|
1. |
Front cover |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 013-014
Preview
|
PDF (120KB)
|
|
ISSN:0306-0012
DOI:10.1039/CS98514FX013
出版商:RSC
年代:1985
数据来源: RSC
|
2. |
Back cover |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 015-016
Preview
|
PDF (32KB)
|
|
摘要:
ISSN 0306-0012 CSRVBR 14(4) 357473 (1985) Chemical Society Reviews Vol 14 No 4 1985 Page HAWORTH MEMORIAL LECTURE The Sweeter Side of Chemistry By Leslie Hough 357 Guanidine Derivatives Acting at Histaminergic Receptors By G. J. Durant 375 The Chemistry of Peroxooium Ions ad Dioxygen Ylides By J. C. Mitchell 399 Brownian Dynamics with Hydrodynamic Interactions: The Application to Protein Diffusional Problems By Eric Dickinson 421 1985 Indexes 457 The Royal Society of ChemistryLondon
ISSN:0306-0012
DOI:10.1039/CS98514BX015
出版商:RSC
年代:1985
数据来源: RSC
|
3. |
Contents pages |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 017-020
Preview
|
PDF (147KB)
|
|
摘要:
CHEMICAL SOCIETY REVIEWS VOLUME 14, 1985 0 Copyright 1985 LONDON THE ROYAL SOCIETY OF CHEMISTRY TURES. By Peter M. A. Sherwood 1 AND C-C, C-N, AND C-0 TRIPLE BONDS. By Malcolm H. Chisholm, David M. Hoffman, and John C. Huffman 69 HALIDESAS SELECTIVE IN ORGANIC SYNTHESIS. By David C. Billington 93 CENTENARY LECTURE. PHASEEQUILIBRIUM STRUCTURE. By B. Widom 121 MOLECULES.By Ian W. M. Smith 141 ACTIVESITE. By John M. Pratt 161 S-NITROSATION OF S-NITROSOCOMPOUNDS.By D. Lyn H. Williams 171 LIGANDSUBSTITUTION OF SQUARE-PLANARREACTIONS MOLECULES. By Ronald J. Cross 197 GEL ELECTROPHORESIS. By Larry R. Sherman and James A. Goodrich 225 FOR CHEMICAL IN SOLUTION. By Michael J. Blandamer, John Burgess, and Jan B. F. N. Engberts 237 CONTENTS PAGE PHOTOELECTRON STUDIESOF ELECTRODESPECTROSCOPIC AND RELATEDSTRUC-POLYMERIZATION SYSTEMS.By Constentinos M. Paleos 45IN ORGANIZED REACTIONS THE TRIPLE BOND IN DIMOLYBDENUMINVOLVING AND DITUNGSTEN HEXA-ALKOXIDES X-ALLYLNICKEL REAGENTS AND INTERFACIAL TILDEN LECTURE. THE COLLISION DYNAMICS EXCITEDOF VIBRATIONALLY THE B12-D~~~~~~~~ THE PROTEIN CONTROLS ISOMERASEENZYMES:How THE AND THE REACTIONS DEVELOPMENT SULPHATE-POLYACRYLAMIDETHE HISTORICAL OF SODIUMDODECYL ACTIVATIONPARAMETERS REACTIONS R. A. ROBINSON MEMORIAL LECTURE. POTENTIOMETRIC OFTITRATIONS AQUEOUS SOLUTIONS. By A. K. Covington 265CARBONATE AND ELECTRON-TRANSFER OF THETILDEN LECTURE. STRUCTURE REACTIVITY BLUE COPPER PROTEIN By A. G. Sykes 283PLASTOCYANIN. THEDRIFS OF TRANSPORTHARD-SPHERE PROPERTIES.By J. H. Dymond 317 HAWORTH MEMORIAL LECTURE. THE SWEETER SIDEOF CHEMISTRY.ByLeslie Hough 357 GUANIDINE ACTINGAT HISTAMINERGICDERIVATIVES RECEPTORS. By G. J. Durant 375 THE CHEMISTRY IONSAND DIOXYGENOF PEROXONIUM YLIDES. By J. C. Mitchell 399 BROWNIAN WITH HYDRODYNAMIC THE APPLICATION DYNAMICS INTERACTIONS: TO PROTEIN PROBLEMS. By Eric Dickinson 421DIFFUSIONAL 1985 Indexes 457 ISSN 0306-0012 CSRVBR 14(4) 357473 (I 985) Chemical Society Reviews Vol 14 No 4 1985 Page HAWORTH MEMORIAL LECTURE The Sweeter Side of Chemistry By Leslie Hough 357 Guanidine Derivatives Acting at Histaminergic Receptors By G. J. Durant 375 The Chemistry of Peroxonium Ions and Dioxygen Ylides By J. C.Mitchell 399 Brownian Dynamics with Hydrodynamic Interactions: The Application to Protein Diffusional Problems By Eric Dickinson 421 1985 Indexes 457 The Royal Society of Chemistry London Chemical Society Reviews EDITORIAL BOARD Professor K. W. Bagnall, B.Sc., Ph.D., D.Sc., C.Chem., F.R.S.C. Professor B. T. Golding, B.Sc., M.Sc., Ph.D., C.Chem., F.R.S.C. Professor G. Pattenden, Ph.D., C.Chem., F.R.S.C. Professor P. A. H. Wyatt, B.Sc., Ph.D., C.Chem., F.R.S.C. (Chairman) Dr. D. A. Young, Ph.D., D.Sc., C.Phys., M.Inst. P. Editor: K. J. Wilkinson, B.Sc., M.Phi1. Chemical Society Reviews (ISSN 0306-001 2) appears quarterly and comprises approximately 20 articles (ca. 500 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 intended to be 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 Managing Editor, Books and Reviews Section, The Royal Society of Chemistry, Burlington House, Piccadilly, London, W 1V OBN. Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at E15.50 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way.All other orders accompanied with payment should be sent directly to The Royal Society of Chemistry, The Distribution Centre, Blackhorse Road, Letch- worth, Herts. SG6 1HN England. 1985 annual subscription rate U.K. E45.00, Rest of World E47.50, U.S.A. $87.00. Air freight and mailing in the U.S.A. by Publica- tions Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting lnc., 200 Meacham Avenue, Elmont, New York 11003. Second class postage is paid at Jamaica, New York 11431. All other despatches outside the U.K. by Bulk Airmail within Europe, Accelerated Surface Post outside Europe. 0The Royal Society of Chemistry, 1985 All Rights Reserved No part of this book may be reproduced or transmitted in any form or by any means -graphic, electronic, including photocopying, recording, taping, or information storage andretrievalsystems- without writ tenpermission from The Royal Society of Chemistry Published by The Royal Society of Chemistry, Burlington House, London, WlV OBN Printed in England by Richard Clay (The Chaucer Press) Ltd, Bungay, Suffolk
ISSN:0306-0012
DOI:10.1039/CS98514FP017
出版商:RSC
年代:1985
数据来源: RSC
|
4. |
Back matter |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 021-028
Preview
|
PDF (967KB)
|
|
ISSN:0306-0012
DOI:10.1039/CS98514BP021
出版商:RSC
年代:1985
数据来源: RSC
|
5. |
Haworth Memorial Lecture. The sweeter side of chemistry |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 357-374
Leslie Hough,
Preview
|
PDF (856KB)
|
|
摘要:
HAWORTH MEMORIAL LECTURE * The Sweeter Side of Chemistry By Leslie Hough THE C HEM I S TR Y D EP A R T MEN T, K IN G ’S C 0L LEG E LO N D 0N (K Q C), CAMPDEN HILL ROAD, KENSINGTON, LONDON, W8 7AH Sir Walter Norman Haworth’s researches played an important role in the determination of the molecular structure of sucrose and he was fully aware of its potential ‘as a source of new industrial materials and intermediates’. After World War 2, he organized a research programme at the University of Birmingham on ‘The Utilisation of Sucrose’,2 under Dr. Leslie F. Wiggins and supported by the Colonial Products Research Council, with a view to finding a market for surplus sugar. Their efforts were concentrated upon degradation products, such as furfural and its derivatives, and it is noteworthy that the review written by Wiggins in 1947 cites only a few genuine sucrose derivatives, mostly octa-substituted, owing to the difficult experimental problems in handling sucrose.Studies were hampered by the multiplicity of hydroxy-groups on sucrose and its sensitivity to acid, coupled with its limited solubility in organic solvents and the lack of suitable protective groups, since standard procedures usually failed. The energy crisis of the 1970s focussed attention on the economic potential of sucrose as an ubiquitous feed stock for chemical and microbiological exploitation, consequently chemical studies have taken on an added importance. Sucrose (1) is a non-reducing disaccharide with eight hydroxy-groups, arranged in the crystal with a conformation3 in which the a-D-glUCOpyranOSyl unit is 4C, whilst the P-D-fructofuranoside unit is 3T4 (2), and the two units are bridged by two intramolecular hydrogen bonds (3) from 0-6’ to 0-5 and 0-1 to 0-2.The unprimed and primed numbers are used to indicate the carbons, and associated oxygen atoms, in the glucosyl and fructoside units respectively. In solution the overall conformation is similar to that found in the crystal, particularly around the inter-glycosidic linkage (3), as revealed by ‘H-and I3C-n.m.r. ~pectra.~ The original chemical synthesis of natural D-sucrose (1) by Lemieux and Huber in 1953 was preceded by its enzymic synthesis in 1944,6 and followed in 1978 by a synthesis ’of * Delivered at the Spring Meeting ofthe Royal Society ofchemistry Carbohydrate Group on 1st April 1985 at the University of Bristol.* I. Levi and C. B. Purves, Adv. Carbohydr. Chem., 1942, 4, 1. L. F. Wiggins, Adv. Carbohydr. Chem., 1949, 4, 293. M. R. Jenner, in ‘Developments in Food Carbohydrate’, ed. C. K. Lee, Applied Science Publishers Ltd., England, 1980, Vol. 2, p. 9 1. K. Bock and R. U. Lemieux, Carbohydr. Res., 1982, 100, 63. R. U. Lemieux and G. Huber, J. Am. Chem. SOC.,1953, 75,4118. W. Z. Hassid and M. Doudoroff, Adv. Carbohydr. Chem., 1950, 5, 29. ’Queen’s University, 1979, Canada Patent 1 556 007. The Sweeter Side of Chemistry L-sucrose (4) which was also sweet but not metabolized. Hence L-sucrose is a potential non-nutritive sweetener provided that it can be made economically.Combinations of either L-fructose and D-glucose, or D-fructose and L-glucose, termed D,L-sucrose and L,D-sucrose respectively, are also sweet and neither absorbed nor metabolized, hence they are calorie-free (non-nutritive).8 The direct use of the monosaccharides L-glucose or L-fructose for sweetening purposes is complicated by their absorption and consequent circulation in the blood stream for long periods. The 1-thio analogue of sucrose proved to be a competitive inhibitor of yeast invertase, an enzyme that hydrolyses sucrose. cx-L -glucopyranosyl-p-L -fructofuranoside (.=OH) L -sucrose) Attempts were made to insert a covalent intramolecular bridge across the two hexose units of sucrose, as for example methylene or carbonate, with a view to retaining the combination after hydrolysis of the hemiacetal linkage to give a di- reducing disaccharide.As might have been anticipated, reactions of sucrose with phosgene-pyridine, ethylene dichloroformate-sodium hydroxide, and diphenyl carbonate-sodium hydrogen carbonate led to intermolecular bridges with the direct formation of polymers-the poly(sucrose carbonates). lo Treatment of sucrose with ethyl chloroformate-aqueous alkali, a procedure introduced by Allpress and Haworth for monosaccharides, yielded an approximately tri- substituted 0-ethoxycarbonyl sucrose (9,which on heating in uucuopolymerized, 13 A. I. Bakal, 1984, U.S. Patent 4 459 316. J. Defaye,H. Driguez, S. Poncet, R.Chambert, and M.-F. Petit-Glatron,Curbohydr. Res., 1984,130,299. lo R. S. Theobald,J. Chem. SOC.,1961,5359,5365,5370; L. Hough, J. E. Priddle, and R. S. Theobald,Adv. Curbohydr. Chem., 1960, 15, 91. C. F. Allpress and W. N. Haworth, J. Chem. Soc., 1924, 125, 1223. Hough with the elimination of diethyl carbonate and ethanol, to form thermosetting resins by the formation of a network of inter-linked carbonate bridges (6). The acid-0 0 II Et 0-OEt c=o 0 OEt . 60 . (5) . (61 + Et,CO, + EtOH sensitive glucosidic linkage of sucrose was considerably stabilized by the presence of 0-alkoxycarbonyl groups, even a small number, and penta-and octa-0-ethoxycarbonyl sucroses were not significantly hydrolysed by N-HCl.In exploiting the subtle differences in the reactivity of the eight hydroxy-groups of sucrose to produce partially substituted derivatives, it was recognized that whilst a large number of isomers are theoretically possible (Table 1) the preferential reactivity of Table 1 Number of isomers of sucrose derivatives Mono 8 Penta 56 Di 28 Hexa 28 Tri 56 Hepta 8 Tetra 70 Octa 1 the primary hydroxy-groups could simplify these apparently complex reactions. Thus etherification of sucrose with trityl chloride in pyridine 3*12afforded the 6,1’,6’-tri-0-tritylate and 6,6’-di-O-tritylate as major products with the 6,l’-and 1’,6’-di-U-tritylates as minor products, indicating that the 1’-OHis less reactive than the other primary 6-and 6’-hydroxy-groups, presumably due to its neopentyl character.A similar order of reactivity was observed on trimolar-tosylation of sucrose in pyridine, yielding after chromatography a crystalline 6,6’-di-O-tosylate (7),13,14 a mixed tri-U-tosylate 13915*16 and a little of the 2,6,1’,6‘-tetra-U-tosylate(9).17 Detailed studies revealed that the ‘tri’-fractioncontained predominantly the 6,1’,6’-tri-U-tosylate (8), but containing some 2,6,6’-tri-O-t0sylate,’~thereby suggesting l2 R. Khan, Pure Appl. Chem., 1984,56, 833. l3 R. U.Lemieux and J. P. Barrette, Can. J. Chem., 1960,38, 656. l4 C. H. Bolton, L. Hough, and R. Khan, Carbohydr. Res., 1972,21, 133. l5 P.D.Bragg and J. K. N. Jones, Can. J. Chem., 1959,37, 575. l6 D.H.Ball, F. H. Bissett, and R.C. Chalk, Carbohydr. Res., 1977,55, 149. J. M. Ballard, L. Hough, S. P. Phadnis, and A. C. Richardson, Carbohydr. Res., 1980,83, 138. The Sweeter Side of Chemistry the order of reactivity as HO-6, HO-6’ > HO-1’ > HO-2. In accord with studies in the monosaccharide field, sodium methoxide converted the 6,6’-di-O-tosylate (7) into 3,6 :3’,6’-dianhydro-~ucrose by participation of the 3- and 3’-hydroxy-groups in the intra-bimolecular substitution of the 6- and 6’-sulphonyloxy substituents. Likewise, the 6,1’,6’-tri-O-tosylate (8) afforded the 3,6 :1’,4’: 3’,6’-trianhydride (10),16318719 and not, as suggested on previous eviden~e,’~ the 3,6: 2,l’: 3’,6’- trianhydride. @*oTs _____) OTsGo+ (7) I Ts 0 (10) (91 The availability of these primary sulphonate esters enabled the selective synthesis by nucleophilic substitution reactions of a range of mono-, di-, and tri-substituted derivatives of sucrose including azido-,20 hal~geno-,~thio-, deoxy-, and unsaturated derivative^.'^.^^ Greater selectivity during sulphonation was noted using the more bulky 2,4,6-trimethylbenzene (mesitylene or ‘trimsyl’) sulphonyl chloride 25926 and 2,4,6-tri-isopropylbenzene(‘tripsyl’) sulphonyl chloride,27 which had the additional advantage that they gave the 6,1’,6’-tri-O- sulphonate esters directly, without recourse to chromatography, and in >50% yield.Later studies showed that chloro- and bromo-derivatives could be utilized as alternative substrates to the sulphonate esters, often with advantage because of their ready availability.28 Thus, selective bromination of sucrose with carbon tetrabromide-triphenylphosphine in pyridine gives in >90% yield the 6,6’-dibromide (1l), which can be used in the form of its hexa-acetate for a series of displacement reactions, in particular to achieve an original objective of bridging the Is R.Khan, Carbohydr. Rex, 1972, 22, 441. l9 N. W. Isaacs, C. H. L. Kennard, G. W. O’Donnell, and G. N. Richards, Chem. Commun., 1970, 260. 2o L. Hough and K. S. Mufti, Curbohyd. Res., 1973, 29, 291. R. Khan, K. S. Mufti, and M. R. Jenner, Curbohydr. Res., 1973, 30, 183. 22 T. Suami, T. Ikeda, S. Nishiyama, and R. Adachi, Bull. Chem. SOC. Jpn., 1975, 48,1953. 23 L. Hough, in ‘Sucrochemistry’, ed. J.L. Hickson, A.C.S. Symposium Series No. 41, 1977, p. 9. 24 R. Khan and M. R. Jenner, Carbohydr. Res., 1976, 48, 306. 25 S. E. Creasey and R. D. Guthrie, J. Chem. SOC., Perkin Trans. I, 1974, 1373. 26 L. Hough, S. P. Phadnis, and E. Tarelli, Carbohydr. Res., 1975, 44, C12. 27 R. G. Almquist and E. J. Reist, J. Carbohydr., Nucleosides, Nucleotides, 1976, 3, 261; Carbohydr. Res., 1976, 46, 33. 28 R. Khan, C. L. Bhardwaj, K. S. Mufti, and M. R. Jenner, Carbohydr. Res., 1980, 78, 185. 360 Hough 6,6’-po~itions.~~When thiourea reacted with the 6,6’-dibromide (1 1) it gave the 6,6’-dithiouronium salts, which on treatment with sodium metabisulphite were transformed into the 6,6’-disulphide (12) containing an 1 1-membered ring. Likewise, the 6,6’-dibromide (1 1) reacted with potassium 0-ethyl dithiocarbonate to give the 6,6’-sulphide (14),29 and not the expected 6,6’-bis(dithiocarb0nate): the cyclization probably proceeds via the mono-bromo mono-(dithiocarbonate) (1 3).The 10-membered ring in the crystalline 6,6’-sulphide (14) was shown by X-ray II CH/s\ UCH2 analysis 30 to have a boat: chair :chair conformation (14a) with the sulphur exo to the ring-oxygen (attached to C-1 and C-2’). In solution, ‘H-n.m.r. spectra of the 6,6’-sulphide (14a) were consistent with conformational equilibration with a boat :chair: chair conformation (14b) due to the movement of the sulphur atom from exo-to endo-po~itions.~~ An entry into the 1’,6’-and 1’-derivatives of sucrose was gained from a careful study 31 of the nucleophilic substitution of the sulphonyloxy substituents in the 29 L.V. Sincharoenkul, Ph.D. Thesis, University of London, 1981. 30 M. G. B. Drew, University of Reading, unpublished results. 31 M. R. Gurjar, Ph.D. Thesis, University of London, 1980. 361 The Sweeter Side of Chemistry 6,1’,6-tri-O-trimsylate (1 5),which with sodium benzoate progressed from the initial 6-O-benzoate (16) to the 6,6’-di-O-benzoate (17). Replacement of the trimsyloxy substituents in (16) and (17) by a variety of nucleophiles then gave 1’,6’-and 1’- derivatives of sucrose. Using this approach, Guthrie and Watters 32 synthesized 1’- chloro- 1’-deoxy-sucrose (1 8), which was not hydrolysed by the enzyme invertase, a fructofuranosidase, and was more stable to acidic hydrolysis, the rate of hydrolysis being 10 times slower than that of sucrose.An alternate route to the 1’-derivatives is (19; R = H) (20;R = S02Me) HO 32 R.D. Guthrie and J. J. Watters, Aust. J. Chem., 1980,33, 2487. Hough via the 6,6’-di-trityl ether (19), which by selective reaction affords the 1’-sulphonate ester (20). Unlike the 6,1’,6’-tri-O-tosylate (8), the 1’-tosylate (20) was transformed, somewhat surprisingly, into the 2,l’-anhydride (21),33 the ring system being of the strainless cis-decalin type. After de-tritylation this bridged derivative of sucrose was tasteless, suggesting the involvement of the 2-and/or 1’-positions in the sweetness template. The difference in the cyclization of the 1’,6’-disulphonate (8) has been attributed to the prior formation of the 3’,6’-anhydro ring (22) from the latter which brings the 4’-OH and the 1’-sulphonate into close proximity, thereby giving the 1’,4’-anhydro ring (10) in preference to the 2,l’-anhydride (21).The isolation of the 2,6,1’,6’-tetratosylate (9) suggested that if the primary hydroxy-groups are protected, as in the 6,1’,6’-tritrityl ether (23), the 2-OH would then be preferentially esterified, as confirmed by the isolation of the 2-tosylate (24).3 As anticipated, the 2,3-anhydro-~-manno-derivative(25) was obtained by the action of sodium methoxide upon the 2-tosylate (24) and subsequent ring- opening occurred exclusively at the 3-position to give after de-tritylation D-do-sucrose, a non-sweet compound, and its 3-derivatives (26).Modifications at ,OTr mo;i-.-.r & (23) (241 NaOMei ,OTr ,OTr <7 HO-2 and HO-3 can also be achieved by oxidation using both biochemical and chemical oxidants; thus 3-ketosucrose (27) 34 was isolated from the culture medium of Agrobacterium tumefuciens and subsequent reduction with sodium borohydride gave D-do-sucrose (28) and sucrose in the ratio 12: 1. D-do-Sucrose (28) was more readily obtained by oxidation of sucrose with dimethyl sulphoxide-acetic anhydride, then reduction in situ followed by fractionation on a Dowex 50 x 8 33 A. K. B. Chiu, M. K. Gurjar, L. Hough, L. V. Sincharoenkul, and A. C. Richardson, Carbohydr. Rex, 1982, 100, 247. 34 L. Hough and E.O’Brien, Carbohydr. Res., 1980, 984, 95. 363 The Sweeter Side of Chemistry (Ca2’) resin which complexes with the axial-equatorial-axial trio1 at C-2, C-3, and C-4. 2-Ketosucrose (29) 35 arises from the enolization of 3-ketosucrose (27) and reduction of both keto-derivatives with sodium borodeuteride afforded, after chromatographic separation, [2-2H]sucrose and [3-2H]sucrose respectively, thereby enabling the C-2 and C-3 signals of the ’3C-n.m.r. spectrum of sucrose36 to be confirmed. Modifications at C-4 of sucrose were achieved uia the elusive 4,6-acetal derivatives, e.g. (30), which were first prepared in 1974 by Khan,37 despite many previous attempts. Reaction of sucrose with 2,2-dimethoxypropane in N,N-dimethyl formamide (DMF) with tosic acid as catalyst, gave 2,l’: 4,6-di-O- isopropylidene sucrose (5573, another bridged derivative, and the 4,6-0-isopropylidene derivative (30).38 The latter (30) was obtained in higher yield (65%) using methyl isoprenyl ether (2-methoxypropene) from which 2,3,1‘,3’,4’,6’-hexa-0- benzyl-sucrose (3 1) was obtained and thence its 4,6-dime~ylate.~’ Selective displacements of this 4,6-sulphonate ester with fluoride anion (from tetrabutyl- ammonium fluoride), in interplay with benzoate anion, then afforded 6-deoxy-6- fluorosucrose (32), 6 deoxy-6-fluoro- and 4-deoxy-4-fluoro-galacto-sucrose(33), and 4,6-dideoxy-4,6-difluoro-sucroseand -galacto-s~crose.~~Pivaloylation of sucrose4’ is a novel and potentially valuable route to specifically blocked esters ”L.Hough and E. OBrien, Carbohydr. Res., 1981,92, 314. 36 L. Hough, S. P. Phadnis, E. Tarelli, and R. Price, Carbohydr. Res., 1976, 47, 151. ”R. Khan, Carbohydr. Res., 1974, 32, 375. R. Khan and R. S. Mufti, Carbohydr. Res., 1975, 43, 247. 39 L. Hough, A. K. M. S. Kabir, and A. C. Richardson, Carbohydr. Res., 1984, 125, 247. 40 L. Hough, A. K. M. S. Kabir, and A. C. Richardson, Carbohydr. Res., 1984, 131, 335. 41 L. Hough, M. S. Chowdhary, and A. C. Richardson, J. Chem. SOC.,Chem. Commun., 1978, 664. Hough (Scheme 1) using pivaloyl chloride (2,2-dimethylpropanoyl chloride), a reagent that was originally exploited for the synthesis of 5'-esters of nucleosides. Thus a heptapivalate (34) with only one hydroxy-group free at C-4 was isolated directly in 45% yield and then converted into galacto-sucrose (39, its 4-chloro derivative (36), and 4-ketosucrose (37).4143 The lack of sweetness in galucto-sucrose (35) is of considerable interest since it implicates the 4-position in the sweetness template.44 The hexapivalates with free hydroxy-groups at C-3 and C-3', and at C-3 and C-4' are of value since they provide an entry to modifications in the fructofuranose ring, for example via the 3',4'-epoxides.In an alternative approach, selective de- esterification of sucrose octa-acetate on an alumina column yields a hepta-acetate HO ___) HBzO Oq Ho 0.0. 0. Bzo 0-... (30) (31) HO-k$j H:o+ 0.0.0. 00.0.. (33) (32) 30'1. 35'1' 2'1' mono di tri tetra penta hexa hepta octa Scheme 42 A, K.B. Chiu, Ph.D. Thesis, University of London, 1983. 43 R.Khan, Carbohyd. Rex, 1972, 25, 232. 44 M. J. Lindley, G. G. Birch, and R. Khan, J. Sci. Food. Agric., 1976, 216, 480. The Sweeter Side of Chemistry /OPV R It+ PVO PVO““qq0Go.. PVO (34) (35; R = OH) (36;R = CI) in 9% yield, in which the 6’-hydroxy-group is free, thus making the 6’-chloride and 6’-deoxy derivative available.45 The direct replacement of hydroxy-groups by chloride, via their chlorosulphate intermediates, was first observed by Helferich in 1921,46 and extended by Jones et a1.47*48forty years later, for example by the characterization of the product from methyl a-D-glucopyranoside (38) as methyl 4,6-dichloro-4,6-dideoxy-2,3-sulpho-~-D-galactopyranoside (39) wherein the reaction was characterized by inversion of chirality at C-4 (gluco galacto).Residual chlorosulphate groups in the reaction products are readily removed by treatment of sodium iodide in methanol. Application of this multi-centred reaction to sucrose was expected to give the same pattern of derivatization in the D-glucopyranosyl ring of sucrose, together with the introduction of new substituents in the fructofuranoside ring. Initial experiments confirmed these generalizations and gave rise to a complex mixture of products 49 from which three derivatives were isolated, differing only in the furanose ring which was observed to contain a 3’,4’-epoxide (40), a 3’-ene (41), or a 1’,4’,6’-trichloride (42).” The reaction can be controlled to avoid cyclic sulphate formation by reaction at low temperatures, adjusting the ratio of sulphuryl chloride to pyridine and dilution of the reaction with chloroform.The reaction of sulphuryl chloride with sucrose was observed to give the 6’-chloride [43%; (43)] and then proceed progressively via the 6,6’-dichloride [29%; (44)] to the 4,6,6’-trichloro- [SO%; ”J. M. Ballard, L. Hough, and A. C. Richardson, Carbohydr. Res., 1974, 34, 184. 46 9. Helferich, Eer., 1921, 54, 1082. 47 J. K. N. Jones, M. 9. Perry, and J. C. Turner, Can. J. Chem., 1960, 38, 1122. 48 A. G. Cottrell, E. Buncel, and J. K. N. Jones, Can. J. Chem., 1959, 37, 1412. 49 P. D. Bragg, J. K. N. Jones, and J. C. Turner, Can. J. Chem., 1959, 37, 1412.J. M. Ballard, L. Hough, A. C. Richardson, and P. H. Fairclough, J. Chem. SOC.,Perkin Trans. 1, 1973, 1524. 366 Hough -HOT$+HO OMe OMe (38) (39) (40) (411 (45)] and 4,6,1’,6’-tetrachloro-[40%;(46)] ’derivatives of galacto-sucrose, suggesting a sequence of stereoselective reactions where the order of reactivity HO-6’ >HO-6 >HO-4 >,HO-l’,H0-4’.It is noteworthy that the 4-OH reacts more readily than the hindered but primary 1’-OH. The 6,6’-dichloride (44)can be more conveniently prepared in higher yield (>70%) by reaction of sucrose with triphenylphosphine and carbon tetrachloride in ~yridine.’~ Under more forcing L. Hough, S. P. Phadnis, and E. Tarelli, Carbohydr. Res., 1975, 44, 37. ”H. Parolis, Carbohydr.Rex, 1976, 48, 132. 53 C. K. Lee and M. R. Jenner, unpublished results. 54 A. K. M. Anisuzzaman and R. L. Whistler, Carbohydr. Res., 1980, 78, 185; 1978, 61, 511 The Sweeter Side of Chemistry conditions, sulphuryl chloride reacts with sucrose to give the 4,6,1’,4’,6’-pentachloride (47) which on treatment with base is transformed, in common with other chloro-sucroses, into an anhydro derivative, in this case a 3,6;3‘,6‘; 2,l’- trianhydro derivative (48).5 The 6,1’,6’-trichloride (49) was synthesized from the CI n I CI (47) (48) (491 (50) 6,l ,6 -trimesitylene sulphonate (15) by substitution with lithium chloride and as anticipated it reacted with sulphuryl chloride to give the 4,6,1’,6’-tetrachloride of galacto-sucrose [(46) ‘~erendipitose’].~~ Serendipity then played a part when Phadnis 56 tasted this compound for the reasons that galacto-sucrose (35) was not sweet whereas 4,6,4’,6’-tetrachloro-4,6,4’,6’-tetradeoxy-galacto-trehalosewas as Table 2 Relative sweetness of chloro-sucroses3*62 Sugar Relative Sweetness Sucrose 1 1’-Chloro-1’-deoxysucrose(18) 20 4-Chloro-4-deoxy-galacto-sucrose(36) 5 6-Chloro-6-deox ysucrose bitter 6’-Chloro-6-deoxysucrose (43) 20 4,l’-Dichloro-4,1 ‘-dideox y-galacto-sucrose 120 1 ’,6’-Dichloro-1',6'-dideox ysucrose 76 6,6’-Dichloro-6,6’-dideoxysucrose(44) not sweet 4,1’,6’-Trichloro-4,1’,6’-trideoxy-galacto-sucrose(50) 650 4,6,1’,6’-Tetrachloro-4,6,1’,6’-tetradeoxy-galacto-sucrose(46) 200 4,1’,4,6’-Tetrachloro-4,1’,4’,6’-tetradeoxy-galacto-sucrose7(59) 2 200 55 S.P. Phadnis, unpublished results. s6 L. Hough and S. P. Phadnis, Nature (London), 1976, 263,800. Hough bitter as quinine,57 and a sample of (46) required for ‘testing’ was misinterpreted as ‘tasting’. Against all predictions, the tetrachloride (46) proved to be several hundred times sweeter than sucrose 56 (Table 2). Clearly the 4-position of sucrose plays an important role in the enhancement of its sweetne~s,~ coupled with the increased lipophilicity of the molecule. This behaviour was difficult to predict since there was no precedent in carbohydrate chemistry. Indeed, modification usually results in a loss of sweetness,58 as in sucrose mono-a~etate,~~ or even bitterness as in the benzoates,60,61 and sucrose octa-acetate, a natural bitter principle of Clematis japonica, is a well known denaturant.A large number of chloro derivatives, ranging from mono- to penta-chloro substituents, were then synthesized from sucrose in order to investigate the structure-activity relationship^.^^.^^ The 4,l ,6 -trichloro-4,l ,6 -trideoxy-galacto- sucrose (50) emerged as the sweetest compound at this stage (Table 2). This trichloride (50) is synthesized 64 from the penta-acetate of the 6,l ,6 -tri-0-trityl ether (23) which on de-tritylation undergoes a 4 --+6 acetyl migration via the 4,6- orthoacetate to yield the 2,3,6,3’,4’-penta-acetate.Clearly chloro substituents at C-4, C-1 ,or C-6’ induce extra sweetness in galacto-sucrose, whereas substitution is disadvantageous at C-6, and a combination of two of the favourable substituents is synergistic and raises the sweetness by an order of magnitude, whilst the combination of all three gives intense sweetness, 600 x sucrose (Table 2).62963The replacement of the l’-OH by chloride appears to be a key factor in the intensification phenomena.The 4,1’,6’-trichloride (50) and 4,6,1’6’-tetrachloride (46) are resistant to hydrolysis by a-galactosidase and P-fructofuranosidase (‘invertase’) and more resistant to acid hydrolysis than suc~ose.~*~~,~~The favourable properties of the trichloride (50), especially its low toxicity, have singled it out for development as a high intensity sweetener that is non-nutritive, non- cariogenic, and safe for human use. Our interest in sweetness had been aroused, so we examined the sweetness property of sucrose in terms of its structure and that of related compounds. Methyl a-D-ghcopyranoside (38) is only l/lOth as sweet as sucrose whilst methyl p-D-fructofuranoside is tasteless, hence it was logical to suggest that hydroxy-groups on both the fructose and glucose components of sucrose act in harmony to initiate the sweet response.62 The ring oxygens can be discounted since pseudo-glucose (51) and pseudo-fructose 66 are equisweet with glucose and fructose respectively.Sweetness is induced by a wide variety of chemical structures (Table 3) but 5’ G. G. Birch, Olfaction and Taste VI, 1977, 27. 58 G. G. Birch and C.K. Lee, J. Food. Sci., 1976,41, 1403. 59 0.K. Konenko and I. L. Kestenbaum, J. Appl. Chem., 1961, 11, 7. 6o D. M. Clode, N. A. Laurie, D. McHale, and J. B. Sheridan, Curbohydr. Res., 1985, 139, 147, 161. 61 D. M. Clode, D. McHale, J. B. Sheridan, G. G. Birch, and E. B. Rathbone, Carbohydr. Rex, 1985, 139, 141. 62 L. Hough and R. Khan, Trends Biochem. Sci., 1978, 3, 61. L. Hough, S. P. Phadnis, R. Khan, and M. R. Jenner, 1979, British Patents 1543 167; 1 543 168. 64 P. H. Fairclough, L. Hough, and A. C. Richardson, Carbohydr. Res., 1975,40, 285. 65 T. Suami, S. Ogawa, and T. Toyokuni, Chem. Lett. (Japan), 1983, 611. 66 T. Suami, S. Ogawa, M. Takata, K. Yasuda, A. Suga, K. Takei, and Y. Uematsu, Chem. Lett. (Japan), 1985. 369 The Sweeter Side of Chemistry A-H---O B---H-N Swcet OH co mpound \ Taste bud protein (51) (52) 4-n h 71 0 I H I' I I +-" Shallenberger and Acree ''noted a common feature of two electronegative atoms, A and B, separated by 2.5-4.0& with an hydrogen atom covalently linked to A, thus giving an AH,B system (52).They postulated an interaction of the AH,B system with a similar system (-NH-CO-) on the proteinaceous, cell membrane receptor of the tongue. Application of this theory to sucrose, coupled with the behaviour of the chloro-sucroses led to the suggestion that the AH,B is situated at the 2-OH and the oxygen of the 1'-OH respectively (53), and the latter can be replaced by a chloro group with intensification of the sweet response.62 The AH,B theory was clearly inadequate since many organic compounds with this requirement were not sweet.In 1976, Kier 68 extended the theory from a study of a series sweet l-alkoxy-2-amino-4-nitrobenzenesby introducing the concept of a binding site X, that is hydrophobic (lipophilic) in nature and located 3.5A and A and 5.5 A from B to give an AH,B,X triangle. Two such triangles were recognized in the sweet 4,l ,6 -trichloride (50),involving lipophilic groups on the upper face of the molecule at the axial C-4 and C-6' respectively, which bind the molecule to the receptor [(54) and (55)]. The adverse effect of the 6-chloro group could be due '* to its competition for the dispersive locking site on the receptor resulting in a misfit of the AH,B at C-2,C-1 .It is significant that the immunoassay for the sweet protein thaumatin ('Talin') also reponds to other high intensity sweeteners, including the chloro-sucroses, and a direct relationship was observed between the immunoassay 67 R. S. Shallenberger and T. E. Acree, Nature (London), 1976, 216, 480. '* L. B. Kier, J. Pharm. Sci., 1976, 61, 1394. Hough and the sweetness response, thereby suggesting a common gl~cophore.~~ The AH,B,X tripartite theory will undoubtedly be further refined, taking into account a trio of physical parameters, lipophilicity, electron distribution, and molecular conformation in exploring the structure-activity relationships in conjunction with the associated neuro-physiological mechanisms. Table 3 Relative sweetness of organic substances (sucrose = 1) Cyclamate (sodium cyclohexylsulphamate) 30-80 Glycyrrhizin 50 L-Aspartyl-L-phenylalaninemethyl 100-200 ester (Aspartame) Acesulfam-K 150 6-Chlorosaccharin 100-3 50 Sodium saccharin 200-700 Stevioside 300 4,1’,6’-trichloro-galacto-sucrose(50) 650 Neohesperidin dihydrochalcone 2 000 l-n-Propoxy-2-amino-4-nitrobenzene 4000 Thaumatin (Talin) (A13’) 34 OOO D-Tryptophan 35 6-Chloro-~-tryptophan 1000 Fenchyl derivative of aspartame 25 000 (methyl fenchyl L-aspartylaminomalonate) ASPARTAME (R = CH, Ph) ACESULFAM -K 0 Na+oNH‘So;L SACCHARIN CYCLA MATE NEOHESPERIDIN DlHY DROCHALCONE 37 1 The Sweeter Side of Chemistry Derivatization at C-3’ and C-4’ of sucrose has led to an interesting group of compounds. Guthrie et aL7’ synthesized the 3’,4’-lyxo-epoxide (56) directly from sucrose in 42% yield by the agency of triphenylphosphine-diethylazodicarboxylate and incorporating acetic acid to prevent 3,6- and 1’,4’-anhydro formation.When the primary positions were protected, as in the 6,1’,6’-tri-O-tritylate (23) and the 4,6;2,lr-0-isopropylidenederivative, the corresponding 3’,4’-lyxo-epoxides were obtained in > 80% yield.’ * Ring-opening of these epoxides with various nucleophiles occurred exclusively at the 4’-position (57), thus reverting back to the fructofuranose configuration.12 When the 3’,4’-lyxo-epoxide (58) derived from 4,1’,6’-trichloro-4,1’,6’-trideoxy-galacto-sucrose(50), was opened with lithium chloride in DMF, it gave the 4,1’,4’,6’-tetrachloride(59), which on tasting proved to 0CI (54) r.Qo+ 0 - l-abob & (56) (57) Qo* Icqo+cl CI (60) (611 69 C. A. M. Hough and J. A. Edwardson, Nature (London), 1978,271, 381. ’O R. D. Guthrie, I. D. Jenkins, S. Thang, and R. Yamaski, Curbohydr. Rex, 1980,85, C5; 1983, 121, 109. Hough 2200 times sweeter than sucrose (Table 2), the introduction of a 4’-chloro group having quadrupled the sweetness of its precursor (50).7’ The 4,1’,4’,6’-tetrachloride (59) was also isolated7’ as a minor component from the reaction of sucrose 6- acetate (60) with sulphuryl chloride, the major component being the sorbo-isomer (61), and they probably arise from the opening of the corresponding 3’,4‘-Iyxo- and ribo-epoxides.Another tetrachloride, namely the 2,6,1‘,6‘-tetrachloro-2,6,1‘,6-tetra-deoxy-rnanno-sucrose (63) was synthesized72 from the 6,6’-dichloro-2,1’-iso-propylidene derivative (62), and it proved to be as bitter as quinine, thus supporting the view that an equatorial 2-OH is essential for sweetness. The fructofuranose ring of sucrose can be expanded to a pyranoside by selective oxidation with lead tetra-acetate73 to give the dialdehyde (64) which can then be cyclized by condensation with nitromethane to give the 4’-nitro-4‘-deoxy-P- D-glucoheptulopyranoside (65), and thence to 4’-amino and 4’-acetamido derivative^,^^.^ Extension of these reactions to the tetra-aldehyde (66), produced by oxidation of sucrose with periodate, gave the 3-nitro-3-deoxy-a-~-glucopy-ranosyl 4 -nitro-4’-deoxy-P-~-heptulopyranoside(67) as the major product.75 OH C CHO CHO (66) (67) C.K. Lee, 1982, U.K. Patent 2 088 855A. 72 R. A. Khan and M. R. Jenner, 1980, U.K. Patent 2 037 561A. 73 A. K. Mitra and A. S. Perlin, Can. J. Chem., 1959, 37, 2047. 74 H. H. Baer and A. Ahammed, Can. J. Chem., 1966,44,2893. 75 L. Hough, K. J. Hale, and A. C. Richardson, unpublished results. 373 The Sweeter Side of Chemistry Acknowledgements. I am indebted to my research colleagues, many of whom are cited in the references, who in the context of this lecture carried out the skilled experimental work and interpretation of results, and I am especially grateful to Drs.A. C. Richardson and R. Khan for their continued interest and enthusiasm in these sucrochemical studies. These researches have been supported continuously since 1956, initially by the International Sugar Research Foundation (now the World Sugar Research Organization) and currently by Tate and Lyle Ltd., and I am grateful for their sustained interest in the fundamental chemistry of this ubiquitous carbohydrate.
ISSN:0306-0012
DOI:10.1039/CS9851400357
出版商:RSC
年代:1985
数据来源: RSC
|
6. |
Guanidine derivatives acting at histaminergic receptors |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 375-398
G. J. Durant,
Preview
|
PDF (1113KB)
|
|
摘要:
Guanidine Derivatives Acting at Histaminergic Receptors By G. J. Durant SMITH KLINE & FRENCH RESEARCH LIMITED, THE FRYTHE, WELWYN, HERTFORDSHIRE 1 Introduction This review concerns compounds that are derived from the two biologically important organic bases, imidazole and guanidine. These bases are contained in the essential amino-acids histidine and arginine, respectively, and in many naturally occurring polypeptides derived therefrom. Guanidine itself is one of the strongest known organic bases (pKa 13.6) and at a physiological pH of 7.4 it will exist almost exclusively as the cationic species [(la), Z = H, Figure 13. Imidazole is a tautomeric base with pKa = 7.03; for 4-substituted imidazoles significant populations of cationic (2a) and tautomeric neutral species (2b and 2c) are likely to be present at a physiological pH of 7.4 (Figure 2).Imidazole is the strongest base Figure 1 Guanidine species equilibria that can exist in the free base form in the pH range at which most enzymes act. This property may contribute to the unique properties of histidine units at the active site of most enzymes where they can act as either proton donor or acceptor. Compounds containing guanidine groups in their structure have been shown to exert a variety of physiological or pharmacological effects. 374 For example, the S.J. Angyal and W. K. Warburton, J. Chem. SOC.,1951,2492. A. H. M. Kirby and A. Neuberger, Biochem. J., 1938, 32, 1146. G. J. Durant, A. M. Roe, and A. L. Green, Prog. Med.Chem., 1970,7(1), 124. G. J. Durant, M. E. Parsons, and J. W. Black, J. Med. Chem., 1975, 18, 830. Guanidine Derivatives Acting at Histaminergic Receptors naturally occurring guanidine derivative creatine is important to muscle activity. Some guanidines affect the cardiovascular system, and many guanidines possess antimicrobial activity. Many imidazoles also exert important biological actions, particularly histamine (3), which is the decarboxylated product from histidine. “Y dR FtR N NH HNdNv (2b) (2c) Figure 2 Imidazole species equilibria + (3) This review primarily concerns molecules containing both imidazole and guanidine or closely related groups in their structures and which exert pharmacological effects at histaminergic receptors.2 The Development of Histamine H,-Receptor Antagonists Histamine stimulates contraction of smooth muscle from various organs such as the gut and bronchi and this effect can be suppressed by low concentrations of conventional antihistaminic drugs. These latter compounds, typified by mepyr- amine (4), comprise a group of lipophilic tertiary amines that were originally discovered in the 1940s and subsequently produced a large family of histamine antagonists. This field has been reviewed recently. 576 The pharmacological receptors involved in these mepyramine-sensitive responses have been defined as histamine H ,-receptors.’ Histamine also stimulates secretion of acid by the stomach and increases the heart rate. The failure of conventional antihistamines to C.R. Ganellin in ‘Pharmacology of Histamine Receptors’, ed. C. R. Ganellin and M. E. Parsons, J. Wright and Sons, London 1982. G. J. Durant, C. R. Ganellin, R. Griffiths, C. A. Harvey, D. A. A. Owen, and G. S. Sach, in ‘Frontiers in Histamine Research’, ed. C. R. Ganellin and J. C. Schwartz, Pergamon Press, Oxford, 1985.’A. S. F. Ash and H. 0.Schild, Br. J. Pharmacol., 1966,21, 427. Durant block these effects of histamine led to the initiation of a research programme at Smith Kline & French Laboratories (Welwyn) in which an antagonist of these non- H ,-receptor (subsequently 'termed H,-receptor) effects of histamine was sought. The discovery of an effective inhibitor of histamine-stimulated gastric acid secretion was envisaged as a potential means of controlling the hypersecretion of gastric acid in patients.The structure of histamine (3) was used as a chemical starting point with the objective of designing a structure that would bind more strongly than histamine but not trigger off the usual physiological response (k,in the pharmacological terminology of receptor occupancy the~ry,~ to reduce efficacy but to retain or enhance affinity). After four years of research, the guanidine derivative of histamine, W-guanylhistamine (9,provided a much sought-after lead following the discovery that it was a weakly active partial agonist at histamine H,-re~eptors.~ In extremely high doses, inhibition of histamine-stimulated gastric acid secretion in the anaesthetized rat was observed (IDso, 800 pmole/kg, i.~.)and high concentrations of N"-guanylhistamine antagonized histamine-induced contractions on guinea-pig right atrium in uitro (Table 1).As an agonist, W-guanylhistamine elicits weak submaximal responses in both of these histamine H,-receptor sy~tems.~This guanidine derivative was originally selected for synthesis based on literature analogies for guanidine structures exerting inhibitory effects at enzymes or receptor sites which are normally responsive to amine~.~ It appeared conceivable that these precedents for guanidines exerting opposing effects to amines could be related to the formation of cyclic hydrogen-bonded structures between guanidinium ions and anionic counter ions (6).Such complexes have been observed in crystal structures of some guanidinium salts with oxyanions. The higher homologue, 3-(imidazol-4- y1)propylguanidine [(7), Table 13was slightly more potent as an antagonist but was also a partial agonist." It appeared likely that the agonist activity of these guanidinium compounds was related to the high basicity of guanidine and that the presence of the cationic charge on these molecules might enable them to mimic histamine under some circumstances. Analogous thioureas e.g. SK&F 91581 (8), which are polar but neutral structures with pK, 'Y 0, were found to be devoid of agonist activity, but antagonist activity was weak (Table 1). However, further chain extension of the thiourea derivative led to an increase in antagonist activity and afforded the first fully characterized histamine H,-receptor antagonist burimamide.' Burimamide (9) was devoid of agonist activity and was shown to be a selective antagonist of histamine on non-H tissue systems (Table l), thereby defining histamine H,-receptors and characterizing burimamide as a histamine H,- receptor antagonist.' The analogous thiourea metiamide (10) is more potent as an antagonist than burimamide (Table 1) and this compound was initially selected to explore the clinical potential of H,-receptor antagonists.' 1*12 Metiamide was * J.W. Black, W. A. M. Duncan, G. J. Durant, C. R. Ganellin, and M. E. Parsons, Nature (London), 1972, 236, 385. R. P. Stephenson, Br. J. Pharmacol., 1956,11, 379.lo M. E. Parsons, R. C. Blakemore, G. J. Durant, C. R. Ganellin, and A. C. Rasmussen, Agents Actions, 1975. 3, 133. l1 J. W. Black, G. J. Durant, J. C. Emmett, and C. R. Ganellin, Nature (London), 1974, 248, 65. l2 J. W. Black, W. A. M. Duncan, J. C. Emmett, C. R. Ganellin, T. Hesselbo, M. E. Parsons, and G. J. Wyllie, Agents Actions, 1973, 3, 133. Guanidine Derivatives Acting at Histaminergic Receptors Table 1 Some key compounds in the development of the H2-receptor antagonist cimetidine H2-Receptor Antagonist Activity Compound (No.) Structure in vitroa PA2 in vivob IDSO ACHZCHZNHCNHZ (Pmo~Ikg) W-Guanylhistamine (5) HNeN II+ NHZ 3.9 800 SK&F 91486 (7) CHZCHz CH 2NH CN H, II+ NH, 4HNvN 4.65 100 SK&F 91581 (8) CH,CHzCH2NHCNHMe II S d eNHN 3.8 C Burimamide (9) CH2CH2CHz CH2NHCNHMe I1 S/=( HN*N 5.11 6.1 C HzSCH ,CH,N H CNH Me Metiamide (10) HNvN II 6.04 1.6S CH, SCH, C H,NH C NH Me Guanidine isostere (1 1) II 4.80 12 HNVN + NHZ C H, SCH ,C HZNHCNHMe Cimetidine (12) II 6.10 1.4 HN+N NC N a pA2 = -log KB where KS is the dissociation constant ( x 10-6M) determined in vitro on guinea-pig atrium against histamine stimulation.Statistical limits and slopes of regressions are omitted. * IDso is the intravenous dose required to produce 50% inhibition of near maximal histamine-stimulated gastric acid secretion in anaesthetized rats using a lumen-perfused preparation (re5 8). No antagonism observed up to an intravenous dose of 256pmole/kg shown to be highly effective clinically in reducing hypersecretion of gastric acid and proved to be of therapeutic value in duodenal ulcer disease.However, the occurrence of a reversible granulocytopenia in a small number of patients limited the use of metiamide and the possibility existed that this effect was related to the presence of the thiourea group. Fortunately, non-thiourea structures had already been c~nsidered,~ including guanidines. The guanidine analogue (1 1) (see Table 3914 l3 R. W. Brimblecombe, W. A. M. Duncan,G. J. Durant, J. C. Emmett,C. R. Ganellin,and M. E. Parsons,J. Int. Med. Res., 1975, 3, 86. G.J. Durant, J. C. Emmett, C. R. Ganellin, P. D. Miles, M. E. Parsons, H. D. Prain, and G. R. White, J. Med. Chern., 1975, 20, 901.Durant 1) was found to be an antagonist but was approximately an order of magnitude less potent that metiamide.I4 Interestingly, this compound differs from the shorter- chain guanidine structures in that it is not a partial agonist. It appears to be a competitive antagonist but of lower potency than metiamide (Table 1). Since guanidine is considerably more basic than thiourea and will exist overwhelmingly as the protonated species at physiological pH, we considered whether activity could be increased by reducing the basicity of the guanidine. HI + H’ Guanidine basicity is highly sensitive to substituent effects. Electron withdrawing substituents increase the acidity of the proton on the adjacent nitrogen atom and stabilize the guanidine base as the imino tautomer [( 1b) Figure 13.Charton” has demonstrated a correlation between guanidine pKa and the inductive substituent constant q.The strongly electron-withdrawing cyano and nitro groups reduce pK, by over 14 pK, units to around zero and approximately to the pKa of thiourea. These guanidine substituents were found to increase antagonist activity and the cyanoguanidine analogue (12) is at least as potent as metiamide as an H,-receptor antagonist I3,l4 (Table 1). This compound cimetidine (Tagamet) was selected for development and was subsequently launched on to the market in the U.K. in 1976, and soon afterwards in most other countries in the world. Cimetidine proved to be a highly effective and successful drug therapy in peptic ulcer disease in many millions of patients.16 Furthermore, cimetidine is free from the side-effect of granulocytopenia that limited the use of metiamide.C H,SC H,C H +.IH C N H Me II HN NCN *N The importance of guanidine structures in the discovery and development of H,- receptor antagonists and cimetidine is evident from Table 1, in which are listed the structures and activities of some of the key compounds synthesized en route to cimetidine. l5 M. Charton, J. Org. Chem., 1965,30,969. l6 For example ‘Cimetidine in the ~OS’,ed. J. H. Baron, Churchill Livingstone, Edinburgh, 1981. Guanidine Derivatives Acting at Histaminergic Receptors The similarity of cimetidine and metiamide in their activity as H,-receptor antagonists was also observed in other pairs of imidazole-derived cyanoguanidines and thioureas and these groups provide an interesting example of chemical equivalence in biologically active molecules l4 (bioisosterism).Some physico- chemical properties of cyanoguanidine and thiourea are compared in Table 2 and the similarity in their acid-base characteristics, polarity, lipophilicity, and geometry is apparent. The crystal structures of cimetidine and metiamide, are near- identical and both form 10-membered intramolecularly hydrogen-bonded ring structures.” It should be added that the bioisosterism of cyanoguanidine and thiourea is not universal and for several other biological actions these groups are not equivalent. 3 Histamine Antagonists Based on Isocytosine Structures Further isosteres of thiourea or cyanoguanidine have been utilized in novel H,- receptor antagonist structures derived from metiamide or cimetidine.One example is the nitrodiaminoethene analogue (13) (see Table 3) which is approximately equipotent as an H,-receptor antagonist. In these carbon isosteres of guanidine it appears that the nitro substituent is essential to act inductively as a powerful electron-withdrawing group and also mesomerically in order to stabilize the negative charge on an adjacent carbon atom. The derivation and utilization of the nitrodiaminoethene group in H,-receptor antagonists has been des~ribed.’~.’~ The properties of this group are compared with those of cyanoguanidine and thiourea in Table 2.This group was subsequently utilized by research workers from Glaxo Laboratories in the development of the drug, ranitidine.” Isocytosine (2-aminopyrimidin-4-one) may be regarded as a guanidine derivative in which substituents on two nitrogen atoms are incorporated into a 6- membered ring. Isocytosine was originally selected for synthesis since it is a planar conformationally restricted neutral guanidine structure with additional sites for potentially increasing affinity by ring substitution. Isocytosine is slightly more basic and more acidic than thiourea or cyanoguanidine but its neutral form should predominate at physiological pH. Properties of isocytosine are compared with those of thiourea and cyanoguanidine in Table 2.The isocytosine analogue (14) is less active as an H,-receptor antagonist than the corresponding thiourea or cyanoguanidine, namely metiamide or cimetidine (Table 3). However, 5-substituted isocytosines have provided a family of structures which are highly effective as histamine receptor antagonists. Examples include oxmetidine (1 5) (see Table 3) which contains a 3,4-methylenedioxybenzylsubstituent in the 5-position of the isocytosine ring and SK&F 93241 (16) (see Table 4) which contains a 6- l7 K. Prout and C. R. Ganellin in ‘Structural Studies on Molecules of Biological Interest’, ed. G. Dodson, J. P. Glusker, and D. Sagre, Clarendon Press, Oxford, 1981. C. R. Ganellin, J. Med. Chem., 1981, 24,913. l9 G. J. Durant, T. H. Brown, J.C. Emmett, C. R. Ganellin, H. D. Prain, and R. C. Young, in ‘The Chemical Regulation of Biological Mechanisms’, ed. A. M. Creighton and S. Turner, R. S. C. Special Publication, No. 42, The Royal Society of Chemistry, London, 1982, p. 27.*’ J. Bradshaw, M. E. Butcher, S. W. Clitherow, M. D. Dowle, R. Hayes, D. B. Judd, J. M. McKinnon, and B. J. Price, in ref: 19, p. 45. Table 2 Comparison of some physicochemical properties of thiourea, cyanoguanidine, 1,ldiamino-2-nitroethene, and isocytosine * 0 thiourea cyano-diamino-isocytosine guanidine nitroethene Proton Dissociation pK, (acid, 25 "C) NH2 15 14 9.6 NHMe 14 pK, (base, 25 "C) NH2 -1.2 -0.4 4.0 NHMe 2.7 Polarity dipole moment NH2 4.89 8.16 P(DebYe) NMe2 7.64 partition (octanol :water) log P (37 "C) NH2 -1.05 -1.15 -0.97 NHMe -0.24 -0.40 -1.28 -0.55 Geometry C-N length NH2 1.34 1.33 1.321; 1.315b 1.32 (1.36) N-C-N angle (degrees) NH2 119 120 119.4 119 restricted bond rotation, AG NHMe 11.8 12.4 (kcal/mole) NMez 9.2 11.8 'See reference 19 for original literature references.* Data for ranitidine hydrogen oxalate (B. Kojic-Prodic, Z. Ruzic-Tolos, and R. Toso, Actu w Crystallogr., Sect. B, 1982, 38, 1837.) Guanidine Derivatives Acting at Histaminergic Receptors Table 3 H2-Receptor antagonist potencies of metiamide, cimetidine, diaminonitroethene, and isocytosine analogues M e H CH2SCH2CHzY HNdN H2-Receptor An tag on ist Activity in vitroa in vivob Compound Y PA2 IDso No.(vmollkg) ; 6.04 1.610 NHCNHMe NCN 12 II 6.10 1.4 NHCNHMe CHNO, 5.85 1.o13 II L NHCNHMe 14 5.13 8.4 H 15 7.69 0.09 n+b See footnotes to Table 1 methyl-3-picolyl substituent in this position. Oxmetidine is about sixteen times more potent than cimetidine as an H,-receptor antagonist and has been shown to be effective in patients with peptic ulcer disease.21*22 SK&F 93479 (17) (see Table 4), which contains a 6-methyl-3-picolyl substituent in the 5-position of the 21 R. C. Blakemore, T. H. Brown, G. J. Durant, J. C. Emmett, C. R. Ganellin, M. E. Parsons, and A. C. Rasmussen, Brit. J. Pharmacol., 1980, 70, 105P. 22 R. Corinaldesi, A. Galassi, C. Borghi, R. Pasquali, M. Miglidi, T. Sacco, and L. Barbara, Hepato-Gastroenterology.1981, 28 (6), 319. 382 Durant isocytosine ring and the 5-dimethylaminomethylfuran-2-ylring in place of imidazole, is a highly potent and selective H,-receptor antagonist in animals and man and has an extended duration of acti~n.~~.,~ MewNC H 2SC H,C H,N HCN HMe II NCN The 3-methylpyrid-2-yl analogue (1 8) of cimetidine is comparable in potency as an H,-receptor antagonist in vitro (atrial PA, 6.0). In isocytosine structures, substituting pyridine for imidazole also leads to a retention of H,-receptor antagonist activity but additionally there is an increase in H,-receptor antagonist activity. Thus, whereas oxmetidine and its picolyl isocytosine analogue (16) (see Table 4) possess only weak H,-receptor antagonist activity, analogues in which the imidazole ring is replaced by pyridine are considerably more active as HI-receptor antagonists.In these structures (19), it has been demonstrated by G. S. Sach and co- workers of these laboratories 25 that both H,- and H,-receptor antagonist activity are dependent upon the bulk of the substituent, R(3), adjacent to the side chain and that these activities have different optima with respect to the ‘Verloop’ steric parameter 26 for R(3). The two activities are nearly in balance where R(3) = OMe, and with this compound [SK&F 93319 (20)],combined antagonism at H, and H, receptors in uiuo has been demon~trated.,~ SK&F 93319 has been studied in man and may have therapeutic utility in conditions, including some inflammatory skin diseases, which require simultaneous antagonism of histamine at H, and H, receptors.27 A particularly interesting finding with SK&F 93319 was its negligible ability to penetrate into the CNS in animals.27 This observation led to a research programme directed to the design of a selective H,-receptor antagonist free from the centrally mediated side-effects which characterized previous compounds in this therapeutic class.For structures (19) it has been shown25 that maximum HI-receptor antagonist activity together with the greatest separation from H,-antagonism occurs where R(3) = Br, Me, or NH2. Further studies showed that activity was also strongly influenced by 5-pyridyl substituents, R(5), in 3,5-disubstituted compounds (19) which enabled HI-receptor antagonist activity to be raised and H,-receptor antagonist activity to be reduced.QSAR analysis indicated the 23 R. C. Blakemore, T. H. Brown, G. J. Durant, C. R. Ganellin, M. E. Parsons,A. C. Rasmussen, and D. A. Rawlings, Br. J. Pharrnacol., 1981, 74, 200P. 24 T. Gledhill, J. G. Mills, A. Clancy, M. Buck, R. H. Hunt, and W. L. Burland, Gut, 1982, 23, A455 25 R. C. Blakemore, T. H. Brown, D. G. Cooper, G. J. Durant, C. R. Ganellin, R. J. Ife, M. E. Parsons,A. C. Rasmussen, and G. S. Sach, Poster Presentation (P7), 2nd SCI/RSC Medicinal Chemistry Symposium, Cambridge, 1983. 26 A. Verloop, W. Hoogenstraaten, and J. Tipker, Drug Design, 1976, 7, 165. ”C. R. Ganellin, R. C. Blakemore, T. H. Brown, D. G. Cooper, G.J. Durant, C. A. Harvey, R. J. Ife, D. A. A. Owen, M. E. Parsons, A. C. Rasmussen, and G. S. Sach, New England and Regional Allergy Proceedings, 1985, in press. 383 Guanidine Derivatives Acting at Histaminergic Receptors 0 H2 +fCII R(3 Me (19) importance of a steric factor for R(3) and electronic factors for R(3) and R(5).28 With SK&F 93944 [(19) R(3) = Me, R(5) = Br], H,-receptor antagonist activity has been increased over the combined antagonist SK&F 93319 by two orders of magnitude and H,-receptor antagonist activity has been reduced by about two orders of magnitude [(21) Table 4].28*29 SK&F 93944 has a potency at least equal to that of the conventional H,-receptor antagonist mepyramine in vitro on guinea- pig ileum and in uivo in antagonizing histamine-induced bronchconstriction in guinea-pigs and also in other in vivo assays for HI-receptor antagonist Negligible penetration of labelled SK&F 93944 into the central nervous system of anaesthetized male rats has been dem0nstrated.j’ In its chemical properties, SK&F 93944 differs from most previously described H ,-receptor antagonists in lacking a tertiary amino-group and being of markedly reduced basicity.28 SK&F 93944 is a completely novel H,-receptor antagonist which offers the prospects of being a truly non-sedative antihistaminic drug and clinical studies are in progress.4 Guanidine Structures Containing Two Heteroarylalkyl Substituents A. Histamine H,-Receptor Antagonists.-The imidazole-containing guanidinium structures described earlier in this review appear to exhibit relatively low affinity for H2-receptors either as partial agonists or as antagonists (Table 1).The guanidinium analogue (11) of cimetidine is an antagonist with PA, = 4.8, on guinea-pig atrium.Additional examples, listed in Table 5, include the des-methyl imidazol-4-yl analogues (22) and (23) and thiazol-2-yl analogues which exhibit similar activity to corresponding imidazoles in the guanidinium structure (24) and in the more active cyanoguanidine (25). In contrast, the 3-(thiazol-Z-yl)-propyl derivatives (26) and (27) are less active. Structures that contain two imidazolyl-(or thiazoly1)-alkyl substituents are considerably more potent as H,-receptor antagonists (Table 6), and the increase in activity due to the second substituent is particularly pronounced for the guanidinium compounds.The symmetrically substituted guanidinium derivative containing two ‘cimetidine side-chain’ sustituents (28), which has PA, = 28 D. G. Cooper, G. J. Durant, C. R. Ganellin, C. A. Harvey, M. L. Meeson, D. A. A. Owen, G. S. Sach, and M. A. Wilnynska, Proceedings of VIIIth International Symposium on Medicinal Chemistry, Vol. 2, Swedish Pharmaceutical Press, Stockholm, 1985, p. 198. 29 G. J. Durant, C. R. Ganellin,R. Griffiths, C. A. Harvey, R. J. Ife, D. A. A. Owen, M. E. Parsons, and G. S. Sach, Br. J. Pharmacol., 1984, 82, 232P. 30 E. A. Brown, R. Griffiths, C. A. Harvey, and D. A. A. Owen, Br. J. Pharmarof., 1985, submitted for publication. 31 E.A. Brown, C. R.Calcutt, R. Griffiths, B. Jackson, B. K. Leigh, D. A. A. Owen, and I. R. Smith, unpublished results. Durant Table 4 HI and H2-Receptor antagonist potencies of some isocytosine derivatives 0 I1 NN‘V CH2wN II Het CH,XC H,CH,N HC,Ny UMe H Antagonist Activity in vitro Compound Het X SK&F pA2 (Hi)” pA2 (H2)b No. No. S 93241 5.49 7.05 17 Me2NCH2as 93479 approx. 4.2‘ 7.78 ’ 20 93319 7.77= 7.49 21 93944 9.55f approx. 5.9’ pA2 = -log KB where KB is the dissociation constant determined against histamine stimulation on guinea-pig ileum (HI) and guinea-pig atrium (Hz)(ref: 8). R. C. Blakemore and M. E. Parsons (SK&F Research Ltd.) -unpublished results.Ref: 23. Ref:27. f Re$ 29, 30. 6.7 on guinea-pig atrium, demonstrates an activity increase of nearly two orders of magnitude at H,-receptors due to the introduction of a second heteroarylalkyl substituent in place of methyl 32733 [cf: (1 l)]. The thiazole-containing substituents also show an increase in potency and in the guanidinium structures (32) and (36) the 3-(thiazol-2-yl)propyl substituent appears to lead to an activity increase, compared with methyl [cf. (1 1 and 26)] similar in magnitude to that of a ‘cimetidine side-chain’ substituent. The magnitude of the increase due to the second heteroarylakyl substituent is generally less marked in the neutral cyanoguanidines. However, it is interesting to compare cyanoguanidines (33) with (31) in which the 3- (thiazol-2-y1)propyl substituent appears to be more effective than the analogous longer chain thioether substituent, although the reverse is true in the N-methylcyanoguanidines (25) and (27) (Table 5).Thus, di-heteroarylalkylguanidines are generally more potent as H,-receptor antagonists than monoheteroarylalkyl- guanidines and the results in Tables 5 and 6 are consistent with the presence of an additional binding site, or accessory binding area, at the H, receptor that can ” C. R. Ganellin, G. J. Durant, J. C. Emmett, D. W. Hills, R. J. Ife, P. D. Miles, and M. E. Parsons, Proceedings of VIIIth International Meeting on Medicinal Chemistry, Uppsala, Sweden, 1985. 33 G. J. Durant and C. R. Ganellin, British Patent, 1431 589, 1976.Guan idine Derivatives A cting at Histam inergic Receptors Table 5 Hz-Receptor antagonist potencies of mono-heteroaryIaIkyIguanidines NZ II Het -X-NHCNHMe Compound Het X Z No. H 4.8 CHzSCHzCHz CN 6.1 1.4 H 4.4 76 HNGN(23) CN 5.9 5.0 (24) H 5.O 8.5 CHzSCHzCHz (25) CN 6.1 2.6 H 4.2 (+ve)f CHzCHzCHz (27) CN <3.3 ( +ve)E *.* See footnotes to Table 1. Weak inhibitory activity observed at doses of 54 pmoles/kg and higher accommodate a second heteroarylalkyl substituent. The magnitude of the increased potency for this substituent in the guanidinium structures indicates the possibility that electrostatic interactions may be involved in the increased affinity of these molecules for the H2-receptor.However, the favourable influence of the 3-(thiazol-2-y1)propyl substituent in the cyanoguanidine (33) indicates that an accessory binding area may also favour interactions of neutral ligands. Questions concerning the presence of additional binding sites at H2-receptors, or even to the proximity of receptor sites, also seem pertinent to the activity of the polymethylene bis-guanidines (Table 7). Linking two of the imidazolylalkyl- substituted guanidinium structures with a polymethylene chain via the two guanidinium groups afforded the series of compounds (38) to (46) and leads to an enhancement in antagonist potency that is dependant upon the length of the bridging hai in.^^.^^ Activity appears to be optimized with a bridge of eight carbon atoms, and the octamethylene guanidinium analogue (43) is approximately three 34 G.J. Durant and C. R. Ganellin, British Patent, 1 493 931, 1977. Durant orders of magnitude more potent than the corresponding monoguanidinium structure (11) (Table 5). An enhancement in potency is also observed with corresponding thiazolyl-substituted structures [compare (47)with (24)], although the magnitude of the effect is less striking than for the corresponding imidazoles. The enhancement in potency which accompanies bridging of these guanidine structures appears to require charged guanidine groups since the corresponding bis- N-cyanoguanidine (48) does not show any increase in activity over the corresponding mono-N-cyanoguanidine [cimetidine (12)].These observations are suggestive of differences in the mode of interaction of bis-guanidinium compounds and N-cyanoguanidines with H,-receptors. The requirement for the charged guanidinium groups in this activity enhancement may indicate that mutual charge repulsion has a beneficial conformational effect on the linking side-chain that is absent in the neutral bis-N-cyanoguanidines. Comparison of the activity of (43) with (49),(50),and (51)(Table 7) suggests that both of the ‘cimetidine side-chains’ and both guanidinium groups are involved in specific interactions with the histamine H,-receptor, and also that the second guanidine group is not functioning merely as an additional cationic head. B. Histamine H,-Receptor Agonists.-3-(Imidazol-4-yl)propylguanidine (7) and its N-methyl derivative (52)(Table 8) are both weakly active partial agonists at H,- receptors.Structural modifications, analogous to those described above for the antagonist structure (ll), lead to a marked increase in potency at histamine H,- receptors (Table 8). Bis-3-(imidazol-4-yl)propylguanidine(53) is about 25 times more potent than (52)as an H,-receptor agonist on guinea-pig atrium and the octamethylene bis-guanidine (54)is more than 150 times as potent. The latter compound retains the partial agonist character of (52) in eliciting sub-maximal responses on guinea-pig atrium which enabled it to be tested as an antagonist of histamine in this H,-receptor preparation. The PA, value (7.7) for this compound indicates that there is an affinity increase at H,-receptors of nearly three orders of magnitude accompanying the linking of these structures by an octamethylene bridge.This factor is similar to that observed in the corresponding antagonists (43) and (11). It is worth noting that the unsymmetrical structure (55), which contains one 3-(imidazol-4-yl)propyl guanidine linked by a chain of eight carbon atoms to a guanidine group containing a ‘cimetidine side-chain’, is a partial agonist at H,- receptors of comparable potency to (54). C. 1mpromidine.-An exceedingly potent H,-receptor agoni~t,~’impromidine (56),has a structure comprising a guanidine group substituted by two different imidazole-containing side-chains. It may be considered as being derived from the partial structures 3-(imidazol-4-yl)propylguanidine or its N-methylguanidine derivative (52) (partial agonist) and 2-[(5-methylimidazol-4-yl)methylthio]ethyl-guanidine or its N-methylguanidine derivative (52) (an antagonist) (Figure 3).As 35 G. J. Durant, W. A. M. Duncan, C. R. Ganellin, M. E. Parsons, R. C. Blakemore, and A. C. Rasmussen, Nature (London), 1978, 276, 403. Guanidine Derivatives Acting at Histaminergic Receptors 2 0 v! 3 n 09 6- 0, 0, X z v x z u “sc XV m I I c X I I Nz Z=VI k 2 u,XVv1? Xu z I X I k 8P u 4 oI x Durant -22 0 z xu X?5 N X u, N X' u,X'uYi3:u nGG 5;33 6'2 6'5 8 N3 'H3'H3SZH3 1'0 6'9 8 H 'H3'H3SZH3 -'68's 21 H 6611'0 L'9 01 H 66EI'O E'l 6H '6PO'O 1'8 8H 6601'0 5'1 LH 6'99'0 1'9 9H 6'9'I 1'9 SH 66E'I 6'5 EH 7,' I 1'9 WH-,x-HN3H N-"t~3I-H N3H N -X-laH II ItZN ZN Durant 2 x - ?- I 2 =?-I I 0 ("?-I 0* I 8m d 39 1 Table 8 3-(Imidazol-4-yl)propylguanidinederivatives h, NH II H N\,/3 Hz-Receptor Activities in vitroa Agonist Activity Compound R Potency relative % Max.Antagonist Activity' No. to Histamine = 1 response pA2 b20.04 31 4.7 3'09 0.2 60 4.9 (53) 4.8 100 -NH(54) II 33 62 7.7 --( CH2I8-NHCNH -(CH,), )=\N NHv NHII (55) -( CH2)0-NHCNH-CH2CHZSCH2 22 39 8.0 N NHv a Determined on guinea-pig atrium. Relative agonist activities were assessed from cumulative dose-response curves.Construction of complete dose-response curves to histamine and test compounds were used to determine maximum responses obtainable and relative potencies were determined from concentration required to elicit 50% of maximal responses. See footnote a, Table 1. Durant NH H,-Receptor Activity II CH, scH,CH,NH c NHCH,C H,C H, h HN N N NHv v Impromidine (56) Potent agonist NH II Me C H ,SCH2CH,N HC N H Me (11) AntagonistHNvN NH II MeNH C HC H,C H,CH Weak partial agonist )-7 NwNH Figure 3 Chemical structures of impromidine and its ‘partial‘ structures an agonist, impromidine is between 10and 50 times as potent as histamine in uitro and in uiuo, and elicits a near-maximal response in most of these test systems.35 Impromidine, which is an exceedingly potent stimulant of gastric acid secretion in animals and man, has been used clinically to study the potency and duration of action of novel histamine H,-receptor antagonist^.^^ It has been suggested that treatment with this compound may offer a new therapeutic approach to patients with catecholamine-insensitive congestive heart fail~re.~’ Numerous analogues of impromidine including modifications to both of the side- chains and to the guanidine group have been synthesized and compared with impromidine3’ (Tables 9, 10, and 11).Analogues (57>--(60) in which the ‘cimetidine side-chain’ of impromidine is replaced by the alternative side-chains listed in Table 9 are also powerful agonists on guinea-pig atrium.These side-chains appear to be associated with affinity for H,-receptors and there are many examples of guanidinium structures containing these side-chains that are antagonists, for example the N-methylguanidines [(22), (24), and (26)] (Table 5) and particularly compounds in the series of bis-heteroarylalkyl guanidine derivatives (Table 6). It therefore appears likely that this side-chain substituent in impromidine and congeners is associated with affinity for H,-receptors. The 3-(imidazol-4-yl)propyl substituent present in impromidine appears to be R. H. Hunt, J. G. Mills, J. Beresford, J. A. Billings, W. L. Burland, and G. J. Milton-Thompson, Gastroenterology, 1980, 78, 505.” G. Baumann, B. Permanetter, A. Wirtzfeld, and H. Blomer, Eur. Heart J., 1984,5, Suppl. 1,34, Abstr. 125; G. Baumann et al., J. Cardiovascular Pharmacol., 1983, 5, 618. 38 G. J. Durant, C. R. Ganellin, D. W. Hills, P. D. Miles, M. E. Parsons, E. S. Pepper, and G. R. White, J. Med. Chem., 1985, 28, 1414. Guanidine Derivatives Acting at Histaminergic Receptors Table 9 H2-Receptor agonist potencies of impromidine analogues NH II Het -X-N HC N HCH *CH 2CH h NvNH H2-Receptor Agonist Activities in vitro Compound Het X Potency relative % Max. No. to Histamine = 1 response Impromidine CHzSCH2CHz 48 99(56) HNvN (57) CH2SCH2CH2 5.4 95 d HNvNCHzCH2CHzCH2 12.4 86 s< (59) CN CH2SCH2CHZ 24.4 91 5% (+N CHzCHzCH2 26.7 87 a*b See footnotes a,b, Table 8 crucial for H,-receptor agonist activity, as indicated by the effect of structural modification (Table 10).The lower homologue of impromidine containing the 2-(imidazol-4-y1)ethyl substituent (6 1) is considerably less active as an agonist and the higher homologue (62) is an H,-receptor antagonist which exhibits a weak sub- maximal response at H,-receptors on guinea-pig atrium (Table 10). Methyl substituents in the 2-and 4-position of the imidazole ring are reasonably well- accommodated and the impromidine analogues containing these substituents [(63) and (64)] are nearly full agonists approximately four to six times less potent than Durant Table 10 H2-Receptor activities of impromidine analogues NH II CH2SCH2C H,N H C N H -XI-Het Compound No.Impromidine (56) (61) (62) (63) (64) HNvN Het' X' 4 CH2CH2CH2HNdN d CH2CH2HNvN d HNyN CH~CH~CHZ Me MeHCHZCH~CH~ HNvN See footnotes a,b,c,Table 8 H2-Receptor Activities in vitro a Agonist Antagonist Activity Activity Potency relative Max. pA2 to Histamine = 1 response 48 99 - 1.9 100 - 0.8 20 5.9 12 100 - 8.1 85 - 0.0 1 11 5.8 2 47 5.6 Guanidine Derivatives Acting at Histaminergic Receptors impromidine. The influence of methyl substituents in the 2- and 4-positions of this imidazole ring is therefore in contrast with that previously observed for histamine where 2-methyl substitution leads to a much greater reduction in agonist activity at histamine H,-receptors than does this substitution at the 4-position of the imidazole ring.8 In impromidine, methylation of the imidazole ring nitrogen atom leads to a marked reduction in agonist activity, the 3-(l-methylimidazol-4-y1)propyl analogue (65) being only a weakly active partial agonist at histamine H,- receptors. As noted previously, the 3-(thiazol-2-yl)propyl analogue (32) is an H,-receptor antagonist and the results for these two compounds suggest that the tautomeric imidazole ring is necessary for agonist activity.The relatively weak activity of the 1,2,4-triazole analogue (66)demonstrates that, unlike for histamine,8 this ring system is not an effective substitute for imidazole in the impromidine series of H,-receptor agonists.Table 11 H2-Receptor activities of impromidine analogues 2 II M~~CH~SCH~CH~-X--C--Y--CH,CH,CH, hHN NHvN N v Hz-Receptor Activities in vitro a Agon is t Activity Antagonist Potency relative % Max. Activity' X Y Z to Histamine = 1 response pA2 -NH NH NH 48 99 --6.2NH NH NCN -NH NH S -5.9 --5.2NH NH 0 -S NH NH 6.9 83 NH S NH --5.5 -NH NH NCH3 0.06 91 NH NH SCH3 0.01 17 4.8 &e footnotes a.b,c,Table 8 Modifications to the guanidine group of impromidine are listed in Table 11. Analogues [(67), (68), and (69)] in which the basic guanidine group is replaced by the neutral groups, cyanoguanidine, thiourea, or urea are H,-receptor antagonists essentially devoid of agonist activity.However, the mere presence of a basic group is insufficient to endow agonist activity. The isothiourea group is strongly basic (S-methylisothiourea, pK, 9.78 at 20 oC)39and will exist predominantly as a protonated species at physiological pH. The isothiourea analogues of impromidine show interesting differences, since the isothiourea (70) resembles impromidine in acting as a strong agonist on guinea-pig atrium whereas the isomer (71) is an 39 A. Albert, R. Goldacre, and J. Phillips, J. Chem. SOC.,1948, 505. Durant antagonist essentially devoid of agonist activity. These results suggest that for agonist activity it is important for the atom bearing the 3-(imidazol-4-yl)propyl side-chain to possess a proton, whereas a proton on the atom bearing the 2-[(5-methylimidazol-4-yl)methylthio]ethyl side-chain is not essential for the agonist activity of impromidine and related compounds.The N-methylguanidine (72) and the S-methylisothiourea (73) are both weakly active as H,-receptor agonists on guinea-pig atrium-results consistent with the view that an amidinium structure containing an -NH, substituent is important for the agonist activity of impromidine (and analogues) at H,-receptors on guinea-pig atrium. Y/R /R' +N-H H-N I'fl,N:HH v (74) (75) Impromidine Congeners Histamine Congeners Cationic amidinium group Cationic ammonium group R(2) and R(4) = H or Me R' = H or Me Y=SorNH X = CH, CMe or N R = 'affinity' group N'-H tautomer of imidazole Figure 4 Chemical properties of impromidine and histamine that may be associated with agonist action at H,-receptors One extrapolation from the limited series of structural modifications listed in Tables 9, 10, and 11 is that the 3-(imidazol-4-yl)propyl amidinium structural fragment (74) is important for H,-receptor agonist activity where R(2) and R(4) are optional 2-or 4-methyl substituents, Y = S or NH, and R is a typical substituent that contributes affinity in H,-receptor antagonist structures (Figure 4).This fragment may be compared with the structure of histamine, and the structural features (75), and chemical criteria that have been considered necessary for its agonist action at H,-re~eptors.~' Clearly there are similarities and the presence of a tautomeric ring system linked by an alkylene chain to a protonated ligand appears to be important for agonist activity in both series of agonists, but there are also differences as indicated in Figure 4.In preliminary attempts to correlate structure and H,-receptor agonist activity, conformational space and molecular surfaces of the monocations of histamine and impromidine have been ~ompared.~' 5 Summary Many biologically active guanidines and imidazoles have been reported. The initial attempt to combine these groups within one molecule by replacing the amino- 40 G. J. Durant, C. R. Ganellin, and M. E. Parsons, J. Med. Chem., 1975, 18,905. 41 E. K. Davies and K. Prout, (Oxford University), unpublished results.Guanidine Derivatives Acting at Histaminergic Receptors group of histamine by guanidine led to the molecule W-guanylhistamine (5) which is a weak partial agonist at histamine H,-receptors. However, subsequent structural elaboration has led to a series of compounds that exhibit high levels of activity at histaminergic receptors. A notable example is the cyanoguanidine derivative, cimetidine (12), which is highly effective clinically as a histamine H,-receptor antagonist. Subsequent developments led to a family of isocytosines, which afforded potent H,-receptor antagonists, e.g. oxmetidine (15) and SK&F 93479 (17), the combined H,-/H,-receptor antagonist SK&F 93319 (20) and the selective H ,-receptor antagonist SK&F 93944 (21).Guanidine structures containing more than one imidazole ring structure have exhibited exceedingly high potency at histamine H,-receptors, either as antagonists or as agonists [e.g. impromidine (56)]. Our understanding of the molecular events underlying the pharmacological actions described in this review is rudimentary at the present time, but in future we hope for a greater comprehension of why the combination of the two moieties imidazole and guanidine within a single molecule leads to compounds that exhibit profound biological effects at histaminergic receptors. Acknowledgements. The work described in this review is the result of a close and long-standing collaboration with many colleagues at Smith Kline & French Research Laboratories. In particular I wish to acknowledge medicinal chemistry colleagues Drs.Robin Ganellin, John Emmett, Tom Brown, Bob Ife, George Sach, and Rodney Young. I wish to acknowledge Sir James Black for originating our interest in histamine and his constant encouragement in the formative years of this research and also thank Dr. Michael Parsons and Bob Blakemore for providing us with all the pharmacological data. The contribution of the chemists involved in the synthesis of compounds described in this review is gratefully acknowledged. These include Derek Hills, Peter Miles, Wasyl Tertiuk, David Cooper, Tony Rawlings, Douglas Prain, Maria Wilczynska, and Dr. Ray White.
ISSN:0306-0012
DOI:10.1039/CS9851400375
出版商:RSC
年代:1985
数据来源: RSC
|
7. |
The chemistry of peroxonium ions and dioxygen ylides |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 399-419
J. C. Mitchell,
Preview
|
PDF (979KB)
|
|
摘要:
The Chemistry of Peroxonium Ions and Dioxygen Ylides By J. C. Mitchell DEPARTMENT OF CHEMISTRY, ROYAL HOLLOWAY AND BEDFORD NEW COLLEGE, UNIVERSITY OF LONDON, EGHAM HILL, SURREY TW20 OEX 1 Introduction The intermediacy of carbonyl oxides (1) in ozonide formation has long been appreciated, but the wider occurrence of these species, or of ‘saturated’ oxy- oxonium ylides (2), and the reactivity of (1) and (2) has been recognized only recent!y. Similarly, discussion of the related alkoxy-oxonium (peroxonium) species (3) and (4)has only begun to appear in the last few years. 0-I 0-IO+RXR R/ ‘R R = H or alkyl OR I I O+RXR OR R’ ‘R (3) (4) Interest in chemical intermediates of this type has been focused primarily in the areas of ozon01ysis,’-~ the singlet oxygenation of alkenes,“l5 the alkylation of peroxides,6-8andthe‘oxenoid’behaviourofsomebiochemical system^.^.'^ Although both dioxygen ylides and peroxonium ions have been recently (1984) observed spectroscopically at low ternperat~re,~” no examples of stable dioxygen * R.Cnegee, Angew. Chem., Inf. Ed. Engl., 1975, 14, 745. A. T. Menyailo and M. V. Pospelov, Russ. Chem. Rev. (Eng. Trans/.), 1967, 36, 284. (a) R.E. Keay and G. A. Hamilton, J.Am. Chem. SOC.,1976,98,6578.(b)R. E. Keay and G. A. Hamilton, J. Am. Chem. SOC.,1975,97,6876. A. A. Frimer, Chem. Rev., 1979,79, 359. ’W. Adam, Chem.-Zrg., 1975,99, 142. N. A. Porter and J. C. Mitchell, Tetrahedron Lett., 1983, 24, 543.’J. C. Mitchell, S. Heaton, and N.A. Porter, Tetrahedron Leu., 1984, 25, 3769. A. J. Bloodworth, J. L. Courtneidge, and H. J. Eggelte, J. Chem. SOC.,Chem. Commun., 1983, 1267. G. A. Hamilton, J. Am. Chem. SOC.,1964, 86, 3391. lo G. A. Hamilton ‘Molecular Mechanisms of Oxygen Activation’, ed. 0.Hayaishi, Academic Press, 1974, p. 405. 0. L. Chapman and T. C. Hess, J. Am. Chem. SOC.,1984, 106, 1842. The Chemistry of Peroxonium Ions and Dioxygen Ylides zwitterions or peroxonium salts have been reported in the chemical literature. In the light of the availability of salts and ylides involving other heteroatoms, it is curious that the dioxygen analogues have proved so elu~ive’~-’~(Table 1). Table 1 Salts and ylides of nitrogen, oxygen, phosphorus, and sulphur Reported methods of generation of these ions as intermediates in chemical reactions range from carbene trapping of oxygen,20*21 to the intramolecular alkylation of peroxides,6V8 and reactions of these species include oxidation of external nucleophiles,2 peroxy-migration, and elimination of carbocati~ns.’~~ This review will attempt to define and interrelate the chemistry of perepoxides, dialkyl peroxonium ions, trialkyl peroxonium ions, dioxygen (0x0-oxonium) ylides, and carbonyl oxides.2 Protonated Hydrogen Peroxide Peroxonium ions can be most simply represented as the protonated ion H,02 + (5). Protonation of hydrogen peroxide has been effected with sulphuric acid23 and by mixtures of hydrogen fluoride and boron trifl~oride.~~ This ‘parent’ peroxonium J. March, ‘Advanced Organic Chemistry’, McGraw-Hill Inc., 1977, pp.337-378 and p. 11 11. l3 A. R.Katritzky and J. M. Lagowski, ‘Chemistry of the Heterocyclic N-Oxides’, Academic Press, 1971, p. 142. I4 A. W. Johnson, ‘Ylide Chemistry’, Academic Press, 1966, pp. 7-131, 251-260, and 304-349. H. Perst, ‘Oxonium Ions in Organic Chemistry’, Verlag Chemie GmbH, Weinheim/Bergstr., 1979. l6 R. A. Eades, P. G. Gassman, and D. A. Dixon, J. Am. Chrm. SOC.,1981,103, 1066. l7 B. J. Walker, ‘Organophosphorus Chemistry’, Penguin Books, Ltd., 1972, pp, 127-138. S. G. Smith and S. Winstein, Tetrahedron, 1958, 3, 317. l9 E. M. Arnett and V. M. DePalma, J. Am. Chem. SOC.,1977,99, 5828. ’O G. A. Bell and I. R. Dunkin, J. Chem. SOC.,Chem. Commun., 1983, 1213.W. Ando, H. Miyazak, and S. Kohmoto, Tetrahedron Lett., 1979, 15, 1317. 22 W. A. Pryor and C. K. Govindan, J. Am. Chem. SOC.,1981, 103, 7681. 23 0.H. Derbyshire and W. A. Waters, Nature, 1950, 165, 401. 24 R.W. Alder and M. C. Whiting, J. Chem. SOC.,1964, 908, 4707. ”G. A. Olah, A. L. Berrier, and G. K. Suryaprakash, J. Am. Chem. SOC.,1982, 104,2373. Mitchell R‘‘OL OR3 RZ/ is reported to be a powerful but unselective oxidant, reacting with benzene and cyclohexane at room temperat~re.~~ The structure of this ion is supported in a recent account by Olah, who reports the protonation of hydrogen peroxide and describes the 170 n.m.r. spectrum of this species.25 3 Perepoxides In these systems the cationic oxygen is incorporated in a three-membered ring, with the exocyclic oxygen usually bearing a proton or electron lone pair [analogous to ions (2) and (4)].Our laboratory has recently extended this class to include perepoxides bearing alkyl groups on the exocyclic oxygen (6). OR^ I R1 and R2 = alkyl +O R3 = H, alkyl, or/\R~ electron lone-pair~1-(6) Perepoxides have been postulated as intermediates in a variety of reactions, including singlet oxygenation of alkene~,~ basic cyclization of P-hydroperoxy bromides,26 and implicated in the silver-mediated ring-closure of P-hydroperoxy bromides.27 A. Singlet Oxygenation of A1kenes.-The existence of peroxonium intermediates in chemical reactions has been vigorously disputed in the scientific literat~re.~~-~~ Sharp2* first proposed perepoxide intermediates in singlet oxygen addition to isolated double bonds of alkenes (Scheme 1).Estimates of the energetics of this perepoxide intermediate (7) relative to the energies of other possible biradical(8) or zwitterionic (9) intermediates suggest that singlet oxygen addition occurs via a stepwise path~ay~~,~~ involving either (8) or (9) rather than (7)35(see Table 2). 26 K. R. Kopecky, W. A. Scott, P. A. Lockwood, and C. Mumford, Can. J. Chem., 1978,56, 1114. 27 K. R. Kopecky, J. E. Filby, C. Mumford, P. A. Lockwood, and J. Ding, Can. J. Chem., 1975,53, 1103. D. B. Sharp, Abstracts, 138th National Meeting, American Chemical Society, 1960, p. 79. 29 Sr. M. Bellarmine Grdina, M. Orfanopoulos, and L.M. Stephenson, J. Am. Chem. Sac., 1979,101,3111. 30 M. S. Dewar and W. Thiel, J. Am. Chem. SOC.,1975,97, 3978. ” C. S. Foote, Acc. Chem. Res., 1968, 1, 104. 32 F. McCapra and I. Beheshti, J. Chem. SOC.,Chem. Commun., 1977, 517. 33 L. B. Harding and W. A. Goddard, J. Am. Chem. SOC.,1980, 102, 439. 34 L. B. Harding and W. A. Goddard, J. Am. Chem. SOC., 1977,99,4520.’’Y. Yamaguchi, T. Fueno, I. Saito, T. Metsura, and K. N. Houk, Tetrahedron, 1981, 22, 749. 401 The Chemistry of Peroxonium Ions and Dioxygen Ylides -0fl( H OOH1HI> /c+L Scheme 1 '0 -0-0 (7) (8) (9) Table 2 Energetics (kJ mol-') of addition of '0, to alkyl-substituted 01efins~~ Zwitterion Perepoxide Biradical Olefin AH(a) AH(b) AH AH(a) AH@) --248 68 42 244 164 64 36 37 k b W 134 58 26 235 76 59 31 34r-130 67 53 22 24 /f-(a) more-substituted carbon; (b) less-substituted carbon These energetics (Table 2) for singlet oxygen addition to alkyl-substituted ethylenes have been calculated by Goddard and Harding.The calculations involve the combination of previous ab initio theoretical studies by the with thermochemical methods for estimating substituent effects. This thermo-chemical method utilizes the group additivities method developed by Benson and 402 Mitchell co-w~rkers.~~When the three proposed intermediates (7), (8), and (9) are considered by this technique, the one with the lowest energy is the peroxy biradical. The energy of the perepoxides averages 25-33 kJ mol-’ above the biradical form while open zwitterions are higher by 40-200 kJ mol-’.In the case of unsymmetrical olefins, zwitterion energetics appear to be dominated by the stability of the resultant carbonium ion. Simply, for zwitterionic intermediates, singlet oxygen addition should occur at the least substituted carbon. For biradical intermediates this regiochemistry of addition is reversed, with oxygen addition preferred to the more substituted carbon. In these cases the authors predict that the resultant less substituted radical would be of slightly lower energy. If, indeed, the perepoxide is of the order of thirty kJ mol-’ higher in energy than the acyclic isoelectronic biradical species in singlet oxygenation of alkenes (Table 2), surely the ring strain (ca.115 kJ mol-’ in ~xirane)~’ must be a contributing factor to the higher energy of formation of this ylide. Although theoretical examination of these reactions may cast doubt on the intermediate existence of perepoxides, experimental evidence for the intermediacy of the peroxonium species has been presented by both Bartlett and Stephenson. When adamantylideneadamantane is reacted with singlet oxygen, dioxetane formation is always accompanied by epoxidation products as reported by Bartlett.38 The production of epoxide occurs in all solvents utilized, with no evidence that any solvent is oxidized in the process. This product distribution is interpreted by Bartlett as involving a perepoxide intermediate (Scheme 2).-0 _//////_/ji O3 + PQkQQiScheme 2 36 S. W. Benson, ‘Thermochemical Kinetics’, 2nd Edn., Wiley-Interscience, 1976. 37 T. H. Lowry and K.S. Richardson, ‘Mechanism and Theory in Organic Chemistry’, Harper and Row, 1976, pp. 75-76. 38 P. D. Bartlett and M. E. Landis, ‘Singlet Oxygen’, ed. H. Wasserman, Academic Press, 1979. The Chemistry of Peroxonium Ions and Dioxygen Ylides Investigations of the reactions of cis- and tran~-[~H,]tetramethylethylene with singlet oxygen by Stephenson indicate the existence of a reaction intermediate2' (Scheme 3). nCH, CD3 = 1.41t0.03k A=1.07i0.7 kD n CD3 CD3 Scheme 3 For the singlet oxygenation of the cis-compound a very small isotope effect is obtained. For the trans-isomer a significant effect is observed.Based on this primary isotope effect, Stephenson rationalizes that a perepoxide, which is formed irreversibly, is this intermediate. Allowing that pyramidal inversion is slow relative to hydrogen abstraction, then an isotope effect should be observed if hydrogen and deuterium can compete (trans) and should be unobserved where there is no competition (cis) (Figure 1). 'CD, trans cis Figure 1 Further evidence for either a perepoxide or zwitterionic intermediate in these reactions has been presented by McCapra and Beheshti in a report describing the only example of a carbonium ion rearrangement as a result of singlet oxygen attack on an 01efin.~~ In a recent report by Schaap, new evidence for the oxidizing potential of intermediates of this type was presented.39 When sulphoxides were included in the 39 A.P. Schaap, S. G. Recher, G. R. Faler, and S. R. Villasenor, J. Am. Gem. Soc., 1983, 105, 1691. Mitchell photoreaction of oxygen with adamantylideneadamantane, the reaction was proposed to proceed via a dioxygen ylide that rearranges to cyclic peroxide or is trapped by sulphur to yield oxidized sulphone products and epoxide (Scheme 4). 0-A'02 * R-RR=R 0-0 0 0 IIaII / \ + PhSMe IIR= R-R R-R 0 Scheme 4 B. Reactions of P-Hydroperoxy Bromides.-Work in this area by Kopecky et al. involving the reactions of 1,2-bromohydroperoxides with methanolic sodium hydroxide resulted not in the production of dioxetane but gave products resulting from peroxy-migration and proton abstraction.26 In these studies it was concluded that these reactions proceeded by way of perepoxide intermediates that collapsed to the observed 'ene' products (Scheme 5).0-0 SO3 YOH 1 CH3 D,C\ / \ ,CH3 D,CC-CCH, + D,C!-CjCH, C-C,I1 / Br CH, CD, Br 3c CH3 \ D3 \C -D3C' OOH HOO iD2 I iH2 C-CCH, + D3CC-I C I I I I CO, CH, CD, CH, 3 1 '1. 6 9 'lo Scheme 5 405 The Chemistry of Peroxonium Ions and Dioxygen Ylides Reaction of these systems with silver acetate in methylene chloride results in both hydroperoxy alkenes and cyclic peroxide^.^' This product distribution supports the formation of a cyclic intermediate that can rearrange to the familiar ene product or dioxetane.4 Intramolecular Alkylation of Peroxides The intermediacy of peroxonium ions, analogous to ion (4),has been proposed and supported by investigation of the intramolecular alkylation of peroxides. To date, these reactions have taken three forms, electrophilic attack on alkylperoxy alkenes, the Lewis-acid-induced ring-closure of alkylperoxy bromides, and the base-induced ring-closure of hydroperoxy bromides. In the case of Lewis-acid-mediated reactions, a variety of silver salts as well as antimony halides have been in~estigated.~' A. Ring Closures of Alkylperoxy Bromides.-We have investigated the reaction of a series of dialkylperoxy bromides with silver trifluoroacetate, silver tetrafluorobor- ate, and antimony pentachloride.Reaction products are dependent on both chain length and nucleophilic character of the reaction solvent (- R=B~,n=l YCH2 In Br R= BU~, n 2 I OOR R= BU~,n=3 (10) R=H,n=2 4-Bromo-2-methyl-2-(t-butylperoxy)butane(R = But, n = 1) reacts with silver tetrafluoroborate in methanol to produce 3,3-dimethyl-1,2-dioxolaneand t-butyl methyl ether in quantitative yield. However, the reaction of 5-bromo-2-methyl-2-(t-buty1peroxy)pentane (R = But, n = 2) with silver salt in methanol produces a peroxy-migration product, while reaction in methylene chloride produces 3,3-dimethyl- 1,2-dioxane. The intermediacy of two different cyclic peroxonium ions has been enlisted to explain these results (Scheme 6). Low-temperature n.m.r. investigation of this reaction in methylene chloride does Scheme 6 40 J.C. Mitchell and N. A. Porter, unpublished results. Mitchell indicate production of ionic intermediates as well as t-butyl cation’ (see Figure 2). When 6-bromo-2-methyl-2-(t-butylperoxy)hexane(R = But, n = 3) was com- bined with silver salt under the same conditions as the 1,3- and 1,4-peroxy bromides, no reaction to produce any peroxonium intermediate was detected.6 This work, when considered in combination with the earlier work of Kopecky, necessitates a mechanistic consideration of enthalpic and entropic effects in these small-ring closure^.^^^^^ In all cases a preference for ‘kinetic’ ring-closure to OOH A*ROOH-RBr hH+H 0-0 OOH FOOBut -&CB~%%n OMe OOBu 0-0 POOH ACBrSfi OMe OOH 0-0 9 GOMe 0OBUt Reagents A-NaOMe in MeOH C-AgBF, in MeOH* A-NaOH in MeOH D-AgBF4 in CH,CL, €3-NaH in pentane D*-AgOCOMe in CH,Cl, Scheme 7 41 E.L. Eliel, ‘Conformational Analysis’, Interscience Publishers, 1965, pp, 189-197. O2 C. Galli, G. Illuminati, L. Mandolini, and P. Tamborra, J. Am. Chern. Soc., 1977,99, 2591. The Chemistry of Peroxonium Ions and Dioxygen Ylides peroxonium ions is observed. That is, attack from the peroxy-oxygens will favour the formation of three- or five-membered rings, and products will result from these kinetic intermediates. As little difference in nucleophilicity between the two peroxy- oxygens can be this phenomenon can be exploited to generate products from peroxonium intermediates (with one oxygen exocyclic), or peroxides with both oxygens endocyclic (Scheme 7).It is perhaps surprising that no peroxonium chemistry was observed for the reaction of our acyclic 1,5-peroxy bromide with silver salt, as a six-membered ring peroxonium intermediate appears ac~essible.~' In a related reaction, the silver-mediated ring-closure of l-bromomethyl-8-(t- buty1peroxy)methylnaphthalene yields naphtho-pyran and trapped 2-methoxy-2- propyl cation.' A mechanistic scheme involving the intermediacy of a peroxonium ion can be written to describe production of naphtho-1,8-pyran from the starting peroxy bromide (Scheme 8). (& A9+, + Me0 -(+\ # I-1")" Me,C=O + MeOH Scheme 8 Formation of both the intermediate peroxonium ion and 2-methoxy-2-propyl cation is strongly supported by low-temperature n.m.r.spectroscopy. When this peroxy bromide is reacted with silver tetrafluoroborate or antimony pentachloride at low temperature, low-field 'H (or 2H from a deuterated precursor) resonances can be observed (Figure 3). B.Electrophilic Attack on Alkylperoxy A1kenes.-Another example of this Baeyer- Villiger type 0-0cleavage with 1,2-nucleophilic migration of a methyl group may be found in a recent report by Bloodworth et a[.*Here, intramolecular alkylation of dialkylperoxide was accomplished by electrophilic attack on alkylperoxy alkenes. When 5-t-butylperoxycyclo-octeneis reacted with N-bromosuccinimide or bromine in carbon tetrachloride, a mixture (3: 1 ratio of A to B) of bicyclic ethers is produced (Scheme 9).43 R. Hiatt, 'Organic Peroxides', Wiley-Interscience, 1971, Vol. 1, Chap. IV Mitchell 5 4 ppm 2 1 Figure 2 Low temperature 'H n.m.r. of 5-bromo-2-methyl-2-(t-butyiperoxy)pentane with antimony pentachloride in sulphuryl chlorojluoride (-60 OC): (a) bromoperoxide, (b) t-butylcation, (c) cyclic peroxide, (d,e) transient signals due to interconverting carbocation and peroxonium ion? (see Scheme 6) I I I I I 8 7 65 4ppm Figure 3 Low temperature 'H n.m.r. of l-bromomethyl-8-t-butyiperoxy-[2H,lmethyl-naphthalene with antimony pentachloride in methylene chloride (-55 "C):(a) naphtho-pyran, (6)peroxy bromide, (c) peroxonium ion? (d) CDCl, (see Scheme 8) Scheme 9 409 The Chemistry of Peroxonium Ions and Dioxygen Ylides The reaction mechanism is reported to be electrophilic attack on the double bond, followed by intramolecular alkylation of the peroxide, and subsequent methyl migration to produce the ether product and 2-methoxy-2-propyl cation (Scheme 10).OO8Ut Me \ Br Qr Scheme 10 Evidence that internal nucleophilic attack at oxygen (Scheme 10) is not a prerequisit for 0-0cleavage in peroxonium ions is also presented by Bloodworth. When 5-hydroperoxycyclo-octene was treated with N-bromosuccinimide or bromine in carbon tetrachloride the same 3 :1 ratio of bicyclic ethers was obtained as in the case of t-butylperoxycyclo-octene.Reactions of this type suggest that peroxonium ions from hydroperoxy precursors may be sources of electrophilicOH, which can be activated by suitable nucleophiles (Scheme 11).NBS6"___) Br i3r Scheme 11 C.Base-induced Ring-closure of Hydroperoxy Bromides.-Aside from electrophilic attack on hydroperoxy alkenes, the production of peroxonium ions from hydroperoxy precursors has been reported by base-induced ring-closures of hydroperoxy bromides. The previously discussed work of Kopecky et al., on the reactions of p-hydroperoxy bromides to produce perepoxide intermediates, is of this t~pe.~~,~ Our curiosity with these reactions stems from the observation that ring-closure of these systems seems not to occur solely through attack by the 'outside' anionic oxygen (Scheme 12), to yield stable peroxides, but involves formation of dioxygen ylides.To investigate further this phenomenon we have studied the reaction of 5-bromo- 2-methyl-2-hydroperoxypentanewith sodium hydride in pentane and with sodium methoxide in methanol. In both cases we isolated 3,3-dimethyl- 1,2-dioxane. These reaction products indicate that upon reaction of this 1,4-hydroperoxy bromide Mitchell NaOMc ___) 0-Br McOH I>ooiT= * Br Scheme 12 with base, attack is made to displace bromide by the ‘outside’ anionic oxygen thus forming the six-membered dioxane. An alternative route, which we have no data to support, is the formation of the ‘kinetic’ five-membered ring peroxonium compound, followed by rapid peroxy-transfer and subsequent ring-closure (Scheme 13).>c70-0,** Scheme 13 5 Carbonyl Oxides The ability of perepoxides and peroxonium ions to oxidize external nucleophiles is coincidental with some recently reported chemistry of carbonyl oxides. Carbonyl oxides have long been recognized as intermediates in the ozonolysis of ole fin^'*^^ (Scheme 14). These oxenoid intermediates have been reported to transfer oxygen atoms to sulphides,4* sulph~xides,~’~~~ alkanesY4’ olefin~,~*~~,~~ and aromatic ring^.^^^^ These species are also believed to be involved in the Baeyer-Villiger decomposition 44 R. Criegee and G. Wenner, Liebigs Ann. Chem., 1949, 9, 564. 45 G. A. Hamilton and J. R. Giacin, J. Am. Chem. SOC.,1966, 88, 1584. L6 H. Kwart and D. H. Hoffman, J. Org. Chem., 1966,31,419.47 T. A. Hinricks, V. Ramachandian, and R. W. Murray, J. Am. Chem. SOC.,1979, 101, 1282. Y. Sawaki, H. Kato, and Y. Ogata, J. Am. Chem. SOC.,1981, 103, 3832. 49 W. Ando, S. Kohmoto, and K. Nishizawa, J. Chem. SOC.,Chem.. Commun., 1978,894. J. W. Jerina, D. M. Jerina, and B. Witkop, Experientia, 1972,28, 1129. 51 S. K. Chandhary, R. A. Hoyt, and R. W. Murray, Tetrahedron Lett., 1976,4235. 411 The Chemistry of Peroxonium Ions and Dioxygen YIides 1 0-0->c,’‘;<0 ->c=o 11+ 01-11 o+ c+ c 1-vcn AA Scheme 14 of furan endoperoxides. 52 Epoxidations of olefins by metal-ion co-ordinated carbonyl oxides have been reported,53 as well as the observation that ozonolysis of olefins in the presence of tetracyanoethylene can result in the production of tetracyanoethylene oxides.54 Ogata has concluded that nucleophilic oxygen transfer is a characteristic reaction of these although electrophilic oxidation of sulphides has also been reported.55 DiFuria and M~dena~~ describe nucleophilic oxygen transfer as a two-step process which involves the addition of the peroxo compound to the substrate and subsequent cleavage of the peroxide intermediate (Scheme 15a).Electrophilic oxygen transfer is described in Scheme 15b. R-00- + Z=X [ROO-Z-X]- -R-0- + 0-Z-X la) ROOR + NU-NU+OR + R-O- == N~O+ ROR (b) Scheme 15 Ab initio and MIND0/3 molecular orbital (MO) calculations have been used to investigate the stabilities of carbonyl oxides and related isoelectronic structures, (1la-c) and (12).Of the four isomers considered, dioxirane (12) has been determined to be 120- 170 kJ mol-’ more stable than carbonyl oxides (ll).57-60 The parent dioxirane has 52 W. Adam and A. Rodriquez, J. Am. Chem. SOC.,1980,102,404. 53 H. S. Ryang and C. S. Foote, J. Am. Chem. Soc., 1980,102, 2129. 54 P. S. Bailey, ‘Ozonation in Organic Chemistry’, Academic Press, 1978, Vol. 1. 55 W. Ando, Y. Kabe, and H. Miyazaki, Photochem. Photobiol., 1980, 31, 191. 56 F. DiFuria and G. Modena, Pure Appl. Chem., 1982,54, 1853. ’’W. R. Wadt and W. A. Goddard, J. Am. Chem. SOC.,1975,97, 3004. L. B. Harding and W. A. Goddard, J. Am. Chem. SOC.,1978, 100, 7180. 59 D. Cremer, J. Am. Chem. SOC.,1979, 101, 7199. 6o K.Yamaguchi, S.Yabushita, T. Fueno, S. Kato, K. Morokuma, and S. Iwata, Chem. Phys. Lett., 1980, 71, 563. 412 Mitchell been synthesized via ozonolysis of ethylene, and has been characterized by mass and microwave spe~trometry.~'-~~ Although the dioxirane was shown to be more stable, a MIND0/3 calculation has shown the energy barrier in the isomerization from (11) to (12) to be on the order of 105 kJ m01-I.~~ When the stabilities of (1 la+) are considered, MIND0/3 analysis predicts the zwitterionic structure (lla) to be more stable than (llb) or (ll~).~~Ab initio calculation suggests, however, that the singlet diradical structure is more stable than (1 la),57 but relative stabilities may be changed by substituents or solvents.58 If the orbital energies of (1 la) versus (12) are considered, the ionization potential as well as the energy of the LUMO of the carbonyl oxide (1 la) structure is lower than that for the dioxirane (12) structure.This suggests both that (lla) may be a much stronger nucleophile than (12), and that the electron affinity of the carbonyl oxide (1la) is also much higher.48 Experimenial support for this difference in reactivity and nucleophilicity of the two isomeric structures (11) and (12) has been recently provided by both Adam65 and Murray.66 Earlier work by Ogata had provided evidence for nucleophilic oxygen atom transfer from carbonyl oxides to sulphoxide (Scheme 16). I R2C+OO-+ >S=O -R,C+OOS--O-I Scheme 16 Murray has reported 0-atom transfer by dimethyl dioxiranes to a series of arenes in good yield, and suggests that the transfer reaction may be electrophilic in nature66 (Scheme 17).Adam has reported an attempt to differentiate between the electrophilic or nucleophilic character of these oxygen transfer agents by incorporation in his reaction mixtures of an oxidizable substrate, containing both nucleophilic sulphide 61 F. J. Lovas and R. D. Suenram, Chem. Phys. Lett., 1977,51,453. 62 R. D. Suenram and F. J. Lovas, J. Am. Chem. Soc., 1978, 100, 5117. 63 R. I. Martinez, R. E. Huie, and J. T. Herron, Chem. Phys. Leit., 1977, 51, 457. 64 L. A. Hull, J. Org. Chem., 1978, 43, 2780. "W. Adam, W. Haus, and G. Sieker, J. Am. Chem. SOC.,1984, 106, 5020. 6b R. Jeyaraman and R.W. Murray, J. Am. Chem. Soc., 1984, 106, 2462. The Chemistry of Peroxonium Ions and Dioxygen Ylides 0\m* Scheme 17 and electrophilic sulphoxide sites.65 When carbonyl oxides and dioxiranes with the same substitution patterns were combined with thianthrene 5-oxide, the carbonyl oxides proved appreciably more nucleophilic than the dioxiranes. These same experimental conditions were also used to corroborate the fact that dioxiranes can epoxidize electron-poor substrates better than peroxy a~ids~~,~*(i.e. m-chloroperbenzoic acid). The formation of carbonyl oxides in ozonolysis reactions, as previously discussed, is well known but these systems are not appropriate for the study of reactivity because of the involvement of other peroxidic specie^.^ Ozone, 1,2,3-trioxolane, and carbonyl oxide are all potential oxygen atom transfer agents, and all three coexist under normal ozonolysis conditions.22 A clean and convenient method for the generation of carbonyl oxides is the reaction of diazo compounds with singlet o~ygen.~ 1,69,70 This method of generation of dioxygen ylides has been exploited both in studies of the mechanism of reaction of these ylides with heteroatom-centred substrates (see above), and in the spectroscopic investigation of the carbonyl oxides, themselves.''JO Chapman has reported the generation, photochemistry, and spectroscopic observation of a carbonyl oxide isolated in an argon-oxygen matrix at 10 K.' 'y7' The carbonyl oxide is photochemically generated from a diazocyclopentadiene precursor, and infrared bands at 1395 cm-' and 1385 cm-' were tentatively assigned to the 0-0stretch in the carbonyl oxide (Scheme 18).This stretching frequency is approximately midway between carbon-oxygen double (ca. 1700 cm-') and single bonds (ca. 1150 cm-'), with the corresponding band in ozone at 11 10 cm-'.1'~72 This spectroscopic evidence in conjunction with 67 J. 0.Edwards, R. H. Pater, R. Curci, and F. DiFuria, Photochem. Photobiol., 1979,30, 63. 68 R. Curci, M. Fiorentino, L. Troisi, J. 0.Edwards, and R. H. Pater, J. Org. Chem., 1980, 45, 4758. 69 D. P. Higley and R. W. Murray, J. Am. Chem. SOC.,1974,%, 3330. 'O J. C. Scaiano, S. E. Sugamori, and H. L. Casal, J. Am. Chem. SOC.,1984, 106, 7623. 71 0.L.Chapman and C. L. McIntosh, J. Chem. SOC.D.,1971, 770. ''M. K. Wilson and R. M. Badger, J. Chem. Phys., 1948, 16, 741. 4 14 Mitchell Scheme 18 the observed photochemistry of this ylide to react further with O2 to produce cy~lopentadienone~and ozone supports the description of these carbonyl oxides as zwitterionic structures. A further complication exists in the study of carbonyl oxides formed by sensitized photo-oxidation of diazo compounds in the presence of sulphides and sulphoxides. That is, the sulphides and sulphoxides can be photo-oxidized directly to persulph~xides~~*~~and persulphones7’ (Scheme 19). RZS=O R2S+OO-II 0 Scheme 19 The reactions of these peroxy sulphides and sulphones resemble those of carbonyl oxides, in that their characteristic reaction is nucleophilic oxygen transfer.Less efficient reactions are the electrophilic oxidation of sulphides and the oxidative C-C cleavage of olefins’’ (Scheme 20). RiSO, + R2S02 Scheme 20 6 Biological Implications of Peroxonium Intermediates A number of chemical systems of biological interest are believed to involve peroxonium ions or oxy-oxonium ylides. Hamilton has suggested that certain reactions catalysed by the mono-oxygenase enzymes involve an oxenoid l3 C. S. Foote and J. W. Peters, J. Am. Chem. SOC.,1971, 93, 3795. 74 M. L. Kacher and C. S. Foote, Photochem. Photobiol., 1979, 29, 765.’’Y. Sawaki and Y. Ogata, J. Am. Chem. SOC.,1981, 103, 5947. The Chemistry of Peroxonium Ions and Dioxygen Ylides mechanism resulting in oxygen atom tran~fer.~,~~ It has also been suggested that carbonyl oxides serve as models for mono-oxygenase enzymes in that they are capable of oxidizing hydrocarbons to alcohols or carbonyl and are capable of oxidizing pyrroles7 and aromatic hydrocarbons.51 The flavin adenine dinucleotides (FAD, FADH,) function as prosthetic groups of these oxidation- reduction enzymes, which are also known as flavoenzymes or flavoproteins (Figure 4). These enzymes catalyse the oxidation of a variety of substrates using molecular oxygen. R RH FAD FADH, Figure 4 Reaction of aromatic compounds in non-enzymatic systems to produce hydroxylated aromatic compounds is known to occur when flavins are present. One mechanism for this hydroxylation is reported to involve a dioxygen ~lide~~ (Scheme 21).Goddard, however, has suggested that, based on molecular orbital calculations, this oxidation results not from a carbonyl oxide like dioxygen ylide, but from a diradical isomer of the ~lide.~~" Also, the recent work of Bruice et al. tends to show that ring-opening (and hence the formation of dioxygen ylides) is not involved in flavin-catalysed oxidation.79b*' The photo-reactions of porphyrins are also believed to involve dioxygen ylides. Both free base and metalloporphyrins are known to sensitize oxygen triplet- to singlet-state conversion in photo-oxidation processes.80-82 Protoporphyrin IX, 76 G. A. Hamilton, 'Progress in Bioorganic Chemistry', Wiley-Interscience, 197 1, Vol.1. 77 H. H. Wasserman and A. H. Miller, Chem. Commun., 1969, 199. 78 A. L. Leninger, 'Biochemistry', 2nd Edn., Worth Publishers, 1975. 79 (a)see: J. L. Fox,Chem. Eng. News, 1978, May 22, p. 28, (b)T. C. Bruice, J. B. Noar, S. S. Ball, U. V. Vankataram, J. Am. Chem. SOC., 1983,105,2452, (c) A. Wessiak and T. C. Bruice, J. Am, Chem. Soc.. 1983, 105,4809. G. Cauzzo, G. Gennari, G. Jori, and J. D. Spikes, Photochem. PhotobioL, 1977, 25, 389. S. Cannistraro, A. von de Vorst, and G. Jori, Photochem. Photobiol., 1978, 28, 257. 82 C. Emiliani and M. Delmelle, Photochem. Photobiol., 1983, 37, 487. Mitchell RH R R I Scheme 21 hematoporphyrin IX,and bilirubin have all been implicated in this photochemical con~ersion.~~*~~The process is known to result in oxidation of amino-acids and in photodynamic membrane damage.85 Porphyrin-O, species have also been implicated in several photobiological disorders, such as erythropoietic protophyria and neonatal jaundice.Although much research has been devoted to clinical studies of these disorders, little is known of the chemical interaction of the relevant porphyrins with oxygen or of the subsequent products. It has been reported that the reaction products of the self-sensitized photo-oxidation of protoporphyrin IX are medium de~endent.~~.'~ In organic (i.e.isotropic) media, products arise from attack of singlet oxygen on the porphyrins8' (Scheme 22). In micelles or vesicles (i.e. organized media) more complex behaviour is observed, with a report of 'superoxide-like' character.88 When photo-oxidation of protoporphyrin IX is carried out in human erythrocyte membranes, which contain saturated and unsaturated lipids as well as a variety of membrane proteins, disruption as well as cross-linking of the membrane is known to OCCU~.~~.~~If the amino-acids methionine, histidine, or tryptophan are included in aqueous microemulsions of the porphyrin, the oxidation of these amino-acids is concurrent with the formation of a porphyrin ep~xide.~~ It is not clear whether the formation of epoxides results from direct reaction of the porphyrins with singlet oxygen, or results from attack on the porphyrins by an 83 J.D. Spikes in 'Porphyrin Photosensitization', ed. D. Kessel and T.J. Doughterty, Plenum Press, New York, 1983. M. Krieg and D. G. Whitten, J. Am. Chem. SOC., 1984, 106, 2477. 85 A. W. Girotti, Biochemistry, 1975, 14, 3377. 86 G. Cox and D. G. Whitten, J. Am. Chem. SOC., 1982, 104,516. 87 H. H. Inhoffen, H. Brockmann, and K. Bliesner, Liebigs Ann. Chem., 1969,730, 173. G. S. Cox, M. Krieg, and D. G. Whitten, J. Am. Chem. SOC., 1982,104,6930. 89 A. W. Girotti, Biochem. Biophys. Acta, 1980, 602, 42. 417 The Chemistry of Peroxonium Ions and Dioxygen Ylides CHO 1 02 ~ CO2H COtH + CHO CO, H C02H Scheme 22 intermediate oxygenated species. Since the thioether group is a well known substrate for singlet ~xygen,'~ it has been suggested that sulphur-containing amino- acids (i.e. methionine) can be oxidized to persulphoxides (R'R2S+OO-) that degrade the photo-sensitizing porphyrins them~elves.~~ The success of photo-therapy in the treatment of neonatal jaundice seems to depend in part on the photo-degradation of bilirubin.Both McDonaghgo and Bonnettg' have suggested that bilirubin can sensitize oxygen triplet- to singlet-state conversion and that oxidation by singlet oxygen may be involved in the mechanism of photo-decomposition of bilirubin. Support for the formation of endoperoxides as intermediates in the photo-oxidation of bilirubin has been presented by Bonnettg2 and Lightner.93*94 More recently, Lightner has considered the dye- sensitized photo-oxygenation of pyrroles as a model for the photo-decomposition of bilirubin.When N-methylpyrrole undergoes photo-reaction with singlet oxygen, the formation of an endoperoxide intermediate that can rearrange to a peroxirane 90 A. F. McDonagh, Biochem. Biophys. Res. Commun., 1971,44, 1306. 91 R. Bonnett and J. C. M. Stewart, Biochem. J., 1972, 130, 895. 92 R. Bonnett and J. C. M. Stewart, J. Chem. Soc., Chem. Commun., 1972, 596, 93 D. A. Lightner and D. C. Crandell, Tetrahedron Lett., 1973, 12,953. 94 D. A. Lightner and G. B. Quistad, FEBS Lett., 1972, 25, 94. 418 Mitchell I IMe Me tie Scheme 23 is ~uggested.’~ This dioxygen ylide can then either internally rearrange or react hydrolytically to form the observed hydroxylactams (Scheme 23). 7 Conclusions Clearly, the production of peroxonium ions and dioxygen ylides in chemical or biochemical systems is of increasing interest to chemical researchers.That formation of these intermediates can result in a wide range of reaction types, including peroxy-migration, peroxidation, epoxidation, formation of ethers, and production of carbocations, makes these ions and ylides a concern of synthetic chemists. Of equal concern to theoretical chemists is the exact nature of charge distribution in these species. The debate over the zwitterionic versus diradical structure of both perepoxides and carbonyl oxides suggests that further considerations of substituent effects and solvation need to be included in future molecular orbital calculations. The intramolecular alkylation of peroxides, whether by reaction of peroxy bromides with silver salts, or by electrophilic attack on peroxy alkenes, appears to be under kinetic control in that a propensity for three- or five-membered ring formation exists.In the case of reaction of hydroperoxy bromides with base, however, although perepoxides appear to be formed via 1,2-hydroperoxy bromides at the expense of dioxetane formation, reaction of 1,4-hydroperoxy bromides appears to form products under thermodynamic control (i.e. dioxanes). Of particular interest is the ability of some of these peroxonium ions and ylides to oxidize external substrates by oxygen atom transfer. This chemistry is similar to oxygen atom transfer reactions reported for carbonyl oxides and may have ramifications in the photochemistry of the biologically important flavins and porphyrins.The photo-degradation of these biological systems appears to involve an oxenoid mechanism, resulting in oxygen transfer. These photodynamic processes may be involved in the oxidation of amino-acids and unsaturated fats, as well as in cellular membrane disruption. 95 D. A. Lightner, G. S. Bisacchi, and R. D. Norris, J. Am. Chem. Soc., 1976,98, 803.
ISSN:0306-0012
DOI:10.1039/CS9851400399
出版商:RSC
年代:1985
数据来源: RSC
|
8. |
Brownian dynamic with hydrodynamic interactions: the application to protein diffusional problems |
|
Chemical Society Reviews,
Volume 14,
Issue 4,
1985,
Page 421-455
Eric Dickinson,
Preview
|
PDF (2102KB)
|
|
摘要:
Brownian Dynamics with Hydrodynamic Interactions: The Application to Protein Diffusional Problems By Eric Dickinson PROCTER DEPARTMENT OF FOOD SCIENCE, UNIVERSITY OF LEEDS, LEEDS, LS2 9JT 1 Introduction What we now call the Brownian motion of microscopic particles was described for the first time in 1828 by the botanist Robert Brown.’ Some sixty years later, Gouy correctly attributed2 the phenomenon to the thermal motion of the surrounding liquid molecules. It was observed at an early ~tage~*~ that Brownian movement is most lively with small particles in liquids of low viscosity, and that Brownian drift velocities are some lo8 times smaller than typical molecular velocities. Roughly speaking, particles can be defined as Brownian if they are larger than normal solvent molecules (or ions), but still small enough to be perturbed appreciably by solvent molecular motion.This puts them in the colloidal size range (1 nm-1 pm). This article is concerned with the behaviour of proteins viewed as small colloidal particles. What follows is a description of how certain aspects of protein dynamics can be treated theoretically as problems which are soluble with the help of a computer. To establish our frame of reference, let us begin by listing some biological processes where we might tentatively expect Brownian motion to be a significant factor: (i) protein adsorption at a cell surface; (ii) the encounter between enzyme and substrate molecules; (iii) the interaction of an antibody with an antigen; (iv) protein mobility in a membrane, or along a fibre; (v) biochemical assembly by monomer aggregation or polymerization; and (vi) protein unfolding and denaturation. The common element in these processes is a kinetic stage which is diffusion controlled; and it is this element which we wish to emphasize here. In chemistry and biology, the complexities of macroscopic change are driven by two types of physical events: time-reversible ones, which obey the classical laws (Newton’s equations of motion), and time-irreversible ones, which obey probabilistic laws having their origin ultimately in the Second Law of thermodynamic^.^ Generally speaking, systems containing a small number of interacting objects are time reversible, and those containing a very large number are time irreversible.The dynamics of a few interacting Brownian particles immersed in an inert fluid medium (millions of molecules) can be regarded as being partly deterministic (reversible) and partly chaotic (irreversible). The deterministic part of the motion arises from interparticle colloidal forces (electrostatic, van der Waals, etc.) and the influence of external fields (magnetic, gravitational, etc.). The R. Brown, Ann. d. Phys. u. Chem., 1828, 14, 294. M. Goiiy, J. Physique (Paris), 1888, 7, 561. F. M. Exner, Ann. Phys., 1900, 2, 843. I. Prigogine and I. Stengers, ‘Order out of Chaos’, Heinemann, London, 1984. 421 Brownian Dynamics with Hydrodynamic Interactions chaotic part of the motion is associated with fluctuating Brownian forces from the apparently random thermal motion of the solvent molecules. The random impacts of surrounding molecules also give rise to frictional forces acting on the particles.Since the size of these frictional forces is dependent on the relative separations of the particles, it is found that the Brownian motions of the different particles, whilst remaining irregular, are in fact statistically coupled via the fluid medium. The description of condensed matter of interest here is one combining Brownian motion with continuum hydrodynamics. The subject of Brownian motion deals with such entities as colloidal particles, diffusion coefficients, and statistical probabilities; hydrodynamics, on the other hand, is concerned with macroscopic bodies, steady flow, and continuum dynamics.Bring the two together and, to coin a phrase, we get Brownian dynamics. This is a kinetic theory essentially diffusive in character, but also including the effects of particle interactions, both hydro- dynamical and colloidal. Brownian dynamics is appropriate for describing protein motions over distances which are large compared with the solvent molecular size, and times which are long compared with the interval between successive solvent impacts. In practice, most Brownian dynamics problems of chemical interest are not amenable to analytic solutions, but can be solved numerically using a computer. The usefulness of computer simulation in describing the dynamics of proteins is becoming increasingly recognized,’-’ if not quite yet universally accepted.* A great strength of simulation, sometimes called ‘computer experiment’, is that it enables one to follow the consequences of changing certain variables independently in a way not possible often in a real experiment.In outlining suitable models for protein simulation, we shall be concerned here with emphasizing the underlying physical features. Necessarily, this will be at the expense of omitting some of the biochemical ramifications-although, in principle, the approach is sufficiently general to include all detailed aspects if the time and trouble are taken to put them in. 2 Basic Principles Although Brownian-dynamics computer simulation is a relatively new field of study, it is based on some old and well-established principles.We begin our discus- sion of the theoretical background by mentioning the major historical contributions. A. Einstein’s Equation.-In his classic paper on Brownian movement, published in 1905, Einstein showed that the average displacement x,, of a tagged particle in one-dimensional projection follows equation 1 x,, = ((X2))t = (2Df)t J. A. McCammon and M. Karplus, Ann. Rev. Phys. Chem., 1980, 31, 29. M. Karplus, Ber. Bunsenges. Phys. Chem., 1982, 86, 386.’J. A. McCammon, Rep. Prog. Phys., 1984, 47, 1. A. Cooper, Prog. Biophys. Molec. Biol., 1984, 44, 181. A. Einstein, Ann. Phys., 1905, 17, 549 (English translation: ‘Albert Einstein, Investigations on the Theory of Brownian Movement’, ed.R. Fiirth, Dover, New York, 1956). See also: A. Pais, “‘Subtle is the Lord. . .”, The Science and Life of Albert Einstein’, Oxford University Press, New York, 1982, chap. 5. Dickinson where t is time, D is a diffusion coefficient, and (x2) is the mean-square displacement in the x-direction. As each Cartesian direction is equivalent, it follows that (rZ> = (x') + (y') + (z2>= 3(x2> (2) where Y is the total instantaneous distance travelled by the particle in time t. The crux of the derivation of equation 1, set out briefly below, is the recognition that the time-dependent probability distribution for random movement of a single particle is mathematically equivalent to the development of the concentration profile in bulk diffusion.Imagine a large number n of identical non-interacting* particles. They are accumulated at time t = 0 in the immediate vicinity of the plane at x = 0, and then left to themselves. The change in local particle density p(x,t)at position x and time t is described by the differential equation known usually as Fick's Second Law of Diffusion." The density profile is found by solving equation 3 subject to boundary conditions, (x # 0, I = 0) (4)p(x,t) = 0, {(x-+ fm,t > 0) and a normalization condition, equation 5. p(x,~)dx= nLW As material has an equal chance of diffusing to the left (-x) or right (+x), the mean displacement (x) is obviously zero. The developing profile from equations 3-5 is a normal distribution centred at x = 0: Combining equation 6 with the definition of the mean-square displacement, equation 7, +a, (x') = n-l J-m p(x,t)x2 dx (7) gives Einstein's equation (equation 1) after integration.* All real particles do, of course, interact strongly at close range. The theoretical position can be realized in the laboratory, however, if we imagine that the particles are so widely distributed in the y-z plane that pairs have negligible chance of colliding during the time-scale of observation. lo A. E. Fick, Phiios. Mag., 1855, 10, 30. Brownian Dynamics with Hydrodynamic Interactions A weakness in Einstein's original derivation, pointed out by Fiirth," is the necessity to invoke a time interval T which is small compared with t, but nevertheless of such magnitude that movements executed by a particle in two successive intervals T are considered as mutually independent.When the time for particle motion is short, this assumption is no longer valid. Under these circumstances, (x') is properly given by' '*12 equation 8 (x') = 2D[t -rnp + exp(-t/rnp)] (8) where m is the particle mass, and p is a coefficient of mobility defined below. Einstein's equation holds for t >> mp.This lower limit of time validity increases with the square of the particle size (as m K d3 and p K a'); it is ca. lo-' s for a 1pm neutrally-buoyant particle in water at room temperature. B. Friction and Mobility Coefficients.-From chaotic Brownian motion, we now turn to the subject of steady hydrodynamic flow.13 When a small constant force F is applied to a macroscopic body immersed in a hydrodynamic fluid, it rapidly attains a constant velocity v given by v = pF (9) where p is a mobility coefficient.Under steady-state conditions, the applied force is exactly counterbalanced by a frictional force J So the friction coefficient 6 is simply the reciprocal of p. Both are related to the size and shape of the body, and the viscosity of the fluid, and the above equations provide the physical basis for determining macromolecular size and shape from techniques such as centrifugation or electrophoresis. Although strictly applicable only to objects in the macroscopic domain, there is a long history of successful application of hydrodynamic theory down to particles of the size of sucrose molecules.Whether one chooses to work in terms of friction or mobility coefficients is largely a matter of convenience. The distinction is trivial in equations 9 and 10,but more complicated for objects of arbitrary shape in the vicinity of other like objects. The magnitude of the friction (or mobility) coefficient then depends on the forces acting on all the objects immersed in the fluid, and mathematically this means that scalar quantities p and < must be replaced by tensors p and <. Analogously, the scalar diffusion coefficient is replaced by a diffusion tensor D. C. Langevin's Equation.-The one-dimensional motion of an isolated Brownian R. Fiirth, Z. Phys., 1920, 2, 244. l2 L. S. Ornstein, Proc. Amsf.,1918, 21, 96.l3 J. Happel and H. Brenner, 'Low Reynolds Number Hydrodynamics', 2nd edn., Noordhoff, Leiden, 1973. Dickinson particle is described by the simple Langevin equation14 m(dv/dt) = -6v + R(f) (1 1) where R(t)represents a random force due to solvent collisions. Equation 11 is an example of a stochastic equation of motion. It is nothing more than Newton’s equation of motion (mass x acceleration = force) plus a random term. The latter is normally assumed to satisfy two conditions: firstly, that the process R(t) is Gaussian, and, secondly, that its correlation time is infinitely short, i.e., where t, and t, are two times, G, is a constant, and 6(x) is the Dirac function [defined as sG(x)dx = 1 for x = 0, 6(x) = 0 for x # 01. It turns out that the Gaussian assumption holds for a particle of mass much larger than that of the solvent molecules, a condition which is readily satisfied for a globular protein in water.The Langevin approach enables a link to be formed between a statistical quantity R and a hydrodynamic quantity 4,leading to a formal expression for the diffusion coefficient (in velocity space) D,: lltzDv = 1; (R(O)R(t)) dt = kq (13) From equations 12 and 13 we see that G, = (kT/n)c,and so the stochastic term is given by Equation 14 is one of the most fundamental relationships in statistical mechanics. Commonly known as the fluctuation4issipation theorem,’ it expresses a relationship between hydrodynamic dissipation and an ensemble average over fluctuations in the system.It is readily generalized to three dimensions, and the nature of the random force is independent of any systematic forces acting on the particles, whether they arise from interactions with other particles or from the influence of an external field. It also applies to rotational motion.15 3 Diffusion A. Phenomenological Coefficients.-As applied to one-dimensional mass transport, Fick’s First Law of Diffusion states6 that the mass flux J, in direction xis given by l4 S. Chandrasekar, Rev. Mod. Phys., 1943, 15, 1 (reprinted in: ‘Selected Papers on Noise and Stochastic Processes’, ed. N. Wax, Dover, New York, 1959). R. Kubo, Rep. Prog. Phys., 1966, 29, 255. For rotational applications see: J. McConnell, ‘Rotational Brownian Motion and Dielectric Theory’, Academic Press, London, 1980.Brownian Dynamics with Hydrodynamic Interactions where D is a phenomenological coefficient, and p is the local density of the diffusing species. In three dimensions, we have where V is the del operator defined by V = i(a/ax)+ j(a/i?y)+ k(i?/az) (1 7) and i,j, and k are unit vectors in x, y, and z directions. Consider a large number of identical particles far enough apart in a fluid that they do not interact in any way. Each particle requires six independent parameters [qi(i = 1,6)] to specify its instantaneous configuration. Three parameters define position (x,y,z), and three define orientation (cp,, (p,, cp,). When Fick’s Law is generalized to position-orientation space, each diffusive flux Ji (i = 1,6) is linearly related to each density gradient (ap/8qj)(j = 1,6) by the phenomenological equation where Dijis now an element in a 6 x 6 matrix called the diffusion tensor, and p is the instantaneous particle density in position-orientation space.We represent the change in particle position during a small time interval dt by the vector dr = idq, + jdq, + kdq, = idx + jdy + kdz (19) where dx, dy, and dz are the projections of the displacement vector onto the Cartesian axes. Similarly, the change in orientation is represented by the vector * dq = idq, + jdq, + kdq, = idq, + jdq, + kdcp, (20) where dq,, dcp,, and dq, denote projections onto the Cartesian axes. In the same way that D in equation 15 has a statistical interpretation in terms of the Einstein equation (equation l), we shall find that each of the elements in the diffusion tensor D defined by equation 18 also has a statistical interpretation.B. An Isolated Spherical Particle.-The simplest possible system consists of a single, hard, spherical particle immersed in a quiescent hydrodynamic medium. Let us suppose that, during some small but finite time interval, the co-ordinates of the particle centre have moved from (x, y, z) to (x + Ax,y + Ay,z + Az). In the absence of any external forces, the mean-square Brownian displacements in the three directions are given by ((AX),) = ((Ay)’) = {(Az)’) = 2DTA.t (21) * Unlike dr,dp is a vector only for infinitesimal displacements. Dickinson where DT is a scalar translational diffusion coefficient.For a sphere of radius a in a medium of Newtonian viscosity q,DT is given by the well-known Stokes-Einstein equation where cTis the translational friction coefficient. During the same time interval At, the Brownian sphere will also have rotated through angles Aqx,AT,, and Aq, about axes parallel to the x, y, and z directions, respectively. Mean-square displacements are given by where DRis the rotational diffusion coefficient: DR= kT/8qa3 (24) Comparing equations 22 and 24, we see that rotational motion decreases much more strongly with increasing particle size than does translational motion. In position-orientation space, the generalized Einstein equation has the form (AqiAqj> = 2DijAt (25) In matrix notation, the diffusion tensor D = [Dij]for a single sphere has non-zero elements only along the leading diagonal, i.e., The off-diagonal elements are all zero because each of the three translational and three rotational degrees of freedom is completely independent for a single spherically-isotropic particle: (AriArj> = <AriA(cpi) = (AcpiAcpj> = 0 (1 < i # j < 3) (27) C.An Isolated Non-spherical Particle.-Irrespective of the shape of a particle, the diffusion matrix [Dij] is always symmetric and positive-definite (i.e., D = Dt, where -f denotes the matrix transpose).* The 6 x 6 diffusion matrix in position- orientation space can be partitioned into four 3 x 3 submatrices: * A symmetric tensor (matrix) is a square matrix which is equal to its transpose.The franspose matrix is formed by interchanging rows and columns. A symmetric tensor T is positioe-definite if Y' T.Y > 0 for any non-zero vector Y. (See Appendix for summary of rules of tensor multiplication.) Brownian Dynamics with Hydrodynamic Interactions D' Dct = LDc D' D' is a (symmetric) translational diffusion tensor of the form: It describes the translational Brownian motion of some point P rigidly fixed in the particle according to the equation (AriArj> = 2D::At (ij= 1,3) (30) where the superscript P on D$ now denotes the fact that, since a body rotates as it moves, the translational diffusion coefficient depends on the location of P. Analogously, the (symmetric) rotational diffusion tensor D' governs the rotational Brownian motion according to (AqiAcpj) = 2DijAt (ij= 1,3) (31) And a coupling tensor D" describes the correlation between translational and rotational displacements: (AcpiArj) = 2Dc!At (ij= 1,3) (32) Again the superscript P on D$ denotes a dependence of the tensor on the location of point P.The values of D' and D"corresponding to two different points on the particle, P and Q, are related byI6 where rpQis the position vector from P to Q, and x represents the usual vector cross-product (see Appendix). Any rigid particle possesses a unique point 0,its centre of diffusion, for which D' is symmetric (i.e.,Dco= Pot).The exact location of 0 is given by rw = [(TrD')/-Or]-' E: DcQ (35) where TrD' represents the trace of tensor D' (the sum of terms on the leading diagonal, i.e., D; + D;, + Di3).The quantity / in equation 35 is the unit tensor (identity matrix) defined by Dickinson (36) The quantity c is a totally antisymmetric third-rank tensor (3 x 3 x 3) known as the Levi-Civita density; in matrix notation, E~~~is defined by ~123= ~312= ~231= 1, ~131= ~321= ~213= -1, and zero elsewhere.For a particle that is spherically isotropic,* Brownian diffusion is completely described by just two scalar coefficients DT and DR: 0'0 = DT/, D' = OR/, Dco= O (37) But, even for a sphere, we note that D"vanishes only at the centre. Many globular proteins are roughly ellipsoidal in shape with diffusion tensors of the form: 3 Dco= 0 (38) 0 0 0% As with spheres and ellipsoids, so for a large class of particle shapes: at the centre of diffusion, translational and rotational motions are uncoupled, with corresponding fluxes, J' and J', given by J' = -0'' (dp/d~), J' = -0' (dp/d~p) (39) A non-vanishing coupling tensor is associated with a screw-like behaviour of the diffusing particle.With biopolymers possessing helicoidal symmetry about a single axis (say the x-axis), the coupling of translational and rotational motions is described by a single scalar coefficient Dc defined byI6 where the algebraic sign on the right of equation 40 depends on the handedness of the enantiomorph. Interesting academically, but apparently not biologically, is the class of isotropic helicoids,' which includes the spherical isotropic particle as a special case (Dc = 0).D. A Spherical Particle Near a Plane Surface.-Moving from the unbounded fluid medium, let us now consider the case of a spherical Brownian particle near a solid plane wall. Making the spatial environment of the particle asymmetric has its effect * In this special category of geometrical objects are the sphere and the five regular polyhedra (tetrahedron, cube, efc.). l6 H. Brenner, J. Colloid Interface Sci., 1967, 23, 407. Brownian Dynamics with Hydrodynamic Interactions on the diffusion tensor D. If the x-axis passes through the particle centre 0 perpendicular to the surface (see Figure l), diffusion in the x-direction is slower than in the y-or z-directions (D\? < D:: = D\%, and depends on the distance I between 0and the surface.At the same time, rotational diffusion about the x-axis is faster than rotational diffusion about the y-or z-axes (D;l > D;2 = Dj,). The origin of these differences lies in a spatially-dependent slowing down of diffusive motion, translational, and rotational, due to hydrodynamic forces between the surface of the colloidal particle and the boundary wall (vide infra). In crude physical terms, we might say that the particle moves slower than it would do in an unbounded fluid because some extra thermal driving energy has to be used to push fluid from (or pull fluid into) the region between the particle and the surface. This extra work against the fluid resistance becomes greater with decreasing separation Y Figure 1 Sphericalparticle with centre 0 at dislance 1from a plane surface.The translational displacement Az is positively correlated with the rotational displacement A(py With the isolated sphere, translational and rotational Brownian motion were completely independent (see equation 37). This is no longer the case in the presence of the surface. For example, a positional displacement Az becomes positively correlated with an angular displacement Aq,,, and the correlation is total in the limit of the sphere touching the plane (/+a), when the particle can only move in a rolling motion. The single-particle diffusion tensor therefore has the following form: Dickinson with D,, = D,,, D,, = D,,, D,, = D,, = -D3, = -D53.4 Hydrodynamics In connection with the Brownian movement of small particles, we are nearly always concerned with hydrodynamics at low Reynolds number. That is to say, viscous forces from local shearing motions of the fluid are assumed to predominate over inertial forces associated with acceleration of the fluid elements. Strictly speaking, inertial forces exist to some small finite extent in all moving systems, but they can be neglected for the systems considered here. A. Isolated Particles.-Frictional forces and torques exerted on a rigid macroscopic body are linearly related to translations and rotations by the friction tensor 5. This 6 x 6 matrix can be decomposed into a 3 x 3 translational tensor c', a 3 x 3 rotational tensor c', and a 3 x 3 coupling tensor <':I7 (43) If we consider a point P on an isolated particle moving in an otherwise undisturbed fluid with rotational velocity o and translational velocity vp, the frictional force on the particle is given by where as before the superscript P denotes a dependence on the location of P.* The frictional torque about P is given by where <',like <", is position-dependent. Every isolated rigid particle has a unique geometrical point, 0,its centre of reaction, for which the coupling friction tensor is symmetric, i.e., CEO = rot (46) * When the particle moves irrotationally, all points necessarily have the same translational velocity.Asfis also independent of where P is located, this means that c' has no position dependence.(Note the contrast here with 0' which is position-dependent.) l7 H. Brenner, Gem. Eng. Sci., 1964, 19, 599. 43 1 Brownian Dynamics with Hydrodynamic Interactions But the centre of reaction is not necessarily located in the same place as the centre of diffusion. At low Reynolds number, diffusion and friction tensors are related by a generalized Stokes-Einstein equation D = kT5-l = kTp (47) where IJ. is the inverse oft;. In terms of submatrices, we have: The individual components of D and t; are related by18 We note in equation 51 that D' and ccvanish at the same location. For a nonskew particle (GCo = 0), the centres of diffusion and reaction coincide exactly. For screwlike particles (cco# 0),it is shown by WegenerI8 that the centres of diffusion and reaction are generally in different places.There are just two scalar friction coefficients for spherically isotropic particles: For a sphere of radius a, they take the simple form ST = 67tqa, GR = 8nqa3 (53) consistent with equations 22 and 24 for DT and DR.In practice, it is found that translation-rotation coupling effects for a single particle are not too important (less than a few per cent), so long as the object has some symmetry elements.Ig Only for highly asymmetric particles (e.g.,a half-turn of helix) is the effect very significant. With more than one particle, however, translation-rotation coupling is important even for spheres, as we shall see below. B. Boundary Conditions: Stick versus Slip.-The Stokes formulae for the translational and rotational coefficients of a spherical particle (equation 53) are implicitly based on so-called 'stick' ('non-slip') boundary conditions.That is, the relative tangential velocity component of fluid in contact with the rigid particle surface is taken as zero. In practice, it is found that stick boundary conditions are correct for large, solid, and impermeable particles immersed in a viscous medium. '' W. A. Wegener, Biopolymers, 1981, 20, 303. l9 J. M. Garcia Bernal and J. Garcia de la Torre, Biopolymers, 1980, 19, 751. Dickinson For a smooth spherical particle, at whose surface perfect slip occurs (e.g., a gas bubble), the friction coefficients are given by For a liquid drop of finite viscosity qi immersed in an unbounded continuum of viscosity qo, the value of cT lies between the pure slip and pure stick limits: In many real situations, the presence of surface-active agents makes the drop surface viscoelastic, in which case cTand cR approach the non-slip limits.20 Protein molecules are relatively small, non-spherical, flexible, and partly permeable to solvent; and so one might at first sight presume that rigid-sphere, stick boundary conditions would be out of the question.This is fortunately, however, not the case. In fact, the hydrodynamic properties of many protein molecules are adequately represented by effective hard-sphere models, and the roughness of the macromolecular surface usually means that stick boundary conditions are appropriate.Nevertheless, it needs to be pointed out that, as over half the water in contact with the protein surface is indistinguishable experimentally from pure solvent, it has been suggested2' that a large proportion of the slip surface (sic) is between the vicinal water molecules and the protein, and not outside the first monolayer as commonly assumed. Globular proteins are somewhat porous due to imperfect packing of subunits and topological surface irregularities. One possible wayt2 of mimicing porosity is to relax partially the stick condition through a 'slipping length' defined by where u, is the tangential velocity component, and du,/dn is its derivative normal to the surface. The effective hydrodynamic radius aeffof a sphere with this boundary condition is where 5 -, co in the pure-slip limit (cf: qi = 0 in equation 55).According to Wolynes and McCammon,22 the ratio 5/a is typically 0.15 for a porous protein. Summarizing, then, we should treat stick boundary conditions as the norm, with full slip conditions only considered for species as small as solvent molecules, simple ions, or polymer segment^.^ Use of semi-empirical partial-slip boundary conditions has some intuitive appeal, but it is not rigorous except at a fluid-fluid interface. With solid particles, any relaxation of normal stick boundary conditions 2o V. G. Levich, 'Physicochemical Hydrodynamics', Prentice-Hall, Englewood Cliffs, N.J., U.S.A., 1962. F. M. Richards, Ann. Rev. Biophys. Bioeng., 1977, 6, 151.22 P. G. Wolynes and J. A. McCammon, Macromolecules, 1977, 10, 86. 23 P. G. Wolynes and J. M. Deutch, J. Chem. Phys., 1976,65,450, 2030. Brownian Dynamics with Hydrodynamic Interactions is really an admission that continuum hydrodynamic theory has broken down for the problem under investigation. C. Hydrodynamic Interactions between Particles.-Consider two freely-rotating particles 1 and 2 acted on by forces F,and F, respectively. The particle velocities are given by where pij(ij = 1,2) are translationalpair mobility tensors. The tensors pi and pi2 are equivalent to the single-particle tensors described above; p:, and pil are new tensors arising specifically from the hydrodynamic pair interaction. For a pair of identical spheres with centres at separation r, the tensors have the general form pij(r) = aij(r)(rr/r2)+ Pij(r)[I -(rr/r2)] (60) where rr denotes the 3 x 3 dyad corresponding to vector r, and afj(r) and p:j(r) are scalar analytic functions of r = Irl.Mobility expressions are known24 as a function of r to high accuracy with stick boundary conditions. The functions at and p' can be expanded as series in powers of (a/r),24 but at very close separations the series converge slowly and asymptotic expressions must be used instead.' The functions aij(r) and pij(r) in the form25 a:,{r) = p:j(r) = Values of the expansion coefficients [a," (= at2), at2 (= a,"), b," (= bi2) and bt2 (= b:')] are listed in Table 1 up to n = 5. The leading selJlterms, a;' and b; ', are simply the reciprocals of the single-particle friction coefficients mentioned previously. Higher order self-terms represent the influence of the second particle on the single-particle friction coefficient of the first.Notice that some of the coefficients are exactly zero. It is the cross-terms which are most important in Brownian kinetics, since these determine the relative motions of the diffusing species. Under stick boundary conditions, the leading cross-terms, aA2 and bA2, combine to give what is commonly described in hydrodynamics as the Oseen tensor:26 D\2 = kTpi2 = (kT/8nqr)[/+ (rr/r2)] (63) 24 D. J. Jeffrey and Y.Onishi, J. FluidMech., 1984,139,261. R. Schmitz and B. U. Felderhof, Physica, 1982, 113A, 90, 103; 1982, 116A, 163.25 R. B. Jones and G. S. Burfield, Physica, 1985, 133A, 152. 26 C. W. Oseen, 'Hydrodynamik', Akademische Verlagsgellschaft, Leipzig, 1927. Dickinson Table 1 Mobility expansion coeflcients a? and b? (n = 0,5) for (A) stick and (B) slipboundary conditions a: ’ a,! b,!’ b,!’ n A B A B A B A B 0 213 1 1 1 213 1 1/2 1/2 1 0 0 -213 0 0 0 113 0 2 -5/2 -1 0 0 0 0 0 0 3 1113 -1 2512 2 -17124 118 0 0 4 7 -1 -5 6 -516 115 0 0 5 -16713 -5 -13112 14 -2318 114 189164 3/16 In the original derivation of equation 63, interacting elements were represented as point sources of friction; better approximations can be regarded as allowing for the effect of finite particle size on the hydrodynamic flow field. Extension of equation 63 to include expansion coefficients a:’ and biz leads to the equation of Rotne and Prager:’’ D:’ = (kT/8qr)([/ + (rr/r2)] + (2a2/3r2)[/ -3(rr/r2)]) (64) While the Rotne-Prager tensor is a better approximation than the Oseen tensor, both are unsatisfactory at close separations (r +24.They overestimate the tendency of particles to move towards (or away from) one another along the line of centres; equation 63 is out by a factor of 10 for r -2.01~.In the limit r+ 24 the component * of 0:’ along the line of centres actually vanishes (the relative friction coefficient diverges to infinity). This means that, in the absence of colloidal attractive forces, perfectly hard spheres can never touch (coagulate) in a continuum solvent! The vanishing relative diffusion coefficient is due to very large velocity gradients in the gap between spheres with stick boundary conditions.The divergence in the friction coefficient with slip boundary conditions is much weaker, having the form of a logarithmic ~ingularity.’~ With more than two particles, things get more complicated, but the results can be expressed in a formally similar way. Mazur and van Saarloos have givenz8 a general scheme for an arbitrary number of spheres to any order in (a/r).Explicit expressions to order (~/r)~are obtained for the mobility tensors, rotational as well as translational, and to this order three- and four-body interactions are included. The dominant contributions to translation from clusters of N spheres (N 2 2) are of order r43N-5).As with the two-sphere case, certain powers of (a/r)are completely absent [e.g., there is no term of order (a/r)’ in the expression for p’]. D. Translation-Rotation Coupling.-When several bodies are immersed in a * It is probably worthwhile emphasizing here that subscripts on D‘,2refer to particles 1and 2, and not, as earlier (see equations 26 and 29), to directional components of the diffusion matrix. 27 J. Rotne and S. Prager, J. Chem. Phys., 1969,50, 4831. 28 P. Mazur and W. van Saarloos, Physica, 1982, 115A,21. Brownian Dynamics with Hydrodynamic Interactions hydrodynamic medium, the frictional forces and torques on each one of them depends on the translational and rotational movements of all the others.Generalizing equations 44 and 45 to an N-sphere system, the frictional force and torque on particle i is given by f; = -cN (gj -Yj + 4fj ' Wj) J N 'ti = -1(q;j vj + 4:j Wj) i where the superscripts t, c, and r denote hydrodynamic couplings between translation and translation, translation and rotation, and rotation and rotation, respectively. As with the single particle (equation 48), the combination of friction submatrices leads to a grand friction matrix related to grand mobility and diffusion matrices:29 Individual components of [Di.i] and [Lij] are related by sets of equations equivalent to those connecting the single-particle diffusion and friction tensors (equations 49-51).The leading cross-terms in the diffusion tensor for a pair of rigid spheres are as follows: We note that interparticle translation-rotation coupling (equation 69) is of shorter range than rotation-rotation coupling (equation 70), but longer range than translation-rotation coupling (equation 68). With slip boundary conditions, the particles rotate freely, and so we have D:2 = Di2= 0. E. Particles Near a Plane Surface.-Since hydrodynamic interactions are of long range in comparison to particle size, the properties of particulate systems are strongly affected by boundary walls. The problem of one sphere in the vicinity of a plane surface is a limiting case of the two-sphere problem. It therefore provides a convenient system for testing the theories of hydrodynamic interaction in the laboratory.Recently, Ambari and co-workers have measured 30 the magnitude of the modified Stokes forcef, exerted on a macroscopic sphere (a = 0.435 mm) with centre 0kept in magnetic levitation at a fixed distance lfrom the surface (see Figure 29 D. W. Condiff and J. S. Dahler, J. Chern. Phys., 1966,44, 3988. 30 A. Ambari, B. Gauthier-Manuel, and E. Guyon, J. Fluid Mech., 1984, 149, 235. 436 Dickinson 2) using an optical feedback system. As the sphere approaches the wall with speed u,, there is a change in frictional force given by f,= -6qau,6(~) (71) where &a= 1 -a is the surface-to-surface separation. Within the experimental uncertainties, the data are found to agree exactly with theoretical equations derived for the cases of smal131 and large32 separations: 1/~-(lns)/5 + 0.9712 + .. . (E --+ 0) (72)1 + 9/8(1 + E) + ... (E % 1) The expression for close separations comes from lubrication theory,' and the long- range formula is an Oseen-type representation. With the macroscopic sphere studied e~perimentally,~' the smallest value of E corresponded to a surface-to- surface separation of ca. 8 pm,well beyond the effective range of London/van der Waals attractive forces. (With particles of colloidal size, of course, this would not be the case.) At infinitesimally close separations, the frictional force diverges to infinity and the corresponding diffusion coefficient vanishes. fx Figure 2 Macroscopic sphere of radius a with centre 0 at distance 1from a plane surface.As the sphere approaches the wall at speed v,., it experiences a frictional force f,. at surface-to- surface separation &a There is less viscous resistance parallel to a plane wall than perpendicular to it. For a sphere moving parallel to a plane surface with stick boundary conditions, the frictional force& is related to the speed uy by33 f,= -67cqavy/[1-(9a/161) + (a/21)3-. . .] (73) 31 R. G. Cox and H. Brenner, Chem. Eng. Sci.,1967, 22, 1753. 32 H. A. Lorentz, Abhandl. Theor. Phys. (Leipzig), 1907, 1, 23. 33 H. Faxen, Arkiv. Mat. Astron. Fys., 1923, 17, No. 27. 437 Brownian Dynamics with Hydrodynamic Interactions The particle rotates with an angular velocity 0, = (3vY/32a)(a/l)*+ . .. (74) the direction being the same as that for simple rolling along the wall. We see from equation 74 that the strength of single-particle hydrodynamic translation-rotation coupling drops off rapidly with the particle-surface separation. Expressions have recently been presented 34 for an arbitrary number of spheres in the vicinity of a plane wall. The main point to note is that, for the same surface-to- surface separation, the hydrodynamic effect of the wall is considerably greater than that of another spherical particle. And even more so for a cluster of particles between two plane walls.13 5 Brownian Dynamics Simulation Just as the motion of atoms in a simple liquid can be simulated by molecular dynamics, the motion of particles in a colloidal dispersion can be computed numerically by Brownian dynamics.(In the purely hydrodynamic regime, where there is no Brownian movement, the term Stokesian dynamics *seems appropriate.) A. Algorithm of Ermak and McCammon.-In a system with colloidal inter- particle forces, the Langevin-type equation of translational motion has the form3' where miis the particle mass associated with index i, uiis the velocity component in direction i, Fiis the sum of external and interparticle forces acting in direction i, and the sum is over all 3N translational degrees of freedom (cf: the one-dimensional Langevin equation, equation 11). The right-hand-side of equation 75 is a sum of three terms: a frictional force, a systematic force, and a stochastic force.The stochastic term depends on a set of coefficients {a:j},defined by and a set of random numbers {xj>with a Gaussian distribution, (xi(0)xj(t))= 26ij6(t) (77) where ?jij is the Kronecker delta (= 1 for i =j,otherwise zero). * The author first heard this term used in public by Professor J. F. Brady at the Euromech symposium in Cambridge in April 1985 (see G. Bossis and J. F. Brady, J. Chem. Phys., 1984, 80, 5141). 34 C. W. J. Beenakker, W. van Saarloos, and P. Mazur, Physica, 1984, 127A,451. 3s J. M. Deutch and I. Oppenheim, J. Chem. Phys., 1971,54, 3547. Dickinson A Brownian dynamics algorithm based on equations 75-77 was derived by Ermak and M~Cammon.~~ Particle displacements are given by 3N 3N Ari = C(dD:'j/&j)At + (kT)-'CDiFYAt + Ri(Dl'j,At),(i = 1,3N) (78)j= 1 j= 1 where Ar = ri -ro is the change in particle co-ordinate during time-step At, Riis the stochastic displacement in direction i, and superscript 0 denotes that the quantity is evaluated at the beginning of the step.It is important to note that in the algorithm defined by equation 78 instantaneous particle velocities are not specified as such. Although the interval At is long compared with the characteristic time associated with the solvent molecule motion, it must be short enough for quantities Dli and Fjto be effectively constant during the simulation time-step. The stochastic displacements are calculated from the set of equations: I Ri(A,t)= 10ijXj j= 1 (Xi) = 0, (XiXj) = 26ijAt (82) Because of the square root in equation 80, the calculation of uij from 0::is generally the most time-consuming part of the simulation.The application of the algorithm of Ermak and McCammon to a particular problem requires specification of the configuration-dependent non-hydrodynamic forces {Fj) associated with the various physico-chemical interactions between the Brownian particles. For instance, with electrostatically-stabilized colloidal particles, the contributions to {Fj>come from derivatives of the DLVO potentials of mean force at the appropriate pair separation^.^^ If a particle is also subject to an external force (e.g., gravity), this is added in as well. A DLVO-type potential is suitable for describing the spherically-symmetric part of the protein-protein interaction, but there may also be asymmetric contributions to the protein potential arising, say, from highly charged, localized patches on the folded macromolecular surface.Once spherical symmetry is lost, particles are subject to torques as well as forces, which means that we must also consider the rotational Brownian motion (see next section). There are some important technical differences between molecular dynamics and Brownian dynamics simulations. Molecular dynamics is based on Newton's equations of motion: energy is therefore conserved, and trajectories are time- reversible. On the other hand, a stochastic equation of motion like equation 78 36 D. L. Ermak and J. A. McCammon, J. Chem. Phys., 1978,69,1352.37 J. Bacon, E. Dickinson, and R. Parker, Faraday Discuss. Chem. Soc., 1983,76, 165.G.C. Ansell, E. Dickinson, and M. Ludvigsen, J. Chem. Soc., Faraday Trans.2, 1985,81,1269. 439 Brownian Dynamics with Hydrodynamic Interactions neither has a definite solution nor does it conserve energy. So, whereas total energy fluctuations can be used to monitor the efficacy of a molecular dynamics calculation, there is no such consistency check with Brownian dynamics. To ensure that particle trajectories are sufficiently accurate for the purpose in question, all one can do is demonstrate that average statistical properties are independent of the size of the integration time-step. In both types of simulation, contributions to forces (Fj} can be truncated at pair separations beyond a certain ‘cut-off’ distance.However, in Brownian dynamics, because the hydrodynamic interactions are of long range, it is difficult to justify a particular ‘cut-off’ distance beyond which they may be neglected. B. A Generalized Algorithm.-It is straightforward to generalize the above algorithm to include rotational Brownian motion and translational-rotation coupling. Proceeding along the lines of equations 65 and 66, one can write down a set of translational and rotational Langevin equations:38 In equation 84, Ii is the moment of inertia associated with index i, and Tiis the sum of external and interparticle torques acting in direction i. The equations 83 and 84 are interdependent since they share the same set of aij coefficients defined by Translational and rotational motions are only fully decoupled when <:j = 0 for all pairs of particles in the system.Let us now switch to a set of generalized co-ordinates qi (i = 1,6N) in 6N-dimensional position-orientation space (see equations 19 and 20). Combining equations 83 and 84 into a single expression, we get a generalized Langevin equation from which can be derived 39 a generalized ‘moving-on’ routine: 6N 6N Aqi = (dD$/8qj)At + (kT)-’ 1D$F:At + Ri(D$,Al) (i = 1,6N) (86) j= 1 j= 1 where Aqi = qi -qo is the change in generalized co-ordinate during At, and Fiis a generalized force component in direction i. Indices i andj from 1 to 3N refer to translation; those from 3N + 1 to 6N to rotation.For spheres of uniform surface roughness, Dijis independent of orientation, and so we have 38 P. G. Wolynes and J. M. Deutch, J. Chern. Phys., 1977, 67, 733. j9 E. Dickinson, S. A. Allison, and J. A. McCammon, J. Chem. SOC.,Faraday Trans. 2, 1985,81, 591. Dickinson In a system with rotational Brownian motion and translation-rotation coupling, the stochastic displacement terms are given by equations 79-82 as before, but Di; is replaced by the grand diffusion tensor D$. C. Choice of Hydrodynamic Approximation.-A few words seem appropriate on the forms to be adopted for pij(and therefore Dij)in a simulation of spherical Brownian particles. The Oseen and Rotne-Prager approximations, equations 63 and 64, are computationally convenient, but they break down at close separations (Y --+ 2a) where lubrication theory must be used.Fortunately, in many systems of realistically modelled Brownian particles, the problem is less severe than with simple hard spheres; this is because pairs of particles are in fact never allowed to get very close due to the influence of electrostatic interparticle repulsive forces. One disadvantage of crude pairwise-additive Oseen hydrodynamics is that it sometimes leads to a N-particle diffusion tensor that is not positive-definite. This is disastrous from the simulation standpoint, since it implies that the stochastic weightings from equations 79 and 80 are mathematically complex, and therefore physically absurd. One way round the difficulty is to use3’ a truncated Oseen interaction with pi2 = 0 for r > rC,where the effective cut-off distance rc is a decreasing function of the local particle concentration. The Rotne-Prager tensor is well-behaved, insofar as it does not suffer from non-positive-definiteness.And, like the Oseen tensor, it gives a computationally convenient divergenceless diffusion tensor (V Dij = 0), so that gradient terms in equations 78 and 86 need not be evaluated, thus saving some calculation time. As mentioned previously, explicit expressions for pij are available2* to order (~/r)~,but their widespread use within Brownian dynamics simulations is likely to be restricted owing to the computational expense of having to sum over all groups of 3 and 4 particles for each time-step. In any case, recent cal~ulations~~ cast doubt on whether in practice expressions to order (a/r)’ give results in multi-particle systems that are any more reliable than those to order (a/~)~(Rotne-Prager approximation).One compromise solution41 is to use an effective hydrodynamic pair tensor which allows for multi-body interactions implicitly via one or more empirical screening constants which depend on the local particle concentration. The idea here is that instead of having a sharp hydrodynamic c~t-off,~~ one has a more gradual screening of the normal pair interaction. We note, however, that the concept of hydrodynamic screening is only strictly applicable to immobile particles immersed in a viscous medium.42 The justification for using simple hydrodynamic approximations in many Brownian dynamics problems comes from the fact that the rate processes are often only weakly affected by changes in the hydrodynamic expressions.Only when 40 G. D. J. Phillies, J. Chem. Phys., 1984, 81, 4046. 41 I. Snook, W. van Megen, and R. J. A. Tough, J. Chem. Phys., 1983,78, 5825. 42 C. W. J. Beenakker, Faraday Discuss. Chem. SOC.,1983, 76, 240. Brownian Dynamics with Hydrodynamic Interactions particles spend most of their time very close together (r -2a 4 a) does one need to be particularly careful about the exact form of the hydrodynamic intera~tion.~~ 6 Protein Dynamics So much for the principles; now to the practice. Most of what follows is forward- looking: its aim is to point out what seems feasible in connection with the application of Brownian dynamics simulation to protein diffusional motion.In the case of the enzyme-substrate problem, some progress has already been made, but, for the most part, the field is still virgin territory. The topics described below are not meant to form a complete list. They just represent a few relevant and interesting problems about which the author has become recently aware. The unifying theme is Brownian dynamics with hydrodynamic interactions. A. Enzyme-Substrate Encounters.-The enzyme-substrate combination is just one of several types of ligand-receptor pairs involved in biological action.43 In the simplest possible model, enzyme and substrate molecules are represented as spherical Brownian particles with ‘reactive patches’ on parts of their surfaces.The overall rate of many biochemical processes is determined by the kinetics of an initial diffusional encounter between enzyme and substrate molecules, and the reaction is said to be ‘diffusion controlled’. Amongst the factors that can affect the rate of reactive binary collision are the charge distributions on both molecules, the hydrodynamic interactions between the particles, the orientational dependence of reactivity, and intramolecular structural fluctuations at and near the ‘active site’. Some limited progress has been made in incorporating these effects into analytic kinetic but it seems likely that the detailed distinguishing features of complicated biochemical processes will be amenable only to numerical simulation methods.An appealing feature of the simulation approach is the ability to make steady and systematic progress by successively refining the assumed model. Let us consider the reaction sequence E + S ES 5products (88) where E and S stand for enzyme and substrate respectively, and k, k’ and k” are rate constants. Under steady-state conditions (d[ES]/dt = 0), the transformation rate into products is described by an effective rate constant keff = kk”/(k‘ + Id’) (89) We have keffz k for a diffusion-controlled reaction (k” + k’). When E and S are spherically-symmetric, non-interacting particles, the reaction is described by the bimolecular Smoluchowski rate constant45 k, = 4xr,D (90) 43 J.A. McCammon, S. H. Northrup, and S. A. Allison, Com. Molec. Cell. Biophys., in press. 44 D. F. Calef and J. M. Deutch, Ann. Rev. Phys. Chem., 1983, 34,493. 45 M. V. Smoluchowski, Phys. Z., 1916,17,557. Dickinson where D is a diffusion coefficient, and re is an encounter distance (roughly equal to the sum of particle radii). Putting in some allowance for hydrodynamic effects, together with a general centrosymmetric potential of mean force u(r)between E and S, leads to the modified Smoluchowski expre~sion,~ 3*46 where D is now a function of the pair separation r (equivalent to the component of the pair diffusion tensor D:2 along the line of centres). Equation 91 represents more-or-less the limit of the analytic approach.When the E-S interaction is more complex than that assumed above, the rate constant must be evaluated numerically by averaging over diffusional trajectories of the substrate in the field of a fixed enzyme target.43 To avoid having to simulate substrate paths which wander well away from the enzyme, the diffusion space around E is divided into two regions4' (see Figure 3). In the outer region (r > p), E and S are far enough apart for diffusion to be described adequately by equation 91; in the inner region (r < p), however, interactions have a more complicated orientation dependence, and therefore must be handled numerically. If each Brownian collision of S with the active site on E leads to reaction, then it can be shown47 that the rate constant is given by where k(p) and k(q) are the values of k from equation 91 with re = p and re = q, respectively.The quantity a represents the probability that a substrate molecule, starting at r = p, and free to diffuse in inner and outer regions, will react before reaching r = q. In the actual simulation, trajectories begin at r = p and terminate on reaction or at r = q. From the fraction of events leading to reaction is calculated the bimolecular rate constant k. Detailed analysis of particle trajectories provides information about the reaction mechanism, i.e. whether or not S is 'steered' into productive collisional orientations during the diffusional encounter. To take a particular example, McCammon and c~-workers~~*~~ have initiated a series of simulations of increasingly realistic models of the diffusion-controlled reaction of superoxide (09catalysed by the enzyme superoxide dismutase.The enzyme molecule was represented as a sphere of diameter 6 nm having two small reactive patches on opposite sides covering ca. 1& of the total surface area. A set of five charges within the model enzyme particle was used to produce an electrostatic field with monopole, dipole, and quadrupole components equivalent to those generated by all 76 charged groups on the real protein. The Oimolecule is modelled as a sphere of diameter 0.3 nm carrying a unit charge. Rate constants O6 S. H. Northrup and J. T. Hynes, J. Chem. Phys., 1979,71, 871. O7 S. H. Northrup, S. A. Allison, and J. A. McCammon, J.Chem. Phys., 1984,80, 1517. 48 S. A. Allison and J. A. McCammon, 1985,89, 1072. 49 S. A. Allison, G. Ganti, and J. A. McCammon, Biopolymers, 1985, 24, 1323. Brownian Dynamics with Hydrodynamic Interact ions Figure 3 Simulation diffusion space for the encounter between enzyme E and substrate S. Distances p and q are the radii of the inner and outer diffusion regions (see text). The shaded portion of E denotes the ’active site’ calculated from equation 92 were based on averages over several thousand trajectories withp = 30 nm and q = 50 nm, and the results were found to reproduce successfully the qualitative experimental features of the enzyme-catalysed reaction.” Using Debye-Hiickel theory to allow for electrostatic screening, it is calculated in agreement with experiment that the rate constant first increases, and then decreases to a plateau, as the ionic strength is increased (i.e. as electrostatic interactions become of shorter effective range).It is postulated43 that the initial increase in k is due to screening of long-range E-S repulsion, whereas the subsequent decrease arises from screening of the shorter-ranged non-central forces which act to steer the substrate into the active site. The above results for 0,+ superoxide dismutase refer to simulations in which hydrodynamic interactions were neglected altogether (DI2= 0). Using the Oseen tensor with slip boundary conditions,* it has been that including hydrodynamic interactions can lead to a reduction in simulated rate constant of ca.30% in the absence of E-S attractive forces. In a separate Brownian dynamics simulation of encounters between a spherical enzyme particle and a dumbell-dimer substrate particle using a constraints algorithm (see later), it was found 52 that the presence of hydrodynamic interactions does not much affect the steering enhancements, but does lead to a fairly uniform decrease in the overall reaction rate. As the structural complexity of enzyme and substrate molecules increases, it is clear that the reaction kinetics is increasingly affected by orientational considerations, as determined by the rotational Brownian motion and (with stick boundary conditions) the coupling between translational and rotational motions. * With a small substrate ion of solvent molecule dimensions, as is the case here, anything other than slip boundary conditions would seem inappropriate (see re$ 23 and 51).50 A. Cudd and 1. Fridovich, J. Biof.Chem., 1982,257,11443. E. D. Getzoff,J. A. Tainer, P. K. Weiner, P. A. Zollman, J. S. Richardson, and D. C. Richardson, Nature (London), 1983, 306,287. 51 R. Zwanzig and M. Bixon, Phys. Rev. A, 1970, 2, 2002. 52 S. A. Allison, N. Srinivasan, J. A. McCammon, and S. H. Northrup, J. Phys. Chem., 1984, 88, 6152. Dickinson B. Proteins at Electrode (and Related) Surfaces.-Protein electrochemistry offers the opportunity for controlled electronic communication with a wide range of biochemical processes. Using enzymes with redox-active sites, there is the possibility of converting electron movement into specific substrate transformations.The combination of immobilized glucose oxidase and a graphite electrode, for instance, has potential application in the amperometric determination of glucose in blood. Proteins appear to adsorb irreversibly at both synthetic and biological surfaces, and it has long been held the view that reversible electrochemistry involving proteins is not possible at conventional electrode surfaces. But, nevertheless, it is knowns4 that reversible protein adsorption can occur if the protein is rigid and the surface is hydrophilic, conditions which ought to be fulfilled in many electrochemical situations involving redox proteins. In fact, recent work with cytochrome c at a gold electrode has shown” that ‘good’ electrochemistry is promoted in the presence of certain bifunctional organic compounds at the electrode surface.Cytochrome c is an example of a robust low-molecular-weight globular protein whose biochemical function is to carry electronic charge between the catalytic and energy transduction sites on the membrane of an organism. Efficient kinetics of electZon transfer depends on the establishment of relatively long-lived, yet freely reversible, interactions of the protein, in uiuo with its physiological redox partners, and in vitro with the electrode surface. As a specific example, let us consider the conversion offerrocytochrome c (A) to ferricytochrome c (B) at a rotating disc ele~trode.’~ In the limit of fast mass transport, the reaction is represented by the scheme where k, is the rate of adsorption of reduced and oxidized forms, k, and k, are the potential-dependent rate constants for the forward and backward electron-transfer reactions, k, is the rate of desorption of reduced and oxidized forms, and p is the areal concentration of adsorption sites on the modified electrode.For reaction at a positive gold electrode (k, % k,), values of the kinetic parameters are estimated 56 to be: k, = 3 x 10-4 m s-’, k: = 50 s-’, k, = 50 s-l, and p = 1.2 mol m-2. Reversible protein binding enhances the overall rate of the electrode reaction at the modified electrode, but the reaction is very slow at the unmodified electrode. The importance of the chemical nature of the electrode surface in inducing reversible binding was demonstrated 57 by comparing electrochemistry at the polished ‘edge’ surface of pyrolytic graphite with that at the freshly-cleaved ‘basal plane’.Cytochrome c electrochemistry at the hydrophilic edge is well-behaved, but at the hydrophobic basal plane it is essentially irreversible. ’ 53 A. E. G. Cass, G. Davis, G. D. Francis, H. A. 0.Hill, W. J. Aston, I. J. Higgins, E. V. Plotkin, L. D. L. Scott, and A. P. F. Turner, Anal. Chem., 1984, 56, 667. 54 J. Lyklema, Colloids SurJ, 1984, 10, 33. 55 P. M. Allen, H. A. 0. Hill, and N. J. Walton, J. Electroanal. Chem., 1984, 178, 69. 56 W. J. Albery, M.J. Eddowes, H. A. 0.Hill, and A. R.Hillman, J. Am. Chem. Soc., 1981, 103, 3904.57 F. A. Armstrong, H. A. 0.Hill, and B. N. Oliver, J. Chem. Soc.. Chem. Commun., 1984, 976. Brownian Dynamics with Hydrodynamic Interact ions To simulate the redox protein +electrode problem by Brownian dynamics, one might proceed as follows. Assume that the redox protein is spherical (diameter -4 nm) and has two ‘patches’ on its surface: one for electron transfer (patch PE),the other for electrostatic binding (P,). Describe the protein interaction with the surface as a sum of (a) a long-range, centrosymmetric, screened electrostatic interaction and (b) a short-range, specific interaction between protein patch P, and binding sites SBon the surface (see Figure 4). Electron transfer is deemed to occur when protein patch P, gets within some distance 6 of the surface.Trajectories can be started with the particle centre at a distance 1 =p from the surface, and terminated upon reaction or when 1 2 q. The protein model just described can be thought of as a crude representation of, for instance, spinach plastocyanin, a photosynthetic ‘blue’ copper protein, much of whose net negative charge (at neutral pH) is taken to be conservatively localized at the side of the molecule.58 [By way of contrast, mitochondria1 cytochrome c has an overall positive charge located in close proximity to the electron-transferring haem edge (i.e.,for this protein P, and P, are coincident).] As far as the surface binding sites are concerned, these could easily represent positively-charged domains of stable chromium(rxx) complexes, since it has been shown58 that, even at low background electrolyte concentrations (< 0.01 mol dm-3), a chromium-modified graphite electrode is active towards plastocyanin.A reasonable value for the electron-transfer distance 6 probably lies in the range 0.5-1.5 nm. Figure4 Representation of spherical redox protein P in vicinity of plane electrode surface with binding sites S,. The two patches on P are associated with electron transfer (PE)and speciJic electrostatic binding (PB).The distances p and q are equivalent to the same quantities in Figure 3 As well as giving rate data, a Brownian dynamics simulation along the lines of that described above could be used to determine the importance of translation- rotation coupling effects as the redox protein diffuses at the interface.It is well- kn~wn~’-~’that rates of diffusion-controlled biological processes are faster in two dimensions than in three. In diffusion towards a small target of diameter d within a F. A. Armstrong, P. A. Cox, H. A. 0.Hill, and A. A. Williams, J. Chem. Soc., Chem. Commm., 1985, 1236. Dickinson large space of dimensionality n and diameter d,, Adam and Delbriick have expresseds9 the time to capture as where the function f,(d,/d) depends on the dimensionality n. For d,/d >> 1, f, is linear in d,/d for n = 3; it has the form ln(d,/d) for n = 2; and it is independent of d,/d for n = 1. So, for a constant diffusion coefficient D, there is a marked enhancement on going from n = 3 to n = 2, but little change in going from n = 2 to n = 1.To permit protein motion on a one- or two-dimensional biopolymer surface, the forces between protein and surface must be strong enough to guarantee adsorption, but weak enough to enable the molecule to diffuse. In this connection, small conformational fluctuations may play a role in the sliding of enzymes on the surface of linear or planar biopolymers.61 There are clearly similarities between diffusional processes at electrode and membrane surfaces. The electrostatic aspects of redox-protein binding to a negatively charged membrane surface has been demonstrated in a study62 of the oxidation kinetics of cytochrome c2 by bacterial photosynthetic reaction centres in unilamellar phosphatidylserine vesicles.In NaCl solution of ionic strength 0.1 mol dm-3 or less, the kinetic data suggest that the protein is restricted to the surface of a single vesicle, and encounters reaction centres by two-dimensional diffusion. The retarded oxidation rate at low electrolyte concentrations suggests that electrostatic interaction between the positive haem-cleft face of the protein and the negative membrane is sufficiently strong to restrict protein mobility. With increasing ionic strength, however, mobile counter-ions shield the electrostatic interaction, and so the protein diffuses more rapidly, though still mainly across the surface of the vesicle. Above 0.1 mol dm-3 NaCl solution, there is little protein-membrane association, and, since the binding regions are oppositely charged, the reaction rate falls-as it also does in solution, and in neutral phosphatidylcholine vesicles.62 The mechanism of protein diffusion at a membrane or electrode surface will depend on the nature of the protein-surface interaction.If the protein is only weakly bound, one would expect a ‘hopping’ mechanism. With stronger binding, a ‘rolling’ or ‘sliding’ mechanism would be more likely, the former being favoured by non-specific electrostatic protein-surface interactions, and the latter by specific interaction with a mobile entity at the interface. From equation 74, we note that the separation between protein and surface must be about one solvent molecule diameter (-0.3nm) or less for there to be appreciable translation-rotation coupling.C. Antibody Mobility and Antigen Binding.-Animals react adaptively against foreign bodies (‘antigens’) by synthesizing specific neutralizing agents (‘antibodies’). 59 G. Adam and M. Delbriick, in ‘Structural Chemistry and Molecular Biology’, ed. A. Rich and N. Davidson, Freeman, San Francisco, 1968, p. 198. 6o F. W. Wiegel and C. DeLisi, Am. J. Physiol., 1982, 243, R475. 61 E. Katchalski-Katzir, J. Rishpon, E. Sahar, R. Lamed, and Y. I. Henis, Biopolymers, 1985, 24, 257. 62 R. E. Overfield and C. A. Wraight, Biochemistry, 1980, 19, 3328. Brownian Dynamics with Hydrodynamic Interactions The commonest class of human antibody is immunoglobulin G (IgG), a Y-shaped glycoprotein (ca. 1.5 x lo5daltons) whose structure is illustrated schematically in Figure 5(a).Two identical globular regions known as Fab (after ‘fragment antigen binding’) are connected flexibly to a third globular region Fc (after ‘fragment crystallizable’). It appears that the hinge angle 8 can take any value in the range 10-180”. Binding can take place at two separate antigen sites, either on a single particle (bacterium or virus) or on two different ones [see Figure 5(b)]. In terms of protein structure, IgG consists of two equivalent ‘light’ protein chains (-2.3 x lo4 daltons) and two equivalent ‘heavy’ chains (-5.0 x lo4 daltons) linked by disulphide bridges and non-covalent interaction^.^^ Conventionally, the chains are subdivided into variable and constant domains as shown in Figure 5(c).When an antigen binds to the antibody, it nestles in a groove or cleft formed at the contact of the light and heavy chain variable domains. About 10or so amino-acid residues are thought to be involved in the binding region.64 A Figure 5 Representation of the structure of the antibody molecule IgG. (a)Simple three-centre hydrodynamic model. At the hinge, which flexibly connects globular fragment Fc to bindingfragments Fab, the angle 0 can take up a wide range of values. (b)The binding of IgG to two sites B on different antigen particles. (c) More detailed model showing light (L) and heavy (H)polypeptide chains. The light chain has one constant region (C,) and one variable region (VL);the heavy chain has three constant regions (Cnl, Cn2, and Cn3)and one variable region (VH).The hinge H consists of one or more disulphide interchain bonds. A is the antigen binding site Antibody flexibility has been demonstrated experimentally using nanosecond fluorescence spectro~copy.~~~~~ A fluorescent chromophore, specifically located at 63 G.W. Edelman and W. E. Gall, Ann. Rev. Biochem., 1969,38,415. 64 M. W. Steward, ‘Antibodies: Their Structure and Function’, Chapman & Hall, London, 1984. J. Yguerabide, H. F. Epstein, and L. Stryer, J. Mol. BioI., 1970, 51, 573. 66 C. L. Lovejoy, D. A. Holowka, and R. E. Cathou, Biochemistry, 1977, 16, 3668. 448 Dickinson the antibody binding site, is excited with a short pulse of y-polarized light, and fluorescence intensities polarized in the x-and y-directions, t;,and Fy respectively, are measured as a function of time t.A function expresses how much the orientation of the transition moment has changed between absorption and emission. It was found by Yguerabide et that A(t) could be approximated as a sum of two exponential terms: In equation 96, cps and cpL are short and long rotational correlation times, and A,,& andf, are constants. Taking cps = 33 ns as the correlation time of an isolated Fab fragment, a fitted value of cp, = 168 ns was interpreted as being due to the rotational motion of the antibody molecule as a whole. [An unhydrated rigid sphere with the volume of IgG is estimated to have a rotational correlation time of cp = (soR)-' = 44 ns.] It was inferred65 that the Fab portions of the intact antibody are free to rotate over an angular range of ca.33" in times of nanoseconds. More recently, Lovejoy and co-workers have found 66 similar correlation times, cps = 33 ns and cp, = 131 ns, again attributed to Fab segmental flexibility and global antibody rotation, respectively. The flexibility of the IgG molecule is related inter alia to the number of disulphide bonds in the hinge region [see Figure 5(c)]. In a comparison of intact and reduced antibodies, it was found 67 that reduction of the inter-heavy-chain disulphide bond increases significantly the internal flexibility of the IgG molecule. In modelling the IgG molecule as a flexibly connected three-sphere entity [as in Figure 5(a)], torsional interactions between Fab and Fc fragments must be consistent with the above correlation times, which can be computed directly in a Brownian dynamics simulation.The precise structures of antigenic determinants on most protein molecules are not known,68 but it does appear that interactions normally extend over some 3-44 nm2 of protein antigen surface. (See the report of an X-ray crystallographic determinati~n~~ of the complex between egg-white lysozyme and the Fab fragment of a monoclonal anti-lysozyme antibody.) In simulating antibody-antigen encounters, as with the enzyme-substrate problem, we can determine how sensitive is the rate constant to such factors as particle size and shape, specific and non-specific electrostatic forces, and so on.D. Other Protein Diffusional Processes In biological membranes, various lipids and proteins are able to undergo lateral "L. M. Chan and R. E. Cathou, J. Mol. Biol., 1977, 112, 653. 68 D. C. Benjamin, J. A. Berzofsky, I. J. East, F. R. N. Gurd, C. Hannum, S. J. Leach, E. Margoliash, J. G. Michael, A. Miller, E. M. Prager, M. Reichlin, E. E. Sercarz, S. J. Smith-Gill, P. E. Todd, and A. C. Wilson, Ann. Rev. Imrnunol., 1984, 2, 67. 69 A. G. Amit, R. A. Mariuzza, S. E. V. Phillips, and R.J. Poljak, Nature (London), 1985, 313, 156. Brownian Dynamics with Hydrodynamic Interactions diffusion within the two-dimensional bilayer str~cture.~’ There are two classes of membrane protein: peripheral and integral. The former associate with membranes predominantly through electrostatic interactions, and their diffusional motion resembles that near a charged electrode surface (vide supra). Integral proteins lie partially within the bilayer, where they exhibit extensive hydrophobic and electrostatic interactions with each other and the surrounding lipid molecule^.^ Depending on the conditions, membrane proteins can exist in various states of aggregation.Pair distribution functions derived from freeze-fracture pictures of lipid bilayers resemble7 those from theoretical models of two-dimensional fluids. Effective protein-protein potentials calculated from experimental pair distribution functions are available73 for use in simulations. The function of a membrane appears to be intimately related to its fluidity.74 In the protein dynamics context, the temperature-dependent viscosity of the bilayer modulates the activity of enzymes and transport-proteins by affecting their lateral and rotational motions.To simulate the Brownian dynamics of membrane proteins, an external potential field could be used to confine the particles to motion in a plane. It would be interesting to compare simulated diffusion coefficients with those measured in fluorescence and phosphorescence decay experiment^.^ 5*76 Diffusion-controlled encounters occur in a wide range of assembly and polymerization processes involving proteins. An important and well-studied example is the assembly of monomeric G-actin (4.2 x lo4 daltons) into polymeric F-actin, a fibrous building block of muscle tissue.The mechanism is supposed to involve a nucleation stage (trimers are the most likely candidates as nuclei), followed by a polymerization stage to give a helical stru~ture.’~~~~ Representations of monomeric and polymerized actin suitable for use in a simulation study are illustrated in Figure 6. Bonding between the roughly spherical G-actin molecules occurs through specific interactions of the type a-b and c4. Solvent conditions sensitively affect the position of the F-G eq~ilibrium,~~ with electrostatic factors particularly important. There is a ‘critical’ actin concentration for helical polymerization which decreases with increasing ionic strength, reaches a minimum at an optimum ionic strength of 0.1 M NaC1, and then goes up again with further addition of electrolyte.At pH values close to the isoelectric point (pH x 4.7), random globular aggregation is superimposed on regular polymerization to F-actin. Divalent cations appear to have both specific and non-specific effects on the 70 M. D. Houslay and K. K. Stanley, ‘Dynamics of Biological Membranes’, Wiley, Chichester, 1982. 71 G. Benga and R. P. Holmes, Prog. Biophys. Molec. Biol., 1984, 43, 195. 72 L. T. Pearson, S. I. Chan, B. A. Lewis, and D. M. Engelman, Biophys. J., 1983,43, 167; L. T. Pearson, J. Edelman, and S. I. Chan, Biophys. J., 1984, 45, 863. 73 J. Naghizadeh, in ‘Lecture Notes in Physics No. 172’, ed. K.-H. Bennemann, F. Brouers, and D. Quitmann, Springer-Verlag, Berlin, 1982, p.247. 74 ‘Membrane Fluidity’ (Biomembranes, vol. 12), ed. M. Kates and L. A. Manson, Plenum, New York, 1984. 75 R. Peters and R. J. Cherry, Proc. Natl. Acad. Sci. USA, 1982, 79, 4317. 76 C. J. Restall, R. E. Dale, E. K. Murray, C. W. Gilbert, and D. Chapman, Biochemistry, 1984, 23, 6765. 77 F. Oosawa and M. Kasai, in ‘Subunits in Biological Systems’, ed. S. N. Timasheff and G. D. Fasman, Marcel Dekker, New York, 1971, part A, p. 261. 78 E. Korn, Physiol. Rev., 1982, 62, 672. 79 M. Kasai, S. Asakura, and F. Oosawa, Biochim. Biophys. Acta, 1962, 57, 13. 450 Dickinson thermodynamics and the kinetics. At low ionic strengths, G-actin can be polymerized below the critical actin concentration by application of a shear flow field which presumably acts to promote nucleation." Figure 6 Equilibrium between monomeric (G)and oligomeric (F)forms of actin.Zn this model, aggregates are held together by specific attractive interactions of the type a-b and c-d Another possible area for Brownian dynamics is in modelling the folding and unfolding of globular proteins.81,82 Here we have in mind the long-time diffusional motions, rather than the rapid conformational fluctuation^.'^*'^ That is, the refolding of a denatured protein molecule may be envisaged as the merging of embryo nuclei by a diffusion-collision process. This type of mechanism is consistent with a model 85m of a globular protein consisting of hydrophobic clusters loosely connected by covalent bonds, and held in fixed spatial orientations by interacting polar groups on the cluster surfaces.Such a model protein would thermally denature in two stages: an initial phase involving movement of intact clusters relative to one another, followed by a second phase involving disruption of hydrophobic clusters. Co-operativity would come predominantly from the second phase. Also of interest, in addition to thermal denaturation, is protein unfolding at a solid or fluid interface, the kinetics of which is important in the field of food colloid^.^' 7 Simulation of Subunit Models Complex biological structures can be modelled as a collection of connected subunits. To simulate the dynamics of structures which possess some degree of rigidity, it is necessary to place constraints on the relative motions of different subunits within the total structure.Allison and McCammon have described" a J. Borejdo, A. Muhlrad, S. J. Leibovich, and A. Oplatka, Biochirn. Biophys. Acta, 1981, 667, 118. 'Protein Folding', ed. R. Jaenicke, Elsevier/North-Holland, Amsterdam, 1980. 82 N. G6, Ann. Rev. Biophys. Bioeng., 1983, 12, 183. 83 R. J. P. Williams, Biol. Rev., 1979, 54, 389. 84 CIBA Foundation Symposium No. 93, 'Mobility and Function in Proteins and Nucleic Acids', Pitman, London, 1982. K. Wiithrich and G. Wagner, Trend Biochem. Sci., 1978, 3, 227. 86 K. Wiithrich, H. Roder, and G. Wagner, in ref: 81, p. 549. E.Dickinson and G. Stainsby, 'Colloids in Food', Applied Science, London, 1982. S.A. Allison and J. A. McCammon, Biopolymers, 1984, 23, 167.45 1 Brownian Dynamics with Hydrodynamic In term tions rigorous method of imposing constraints in Brownian dynamics. The procedure is based on the SHAKE molecular dynamics algorithm devised by Ryckaert et and subsequently improved by Ciccotti et aL90 In a rigid body of N spherical subunits, N(N -1)/2 inter-subunit distances are invariant. Neglecting gradient terms, the unconstrained Brownian dynamics step is [see equation 78 and note change from scalar to vector notation (3N-N)]: where the prime denotes the new, unconstrained co-ordinates of subunit i, and fl is the total force acting on subunitj, but excluding forces of constraint. The corrected co-ordinate ri is given by N 6ri = ri -ri‘ = (At/kT) 0:: @ j= 1 where @ is the net force of constraint acting on subunit j.Allison and McCammon have shown88 that 6ri can be represented as where rk,, = r’, -ri, the labels m and n refer to subunits restricted by the pth constraint,d,, is the constrained distance between rn and a, NC is the total number of constraints, and Enforcement of the pthconstraint partially destroys those enforced previously. So it is necessary to repeat the cycle of enforcing all constraints until they are satisfied within a specified tolerance level. The procedure reduces to that of Ryckaert et al.89 in the absence of hydrodynamic interactions, i.c. when The Brownian dynamics algorithm with constraints has been tested for an isolated wormlike chain” and a pair of rigid cubic octamer particles.38 The method offers the opportunity for modelling globular proteins and bacterial viruses as multisubunit structures like those described by Garcia de la Torre and Blo~mfield.~~.~~Using the above algorithm, we can see how, for instance, a model 89 J.-P.Ryckaert, G. Ciccotti, and H. J. C. Berendsen, J. Comput. Phq’s., 1977, 23, 327. 90 G. Ciccotti, M. Ferrario, and J.-P. Ryckaert, Mol. Phys., 1982, 47, 1253. 91 S. A. Allison and J. A. McCammon, Biopolymers, 1984, 23, 363. 92 J. Garcia de la Torre and V. A. Bloomfield, Eiopolymers, 1977, 16, 1779; 1978, 17, 1605. 93 J. Garcia de la Torre and V. A. Bloomfield, Quart. Rev. Eiophq’s., 1981, 14, 81. 452 Dickinson of the antibody-antigen encounter could be successively refined by considering a sequence of multisubunit structures of increasing complexity.8 Concluding Remarks Research in colloid science has led to a resurgence of interest in the Brownian motion of small interacting particle^.'^ This article has tried to show that the concepts used in colloid science have a broader biological relevance. In particular, from statistical mechanics and fluid mechanics is derived a Brownian dynamics computational algorithm suitable for simulating diffusional processes involving entities like enzymes, immunoglobulins, and redox proteins. A protein is modelled as a single Brownian sphere, or a cluster of connected spheres, and account is taken of electrostatic and other forces to whatever level of complexity is feasible under the circumstances.The development of reliable electrostatic potentials of mean force between proteins and their subunits is a requirement for substantial progress in this field. The electrostatic interaction between closely approaching proteins can be represented 95 by so-called 'high-dielectric' models, and numerical results of this type have been recently reported by Matthew and co-workersg6 for the putative reaction complex between flavodoxin and ferricytochrome c. Their calculations show that the two molecules begin to become orientated significantly by the electrostatic field at separations closer than ca. 0.7 nm, when the interaction free energy is some 2 kT less than the sum of free energies of the isolated molecules.An allowance for electrostatic screening accounts for the experimental increase in flavodoxin-cytochrome c association rate at lower ionic strengthsg7 Cases where electrostatic interactions between protein subunits are important include salt bridges in haemoglobin and the superstructure of virus-coat proteins.98 Computer experiments are most useful when they can be compared directly with real experiments. Amongst the techniques available for studying colloidal particle m~tion,~quasi-elastic light scattering is particularly useful. It is encouraging to note, therefore, that a recent light-scattering studyg9 of aggregating proteins gives information on aggregate structures and rate constants which are suitable for comparing with Brownian dynamics simulations.The techniques of nuclear magnetic resonance1" and quasi-elastic neutron scatteringlo' are also being increasingly applied to protein dynamics, and it seems likely that they will also provide useful data for comparing with the computer simulations, and with analytic theories of the many-body hydrodynamic problem.' O2 94 E. Dickinson, Annu. Rep. Prog. Chem.. Sect. C, 1983, 80, 3. 95 A. Warshel and S. T. Russell, Quart. Rev. Biophys., 1984, 17, 283. 96 J. B. Matthew, P. C. Weber, F. R. Salemme, and F. M. Richards, Nature (London), 1983, 301, 169. 97 R. P. Simondsen, P. C. Weber, F. R. Salemme, and G. Tollin, Biochemistry, 1982, 21, 6366. 98 M. F. Perutz, Science (New York), 1978, 201, 1187. 99 J. Feder, T. Jsssang, and E. Rosenqvist, Phys.Rev. Lett., 1984, 53, 1403. loo G. R. Moore, R. G. Ratcliffe, and R. J. P. Williams, Essays Biochem., 1983, 19, 142. lo' H. D. Middendorf, Ann. Rev. Biophys. Bioeng., 1984, 13,425. lo' P. Mazur, Can. J. Phys., 1985, 63, 24. 453 Brownian Dynamics with Hydrodynamic Interactions Acknowledgements. This article was written while the author was on study leave in Oxford supported by a grant from the Leverhulme Trust. I thank Professor R.J. P. Williams and the Oxford Enzyme Group for their kind hospitality and many stimulating discussions. I am particularly grateful to Dr. H. A. 0. Hill for introducing me to protein electrochemistry, and Dr. F. Armstrong for several helpful comments during the preparation of the manuscript. Dickinson Appendix Tensor Multiplication *.-The dot product of tensor 7 and vector w, in matrix notation, is given by where Tt is the transpose of T.The dot product is sometimes written more concisely using the summation convention, i.e. (T V)k = TkiVi (A21 where it is agreed that the indices run from 1 to 3. The inner product of two second- order tensors S and T is itself a second-order tensor: (s‘ T)ik = SimTmk (A31 The double inner product of S and 7 is given by S :T = Tr(S 7)= SimTmi (A41 where Tr is the trace. The double inner product of a third-order tensor B and a second-order tensor T is a vector: (B:7)i = BikmTmk (T:B)i = TkmBmki (A51 The cross product of tensor T and vector w is a tensor: (T X v)ik = -T~,,E,,L~V~ (V X T), = EiWVpTqk (A61 In equation A6, B is the Levi-Civita density. * For further details see, for instance, the book ‘Cartesian Tensors’ (Ellis Horwood, Chichester, 1982) by A. M. Goodbody, from which the nomenclature adopted here was taken. (There is a summary of tensor manipulation, with particular reference to fluid mechanics, in the Appendix to ref: 13.)
ISSN:0306-0012
DOI:10.1039/CS9851400421
出版商:RSC
年代:1985
数据来源: RSC
|
|