarity, and that carbanions have a strong preference for non-planarity. Thus chirality due to non-planarity of the trisubstituted centre is probably restricted to carbanion systems. A barrier to racemization of at least 16-20 kcal mol-l is necessary if a system is to be resolved at room temperat~re.~Thus unless the barrier to inversion (lb) -(3) can be raised above the usual 5 kcal mol-l range or the barrier to rotation (4b) -(5) can be raised above the usual 3 kcal mol-l range,s chiral carbonium ions, carbanions, and radicals will exist as racemic mixtures at room temperature.There are several systems in which such increased barriers have been observed. These systems and the mechanisms proposed to account for the increase in the inversion or rotation barrier of each are the topics of this review. In some cases, the ability of an intermediate to retain its configuration has been measured directly, either by observing its racemization rate or by deter- mining the barrier to the racemizing process by dynamic n.m.r. (d.n.m.r.).c However, in most studies, the optical activity of the intermediate has been detected by the product rotation method, i.e. by isolation of an optically active product from an optically active starting material.The observed rotation of the product of such a reaction reflects not only the ability of the intermediate to retain its configuration but also the stereoselectivities of the formation and capture reactions. Unless these reactions are 100% stereoselective, the net racemization observed in the product will be greater than the racemization of the inter- mediate during the reaction. '(a)V. Buss, P.von R. Schleyer, and L. C. Allen, submitted to J. Amer. Chem. Soc.; (6) W. A. Lathan, W. J. Hehre, and J. A. Pople, J. Amer. Chcm. SOC.,1971.93, 808; (c) R. E. Kari and I. G. Csizmadia, J. Chem. Phys., 1969.50, 1443; (d) M. J. S. Dewar and M. Shanshal, J. Amer. Chem. SOC.,1969, 91, 3654; (e) T.P. Lewis, Tetrahedron, 1969, 25, 41 17; cf) M. S.Gordon and H. Fischer, J. Amer. Chem. SOC.,1968, 90, 2471; (R) K. B. Wiberg, ibid., p. 59; (h) P. Millie and G. Berthier, lnternaf. J. Quantum Chem., 1968, 2S, 67. *(a) G. A. Olah, J. R. DeMember, A. Commeyras. and J. L. Bribes, J. Amer. Chem. SOC., 1971, 93, 459; (6) 0.Simarnura in 'Topics in Stereochemistry', ed. N. L. Allinger and E. L.Eliel, Interscience, New York, 1969, Vol. IV. pp. 1-37, and references therein. a R.C. Fort,jun., and P. von R. Schleyer in 'Advances in Alicyclic Chemistry', ed. H. Hart and G.J. Karabatsos, Academic Press, New York, 1966, Vol. I, pp. 283-370; T. Kawamura, T. Koyama, and T. Yonezawa. J. Amer. Chem. SOC.,1973, 95, 3220. E. L. Eliel, 'Stereochemistry of Carbon Compounds', McGraw-Hill Book Co., New York, 1962, p.156. M.Hanack, 'Conformation Theory', Academic Press, New York, 1965, pp. 22-41.'G. Binsch in 'Topics in Stereochemistry', ed. N. L. Allinger and E. L. Eliel, Interscience, New York, 1968, Vol. 111, pp. 97-192. Henderson A recent review' of pyramidal atomic inversion points out the problems introduced into the study of inversion barriers in carbanions by the covalent nature of many carbon-metal bonds and by the existence of many organomet- allic compounds as molecular aggregates in solution. These factors also affect studies of barriers to racemization by bond rotation. Ion pairing introduces similar problems into the study of carbonium ions whereas reactions within the solvent cage introduce similar problems into the study of radicals.All these effects are solvent-dependent as was shown8 for the stereochemistry of base-catalysed hydrogen-deuterium exchange via the carbanion, which varied from 99% net retention to 60% net inversion with changes in solvent composition. This review includes only systems for which there is reasonable evidence that barriers to racemization are independent of interactions with the solvent. 2 Cyclopropyl Anion and Radical In the planar transition state (2) for inversion, and thus for racemization, of (l), the a-X-b angle is larger than in non-planar (l), hence the barrier to inversion can be raised by fixing one angle at much less than 120"by incorpora- ting it into a small ring.Ab initio calculationsla on methyl cation, anion, and radical show that if one H-C-H angle of the carbanion is fixed at 120"the ion is non-planar with an inversion barrier of 4.6 kcal mol-l. If the angle is fixed at SO", however, the inversion barrier is increased to 19.9 kcal mo1-l. Similarly, with one angle fixed at 120", the radical is planar; but when the angle is fixed at go", it is non-planar with an inversion barrier of 1.2 kcal mol-I. For the carbonium ion, however, planarity is favoured in both cases and favoured more heavily with one angle fixed at 90". Ab initio calculations on cyclopropyl anion predict a non-planar structure and a 20.85 kcal mo1-1 inversion barrier.$b OneBa of two semi-empirical calculations on the cyclopropyl system predicts a non-planar carbanion with a 14.2 kcal mo1-1 inversion barrier.The other'd predicts a non-planar radical with a 4.8 kcal mo1-I inversion barrier and a non-planar carbanion with a 36.6 kcal mo1-l inversion barrier.1° In agreement with these calculations, experimental results imply a non-planar cyclopropyl anion with an inversion barrier high enough to prevent racemization at room temperature and a cyclopropyl radical which is either planar or non- planar with a low inversion barrier. The carbonium ion undergoes facile ring- opening to the ally1 cation12 and thus cannot be studied. J. B. Lambert in 'Topics in Stereochemistry', ed. N.L. Allinger and E. L. Eliel, Inter- science, New York, 1971, Vol. VI, pp. 19-105. D. J. Cram, J.L. Mateos, F. Hauck, A. Langemann, K. R. Kopecky, W. D. Nielsen, and J. Allinger, J. Amer. Chem. Soc., 1959, 81, 5774. (a) A. Rauk, J. D. Andose, W. G. Frick, R. Tang, and K. Mislow, J. Amer. Chem. Soc., 1971, 93, 6507; (b) D. T. Clark and D. R. Armstrong, Chem. Comm., 1969, 850. lo Calculations'bs l1 indicate that the barrier to inversion of the cyclopropenyl carbanion should be higher than that of the cyclopropyl carbanion owing to the anti-aromatic nature of the planar transition state. However, there is no experimental evidence that this is the case. llD. T. Clark, Chem. Comm., 1969, 637. 12R. Breslow in 'Molecular Rearrangements', ed. P. de Mayo, Interscience, New York, 1963, Part I, pp. 233-294. Chirality in Carbonium Ions, Carbanions, and Radicals The 1-methyl-2,2-diphenylcyclopropylanion (6) has been produced with a variety of leaving groups in numerous solvent-base systems with reaction times up to 7.5 h.Net retention was 46-100%.13 All comparable reactions in open-chain systems yielded racemized products except two reactions at low tempera- tures in hydrocarbon solvents.l* Since many of the results on (6) were obtained at high temperatures in more dissociating ethereal solvents, the cyclopropyl anion must have a much higher barrier to inversion than the open-chain carban- ion. However, since the solvents were not highly dissociating, the species involved was almost certainly not the free ion, so no estimate of the magnitude of the inversion barrier is possible. Studies in more highly dissociating solvents reveal a type of solvent inter- action other than ion-pair dissociation, which can complicate studies of carban- ion chirality.Replacing the methyl group of carbanion (6) with groups capable of stabilizing the negative charge by resonance lowers the energy of the planar transition state for inversion. Thus the carbanion generated from 1-cyano-2,2- diphenylcyclopropane (8) racemized in less than 15 s; under the same conditions (6) showed high retention of configuration.16* However, (8) underwent base- cataiysed hydrogen-deuterium exchange in methanol, t-butyl alcohol, or DMSO-methanol with 99.9-97.2 % net retention. 15a Unlike the order expected from a dissociation effect, retention was highest in methanol and lowest in DMSO-methanol.The difference between the complete racemization observed in non-dissocia- ting, aprotic solvents and the high net retention observed in dissociating, protic solvents is explained16a in terms of an increased barrier to inversion imposed by hydrogen bonding of the protic solvent with the charge-bearing sp3 orbital, as observed in other systems.? The increase in racemization in t-butyl alcohol is due to the lesser ability of that solvent to form hydrogen bonds.16a DMSO cannot form hydrogen bonds with the carbanion but increased barriers to inversion due to the presence of non-hydrogen-bonded complexes are knom.' Since hydrogen bonding does not seem to affect barriers to rotation,? this effect lSJ. B. Pierce and H.M. Walborsky. J. Org. Chern., 1968, 33, 1962; H. M. Walborsky, J. F. Impastato, and A. E. Young, J. Amer. Chern. Soc., 1964, 86, 3283; F. J. Impastato and H. M. Walborsky, ibid., 1962, 84, 4838. l4 D. Y.Curtin and W. J. Koehl, jun., J. Amer. Chem. SOC.,1962, 84, 1967; R. L. Letsinger,ibid., 1950, 72, 4842. lK (a) H. M. Walborsky and J. M. Motes, J. Amer. Chem. SOC.,1970, 92, 2445; (b) H. M. Walborsky and F. M. Hornyak, ibid., 1955,77, 6026. Henderson should be much less in other chiral carbanion systems which racemize by rotation or simultaneous inversion-rotation mechanisms. In the decomposition of ( -)-(R)-I -methyl-2,2-diphenylcyclopropanoyl peroxide, all products of the I-methyl-2,2-diphenylcyclopropylradical (7) formed outside the solvent cage are racemic.lg Within the solvent cage, l-methyI-2,2- diphenylcyclopropane formed by the disproportionation of the two radicals to alkane and alkene shows 31-37"/, net retention.The rate constant for inversion of the cyclopropyl radical determined by e.s.r. is 108-1010 s-l.'' Since the disproportionation rate constant is probably not that high, much of the optical retention of the product is probably due to constraints placed on the radical by the solvent cage. Both cis-and trans-1-bromo-2-methylcyclopropanereact with n-butyl-lithium to yield 100% retained product after reaction with ethylene oxide and react with metallic lithium to yield partially racemized products with 8-38% retention of configuration.18 The racemization with metallic lithium occurs at the intermediate radical formed in the first of two successive one-electron trans- fers.An interpretationls claiming that the observed retention is due to the intrinsic optical stability of that radical requires a much higher barrier to inversion than would be expected from the data on the reaction of (7). More likely, the retention is due to surface*s and solvent-cage effects which hold the radical in such a position that the second electron transfer takes place on the same side of the cyclopropyl ring as the first. 3 Triaryimethyl Cations, Anions, and Radicals Because overlap with the delocalized electrons of the aryl rings is greatest with p-orbitals, a planar structure for triarylmethyl carbanions and radicals will be greatly stabilized relative to the non-planar structures whose sp3 orbitals overlap less well.Thus the ionic or radical centre of triarylmethyl anions and radicals is either planar or inverting so rapidly between its two non-planar conformers that it is effectively stereochemically planar. The structure of tri- phenylmethyl perchlorate has been determined by X-ray crystallography and the carbonium ion was planar.20c The structure of triphenylmethyl radical has been determined by electron diffraction in the gas phase20b and that of tri-p-nitrophenylmethyl radical has been determined by X-ray crystallography.*OU In both cases, the radical centre was nearly planar with a bond angle of 118.0". There have been no such studies of a triarylmethyl earbanion.Since the three aromatic rings of triarylmethyl systems neither occupy the same plane as the three bonds to the central carbon (hereafter called the l6 H. M. Walborsky and J.-C. Chen, J. Amer. Chem. SOC.,1971,93,671. l7 R. W. Fessenden and R. H. Schuler, J. Chem. Phys., 1963. 39, 2147. M. J. S. Dewar and J. M. Harris, J. Amer. Chem. SOC.,1969, 91, 3652. lSH. M. Walborsky and M. S. Aronoff. J. Organometallic Chem., 1973, 51, 55; H. M. Walborsky and A. E. Young, J. Amer. Chem. SOC.,1964, 86, 3288.*IJ (a) P. Andersen and B. Klewe, Actu Chem. Scand., 1967, 21, 2599; (b) P. Andersen, ibid., 1965, 19,629; (c) A. H. Gomes de Mesquita, C. H. MacGillavry and K. Eriks, Acru Crysr., 1965, 18, 437. Chirality in Carbonium Ions, Carbanions, and Radicals reference plane) nor are perpendicular to that plane, but instead have propeller- like structures,z0 symmetrically substituted triarylmethyl anions, cations, and radicals can exist in the enantiomeric right- and left-handed propeller forms (9a) and (9b) which are interconvertible by rotation of the aryl groups through a transition state in which they are parallel or perpendicular to the reference plane.21 Thus the barrier to these three simultaneous rotations is the barrier to racemization.For the substituted carbonium ion the barrier to inter- conversion of the enantiomers determined by d.n.m.r. is 12.5 kcal mol-'. For the substituted radical (11),24 the barrier to interconversion of the enantiomers determined by electron nuclear double resonancezs is 5.5 kcal mol-l.These barriers are too low to prevent racemization at room temperature. Substituents in the ortho-position are unlikely to have a dramatic effect on the barrier since the steric interactions that the substituents would impose on a transition state in which the ortho-substituted ring rotates through the reference plane can be avoided by a transition state in which the ring becomes perpendicu- lar to the reference plane.22 However, triaryl systems in which the three aryl groups are different and at least one aryl group has an ortho-or rnetu-substituent *l The possible mechanisms for interconversion of isomers in this and other triaryl systems have recently been analysed. For a discussion of which mechanism actually operates, see ref.23.** D. Gust and K. Mislow, J. Amer. Chern. Sac., 1973, 95, 1535. m3 J. W. Rakshys, jun., S. V. McKinley, and H. H. Freedman, J. Amer. Chem. SOC.,1971, 93, 6522. IrlJ. S. Hyde, R. Breslow, and C. DcBoer, J. Amer. Chem. Sac., 1966, 88,4763. t6 J. S. Hyde in 'Magnetic Resonance in Biological Systems', ed. A. Ehrenberg, B. G. Malstrom, and T. Vannard, Pergamon Press, Oxford, 1967, pp. 63-84; J. S. Hyde, J. Chem. Phys., 1965, 43,_1806. Henderson possess an element of chirality separate from the propeller chirality described above.aa System (12a) is non-superimposable on its mirror image (12b). If (12b) under- goes rotation of the three aryl groups through a transition state in which the ring bearing the meta-(or ortho-) substituent is parallel to the reference plane, structure (13) results which is superimposable upon (12a), the enantiomer of (12b).However, if (12b) undergoes rotation of the three aryl groups through a C C x X Y X1’___)f---C --I$X J. [, Chirality in Carbonium Ions, Carbanions, and Radicals transition state in which the ring bearing the meta-substituent is perpendicular to the reference plane, structure (14) results, which is non-superimposable upon (12a) even though they have the same configuration in terms of propeller chirality; that is, (12a) and (14) are diastereomers. Thus (12) must racemize through a transition state in which a ring with an ortho-or rneta-substituent becomes parallel to the reference plane, a transition state which should be more sensitive to steric hindrance to rotation than one in which the ring is perpendicular to the reference plane.It is clear that this chirality is independent of propeller chirality if we consider the case in which the three aryl rings of (12) are perpen- dicular to the reference plane. This destroys propeller chirality, yet (12a) is not superimposable upon (12b). This element of chirality also operates in triaryl systems with four bonds to the central carbon,za and racemization barriers of greater than 26 kcal mol-l have been observedz6 in such systems by d.n.m.r. However, the only triaryl systems with a trisubstituted centre for which this chirality has been found are pheny1bipheny I-a-naph thylmethyl cat ion (15) and phenyl biphenyl-1 -(&methyl- naphthy1)methyl cation (16)2* and anion (17).z9 (15) R = H,* = + (16) R=Me,# = + (17) R =Me,* = -Observation of a mass-law effect shows that the hydrolysis of phenylbiphenyl- a-naphthylmethyl benzoate takes place through free dissociated ion (1 5).Yet the solvolysis of optically active benzoate yields alcohol with 33--52% net retention.27 The barrier to racemization of ion (15) is the barrier to rotation of the naphthyl group. However, the optical activity of the product depends not only 011 the rate of racemization of ion (15) but also on the ability of the starting material to form l6 S. V. McKinley, P. A. Grieco, A. E. Young, and H.H. Freedman, J. Amer. Chem. Soc., 1970, 92, 5900, and references therein. B. L. Murr and C. Santiago, J. Amer. Chem. SOC.,1968.90, 2964; 1966, 88, 1826. p* (a) B. L. Murr and L. W. Feller, J. Amer. Chrm. Soc., 1968, 90, 2966; (b) L. W. Feller, Ph.D. Thesis, The Johns Hopkin University, 1968. In J. W. Henderson, Ph.D. Thesis, The Johns Hopkin University, 1971. Henderson one enantiomer stereospecifically and on the ability of the naphthyl group to direct the capturing reagent to only one face. Assuming that water attacks both faces of (15), the barrier to racemization is >9 kcal ~oI-?~' The substitution of a methyl group at position 8 of the naphthyl ring increases the barrier to rotation of the naphthyl group. Carbonium ion (16) is so optically stable that its rotation can be observed directly.2s The carbonium ion produced by ionization of d-phenylbiphenyl-l-(8-methyInaphthyl)methanolin sulphuric acid showed [a]3851940", [aL3s3160" (1.24 x moll-l).The ion showed no racemization after 2 h at -20 "C. At room temperature in sulphuric acid, ion (16) underwent ring sulphonation to yield an ion with the opposite sign of rotation. However, once the reaction was complete, the rotation of the solution remained constant apparently indefi- nitely at room temperature.28b Thus the barrier to racemization of (16) is in the same range as that of compounds which are chiral by virtue of an asymmetric carbon. A preliminary study20 of phenylbiphenyl-l-(8-methylnaphthyl)methylanion (17) indicates the presence of a similar but smaller barrier to racemization.Results on the st ereochemis try of p hen yl bip heny 1-I-( 8-met h ylnap ht hy1)me thy1radical are not yet available. 4 Ferrocenylmetbyl Cation In the formation of a carbonium ion a to a ferrocenyl group, the leaving group departs anti to the iron Thus (-)-(R)-l-ferrocenylethanol (1 8) ionizes stereospecificallyin acidic solutions to form only enantiomer (19) of the chiral Fe ____c Fe b'.. (18) (19) I-ferrocenylethyl cation. The barrier to racemization of (19) is the barrier to rotation about the carbon-ferrocenyl bond which is increased by resonance interaction with the ferrocenyl group. Ion (19) and its enantiomer show [a],":,= f 395" (C = 0.1 in CF3C02H)with a 24.5 kcal mol-1 barrier to racernizati~n.~~ so G.W. Gokel, D. Marquarding, and I. K. Ugi, J. Org. Chem., 1972, 37, 3052; E. A. Hill and J. H. Richards, J. Amer. Chem. SOC.,1961, 83,4216. *lT. D. Turbitt and W. E. Watts, J.C.S. Chem. Comm., 1973, 182. 405 Chirality in Carbonium Ions, Carbanions, and Radicals 5 Carbanions a to Sulphur Carbanions formed a to sulphonyl (20),8a sulphinyl (21),88 sulphenyl (22),34 and sulphonium (23)35 groups are capable of yielding optically active products under conditions which produce racemic products in other systems. In all these systems, the barrier to racemization is the barrier to C-S bond rotation if the carbanion centre is planar or the barrier to carbanion inversion with C-S bond rotation if the carbanion is non-planar. L.0. R' (22) If the carbanion centre is planar, (24) is the only chiral staggered conformer. Racemization takes place via rotation about the C-S bond, (24a) -+ (24b). If 32 (a) E. J. Corey and T. H. Lowry, Tetrahedron Letters, 1965, 803; (b) E. J. Corey and T. H. Lowry, ibid., p. 793; (c) E. J. Corey and E. T. Kaiser, J. Amer. Chem. SOC.,1961, 83, 490; (d)J. N. Roitman and D. J. Cram, ibid., 1971,93,2225; (e) D. J. Cram, R. D. Trepka and P. St. Janiak, ibid., 1966, 88, 2749; cf) D. J. Cram, D. A. Scott, and W. D. Nielsen, ibid., 1961, 83, 3696; (g) M. D. Brown, M. J. Cook, B. J. Hutchinson, and A. R. Katritzky, Tetrahedron, 1971, 27, 593; (h) R. R. Fraser and F. J. Schuber, Chem. Comm., 1969, 1474.33 (a) R. Viau and T. Durst, J. Amer. Chem. SOC.,1973,95, 1346; (b) T. Durst, R.Viau, and M. R. McClory, ibid., 1971, 93, 3077; (c) T. Durst, R. R. Fraser, M. R. McClory, R. B. Swingle, R. Viau, and Y. Y. Wigfield, Canad. J. Chem., 1970, 48, 2148; (d) R. R. Fraser, F. J. Schuber, and Y. Y. Wigfield, J. Amer. Chem. SOC.,1972, 94, 8795; (e) K. Nishihata and M. Nishio, Tetrahedron Letters, 1972, 4839; cf) K. Nishihata and M. Nishio. J. C. S. Perkin 11, 1972, 1730; (g) R. Lett and A. Marquet, Tetrahedron Letters, 1971, 3255; (h) B. J. Hutchinson, K. K. Anderson, and A. R. Katritzky, J. Amer. Chem. SOC., 1969, 91, 3839; (i)J. E. Baldwin, R. E. Hackler, and R. M. Scott, Chem. Comm., 1969, 1415; (j)E. Bullock, J. M. W. Scott, and P. P. Golding, ibid., 1967, 168; (k) S.Wolfe and A. Rauk, ibid., 1966, 778; (I) D. J. Cram and S. H. Pine, J. Amer. Chern. SOC.,1963, 85, 1096. 84 (a) E. L. Eliel, A. Abatjoglou, and A. A. Hartman, J. Amer. Chem. Soc., 1972, 94, 4786; (6)A. A. Hartman and E. L. Eliel, ibid., 1971, 93,2572; (c)E. L. Eliel, personal communica- tion. See also R. T. Wragg, Tetrahedron Letters, 1969, 4959. 85(a) G. Barbarella, A. Gabesi, and A. Fava, Hefv. Chim. Acta, 1971, 54, 2297; (b) G. Barbarella, A. Gabesi, and A. Fava, ibid., p. 341. Henderson the carbanion centre is non-planar, all conformers are chiral. Racemization involves both carbanion inversion and C-S bond rotation, either simultaneously or stepwise. Stepwise racemization is shown for non-planar conformer (25).;c (25~) (256) If the first step in racemization is inversion, this structure is formed; if the first step is rotation, the structure with R1 and R2reversed is formed. Y = 0 or an electron pair Findings on the stereochemical properties of other systems which, like (20)-(23) have electron pairs or pofar bonds on adjacent atoms have been generalized into two rules. 36 First, electron pair-electron pair, electron pair-polar bond, and polar bond-polar bond interactions significantly increase the rotation-inversion barrier in such systems. Second, the most stable conformer in such systems is the one which contains the maximum number of gauche interactions between the electron pairs or polar bonds. The first rule has been experimentally verified for numerous including sulphonamides (26), sulphinamides (27), and sulphenamides (28), the isoelec- (26) (27) (281 S. Wolfe, A.Rauk,L. M. Tel, and I. G. Csizmadia,J. Chem. Suc. (E), 1971, 136, and refer-ences therein. Chirarity in Carboniurn Ions, Carbaniom, and Radicals tronic nitrogen analogues of (20)-(22), respectively. The action of the second rule, known as the gauche effectY3'can be seen in the preference3* of hydrazine for conformer (291, the only conformer which contains an electron pair-electron pair gauche interaction. The barrier to rotation or rotation-inversion in (23)-(23) is undoubtedly the same as that found for (26)-(28) and other systems with electron pairs or polar bonds on adjacent atoms. A theoretical explanation of the nature of this barrier and of the interactions which produce the gauche effect has recently been provided.37 Ab initio calculations on a-sulphonyl,3s a-s~lphinyl,~~and a-~ulphenyl~~ carbanions predict that the most stable conformer in each case is that predicted by the gauche effect. The experimental verification of this predicted confor- mational preference is discussed below for each system, 0 0 OH-~ 0 '0 (30) (31) r"--0,SCHMePh (32)+ CH*=CHCHO 87 S. Wolfe, Accounts Chem. Res., 1972, 5, 102. ssT. Kasuya and T. Kojima, Proc. Internat. Symp. MoI. Struct. Spectroscopy, Tokyo, 1962, C404; A. Yamaguchi, I. Ichishima, T.Shimanouchi, and S.4. Mizushima, J. Chem. Phys., 1959, 31, 843. 38 S. Wolfe, A.Rauk, and I. G. Csimadia, J. Amer. Chem. SOC., 1969, 91, 1567. O0 A. Rauk, S. Wolfe, and I. G. Csizmadia, Canad. J. Chem., 1969, 47, 113. 41 S. Wolfe, A. Rauk, L. M. Tel, and 1. G. Csizmadia, Chem. Comm., 1970, 96. Henderson A. a-Methylsulphonyl Anions.-Base-catalysed cleavage of 3-hydroxy-2-methyl- 2-phenylthiolan 1,l-dioxide (30) via the sulphonyl carbanion (31) yielded 1-phenylethane sulphinate (32) of 90-100 % inverted configuration. 32b Base-catalysed cleavage in open-chain systems proceeded with 98-1 00% retention of configurati~n.~~a This complete change in stereochemistry was taken32b as proof that the observed inversion in the reaction of (30) is not due to blocking of one face of the ion by the leaving group but reflects the preference of a-sulphonyl carbanions for a conformation in which the electron pair of the carbanion is directed along the bisector of the 0-S-0 angle, as predicted by the gauche effect.Planar conformer (33)3aCsf and non-planar conformer (34),39 both of which have their electron pair directed along the bisector of the 0-S-0 angle, have been proposed as the preferred conformer of the a-sulphonyl carbanion. Ab initio calculationsaDon the hypothetical hydrogen methyl sulphonyl anion (36) (361: (331, (341,or(351, R1= R2= $= H predict that (34) with an H-C-H angle of 115” is 2.4 kcal mol-l more stable than (33) but there is no experimental verification for this preference. Rather there is that an increase in the planarity of an a-sulphonyl carbanion does not decrease its barrier to racemization.The rate of exchange of 1-phenylethyl phenyl sulphone (37) exceeds that of 2-octyl phenyl sulphone (38) by a factor of about lo*,implying that the carbanion centre of (37) is more PhSOZCHMtPh PhS02CHMeBu” (371 (38) nearly spahybridized to permit the favourable n overlap of the phenyl groups with thep-orbital. However, in spite of this increase in planarity, the barrier to racemization is slightly higher in the carbanion formed from (37) than in the carbanion formed from (38). The difference between (33) and (34) is subtle and refers only to the conforma- tion of the ion and not to the barrier to its racemization. The potential energy surface calculatedaD for (36) indicates that it ‘racemizes’ by simultaneous rotation and inversion via the same transition state (35) through which (33) racemizes.Chirality in Carbonium Ions, Carbanions, and Radicals Probably the hybridization at the carbanion centre in a-sulphonyl carbanion varies netween sp3 and sp2 depending on the nature of the substituents R1and R2. B. a-Methyl Sulphinyl Anions.-Since an a-sulphinyl carbanion contains a chiral sulphoxide group, a-sulphinyl carbanions with opposite configurations at carbon are diastereomeric rather than enantiomeric if the configuration at sulphur remains constant. This intrinsic diastereomerism makes them and the a-methyl sulphonium anions (see below) intrinsically different from the other systems covered by this article. They are included because of the close relationship of the barrier to interconversion of their diastereomeric carbanions with the barrier to interconversion of the enantiomeric carbanions of the sulphonyl and sulphenyl systems.All di~cu~~ions~~~~~of the stability order of a-sulphinyl carbanions have been in terms of the three diastereomeric non-planar conformers (39)-(41). A strong has been made in favour of a non-planar carbanion centre rather than a planar one but there are no experimental data that cannot be explained in terms of either a planar or non-planar carbanion centre. R' R2 (421: (391, (401,or (411, R' =I?*=R3=H Ab initio calculation^^^ on the hypothetical hydrogen methyl sulphinyl anion (42) are in agreement with the prediction of the gauche effect.Conformer (39), which experiences the maximum number of gauche interactions, is predicted to be 1.6 kcal mol-1 more stable than (40)and 12 kcal mol-1 more stable than (41), which represents an energy maximum. Experimental determinations of this stability order have centred on the study of compounds of type (43) by two approaches: measurements of the relative rates of base-catalysed hydrogen- deuterium exchange of HA and HB by n.m.r. arid quenching with electrophilic reagents of the carbanions formed by the reaction of (43) with alkyl-lithium. Henderson To interpret such results, several factors must be kno~n:~~(i) the absolute stereochemical relationship of the chiral centre at sulphur and the chiral centre created at carbon; (ii) the chemical shift assignments of HA and HB (in the n.m.r.experiments only); and (iii) the conformation in which the system reacts. Although attempts have been made to determine the conformation through which (43) reacts in conformationally free systems, the most reliable results are expected to be those obtained on conformationally rigid systems in which the conformation in which (43) reacts is unambiguously known.42 The experimentally determined relative stabilities of conformers (39)-(41) are shown in the Table along with the theoretical stability order. Table Relative stabilities of a-sulphinyl carbanions Method Solvent Order Ref. Theory gas phase (39) > (40) > (41) 40 H-D exchange ButOD (41) > (40)" 33d H-D exchange ButOD (39) N (40) N (41)' 33h H-D exchange CD,OD (40) > (41)a 33d H-D exchange CD30D (40) > (39) > (41)" 33h H-D exchange H-D exchange D20 D20 (40) > (41),a (40) > (39)a (40) > (39) > (41)a 33g3312 H-Dexchange DMSO-CD,OD (39) N (40) N (41)" 33h quenching THF (40) > (39) > (41) 33e quenching THF (39) > (40) > (41) 33a UDetermined for conformationally rigid systems.Some correlations among the results are discernible but they are not consistent and agreement with the theoretical order is rare, apparently because solvent effects, which are neglected in the theoretical calculations, are the predominant factor in the determination of the preferred conformation for a-sulphinyl carbanions.In one the ratio of the diastereomeric quenching products was reversed from 14.2 :1 to 0.22 :1 by changing the solvent. In another ~ysfem,~~dthe relative stability of the carbanions was changed from (41) > (40) to (40) > (41) by a change of solvent. Although the theoretically predicted stability order has not been confirmed experimentally, the predicted high angular dependence for a-sulphinyl carbanion stability has been ~erified.~~d A difference of 40" in the orientation of two otherwise identical protons in a rigid system caused a thousand-fold difference in their exchange rates even though both of the resulting carbanions would formally be in conformation (39). C. a-Methyl Sulphenyl Anions.-The gauche effect and ab initio calculationsq1 on the hypothetical hydrogen methyl sulphenyl anion (46) predict that conformer (44) is preferred over conformer (45).This prediction has been verified experi- Results in open-chain systems have, however, important applications in the synthesis of new chiral centres via optically active sulphoxides. See ref. 33bf. Chirality in Carbonium Ions, Carbanions, and Radicals 0-• mentally by studiess4 on the ionization of the C-Zsubstituted cis-4,6-dimethyl-1,3- dithianes (47)-(52) with n-butyl-lithium followed by quenching with electro- philic reagents. In this system, the carbanion with the charge-bearing orbital equatorial is in conformer (44)(R2and R3being members of the six-membered ring) with respect to both sulphur atoms.The carbanion with the charge-bearing orbital axial is in conformer (45) with respect to both sulphur atoms. Ionization of (47) followed by quenching with deuteriated hydrochloric acid yielded >99 % equatorially deuteriated (48).s*0 Similarly, ionization of (49) yielded only (50) after quenching with hydrochloric a~id.~~b A lower limit can be set for the magnitude of the equatorial preference of the carbanion since ionization of the dithiane (51) with a t-butyl group in the equatorial position yielded, after quenching, over 99% (52) with the t-butyl group in the axial po~ition,~*cSince the equatorial preference of the t-butyl group at C-2in a 1,3-dithiane is > 2.7 kcal rn~l-l,~~ the equatorial preference of the charge-bearing orbital must be at least 5 kcal mol-1 to account for the observed product distrib~tion.~*C D.a-Methyl Sulphonium Anions.-The diastereotopic a-protons HAand HB in 1-methylthiolanium iodide (53) undergo base-catalysed hydrogen-deuterium E.L.Eliel and R.0. Hutchins, J. Amer. Chem. Soc., 1969,91,2703. Henderson exchange in D20at relative rates of 30 :1.86~~s4~However, the conformational mobility of the system between the envelope and half-chair conformers does not alpow unambiguous determination of the preferred conformation of the inter- mediate u-sulphonium carbanion. No rate difference was observed in the analogous six- and seven-membered ring systems. 854 413

ISSN:0306-0012    

DOI:10.1039/CS9730200397

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Chemistry of Azidoquinones and Related Compounds By H. W. Moore DEPARTMENT OF CHEMISTRY, UNIVERSITY OF CALIFORNIA, IRVINE, CALIFORNIA, 92664, U.S.A. 1 Introduction Azidoquinones constitute a remarkably versatile class of synthetically useful reagents. They are easily prepared and can function as penultimate precursors to a large variety of other compounds, among which are azidohydroquinones,ly2 aminoq~inones,1-~y-cyanoalkylidene-A5~~-butenolides,42-cyanocyclopent-4-ene-l,3-diones,6 azepine-2,5-diones,6 diacyl cyanides,’ 3-cyano-2-aza-l,4-quinones, 4-acetoxy- 1,2-quinone-2-(N-a~etyl)imines,~trans,trans-1,4-diacetoxy-cis,cis-1,4-dicyanobuta- 1,3-diene~,~ 2-alkenyl-2,3-dihydroindole-4,7-dione~,’~ benzo[ flindole-4,9-diones,” and cyanoketens. l2 Many of these compounds are themselves members of new or relatively unexplored classes of compound and should find synthetic utility in their own right. It is the purpose of this review to discuss the synthesis and chemistry of azidoquinones as well as certain of those compounds to which they are structurally and chemically related.2 Synthesis of Azidoquinones The literature contains over seventy examples of variously substituted mono-, di-, and poly-azido-l,4-benzo-, -1,4-naphtho-, and -1,2-naphtho-quinones. Almost without exception these compounds are prepared by the reaction of halogeno- or acetoxy-substituted quinones (1) with inorganic azide in aqueous alcohol, being generally obtained as highly coloured crystalline solids which are safely manipu- lated under normal laboratory conditions.One exception is tetra-azido-l,4- benz~quinone,~*~~J~a beautiful deep-purple solid which is extremely shock and thermally sensitive and should be handled with great caution. The fact that 1 L. F. Fieser and J. L. Hartwell, J. Amer. Chem. SOC.,1935, 57, 1482. H. W. Moore and H. R. Shelden, J. Org. Chem., 1968,33,4019. E. Winkelmann, Tetrahedron, 1969, 25, 2427. H. W. Moore, H. R. Shelden, D. W. Deters, and R. J. Wikholm, J. Amer. Chem. Soc., 1970, 92, 1675. W. Weyler, jun., D. S. Pearce, and H. W. Moore, J. Amer. Chem. SOC.,1973, 95, 2603. H. W. Moore, H. R. Shelden, and W. Weyler, jun., Tetrahedron Letters, 1969, 1243. 7 J. A. Van Allen, W. J. Priest, A. S. Marshall, and G. A. Reynolds, J.Org. Chem., 1968,33, 1100.* D. S. Pearce and H. W. Moore, unpublished results. D. S. Pearce, M. S. Lee, and H. W. Moore, J. Org. Chem., in the press. 10 P. Germeraad and H. W. Moore, J. Org. Chem., in the press. l1 P. Germeraad and H. W. Moore, J. Org. Chem., in the press. l* H. W. Moore and W. Weyler, jun., J. Amer. Chem. SOC..1970, 92, 4132. la A. Korezynsky and St. Namylslowski, Bldl. SOC.chim. France, 1924, 35, 1186. lP K. Fries and P. Ochwat, Ber., 1923, 56, 1291. 415 Chemistry of Azidoquinones and Related Compound3 quinones bearing halogenoJ6 as well as acetoxy-le groups are readily available translates to a versatile synthesis of azidoquinones (2). Yields are usually high, particularly when the leaving group is a halide ion.(1 1 X = C1, Br, or OCOMe 3 Reactions of Azidoquinones A. Reduction.-Azidoquinones are reduced under a variety of conditions to the corresponding primary aminoquinones (Na2S201, H2/Pd-C, HJPt,O, H2/Pt-C).1s2J7-1B The scope of this reaction has not been extensively explored. However, it does appear to provide potentially one of the best methods of intro- ducing an amino-substituent on to the quinoid nucleus. The yields are high and the conditions mild. The mechanism of the sodium dithionite reduction is quite interesting in that the quinone nucleus is initially reduced to the hydroquinone which then appar- ently disproportionates to the corresponding aminoquinone and nitrogen. This latter step was proposed by Fieser and Hartwell' to explain the smooth conver- sion of 1,Qnaphthoquinone (3) into 2-amino-1,4-naphthoquinone(4)upon its treatment with hydrazoic acid in glacial acetic acid.Subsequent investigations showed that azidohydroquinones do indeed undergo a thermally induced oxidation-reduction to the corresponding aminoquinones (5).2*20 Whether this is an intra- or inter-molecular process is not clear and awaits further work. An implication that the latter is IikeIy comes from the fact that the unsymmet- rical diazide, 2,5-diazido-3-methyl-5-isopropyl-l,4-benzoquinol(6) gives 2-amino-5-azido-3-methyl-6-isopropyl-1,6benzoquinone (7) and 2-amino-5-azido-6-methyl-3-isopropyl-1,4-benzoquinone(8) in the ratio of 2: 1. Closer to a 1:1 mixture might be anticipated for an intramolecular reaction. The thermal decomposition of symmetrical diazidohydroquinones, (9) and (1 l), gives a 15 H.W. Moore. D. L. Maurer. D. S. Pearce. and M. S. Lee, J. Org. Chem., 1972, 37, 1984. 16 J. F. W. McOmie and J. M. Blatchly. Org. Reactions. 1972. 17, 199. 17 R. J. Wikholm and H. W. Moore, Ch~m.Comm., 1971, 1070. I* Z. Cheng. K. Yuen. and C. C. Cheng, J. Medicin. Chem., 1970, 13, 264. 19 Y. Waianabe, K. Nakajima, T. Seki, and H. Ozawa. Chem. and Pharm. Bull. (Japan), 1970, 18, 2208. H. W. Moore, H. R. Shelden, and D. F. Shellhamer, J. Org. Chem., 1969, 34, 1999. 416 Moore specific reduction of only one azide group and provides an efficient route to 2-amino-5-azido-l,4-benzoquinones(10) and (12), respectively.OH OH OH OH 0 OH 0 (9) (10) OH 0 417 3 Chemistry of Azidoquinones and Related Compounds An additional use of this reaction has recently a~peared.'~ The naturally occurring aminoquinone rhodoquinone-9 (13)a1 was obtained from the azido- quinone (14) via sodium dithionite reduction to the azidohydroquinone (15) and subsequent thermal decomposition. Na2S204 Me ___) Me0 Me0 (CH*CH=C -CH2)gH 0 OH (14) (15) 0 M.e Me0 (CH.2 CH =C-CH2 )gH 0 One can envisage other non-reductive routes to azidohydroquinones. For example, the 1,4-addition of hydrazoic acid to 1,4-benzo- and 1,4-naphtho- quinones has been considered. As mentioned above, such an addition was proposed by Fieser and Hartwelll to explain the formation of 2-amino-1,4- naphthoquinone (4) from 1 ,4-naphthoquinone and sodium azide in glacial acetic acid. Under the same conditions, benzoquinone reacts to give a 35% yield of 2,5-dia~idohydroquinone~~3 22 (9).However, 2-me thyl- 1,4-nap hthoquinone and 4-methy1-ly2-naphthoquinonefailed to react.' Under strongly acidic conditions, cold concentrated sulphuric acid, the reaction takes an entirely different course; variously alkyl-substituted quinones (1 6a-e) react with hydrazoic acid to give azepinediones (17a-e) rather than azidohydroquinone~.~~-~~Under the same conditions 2-hydroxy-l,4-naphthoquinone(18) undergoes a deep-seated re- arrangement to 3-oxo-A1a-isoindolineaceticacid (1 9). 28 Thymoquinone (20) reacts with sodium aide in trichloroacetic acid at 65 "Cto give the butenolide (21).2,a7328It is shown below that this last reaction does involve the initial 1,4-addition of hydrazoic acid to the quinone giving an azidohydroquinone intermediate.Quinone mono- and di-imines appear to add readily hydrazoic R. Powls and F. W. Hemming, Phytochemistry, 1966, 5, 1235. a* E. Oliveri-Mandala and E. Calderao, Gazzetta, 1915, 45, 307. s9 D. Misiti, H. W. Moore, and K. Folkers, Tetrahedron 1966,22, 1201. '4 R. W. Richards and R.M. Smith, Tetrahedron Letters, 1966, 2361. 16 G. R.Bedford, G. Jones, and B. R. Webster, Tetrahedron Letters, 1966, 2367. I* H. W. Moore and H. R. Shelden, J. Org. Chem., 1967,32,3603.A. H. Rees, Chem. and Id., 1964,931. A. H.Rees, Chem.and Id., 1965,1298. Moore 0 0 R3f4R2 NaN3 4 RQ;RL R' "2SOL RL 0-5OC 0 "0 (16) (17) R' R' R' Me Me Me =Me, H Me Me -cH=cH-cH==cH-Me H Me Me Me Me acid (NaNS-MeCO,H), giving the corresponding aryl azides in good yield (68-99 %).2B For example, 1,4-benzoquinonedibennesulphonimide(22) and 1,4-benzoquinonebenulphonimide (24) react with sodium azide in aqueous acetic acid to give, respectively, (23) and (25) in greater than 90% yield. 0 @OH @NH 0-5 OC 0 CHC02H (18) (19) 0 0 NaN3 *CI3CCO2H CHMe2 65 OC aD R. Adams and W. Reifschneider, Bull. Soc. chim. France, 1958, 23. Chemistry of Azidoquinones and Related Compounds PhOzS, N PhO2SNH -b”’MeC02HNoN3 H208 NHS02PhN\ S02Ph (22) (23) OH _____)MeC02NaN3ti 6”H20 NS02Ph NHS02Ph (24) (25) B.Synthesis and Therrnolysisof 1,4Diacetoxyazidobnes.-The facile sodium dithionite reduction of azidoquinones to the corresponding hydroquinones not only provides a route to aminoquinones, but also potentially allows the synthesis of a large variety of other aryl azides by functionalization of the phenolic hydroxy-groups. For example, the diacetates (26) of a variety of mono- and di-azidoquinones have been prepared by the reaction of the corresponding azidohydroquinones (27) with acetic anhydridepyridine.e~20 OH OCOMe R2R3QN3‘ R’ (MeCOIZO %“5N - OH OCOMe (27) (26) R1 Me Me H CMe, Ph H CMe3 N3 N3 N3 N3 420 Moore The thermal chemistry of certain of these 1,4-diacetoxyazidobenzeneshas been studied.s The monoazides, (26a, b, and d) smoothly rearrange with nitrogen loss in refluxing chlorobenzene to the respective N-acyl-1 ,2-quinoneimines (28)-(30) whereas (26e)gives the carbazole (31) when decomposed under the same conditions.The formation of the N-acyl-l,2-quinoneiminesis unusual since it involves an acyl migration to an azide nitrogen, a rarely observed process.3o Other known OCOMeWN3Me A (89%) 0WNCoMeMe OCOMe OCOMe (26a) (28) OCOMe 0 A NCOMe Me (15%) > Me OCOMe OCOMe (26b) (29) OCOMe A - NCOMe Me3cq:., (69’/01 Me3c+ OCOMeCMe3 OCOMQ (26d) (30) OCOMe OCOMe A ph+N3 Ph (61%) > OCOMe OCOMe (26e) (31) SO W.Lwowski,‘Nitrenes’, Interscience, New York, 1970, p. 72. Chemistry of Aziabquinones and Related Compowtdr examples in which such a migration takes place are found in the photolytic decomposition of dimethyl diazid~malonate~l and the pyrolytic decomposition of 2,2-diazido- and 2-azid0-2-aryl-indane-1,3-dione.~~ Unlike the monoazide series, the 1,4-diacetoxy-2,3-diazidobenzenes(26i-k) smoothly undergo thermally induced ring-cleavage in reflwing o-dichlorobenzene to give the trans,trans-l,4-diacetoxy-cis,cis-l,4-dicyanobuta-l,3-dienes(32)-(34),respectively.* These highly functionalized dienes, which can be regarded as the acylated cyanohydrins of bis-ketens, are masked 1 ,4dicarbonyl moieties and may find according synthetic utility.The formation of 1,3-dienes from o-diazidobenzenes appears to be a general reaction, the first examples having been reported by Hall and Patterson.** Related transformations have been observed for the lead tetra-acetate oxidation of o-phenylenediamines." OCOMe OCOMe N3mN3( 77%1 A * ($:: OCOMe OCOMe (26i) (32) OCOMe OCOMe -A -(72X) OCOMe OCOMe (26j1 (33) OCOMe A * ( 94%)Me OCOMe (26k) R. M.Moriarty and P. Serridge, J. Amer. Chcm. SOC.,1971,93, 1534. H. W. Moore and D. S. Pearcc, Tetrahedron Letters, 1971, 1621. *a J. H. Hall and E.Patterson, J. Amer. Chem. SOC.,1967, 89, 5856. K.Nakagawa and H.Onove, Chcm. Comrn., 1965,396. Moore A particularly interesting example of this pyrolytic cleavage was observed when 1,4-diacetoxy-2,3-diaidonaphthalene(26h) was thermally decomposed.The presumed intermediate quinodimethane collapsed to the truns-and cis-benzo- cyclobutenes (35) and (36). In addition, the unexpected isoquinoline (37) was isolated as the major product. The formation of 1,4-diacetoxy-3-cyanoiso-quinoline (37)from (26h) is most intriguing and must result from a very deep- seated rearrangement. An attractive possibility for such a mechanism is based upon the fascinating gas-phase equilibrium of phenylnitrenes and cx-pyridyl- ~arbenes.~~In the case at hand, the nitrene (38) could rearrange to the azido- carbene (39), which upon nitrogen loss would give (37). OCOMe OCOMe I I 6COMe CN OCOMe OCOMe OCOMe OCOMe (36) <gO/o (37) 39% OCOMe OCOMe OCOMe OCOMe (38) (39) W.D. Crow and C.Wentrup, Tetrahedron Letters, 1968, 6149. Chemistry of Azidoquinones and Related Compounds C. Reactionsof Azidoquinones with Nucleophi1es.-The reactions of azidoquinones with nucleophilic species have not received detailed study. This should be a worthwhile area for investigation since both quinone nuclei and azide groups are susceptible to nucleophilic attack. The former would give azidohydroquinones and their transformation products, e.g. substituted aminoquinones, and the latter could result in diazo-transfer reaction^.^^ Examples related to each have been observed. 2-Azido-5-t-butyl-l,4-benzoquinone(40) and 2-azido-l,4-naph- Me3C4"3 Me$ @ToHSCH2CO2Et 5Me3COHMe3CO-K' ~ 0 0 (40) (42) 83% PhSHMe3COHN3 MqCO-K+ * SPhWNH2 0 0 (41) (43) 73% (44) 70% (461 0 0 38 M. Regitz, Synthesis, 1972.351. 424 Moore thoquinone (41) readily react with thiol nucleophiles. For example, the former reacts with ethyl mercaptoacetate to give the heterocyclic quinone (42) and the latter reacts with thiophenol to give the aminoquinone (43).37 More interestingly, the enolate anion of diethyl malonate reacts with (40) giving a 70% yield of the indolequinone derivative (44),presumably arising via the azidohydroquinone (45) and aminoquinone (46) intermediate^.^? Mosby and Sil~a~~-~O have investigated the reactions of certain 2,3-diazido- 1,4-quinones with phosphines and phosphites. When two molar equivalents of triphenylphosphine were added to a solution of 2,3-diazido-l,4-naphthoquinone (47), 2,3-triphenylphosphoranylidenearnino-1,6naphthoquinone (48) and the interesting and unanticipated triazoline (49) were isolated.The ratio of these products was markedly dependent upon the solvent employed; in benzene the ratio (48): (49) was 1.0: 0.8 whereas in dichloromethane it was 0.09:l.O. There (47) J 111 N 0 0 QQN=pph3 N3 0 0 PPk* (49) 0 3f G. Cajipe, D. Ratolo, and H. W. Moore, Tetrahedron Letrers, in the press. 38 W. L. Mosby and M. L. Silva, J. Chem. SOC.,1964, 3990. sB W. L. Mosby and M. L. Silva, J. Chem. SOC.,1965, 1003. W.L.Mosby and M.L. Silva, J.Chem. Soc., 1965,2727. 425 Chemistry of Azidoquinones and Related Compounds 0 R20 &:>1R2 R3 R4 P N-N=P-R3 ~a:: I. R’ 0 R’ 0 R4 (52) R’ R’ R* R4 Yield % a; c; d; b; NOI NHCOMe H NHS Ph Ph Ph Bu Ph Ph Ph Bu Ph Ph Ph Bu 53 37 57 21 e; H 50 f; H 49 appears to be a strong tendency for triazoline ring formation, since the other vicinal diazides (50), (51), (52a-f), (53), and (54) also gave the heterocyclic ring system upon reaction with phosphines. However, o-diazidobenzene (55) reacts with triphenylphosphine in a completely ‘normal’ fashion to give the mono- (56) and bis- (57) triphenylphosphoranylideneamino-compounds. When one molar equivalent of triphenylphosphine was treated with 2,3-diazido-1,4-naphthoquinone(47) a 65% yield of the triazoline (49) along with a 7 % yield of 2-azido-3-triphenylphosphoranylideneamino-l,4-naphthoquinone was obtained.3e In contrast, when (47) was treated with one equivalent of trimethyl phosphite, (58) was formed in 87 % yield and upon acid hydrolysis it - 0 0 N3GNPh P~I~P=N-N:~GNP~ N/ N3 0 0 gave the phosphoramidate (59) in 84 % yield.POPercMoro-2,3-diazido-l,4-naphthoquinone behaved similarly.The phosphorimidate(58) did not react with anotherequivalent of trimethylphosphite but it didreactwith triphenylphosphine to give (60) and (49). 0 0 Chemistry of Azidoquinortes and Related Compounds 0 0 Me 0 (61) A most interesting transformation was observed when (59) was treated with one equivalent of trimethyl phosphite; a 33 % yield of 1-methyl-1H-naphtho-[2,3-dJtriazole-4,9-dione(61) was obtained.It is proposedqo that (61) arises via a mechanism in which the azide group suffers the loss of the single terminal nitrogen, a reaction which appears to have no precedent in azide chemistry. D. Acid-catalysed Rearrangements of Azidoquinones.-2-Azido-l,4-benzo-and -1,4-naphtho-quinones and certain 2,5-diazido-l,4-benzoquinonesundergo a stereospecific rearrangement to y-cyanoalkylidene-Aa*b-butenolides(62) when decomposed in cold (0-5 "C)concentrated sulphuric a~id."~J~ The general structures (62) and (63) illustrate the overall chemical transformation. This reaction generally proceeds in high yields and gives the butenolide in which the cyano-substituent is trans to the lactone oxygen.This reaction has been shown4 to involve an initial protonation on that carbonyl oxygen which is in direct conjugation with the azide group to give the iminodiazonium ion (64).This intermediate suffers heterolytic cleavage with nitrogen loss in the rate-determining step to give (65), which then undergoes o-acylation to the butenolides (62). These butenolides (62) are actually masked active methylenes, i.e. (66), and should find synthetic utility as such. 0 0 0 OH (63) (64) 0 ':OH (62) (65) For R1-R3, see Table 1 428 Moore 0 0 NC k;1(63) An example of this rearrangement is the synthesis of vulpinic acid, a natural product occurring in a number of lichens.Commercially available 2,5-diphenyl- 1,4-benzoquinone (67) was converted into the natural product (68) as shown. 0 ph* Ph 0 0 (67) 69% HCI -dioxan1 0ph0N3< N3- 90% HO Ph HOph+cl Ph 0 0 (69) HCI -McOH -HO 95 % HO (62i) Chemistry of Aziabquinones and Related Compounds Substituents in compound (62) R1 Rg R8 Yield % H Me H 65 H CMe, H 95 H Ph H 80 H NHS H 44 Me H Me 70 Ph H Ph 30 Me NH, CHM% 94 CHMQ NHS Me 95 Ph OH Ph 65 H Ns H 73 CHMe, N3 CHMe, 87 Me N8 Me 87 H -CH=CH-CH=CH- 59 Me --CH=CH GH=CH- 95 CMe, Br H CMe, CM% H 87 86 Me H CHMes 87 The conversion of (69) into (62i) required no external source of acid.Simply refluxing a solution of the azidoquinone (69) in chloroform for a few minutes induced its rearrangement to the butenolide in 65% isolated yield. Here, an intramolecular acid-catalysed process can be envisaged, as shown. This rearrange- ment is not simply a thermal process since, as is discussed below, azidoquinones thermally ring-contract to 2-cyanocyclopent-4-ene-l,3-diones. 0 h Ph OH (69) HO Oy" NC Ph Moore Earlier, it was pointed out that thymoquinone (20) rearranges to the butenolide (21) when treated with hydrazoic acid in trichloroacetic acid at 64 0C.z7028One of the key steps in this reaction has been shown to be an example of the acid- catalysed rearrangement of an azidoquinone, namely, the rearrangement of (70).a HN3 -CHMe2 CHMe2 0 OH (20) HN3 H2N CHMQ~ = CHMe2 OH 0 The rearrangement of azidoquinones to butenolides appears to be quite generaI. Exceptions which have been reported are for 2,5-diazido-l,4-benzo- quinones (71) with bulky substituents (t-butyl and t-pentyl) in the 3-and 6-positions, and for 2-azido-3-vinyl-l,4-naphthoquinones(72).The former react to give the tetrazoles (73)*l via the butenolide intermediate (74). The latter do not rearrange, but rather undergo ring-closure to the respective indolequinones (75).'l W. Weyler, jun.,P. Germeraad, and H. W. Moore, J. Org. Chem., in the press. 431 Chemistry of Azidoquinones and Related Compounds h250h N3 y33 0-50cR 0 (71) N'II88N (74)J N-N R = CMe, 76% R = CMe,Et 80% @7JR 0 0 (72) 6-NR -0 H 0 R-Me67% R = Pr 94% R = Ph24% 432 Moore The chemistry of azido-1 ,Zquinones has barely received attention.The only known member of this series is 4-azido-1 ,2-naphthoquinone (76). Interestingly, this azide undergoes facile ring-expansion to 2,5-H-4-hydrozybenzoazepine-2,5-dione (77) when treated with cold concentrated sulphuric acid.e In passing, it is interesting to note the marked difference between chemistry of the iminodiazon- ium ions (am) and (78); the former ring-contracts and the latter ring-expands. 0 0 (76) E. Thermal Rearrangements of 2-Azido-l,4quinones.-Thermal decomposition of azidoquinones in refluxing benzene or toluene results in their smooth re- arrangement to 2-cyanocyclopent-4-ene-l,3-diones(79).6 This constitutes an efficient entry into this carbocyclic ring system which is of significant importance. For example, various natural products are 2-acylcyclopent-4-ene-l,3-diones,i.e., linderone,42 methyl-linderone,42 Iu~idone,4~ ~alythrone,4~methyl-l~cidione,4~ and a number of hop ~~n~tituent~.~~ Various 2-substituted indane-l,3-diones show marked pharmacological activity as anticoagulant^.^^ Pyrethrins*' are among the most important natural insecticides and are related structurally to the cyclo- pentenedione ring Indeed, even the prostaglandins,4@ which are of 41 A.K. Kiang, H. H. Lee, and K.Y. Sim, J. Chem. SOC.,1962,4338. 43 H.H.Lee, Tetrahedron Lerters, 1968, 4243. R. 0. Hellyer, Austral. J. Chem., 1968,21,2825. 46 R.Stevens, Gem. Rev., 1967.67,19. 46 R.Biggs and R. G. MacFarlane, 'HumanBlood Coagulation', Oxford, 3rd Edn., Oxford University Press, Oxford, 1962. L. Crombie and M. Elliot,Fortschr. Chem. org. Nutursrofe, 1961, 19, 120. R.A. Lee Mahiew, M. Carson, and R. W. Kierstead,J. Org. Chem., 1968,33,3660. 4s R.Clarkson, Progr. Org. Chem., 1973,8, 1. 433 Chemistry of Azidoquinones and Related Compoundr pivotal biological importance, can be viewed as derivatives of the partially reduced cyclopentene-1,3-dione ring system. Of importance here is the fact that this basic ring system can be conveniently prepared in good yield from the readily available 2-azido-l,4-benzo- and -ly4-naphtho-quinones.The general structures (63) and (79) illustrate the synthetic scope of this transformation. For R1-R8, see Table 2. Table 2 Substituentsfor compound (79)R' Ra R8 Yield % Me H Me 92 Ph H Ph 70 Me NHa CHMel 89 CHMe, NHl Me 92 CHMel CHMe, c1 31 CMe, H CMe, 95 Me NHa Me 89 NC7HlO H CMe, 78 Br H CMeS 75 Me CMe, H 80 Me Ph Me 82 Me p-MeOC& Me 87 Me P-No8cSH4 Me 65 Me <H=CH-CH=CH- 95 OMe -CH=CH-CH=CH- 70 The mechanism of this reaction has been studied in some detail? Based upon product analysis, activation parameters, and the absence of kinetic solvent effects and substituent effects, the mechanism shown in Scheme 1 has been presented.The synthetic limitations of this reaction thus far reported are for those azidoquinones in which the substituent adjacent to the azide group is a proton, a vinyl group, or a substituted amino-function. Those which are unsubstituted give a complex mixture of products upon pyrolytic decomposition, whereas those which are vinyl substituted are thermally converted into indolequinones,ll i.e., Moore slow j$4N3 ___I) -N2 0 Scheme 1 (80) -+ (81). The azidoquinones having an aryl-or alkyl-substituted amino-group in the 2-position pyrolytically decompose to give polynuclear heterocyclic quinones and are discussed later. 0 0 0 (80) R' R' RB Yield % a; Me KH=CH-CH=CH- 90 b; Pr <H=CH-CH=CH- 81 c; CH,(CHA,Me -CH=CH-CH=CH- 87 d; Ph -CH=CH<H=CH- 92 e; f; CH,(CHJOCOMe Ph --CH=CHKH=CH-Me Me 88 66 The synthesis of indolequinones as outlined above constitutes one of the best routes to this class of compound. An illustration of the synthetic utility of this reaction is the construction1' of 6,7-benzo-2,3-dihydro-5,8-dioxo-1H-pyrolo-[1,2u]indole (82), the naphthoquinone analogue of the biologically important 435 Chemistry of Aziabquinones and Related Compounds mitosene ring systems.so Hydrolysis of 2-(3-acetoxypropyl)benzoLflindole-4,9-dione (8le) in reflwing aqueous methanolic hydrogen chloride gave the alcohol (83) in 94% yield.Reaction of this alcohol with toluene-p-sulphonyl chloride in pyridine gave the tosylate (84)in 57 % yield which upon reaction with potassium t-butoxide in t-butyl alcohol gave (82) in 88% yield.0 0 OCOMe A ~ C6H6 88 7* 0 (elel bCOMe 0 0 " (83) OH N 0 A most interesting reaction has been reported by Van Allen, Reynolds, and Ade151@ who have observed heterocyclic quinone (85) formation when 2-arylamino-3-chloro-1,4-naphthoquinones(86)are treated with sodium azide in dimethylformamide at 90-100 "C. The corresponding azidoquinones (87)are the presumed intermediates in this reaction and the products are assumed to arise by nitrene insertion followed by dehydrogenation. so G. R. Allen, jun., J. F. Poletto, and M. J. Weiss, J. Org. Chem.. 1965, 30,2997.J. A. Van Allen, G. A. Reynolds, and R. E. Adel, J. Org. Chem., 1963, 28, 520. J. A. Van Allen, G. A. Reynolds, and R. E. Adel, J. Org. Chem., 1963, 28, 524. Moore (86) (87) J 0 R = OMe (46%), C1, Me, or OH The same authors also observed that 2-chloro-3-alkylamino-1,4-naphtho-quinones(88)-(93)react under thesameconditions togive higher yields of hetero-cyclic quinones than the arylamino-derivatives. In one case an azidoquinone was isolated and shown to an intermediate, i.e., 2-azido-3-morpholino-1,4-naphtho-quinone (94) smoothly decomposed under thermal conditions to give 1,2,3,4,5,10- hexahydro-5,10-dioxo-2-oxa-4a,1 1-diazabenzo [bIfluorene (95) in 49 % yield. (88) 0 437 Chemistry of Azidoquinones and Related Compounds 0 0a:>Me DMFN3-, 0 OMF@r3 0 CH,Ph 0 CH2Ph (91) Moore F.Photolysis of Azidoquinones.-Photolysis of azidoquinones in benzene with 3600A light results in transformations which are analogous to those observed for their thermal decompositions. That is, 2-azido-3-alkyl(aryl)-l,4-quinones ring-contract to 2-cyano-3-alkyl(aryl)cyclopent-4-ene-l,3-dionesss and 2-azido- 3-vinyl-l,4-quinones ring-close to the corresponding indolequinones.ll The photolytic ring-contraction has particular advantages for the synthesis of 4-azido-2-cyanocyclopent-4-ene-l,3-diones(96) from the corresponding 2,s-diazido-l,6benzoquinones(97). As discussed below, these diazidoquinones also thermally ring-contract to the azidocyclopentene-l,3-diones.However, such products readily cleave under the reaction conditions giving two molecules of the correspondingly substituted cyanoketen.No such cleavage is observed when the photolytic ring-contraction is carried out in benzene with 3600 A light.l2sS4 0 0 U (97) (96) R = CMe8(7S%), CHMe2(75%), or Me(65%) 2-Azidoquinones which are unsubstituted at the 3-position decompose photolytically as well as thermally to give a complex mixture of products. However, if the photolysis is carried out in the presence of dienes, 2-alkenyl-2,3- dihydroindole-4,7-dionesare obtained in generally good yields.1° Specifically the heterocyclic quinones (98)-(102) have been prepared in this manner. (63b) (99) 53% 63 P. Germeraad, W.Weyler, and H. W. Moore, unpublished results. 64 H. W. Moore and W. Weyler, jun., J. Amer. Chem. SOC.,1971, 93,2812. Chemistry of Azidoquinones and Related Compounds (63a) (100) 37% ”0 0 (63m1 (1011 74% ? R6 (63) (102) R’ Ra R3 R4 R6 R6 Yield % H CMe, Me H H H 96 H CMe, H Me H H 73 H CMe, Me H H Me 66 H CMe, H Me Me H 53 H Me Me H H H 40 H Me H Me Me H 84 -CH=CH-C H=CH -Me H H H 72 -CH=CH-CH=CH -H Me Me H 82 On the basis of kinetic and stereochemical studies, the mechanism shown in Scheme 2 has been presented. Pertinent data which are consistent are the facts that the rate of nitrogen evolution is dependent upon diene concentration and that regiochemistry and stereochemistry are nicely accounted for in terms of the proposed intermediates, i.e.,the reaction is regiospecific, giving only the 2-alkenyl isomer, and stereoselective, giving as the major isomer the one having a cis relationship between the substituents at the 2,3-positions regardless of the stereo- chemistry of the starting diene.Moore 0 bv * MeMe-\ 0 0 CHMe CHMe Scheme 2 One novel feature of this transformation is the photolytic cycloaddition of the organic azide to the carbon-carbon double bond of the diene. Such a reaction is certainly well known in the thermal chemistry of organic a~ides,~~ but appears to be without precedent under photolytic conditions. Cycloadditions of this type may be limited to those azides which can accept light of relatively low energy (>3600 A) such as the highly coloured azidoquinones.Light of higher energy may result in nitrene formation, thus leading to other products. G. Thermal Cleavage of 2,5-and 2,6-Diazido-l,4quinones.-It has been shown that the thermal ring-contraction of monoazidoquinones to 2-cyanocyclopent- 66 G. L’Abbe, Chem. Rev., 1969,69,345. Chemistry of Azidoquinones and Related Compoundr 4-ene-1,3-diones (79) results from electrocyclic ring-closure of a zwitterionic intermediate (see Scheme 1). On the basis of this mechanism one would predict that 2,5-diazido-l,4-benzoquinoneswould thermally generate an analogous ring-opened intermediate which could partition itself between electrocyclic ring-closure and cleavage to two molecules of a cyanoketen.In fact, when 2,5-diazido-3,6-di-t-butyl-l,4-benzoquinone(103a) was refluxed for a few minutes in anhydrous benzene, t-butylcyanoketen (104a) was formed in >95% yield as a stable cumulene in s01ution.~~~~~ When the reaction was closely monitored by t.l.c., 4-azido-2,4-di-t-butyl-2-cyanocyclopent-4-ene-1,3-dione(105a) was also detected. As pointed out earlier, photolysis of the diazidoquinone (103a) with 3600A light in benzene gave a 75 % yield of the cyclopentenedione (l05a) and no keten. However, (105a) was quantitatively converted into t-butylcyanoketen in refluxing benzene. The scope of this reaction has not yet been extensively probed, but it has been shown that t-pentyl, isopropyl-, methyl-, and phenyl-cyanoketen canbe generated in an analogous fashion. The t-pentyl homologue, like t-butyl- cyanoketen, is stable in solution; the others are not and were isolated as their methyl esters by trapping with methanol.0 0 (103) It 0 2 “>=,=o N3Rf&N NC 0 (104) a; R = CMe, b; R = CMe,Et c; R= CMe2H d; R=Me e; R= Ph E.~H. W. Moore and W.Weyler, jun., J. Amer. Chem. SOC.,1970, 92, 4132. Moore 2,6-Diazido-l,6quinones (106) also thermally cleave to cyanoketens.'g An initial ring-contraction to 2-cyano-4-azidocyclopent-4-ene-l,3-diones(105), the same intermediate as is formed from the 2,5-diazido-isomers, is followed by electrocyclic ring-opening and subsequent cleavage to give the ketens (104).C6"6 RA , "'eR 0 0-(106) 1 0 c N3vNR R 0 2 "+o NC (104) R = CHMe, or Me Only two previous reports have appeared regarding the synthesis of cyano- ketens. De Selms6' and Schmidt and Reids8 have independently reported the unique formation of phenylcyanoketen upon reaction of 2-halogeno-1 -phenyl- cyclabut-l-ene-3,4-dione(107) with sodium azide. CI _LMeCN ph)=c=oN3-NCPh (107) (104) 67 R. C. De Selms, Tetrahedron Letters, 1969, 1179. A. H.Schmidt and W. Reid, Tetrahedron Letters, 1969, 2431. 443 Chemistry of Azidoquinones and Related Compounds The generation of cyanoketens by the classical dehydrohalogenation of the corresponding acid chlorides may be difficult.This is csrtainly true for t-butyl-cyanoketen (104a) since reaction of a benzene solution of 2-cyano-3,3-di-methylbutyrylchloride (108) with a catalytic amount of triethylamine gives a good yield of 1,3-di-t-butyIallene( The same product was immediately formed when a solution of the keten was similarly treated. The reaction was shown to Me3CCH-COCI [Et3NI > Me3c>c=o]NCCN C6H6 (108) (104a1 II t 0 C-NEt NC 0-NC CMe3 CMe3 NC-HNEtt + NEt3 NCTo NC CMe3 CMe3 (1 10) 0 II + 0/C-NEt3 (109) 60 H. W. Moore and W. G.Duncan, J. Org. Chem., 1973,38,156. Moore involve a triethylamine-catalysed dimerization of t-butylcyanoketen to the p-lactone (1 10) which then reacted further with the amine, as indicated, to give the allene (109).On the other hand, when t-butylcyanoketen is generated by pyrolysis of 2,5-diazido-3,6-di-t-butyl-l,4-benzoquinone(103a) it is stable for days in benzene even at the reflux temperature. Even though t-butylcyanoketen is reluctant to self-condense in benzene, it quite readily undergoes cycloaddition to other substrates. Addition to alkenes,12 a1 kynes,6O allenes,6l and carbodi-irnidesl2 have been reported, and selected examples are outlined in Scheme 3. NC' 'H Scheme 3 '* M. D. Gheorghiu. C. Draghici, L. Stanescu, and M. Avram, Tetrahedron Lnrers, 1973, 9. O1 W. Weyler, jun., L. B. Byrd, M. C. Caseria, and H. W. Muore, J. Amer. Chem. Soc., 1972, 94, 1027. Chemistry of Azidoquinones and Related Compounh H.Thermal Rearrangement of 2,3-Diazido-l,4quinones.-2,3-Diazido-l,4-quinones (1 11) undergo a fascinating rearrangement to 2-aza-3-cyano-l,4-quinones (1 12) when decomposed in refluxing chlorobenzene.* This reaction constitutes the first unambiguous synthesis of the new heterocyclic azabenzo- quinone ring system. All other reported examples are hydroxy-derivatives which have several tautomeric possibilities, the azaquinone form being only one, and no evidence has been presented which would allow one to determine which isomer or isomers In the azanaphthoquinone series one example has recently been described; 2-aza-3-phenyl-lY4-naphthoquinonehas been prepared by two different r~utes.**~~~~~* PhCl A' CN 0 0 R1 R* (112) CM% H H CMe, Ph H H Ph Me Me --CHICH&H&HS-<HSHXH=CH-The conversion of 2,3-diazido-l,4-naphthoquinone(1 1lg) into 2-aza-3-cyano- l,4-naphthoquinone (1 12g) warrants further comment.Van Allen, Priest, Marshall, and Reynolds7 have reported that 2,3-diazido-1 ,4naphthoquinone (1 1 lg) is pyrolytically converted into phthaloyl cyanide (1 13) in refluxing toluene. 0 PhMcWN3N3 A* 0 0 H. J. Knackmuss, Angew. Chem., 1973,85,163."J. A. Moore and F.L. Marascia, J. Amer. Chem. SOC.,1959,81,6049. J. H. Boyer and S. Kruger, J. Amer. Chem. SOC.,1957,79, 3552. ab H. Ost, J. prakr. Chem., 1890, 27, 260. R. Kudernatsch, Monarsh., 1897,18, 613. a' K.Schedcer, Helv. Chim. Acru, 1968,51,413.I. Felnet and K. Schenker, Hefv. Chfm.Acta, 1969,52, 1810. Moore This has been confirmed, and it has been shown that, in addition, the azaquinone (112g) is also generated.8 The propensity of azidoquinones in general to undergo thermal ring- contraction to 2-cyanocyclopent-4ene-1,3-dionessuggested that such a process is also involved in the conversion of 2,3-diazidoquinones into azaquinones. Indeed, this was shown to be true. Thermolysis of 2,3-diazido-5-t-buty1-1,4-benzoquinone (1 1la) in refluxing toluene gave predominantly 2-azido-4-t-butyl- 2-cyanocyclopent-4-ene-1,3-dione(1 14) in 80% yield. This compound, upon subsequent thermolysis in refluxing chlorobenzene, gave the azaquinones (1 12a) and (1 12b) as a 1:1 mixture. This last transformation finds a precedent in the previously reported conversion of 2-azido-2-phenylindane-1,3-dioneinto 2-aza- 3-phenyl- 1,4-naphthoquin0ne.~~ PhMe A N3 Me$Me$ &3 0 0 0 0 0 The reactivity of the azaquinones centres around the very electron-poor imine double bond.This is illustrated for 2-aza-3-cyano-5-t-butyl-1,4-benzoquinone (1 12a), which readily undergoes nucleophilic additions as well as cycloadditions. 447 Chemistry of Aridoquinones and Related Compounds 0 0 0 pp?Me$ CN0 4 Reactions of Related Systems A large number of cyclic azidoeneones and related compounds can be envisaged which might be chemically similar to the azidoquinones. A few of these are shown here [(I 15)-(125)], and some have received limited study.It is also possible that reactions analogous to those described for azidoquinones might be induced from related vinylogous amides, such as aminoquinones, under oxidative conditions, e.g. the generation and chemistry of nitrenoid species from aminoquinones upon lead tetra-acetate oxidation. The following sections summarize the limited results in these areas. A. Thermolysis and Pyrolysis of 4-Azido-1,2-pyridazine-3,ddione.4-Azido-l,2-dimethylpyridazine-3,6-dione(126) was subjected to pyrolytic and photolytic decomposition in the nucleophilic solvents methanol and diethylamine.6@ Its thermal decomposition in the alcohol at I30 "Cgave a mixture of the amino- derivatives (127) and (128) in 25 and 10% yields, respectively. On the other hand, photolysis in methanol gave (129), (127), and (130) in 44, 20, and 5 % yields, respectively.Thermolysis of (126) in diethylamine gave an 8 1 % yield of 4-amino- 5-diethylamino-l,2-dimethylpyridazine-3,6-dione(131). A nitrene intermediate has been suggested as the precursor of these products. Formation of the ring-contracted product (1 30) is anaIogous to that observed for the thermolysis and photolysis of monoazido-l,4-quinones. As pointed out earlier, such a process in the quinone series is not facile when the position 60 T.Sasaki, K. Kanematsu, and M. Murata, Tetrahedron, 1973, 29, 529. Moore adjacent to the azide group is unsubstituted. This is apparently true also in the pyridazine series since (1 30) was formed in only 5 % yield.Thus, 4-azido-5-alkyl (aryl)-l,2-pyridazine-3,6-diones should be prepared and their thermolysis and photoIysis investigated. CN CNg3 449 4 Chemistry of Azidoquinones and Related Conipounds 0 0 0 MeOH Me" OMe 0 0 0 0 Me" NEt* 0 Unlike the azidoquinone series, (126)is reported to be stable in strong acid, even at 100°C.69 B. Thermal Cleavage of 2.3-Diazido-l,4-naphthoqu~onedibenzenesulphonimide.-The thermal decomposition of 2,3-diazido-1,4-naphthoquinonedibenzene-sulphonimide (1 32) in refluxing benzene gives phthaloyl cyanide dibenzene- sulphonimide (133). 'O The a-cyanobenzenesulphonimide groupings in (133) are very reactive towards nucleophilic reagents. For example, water very readily 'O H.W. Moore and M. S. Lee, Tetrahedron Letters, 1971, 3645. Moore reacts with (133) to give the N-benzenesulphonyl-lactam(134) in 85% yield. This cleavage reaction finds analogy in the previously described thermolysis of 2,3-diazido-1,4-naphthoquinone(11 lg) to phthaloyl cyanide (113). As a result, it is possible that the readily a~ailable~l-’~ azido-l,4-quinonedi-iminesmay be chemically similar to the azidoquinones, and their detailed study is thus warranted. ,S02Ph S02Ph N N’ N ‘S02Ph N\ SO,Ph (132) (133)p NHS02PhNC (134) C. Thermal Rearrangement of 3-Azido-2,5H-azepine-2,5-diones.-2,5H-Azepine--2,s-diones are readily prepared from 1 ,dbenzo- and 1,4-naphtho-quinones upon their reaction with hydrazoic acid in cold concentrated sulphuric acid, e.g., (16) -+ (17).2a-aaAn azido-derivative in this series has recently been subjected to thermal decomp~sition,~~ arid again ring-contraction was observed.Specific- ally, 3-azido-2,SH4methylbenzoazepine-2,5-dione(135) rearranged in 80 % yield to the quinoline derivative (136) when a chlorobenzene solution was refluxed for two hours. 71 R. Adams and D. C. Blomstrom, J. Amer. Chem. SOC.,1953,75,3405. R. Adam and W. Moje, J. Amer. Chem. SOC.,1952,74,5560. 73 R. Adams and W. P. Samuels, J. Amer. Chem. SOC.,1955,77,5357. R. Adams and J. W. Way, J. Amer. Chem. SOC.,1954,76,2763. 7s R. Adams and L. Whitaker, J, Amer. Chem. SOC.,1956,78, 658. R. Adams and E. L. DeYoning, J. Amer. Chem. SOC.,1957,79,417.77 G. Landen and H. W. Moore, unpublished results. 451 Chemistry of Azidoquinones and Related Compounds WN3 A, Me PkCI 0 D. Lead Tetra-acetate Oxidation of Aminoquinones.-The reaction of primary amides with lead tetra-acetate has been investigated and found to parallel the Hofmann rearrangement. * Therefore, primary aminoquinones, being vinylogous amides, might be expected to react with lead tetra-acetate to give 2-cyanocyclo- pent-4-ene-l,3-diones via an intermediate nitreneoid species. That is, such reactions might be similar to those observed for the thermal reactions of azido-quinones. Limited studies indicate that there is some foundation for such an analogy. For example, 2,5-diamino-3,6-di-t-butyl-l,4-benzoquinone(137) gives a 75% yield of 4-amino-2,5-di-t-butyl-2-cyanocyclopent-4-ene-1,3-dione(1 38) upon lead tetra-acetate oxidation in chloroform.79 Under the same conditions, the 2-amino-5-anilino-3-acyl-l,4-benzoquinones(1 39a-c) give good yields of the corresponding isoxazoles (140a-c). 79 This latter transformation is analogous to that observed when 2-acetyl-3-chloro-6-methoxy-1,4-benzoquinone(141) was treated with sodium azide to give the isoxazole (142).*O 0 0 0 (1371 (138) 0 0 phHN$$; 0 (139) (140) a; R= Me a; 75% b; R=OMe b; 72% c; R = NMe, c; 65% J. B. Aylward, Quart. Rev., 1971,407. 79 H. W. Moore and W. Schiifer, unpublished results. 80 W. Schllfer and Hj. Schlude, Tetrahedron Letters, 1967, 4313.Moore MeoWMe MeOHNaN3-[ ] 0 (142) In the past thirty-five years alone well over one hundred and fifty richly substituted primary amino-1 ,dbenzo- and -1,dnaphtho-quinones have been reported in the literature. The plethora of such readily available starting materiaIs along with the rich chemistry of the amino-group and the quinone nucleus should make a study of their oxidative rearrangements most worthwhile. 5 Conclusions It is apparent from this review that the chemistry of azidoquinones and particu- larly of the related azidoeneones is still in its early stages of development. Many of their rearrangement, fragmentation, and cleavage reactions can be formally outlined according to Scheme 4. The penultimate precursor to the zwitterionic species (143) may be the azide (144), a nitrene, or an azirine.The group X is cation-stabilizing and Y and/or 2 are anion-stabilizing substituents. Such a scheme adequately rationalizes the observed conversions of azidoquinones into y-cyanoalkylidene-Aasfl-bu t enolides (62),2-cyanocyclopen t -4-ene- 1,3-diones(79), 2-aza-3-cyano-l,4-quinones(112), and cyanoketens (104). It is also in agreement with the observed ring-con tractions of 4-azido-1,2-dirnethylpyridazine-3,6-dione (126) and 3-azido-2,5H-4-methylbenzoazepine-3,5-dione(135). Certainly a large number of other cyclic and acyclic vinyl azides meeting the structural requirements outlined above should be investigated. Some such work has already been done and is outlined be lo^.^^-^* J.D. Hobson and J. R. Malpass, J. Chem. SOC.,1967, 1645. D. Knittel, H. Hemetsberger, R. Leipert, and H. Weidmann, Tetrahedron Letrers, 1970, 1459. 83 S. Sato, Bull. Chem. SOC.Japan, 1968, 41,2524. S. Maiorana, Ann. Chim. (Italy), 1966, 56, 1531. 453 Chemistry of Azidoguinones and Related Compounds N -[(X,] (1441 (143) Scheme 4 R'CH=CN~COR~ A R'--CH-COR~ I CN Yield % 73 65 74 80 75 70 74 45 Moore 0 .Me 0I/ II /MeCH=C ___) I \ NL Ph + PhCCH\ I 'COPh qMe'C NN3--d I wish to express my sincere thanks to the National Science Foundation and the National Institute of Health, who provided financial support for much of the work described here. Also I am most grateful to Dr. W. Schafer, of the Max Planck Institut fiir Biochemie, who provided space and facilities as well as stimulating discussions during the preparation of this review.455

ISSN:0306-0012    

DOI:10.1039/CS9730200415

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Chemistry-a Topological Subject By R. W. Jotham THE KESTEVEN COLLEGE OF EDUCATION, STOKE ROCHFORD, LINCOLNSHIRE, NG33 5EJ 1 Introduction Metric geometry is governed by the concepts of congruence of line segments and angles. In contrast, tqology is the geometry of properties which remain invariant under topological mappings, i.e. mappings such that they and their inverses are single-valued and continuous. These maps may be expressed by groups of continuous functions. Topology is therefore a major division of abstract mathematics with applications in concrete situations that range from knots and knitting to systems of connected points such as railway networks, trees, and chemical formulae and to the properties of strange and complex sur- faces such as the Moebius strip.l Various aspects of topology have been applied to structural and other problems in chemistry, but the aspect which has so far proved to be most useful is the theory of mathematical graphs.2 Graphs are collections of points and lines, drawn so that pairs of the points are connected together.Such a graph may, for example, illustrate the operations of any symmetry group. It may also be a concise state- ment of our preconceptions about the connections (which may be independent of symmetry) within a system based, hopefully, on reasonable premises. To emphasize this dependence of the connectivity graph on our prejudices we reproduce in Figure 1 a religious graph which symbolically represents the Christian Trinity. This diagram expresses relationships in close accord with those embodied in a valence-bond description of ammonia-in both cases there is a concise statement of unprovable but established understanding of the main features of the system of interest, The special value of chemical graphs, however, is they may often also be used to obtain additional information about the system.Indeed, very powerful semi-empirical conclusions may frequently be obtained by the sympathetic use of the symmetry and topology of the pr~blem,~~~ par-ticularly in the cases of polygonal and polyhedral molecules and other chemical sy~tems.~~ B. Mendelson, ‘Introductionto Topology’,Allyn & Bacon, Glasgow, 1968. D. H. Rouvray, R.1.C. Rev., 1971, 4, 173. a H. H. Schmidtke, Co-ordination Chem.Rev., 1967, 2,3.H. H. Schmidtke, Internat. J. Quantum Chem., Symp., 1968, ZS,101.‘R.B. King, J. Amer. Chern. Soc., 1970,92, 7211, 7217. H. H. Schrnidtke, Theor.Chim. Acra, 1968,9, 199. A. Graovac, I. Gutman, N. TrinajstiC, and T. ZirkoviC, Theor. Chim.Acra, 1972,26,67. S. F. A. Kettle and V. Tomlinson, Theor.Chim. Acta, 1969,14, 175. S. F. A. Kettle, Theor. Chirn. Acta, 1966, 4, 150. 457 Chemistry-a Topological Subject Figure 1 We shall now show, mainly by reference to square-planar and tetrahedral molecules, how the general solution of some topological problems may be applied to a variety of situations of chemical interest (Section 2). In Section 3 we give detailed examples designed to help the reader to apply similar methods to other systems of interest.The information which we shall obtain consists of energy levels that refer to equivalent components of the system. These are calculated algebraically in terms of a limited number of connectivity parameters variously called ‘resonance integrals’, ‘interaction constants’, ‘coupling constants’, etc. according to the context. In the simplest cases only a single connectivity para- meter, K,may be needed, and the calculated energy levels take the form Eg = Eo + CZK.Much may be deduced even from knowledge of the sign and order of magnitude of K. Secondly,in a case with a manifold of energy levels, the empirical determination of a single energy level separation suffices to determine the energies of all levels in the manifold.In general, useful information can always be obtained so long as the number of important and distinctive parameters does not exceed the number of relevant experimental data. 2 Connectivity in Familiar Chemical Ideas The manner in which a molecule is thought to be connected together enables us to select those interactions which are important and those which are of little significance. Thus both CHI and P4have the symmetry of a regular tetrahedron, Jotham but the topology of the bonding schemes which are intuitively required by the chemist are quite different. The topology of the valence bonds in phosphorus is, however, the same as that of the bonding-pair electron repulsions and the proton nuclear spin coupling in methane.In Figure 2 we show the structure of these and other tetrahedral molecules BehO( MeC02)G (simplified 1 Figure 2Topological diagrams for tetrahedral molecules and ions with a diversity of bond connectivity graphs. The high, Td, symmetry of each compound enables us to solve problems within each set of equivalent Chernistry-a Toporogical Subject atoms exactly in terms of the interaction between any pair of equivalent atoms. Nevertheless, we can see immediately that, in some cases such as the [CU~OC~~,,]~- ion, the interactions between equivalent atoms are unlikely to be the most im- portant! Rather we would expect that the important interactions affecting the properties of the [CU~OC~~~]~- ion are those between the set of four copper atoms and each of the other sets of atoms.Clearly topology will not help us much in this case, whereas it is likely to dominate the discussion of all of the properties of P4.There are many other cases in which topologically distinct figures have a common symmetry.s A familiar example is provided by the pairs of Platonic solids of oh and Ih symmetry. A decision on the topology of a chemical problem may be of more funda- mental importance than a decision on its real or approximate symmetry, since these decisions must be based on a realistic appraisal of energetic factors. Consider the bonding in ‘octahedral’ transition-metal complexes. True octahedral complexes are not common but, nevertheless, the most important properties of the majority of distorted complexes may be interpreted in terms of Oh symmetry, to which modifications due to small distortions are added if required.The recog- nition of a superclass of ‘octahedral’ complexes is itself essentially a topological perception. In these metal complexes the body of evidence leads us to argue that the magnitude of the metal-ligand a-bonding is such as to give rise to interesting and measureable effects with decreasingly important roles assigned to mbonding, distortions, and ligand-ligand bonding interactions.lOsll The relative order of magnitude of these interactions is similar in unsaturated organic molecules, but in these cases the bonding is generally much stronger and we rarely encounter a partially filled a-orbital.Rather it is the v-orbital set which gives rise to the bulk of the properties of interest and it is the topology of the n-bonding which forms the basis of Huckel theory in two or three dimensions.* Although many chemical bonds are of similar strengths, their stretching frequencies often differ widely because of kineticenergy effects. This leads to the recognition of characteristic group frequencies in infrared spectroscopy. For similar bonds in a variety of molecules, modifications to the characteristic frequency may be meaningfully interpreted in terms of the relative strengths of the bonds. Conversely, the restoring forces for most bending and deformation modes are dominated by electron-repulsion effects of a different topology. Thus the in-plane C- H deformation frequencies of aromatic compounds are much higher than those associated with the out-of-plane modes.Molecular vibrations may be of the same intrinsic energy either accidentally or as a necessary consequence of symmetry. These vibrations will then couple together relatively strongly if the topological relationship between the groups is sufficiently important for them to ‘know’ one another, and the coupling may be recognized by the appearance of a manifold of frequencies centred on the group frequency;l2 indeed, the average- loH. B. Gray, J. Chem. Educ., 1964,41, 2. l1 S. F. A. Kettle, J. Chem. Educ., 1966, 43, 652. Is G. Herzberg, ‘Molecular Spectra and Molecular Structure: Volume 11, Infra-red and Raman Spectra of Polyatomic Molecules’, Van Nostrand, New York, 1945.Jotham frequency rule is a valuable aid to vibrational assignments. l3 Furthermore, there are certain groupings of symmetry-related vibrations which have a characteristic topology determined by their number and geometrical relationship. Thus, for example, even the vibrations of a methyl group may often be readily dis- tinguished from related vibrations of a methylene gr0~p.l~ An intrinsically topological topic is the interpretation of the ‘coupling’ between nuclear-spin moments, which is an essential feature of the analysis of high-resolution n.m.r. spectra. l5 Each significant coupling constant defines a connection which is a component part of the overall topology of the problem. This connection has a peculiar form, because it has (2Sz + 1) (2Sj + 1) sub-components which lead to the familiar dipolar coupling term, -Jzj Sdj, in the effective Hamiltonian used to interpret the system.* Analogous descriptions apply to the hyperfine interactions in e.s.r. spectroscopy and to the bonding process itself, which may also be thought of in terms of a dipolar electron-spin coupling. Unfortunately, these spin-coupling problems are difficult to illustrate concisely because of the large number of spin functions required to describe systems of interest, and since we have previously given the energy levels arising from spin- spin coupling for a large range of molecular symmetries and topological situa- tions,ls we shall omit examples of this type from Section 3 below.It should also be noted that there are many similarities within all systems in which the coupling of angular momenta is involved. For example, an effective connectivity scheme underlies the formulation of Russell-Saunders coupling, spin-orbit coupling, jj-coupling, etc. Finally, we comment on the fact that, after the recognition of elements and atoms, the most powerful perception in chemistry has been the recognition of the chemical bond. Clearly then, at least implicitly, chemists like to think of chemistry as a topological subject. 3 General Solution of Some Simple Topological Problems In this section we shall be concerned with the solution of some simple problems associated with tetrahedral and square-planar molecules and in illustrating the more general applicability of these solutions with a few other examples.We arbitrarily choose to discuss vibrations of tetrahedral molecules and the Hiickel energies of the square system in the most detail. *Aninteresting feature of the coupling constants in n.m.r. spectroscopy concerns the nature of the coupling between equivalent nuclei. This can be neglected because the selection rules appropriate to the normal n.m.r. experiment rigidly exclude transitions between energy levels which differ in this coupling energy. The complexity of the problem is thereby often strikingly reduced by neglect of these connections although, in many cases, the neglected interactions are amongst the strongest for that molecule. l3W.J. Lehmann, J. Mol. Spectroscopy,1964, 7,1. l4 A. D. Cross, ‘An Introduction to Practical Infra-red Spectroscopy’, 2nd Edn., Butterworths, London, 1964. Is J. W. Emsley, J. Feeney, and L. H. Sutcliffe, ‘High Resolution Nuclear Magnetic Resonance Spectroscopy’, Pergamon, Oxford, 1965. l6 R. W. Jotham and S. F. A. Kettle, Inorg. Chim.Acta, 1970,4, 145. 461 5 Chemistry-a Topological Subject A. Tetrahedral Systems.-The Td character table (Table 1) will be required for the problems discussed in detail below. Table 1Td Character table Ta E SC3 3c2 6S4 60d Rotations Vectors Vector products A1 1 1111 x2 + y2 + z2 A2 1 1 1 -1 -1 XYZ E 2 0200 (222 -x2 -y2 x2 -y2) Ti 3 -1 -1 -1 1 (R,,R,,Rz) T2 3 -1 -1 1 -1 (X,Y,Z) (XY,XZ,YZ) (i) C-H Stretching Vibrations of CH,.The C-H stretching vibrations of me-thane are readily shown to transform as Al + T2.The non-degenerate mode is active only in the Raman spectrum whereas the triply degenerate mode may be observed in both the i.r. and the Raman spectra. The observed modes of T2 symmetry are not quite pure stretching vibrations because deformation modes of the same symmetry are mixed in slightly and the vibrations must be ortho- gonal to the T, translations, but to a very good approximation the modes described as the Al and T, C-H stretching vibrations are observed at 2914 and 3020 cm-l, respectively.12 If wewrite the four individual C-H stretching motions as rl-r4 we may readily show that the appropriate orthonormal combinations of these unit vectors are given by (1).If a negative coefficient indicates a compression, we can readily see that these expressions are not exact, for each of the T2modes should include a slight oppos- ing motion of the carbon atom to maintain the centre of gravity, and we empha- size that the observed frequencies are those of the normal modes rather than the symmetry modes. Nevertheless, the functions #1-$4 may be used to extract useful information from the observed frequencies. To do this we must consider the energy appropriate to a single C-H stretching motion in CH4which we shall call v and the interaction energy between each pair of stretching motions which we shall call v’. In tetrahedral CH4 all the interactions are of the sametype.We may represent this set of interactions by the topological matrix (2). r1 r2 r3 r4 rlO1ll r,1011 r31101 r4 1 130 Jothum It follows, then, that the eigenvalues (i.e.the observed frequencies) of this prob- lem may be obtained by solving the secular equation (3). V-E V’ V’ V’ V’ V‘ v -E V’ V‘ V-E V’ V‘ =o (3) V’ V‘ V’ V-E This equation is grossly simplified if we utilize the symmetry functions $1-$4. We first write out a new energy matrix (4) for the (ri 11 r3) in terms of v and v’. We now transform the matrix (4) into the matrix (6)by either a simple direct expansion technique or, more rapidly, by proceeding through the inter- mediate matrix (5). Thus, for example: = 0 [after substituting for each (ri 11 rj) the appropriate element of matrix (4)] ($11 ($21 <$3I ($41I $1) v + 3v’ 0 0 0 I *2) 0 v-v’ 0 0 0 0 v-v’ 0 (6) 0 0 v-u’ Chemistry-a Top0logical Subject The solution of the corresponding secular equation (7) is trivial.v+3v’-E 0 0 0 0 0 v-v’-E 0 0 v-v’-E 0 0 = 0 (7) 0 0 0 v-v‘-E Thus the eigenvalues v + 3v‘ and v -v’ (thrice) correspond to the Al and T2 modes, respectively. If these are equated to the observed frequencies, we find that, for CHI, v = 2994 cm-l and v’ = -26 cm-l. The quantity v should be comparable for all types of C-H bond in similar molecules. Weseealso that the interaction between two stretching motions is not very great in this case (as in all hydrocarbons),reflecting the very high localization of the electrons in the well separated bonds.The sign of v’ is often of great interest ;a negative sign indicates that the out-of-phase motion is of higher frequency than the in-phase motion (i.e. the stretching of one bond facilitates the stretching of the adjacent bond in the case of CH,). (ii) The Vibrations of P4. The number of vibrations of CH, is nine, whereas P4 has only six vibrations which classify as Al + E + T2. These are observed at 606 (Raman, polarized), 363 (Raman), and 465 cm-l (Raman, i.r.), respectively. It is clear that we could treat the Al and T2modes as the radial motions of the phosphorus atoms with respect to the centre of gravity. This discussion would parallel that given in (i) and we calculate v = 500 cm-1 and v’ = 35 cm-l.The interaction between two of these radial stretches now, however, represents a stretching of a P-P bond and it is not surprising that these radial stretches hinder one another. To completea discussion on this basis the Emode would have to be described independently in terms of the tangential motions of the phos- phorus atoms. Alternatively, it is possible to set up this problem in terms of an edge basis, for the six P-P bonds transform as Al + E + T2.The same vibrations may be described in terms of either these P-P stretches or the radial and tangential motions of the vertices both for CH, and for P4. Intuitively, however, we prefer an edge topology for P, and a radial one for CH4.The interaction matrix for the six P-P bond stretches s12-&4 is given in (8).Jotham From standard group-theoretical arguments we find that the symmetry modes are as shown in (9). 1 A1 $1 = -(s12 fs13 + s14 f s23 f s24 fs34)46 1 $2 = 2/12(2s12 -sl3 -s14 -s23 -s24 + 2s34) 1 $3 = T(s13 -s14 -s23 + S24) 1 944 = -p12 -S34) (9) If we use these functions to diagonalize the matrix (6) and equate the solutions to the observed frequencies we find A, v + 4vc + vt = 606 cm-l E Y -2vc + vt = 363 cm-l T2 v -vt = 465 cm-l Solution of these equations gives v = 454, vc = 41, and vt = -11 cm-l. In this case, if the T2translation is neglected (and the T2 functions 944-948 are readily seen to be virtually free of a translational element), a complete description is obtained in terms of three independent parameters v, vc, and vt by the application of symmetry to the topological matrix.As there are three observable frequencies a complete semi-empirical description is obtained for this problem. (iii) Molecular Orbitals of CHI. If the 1s electrons of the carbon atom are neg- lected, the basis set of atomic orbitals for CH4 consists of the 2s and 2p orbitals of the carbon atom, which transform as Al and T2 respectively in Td symmetry. The four 1s ligand group orbitals of the hydrogen atom also transform as Al + T2.l1 This set of four orbitals bears the same symmetric and topological relationships as the four C-H stretching motions of CHI. If we call the four 1s orbitals $144, then the A, and T2 ligand group orbitals are isomorphous to the set (l), and we may parallel the whole of the discussion in (i) by replacing each vij by a Hij.This calculation is, however, of little significance in this case because the off-diagonal Hi1 are virtually zero unless the pairs of 1s orbitals overlap appreciably. As a result of the negligible values of the off-diagonal elements the Al and T2 ligand group orbitals are essentially isoergic. Instead, the analysis of this problem is dominated by the interactions between the 2s and 2p orbitals of the carbon atom and the hydrogen orbitals. As the set of ligand group orbitals Chemistry-a Topologica1 Subject isomorphous to (1) contain equal contributions from each atom, the A, and the three T, bonds will also involve these equal contributions so that the tetrahedral symmetry is maintained in the delocalized description.The valence-bond descrip- tion, involving sp3 hybrid orbitals on the carbon atoms, remains useful and popular precisely because it provides a simple concept which may be directly related to the intuitively perceived topological requirements of this problem and related ones. (iv) Molecular Orbitals 0fP4. Unlike the case of CHI, topological considerations play a great part in the discussion of the molecular orbital diagram of P4,because the overlap of the symmetry-related atomic orbitals in this molecule is not negligible. If we consider each phosphorus atom separately and select as a basis set of atomic orbitals the 3s and 3p orbitals only, we can consider each of these sets to be subject to a perturbing field of CW symmetry due to the other three phosphorus atoms.Defining local z-axes accordingly we see that each 3s and 3pz orbital may be classified as Al and each pair of 3px and 3p, orbitals as E in the CsV symmetry. One spz hybrid lies inside the tetrahedron and the second lies outside. The latter hybrids accommodate the two essentially non-bonding electrons on each phosphorus atom. The other set of four spz hybrid orbitals and the 3px and 3py orbitals are responsible for ‘central’ and ‘edge’ bonding of the phosphorus atoms in the tetrahedron.The set 4 x (1ocalAJtransforms asAl + T2 in Td symmetry, whereas the set 4 x (local E) transforms as E + TI + Tz. Within each of these sets the splitting may be decided entirely on topological grounds, and explicit use of functions derived by symmetry leads to a rapid solution of this part of the problem.The assessment of the interaction between the T, states can only be decided on energetic grounds. This general solution of topological matrices by symmetry arguments forms the basis of Huckel theory in two dimensions and is readily extended to three, We have already solved the central bonding portion twice! The solutions are contained in the matrix (6) if we replace v and v’ by the Hiickel parameters ac and pc (c indicating cmtral and $1-$4 are appropriate combinations of the spz hybrid orbitals isomorphic to the equations (1). a and 16 are regarded as intrinsically negative (i.e.binding) quantities so that the Al orbital at aC + 3Pc is more stable than the three T2 orbitals at ac-pc. (The latter should not be regarded as antibonding, but as less stable than their parent orbitals.) In this case we may reasonably assume the topological basis to be mainly edge bonding, although the localgs andpy orbitals lie largely outside the tetrahedron. Although we may readily derive the repre- sentation of the set of eight tangential orbitals, it is difficult to relate the pairs of orbitals at each corner neatly to the three edges that meet there. We refer the interested reader to the discussion of this peripheral and rr-bonding problem by Schmidtke for its detailed solution.6 He found that the complete orbital energy sequence for the twelvep orbitals is A, < T, < E < Tl < T2.Inclusion of the 3s orbitals stabilizes this A, orbital further but destabilizes the T2 orbitals slightly.The calculation shows very simply that, in addition to the non-bonding Jotham electrons in external A, and T2orbitals, the remaining twelve valence electrons of P4are accommodated in bonding orbitals of Al, E, or T2symmetry. The Al bonding orbital has an entirely central character whereas the E orbitals are peripheral. The T2bonding orbitals have a mixed central and peripheral nature. The particular admixture is determined by the relative magnitudes of the ‘resonance’ integrals, pi. Finally we may remark that the six ‘valence bonds’ of P4transform as the six edges of a tetrahedron, i.e.as Al + E + T2.The repro- duction of six bonding orbitals spanning just these representations reflects, ultimately, the intuitive choice of an edge topology for the important interactions. Kettle has given a very similar discussion of P4in which 12 bonding electrons are found in edge orbitals of types A, + E + T2whereas the remaining eight electrons are found in face-bonding orbitals of type Al + T2?These face orbitals correspond to a mixture of our central bonding orbitals and external non- bonding orbitals. It is important to realize that the topological basis of the prob- lem can be ambiguous for many three-dimensional molecules.* (v) Stretching Vibrations of Deuteriomethanes. A small distortion of a tetrahedral molecule which does not destroy one of the three-fold axes usually leads to CsV symmetry.The topology of the system may well be invariant to this distortion, so that we must explore the consequences of introducing an energetic inequiva- lence into the same topological framework. A particularly interesting example is provided by the vibrational spectrum of CH3D.The reduced mass of a C-D bond pair is almost ,/2 times as large as that of a C-H bond pair [more accurately ~D/PH = ,/(13/7)], and C-D stretching vibrations are observed near 2200 cm-1 whereas the corresponding C-H modes occur at about 3000 cm-l. Nevertheless, the electronic forces in CH,D cannot differ greatly from those in CH4;in particular it seems very likely that the coupling between C-H stretching motions and C-D stretching motions will be very similar to that between two C-H stretching motions.We therefore replace the matrix (4) of the CH4 problem by (10) in which we expect V’H-H M V’H-D but VHZ ,/(13/7)~~. (rl I (r2 I (r3 I (r4 I I PI> VH V’H-H V’H-H V’H-D 1 r2> V‘H-H VH V’H-H V’H-D (10)I r3) V’H-H V’H-H VH V’H-DI r4) V’H-D V’H-D V’H-D VD Linear combinations of r1-~4 of 2A1 + E symmetry are readily constructed from the CsVcharacter table (Table 2) and are given by (11). Table 2 C3vCharacter table c3v E 2C3 3oV Rotations Vectors Vector products A1 1 1 1 2 222 -x2 -y2, x2 + y2 + z2 A2 1 1 -1 Rz E 2 -1 0 (Rz, RY) (x, Y) (XZ, YZ)?(XY, x2 -Y2) 467 Chemistry-a Topological Subject Regrouping the terms of matrix (10) in terms of the functions (11) we obtain the matrix (12).(*l I ($2 I 043 I ($4 II $1) VD 43V’H -D 0 0 I$d J3v’H-D VH + 2V’H-H 0 0 (12)1w 0 0 VH -V’H-H 0 I $4) 0 0 0 VH -V’H-H Applying standard perturbation arguments,* we may immediately write the 3(V’H -D)2 3(lr’H-D)2eigenvalues of (12) as VD --(Ai), VH f 2V’H-H -t--(A1) ,and-VD VH -VD *The ‘repulsion’ of interacting energy levels and the particular form of the zeroth- and first-order perturbation energies may be neatly illustrated by a calculation similar to many of those in this article. Suppose that we have to solve the general two-eigenvalue problem summarized in the determinant A quadratic equation is obtained on expansion of the determinant: (A -E)(B-E) -Ca=O E8 -(A + B) E + (AB -C’)= 0 Then : 2E = (A + B) f.\/ [(A + B)’ -4(AB -C’)] = (A + B) AZ .\/“(A + BIB+ 4ca1 NowifA=B,thenE=A& C but if A z=-B, then we may write: = (A + B) f(A -B) [l + -+2ca .....-1 Whence Note that, since A > B, the perturbations automatically cause a repulsion oflevels.Intermediate cases, where A -B z C, must be solved exactly to give eigenvalues between the above limits. Jotham VH -Y’H-H(E) if VH -VD V’H-D. The expression for the eigenvalues of the type-E functions are identical to those for the T2functions of CH,, but the Al functions are different. The reduction in symmetry leads to more mixing with other motions than in CH, but this is again neglected. Furthermore the eigen- values of (12) incorporate four different parameters and only three frequencies are available experimentally.It is precisely because the topology of the inter- actions of CH, and CH,D vibrations are almost identical, and therefore V’H-H z U’H-D, that we can attempt to solve the problem in the same way as the corresponding CH, problem, The reader will appreciate immediately the greater difficulties in treating the isomorphous problem of CH,CI for which two topo- logical parameters are needed. The relevant frequencies observed for CH,D are 2982 (Al), 3030 (E),and 2205 cm-l (A1),12 which may be compared with the corresponding modes of CH,, namely the Al mode at 2914 cm-I and the T2mode at 3020 cm-l. If we equate the three observed frequencies to VH + ZV’H-H, VH -V‘EI-H, and vD,respectively, we obtain values of 3016,2205, and -16 cm-I for VH, YD, and V’H-H.The cor- responding values of VH and V’H-H of CH, are 2994 and -26 cm-l. If we assume that V’H-DE -16 cm-l also, then the term ~(V’I~-D)~/VEI -VD,which we have neglected, may be seen to have a value only of the order of 1 cm-l. The value of V’H-H for CH3Dobtained in this way appears to be strikingly different from the value obtained for CH,. The difference in the efective V’H-H values reflect the limitations of some of our approximations in which deformation modes are neglected. In particular, there is an important interaction in CH, and CH,X compounds between the Al C-H stretching fundamental and the overtone of one of the asymmetric C-H deformations.In CH,D the latter is found at 1477 cm-l (2 x 1477 = 2954) and therefore perturbs the A1 fundamental to higher energy, whereas in the corresponding case of CH, (2 x 1526 = 3052) the Al fundamental is only slightly perturbed to lower energy by this interaction. The values of rn for CH2D2 and CHDB are 2997 and 2992 cm-l, respectively,12 (cf. 2994 and 3016 cm-l for CH, and CH3D). The approximations which are required increase rapidly with molecular complexity and with any reduction of symmetry. Under these conditions, the number of independent energy parameters required to describe the system rapidly exceeds the number of independent experimental data, and the informa- tion obtained from symmetric and topological arguments is correspondingly diminished.B. Planar Systems. (i) Hiickel Energies of Cyclobutadiene. The D4h character table appropriate to this and later problems is given in Table 3. We associate with each point in the system the Coulomb integral, a. In the topological matrix it is necessary to distinguish between the pairs of points which are cis and trans to one another. We therefore require two resonance integrals and st to distinguish the two cases. The basis set is taken to consist of the four out-of-plane (local py,molecular pz) orbitals of the carbon atoms which are numbered sequentially around the square. The remaining Chemis try-a Topological Subject c3 A n2I NY n h) %Y 470 Jotham orbitals all lie in the plane of the molecule and are regarded as ‘framework‘ and/or non-bonding orbitals.The energy matrix is given in (1 3). The set of four py orbitals transform as Azu + Bzu + Eg.The orthonormal combinations of are given by (14). The reader should note that this set is also isomorphous to (1). With these functions we immediately obtain the eigenvalues of this general topological problem as a + 2Pc + /It (Azu),a -2Pc + Pt (Bzu)and a -fit (twice, Eg).In Hiickel theory it is conventional to set the resonance integrals between non-adjacent atoms equal to zero and, therefore, for the case of cyclo- butadiene, the four n-electrons are allocated to orbitals of the lowest energy to give the configuration (a2u)z(eg)z. This configuration does not confer great stability as in the corresponding case of the hexagonal system.The spectro- scopic transitions Azu c-)EB(Eu)and Eg t)BW (Eu)are both electric-dipole allowed in (x,y)polarization,whereas the AzucsB2, transition (Big) is forbidden. Neither of the allowed transitions would involve a change in electron-repulsion energy so that it is hypothetically possible to determine the quantity P, exactly for this unknown molecule. (ii) Stretching Vibrations of XeF,. This problem is isomorphous to the previous one, but the interaction constant v’t can no longer be neglected. Indeed, as the trans-ligand atoms in such molecules tend to form bonds involving the same orbitals of the central atom, it is very probable that 1 v’t 1 > 1 vtcI.The four stretching modes #1-#4 which are isomorphous to the set (14) transform as Alo, Blg, and E,, respectively. From the observed1’ Raman and i.r. frequencies of XeF, [543 (R, p), 502 (R, dp), 586 cm-l (i.r.)] we calculate the values of the three parameters as v = 554, v’, = 10, and v’t = -32 cm-l. (iii) Other Polygons. In Table 4 we give general solutions of topological matrices for polygons up to the hexagon, distinguishing between the central, peripheral, and the out-of-plane interactions. The local z-axes point directly to the centre l7 K. Nakamoto, ‘Infra-red Spectra of Inorganic and Co-ordination Compounds,’ 2nd Edn., Wiley, London, 1970. Chemistry-a Topological Subject Table 4 General expressions for the eigenvalues of topological matrices for regular polygons System Central interactions (sp z) Peripheral interactions (pz) r Eigenvalue r Eigenvalue Linear dimer, Dcoh cQ+a + /3 cu+a -/3 Equilateral triangle, Al' a + 2P Az) 01' -2/3' D3h E'*aO1-/3 E'* a' + 18' Square, D4h A1g 01 + 2/3c + fitb A2g 01' -2pc' -/3t' Big cx -2/3c + Pt B2g a' + 2/3c' -Pt' Eu* 01 -/% Ez,* 01' + /3t' Pentagon, D5h Al' a + 2P0 + 2Pm A2' 01' -2/30' -2&' [g = +(1/5 + I), El'* 01 + g;So -gPm El'* a' -g'po' + g/3m' g' = *(1/5 -11 E2'* a -gPo + g'Pm E2'* a' + gPo' -g'/3m' Hexagon, D6h AIQ a + 2/30 4-2Pm + /3p A2g 01' -2/30' -2pm' -/3p' E2g* a -/30 -Pm + Pp E,g* a + Po' + Pm' -/3p' B1u 01 -2/30 + 2pm -/3p B2u 01' + 2/30' -2Pm' + Pp' Elu* 01 -k Po -Prn -/3p Elu* a' -Po' + Pm' -k /3p' An asterisk indicates the existence of an additional cross-term (usually small).0 c = cis; t = trans; o = ortho; m = meta; p = para. of the polygon. Note the similarity of the algebraic expressions of related eigen- values for each set of interactions; the absolute values of the at and the will, however, differ for each case. C. Molecular Orbitals of an Octahedral Transition-metal Complex.-We take as our final example the energy levels of an octahedral transition-metal complex. It is well known that for this system the crystal-field theory predicts a splitting of the ten-fold degenerate 2D state (i.e. dl or dBconfigurations) into 2Egand 2T2g states separated by the frequently parameterized quantity 1ODq. The molecular orbital approach is even more valuable since it provides scope for consideration of both a-and n-bonding effects on these two states and also information on a number of important excited states.loThe pictorial description of the crystal field, which invokes orbital destabilization through repulsive interactions with ligand electrons, is clearly topological in essence and the same is true of the qualitative molecular orbital description. It is also interesting that Schmidtke, perhaps recognizing that the a-and n-bonding interactions are topologically equivalent to strong ligand-ligand repulsion interactions along the edges of the octahedron combined with metal-ligand a-bondingasaperturba- tion, has shown that a similar sequence of energy levels may be obtained by either method.4 Concluding Remarks Before we tackle any problem of high symmetry it is always advantageous and l8 H. H. Schmidtke, J. Chem. Phys., 1968, 48,970. Jotham Out-ofiplane interactions (py) Transformation properties of orbitals at centroid S P d cg+ cu+ +nu Cg++ng+ Ag Al' E' + A2" Al' + E' + E" Al' El' + Az" AX' + EZ' + E," sometimes necessary to decide the basis set for the problem on energetic grounds and to select the topology of the important interactions within this basis set. The topological matrix is then divided into two parts which depend respectively on the symmetry of the problem and the magnitude of the energy terms. The first part of the problem may be solved exactly, but the second part becomes increasingly empirical as the complexity of the problem increases and/or the syinmetry decreases.In some cases the interactions between equivalent members of the basis set are not the most important, but in many others a system may be accurately described by the solutions of a simple topological matrix. Such a case is the familiar n-bonding problem of aromatic hydrocarbons. At the other extreme we note the low significance of many ligand-ligand interactions and of the vibronic coupling in many cases where a static distortion is predicted by the Jahn-Teller theorem. We have given examples of the profound influence of relative energy on the whole basis of a problem and its manner of solution. We may comment addi- tionally that topological considerations are more readily introduced into some problems, such as spin-coupling, than into others such as the 7r-interactions of the tetrahedron. Topological considerations enter neatly into a valence-bond model, whereas the application of symmetry in this case is seriously complicated by frequent non-orthogonality of the basis set of orbitals.Conversely, in a molecular orbital discussion the symmetry is readily introduced but often this is at the cost of a non-intuitive topology in addition to the problems associated with the neglect of configuration interaction in polyelectronic systems. Neither the symmetry nor the topology of a problem is usually introduced explicitly into a self-consistent-field calculation, as the former is not readily used to shorten the process time and the latter is not required as a conceptual aid! Conversely, Chemistry-a Topological Subject the general discussion of any problem is facilitated by distinguishing those parts of the problem which involve the symmetry, the topology, and the energy terms appropriate to the problem in a fully complementary manner.The general solutions of common topological matrices in high-symmetry systems may be used in this way to treat many diverse problems, once the energetic considerations have been clarified.

ISSN:0306-0012    

DOI:10.1039/CS9730200457

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

The Brsnsted Relation -Recent Developments By A. J. Kresge DEPARTMENT OF CHEMISTRY, ILLINOIS INSTITUTE OF TECHNOLOGY, CHICAGO, ILLINOIS 60616, U.S.A. 1 Introduction Fifty years ago, Brarnsted and Pedersen reported1 that a series of rate constants, kg, which they had measured for the base-catalysed decomposition of nitramide, bore a simple relationship to the basicity constants, KB,of the catalysts: kB = (6.2 X (KB)''~~ (1) They then suggested that similar correlations might be found for other proton transfer reactions. This hypothesis has been amply confirmed. In fact, it is now generally accepted that all reactions showing general acid-base catalysis will give relationships of the Brarnsted type :kA =GA(&)" for general acid catalysis and kB = GB(KB)B= GB(KA)-~for general base catalysis.A large number of such cor- relations have been made, and the Brarnsted relation has been used not only to summarize data and make predictions, but also for such diverse purposes as de- tecting changes in reaction mechanism, identifying nucleophile catalysis, and deducing transition-state structure. It should also be noted that the Brarnsted relation is a linear free energy relationship, and that it antedates the better known Hammett equation by more than a decade; the already extensive but still rapidly growing area of chemistry which deals with free energy relationships can therefore be traced back directly to the Brarnsted relation.2 Although the Brarnsted relation is a linear correlation in the free energy sense, there are good reasons for expecting this linearity to hold only over a limited range.Brarnsted and Pedersen recognized this themselves, and they predicted in their original paper1 that log kA would in the general situation be a curved rather than a linear function of ~KA.For many years, however, virtually all examples of the Brarnsted relation proved to be accurately linear, even over extended ranges of ~KA. It therefore came as somewhat of a surprise when sharply curved relationships were produced recently by data made available through newly developed techniques for measuring the rates of very fast reaction^.^ These sharply curved Brarnsted plots use oxygen and nitrogen acids and bases as both the proton donor and the proton acceptor, whereas linear relations deal for the most part with proton transfer to or from carbon.It is nevertheless probable that the identity of the atoms between which the proton is being transferred is only J. N. Brernsted and K. Pedersen, Z. phys. Chem., 1924,108, 185. L. P. Hammett, in Foreword to 'Advances in Linear Free Energy Relationships', ed. B. N. Chapman and J. Shorter, Plenum Press, London, 1972. M. Eigen, Angew. Chem. Znternat. Edn., 1964, 3, 1. The Brmsted Relation-Recent Developments of secondary importance in determining whether a Brarnsted relation will be linear or curved, and that the primary factor is the reactivity of the system; Marcus rate theory4 is especially useful in understanding this connection between reactivity and curvature.Another recent development is the discovery of Brarnsted exponents less than zero and greater than one. This is a phenomenon which Brarnsted and Pedersen did not anticipate: they, as well as a number of subsequent authorities, expected uand always to lie in the range zero to one. Marcus theory offers an explana- tion for anomalous Brarnsted exponents such as these, and they may be under- stood also in terms of substituent effects, particularly those produced by interactions in the transition state. Transition-state interactions have also been shown recently to be the origin of the dispersion according to charge type generally shown by the catalysts in a Brransted correlation. Brarnsted and Pedersen’s own nitramide decomposition reaction provides what is probably the best known example of this phenomenon, but newly discovered cases serve to reinforce the explanation.The regular devia- tions from Brarnsted relations commonly shown by the hydronium and hydroxide ions have also received recent attention and a new explanation. These three phenomena, (i) curvature, (ii) anomalous exponents, and (iii) systematic deviations, together with their implications on certain uses to which Brarnsted relations are often put, form the chief concern of this review. 2 Curvature As Brernsted and Pedersen pointed out,l the rate of proton transfer from an acid to a base [equation (2)] cannot continue to increase in accordance with kA = GA(KHA)~indefinitely. As the acid is made stronger and stronger, the rate will kA HA+ B+ A+ HB kB become faster and faster until eventually reaction occurs at every encounter between the acid and base molecules. Once this limit is reached, further increases in acid strength will have no effect :kA will then be constant and independent of KHA,and awill therefore be equal to zero.In this situation, since kA/kB = K and kA is now constant, kB will change in inverse proportion to K;this, because K = KHA/KHBand only KHAis being varied, requires kB to change as (KHA)-~, i.e. it makes /3 equal to unity. If, on the other hand, the acid strength of HA is continually decreased, kB will increase until the rate of reaction in this direction reaches its encounter-controlled limit.No further increase in rate will then be possible, and 18will be equal to zero; under these conditions, uwill have to be unity. Arguments such as these led Brarnsted and Pedersen to conclude that both u and would have to change regularly between the limits zero and one. Neither exponent could be constant over any extended range of catalyst strength, and R.A. Marcus, J. Ph-vs. Chem., 1968,12, 891. relationships such as that of equation (1) could therefore be valid only over limited intervals. They summarized these ideas in the form of a diagram (Figure l), in which the solid lines represent a simple acid-base reaction [such as that of equation (2) with HA constant and B changing; the abscissa is then log KFIB and kD = kB],and the broken line refers to some other, more complex, base- catalysed process.For many years after Brarnsted and Pedersen’s prediction there was little evidence for curvature in Brarnsted relations. This was due in part to the fact that a and fi often change only very slowly, and large differences in catalyst strength are therefore needed to produce detectable curvature; this requires considerable variation in catalyst structure, which itself is liable to produce deviations from the correlation, Some indication of curvature, however, was obtained from studies in which the substrate as well as the catalyst was varied. Bell and Lidwells~6~ found that Brarnsted relations for the base-catalysed halo- genation of a series of ketones, though each accurately linear, gave values of fi which decreased systematically with increasing substrate reactivity, and a plot of the ~KA’s of the substrates vs.log k for the reactions catalysed by a common base (water) was decidedly Striking confirmation of Brarnsted and Pedersen’s prediction came about ten years ago when techniques for measuring rates of very fast reactions became lnk Figure 1 Brmsted plot curvature predicted by Bronsted and Pedersen in 1924 (Reproduced by permission from Z. phys. Chem., 1924,108, 185) R. B. Bell and 0.M. Lidwell, Proc. Roy. SOC.,1940, A176,88. R. P. Bell, ‘The Proton in Chemistry’, Cornell University Press, Ithaca, New York, 1959 (a)p. 171; (6) p. 163; (c) p. 172; (d)p. 173. 477 6 The Brmsted Relation-Recent Developments available.Eigen pioneered the application of these methods to acid-base reactions in solution; his data for a typical system, proton transfer to ammonia from a series of oxygen acids,s are shown in Figure 2. In this diagram, X represents ammonia and HY the oxygen acids; the abscissa is equal to ~KHX ~KHY-and the ordinate, log k. It may be seen that the logarithm of the protonation rate constant changes from being almost directly proportional to log KHY(0: = 1) for the weakest acids to being nearly independent of log KHY(a= 0) for the strongest acids. The reverse reaction gives corresponding behaviour, with changing from 0 to 1; and the limiting rates agree well with those expected for encoun ter-controlled processes. The sharp curvature shown by these Brsnsted plots stands in marked contrast to the behaviour commonly found in other systems, where linear relations often extend over pKa ranges comparable with, and sometimes even greater than, that of Figure 2.A possible cause of this striking difference is suggested by the fact that the curved relationships almost invariably use oxygen and nitrogen, and in a few cases sulphur, acids and bases as both the substrates and the catalysts. These substances are often inorganic, and they are in general species to which one would normally attach the label ‘acid’ or ‘base’. Eigen, in fact, classifies them as ‘normal’ acids and bases, and he calls the curved Brsnsted plots which they generate, ‘normal’ behavi~ur.~ The systems giving linear plots, on the other hand, almost always involve proton transfer to or from carbon, i.e.although the catalysts may be oxygen and nitrogen acids or bases, the substrates are either carbon acids, such as nitro or carbonyl compounds, or carbon bases, such as I \ -6 -5 -4 -3 -2 -1 0 1 2 3 4 PKHX -PKHy Figure 2 Bronstedplot curvature confirmed by Eigen in I964 (Reproduced by permission from Angew. Chem. Internat. Edn., 1964, 3, 1) 478 Kresge vinyl ethers or alkoxybenzenes. These are substances to which one does not normally attribute acidic or basic properties, and they are therefore often called 'pseudo' acids and bases. This suggests that the identity of the atoms betweeen which the proton is being transferred is important in determining whether a Brsnsted relation will be linear or curved, with linearity associated with carbon atoms and curvature a property of oxygen and nitrogen.Quite recently, however, sharply curved Bransted relations were found for certain carbon acids. This development is due mainly to Long, who discovered that rates of proton transfer from cyano- carbons, such as malononitrile, to oxygen and nitrogen bases give Brsnsted relations similar to that of Figure 2, i.e. they have linear portions of slope 1 and 0 which are connected by curved transition regions.' Subsequent work shows that chloroform* and phenylacetylene@ behave similarily. These studies therefore negate the idea that the identity of the atoms between which the proton is transferred determines whether a Brarnsted plot will be linear or curved, but they also point to another property, reactivity, which very prob- ably is the controlling factor.Proton transfer between the oxygen and nitrogen acid-base pairs which give sharply curved Bransted plots is very fast in the exo- thermic direction, and the same is true of the cyanocarbons, chloroform, and phenylacetylene. Proton transfer involving the carbon acids and bases which give linear Brsnsted plots, on the other hand, is invariably slow, even when exo- thermic. This suggests, then, that rapid proton transfers, whether they involve carbon or not, will give curved Bransted relations, and slow proton transfers, again irrespective of the identity of the atoms involved, will give linear relation- ships.Most proton transfers to or from carbon are of course slow, and this accounts for the preponderance of linear relations involving carbon. Moreover, since very few fast reaction rates were measured before the beginning of the past decade, all but the most recent history of the Bransted relation is dominated by linear relationships. Support for a connection between reactivity and Brmted plot curvature comes from Marcus the~ry.~ This theory relates the free energy of activation for a proton transfer process, dG*, to the standard free energy of reaction, AGO, through the free energy of activation whendC" is zero, AGO* [equation(3)]. The latter is the barrier to reaction when the process is free of any exothermic drive or endothermic impediment; it is therefore a good measure of the intrinsic reactivity of the system, and Marcus in fact calls it the 'intrinsic barrier'. Since the Brsnsted exponent a may be identified with the derivative ddG*/ddGo '(a) E.A. Walters and F. A. Long, J. Amer. Chem. SOC.,1969, 91, 3733; (b) F. Hibbert, F. A. Long, and E. A. Walters, ibid., 1971, 93, 2329; (c) F. Hibbert and F. A. Long,ibid., 1972,94,2647. Z. Margolin and F. A. Long,J. Amer. Chem. SOC.,1972,94,5108; 1973,95,2757. * A. J, Kresge and A. C.Lin,J.C.S. Chem. Comm., 1973,761. The Brmsted Relation-Recent Developments [equation(4)],* and the curvature of a Bransted plot, with the rate of change of awith dG" or second derivative of dG* with respect to dG" [equation (5)], Marcus theory immediately provides a connection between curvature and reactivity. The resulting relationship predicts that intrinsically fast reactions (small AGO*)will show sharp curvature (large dar/ddG"), and intrinsically slow reactions (large AGO*)will show little curvature (small da/ddG").a = adG*/3dG0 = (1 + dG0/4dG0*)/2 (4) da/ddG" = l/8dG0* (5) It is illuminating to evaluate equation (5) for some representative intrinsic barriers. When AGO*= 1 kcal mol-l, achanges from 0 to 1 over the interval dG" = -4 to +4 kcal mol-l, which corresponds to a difference in catalytic strength of 5 pK units at 25 "C. With a barrier as small as this, the rate constant for a thermoneutral process at 25 "C is 1 x JOla 1 mol-l s-l; this is above the encounter-controlled limit, and such behaviour might therefore be typical of proton transfer between an oxygen or nitrogen acid-base pair, which is known to be a diffusion-controlled reaction.When AGO*= 10 kcal mol-l, on the other hand, a difference of 55 pK units is required to change cc from 0 to 1, and when AGO* = 20 kcal mol-l, the range needed becomes 110 pK units. In the latter case, a change in catalyst strength of 5 pK units, a difference typically used in practice, would change aby slightly less than 0.05; a change as small as this would be difficult to detect experimentally and the Bransted plot would therefore appear to be linear. Marcus rate theory was first devised for outer-sphere electron transfer reactions in solution, where it has received considerable empirical support on a quantitative level.10 Its extension to proton transfer reactions in solution is more dubious, but the fundamental ideas of the theory seem at least to be qualitatively correct.For example, Marcus himself has applied equation (3) to literature data for a number of reaction series, most of which had previously been assigned linear Bransted relations.ll It is difficult to say in most cases whether curved relationshipsfit the data any better, but the intrinsic barriers which result from * Although the Brnrnsted relation correlates specific rates of proton transfer between a substrate and a series of catalysts with the ionization constants of the catalysts, and it therefore relates rates and equilibria of two different processes, this is equivalent to cor- relating rates and equilibria of a single, i.e.the catalysed, reaction. This follows from the fact that the equilibrium constant for the substrate protonation reaction, HA + S 3 A+ HS, is equal to the acidity constant of the catalyst divided by the acidity constant of the protonated substrate: K = (A) (HS)/(HA) (S) = KHA/KHS.Since the substrate, and therefore KHS,remains constant along the reaction series, K and KHAmust vary in exactly the same way, and one may be substituted for the other in the correlation: HA = G(KHA~ = G(m.In terms of free energies, then, the Brnrnsted relation correlates the free energy of activation of the proton transfer reaction, AGP, with its standard free energy of reaction, AGO, and the Brnrnsted exponent measures the rate of change of the former with respect to the latter: CY -SdG*/6dG0, which, in the limit, is ddG*/ddG".10 See, for example, R. E. Weston, jun., and H. A. Schwartz, 'Chemical Kinetics', Prentice Hall, New York, 1972, pp. 205-213. A. 0. Cohen and R A. Marcus, J. Phys. Chem., 1968,72,4249. Kresge this analysis do increase as factors known to slow proton transfer reactions, such as heavy atom reorganization and charge delocalization, come into prominence; the expected relationship between curvature and intrinsic reactivity is thus readily apparent. There is also some independent theoretical support for Marcus theory. A relationship of the form of equation (3) can be obtained from a solvent polariza- tion model for proton tramfer,l2 and from the Leffler principle13 relating transi- tion state to reactant and product free energies.14 Equation (3) may also be derived from the expression for an Eckart reaction barrier,16 and it follows as well from a simulation of the proton transfer process in terms of two intersecting parabolic potential energy functions.16 The latter model, however, as well as one based upon the BEBO method" of predicting reaction barriers leads to important quantitative differences from equation (2) (see be lo^).^^^^* Some difficulties also emerge from certain more detailed quantitative applica- tions of Marcus theory to experimental data.These analyses use the extended description of proton transfer as a three-step process, upon which Marcus based his theory'l but which he did not himself use in his own analysis of experimental data.ll This formulation, equation (a,treats the encounter of reactants and the w' AGO AH + B 7AH*BGA*HB WD A + HB (6)encounter proton separationtransfer separation of products as distinct steps, separate from the actual proton transfer itself.Equation (3) is then taken to apply only to proton transfer within the encounter complex and not to the diffusion steps preceding and following it. This makes the observed free energy of activation, (dG*),a,, equal to dG* calculated by equation (3) plus wr, the work expended in forming the encounter complex, or free energy of formation of AH.B from separated AH and B.The observed free energy of reaction, (dGo)ObS,is likewise equal to AGO, the free energy change for reaction within the encounter complex, plus wr minus wp, the latter being the free energy of formation of the product encounter complex from separated products. Recasting equation (3) in terms of observed quantities then leads to: (dG*)obs =W' + (1 4-[(dGo)obs -Wr + ~Pl/4AGo*)~dGo* (7) Experimental data are usually fitted to equation (7)on the assumption that wry WP, and AGO*remain constant along a reaction series. This makes (dG*)obs a quadratic function of (dGo)obs, whose three coefficients are themselves la E. D. German, R. R. Dogonadze, A. M. Kuznetsov, V. G. Levich, and Y. I. Kharkats, J. Res. Inst. Catalysis, 1971, 99, 115.l3 J. E. Leffler, Science, 1953,117,340. l4 J. R. Murdoch, J. Amer. Chem. SOC.,1972,94,4410. 16 C. Eckart, Phys. Rev., 1930,35, 1303. ~IIG. W. Koeppl and A. J. Kresge, J.C.S.Chem. Comm., 1973,371. l7 H. S. Johnston, 'Gas Phase Reaction Rate Theory', Ronald Press, New York, 1966. A. J. Kresgc and G. W. Koeppl, unpublished work. The Bronsted Relation-Recent Developments functions of wr,wp, and AGO*.The latter three quantities may therefore be evaluated from these coefficients, determined, for example, by a least-squares fit of (dG*)obe to (dGo)obs. Quite often, however, values Of (dGo)obs are not avail- able, for these require knowledge of the PKA of the protonated substrate, pKs~+, as well as the PICA'S of the catalysts. In such cases, AGO* and wr may still be obtained from a fit of (dG*),bs to a quadratic expression in AGOHA, the free energy of ionization of the catalyst, or from its equivalent, a fit of log kobe to catalyst PKA.The coefficient of the squared term in the relationship between these pairs of variables is a function of AGO* alone, and the constant term con- tains only AGO* and wr;both AGO* and wr may therefore be evaluated. The coefficient of the linear term, however, is a function of ~KsH+as well as wp, AGO*,and wr; and, if pKs~+ is unknown, wp cannot be determined. This kind of analysis was first applied by Kreevoy and Konasewich to the hydrolysis of diazoacetate ion catalysed by a series of phenols and carboxylic acids,lQ and Kreevoy and Oh have since provided additional data for the same reaction using tertiary ammonium ion catalysts20 (Table 1, Reactions 1 and 2).Proton transfer to other diazo-compounds has been investigated by AIberyzl (Table 1, Reactions 4 and 5), and the protonation of a series of aromatic sub- stances has been studied in the author's laboratory2a (Table 1, Reaction 3). Kreevoy20 has also analysed literature data23 for the enolization of acetylacetone catalysed by the anions of a group of oxygen acids, chiefly phenols and carboxy- lic acids but including also glucose and cacodylic acid (Table 1, Reaction 6), and Albery21has done the same for Bell's data on the halogenation of various ketones catalysed by a common base (the anion of a hypothetical carboxylic acid with p& = 4.00)6c (Table 1, Reaction 7).In addition to these examples already appearing in the literature, the author has carried out an analysis of the very extensive set of data due to Bell and Higginson for the acid-catalysed dehydration of acetaldehyde hydratea* (Table 1, Reaction 8). Most of the Brarnsted relations summarized in Table 1 are not very strongly curved, and in some cases the data fit a straight line as well as they do a quadratic expression. The uncertainty in the coefficient of the squared term, and conse- quently in AGO*, is therefore high. For example, standard statistical methods give AGO* = 9.8 +, 2.0 kcal mol-l for Reaction 3 and AGO* = 5.4 +_ 1.7 kcal mol-1 for Reaction 8; the standard deviations in the work terms, wr and wp, are comparable.These uncertainties, nevertheless, are not large enough to obscure the single most striking feature of these results, namely, that the intrinsic barriers are on the whole very small and the work terms are consequently large. These work terms average 11 kcal mol-l, which is far too much to represent simple encounter of the reactants: the energy needed to localize a catalyst mole- M. M. Kreevoy and D. E. Konasewich, Adv. Chern. Phys., 1971,21,241. M. M. Kreevoy and S.-W. Oh, J. Amer. Chem. SOC.,1973, 95, 4805. W. J. Albery, A. N. Campbell-Crawford, and J. S. Curran, J. C.S. Perkin ZI, 1972, 2206. A. J. Ksesge, S. G. Mylonakis, Y.Sato, and V. P. Vitullo, J. Amer. Chern. SOC.,1971,93, 6181. 23 M.L. Ahrens, M. Eigen, W. Kruse, and G. Maass, Chem. Ber., 1970,74,380. R. P. Bell and W. C. E. Higginson, Proc. Roy. Soc., 1949, A197, 141. Table 1 Resultsa of analysis of some proton tramfer reactions according to Marcus theoryb Reaction AG,* wr WP Reference 1. RC02H or ArOH + CHN,CO,---+RC0,-or ArO-+ CH2+N,C0,-5 8 19 2. RSNH+ + CHN2CO,-+R,N + CH,+N,CO,-1 14 -20 3. H30+ + HAr-H20 + H2Ar+ 10 10 8 22+ 4. RCOaH + CH3CN2COCHg +RCOa-+ CH,CHN&OCH, 2 16 -21 4-5. RC02H + CH3CN2C02C2H,-RC0,-+ CH3CHN2C02C2H, 2 14 -21 6. Oxygen anions + CH,COCH,COCH, -Oxygen acids + CH,COCHCOCH,-3 11 9 20,23 7. RC0,-(~KA= 4) + ketones -RC02H + enolate anions 3 14 7 6c, 21 8. RC02H or ArOH + CH,CH(OH), -jRC0,-or ArO-+ CH,CHO + H20 5 13 -24 =Resultsin kcal mol-I.bBased upon the definition (dG=k)obs= -RT In (kh/kT)where k is a bimolecular rate constant expressed in units of 1 moP s-'. The Brmtsted Relation-Recent Developments cule next to the substrate in aqueous solution is only about 2.5 kcal mol-1 (RT In 55.5). It has been suggested, therefore, that work terms such as these should also include whatever energy is needed to orient the reactants properly within the encounter complex so that proton transfer may take place, i.e. to convert the encounter complex to a reaction complex.1Q~20 Reorganization of the solvent, which is known to contribute significantly to barriers for reactions in ~olution,~6 will be an important part of this orientation. In particular, for reactions between a carbon substrate and an oxygen or nitrogen catalyst in aqueous solution, where it is likely that proton transfer takes place directly with no solvent intervening between catalyst and substrate,as desolvation of the reactants will have to occur.Since the oxygen or nitrogen acid or base serving as the catalyst will very probably be hydrogen-bonded to the solvent, and since this hydrogen bond will not be replaced by another between the catalyst and carbon substrate when these two come into juxtaposition, desolvation will take place at the expense of, among other things ,hydrogen- bond format ion. The strength of the hydrogen bond between a typical acid catalyst and solvent water has been estimated at about 6 kcal mol-l.ls This,when added to the energy of reactant localization, gives a work term consistent with the smaller values of Table 1, and the difference between that and the larger results could easily be made up by further reorganizational effects. Reaction 8 presents an apparent difficulty for this explanation, for its substrate is an oxygen base which can hydrogen bond to the acidic catalyst. Desolvation here should thus be less expensive energetically, but the work term is not reduced accordingly.There is evidence, however, that this reaction OCCUTS through a cyclic transition state which immobilizes a water molecule in addition to the catalyst and the ~ubsfrate.~' Formation of such a reaction complex should be more difficult than usual, and it could well cost enough additional energy to offset that gained back through substrate-catalyst hydrogen-bond formation.A more serious objection can be raised on the basis of whether or not it is correct to separate reagent positioning and solvent reorganization from proton transfer in this way, i.e. whether orientation and proton transfer do in fact occur in separate reaction steps.28 A similar separation is in all probability valid for electron transfer, for which Marcus theory was first derived and where it appears to work well, for here the small mass of the electron ensures that electronic motion will be essentially uncoupled from whatever atomic rearrangements must take place. The mass of the proton, however, is much greater than that of the electron, and a similar kind of Born-Oppenheimer separation for proton transfer Is C.D. Ritchie, in 'Solute-Solvent Interactions', ed. J. F. Coetzee and C. D. Ritchie, Marcel Dekker, New York, 1969, Chap. 4. p6 D. M. Goodall and F. A. Long,J. Amer. Chem. Soc., 1968,90,238; M. M. Kreevoy and J. M. Williams, ibid., p. 6809. R. P. Bell, J. P. Millington, and J. M. Pink, Proc. Roy. SOC.,1968, A303, 1 ;H. Dahn and J.-D. Aubort,Helv. Chem. Acta, 1968,51, 1348; R. P. Bell and J. E. Critchlow, Proc. Roy. SOC.,1971, A32!5,35; R. P. Bell and P. E. Sorensen, J. C. S. Perkin IZ, 1972,1740. R. P. Bell, personal communication. #resge might therefore be inappropriate.This argument maybe however, by the observation that hydrogenic and heavy-atom motion remain largely un-coupled in molecular vibrations.Some information bearing on the question of whether proton transfer and hydrogen-bond reorganization occur in a stepwise or concerted fashion comes from studies of the reaction of hydroxide ion with internally hydrogen-bonded acids, such as substituted hydrogen malonate ions. The strong internal hydrogen bonds in these systems are destroyed in these reactions, and the rates in the exo- thermic direction are consequently several orders of magnitude short of encounter-controlled, even though the proton is transferred between oxygen atoms. This retardation could operate through a stepwise mechanism, in which the internally hydrogen-bonded acid is first converted into an externally bound species with the acidic proton hydrogen-bonded to a solvent molecule; the hydroxide ion would then react with this low concentration intermediate in an encounter-controlled second step, much as it does with an ordinary carboxylic acid.Alternatively, hydrogen-bond breaking and proton transfer might occur simultaneously via a single, concerted transition state, such as that shown in (1); here the breaking and forming bonds are not collinear and the geometry is therefore not optimum for proton transfer. Internally hydrogen-bonded systems such as these have been examined in several laboratories, but the question of mechanism has been attacked most directly by Eyring. On the basis of medium effects,30 isotope effects,31 the effect of changing the atoms involved in the hydrogen bond,3a and the non-zero slope of a Brsnsted correlation based upon a group of different acids,38 Eyring con- cluded that the concerted mechanism was operating. Very recently, however, the question was reopened by Fuen~,~~ who pointed out that a non-zero Brsnsted I0 M.M. Kreevoy, personal communication. ao R. P. Jensen, E. M. Eyring, and W. M. Walsh, J. Phys. Chem., 1966,70,2264. J. L. Haslam, E. M. Eyring, W. W. Epstein, R. P. Jensen, and C. W. Jaget, J. Amer. Chem. SOC.,1965,87,4247; E. M. Eyring and J. L. Haslam, J. Phys. Chem., 1966,70,293. 3a J. L. Haslarn and E. M. Eyring, J. Phys. Chem., 1967, 71, 4470. 33 M. H. Miles, E. M. Eyring, W. W. Epstein, and M. T. Anderson, J. Phys. Chem., 1966, 70, 3490. a4 T. Fueno, 0.Kajimoto, Y.Nishigaki, and T. Yoshioka, J. C. S. Perkin ZI, 1973, 738. The Bronsted Relation- Recent Developments slope is really not inconsistent with the two-step mechanism, inasmuch as differences in strength between internal and external hydrogen bonds along the series of acids could produce regular changes in the first step and thus obscure the zero slope expected for the second. Eyring's evidence from isotope effects, moreover, is based upon a rather approximate separation of primary from secondary effects. Some additional information on the timing of proton transfer and solvent reorganization comes from a comparison of Brsnsted plots for the reaction of a group of hydrocarbon acids with anionic oxygen bases in methanol and in DMSO These reactions are faster in DMSO than in methanol, as expected on the basis of the well-known greater hydrogen-bond-donating ability of the latter solvent and its consequent stabilization of anions.But the data for DMSO also give a more curved Brsnsted plot than do those obtained in methanol, which implies that the intrinsic barrier to proton transfer in these systems is lower in DMSO than in methanol. This suggests that there is a solvent effect on the proton transfer process, and that in turn requires solvent reorganiza- tion and proton transfer to occur simultaneously in a single step. It is possible, however, to interpret this difference in behaviour between the two solvents in another way. The argument is based upon an interesting tendency, significant in its own right, which the diffusive steps of the three-stage mechanism for proton transfer [equation (6)]have of exaggerating Bransted plot curvature.This effect was demonstrated by Murdoch,14 who carried out detailed calcula- tions of overall specific rates using typical diffusion rate constants in combination with specific rates for the proton-transfer step governed by Marcus theory, i.e. ty equation (3). The curvature of Brsnsted plots in the regions near the limiting, i.e. zero and one, values of a,based upon such 'observed' rates was always greater than that provided by the proton transfer step alone, and these regions of exaggerated curvature pushed in toward the centre of the plot as the value of AGO*dropped. The effect may be traced to the fact that, unless AGO* is very large, diffusion and not proton transfer is rate-limiting at the ends of the range of dG" needed to change a from zero to one.Murdoch estimates that, even with AGO* as large as 10 kcal mol-l, a based upon 'observed* rate constants is a reasonably accurate reflection of a for the proton-transfer step only over the middle third of the range zero to one. It is interesting in this connection that, in systems with intrinsic barriers as small as some of the lower values of Table 1 and a free energy of activation for diffusion of 2-3 kcal mol-l, proton transfer can never be fully rate-determining; it can, at best, be only partly rate-controlling, and that only over a rather narrow interval a few kcal mol-l to either side of AGO = 0.Exaggerated curvature of this kind depends not only on the magnitude of the intrinsic barrier but also upon the size of the barrier to diffusion, and making diffusion more difficult has the same effect as lowering the intrinsic barrier. Since DMSO is a more viscous solvent than methanol,36 it is possible that the greater 36 C. D. Ritchie and R. E. Uschold, J. Amer. Chem. Soc., 1968,90, 3415. A. J. Parker, Chem. Rev., 1969,69, 1. Kresge Brransted plot curvature seen in DMSO solution for the reactions described above is simply a viscosity effect and is not caused by a difference in intrinsic barrier. The timing of changes in solvent organization in proton transfer reactions has also been examined from a theoretical point of view recently, with the interesting conclusion that different kinds of behaviour should be observed for proton transfer between bases of different size.s7 When the bases involved are small, e.g.HzO, changes in solvation should be coupled with proton transfer, and the mechanism should be synchronous. When the bases are large, on the other hand, a stepwise mechanism should be favoured, in which the solvent first reorganizes to a configuration appropriate for the transition state, the proton transfers, and the solvent then relaxes to its product configuration. A general solvation rule proposed by Swain and Schowen several years also has some bearing on this matter. This rule states that the proton will always be located in an entirely stable potential energy well at the transition state of a process in which proton transfer between electronegative atoms and heavy atom reorganization both occur; protonic motion here will therefore be uncoupled with heavy-atom movement.The situation, however, now appears to be more complex than was at first anti~ipated,~~ and some exceptions to this rule seem to have been uncovered.40 Some additional information which bears upon the results listed in Table 1 comes from several theoretical studies of the proton transfer process itself. One of thesels is especially interesting in that it develops a familiar model for proton transfer and gives behaviour which reduces to Marcus theory under certain conditions. This model is shown schematically in Figure 3.It takes the reactant and product acids of the proton transfer process, AH + B 3A + HB, to be simple harmonic oscillators with intersecting potential energy functions ; the point of intersection of the two parabolae is the energy of activation of the system, Ea, and their vertical displacement is its energy of reaction, dE. The model leads to an expression for Ea in terms ofdEwhich contains three additional parameters: AH and ~BH,the harmonic force constants of the A-H and B-H bonds, and d, the horizontal distance between the bottoms of the two wells. When km and ~BHare equal and constant, and when d is also constant, the relationship between Ea and dE assumes the simple quadratic form of equation (3), and the model conforms to Marcus theory.* This behaviour is conveniently summarized by the linear relationship between a and dE shown as curve A in Figure 4 [cf.equation (4)]; use of the reduced variable dE/Ea,o gives a single relationship for all values of the intrinsic barrier, Ea,o. It is doubtful, however, whether the conditions which allow this model to * Simple Marcus theory was in fact derived assuming intersecting parabolic potential-energy functions of constant curvature; the Reviewer thanks Professor R. P. Bell-for bringing this to his attention. 37 J. L. Kurz and L. C. Kurz, J. Amer. Chem. SOC.,1972,94,4451. 38 C. G. Swain, D. A. Kuhn, and R. L. Schowen, J. Amer. Chem. SOC.,1965,87,1553. 3s R. L. Schowen, Progr. Phys. Org. Chem., 1972,9,309.40 R. L. Schowen, H. Jayaraman, L. Kershner, and G. W. Zuorick, J. Amer. Chem. SOC., 1966,88,4008. The Brmsted Relation-Recent Developments E Figure 3 Intersecting oscillators model for proton transfer reduce to Marcus theory behaviour will be met by real systems. Force constants are a measure of bond strength, and km will therefore be equal to kBH only when the strengths of the A-H and B-H bonds are the same, i.e. at dE = 0. Inthe more general situation, as the A-H and B-H bond strengths are varied to produce changes in dE, AH and ~BHwill change also. Similarly, the parameter d, which can be identified with the distance the proton must travel when it moves from donor to acceptor within the activated complex, can also be expected to change with dE.In particular, d is likely to be greater for the relatively loose activated complexes of strongly exothermic and strongly endothermic reactions than for the tight activated complexes of thermoneutral systems; this follows, for example, from the Pauling relation between bond length and bond ordeF L.Pading, J. Amer. Chem. SOC.,1947,69, 542. I .o 0.8 0.6 a 0.4 0.2 0 f I I I I I I -6 -4 -2 0 2 4 6 AE/&,o Figure 4 Relationship between aanddE: curve A, simple Marcus theory; curve B, intersecting oscillators model with both EAIIand EBHvaried; curve C, intersecting oscillators model with Em varied and EBHheld constant upon which the highly successful BEBO method of predicting reaction barriers is based. When these features are incorporated into the model, the simple linear dependence of a upon dE gives way to more complex sigmoid relationships, such as that shown by line B of Figure 4.The central regions of these curves, however, are xiearly linear over appreciable ranges, and data conforming to this model in these regions could easily be fitted to Marcus theory. But the slopes of these central linear portions are considerably greater than that predicted by Marcus theory, i.e. by equation (5); for the example given in Figure 4, for instance, the The Bronsted Relation-Recent Developments difference amounts to a factor of two. An analysis of such data by Marcus theory would therefore produce an intrinsic barrier only half the true value.Theoretical examination of proton transfer using the BEBO method for generating reaction barriers leads to a similar conclusion.18 Marcus has derived an extended version of his theory starting with BEBO premise^,^ and this too gives a sigmoid dependence of a upon dE whose central linear region has a slope some 50% greater than the slope predicted by the simple theory. Another model study which uses a modified Sat0 potential energy surface to describe the proton transfer process42 gives the same result, i.e. it too suggests that intrinsic barriers obtained by applying simple Marcus theory to experimental data are apt to be somewhat lower than true values. It should be emphasized, however, that these theoretical studies all give Brernsted relations which are curved.They are also in unanimous agreement with Marcus theory in predicting that the degree of curvature will change with the intrinsic reactivity of the system, and that faster reactions will show more curvature than more slowly reacting systems. These model studies thus support the important qualitative deductions about Brarnsted plot curvature which can be made using simple Marcus theory. It is likely, moreover, that the relative values of parameters obtained using simple Marcus theory are valid even if these quantities themselves are misleading in an absolute sense. It is interesting, therefore, that the greatest intrinsic barrier listed in Table 1 is for aromatic protonation; this reaction destroys a resonance- stabilized ring, and it is therefore certain to be accompanied by much structural reorganization, a feature known to make proton transfer slow.The second greatest barrier is for the dehydration of acetaldehyde hydrate, a reaction with a cyclic transition state which contains a molecule of solvent water in addition to the catalyst and the here again, therefore, considerable heavy- atom motion will occur as the reaction takes place. Another interesting system is diazoacetate ion hydrolysis, where the barrier for catalysis by phenols and carboxylic acids is somewhat greater than that for catalysis by ammonium ions; this difference is consistent with the hypothesis that charge delocalization, which occurs in carboxylate and phenoxide but not ammonium ions, makes proton transfer kinetically more diffi~ult.~d~*~ These model studies also have an interesting bearing on the generally held belief that 01 will be one-half when the proton transfer is between bases of equal strength and therefore dG" = 0.This idea is implicit in simple Marcus theory; it follows, for example, from equation (4). The model studies, however, suggest that awill be one-half at dG"= 0 only when the system is symmetrical. For the intersecting oscillators model, this means that the relationship between force constants and bond strength must be the same for AH as for BH, and that changes in dE along the reaction series must be made up of equal changes (in opposite directions) in the strengths of both AH and BH bonds.It is difficult 41 G. W. Koeppl, unpublished work. a A. J. Kresge and Y. Chiang, J. Amer. Chem. SOC.,1973,95, 803. 490 Kresge to say whether the first of these conditions will be met in general, but the second is certainly contrary to the way Brnmsted relations are commonly set up: the usual practice is to keep one of the reaction partners (the substrate) constant and to vary only the other (the catalyst). When this feature is included in the model, the kind of skewed behaviour illustrated by line C of Figure 4 results; in this case, the value of a at dE = 0 is 0.6 rather than O.5.ls Symmetry for the BEBO model also means that both reaction partners must contribute equally to changes indE, and it requires further that the bond-energy- bond-order relationships which this method uses" be the same for the forming as for the breaking bond.18 These BEBO relationships, however, are sensitive to the identity of the atoms bonded," and there is also some evidence that they depend upon the structure of the molecule in which the bond happens to be located.44In particular, they seem to change regularly with factors such as charge delocalization and heavy-atom rearrangement, which implies that they may be different for bonds to hydrogen in pseudo acids than for bonds to hydrogen in oxygen and nitrogen species.This suggests that a is especially likely to deviate from being one-half at dE = 0 for Brernsted relations generated by proton transfer between pseudo acid substrates and oxygen or nitrogen catalysts.3 Anomalous Exponents During very nearly all of the long history of the Brarnsted relation, it was taken for granted that Brransted exponents would never be less than zero nor greater than one. Brarnsted and Pedersen said as much in their original paper,' and the statement was repeated in many discussions of acid-base catalysis. It is logical, in a sense, to limit Brernsted exponents to the range zero to one, for it is only then that proton transfer reactions can be acid-catalysed in one direction and base-catalysed in the other. Consider, for example, proton transfer from an acid to a substrate [equation (9)],with Brernsted relations for forward ka HA + S+ A-+ HS+ (9) kB and reverse reactions: kA = GA(KHA)" When a eXCeedsand kg = GB(~/KHA)B.unity, the forward reaction still behaves normally in the sense that kA increases with increasing acidity of the proton donor, i.e. it is acid-catalysed. But now, since the sum of aand /3 must be unity,* /?will be negative and kg = GB(~/KHA)-B = GB(KHA)B.Thus the rate of the reverse reaction also increases with the acid strength of the catalyst, i.e. the reverse reaction is acid-catalysed as well. When a is less than zero, similar arguments lead to base catalysis in both directions. * This may be seen by comparing the expression for the equilibrium constant of equation (9), K = (A-))-KHA/KHS+,with the ratio of the Brsnsted relations for the forward(HS)(9and the reverse reactions, K = kA/kB = GA(KHA)u/GB(Ka)-p= (GA/GB)(Km?Q+p.Since KHAis the only quantity varied, its exponent must be the same in both expressions for K,and a -I-p= 1.A. V.Willi, Helv. Chim. Acta, 1971,54, 1220. 491 TIre Brmsted Relation-Recent Developments Brransted exponents outside the range zero to one are also incompatible with the practice of equating the exponent with the fractional extent of proton transfer at the transition state of the reaction being correlated: less than no transfer (a < 0) and more than complete transfer (a > 1) clearly have no meaning. Certain expected relationships between the free energies of initial, final, and transition states of the same reaction also require Brsnsted exponents to lie in the range zero to ~ne.~~~~~ Thisfollows from the fact that structural changes occur smoothly and continuously as a system moves from reactants through transition state to products; the transition state thus has a structure intermediate between reactants and products. A perturbation made upon the system, such as the substituent change in the catalyst commonly used to produce variation along a Brransted correlation reaction series, must therefore affect the free energy of the transition state by an amount which is also intermediate between its effects on the initial and final states.This requires the substituent effect on the change in free energy between initial and transition states, 8~dG*,to be in the same direction as, but not greater than, the substituent effect on the free energy change between initial and final states, 8RdGO; and that, since 8dG*/8RdG0 = a (seefootnote, p.480), limits a to the range zero to one. Despite all of this, a number of Brsnsted exponents less than zero, and others greater than one, have been discovered during the past few years. The first of these were found almost simultaneously by Bordwellq6 and Schechte?' in the reaction of substituted 1-phenyl-1-nitroethanes with hydroxide ion, and by BordwelP in the analogous reaction of substituted 1 -phenyl-2-nitropropanes. Bordwell later provided additional examples using substituted phenylnitro- methanes reacting with hydroxide ion and with several different amine bases.48 Both Bordwell and Schechter, moreover, pointed out that data which had been in the literature for some time on the acidity constants of nitromethane,'* nitro- ethane,'* and 2-1itropropane~~ and the specific rates of reaction of these sub-stances with hydroxide ion,61 when combined gave a negative value of a.These reactions all use nitroampound pseudo acids as the proton donor, but Stuehr has recently found that proton transfer from a series of internally hydrogen- bonded phenols to hydroxide ion also gives a Brsnsted plot with a greater than These examples are summarized in Table 2.In each of these reaction series, the substance held constant, and therefore the one taken to be the substrate, is either the hydroxide ion or an amine. Amines 4c J. E. Lemer and E. Grunwald, 'Rates and Equilibria of Organic Reactions', Wiley, New York, 1963, p.156,235. 4e F. G. Bordwell, W. J. Boyle, jun., J. A. Hautala, and K. C. Yee, J. Amer. Chem. SOC., 1969, 91, 4002; F. G. Bordwell, W. J. Boyle, jun., and K. C. Yee. ibid., 1970, 92, 5926. '' M. Fukuyama, P. W. K. Flanagan, F. T. Williams, jun., L. Frainier, S. A. Miller, and H. Schechter, J. Amer. Chem. SOC.,1970,92,4689. F. G. Bordwell and W. J. Boyle, jun., J. Amer. Chem. SOC.,1971, 93, 511; 1972, 94, 3907. 40 R. G. Pearson and R. L. Dillon, J. Amer. Chem. SOC.,1953,75,2439. bo D. Turnbull and S. H. Maron, J. Amer. Chem. SOC.,1943, 65, 212; G. W. Wheland and J. Farr, J. Amer. Chem. SOC.,1943, 65, 1433. b1 S. H. Maron and V. K. La Mer, J. Amer. Chem. SOC.,1938,60,2588. ba M. C. Roseand J. Stuehr,J. Amer.Chem. SOC.,1971,93,4350; 1970,94,5332. Table 2 Anomalous Brmsted relations Reaction U Reference H.0 1. RzCHNOz + HO-+R,CNO,-+ H,O -0.48 k 0.13 49, 50, 51 (R = H or CH3) H*O 2. ArCH,NO, + HO-+ArCHNO,-+ H,O 1.54 & 0.10 48 H*O 3. ArCH,NO, + B -+ ArCHN0,-+ BH+ 1.29 f 0.10 48 (B = morpholine) H,O4. ArCH,NO, + B __+ ArCHN02-+ BH+ 1.30 k 0.09 48 (B = 2,6-lutidine) HsO 5. ArCHMeNO, + HO---+ ArCMeN0,-+ H20 1.14 k 0.04 48 50% H,O-dioxan 6. ArCHMeNO, + HO------+ ArCMeN0,-+ H,O 1.17 -1.20 47 & 0.05 -0.06 7. ArCHMeNO, + RO-50% HpO-MeOH ArCMeN0,-+ ROH 1.37 f 0.07 46 8. ArCH,CHMeNO, + RO-Hzo-Meo: ArCH,CMeN02-+ ROH 1.61 5 0.11 46 HzO 9. ArOH + HO-__+ ArO-+ H,O 1.21 k 0.09 52 (Phenolic hydroxy-group internally hydrogen-bonded to azo or carboxylate function) The Bronsted Relation-Recent Developments and the hydroxide ion, however, are normally used as catalysts rather than substrates in Brarnsted oorrelations. In all but the last example, moreover, the catalysts are pseudo acids, but pseudo acids generally serve as substrates rather than catalysts.In these anomalous relations, therefore, the catalyst and substrate roles are the reverse of normal practice. It is in a sense somewhat arbitrary, however, which partner in a proton-transfer reaction is taken to be the catalyst and which the substrate, and in that respect this reversal of roles is of no con- sequence. Marcus theory, on the other hand, takes the view that it does make a difference, and the theory in fact predicts that anomalous relations are more likely to be found when the substance varied is a pseudo acid than when it is an oxygen or nitrogen species .63 The argument is based upon an expression for a which is more complete than that given in equation (4).The latter was derived on the assumption that the intrinsic barrier does not change along a reaction series; however, this will not be necessarily true, and, when that possibility is taken into account, equation (10) u= ddG*/dG" = [(l + dG0/4dGo*)/2] + [1 -(dGo/4dG~*)2]ddG~*/ddGo (10) results. For a series with a constant intrinsic barrier, ddG,*/ddG" = 0; the second term of equation (10) then drops out, leaving equation (4). This, when limited to the range of dGo over which simple Marcus theory is valid : -44 Go* < dG" c +44G0*,* confines ato the range zero to one.When the intrinsic barrier is not constant, on the other hand, the second term of equation (10) contributes, and for suitable values of ddGo*/ddGo it may make o( less than zero or greater than one. Some insight into the conditions under which ddGo*/ddGo will be significant may be gained from the Marcus theory expression, equation (ll), which gives the intrinsic barrier to proton transfer between two different bases, equation (12), as the mean of the barriers for the two identity reactions, equations (13) and (14). This makes the intrinsic barrier for proton transfer between a pseudo acid and (~Go*)AB= [(~G*)AA4-(~G*)BB]/~ (11) AH+ B-tA + HB (12) AH+A-tA+HA (1 3) BH+B-B+HB (14) an oxygen or nitrogen base the mean of the barriers for the identity reactions between the pseudo acid-base pair and the oxygen or nitrogen acid-base pair.Inasmuch as proton transfer between oxygen or nitrogen acids and bases is usually very fast whereas that between pseudo acids and bases is slow, one of these identity reactions, that between the oxygen or nitrogen acid-base pair, will * This is the range over which dG*is a single-valued function of dG"and therefore the range over which the theory is physically realistic. Is R. A. Marcus, J. Amer. Chem. SOC.,1969,91,7224. Kresge have a much lower barrier than the other and will consequently contribute little to the intrinsic barrier for the mixed reaction.Changes in the pseudo acid are therefore likely to have a much greater effect on AGO+than changes in the oxygen or nitrogen base, and ddG,*/ddG" may consequently be appreciable in the former case but negligible in the latter. Rates of identity reactions for nitro-compounds have not been measured, but it is likely that they would be both slow and variable. It is known, for example, that nitroalkanes react only slowly with oxygen and nitrogen bases even when the base is as strong as the hydroxide ion, and that the rates of these reactions are sensitive to the structure of the nitr~-compound.~~-~~~~~ This slowness and variability are likely to be accentuated when both reaction partners become pseudo acids or bases. Nitro-compounds, for instance, are poor catalysts for the dehydration of acetaldehyde hydrate when compared with oxygen and nitrogen acids of the same pKA,24 and nitronate ions are likewise poor catalysts in the decomposition of nitramide.' It seems likely, therefore, that barriers for the identity reactions of the nitro-compounds listed in Table 2 will be large and variable, and that intrinsic barriers will thus not remain constant within the reaction series. Marcus has analysed the data for the first reaction in this Table and has found ddGo*/ddGoto be 1.1.O Reaction 9 has also been treated in this way with the result ddG,*/ddG" = 0.8.62It is superficially less obvious why the intrinsic barrier should change significantly along this series, for the proton transfer here is between a group of phenols and hydroxide ion, and both reaction partners are therefore substances whose identity reactions should be fast. The phenols, however, are internally hydrogen-bonded, and their rates of reaction are accordingly slowed.Provided, then, that internal hydrogen-bond breaking is concerted with proton transfer in these systems, i.e. that the synchronous mechanism described above in the dis- cussion of internally hydrogen-bonded systems is operative, the intrinsic barrier for this series could be large and variable. It should be mentioned, however, that some of the phenols used here are rather different from the others, and that a group which makes a structurally more homogeneous subset gives an entirely normal Brnrnsted relation with a = 0.3 k 0.4.A somewhat different approach to understanding Brnrnsted relations with anomalous exponents examines the problem in terms of substituent It begins with the definition a = 8dG*/6dGo,and points out that this requires the exponent to lie in the range zero to one only if the substituent interacts with the reaction zone in just one way. When two or more interaction mechanisms are operative, and when these differ in sign and some lead or lag behind others, their effects may then combine to make 8dC* greater than SdCoor to give these two quantities opposite signs; the first of these conditions would, of course, make agreater than one and the second would make it less than zero. Situations could also exist in which substituent effects contribute to 8dG* without affecting 8RdGo,as a result of interactions which develop in the transition state b4 A.J. Kresge, J. Amer. Chem.SOC.,1970,92,3210. The Brmsted Relation-Recent Developments but are absent from initial and final states. This could happen, for example, in bimolecular reactions where two reagents separated by large distances in the initial and final states come together in the transition state. These ideas are well illustrated by the first reaction series listed in Table 2, where a proton is transferred to hydroxide ion from the successively methyl- substituted nitro-compounds CH ,NO 2, CH ,CH 2N02, and (CH 3) ,CHNO 2. The experimental data give 64G* =+1.O kcal mol-l and 64Go =-2.0 kcal mol-l as the average effects of methyl substitution.Since these quantities are different in sign, a is negative. The change in AGO, moreover, is opposite from that expected on the basis of the acid-weakening polar (inductive)effect commonly shown by methyl groups in acid ionization equilibria, and that suggests that a second interaction is operating. This additional interaction has in fact been identified recently, through the use of secondary isotope effects, as hyper- conjugative stabilization of the carbon-nitrogen double bond in nitronate ions.66 In addition to these two effects, each of which contributes to SdG* as well as to 6RdGo,two others will operate in the transition state alone and thus will add to 8dG* without affecting 6dGO.One of these is a polar (electrostatic) inter- action between the hydroxide ion and the methyl group dipole; this is absent from the initial state because the hydroxide ion and the nitro-compound have not yet come together, and it does not exist in the final state because the hydroxide ion has now beenconverted into a water molecule. The other effect without an initial-state or a final-state counterpart is the polar interaction of the methyl group with the negative charge which builds up on the a-carbon atom. Although the negative charge of fully formed nitronate ions is very probably largely delocalized on to the nitro-group, this is less likely to be the case in a nitronate-ion-producing transition state where the carbon-nitrogen double bond, through which delocalization must take place, is only partly formed; electrostatic attraction between the departing proton and the electron pair of the breaking bond will further inhibit the flow of negative charge away from the a-carbon atom.It is possible, on the basis of reasonable assumptions concerning the structure of the transition state, to make rough quantitative estimates of each of these effects. These lead to a value of a which is negative, in agreement with the experimental result. The model, moreover, may be applied to the other nitro- compound reaction series listed in Table 2 by leaving out the hyperconjugative interaction, inasmuch as this effect will be either absent from or constant along each of the other series.The result now is a positive value of a, which is again consistent with the experimental finding. An important feature of this model is the incomplete delocalization of charge on to the nitro-group in the transition state, and its consequent build-up on the a-carbon atom. This places the charge considerably closer to the substituent in the transition state than in the final state, and that permits a disproportionately large transition-state interaction. It follows from this that removing the sub- b5 A. J. Kresge, D. A. Drake, and Y. Chiang. Canad. J. Cfiem.,in the press. Kresge stituent to some more remote part of the system might restore normal behaviour, and it is significant, therefore, that Brarnsted relations constructed by keeping the nitro-compound constant and changing substituents in the base have com- pletely normal exponents ranging from 0.50 to 0.65.48+68Marcus theory, of course, also predicts that moving the site of substitution away from the pseudo acid part of the system and into the oxygen or nitrogen base in this way should restore normal behaviour.Proton transfer from pseudo acids is usually accompanied by considerable charge delocalization and redistribution, whereas that from oxygen or nitrogen species ordinarily involves little or none. The unusual charge distributions which lead to disproportionate transition state interactions are therefore more likely to occur in pseudo acids and bases. Analysis in terms of substituent effects thus leads to the same general conclusion as Marcus theory, viz.that, in proton transfer between a pseudo acid and an oxygen or nitrogen base, anomalous Brarnsted relations are less likely to occur if the pseudo acid is held constant and the oxygen or nitrogen species is varied than if the pseudo acid is varied and the other species held constant. 4 SystematicDeviations A. Electrostatic.-It is well known that the catalysts in a Brernsted relation must be structurally similar if the correlation is to be a good one. Differences in charge type are especially likely to produce deviations, as is illustrated, for example, by the base-catalysed decomposition of nitramide :here dipositive, neutral, negative, and dinegative catalysts define four parallel lines separated by more than two orders of magnitude in reacti~ity.~,~~Q A more recent example of the effect of charge is provided by the hydrolysis of ethyl vinyl ether, where neutral carboxylic acids and positively charged amino-acids give good parallel correlations separated by a factor of two in rate, and individual negatively charged acids show devia- tions from the neutral carboxylic acid line which approach a factor of ten in reacti~ity.~~These data for ethyl vinyl ether hydrolysis illustrate quite dramatical- ly what may happen when a small number of structurally dissimilar catalysts are used in a Brarnsted relation: taking the catalysts in pairs gives values of a which range from -3 to +15! Systematic deviations such as these may be understood in terms of transition- state interactions not unlike those used above to account for anomalous Brarnsted relations.For example, in the transition state for decomposition of nitramide, the base is removing a proton from a neutral substrate, and the substrate is there- fore taking on negative charge. A positive charge initially situated on the base L6 (a) R. G. Pearson and F. V. Williams, J. Amer. Chem. SOC.,1954,76, 258;M.J. Gregoryand T.C. Bruice, ibid., 1967,89, 2327;J. E. Dixon and T. C. Bruice, ibid., 1970,92, 905; (6) D.J. Barnes and R. P. Bell, Proc. Roy. SOC.,1970, A318,421. 6’ R. B. Bell, ‘Acid-Base Catalysis’, Oxford University Press, London, 1941,(a)p. 86;(6)p. 92; (c)p. 89. 6* A. J. Kresge and Y. Chiang, J. Amer. Chem. Suc., 1973,95, 803. The Bronsted Relation-Recent Developments will undergo an attractive interaction with this negative charge, and that will lower the free energy of the transition state and increase the rate of reaction.A negative charge initially on the base will have the opposite effect, and reactivity should therefore increase as the charge on the catalyst changes from dinegative to negative to neutral or dipositive, just as observed. In the hydrolysis of ethyl vinyl ether, on the other hand, a proton is being transferred from an acid to a neutral substrate, and the substrate is therefore taking on positive charge, which gives a situation just the opposite of that in nitramide decomposition. The effect of charge on reactivity should therefore also be reversed, i.e.positive charge on the catalyst should now slow the rate while negative charge accelerates, which again is just as observed. Dipolar groups in the catalyst might be expected to show similar but smaller effects. Some evidence that this is so comes from the hydrolysis of eight different vinyl ethers catalysed by the same set of seven neutral carboxylic acids.60 A few of the acids which contain strongly dipolar groups, such as cyan0 or methoxy, show small but consistent, i.e.always positive or always negative, deviations from Bransted correlations based upon all of the data. These deviations, moreover, are more pronounced at the strong acid ends of the correlations than at the weak acid ends, and they might therefore influence the slopes of these plots.It is thus significant that those of these hydrolyses which show maximum isotope effects, and in whose transition states the proton is therefore presumably half-trans- ferred,60 give Brarnsted a's greater than 0.5, i.e. of the order of 0.60-4.65. Further differences of this sort may be found in the hydrolysis of ethyl vinyl ether where a = 0.70,69abut isotope effects in H20-D20mixtures suggest only 0.6 proton transfer,61 and comparison of kinetically with competitively deter- mined isotope effects implies the value 0.56.62 It is possible, on the other hand, that these differences may be due to the fact that the isotope effects are for hydronium ion catalysis whereas the Brarnsted correlations are based upon considerably weaker carboxylic acid catalysts :the degree of proton transfer at the transition state should of course decrease with increasing catalyst strength.The Brarnsted plot curvature which this implies corresponds, on the basis of simple Marcus theory, to an intrinsic barrier of ca. 10 kcal mol-l. This is not an unreasonable value, and it is in fact in good agreement with an estimate of 12 kcal mob1 which can be made using the measured free energy of activation of the hydrogen ion catalysed reaction, dG* = 17 kcal and a value of its standard free energy of reaction, dG"= 10 kcal mol-l, based upon an estimate of the PKA of carbon-protonated ethyl vinyl ether.63 It is interesting that the second of these calculations of the intrinsic barrier for 69 (a) A.J. Kresge, H. L. Chen, E. Murrill, M. A. Payne, and D. S. Sagatys, J. Amer. Chem. SOC.,1971, 93, 413; (b) A. J. Kresge, and H. J. Chen, ibid., 1972, 94, 2818. 6o A. J. Kresge, D. S. Sagatys, and H. L. Chen, J. Amer. Chem. SOC.,1968,90,4174. A. J. Kresge and Y. Chiang, J. Chem. SOC.(B), 1967,58. 6a M. M. Kreevoy and R. E. Eliason, J. Phys. Chem., 1968,72,1313. P. Salomaa and A. Kankaanpera, Acta Chem. Scand., 1966,20, 1802. Kresge ethyl vinyl ether hydrolysis assumes the work term, wr, to be zero whereas the first does not. The good agreement between the two estimates might thus be taken as evidence that wr is in fact small and cannot therefore represent reagent positioning or solvent reorganization in addition to reactant encounter.This argument, however, presumes that carboxylic acids and the hydronium ion con- stitute a homogeneous set of catalysts correlated by a single Bronsted relation, but that, as will be seen in the section below, is probably not the case. B. Hydronium and Hydroxide Ions.-It has been recognized for some time that the hydronium and hydroxide ions usually do not conform to Brernsted relations based upon other, non-solvent-derived, catalytic species. In 1941, for example, BelPb listed seven reactions for which the hydroxide ion catalytic coefficient had been measured and for which Brransted relations were also available. In only one of these systems, the mutarotation of glucose, was the rate constant cal- culated from the correlation as close as a factor of five to the observed result; in all of the other cases the discrepancy amounted to at least two, but more often three or four, orders of magnitude.Less information was available at that time for hydronium ion catalysis, but what few data did exist suggested that this species was also in general an anomalous catalyst. Table 3 lists these early examples together with some more recent results. Many more data are now available for hydronium ion catalysis, principally because of the recent positive identificati~n~~ and subsequent detailed investiga- tion of slow proton transfer from acid to substrate, a reaction type not available before 1959. Very few of the reactions listed in this Table show good agreement between observed and calculated hydronium and hydroxide ion catalytic co- efficients.Both ions are in most cases anomalously poor catalysts, but, signi- ficantly, in a few systems they are better than predicted. There is a difficulty in assigning exact acid and base strengths to the hydronium and hydroxide ions in aqueous solution which is not unrelated to this anomalous behaviour. The conventional acidity constant of an acid which ionizes in water according to equation (15) is given as KHA = (H,O+)(A-)/(HA); it m C. A Marlies and V.K. La Mer, J. Amer. Chem. SOC.,1935,57, 1812. O5 J. N.Brransted and E. A. Guggenheim, J. Amer. Chem. SOC., 1927,49,2554.T.M. Lowry and G. L. Wilson, Trans. Faraday SOC.,1928,24,683. 67 R. P. Bell, M. H. Rand, and K.M. A. Wynne-Jones, Trans. Faradny SOC., 1956,52, 1093. O8 R.P.Bell and P. G. Evans, Proc. Roy. SOC.,1966, A291,297. A. J. Kresge, S. Slae, and D. W. Taylor, J. Amer. Chem. SOC., 1970,92, 6309. 'O R. J. Thomas and F. A. Long, J. Amer. Chem. SOC.,1964,86,4770. i1 V. Gold and D. C. A. Waterman, J. Chem. SOC.(B), 1968,849. 72 T. S. Straub, Ph.D. Thesis, Illinois Institute of Technology, Chicago, Ill., 1970. 73 V. Gold and D. C. A. Waterman, J. Chem. SOC. (B), 1968, 839. 7a A. J. Kresge, Y.Chiang, and J. R. Wiseman, to be published. 75 M. M. Kreevoy, T. S. Straub, W. V. Kayser, and J. L. Melquist, J. Amer. Chem. SOC., 1967, 89, 1201. 76 M. M. Kreevoy and R. A. Landholm, Internat. J. Chem. Kinetics, 1969,1, 157. 77 R.E.Barnett and W. P.Jencks, J. Amer. Chem. SOC.,1969,91,2358. 78 M.E. Aldersley, A. J. Kirby, and P. W. Lancaster, J. C. S. Chem. Comnr., 1972, 570. 'O A. J. Kresge and Y. Chiang, J. Amer. Chem. SOC., 1959,81,5509. Table 3 Catalysis by hydroxide and hydronium ions Reaction Referencelog -kcalc Hydroxide Ion 1. Decomposition of nitramide 2 x 104 -3.0 576,64 2. Iodination of acetone 2.5 x 10-l -3.54 5,576 3. Iodination of acetonylacetone 1.7 -3.74 5,576 4. Iodination of monochloroacetone 9.3 -3.43 5,576 5. Iodination of monobromoacetone 2.0 x lo8 -2.92 5, 57b 6. Bromination of 1,l-dichloroacetone 4.5 x lo2 -2.71 5, 576 7. Detritiation of phenyl [2-3H]acetylene 2.3 x lo2 -2.10 9 8. Ionization of ethyl nitroacetate 1.5 x 105 -2.57 566 9. Detritiation of 1,4-dicyano[ 1-SH]but-2-ene 2.1 x 10-1 -3.0 7a 10.Detritiation of p-nitro[~-~H]benzylcyanide 2.6 x 10 -1.65 7c 11. Mutarotation of glucose 6.4 x 10 +0.71 57b, 65,66 12. Hydration of acetaldehyde 8 x lo4 +1.31 67 13. Dehydration of methylene glycol 1.6 x 103 +0.90 68 Hydronium Ion 1. Iodination of acetone 7.4 x 10-4 -1.29 57c 2. Detritiation of 1,3,5-trirnetho~y[2-~H]benzene 6.2 x lo-* -1.53 69 3. Detritiation of [l-3H]azulene 1.8 x 10-1 -1.63 70 4. Detritiation of [3-3H]guaiazulene 6.1 -1-60 70 5. Hydrolysis of ethyl isopropenyl ether 5.8 x lo2 -1.57 59a 6. Hydrolysis of ethyl cyclopentenyl ether 4.5 x loa -1.35 59a 7. Hydrolysis of ethyl cyclohexenyl ether 8.0 x 10 -0.89 59a 8. Hydrolysis of methyl cyclohexenyl ether 4.2 x 10 -1.24 59a 9.Hydrolysis of phenyl isopropenyl ether 6.0 -0.99 59a 10. Hydrolysis of ethyl vinyl ether 1.8 -1.35 59a 11. Hydrolysis of ~-meethoxy-tmethylst~ene 1.7 -1.34 59b 12. Hydrolysis of phenyl vinyl ether 3.3 x 10-8 -1.84 59a 13. Hydrolysis of 2-dichloromethylene-1,3-dioxolan 1.0 x 1v -1.81 71 14. Hydrolysis of dimethylketen dimethyl acetal 2.9 x 10 -1.01 72 15. Hydrolysis of cyanoketen dimethyl acetal 2.4 x 10 -2.07 73 16. Hydrolysis of methylbromoketen diethyl acetal 6.5 -1.12 72 17. Hydrolysis of dichloroketen dimethyl acetal 5.2 -1.12 72 18. Hydrolysis of dichloroketen diethyl acetal 3.4 -1.39 72 19. Hydration of bicyclo[3,3,1]non-l-ene 3.2 x 10 -1.33 74 20. Acid cleavage of allylmercuric iodide 1.4 x 1O-e -1.53 75 21.Acid cleavage of isobutenylmercuric bromide 2.4 x -1.60 76 22. Mutarotation of glucose 2.4 x 10-+0.24 65 23. Hydration of acetaldehyde 9.3 x 108 +0.22 67 24. Dehydration of methylene glycol 2.7 +0.08 68 25. Intramolecular aminolysis of S-acetylmercaptoethylanine 6.5 x 10LO + 1.36 77 26. Intramolecular aminolysis of S-chloroacetylmercaptoethylamine 6.5 x 10'0 + 1.40 77 27. Hydrolysis of di-isopropyl-N-n-propylmaleanicacid 6.5 x lo8 +1.38 78 For footnote references in Table, see page 499. The Brmsted Relation-Recent Developments is therefore equal to the equilibrium constant for equation (15), Kls = (H 30+)(A--)/(HA)(H 20),times the concentration of water: K,,(H 20) = (H ,O+)(A-)/(HA) = KHA.The particular case of equation (15) for HA = H HA + H20 H,O+ + A-(15) is given by equation (16), whose equilibrium constant is of course unity.The H,O+ + H20 H,O+ + H,O (16) acidity constant of the hydronium ion must therefore be unity times the con- centration of water, which is 55 on the molar scale; thus KH,o+,according to this convention, is 55. A similar argument applied to the conjugate acid of the hydroxide ion, water, leads to 10-14/55= 1.8 x 10-l6 as the acidity constant of this species. Although these are the values commonly used in fitting hydronium and hydroxide ion points to Brarnsted relations, it is not at all certain that they should be mixed in with more conventional acidity constants, inasmuch as they involve the concentration of the solvent whereas the other constants refer only to dilute solution solute species.It has been pointed outao that this difficulty would be relieved to a considerable extent if the appropriate species to be used in equation (16), and its analogue for H20 acting as an acid, were monomeric water. Since liquid water has an extensively hydrogen-bonded polymeric structure, the fraction of monomeric molecules is small, and monomeric water is therefore in a sense in relatively dilute solution in the rest of the solvent. The effect of using [H20] < 55 mol 1-1 would be to lower KE,o+ and to raise KH~OThese changes are in the directions required to reduce negative rate deviations for both ions, and negative deviations in both cases make up the bulk of the anomalous behaviour.Several estimates of the acidity constants of H ,O+ and H 20have been made on this basis,80~81 even though the concentration of monomeric molecules in liquid water is not very well known. The idea has more often been applied in the opposite sense, to calculate acidity constants for H,O+ and H20by putting the catalytic coefficients for these species on Brarnsted lines defined by other cata- lyst~.59,7~The values obtained, however, scatter widely, and the hypothesis at any rate can account only for negative and not for positive rate deviations. The anomalous behaviour of hydronium and hydroxide ions has also been discussed in terms of Bransted plot and electrostatic interactions,68 but these explanations again cannot account for both positive and negative deviations.A somewhat different approach has been taken in connection with the reversible ionization of carbon acidsa This explanation focuses its attention on the reverse reaction, which it describes as occurring via a hydrogen-bonded complex between the carbanion and the protonated proton acceptor. It requires solvent water, which will be the protonating agent in the reverse reaction when R. P. Bell, Trans. Faraday SOC.,1943,39,243. A. 3. Kresge, Y. Chiang, and Y. Sato, J. Amer. Chem. SOC.,1967, 89, 4418. Kresge proton transfer in the forward direction is to hydroxide ion, to be relatively ineffective at forming such a hydrogen-bonded complex; this will impede the reverse reaction, which, via the equilibrium, will slow the forward process as well.It is not clear, however, why water should be a poor hydrogen-bond donor relative to its ~KA, and the explanation also does not allow for cases where the reverse reaction does not occur but the anomaly still exists, such as the halogena- tion of ketones. An explanationQ which is free of these objections makes use of the well-known fact that hydronium and hydroxide ions enter into the hydrogen-bonded structure of liquid water unusually well and are consequently more strongly solvated than most other acids and bases. Desolvation of the catalyst will therefore require the input of more energy, and will make a greater contribution to the reaction barrier, when the proton transfer involves hydronium or hydroxide ion than when it involves other acids or bases.It is likely, however, that this rate-retarding effect will operate only when the substrate itself cannot hydrogen-bond to the solvent, for, when it can, proton transfer by the Grotthuss chain mechanism becomes possible and desolvation is not necessary. The Grotthuss chain mechanism, moreover, gives a special advantage to hydronium and hydroxide ions, and positive deviations might therefore be observed in some cases. This hypothesis is supported by the fact that in all but one of the 31 reactions of Table 3 which show negative deviations, the proton transfer is to or from carbon. Carbon acids and bases, of course, form hydrogen bonds to water reluctantly if at all; the Grotthuss chain mechanism will therefore not be avail- able here, and negative deviations will occur.The single exception to this generalization is the nitrogen acid nitramide, whose decomposition catalysed by hydroxide ion gives a strong negative deviation despite the fact that the proton is transferred from an N-H bond which is presumably capable of hydrogen- bonding to the solvent. The deviation here, however, is based upon a linear Brarnsted relation, and a curved ~orrelation,~ which erases the anomaly, fits the data equally well. All of the nine reactions listed in Table 3 which show positive deviations, on the other hand, are systems in which the catalyst operates on a hydroxy- or an amino-group. In each of these cases the substrate will be hydrogen-bonded to the solvent at the reaction site, and in each case, therefore, proton transfer can proceed by the Grotthuss chain mechanism. That, of course, gives positive deviations. The author thanks the National Science Foundation for financial support which made this review possible.

ISSN:0306-0012    

DOI:10.1039/CS9730200475

出版商:RSC    

年代:1973

数据来源: RSC