摘要:

Chemical Society Reviews Editorial Board Professor H. W. Kroto FRS (Chairman) Professor M. J. Blandamer Or. A. R. Butler Professor E. C. Constable Or. T. C. Gallagher Professor D. M. P. Mingos FRS Professor J. F. Stoddart Consulting Editors Dr. G. G. Balint-Kurti Professor S. A. Benner Dr. J. M. Brown Dr. J. Burgess Dr. N. Cape Professor B. T. Golding Professor M. Green Professor A. Hamnett Dr. T. M. Herrington Professor R. Hillman Professor R. Keese Dr. T. H. Lilley Dr. H. Maskill Professor A. de Meijere Professor J. N. Miller Professor S. M. Roberts Professor B. H. Robinson Professor M. R. Smyth Or. A. J. Stace Staff Editors Mr. K. J. Wilkinson Or. J. A. Rhodes Dr. M. Sugden University of Sussex University of Leicester University of St.Andrews University of Basel, Switzerland University of Bristol Imperial College London University of Birmingham University of Bristol Swiss Federal Institute of Technology, Zurich, Switzerland University of Oxford University of Leicester Institute of Terrestrial Ecology, Lothian University of Newcastle upon Tyne University of Bath University of Newcastle upon Tyne University of Reading University of Leicester University of Bern, Switzerland University of Sheffield University of Newcastle upon Tyne University of Gottingen, Germany Loughborough University of Technology University of Exeter University of East Anglia Dublin City University, Republic of Ireland University of Sussex Royal Society of Chemistry, Cambridge Royal Society of Chemistry, Cambridge Royal Society of Chemistry, Cambridge It is intended that Chemical Society Reviews will have the broad appeal necessary for researchers to benefit from an awareness of advances in areas outside their own specialities.Deliberate efforts will be made to solicit authors and articles from Europe which present a truly international outlook on the major advances in a wide range of chemical areas. It is hoped that it will be particularly stimulating and instructive for students planning a career in research. The articles will be succinct and authoritative overviews of timely topics in modern chemistry. In line with the above, review articles will not be overly comprehensive, detailed, or heavily referenced (ca.30 references), but should act as a springboard to further reading.In general, authors, who will be recognized experts in their fields, will be asked to place any of their own work in the wider context. Review articles must be short, around 8-1 0 journal pages in extent. In consequence, manuscripts should not exceed 20-30 A4/American quarto sheets, this length to include text (in double line spacing), tables, references, and artwork. An Information to Authors leaflet is available from the Senior Editor (Reviews). Although the majority of articles are intended to be specially commissioned, the Society always considers offers of articles for publication. In such cases a short synopsis (including a selection of the literature references that will be cited in the review and a brief academic CV of the author), rather than the completed article, should be submitted to the Senior Editor (Reviews), Books and Reviews Department, The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 4WF. @ The Royal Society of Chemistry, 1994 All Rights Reserved No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic or mechanical, photographic, recording, or otherwise, without the prior permission of the publishers. Typeset by Servis Filmsetting Ltd. Printed in Great Britain by Blackbear Press Ltd.

ISSN:0306-0012    

DOI:10.1039/CS99423FX005

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

ISSN 0306-001 2 CSRVBR 23(2) 75-146 (1994) Chemical Society Reviews Volume 23 Issue 2 Pages 75-146 April 1994 Carrier-mediated Transport through Liquid Membranes By Herman C. Visser, David N.Reinhoudt, and Feike de Jong (pp. 75-81 ) A review is presented of work on carrier-mediated transport of (organic) salts and neutral molecules through liquid membranes. Attention is focused on recent studies of diffusion-limited transport through BLM's and SLM's, the description of which is guided by simple mathematical equations. Finally, developments in the direction of ultra-thin membranes are outlined. The Colourless 'Chameleon' or the Peculiar Properties of Zn2+ in Complexes in Solution By Helmut Sigel and R. Bruce Martin (pp. 83-91 ) The adaptability of the coordination sphere of Zn2 + in biological systems is well known.Less appreciated is the fact that equilibria between Zn2 + species with coordination numbers 6 and (5 or) 4 are indeed quite common in solution in the presence of low-molecular-weight ligands. In this essay ways are described that allow a quantification of these equilibria in binary and in ternary (mixed ligand) complexes. For example, the 1:1 and 1:2 complexes formed by Zn2 + and ethylenediamine exist in aqueous solution to about 50 and 90%, respectively, as tetrahedral species; the remaining parts being octahedral. The Diagnosis of Concerted Organic Mechanisms By A. Williams (pp. 93-1 00) The concept of concertedness is paramount in all discussions of reaction mechanisms. The criterion that a concerted mechanism has a single transition state and hence possesses no intermediates enables the unambiguous diagnosis of these mechanisms.This review describes current experimental techniques which provide unequivocal evidence against processes involving intermediates; such techniques therefore indicate concertedness. Contemporary evidence is discussed for reactions, such as displacements at unsaturated centres, not hitherto regarded as capable of supporting concerted mechanisms. \I MELDOLA LECTURE. The Role of Aromatic Interactions in Molecular Recognition By Christopher A. Hunter (pp. 101-1 09) Studies of T-T interactions in covalently linked porphyrin dimers have been used to develop a simple electrostatic model for understanding and computing non-covalent interactions between aromatic molecules.The model has been tested on experimental data from protein X-ray crystal structures and by the successful design and synthesis of a molecular receptor for p-benzoquinone. Applications of the model to sequence- dependent DNA structure show that ideas derived from simple chemical systems can be useful for understanding more complex molecular assemblies. Non-Bonding Molecular Orbitals and the Chemistry of Non-Classical Organic Molecules By Christopher A. Ramsden (pp. 1 1 1-1 1 8) Professor Ramsden explores common features of the bonding and modes of reaction of a large and diverse family of 'non-classical' organic molecules, which includes 1,3-dipoles, hypervalent compounds, betaines, and ylides, and discusses their relationship to the transition states of SN2 reactions. All these molecules can be considered to contain three-centre, four-election bonds and the associated non-bonding molecular orbitals influence structure and reactivity.Two common modes of reaction, which are described as ligand coupling and syn-addition, are recognized and examples of these transformations are discussed. TILDEN LECTURE. Studies on Thymidylate Synthase and Dihydrofolate Reductase -Two Enzymes Involved in the Synthesis of Thymidine By Douglas W. Young (pp. 11 9-1 28) The review describes the author's work on thymidylate synthase and dihydrofolate reductase. A chemical model for the synthesis of thymine followed the biological labelling results more closely than expected. Work on the enzyme dihydrofolate reductase showed a difference in docking at the active site of the enzyme between the natural substrate and the drug methotrexate.Synthesis of stereospecifically deuteriated leucine and incorporation into the enzyme allowed assignment of the diastereotopic methyl groups of the leucine residues of the enzyme in the 'H-NMR spectrum. This will be generally applicable to other proteins. Structure and Dynamics of Electrolyte Solutions. A NMR Relaxation Approach By Antonio Sacco (pp. 1 29-1 36) Investigations on the dynamical and structural properties of electrolyte solutions based on NMR relaxation methods are reported. Using selected examples of electrolytes in pure solvents, the behaviour of the solvent molecules in the first coordination sphere of ions is highlighted.Moreover, by splitting intra- and inter- molecular contributions to the relaxation for some electrolytes in DMSO, the influence of the salts on the reorientational and translational properties of the solvent molecules is shown. Finally, NMR relaxation by quadrupolar interactions to probe the preferential solvation of ionic species in binary solvent mixtures is reported. The Electrophoresis of Semiconductor Particles By Colin Boxall (pp. 137-1 45) The role of colloidal semiconductor particles as photocatalysts in some environmentally important reactions, and the importance of the nature of the particle surface during such reactions is discussed. A summary of semicondutor particle-related electrochemical concepts is presented and the potential of electrophoresis for in situ non-perturbative interrogation of particle surface behaviour is described.The new technique of photoelectrophoresis and its utility in photocolloidal systems is also described. Finally, some recent advances in the field of semiconductor colloid chemistry are presented, with emphasis on those systems for which particle surface charge-related information would prove most useful. Articles that will appear in forthcoming issues include Homo- and Hetero-metallic Alkoxides of Groups 1, 2, and 12 Metals R. C. Mehrotra, A. Singh, and S. Sogani Towards a Laboratory Strategy for the Study of Heterogeneous Catalysis in Stratospheric Ozone Depletion Martin R.S. McCoustra and Andrew B. Horn Trimetallic Units as Building Blocks in Cluster Chemistry D. Imhof and L. M. Venanzi Chemistry in Near-critical Fluids Roberto Fernandez-Prini and M. Laura Japas Polyradicals: Synthesis, Spectroscopy, and Catalysis Joe A. Crayston, Ahmed Iraqi, and John C. Walton Affinity Biosensors Dbnal Leech The Hydrides of Aluminium, Gallium, Indium, and Thallium: A Re-evaluation Anthony J. Downs and Colin R. Pulham Electrophoretic NMR Manfred Hob Protein Structure from Linear Dichroism Spectroscopy and Transient Electric Birefringence Michael Bloemendal Chemical Society Reviews (ISSN 0306-00 12) is published bi-monthly by The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge, CB4 4WF, England.All orders accompanied with payment should be sent directly to The Royal Society of Chemistry, Turpin Distribution Services Ltd., Blackhorse Road, Letchworth, Herts., SG6 IHN, U.K. NB Turpin Distribution Services Ltd., distributors, is wholly owned by The Royal Society of Chemistry. 1994 annual subscription rate E.C. E99.00, U.S.A. $186.00, Canada €1 1l.OO+ GST, Rest of World €106.00. Customers should make payments by cheque in sterling payable on a U.K. clearing bank or in U.S. dollars payable on a US. clearing bank. Second class postage is paid at Jamaica, N.Y. 11431. Air freight and mailing in the U.S.A. by Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. All other despatches outside the U.K. by Bulk Airmail within Europe and Accelerated Surface Post outside Europe. PRINTED IN THE U.K. Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at E30.00 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way.

ISSN:0306-0012    

DOI:10.1039/CS99423FP007

出版商:RSC    

年代:1994

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99423BP009

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Carrier-mediated Transport through Liquid Membranes Herman C. Visser, David N. Reinhoudt, and Feike de Jong" Department of Organic Chemistry, University of Twente, 7500A E Enschede, The Netherlands 1 Introduction The separation of molecules by membranes is not only a very important event in biological systems, it has also important large-scale industrial applications, and therefore, it is of a broad scientific interest. An efficient separation process combines a high transport rate with a high selectivity. In a steady-state permeation experiment, the flux of a species S through a membrane of thickness I, is related to the concentration gradient (dc,) through Fick's first law: High fluxes can only be obtained when a large chemical potential is maintained over a thin membrane in which the diffusivity D, of the species is high.In this review we describe our systematic approach to such systems, which has led us from bulk liquid membranes to very thin supported monolayers. Four different types of liquid membranes can be dis-tinguished: bulk, supported, emulsion, and polymer composite membranes. Bulk liquid membranes (BLM) consist of a source and receiving phase separated by an immiscible membrane phase. In most cases, the source and receiving phase are aqueous and the membrane organic, but the reverse configuration can also be used.z The thickness of the diffusion layer I is of the order Herman C. Visser was born in Rotterdam, The Netherlands. He obtained his M.Sc. (1989) from Utrecht University and is now a Ph.D.student in the department of Organic Chemistry at Twente University. His Ph.D. work is concerned with application of macrocyclic receptors as carriers in supported liquid membranes and elongation of the life-time of these membrane systems. He is also interested in the kinetics of membrane transport. Feike de Jong was born in Groningen, The Netherlands. He obtainedhis M.Sc. (1968) and Ph.D. (1972) from the University ofGroningen under Professors M. J. Jansen and H. Wijnberg. His interest in the physical organic aspects of macrocyclic chemistry was aroused during his post-doctoral work at the University of California in Los Angeles with Professor D. J. Cram. In 1974 he started as a research scientist at the RoyallDutch Shell Laborator- ies in Amsterdam.Since 1990 he has combined his position in industry with the position of part-time Professor of Physical Organic Chemistry at the University of Twente. of the Nernst layer,3 which is dependent on the experimental conditions, but is typically in the order of 50-500 pm. Supported liquid membranes (SLM) have essentially the same configuration as BLMs, but now the organic phase is contained in the pores of a macroporous (pore size 0.1-1 .O pm) polymer sheet, of which the thickness is in the range of 10-100 pm. For practical applications, the SLM-concept has been developed into practical hollow fibre mod~les.~ Polymer composite mem- branes are another SLM-variant: thin (10-50 pm) films of a polymer-supported liquid phase can be obtained by solvent casting of mixtures of polymer and organic phase.5 A further reduction in membrane thickness can be accom- plished in emulsion liquid membranes (ELM).With the aid of a surfactant, relatively stable water-(receiving phase)-in-oil(mem- brane phase)-in-water(source phase) emulsions can be created.6 On average, ELMShave a thinner membrane separating source and receiving phase, although membrane thickness changes with the amount of the material transported.' Still thinner (5-10 nm) are membranes composed of bilayers (vesicles, liposomes). In many cases, however, the thickness of the separating layer has become so small that diffusion is no longer rate-limiting. In the approach pictured above, transport rates are increased by going to ever thinner membranes.The transport selectivity, however, generally does not change. In order to achieve high selectivity, a substrate-specific receptor must be present in the membrane phase, in which it can act as a carrier between source and receiving phase. Whereas in biological membranes this task is fulfilled by ionophores such as valinomycin (l), in artificial membranes we rely on the realm of synthetic macrocyclic receptors developed during the past two decade^.^ 2 Bulk Liquid Membranes 2.1 Transport Model in Bulk Liquid Membranes Ever since the first transport experiments, theory and practice have developed side-by-side. A theoretical model for transport in a simple BLM system was formulated by Reusch and Cussler' O as early as 1973, and still serves as the basis of more David N.Reinhoudt was born in Wolfaartsdijk, The Netherlands. He obtained his MSc. (1966) and his Ph.D. (1969) from Delft University of Technology. In 1970 hejoined the Royal Dutch fShell Laboratories in Amsterdam as a research chemist in the depart- ment of Organic Chemistry. In 1975 he was appointed as a part- time professor (extraordinar- ius) at Twente University. His appointment as a full professor followed in 1978. The major part of his research deals with supramolecular chemistry, e.g. the synthesis of crown ethers, calixarenes, and (hemi)spher- ands, complexation studies with neutral guest molecules, and the application of supra- molecular chemistry in mem- brane transport, in thefields of electronic or optical sensor systems, catalysis, and molecu- lar (NLO) materials.75 complicated systems. The concentration profile in a BLM consisting of an organic phase containing a carrier L, separating two aqueous phases containing the substrate S, is schematically shown in Figure 1. MEMBRANE ~-RECEIVING PHASESOURCEPHASE I Figure 1 Concentration profile in a BLM. The transport from the aqueous source phase through the organic membrane into the aqueous receiving phase can be dissected into the following discrete steps: (i) diffusion through the aqueous boundary layer of the source phase: (ii) uptake at the source/organic interface (3)(iii) diffusion through the stagnant membrane layer at the source-side of the membrane: (iv) transport by convection through the stirred region of the membrane phase (5)(v) diffusion through the stagnant membrane layer at the receiving side of the membrane: (vi) release at the organic/receiving phase interface (7) (vii) diffusion through the aqueous boundary layer of the receiving phase: Once the extraction reactions 3 and 7 have been defined, the system can be solved assuming: (i) steady-state conditions (then all fluxes are necessarily the same), and (ii) thermodynamic equilibrium at the uptake and release interfaces.2.2 Basic Features of Carrier-mediated Transport In the most simple case, there is no ligand present in the organic CHEMICAL SOCIETY REVIEWS, 1994 membrane, and the species to be transported is simply distri- buted over the phases.The partition coefficient Kpis defined as: This leads to the following relation for the flux J, expressed in the form chemical-potential/flux = resistance: For particles of the same size, transport rates depend on the partition coefficient Kp only, and consequently, the transport selectivity arises from differences in solubility (ratios of Kp's), and hence, transport selectivity only depends on the solvents used. When the membrane contains a carrier that is able to form a complex with the substrate in the organic phase: the extraction steps 3 and 7 discussed above become: In combination with the diffusive processes, this leads to the following equation for the flux: in which The effect of carrier addition is contained in the term F, and since F> 0, the presence of the carrier lowers the resistance of the membrane.Behr et a1.l used an equation similar to 13 to illustrate the most important basic features of carrier-mediated transport: (i) the flux is proportional to the carrier concentration (ii) the initial (when S, = 0) flux shows typical saturation behaviour, J being proportional to S, for low substrate concent- rations, and J being independent of S, for high substrate concentration (iii) the flux J versus log(&.,) shows a maximum, the position depending on Ss and SR. A too strongly complexing carrier becomes fully loaded with substrate even for very low substrate concentrations, and therefore does not produce a gradient over the membrane.This leads to the conclusion, that the best carrier is not necessarily the one with the strongest complexing behav- iour. Apart from this thermodynamic reason, by which strong complexing agents retain the metal ion in the membrane, there may also be a kinetic reason: strong complexing agents fre- quently show a slow rate of decomplexation. 2.3 Transport of Ion-pairs in Bulk Liquid Membranes The choice of membrane solvent in BLMs is often governed by factors such as low viscosity and low water solubility. In practice, chloroform and dichloromethane are used most fre- quently, and as a consequence of their low dielectric constant, univalent cations will be extracted as ion-pairs: Izatt et al.have studied this type of BLM extensively for alkali metal ions, and have derived lengthy expressions for the flux for single and multiple-cation transport. A simplified equation has been used by us3 and others," by taking the experimental CARRIER-MEDIATED TRANSPORT THROUGH LIQUID MEMBRANES-D. N. REINHOUDT ET AL. finding that the concentration drop over the aqueous boundary layers is small compared to the drop over the organic boundary layers. This leads to a simple expression for the flux J: The features of ion-pair transport are essentially the same as for neutral compounds: the flux J is proportional to the carrier concentration, it shows saturation behaviour with respect to aqueous substrate concentration, and an optimum in the rela- tion of J and K,, when S, > 0.All these relations have been experimentally verified. The effect of carrier structure on the flux, is contained in the complexation constant K,. Although this parameter is known for many receptor/substrate combinations in solvents such as methanol, they are frequently unknown in the solvent system used for transport. This problem can be circumvented by assuming that (Ka)solventand (QMeOH are linearly related, in which case equation 15 can be rewritten as (initial transport, s, = 0): Lo/J = 21,,,/D + 21,/DK,Kp[Ss]2 (16) Although the thermodynamic basis for this assumption is not always valid,I4 the relationship holds in many cases. For example, Figure 2 shows the relation for the flux of guanidinium thiocyanate as a function of (Ka)MeOHthrough a chloroform BLM containing various benzo-(2) and dibenzo-(3) crown ethers as carrier^.^ 1 I 0 0.01 0.02 0.03 0.04-K-' (W') Figure 2 Relation between the flux J through a chloroform liquid membrane and the association constants K in methanol for guanidi- nium complexes of carriers (2) (n = 7,8,9) and (3) (n = 10,11,12).The agreement is excellent in this case; from the intercept (lm/D)and a literature value for D a thickness of 38 pm for the stagnant chloroform layer can be obtained. When the source phase contains a mixture of two cations M,+ and M2+,the individual initial (S, = 0) fluxes J, and J2 can also be related to the fundamental parameters: Since the diffusion coefficients D,and D,are very similar, the selectivity is determined by the ratio of the extraction constants.Indeed, there is abundant proof for this statement in the litera- ture.15 We also found a good correlation for the selectivity of valinomycin (1) and crown ethers (7) and (8) in the competitive transport of guanidinium and imidazolidinium thiocyanate' through a chloroform bulk liquid membrane. 3 Supported Liquid Membranes 3.1 Stability of Supported Liquid Membranes Supported liquid membranes consist of an organic carrier solution immobilized in a porous polymer sheet, separating the aqueous source and receiving phases (Figure 3). The support is often a hydrophobic polymer such as polyethylene, polypropy- lene, polytetrafluoroethylene, or polysulfone.For fast trans- port, the exchanging surface between source and membrane should be large, and therefore, a thin film with high porosity is attractive. Electrode 3+ Magnet I \ Membrane 10.1 pm ca. 100pm Figure 3 Supported liquid membrane. (Reproduced by permission from Red. Trav. Chim.Pas-Bas, 1993, 6, 317.) The small membrane volume compared to the large exchang- ing interface is an important advantage of SLMs over BLMs, since much less of the more expensive carrier component is needed. However, at the same time precautions have to be taken to ensure that both the solvent and the carrier are retained in the polymer sheet. Several factors come into play in the loss of solvent: solubility in water, solubility of water in the solvent, volatility, interfacial tension, ability to form emulsions in the aqueous phase, and the pore size of the support.' Polypropylene membranes Celgard@ and Accurelo, in combination with solvents such as o-dichloro- benzene, phenylcyclohexane, or o-nitrophenyl octyl ether (NPOE), give SLMs that are stable under most laboratory conditions. For practical purposes however, the stability is still unsatisfactory. The loss of carrier is related to the partition coefficient (P)of the carrier itself: La F? L, The partition coefficients can either be determined experimen- tally, or be calculated using an empirical relationship based on the group additivity approach introduced by Rekker:' in which f is the 'hydrophobic fragmental constant', N the number of fragments, and p the 'proximity effect', which accounts for neighbouring group interactions.For various fragments, $values are known for the water/l-octanol system. CHEMICAL SOCIETY REVIEWS, 1994 In the same solvent system, we have determined the partition coefficients for the crown ethers (2) and (3) According to equation 18, the fragmental constant for the oxyethylene frag- ment is obtained from a plot of log(P),,, against N The slopes of the relation for benzo and dibenzo-crown ethers are equal (Figure 4),and give af-value of -0 17 for the -OCH,CH,O- unit Using this value, partition coefficients for a large number of carriers can be calculated 2o These values can be converted into the NPOE/water system frequently used in SLMs by the empiri- cal relation 2o 21 It can be calculated that in order to retain the carrier in the NPOE layer, bg(P)NpOE values should exceed 5, and hence log(P),,, > 5 2 It is clear from Figure 4 that the hydrophobicity of all common crown ethers is too low Transport studies with these carriers that do not explicitly take into account the differences in hydrophobicity are unlikely to yield reliable values for the extraction constant and diffusion coefficient For exam- ple, transport of guanidinium thiocyanate through a NPOE/ Accurel -supported liquid membrane with benzo and dibenzo- crown ethers, (2) and (3), was more affected by differences in lipophilicity than by differences in complexing ability 28 20 10 0 -N Figure 4 Experimental partition coefficient, log(P)NPOE vs the number of oxyethylene units for the benzo [(2), N = n, 01and dibenzo [(3), N =m +n, +]crown ethers Obviously, in order to create a stable SLM system, the lipophilicity of the carriers needs to be improved We have used the following three approaches (1) attachment of long alkyl chains, (11) attaching the carrier onto a polymer, (111) attachment of one or more NPOE-moieties The first approach is illustrated by a series of diaza- 18-crown-6 carriers, (4) The unsubstituted carrier has log(P)NpOE = -0 15, and hence leaks from the membrane extensively Providing the carrier with long alkyl chains improves the lipophilicity into the desired range, and consequently, the membranes are stable2, (Table 1) In the second approach, a hydrophilic crown ether such as benzo-18-crown-6 [(2),n= 41 was coupled to the terminal positions of a polysiloxane backbone to give carriers such as (5) Table 1 Partition coefficients of some selected carriers for the system NPOE/water Carrier log(P)NPOE Carrier 10g(p)NPOE (2) n = 3 n=4 0 91 0 58 (5) p = 6, n = 4 p= 46,n = 2 47 27 (3)n= l,m= 1 33 (7) 79 (4) R = H R = C,H,, -0 15 4 05 (8) 13 R=C,,H,, 825 R=C,,H,, I1 6 n R-N co O3 LooJ-R W (4) In agreement with the very large calculated log(P) values (Table l), membranes with these carriers retain their properties even after contact with large volumes of water 23 Although the two approaches discussed above lead to carriers with the desired hydrophobicity, they still are not perfect Long- alkyl and polymer-modified carriers generally do not show high solubility in NPOE (the solubility of didecyl-diaza-18-crown-6is 0 35 M at 25 “C) Furthermore, polymer-bound carriers show reduced rates of diffusion These drawbacks can be avoided by providing the carrier with one or more ‘solvent-like’ substi- tuents For example, NPOE-modified carriers such as (6) are completely miscible with NPOE, and allow very stable mem- branes with high permeability to be constructed 24 3.2 Transport Model for Supported Liquid Membranes The mechanism of ion transport through a liquid contained in the pores of a polymer sheet is not fundamentally different from transport through a bulk liquid as depicted in Figure 1 the stirred part of the membrane phase is now lacking, and the stagnant layers form the whole polymer sheet Therefore, as a first approximation, the flux equations for SLMs can be obtained by replacing the term 21, in BLMs by d, the thickness of the polymer sheet However, corrections have to be made for the morphological characteristics of the sheet The apparent diffusion, Dapp,coeffi-cient is related to the true coefficient of diffusion, D,, through E is the membrane porosity, and T the tortuosity (which is related to 8, the tortuosity factor defined as the average pore length/ sheet thickness) Values of E and T are available for commonly used macroporous membranes such as Celgard” and Accurel” It is customary in literature to assume that D, = Do,Dobeing the diffusion coefficient of the species in bulk solution By doing so, we are in the comfortable position to estimate D, by empirical methods such as the Stokes-Einstein relation 7 being the viscosity of the solvent, and rthe radius of the solute Alerted by the fact that diffusion in confined geometries can be much slower than in bulk solution,25 we have compared diffu- sion rates in bulk NPOE and in NPOE confined in the pores of CARRIER-MEDIATED TRANSPORT THROUGH LIQUID MEMBRANES-D N REINHOUDT ET AL AccureP (pore diameter 0 1 pm) and in CelgardO (pore diameter Flux T 0 04 pm) using pulsed-field-gradient NMR 26 It turned out that the diffusion constant in AccureP was reduced by a factor of 2, and in Celgarde even more, compared to bulk NPOE There- fore, diffusion constants obtained from transport experiments, even when corrected for porosity and tortuosity, remain depen- dent on the morphology of the membrane 3.3 Transport of Free Ions through Supported Liquid Membranes The volatile solvents mostly used in BLMs (chlorinated hydro- carbons) cannot be used in SLMs since they are washed out of the support ’Therefore, non-volatile solvents of low dielectric constant such as hexylbenzene have frequently been used In this type of solvent, salts are transported as ion-pair~,~~ and fluxes are given by an expression similar to equation 15 However, polar lipophilic solvents such as NPOE (E, = 24) can also be retained in polypropylene membranes A mechanistic study of potassium perchlorate transport through NPOE contained in an AccureP membrane, mediated by dibenzo-18-crown-6 [(3),m= n = I], showed that the flux could not be described by equation 15 for the transport of ion-pairs 28 Instead, a model assuming fully separated ions in the membrane phase was found to predict the observed fluxes quantitatively (K+),, + (ClO;),, + L, e(KL+), + (ClO,), J= D,,,/2d [-1 + J(1 + 4L,/KexS’)] (22) This model was experimentally verified for the transport of alkali cationsz9 30 and guanidinium cation30 using a wide variety of carriers the natural potassium ionophore valinomy- cin (I), crown ethers (3), calixcrowns (7), and calix[6]arenes (8) (1o-~mot m-2 s-1) 3 t 21Q”1 0 0 002 004 006 008-4 01 (MI Figure 5 KClO, (0)and NaClO, (+) flux as a function of the salt activity in the source phase for calixcrown (7), [carrier]; = 10-’M, T = 298 K Solid lines are calculated according to equation 22 4 Flux (1o-~mol m-2 S-’) t 3 2 1 0 0 5 10 15-Carrierconc. (mM) Figure 6 KClO, (0)and NaC10, (+) flux as a function of the calix crown (7) concentration in NPOE, [alkali perchlorate] = 0 1 M, T= 298 K phase is the same Obviously, in practice one would like to AN 001(re)move one component completely from the source phase In order to do so, we need an extra driving force to transport the cation against its concentration gradient According to Figure 1, the gradient over the membrane, dcs, is given by [LS,] -[LS,], and the problem becomes how to keep this term positive, even in situations where the concentration in the receiving phase, S, is higher than in the source phase, S, The complex concentration at the source interface, LS,, can be manipulated by the addition to the source phase of a salt having a common anion and a cation which is not extracted (eg Li +) The common-anion effect is accounted for through equa- tion 22, and it can be shown that at equilibrium, the concent- rations in source, S,, and receiving, S,, phase are related to the lithium concentration, A, Table 2 Calculated diffusion coefficient, D,, and extraction constant, Kex,for carriers in NPOE/Accurel@ at 298 K Carrier Dm (IO-ll m2s-l) Kex (M-I) K+/Na+ selectivity (3)m=n=l K+ Na+ 2 2 13 19 x lo-’ 66 (7) K+ Na+ 067 067 2 3 103 45 x lo-’ 5100 (8) K+ Na+ 028 028 56x 10’ 42 x lo-’ 1300 (9) K+ Na+ 13 16 32 x 10-l 2 1 x 10-1 Gu+ 10 29 (1 1) (10) urea urea 0 61 097 20 x 10’ 2 2 x 10’ (7) The basic features of free-ion transport through SLMs are illustrated in Figures 5 and 6 The transport of KClO, and NaClO, by the calix[4]crown-5 (7) is shown in Figure 5 as a function of the cation concentration in the aqueous phase The transport rate is first order in cation only at very low concent- rations, at high concentration, typical saturation behaviour is observed The dependence of the flux on the carrier concent- ration is linear under conditions of full saturation at the source interface (high aqueous salt concentration and/or strong com- plexing behaviour), but the order in carrier is less than unity for weakly complexing carriers (Figure 6) By simple curve-fitting procedures applied to the relations as shown in Figures 5 and 6, K,, and Dappcan be obtained (Table 2) The values obtained for K,, in single-ion transport experiments can subsequently be used to predict the selectivity in competitive transport experiments The agreement between theory (with equation 17) and experi- ment is excellent, both for potassium/sodium as well as for combinations with guanidinium cations As for BLMs, we found that transport selectivity in SLMs is determined by extraction 3.4 Uphill Transport of Cations From Fick’s law (equation 1) it follows that cation transport will cease when the concentration of the ion in source and receiving SJS, = 1 + Ao/So (23) In full agreement with these predictions, we were able to extract virtually all potassium from a mixture of potassium and sodium perchlorate to which an excess of lithium perchlorate was added.29 The complex concentration at the receiving interface can be kept very low by applying a favourable exchange reaction with a different cation, for example a proton: [LS,], + H+ e[LH+l, + PI, The counter-transport of protons, from the acidic receiving phase to the basic source phase, is now the driving force for cation transport.Using this mechanism, potassium perchlorate can be transported uphill by using diaza-crown-6 (4) derivatives.22 3.5 Transport of Urea through SLMs Selective removal of urea from blood is of great importance in medicine.Very recently, lipophilic macrocyclic receptors for urea have become available in our group. Metallo-macrocycles (9) and (10) complex urea by coordination of the urea carbonyl to UO,. Dissolved in NPOE, these carriers transport urea through an AccureP membrane.31 Transport rates can be described by the SLM equivalent of equation 13 (neglecting the aqueous boundary layers, and putting 21, = 6).Figure 7 shows the predicted first order dependence on the carrier concentration for binaphthyl carrier (9).Curve-fitting of experiments with varying urea concentration in the aqueous phase32 yields the diffusion coefficient D, and the extraction constant K,, (Table 2). The transport selectivity for a mixture of urea/N-methylurea was 8.2, close to the selectivity observed in extraction experi- ments (viz.,8.7). In competition with K+, urea was extracted exclusively. Several polyazareceptors were found to transport urea as well, albeit slightly less effi~iently.~~ Assisted flux 1 I (1o-~mol m-2 s-') t 6 8 100 2 4 -Carrier conc. (mM) Figure 7 Dependence of urea flux on the concentration of carrier (9) through a NPOE/Accurel@ SLM. 4 Supported Monolayers With the possible exception of certain calixarene~,~~ the mem- brane transport discussed above is limited by diffusion.Accord- ing to Fick's law, the flux is inversely proportional to the thickness of the membrane. The desire to make membranes as thin as possible has guided us to the use of monolayers. As such, these layers are mechanically too weak to be used as stable membranes, hence they are supported by a cation-transparent Nafionm film. Nafionm (1 1) is a perfluorinated polymer contain- ing sulfonic acid groups, many of which occupy the surface when a Nafion film is in contact with water. We envisaged the use of these surface sulfonate groups to bind alkyltriphenylphospho- nium amphiphiles (12) via an ion-exchange process. The conductivity over the Nafion* membrane decreased CHEMICAL SOCIETY REVIEWS, 1994 dramatically by the presence of 25 mM of amphiphiles (Figure 8).34Subsequent analysis of the adsorbed layer by UV-spectro- scopy, attenuated total reflection IR, XPS, and SIMS, has led us to conclude that the amphiphile is indeed ionically linked to the membrane, to form an incomplete monolayer, most of which is located at or very near to the surface of the membrane.The next step will be to equip these amphiphiles with macrocyclic carriers to obtain a stable and thin membrane. Conductivity is (mSkm) Aqueousphase f lo\ 0 20 40 60-Time (hours) Figure 8 Conductivity experiments of the adsorption of n-alkyltri-phenylphosphonium amphiphiles (12) (aqueous solution of 25 mM of amphiphile and 0.1 M KCl as electrolyte) on a Nafiona 117 (1 1) membraneat298K.(l2)n=8+, U;n= IOA, O;n= 160, x.I CF3 k = 65, I = 1. m = 1. n-ca 1000 (1l)Nafiona 117 (12) 5 Conclusions The (neutral) carrier-mediated transport of alkali cations and neutral species through bulk liquid membranes is well-under- stood on the basis of a model in which exchange processes at the interfaces are fast, and diffusion through stagnant solvent layers is rate-limiting. This model predicts that transport selectivity is based on differences in extraction. Supported liquid membranes have a better chance of being applied, but the concurrent CARRIER-MEDIATED TRANSPORT THROUGH LIQUID MEMBRANES-D N REINHOUDT ET AL requirements for fast and selective transport as well as a good membrane stability, necessitate a careful choice of carrier/ solvent/support combination Studies on these systems have revealed that transport through porous polymer membranes is still often diffusion-controlled, but the specific morphology of the membrane affects the properties of the supported liquid as well In order to increase transport rates still further, very thin membranes are needed This will take us into the area of kinetic rather than diffusion control 6 References 1 P R Danesi, Sep Sci Technol, 1984-85,19,857 2 F Diederich and K Dick, J Am Chem Soc , 1984,106,8024 3 T B Stolwijk, P D J Grootenhuis, P D van der Wal, E J R Sudholter, D N Reinhoudt, S Harkema, J W H M Uiterwijk, and L Kruise, J Org Chem , 1986,51,4891 4 R M Izatt, D K Roper, R L Bruening, and J D Lamb, J Membr Sci 1989,45, 73 5 T Kajiyama, Top Inclusion Scz , 1991, 2, 1 1 1 6 M P Thien and T A Hatton, Sep Sci Technol, 1988,23,819 7 J Draxler and R Mar, Chem Eng Process, 1986,20,319 8 T M Fyles, Bioorg Chem Frontiers, 1990, 1, 71 9 S R Cooper, ‘Crown Compounds, Toward Future Applications’, VCH Publishers Inc , New York, 1992 10 C F Reusch, and E L Cussler, AIChE J, 1973,19,736 11 J P Behr, M Kirch, and J M Lehn, J Am Chem Soc , 1985,107, 24 1 12 R M Izatt, G A Clark, J S Bradshaw, J D Lamb, and J J Christensen, Sep Purf Methods, 1986, 15, 2 1 13 J D Lamb, R M Izatt, D G Garrick, J S Bradshaw, and J J Christensen, J Membr Sci , 1981,9, 83 14 T M Fyles, Can J Chem , 1987,65,884 15 J D Lamb, J J Christensen, S R Izatt, K Bedke, M S Astin, and R M Izatt, J Am Chem Soc, 1980,102, 3399 16 T B Stolwijk, E J R Sudholter, D N Reinhoudt, J van Eerden, and S Harkema, J Org Chem , 1989,54, 1000 17 W F van Straaten-Nijenhuis, F de Jong, and D N Reinhoudt, Recl Trav Chim Pays-Bas, 1993,112,317 18 (a) A M Neplenbroek, D Bargeman, and C A Smolders, J Membr Scz , 1992, 67, 133 (b)P Deblay, S Delepine, M Minier, and H Renon, Sep Sci Techno1 , 1991,26,97 (c) H Takeuchi, K Takahashi, and W Goto, J Membr Sci , 1987,34, 19 19 R F Rekker, ‘The Hydrophobic Fragmental Constant’, Vol 1, Elsevier Scientific Publisher, Amsterdam, 1977 20 T B Stolwijk, L C Vos,E J R Sudholter,andD N Reinhoudt, Reel Trav Chim Pays-Bas, 1989, 108, 103 21 T B Stolwijk, E J R Sudholter, and D N Reinhoudt, J Am Chem Soc , 1989,111,632 1 22 W F Nijenhuis, J J B Walhof, E J R Sudholter, and D N Reinhoudt, Reel Trav Chim Pays-Bas, 1991, 110,265 23 M M Wienk, T B Stolwijk, E J R Sudholter, and D N Reinhoudt, J Am Chem Soc , 1990,112, 797 24 H C Visser, Ph D Thesis, University of Twente, 1994 25 W M Deen, AIChE J, 1987,33,1409 26 E G Buitenhuis, Ph D Thesis, University of Twente, 1994 27 R M Izatt, R L Bruening, M L Bruening, and J D Lamb, Zsr J Chem , 1990,30,239 28 T B Stolwijk, E J R Sudholter, and D N Reinhoudt, J Am Chem Soc , 1987,109,7042 29 W F Nijenhuis, E G Buitenhuis, F de Jong E J R Sudholter, and D N Reinhoudt, J Am Chem SOC, 1991,113,7963 30 A Casnati, P Minari, A Pochini, R Ungaro, W F Nijenhuis, F de Jong, and D N Reinhoudt, Isr J Chem , 1992,32,79 31 W F Nijenhuis,A R van Doorn, A M Reichwein, F de Jong,and D N Reinhoudt, J Am Chem Soc , 1991,113,3607 32 W F van Straaten-Nijenhuis, A R van Doorn, A M Reichwein, F de Jong, and D N Reinhoudt, J Org Chem , 1993,58,2265 33 W F van Straaten-Nijenhuis, F de Jong, D N Reinhoudt, R P Thummel, T W Bell, and J Liu, J Membr Scr , 1993,82,277 34 W F van Straaten-Nijenhuis, E J R Sudholter, F de Jong, D N Reinhoudt, and J W G Mahy, Langmuir, 1993,9, 1657

ISSN:0306-0012    

DOI:10.1039/CS9942300075

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

The Colourless 'Chameleon' or the Peculiar Properties of Zn2+ in Complexes in Solution. Quantification of Equilibria Involving a Change of the Coordination Number of the Metal Ion" Helmut Sigel Institute of Inorganic Chemistry University of Basel Spitalstrasse 51 CH-4056 Basel Switzerland R. Bruce Martin Chemistry Department Charlottesville University of Virginia VA 22903 U.S.A. 1 Introduction The metal ion Zn2 + plays a crucial role in the active site of over 300 metalloenzymes (e.g.' -3). However its role is not limited to catalysis and gene expression Zn2 + also stabilizes the structure of proteins and nucleic acids preserves the integrity of subcellu- lar organelles engages in transport processes and plays import- ant roles in viral and immune phenomena.'J Thus an under- standing of its basic coordination properties is desirable. Zn2 + coordinated in proteins usually has a tetrahedral or distorted tetrahedral coordination with the low coordination number 4. During the course of an enzymatic turnover this inner-sphere coordination number of catalytically active Zn2 + may increase to 5,4but to our knowledge coordination number 6 has never been observed for a protein-bound Zn2 +. This situation contrasts with that in low molecular weight complexes a recent evaluation of 490 crystal structures showed that Zn2+ has coordination number 4 in about 58% of the complexes and coordination numbers 5 and 6 in about 13 and 27% respectively. The percentages for the related metal ions Co2 and Cd2 are for the coordination numbers 4,5 and 6 for+ + Co2+,13 10 and 76% (from 1500 structures); and for Cd2 + 19 8 and 56% (from 200 structures; the difference to 100% accounts for other coordination numbers). Although coordi- nation number 6 for Zn2 + in the solid state apparently is not as important as for Co2+ or Cd2+ Zn2+ often has a greater coordination number in low molecular weight complexes than in proteins. In aqueous solution coordination number 6 is probably even more important for Zn2 +:e.g. depending on the concentration * Abbreviations Bic- abbreviation for Bicinate as used in formula; Bicinate monoanion of Bicine; Bicine N,N-bis(2-hydroxyethyl)glycine;CN coordination number (r.g Zn& means Znz+ with coordination number 6); En ethylenedia- inine ( = I ,2-diaminoethane); Gly- glycinate; Ha histamine; HisMe histidine methyl ester; Im imidazole; Nta3-,nitrilotriacetate; Oxz-,oxalate; Ser- serinate; UTP4- uridine 5'-triphosphate. Bruce Martin is Professor (1965) of Chemistry at the University of Virginia where he joined the faculty in 1959. He was born in Chicago and earned his B.S. degree at Northwestern University in 1950 and his Ph.D. degree in physical chemistry at the Uni- versity of Rochester in 1953. He held postdoctoral appoint- ments at both Caltech and Har- vard and has twice been on leave from Virginia at Oxford University. Dr. Martin has published widely in thefields of biophysical and bioinorganic chemistry. His research with metal ions has includedstudy of their interactions with amino acids peptides proteins lipids nucleosides and nucleotides by a variety of techniques. 83 dissolution of ZnC1 in water leads to [Zn(H20)J2+ [ZnCl(H,O),] + [ZnCl,(H,O),] and [ZnCI,(H,0),]2 -spe-cies.6 In aqueous 2.88 M ZnC1 in 9.35 M HC1 or in more diluted aqueous ZnC1 solutions upon addition of organic solvents [DMF CH30H (CH,),SO] an equilibrium between octahedral-tetrahedral geometries is suggested.' Evidently in solution well-defined coordination numbers especially the lower ones are difficult to obtain,8 though by employing well-devised ligands coordination numbers 4 or 5 can be enforced (e.g. see reference 9). Znz+ is clearly among those metal ions having a most flexible and adaptable coordina- tion sphere truly a 'chameleon'! However the switch from one coordination geometry to another cannot easily be studied and quantified in solution as Zn2 + is diamagnetic and colourless. Having noted the peculiar coordinating properties of Zn2 + for some time we recently made a fruitful observationlo in ternary systems consisting of N,N-bis(2-hydroxyethy1)glycinate ( = Bicinate) imidazole (Im) and a divalent metal ion (M2+) i.e. Co2+,Ni2+,Cu2+,Zn2+,orCd2+.Bicinateisaderivativeof the amino acid glycine (Figure I) yet owing to its two hydroxy- ethyl groups it is potentially a tetradentate ligand. Indeed for the aforementioned M2+ only about 0.5 percent of M(Bici- nate) + exist with a twofold coordinated glycinate-type binding mode whereas in the case of Co2 +,Ni2+,and Zn2 + about 85 to 90% are bound in a tetradentate mode;11 in the remaining 10 to 15% only one of the two hydroxyethyl groups binds. Owing to the Jahn-Teller effect the tetradentate mode drops to about 60% for Cu(Bicinate) . Surprisingly despite this dominating + action of Bicinate as a tetradentate ligand," the ternary Zn(Bicinate)(imidazole) complex shows a very high stability + the affinity of imidazole toward Zn(Bicinate) is significantly + Helmut Sigel is ao Professor (1978) of Inorganic Chemistry at the University of Basel where he became Dozent in 1967 and also earned his Ph.D. in 1964. He was a Visiting Stag Member at Cornell University (Ithaca N.Y.) in 1968169 the 1985 Brother- ton Visiting Research Professor at the University of Leeds (U.K.),andfor 1992-6 he is named Visiting Professor at Zhong- shun University in Guangzhou (China). Dr. Sigel is the reci- pient of the 1977 AIfred Werner Award of the Swiss Chemical Society a member of the Steering Committee of the European Science Foundation Programme 'Chemistry of Me-tals in Biological Systems' and of various Editorial Boards of international journals; he has published widely in the field of bioinorganic chemistry and -together with Astrid Sigel- he is editor of the series 'Metal Ions in Biological Systems'. 84 ">C-CH2-NH2 -0 Glycinate Figure 1 Chemical formula of glycinate i.e. of the anion of the amino acid glycine as well as of its derivatives N,N-bis(2-hydroxyethyl)glyci-nate ( = Bicinate or Bic-) and nitrilotriacetate (Nta3-). larger than the one toward Zn(aq)2 + . O This may be explained by a reduction in coordination number; i.e.,upon coordination of one imidazole to Zn(Bicinate)+ two water molecules (and possibly one hydroxyethyl group) are released from the coordi- nation sphere of Zn2 +,giving rise to a reduced coordination number of 5 (or even 4). Similarly nitrilotriacetate (Nta3 -) is also like Bicinate a derivative of glycinate (Figure 1). Considering the iminodiace- tate moiety as the central binding core in M(Nta)- complexes,' one calculates that the third acetate residue coordinates in more than 99.8% of the M(Nta)- species proving that Nta3- also acts overwhelmingly as a tetradentate ligand. Furthermore like +Zn(Bicinate)(imidazole) the Zn(Nta)(imidazole)- complex shows high stability.I3 We believe that these ob~ervations~~~~~ suggest a changing coordination number of Zn2 + in certain complexes in aqueous solution.lO Herein we describe in more detail the connected equilibria and quantify the degree of formation of the species containing Zn2 + with various coordination numbers. 2 Evidence for a Reduction of the Coordination Number of Zn2+ in the Ternary Complex Formed in the Zn2+-Bicinate- lmidazole System and Relevant Equilibria The stability of mixed ligand complexes can be quantified by comparison with that of their binary parent complexes; i.e. by finding the position of equilibrium 1:14,15 M(Bic)+ + M(Im)z + S M(Bic)(Im) + MZ (1)+ + Its constant is defined by equation 2 and values for it follow from relation 3:14y1 1Od"JgKM = [M(Bic)(Im)+][M2 +]/[M(B~C)+][M(I~)~ (2)+I The definition of the stability constants of the binary and ternary complexes occurring in equation 3 is evident from the following two examples M2+ + Im*M(Im)2+ (44 M(Bic)+ + Im*M(Bic)(Im)+ (54 CHEMICAL SOCIETY REVIEWS 1994 Where further identification of d log& (equation 3) is desir- able additional subscripts are given like d log KZnlBlc,lrn,mean-ing that the value refers to the Zn2 +-Bicinate-imidazole system. According to the general rule for binary complexes K > K2.. . (e.g.,see reference 16) equilibrium 1 would be on its left side; i.e. for d log& (equation 3) negative values are expected. Indeed statistical (st) leads to the same conclusion a monodentate ligand entering an octahedral (oh) coordination sphere has six possibilities for binding while only two remain if four are already occupied by Bicinate. As the probability for dissociation is the same namely 1 in the binary and ternary complex the statistical value for d logKsti0h is log (2/ 6) = -0.5. For Cu2 the value is more uncertain:l4 assuming a + square-planar (sp) or a distorted octahedral (do) coordination sphere one may conclude' that the value is log (1 /4)or log (1/6) and hence d log Kstispido= -0.6 or -0.8; i.e. log K,,, = -0.7 appears as a reasonable estimate. Comparison of these statistical values d log& with the experimental results'O d logKM/B,c/Im= -0.05 -0.08 -0.11,+ 0.31 and -0.50 for the complexes with Co2+ Ni2+ Cu2+ Zn2+,and Cd2 +,respectively indicates that only the Cd2 +/ Bicinate/Im system behaves as predicted by the statistics. The ternary complexes with Co2 +,Ni2+,Cu2+,and Zn2 + are more stable than expected. This observation agrees with the previous 591evidence' 3-1 that mixed ligand complexes formed by a metal ion of the second half of the 3d series an 0-donor (as Bicinate is to a large part) and a heteroaromatic N-base with 7~-accepting properties like imidazole are especially stable. * However the most unexpected result is the high relative stability of Zn(Bicinate)(Im)+ a positive value for d log KZnlB,cilrnis observed meaning that equilibrium 1 is on its right-hand side. This observation suggests that a reduction of the coordination number of Zn2 occurs upon the coordination of imidazole to + Zn(Bicinate)+ as this is the only way to explain the increased complex stability. 3 The Special Properties of Zn( Bicinate)( Im) + and of Related Ternary Complexes To assist further comparisons and to decide whether Zn(Bicin- ate)(Im)+ is an isolated case we collected the results of Table 1 which include Bicinate Nta3- (Figure I) and uridine 5'-triphosphate (UTP4-; Structure I) as multidentate ligands.10.13319At first we compare the values ford log& given in rows 1 and 2 The relative stabilities of the complexes with Coz+ NiZf or Cuz+ and Nta3-/Im are slightly larger than those with Bicinate and imidazole probably due to coordination of the negatively charged oxygens of the carboxylate groups of Nta3-(Figure l) which should make the n-accepting properties of imidazole somewhat more effective (Section 2). However most importantly a rather high relative stability is also observed for Zn(Nta)(Im)- especially if compared with the value for d log KcolN!alImfor the similarly sized Co2 or the statistical + value (Section 2). From rows 1 2 and 4 of Table 1 it is clear that the relative stability of the ternary Zn2+ complex is always pronounced especially when Cu2 + is excluded because of its irregular (tetra- gonal) coordination sphere and when emphasis is put on the comparison with Co2+ (or Ni2+) which usually prefers a Q hhh -0II -0p-0-0"-o-F;,-o -u I;,-o-cH,4:@; 0 H3' I THE FLEXIBLE COORDINATION SPHERE OF ZINC(I1) IN SOLUTION-H. SIGEL AND R. B. MARTIN Table 1 Comparison of the relative stabilities of various mixed ligand complexes in analogy to equilibrium 1 and equations 2 3 for aqueous solutions at 25 “Cand I = 0.1 Mu A logKM + +No Equilibrium co2 NiZ cu*+ Zn2+ Cd2+ Ref. I M(Bic)+ + M(Im)2+ =M(Bic)(Im)+ + M2+ -0.05 f0.04 -0.08f0.01 -0.11 f0.03 0.31 f0.04 -0.50 f0.04 10 2 M(Nta)-+ M(Im)2+ =M(Nta)(Im)- + M2+ 0.01 f0.02 -0.01 f0.01 0.26 f0.01 0.22 f0.02 13,19b 3 M(Nta)-+ M(NH3)2+ =M(Nta)(NH,)- + M2+ -0.26 f0.04 -0.20 f0.06 -0.39 f0.08 -0.09 f0.09‘ 13 4 M(UTP)2-+ M(Im)2+ *M(UTP)(Im)2- + M2+ -0.36 f0.04 -0.40 f0.03 -0.37 f 0.02 0.03 f 0.02 -0.38f 0.03 19 5 M(UTP)2-+ M(NH3)Z+ *M(UTP)(NH,)’ -+ M2+ -0.58 f 0.12 Q-0.1 6-0.4 19 i The error limits (3 U) were calculated according to the error propagation after Gauss they are based on the errors of the stability constants measured for the binary and terndry complexes (~,fequations 3-5) h The values for A log KMin this row were calculated according to equation 3 with the stability constants of the ternary M(Nta)(Im)- complexes given in reference 13 and those of the binary M(Im)2+ complexes given in reference 19 (regarding the latter values see also reference 10) Given above is the exact (unrounded) result of the previous experiments in reference 13 A log& % -0 1 is listed. regular octahedral sphere (Section 1).20 That this effect of a relative stability enhancement for the Zn2 + complexes is beyond that attributable to the combination of an 0-donor and a heteroaromatic n-accepting aminel 5,1 7,1 (Section 2) is evident from a comparison of the data between rows 2 and 3 or 4 and 5. Ternary complexes with ammonia are always less stable than the corresponding ones with imidazole yet even among the ternary complexes containing NH those with Zn2 + show an increased stability. We are convinced (Section 2) that the explanation for these observations is linked to a varying coordination sphere of Zn2 + which switcheseasily from coordination number 6 to 4or 5.23,20 The prime example for this behaviour is the Zn2 +-ethylenedia- mine (En) system cited in many textbooks (e.g. page 42 in reference 2 page 71 in reference 6a,or page 45 in reference 66) the Zn(En):+ complex is relatively stable while Zn(En)i+ is unstable.2 This difference is explained by attributing to Zn2 + in Zn(En)2 an octahedral coordination sphere (but see Section 5)+ which switches to a tetrahedral one upon coordination of the second En releasing four water molecules;? binding of the third En requires again an enlargement of the coordination sphere. In contrast with Co2 + the coordination sphere remains octa- hedra122 throughout the coordination process with En (for the revised discussion of the Zn2 +-En systems see Section 5). The results of Table 1 may be similarly explained. If Bicinate and Nta3 -bind to an octahedral Zn2 + and if upon the further binding of imidazole or ammonia the coordination number is reduced to 4 or 5 two water molecules are released upon the coordination of the monodentate ligand and such a switch in the coordination geometry is favoured. Clearly for Co2 +,Ni2+9 or Cd2+ with their more well-defined octahedral spheres such a process is less probable. The analogous explanation holds for the UTP4 -systems (Table 1). Considering UTP4- (Structure I) as a tridentate ligand coordinating via its triphosphate chaint3 one arrives14 at the following statistical values d logKs,,,h = -0.30 (cf. also reference 19) and d logKs,,c N -0.7. Comparison with the data of Table 1 indicates a steric hindrance of about 0.1 log unit at least for M(UTP)(Im)’ -with Co2+,Ni2+,and Cd2 + (neg-lecting any promotional effects due to the 0-donor/heteroaro- matic amine combinati~nl~). This hindrance is probably due to hydrogen-bonding between coordinated water and oxygens of the also coordinated triphosphate chain which would render P It is difficult to predict whether the driving force for the reduction of the coordination number IS enthalpic and/or entropic Owing to the opposingeffects of replacing six long bonds (0 74 pm) by four short bonds (0 60 pm see page 28 of reference 20) one may in a first approximation assume that the overall bond energies are similar in both geometries and that therefore the enthalpy change (AH) IS smdll hence the process appears as entropy-driven due to the release of extra solvent molecules from the metal ion-coordination sphere upon reduction of the coordindtion number (see also S Ahrland Struct Bonding,1973 15 167 Pure Appl Chem ,1979,51,2019) the replacement of a water in the coordination sphere of the metal ion by an incoming ligand more difficult. In any case Zn(UTP)(Im)2-has an increased stability (Table 1). 4 To What Extent is the Coordination Sphere of Zn2+ Altered upon Binding of lmidazole or Ammonia to Zn(Bicinate)+ Zn(Nta)- or Zn( UTP)2-? At this stage there is no final quantitative answer to this question yet an estimate is possible. If we define an octahedral sphere with coordination number 6 as ‘CN6’ and a tetrahedral one as ‘CN4’ and if we further express our uncertainty about the extent of the reduction of the coordination sphere of Zn2 by+ ‘CN4(5)’ indicating thus that a species with coordination number 5 could also be involved equilibrium 5a and the definition 5b may be rewritten Zn(B1c)+ + Im =ZncN,(Bic)(Im) *ZncNqn(Bic)(Im)+ + (ha) Zn(Bic)KZn(Blc)(lm) is the measured constant,1° but there follow two further definitions Zn(Bic) -[ZnCN6(Bic)(1m)+l (7)KZn(B~c)(lrn)/CN6 -[Zn(Bic) +][Im] This latter (intramolecular) dimensionless** equilibrium con- stant KI is of interest and based on equations 6b 7 and 8 the corresponding relations 9 and 10a are obtained (in analogy to other considerations on intramolecular equilibria) *12~24 ** K,IS dimensionless and attnbuted to an intramolecular equilibrium despite ZncNdc5,( +Bic)( Im) and ZncN6( Bic)(Im) + being no true isomenc complexes because they differ in the number of coordinated water molecules; yet the concentration of water in these diluted aqueous solutions is of course not significantly altered by this process CHEMICAL SOCIETY REVIEWS 1994 Instead of expanding 5a into 6a equilibrium 1 could have been re~onsidered,~to obtain equation lob which may be trans- formed into 1Oc with the definition 11 The expressions d log Kz in equations 10b and 1 1 are defined in equations 2 and 3 but now also additional subscripts are given for sure identification of the terms Kl may now be calculated with equations 1Oa or lob provided that values for Kg$&m)/CN6 or A log KZn/Blc/lm/CN6 can be obtained It is easier to estimate d log KZn/Blc/lm/CN6 and therefore equations lob and 10c are used in the following Of course d log KZnlBlcjlm(equation 11) is already determined (Table 1) The well-known enhanced stability of mixed ligand complexes formed by a metal ion of the second half of the 3dseries including Zn2+ an 0-donor and a heteroaromatic N-base with T-accepting properties (see Section 2) should be more pronounced with Co2 + than with Zn2 +,owing to the completed 3d shell of Zn2 Therefore d bgKCo/B,c/lm of the octahedral Co2+ may be + considered as the upper limit for d log KZn/Blc/lm/CN6 (equation 11) On the other hand the lower limit for d logKZn/BlcImlC-6 seems to be well represented by the statistical value d log Ksti oh = -0 5 (Section 2) which is also identical with that deter- mined for d lOgKCd/&c/lm/ (Table 1) This result is comforting because one could argue that the experimental values for the ratios K,/K etc are usually somewhat smaller than the statistical result due to steric effects and the like (e g ,see page 70 in reference 6a or page 44in reference 6b),however should this also be true for the systems presently considered the formation degree of the Zn&!j4 species would be even larger Hence we conclude that the lower and upper limits for Kl according to equations 8 and 1Oc may be calculated as well as the percentage of ZncN4(5)(Bicinate)(Im) (equation 13) in the (intramolecular) + equilibrium 12 (which is a part of equilibrium 6a) The same reasonings may be applied to the other ternary Zn2+ complexes listed in Table 1 for which the general 'intra- molecular' equilibrium 14 may be written This analysis assumes insignificant tendencies toward tetrahed- ral Zn2+ with the first ligand bound -which indeed appears valid at least for Zn(Bicinate) + However to the extent that there is significant (> 15%) Zn66 upon binding of the first ligand including it in the analysis would result in an increase in the calculated Kl and would narrow the difference between the upper and lower limits in Table 2 The calculations summarized in Table 2 show that an equili- brium exists between Zn&A6 and Zn&A4(j) In addition the values for d log Kzn (Tables 1 and 2) are such that equilibrium 1 as well as the analogous equilibria in Table 1 are in general weakly on the side of the ternary complexes which are formed only between approximately 50 and 6O% based on the total amount of Zn(Bicinate)+ Zn(Nta)- or Zn(UTP) -present For Zn2 +-Bicinate-imidazole this weakness means that about 60% are present as Zn(Bicinate)(Im) + and 40% as Zn(Bicinate)+ conse- quently about two thirds (70 f 15% Table 2) of 60% Zn(Bici- +nate)(Im) occur with Zn$A4(j) The equilibrium Zn&A6= Zn$&4(5) may shift to either coordination mode depending on the ligands binding to Zn2 +,though at this stage too little information is available to identify systematic trends 5 The Old Example 'par Excellence' Considerations on the Zn2 +-Et hylenediamine System 1,2-Diaminoethane or ethylenediamine (En) combines with Zn2+ to form 1 1 2 1 and 3 1 complexes with the observed successive stability constants logK$;[E? =log K = 5 92 logKt::E$ = IogK = 5 15 and logK&$ -logK = 1 86 (I= 1 M 25 "C)21 As indicated in the third paragraph of Section 3 the steep drop in the last constant has long been taken as strong evidence for a tetrahedral disposition of the four nitrogens in the 2 1 complex Zn(En); + Below we estimate the amount of tetrahedral Zn2 +,Zn& in the Zn2 + En = 1 1 and I 2 complexes by comparison with the successive En stability constants for the similarly sized Co2 + log K = 5 89 log K = 4 83 and log K = 3 10 (I = 1 M 30°C) 22 Comparison of the constants for Zn2+ and Co2+ reveals similar K values and a stronger K2 and a much weaker K3 for Zn2+ Without a heteroaromatic amine nitrogen no synergism among the ligands is expected hence we may simply use the Co2+-En system to represent an octahedral coordina- tion mode M$&6++ One may add that the statistical values'4 expected for an octahedral coordination sphere and binding of a bidentate hgand are log K -log K = log(5/ 2) -log( 12/1) = -0 68 and log K -log K = log( l/3)(4/5)-log(5/2) = -0 97 Comparison of these statistical differences with the experimental values obtained for the Co2+ system reveals that these are smaller by 0 38 and 0 76 log units respectively This is typical for dmines and shows that other factors like steric effects are also important (see e g page 44 of reference 6h) for this reason the statistical values are not employed in the evaluations given in Section 5 Table 2 Information on the intramolecular equilibria 12 and 14 regarding Zn&!j6 and Zn&&4(5) as calculated from equations 6 to I 1 together with the estimated percentages in which the Zr~&!j~(~) (equation 13) and Zn&!j6 species occur in aqueous solution at 25 "Cand I = 0 1 M Complex system limit A logKz," A log KZn CN6 A A log Kzn KI ?4 ZnE64c5 % Zn&' +Zn(Bic)(Im) lower 0 31 -0 05" 0 36 1 29 ;:}70 f15d 30f 15 upper 0 31 -0 50' 0 81 5 46 Zn(Nta)(Im) -lower 0 22 0 01' 0 21 0 62 upper 0 22 -0 50' 0 72 4 25 ;:}6O f25d 40 f25 Zn(N ta)( NH3) lower -0 09 -0 26" 0 17 0 48 f1.P 55f 15 upper -0 09 -0 50 0 41 157 Zn(UTP)(Im)' lower 0 03 -0 36' 0 39 145 ::}60 flod 40* 10 upper 0 03 -04Of 0 43 1 69 Values from Table 1 cf equations 2 3 Lower limit regarding Ki (equations 8 10) the value corresponds to A log K of Table 1 as Justified in the text of Section 4 it is needed in equations 10 1 1 Upper limit regarding K the value corresponds to d log K ah and in the case of the Bic system also to A log KCdof Table 1 (see also b) These estimates are the rounded averages of the lower and upper limits together with also estimated error limits These estimates follow from 100 -Yo Zn&& The statistical value A log K oh = -0 30 was in this case (~falso L) reduced by 0 1 log unit to account for steric effects (see the last pardgraph in Section 3) THE FLEXIBLE COORDINATION SPHERE OF ZINC(I1) IN SOLUTION-H. SIGEL AND R. B. MARTIN 5.1 Calculation of the Percentage of Zn& in En Complexes The binding of En or any bidentate ligand to a hexacoordinate metal ion designated as McN~ involves three successive equili- bria (charges are ignored) The 1:l and 2:l complexes are in (an intramolecular) equili- brium with their four coordinate MCN4,counterparts For such a system the observed stability constants are given by equations 20 to 22 where the right-hand part follows from combinations with equations 15 to 19 Owing to the inherent differences in binding strengths between + +the Zn2 and Co2 complexes these equations cannot be used directly. However by taking the ratios of the observed stability constants we normalize for these inherent differences -= KliCN6(1+ K$M/En)(l + Kc6/2En) (25)Kl K3 K3/CN6 The stability constants KlICN6 K2IC-6 and K3/CN6 for the hexacoordinated octahedral complexes are taken from the Co2+-En system. These constants also appear as ratios in equations 23 to 25,as is the case for the constants K K2,and K3 attributed to the Zn2+-En system. Any two of the last three equations may be solved simultaneously for 1 + KfiMiEn and -t-Kfih/2En-With the stability constants given in the first two paragraphs of Section 5 we calculate that KfiZn/En = 1.12 for Zn(En)2 + corresponding to 53% ZnCN4(En)2(cJ equation 13) and Kfi2,,+ 2En = 7.8 for Zn(En)$+ corresponding to 89% ZncN4(En)$+. Thus about half of the 1 :1 and about 90% of the 2:1 complexes are tetrahedral. 5.2 Conclusions from the Evaluation of the Zn2 +-En System Three main points follow from Section 5.1 and we list these in order of their decreasing importance (I) The formulation regarding MCN6'- MCN4 for the Zn2 +-En system applies not only to other bidentate ligands but may easily be generalized to the successive additions of any number of identical ligands to any metal ion in any pair of coordination numbers. The only restriction is that to apply it successfully one must be able to estimate the ratios of the KnICN6 constants or of their analogues. (2) Our calculations show that for Zn(En)2 about half of the + species are octahedral and half tetrahedral (equation 18).This evidence that about 50% of the Zn(En)2 + species already exist with a lower coordination number in aqueous solution is an important new insight; previously it was believed (Section 3) that tetrahedral complexes occur only with Zn(En)$ +. (3) The result of a degree of formation of about 90% for ZncN4(En)$ agrees with the previous assumption (Section 3)+ that Zn(En)$ in aqueous solution is tetrahedral. However + the presence of about 10% of Zn(En)$+ as an octahedral species is consistent with the formation of Zn(En) + by a large excess of En over Zn2 + . 6 Considerations on Some Other Zn2+ Systems with Bidentate Ligands 6.1 The Oxalate and Glycinate Systems are Octahedral! After evaluation of the Zn2 + system with the bidentate ethylene- diamine an N-donor forming 5-membered chelates it seemed appropriate to deal with a bidentate O-donor that also forms 5-membered rings; hence we considered oxalate (Ox2 -). The stability constants given in the literature' 6a-c.g,h for the com- plexes with Zn2 + and Co2 + disagree making it difficult to select the most reliable values; therefore we averaged all the available constants (close to 25°C and I= 0-2 M) and obtained log K$Ejoxi= 4.30= logK (averaged from 11 values),logKzn(ox)2ZnOx = 2.52= logK (lo) and logKg::gi{; = 0.98= lo K (3) as well as log&& = 4.13 (from 17 values) log K&&2 = 2.32 (12) and log&^^^^^; = 1.47 (1). Comparison shows that the K1/K2ratios are very similar for Zn2+ and Co2+ (101.81),indicating that there is no pronounced tendency towards a tetrahedral geometry in Zn(0x)f -; this conclusion agrees with the K2/K3 ratios though their compari- son is less meaningful since for Co(0x):- only a single stability value is available. A ligand that fits into a comparison with ethylenediamine and oxalate is glycinate (Gly -). 6a-c*e Here we could rely on a recent 'Critical Survey of Stability Constants of Complexes of Gly- ZnGly -Cine':25 logk&,y) = 5.03 logKZn[Gly]z-4.20 and =lo KZn(G'y)z2.54 as well as log = 4.66 log K-Co(Gly)z-3.85,and logh$~~~$~~ c0?Gly7n(_G1y)3 = 2.32 (25 "C; I= 0.1-0.2 M). For glycinate the Kl/K2and K2/K3ratios are closely similar for both theZn2+ and and Co2+ and complexes suggesting that this bidentate N,O-donor ligand preserves octahedral hexacoordinate Zn2+ through all complexes. 6.2 The Zn2 +-Histamine System has a Significant Tetrahedral Portion For biological systems histamine (Ha; Structure 11)is of special interest. From all the stability constants listed for its Zn2 + and Co2+ complexes in16a-cf,h,i the values given in26 appear the best these constants refer to 37"Cand I = 0.15 M,and the differences between two sources16i,26 for the logK and logK values are < 0.08log unit for the Zn2 + and Co2 + complexes. We therefore use the following averages log K$:(Ha)= 4.99= log K and log Kg,"l::{2= 4.78= log K2; log A$&,) = 4.90 and log&$$~~,3.54.The values for the 1:3complexes are in each case = given only once logK$:# = 2.2826= logK and log&$$Ei;; = 2.27.16' The ratio K1/K2= for the Zn2+ complexes is much smaller than that of for the Co2+complexes while the ratio KZ/K3 = 102.50is much larger for the Zn2 + species than for those of Co2+with This comparison suggests a pro- nounced degree of formation of tetrahedral Zn2 + complexes. Application of the procedure according to equations 23-25 of Section 5.1 leads to a negative value for KfizniHa(equation 1 8) which can at the minimum be zero meaning that no ZncN4(Ha)2 species are formed. This negative result further + indicates that something must be wrong. A careful evaluation of the mathematical calculation shows that the negative value for Kfizn Ha is due to an oversized logK,; i.e.,the stability constant logK&~~:= 2.28 is too large! Indeed none of the other studies listed in16u-Cdh,i give a stability constant for the Zn(Ha):+ complex; thus the degree of formation of this species must be low. We suggest that unrecognized hydroxo-complex formation led to the above irregular result. The mathematical evaluation shows further that the upper limit for log K3 = 1.21;only then K$zn/Haturns from a negative value to zero. With this value also the lower limit for equation 19 is obtained; i.e.,Kfi$n/2Ha > 3. Hence the degree of formation of ZncN4(Ha)$+ is larger than 75%. Moreover based on our experience and comparisons of related data we estimate for Zn(Ha)g+ the stability constant logK$:$:{ = 0.7 = logK a value also in accord with the Zn2 +-En system (Section 5). This constant together with the other values listed above and appli- cation of equations 23-25 gives the following results Kfiznl Ha = 0.48 and 32% of ZIICN~(H~)~+,as well as Kfi2n/2Ha = 30 and 97% of ZncN4(Ha)$ +. 6.3 The ZnZ +-Histidine Methyl Ester System Confirms the Importance of Zn;A4 Species for Histamine-type Ligands Considering the importance of the imidazole group as binding site in Zn2+ enzymes we included in our evaluations the bidentate histidine methyl ester (HisMe; Structure 111). For the Zn2 +-HisMe system the first two stability constants of the Zn2 + complexes differ only slightly log @;(HlsMe) = 4.46 = log K and logKZn(HisMe) -4.20 = l~gK,;~'Zn(H1sMe) -in line with our conclusions at the Zn2 +-Ha system (Section 6.2) K3 is indeterminately No experiments with Co2+ were conducted but the same reports for Ni2+ the stability constants logK = 6.19 10gK2 = 4.91 and 10gK3 = 2.90. We compare these constants despite the indication that a square-planar Ni(HisMe)$+ species might occur to some extent; in fact for Zn2+-HisMe we aim only for the limiting conditions of the Zn& formation. 0 As before in the Zn2 +-Ha system (Section 6.2) and now with Zn2 +-HisMe the value for KfiZn/HisMe is difficult to obtain. For the necessary condition that KfiZn/HisMe >0 it may be shown that relation 26 holds (in fact this condition was indirectly applied already in Section 6.2 in the arguments about the validity of logK$:$:i = 2.28 and the replacement of this value by an estimate) With condition 26 and using the Ni2 + results as reference for K CN63 K2/CN6 and K3/CN6 we find logK3<0.2 and K$?k/ 2HisMe > 10 meaning Zn5A4 must comprise more than 90% of the Zn(HisMe)$ species. These conclusions are consistent both + n(HisMe),with the experimental difficulty of determining K3 = Zn(HisMe) as well as with the results for the Zn2 +-Ha system. CHEMICAL SOCIETY REVIEWS 1994 6.4 Conclusions from the Evaluation of the Various Zn + Systems Containing Bidentate Ligands From the results presented in Sections 6.1 to 6.3 it follows (1) Bidentate ligands like oxalate or glycinate do not enforce a significant formation degree of tetrahedral Zn$$j4 in any of the 1:1 or 1:2 complexes. (2) The results of Sections 6.2 and 6.3 allow the conclusion that 1:l complexes of histamine-type ligands exist to about 30% with a Zn$!j4 coordination sphere. In the 1:2 complexes the tetrahedral portion increases to more than 90%. Conse- quently the stability of 1:3 complexes is low and (if at all) difficult to determine. (3) A small logK,/K2term requires a large 10gK2/K3 term of a Zn2+ system in comparison with a 'normal' set of reference constants for an octahedral system on the right-hand side of the inequality relation 26. This means that the more strongly the 1:2 Zn2+ complex goes tetrahedral the greater the diffi- culty of adding a third bidentate ligand to yield again a hexacoordinate Zn2 +. 7 Estimations on the Extent of Zn& in Some Mixed Ligand Complexes Containing Bidentate Ligands The stability of any mixed ligand M(A)(B) complex may be quantified analogously to equilibrium 1 (Section 2); its general form (neglecting charges) is M(A) + M(B)*M(A)(B) + M (27) and the corresponding equilibrium constant 1Od logKM (equation 3) may be calculated via equation 28 Table 3 lists results for mixed ligand complexes formed with histamine (Ha) ethylendediamine (En) and the glycinate-like serinate (Ser-); i.e. for ligands or ligand-types we have con- sidered in their binary parent complexes in Sections 5 and 6. Comparison of the equilibrium constants in Table 3 for the complexes containing Ha confirms the well-known discriminat- ing properties and the stability enhancing effect (in the presence of O-donor sites) of an imidazole group i.e. of a heteroaromatic amine;' 531 * a topic extensively discussed' -l 5.1 '-19 (c$ also Sections 2 and 3) and not considered further here. In the present context the increased stability of all the ternary Zn2 + complexes compared to the corresponding Co2 complexes is of interest. + This is again evidence for the formation of tetrahedral Zn$$j4 species! In view of the difficulties encountered with log K::iE; for the octahedral Zn(Ha)$ + complex (Section 6.2) we attempt here Table 3 Comparison of the relative stability of various mixed- ligand complexes according to the equilibrium M(A) + M(B)=M(A)(B) + M and the corresponding constant A log KM(equation 28) for aqueous solutions at 37 "Cand I = 0.15 M (KNO3Y A log KM + +A B coz NiZ cu2 + Zn2+ Ha En -0.88 -1.20 -1.43 -0.19 Ha Ser-0.48 -0.83 -0.58 + 0.17 En Ser-0.46 -0.41 -0.87 -0.14 (1 The values were calculated from the stability constants provided by Perrin et ul. Coz+,Znz+;26 NiZ+ (D. D. Perrin and V. S. Sharma J. Chem. SOC.(A) 1968,446);Cuz+(D. D. Perrin I. G. Sayce and V. S. Sharma J. Chem. SOC. (A) 1967 1775). THE FLEXIBLE COORDINATION SPHERE OF ZINC(I1) IN SOLUTION-H SIGEL AND R B MARTIN only to obtain an estimate for the degree of formation of ZncN4(A)(B) by applying equations analogous to equations 6-11 I e we will proceed as in Section 4 However we now consider the equilibrium constants of the Co2 + complexes rather as the upper limits or even the 'true' values for the Zn&!& complexes assuming because of the similar size of Co2+ and Zn2+,a comparable steric hindrance among the ligands in the coordination sphere of the two metal ions and that this also applies for the already mentioned 'heteroaromatic amine effect' where appropriate The louer limit is most probably well represented by the statistical value (st) for dlogKM for the coordination of the two different bidentate ligands A and B Ligand B has 12 possibilities for binding to an octahedral (oh) coordination sphere,I4 yet only 5 for binding to an octahedral M(A) species but the probability for dissociation is of course 1 in both cases hence dlogKoh,s,= log (5/1) -log (12/ 1) = -0 38 We attribute the lower limit to dlogKoh because the statistical ratios for K2/K,(etc ) are often larger than those found experimentally (cf footnote p 86) The preceding reasoning leads to the results in Table 4 These data are only estimations and should not be overinterpreted because in this evaluation the presence of Zn,5h4 in any of the binary Zn(A) or Zn(B) complexes is ignored However consi- dering them would lead to even higher percentages for ZncN4(A)(B). hence the results listed provide str6ng evidence that in all these mixed ligand complexes equilibrium 14 is operating and that significant portions of the Zn(A)(B) species have a tetrahedral geometry In this sense these results further generalize the observations described in the preceding sections In addition they indicate that a N20zdonor set as present in Zn(En)(Ox) gives rise to a significant portion of a Zn6h4 coordination sphere while this is not the case for the same donor set in Zn(Gly) (Section 6 1) The difference is evidently that in Zn(En)(Ox) the two crucial N-donor sites are within the same ligand and thus lead to a czs-type coordination in an octahedral coordination sphere Moreover the present result for ZncN4 (En)(Ox) agrees with that obtained in Section 5 for znCN4(En)(H20)' + 8 Imidazole and the Zn& Coordination Sphere which Facilitates Hydroxo-Complex Formation We have already seen in Section 4 that upon its coordination to ZnZ+,imidazole (Im) initiates a tetrahedral geometry in mixed- ligand complexes Now we ask What are the properties of the binary Zn2 +-imidazole system' For our evaluation we use the following stability constants which refer to 25 "Cand I = 0 16 M 28 logK$:,i = 2 57 = logK lo K$$:E{ = 2 36 = 10gK2 logK~~{~~{:= 2 22 = logK and logK!:if;{; = 2 01 = 10gK,,~~" for the Co2+-Im system the values are logK = 242 10gK2= 1 95 logK = 1 58 and logK = 1 2 28h These stability constants yield for the Zn2 + system the ratios K1/K2= loo 21 K2/K3= loo 14 and K3/K4= loo 21 for the Co2+ system they yield loo 47 loo 37 and loo 38 respectively These ratios are significantly smaller throughout for the Zn2 +-Im system than for the Co2 + one clearly indicating a progress- ing tetrahedral geometry with an increasing amount of imida- zole in the coordination sphere of Zn2 + Unfortunately an exact evaluation (cf Section 5) is not possible because the log K value for the Zn2 +-Im system is not known though for Co(1m)z and+ Co(1m); + estimates do exist 66f Rough estimations however suggest that Zn(Im)2 + occurs as about 20% ZncN4(Im)2+ and Zn(Im)f+ as about 50% ZncN4(Im)f+,the tetrahedral species is formed to approxima- tely 90% for Zn(Im)$ + and to more than 98% for Zn(Im)i+ In fact the latter two degrees of formation are most probably even more pronounced at high ionic strength as revealed by the following stability constants which refer to 25 "Cand I = 3 M (NaClO,) 29 log K = 2 92 logK = 2 01 logK = 3 84 and log K4 = 2 64 Under these conditions the stability of Zn(1m)Z + exceeds that of Zn(Im)2 + and Zn(1m)t + by factors of about 8 and 70 respectively this implies that already the formation degree of ZncN4(Im)5+ is large and that apparently the switch from Zn& to Zn&A4 occurs mainly upon addition of a further imidazole to Zn(Im)$ + This conclusion confirms the observation that under the same conditions (25 "C,I = 3 M NaClO,) Zn(Im)3(H20)2+ loses a proton with pK = 8 0 29 30 This impressive acidification of a Zn2 +-bound water by the coordinated imidazoles is of biologi- cal significance 30 31 From the Zn2 + complexes and the various pK values assembled in Figure 2 it clearly follows that this acidification is connected with the formation of Zn&k4 No ligands with a special structure have to be devised since ligands with a single coordinating atom are enough for this effect provided that upon their coordination a tetrahedral coordina- tion sphere at Zn2 + is enforced Imidazole is such a Iigand and nature makes plentiful use of this ligating moiety in many Zn2 +-enzymes That hydroxo-complex formation and tetrahedrality facilitate each other is also apparent from the tetrahedral Zn(OH),(H,O)- which is the principal zincate ion in aqueous solution (see page 598 in reference 6a or page 605 in reference 6b) though in strongly alkaline media the (also) tetrahedral Zn(0H);-may be formed 32 9 Conclusions and Outlook To our knowledge this is the first time that increased relative complex stabilities in low-molecular-weight ternary Zn2 + com-plexes have been explained quantitatively in terms of decreasing coordination numbers of Zn2 + Observation of posrtrve values for dlogKz (equations 2 3) and correspondingly the shift of equilibrium 1 to the side of the mixed ligand complex leave no other reasonable explanation Indeed consideration of the intramolecular equilibria 12 and 14 and estimations for the corresponding species distributions prove that in Zn(Bicina- Table4 Information on the intramolecular equilibrium 14 regarding Zn&A4 as calculated in analogy to equations 6 to 11 together with the estimated percentages in which ternary species Zn&A)(B) occur in aqueous solution at 37 "C and I = 0 15 M Complex system limit A log Kznu A log KZn CN6 Ad log Kz Kl YOZn$A,(A)(B) +Zn(Ha)(En) lower -0 19 -0 3Sh 0 19 0 55 :;}60 f30 upper -0 19 -0 88' 0 69 3 90 +Zn(Ha)(Ser) lower 0 17 -0 3Sh 0 55 2 55 ;;}75 f 10 upper 0 17 -0 48' 0 65 3 47 -Zn(En)(Ser)' lower -0 14 0 3V 0 24 0 74 upper -0 14 -0 46' 0 32 1 09 Zn(En)(Ox) lower 0 05" -0 38' 0 43 1 69 63 2 60 Values from Table 3 * Lower limit regarding K the value corresponds to A log K oh for bidentate ligands see Section 7 Upper limit regarding K the value corresponds to A log K of Table 3 as justified in the text of Section 7 This value refers to 25 "Cand I = 1 M (KNO,) it is calculated from the stability constants provided by Y Kanemura and J I Watters J Inorg Nu~lChem 1967 29 1701 CHEMICAL SOCIETY REVIEWS I994 -L pKa = gCa’ pK = 8.3cb’ p/d = 80‘‘’ pK = 73cd1 Figure 2 Effect of the coordination sphere on the deprotonation of H20 bound to Zn2 The pKd values given refer to the reaction Zn”(X)- + (H,O)*Znll(X)(OH-) + H+ These acidity constants are taken from the following sources pKd = 8 95 f0 15 is the average of selected values from references 16a,b,d at 25 “C and Z= 0 -2 M The above value is from kinetic experiments and refers to a 33% ethanol/H,O mixture at 25 “C with Z= undefined (probably in the order of 0 1 M) potentiometric pH titrations gave pK = 8 16 f0 10 from R G Clewley H Slebocka-Tilk and R S Brown Znorg Chim Acta 1989 157 233 We estimate that the corresponding pK value in aqueous solution would be lower and close to 7 8 25 “C and Z= 3 M (NaClO,) from references 29,30 It may be added that at an ionic strength lower than 3 M the pK value of Zn(Im),(H,O)* + is most probably somewhat below 8 as is indi-cated by the pK values for Zn(H20)i+ available at various ionic strengths 16rr 25 “Cand I = 0 1 M (NaCIO,) from reference 9a te)(Im) + Zn(Nta)(Im)- Zn(Nta)(NH,) and Zn(UTP) (Im)2-at least two isomers with different coordination numbers of Zn2 + are present simultaneously (Table 2) Moreover the results described including those for the binary Zn2 +%thylenediamine -histdmine and other systems (see Sections 5-8) indicate that in principle in any Zn2+ system in (aqueous) solution species are present which contain Zn2+ with coordination number 6 as well as further species having Zn2 + in lower coordination numbers mainly 4and with 5 as intermediate Depending on the conditions especially the ligands present the intramolecular equilibrium can be largely on the side of 6-fold coordinated Zn2+ but also on that of 4-coordinated Zn2+ giving the impression that only a certain (single) Zn2 +-coordination sphere occurs in the system Of course with sterically more rigid ligands like macrocycles certain coordination numbers for Zn2+ may be enforced9 to some degree However the common claim that the stereochemistry of a metal ion in an enzyme is ‘undoubtedly’ controlled by the ligand geometry (e g ,references 8,33) is only partly correct Certainly Zn2+ submits readily to structural demands of a ligand,’ but the present evaluations demonstrate that a tetrahedral Zn2 + is also created by flexible and highly adaptable ligands like ethylenedia- mine or histamine in fact the best example here is the monoden- tate imidazole Overall it appears that the formation of ZntA is driven by the Lewis basicity of the donor atoms if the coordinat- ing ligand is a strong Lewis base,34 the coordination number of Zn2+ drops and the bond length shortens I e ,the metal-ligand bond becomes more of the covalent type This explains why an imidazole group promotes Zn&k4 formation more effectively than an amino group and this group more than an 0-donor Considering the adaptability of Zn2 +-coordination spheres in biological systems the results described and their explanations are also meaningful An incoming substrate is favourably bound and also enhances the Lewis acidity of the Zn2 + (cf also Figure 2) if the coordination number is reducedand thus more water (or other bound sites or) molecules are released in addition this property allows that during a reactive transition a further donor site can easily be bound wiihout releasing another group from the coordination sphere As indicated above the reduction from coordination number 6 to (5 or) 4is favoured by the participa- tion of N-donor ligands (or sites) This reduction accords with the situation in biological systems where the extremely versa- tile‘sh imidazole group is often involved ’ ’ To conclude the strategies described here for the quantifica- tion of the varying coordination numbers of Zn2+ in its com- plexes in aqueous solution may also be applied to other metal ions such as A13+,Ag+ and Hg2+ and ions that undergo a change of spin state such as Ni2+ Further insights into the ‘flexibility’ of metal ion coordination spheres in solution are thus to be expected a topic so far not appropriately considered -possibly due to the lack of methods -in coordination chemistry Acknowledgements The support of this research by a grant from the Swiss National Science Foundation (H S ) and the compe- tent technical assistance of Ms Rita Baumbusch during the preparation of this manuscript are gratefully acknowledged 10 References 1 B L Vallee and D S Auld Biochemistry 1990,29 5647 and 1993 32,6493 2 J J R Frausto da Silva and R J P Williams ‘The Biological Chemistry of the Elements’ Clarendon Press Oxford 1991 p 299 3 (a) ‘Metal Ions in Biological Systems’ Vol 15 Zinc and Its Role in Biology and Nutrition ed H Sigel and A Sigel Marcel Dekker Inc New York and Basel 1983 (b) ‘Progress in Inorganic Bio- chemistry and Biophysics’ Vol 1 Zinc Enzymes ed I Bertini C Luchinat W Maret and M Zeppezauer Birkhauser Verlag Basel 1986 4 D W Christianson Adv Protein Chem 1991,42 281 5 J P Glusker Adv Protein Chem 1991,42 1 6 (a) F A Cotton and G Wilkinson ‘Advanced Inorganic Chemistry’ 4thEdn ,J Wiley New York 1980 p 599 (b)SthEdn 1988 p 607 7 H B Silber D Simon and F Gaizer Inorg Chem 1984,23,2844 C 0 Quicksall and TG Spiro Znorg Chem 1966,5,2232 8 R S Brown J Huguet and N J Curtis Met Ions Bid Syst 1983 IS,55 9 (a) E Kimura T Shiota T Koike M Shiro and M Kodama J Am Chem Soc 1990 112 5805 (b) E Kimura H Kurosaki T Koike and K Toriumi J Inclusion Phenom Mu1 Recogn Chem 1992,12,377 M Shionoya E Kimura and M Shiro J Am Chem Soc 1993 115 6730 (c) A Looney G Parkin R Alsfasser M Ruf and H Vahrenkamp Angew Chem 1992 104 57 (Angeu Chem Znt Ed Engl 1992,31,92) R Alsfasser S Trofimenko A Looney G Parkin and H Vahrenkamp Znorg Chem 1991 30 4098 10 L -n Ji N A Corfu and H Sigel Inorg Chim Acra 1993,206,215 11 H Sigel Coord Chem Rev 1993 122,227 12 R B Martin and H Sigel Comments Inorg Chem 1988,6 285 13 D Banerjea TA Kaden and H Sigel Inorg Chem ,198 1,20,2586 4 H Sigel Angeu Chem 1975,87,391 Angew Chem Int Ed Engl 1975 14,394 5 H Sigel ‘Stability Structure and Reactivity of Mixed Ligand Complexes in Solution’ in Coordination Chemistry -20 ed D Banerjea IUPAC Pergamon Press Oxford and New York 1980 p 27 6 (a) L G Sillen and A E Martell ‘Stability Constants of Metal-Ion Complexes’ Special Publication No 17 The Chemical Society London 1964 (b)L G Sillen and A E Martell ‘Stability Constants of Metal-Ion Complexes’ Suppl 1 Special Publication No 25 The Chemical Society London 1971 (c) D D Perrin ‘Stability Con- stants of Metal-Ion Complexes’ Part B Organic Ligands IUPAC Chemical Data Series No 22 Pergamon Press Oxford 1979 (6)E Hogfeldt ‘Stability Constants of Metal-Ion Complexes’ Part A Inorganic Ligands IUPAC Chemical Data Series No 21 Pergamon Press Oxford 1982 (e) A E Martell and R M Smith ‘Critical Stability Constants’ Vol 1 Amino Acids Plenum Press New York 1974 (f) R M Smith and A E Martell ‘Critical Stability Con- stants’ Vol 2 Amines Plenum Press New York 1975 (g) A E Martell and R M Smith ‘Critical Stability Constants’ Vol 3 Other Organic Ligands Plenum Press New York 1977 (h)A E Martell THE FLEXIBLE COORDINATION SPHERE OF ZINC(I1) IN SOLUTION-H SIGEL AND R B MARTIN and R M Smith ‘Critical Stability Constants’ Vol 5 1st Suppl Plenum Press New York 1982 (I) R M Smith and A E Martell ‘Critical Stability Constants’ Vol 6 2nd Suppl ,Plenum Press New York 1989 17 H Sigel B E Fischer and B Prijs J Am Chem SOC 1977 99 18 (a)P R Huber R Gnesser and H Sigel Inorg Chem 1971 10 945 (6) H Sigel Inorg Chem 1980 19 I41 1 19 N Saha and H Sigel J Am Chem Soc 1982,104,4100 20 R B Martin Met Ions Biol Syst 1986 20 21 21 J Bjerrum Chem Rev 1950,46 381 22 J Bjerrum ‘Metal Ammine Formation in Aqueous Solution’ P Haase and Son Copenhagen 1941 & 1957 (taken from ref 16a) 23 (a)R B Martin and Y H Mariam Met Ions Biol Sjst ,1979,8,57 (b)H Sigel Eur J Bzochem 1987 165 65 H Sigel Chem SOC Rer 1993,22,255 24 B E Fischer and H Sigel J Am Chem Soc 1980,102,2998 S S Mdssoud dnd H Sigel Inorg Chim Acta 1989 159,243 25 TKiss I Sovago and A Gergely Pure Appl Chem 1991,63,597 26 D D Perrin and V S Sharma J Chem SOC (A) 1969,2060 27 H L Conley Jr and R B Martin J Phys Chem ,1965,69,2923 28 (a) Y Nozaki F R N Gurd R F Chen and J T Edsall J Am Chem Soc 1957,79,2123 (b)R B Martin and J T Edsall J Am Chem SOC 1958,80 5033 29 W Forsling Acta Chem Scand 1977,31A 759 30 R B Martin Met Ions Biol Syst 1979,9 I see page 29 31 D K Wilson and F A Quiocho Biochemistry 1993,32 1689 32 N Wiberg ‘Holleman-Wiberg Lehrbuch der Anorganischen Che- mie’ 9ISt-10Oth Edn ,W de Gruyter Berlin and New York 1985 p 1038 33 R H Prince Adv Inorg Chem Radiochem 1979,22,349 34 I D Brown Acta Cryst 1988 B44,545 35 R J Sundberg and R B Martin Chem Rev ,1974,74,471 H Sigel and R B Martin Chem Rev 1982,82 385 H Sigel R Tribolet and 0 Yamauchi Comments Inorg Chem 1990,9,305

ISSN:0306-0012    

DOI:10.1039/CS9942300083

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

The Diagnosis of Concerted Organic Mechanisms A. Williams Department of Chemistry University of Kent Canterbury Kent CT2 7NH U.K. ISetting the Scene Almost all discussions of mechanism involve the concept of concertedness. The relative sequence of bonding changes predi- cates the charge distribution in a reaction path and its revelation is therefore of the utmost importance to our understanding of mechanism. Despite intensive study it is surprising how little evidence is available which unambiguously excludes stepwise paths for the majority of reactions. The absence of evidence for intermediates (such as stereochemical integrity) while consistent with the postulate of a concerted mechanism does not exclude stepwise processes except in special cases which are the subject of this review. A concerted mechanism has no intermediates and thus pos- sesses only a single transition state. lo Although other definitions have been applied,lb the above is probably the most universally accepted especially as it enables unambiguous diagnosis. A definition involving partial bond formation and fissionlb pres- ents an operational problem of distinguishing between hybridi- zation changes and bond fission; moreover it tends to exclude those mechanisms where a bond change is almost or only partially complete in the transition state. A two-step consecutive process would give rise to an observation of partial bond formation and fission when neither step is rate limiting because both steps would then contribute to the rate. The preferred definition allows of concerted mechanisms where a particular bond change may not be very far advanced and thus encom- passes the concept of variable transition states. The term ‘synchronous’ (see Figure 1) describes concerted mechanisms where the extents of bond fission and formation are equally advanced in the transition state. Concertedness refers to the major bonding changes in the reaction and does not include any initial or final steps (such as formation of encounter complexes) required to complete the overall process. This review discusses contemporary work on some reactions hitherto regarded as not involving concerted mechanisms. The review emphasizes that out knowledge of whether or not a reaction proceeds via a concerted mechanism as opposed to a stepwise one is not firmly established in the majority of cases and it covers techniques developed over the past ten years for excluding stepwise or concerted mechanisms2 Since they are already covered at length in many general texts reactions Andrew Williams obtained his D.Phi1. under Gordon Lowe at Oxford in 1964 and spent a postdoctoral year with Myron Bender at Northwestern before being appointed lecturer at the Uni- versity of Kent. He worked in Bill Jencks’ laboratory at Brandeis during 1972-73 and was a senior Ciba-Geigy Fel- low at Genoa in 1984; he was elected Professor of Organic Chemistry at Kent in 1987 and has served as Chairman of the Department. He is interested in the fundamental mechanisms of bio-organic reactions in solu-tion including enzyme- and polymer-catalysed reactions. 93 NU-X-Lg NU-A LZi Figure 1 Reaction map for a general transfer reaction involving two major bond changes; the concerted path can take any route from reactants (A-Lg + Nu-) to products (Nu-A + Lg-). I and I1 refer to stepwise associative and stepwise dissociative mechanisms respect- ively. I11 and IV are transition states for enforced concerted mechanisms. classically assumed to involve concerted mechanisms namely E2 SN2,and cyclical rearrangements are not dealt with in depth. Interconversion between various molecular structures is a dynamic phenomenon. The mechanism for a particular inter- conversion describes the structures of intermediates (that is discrete compounds) and the structures between the inter- mediates on the reaction path. The use of explicit ‘KekulC’ structures has served very well to enable organic chemists to visualize mechanism graphically and also to enable predictions to be made. This method has the following shortcomings (a) it does not display the solvent-solvent and solvent-solute interac- tions in a solution reaction; (b) it does not pay due regard to the fact that the reaction involves large collections of reacting molecules and that the Kekule structure is only an average; (c) the application of Kekule structures to the reaction path between reactants products and intermediates is not strictly valid as these are not of discrete compounds except at energy minima. Progress in elucidating reaction mechanisms was dra- matically advanced when reactions were considered as intercon- versions of states (rather than of single molecules) and the use of the term transition STATE should continually emphasize this approach. The determination of transition-state structure does not easily follow from various kinetic parameters (which refer to collections of molecules) and a major challenge is to improve the techniques by which these parameters are translated into the Kekule idiom. Reactions in solution involve the collision of solvated mole- cules to form an arrangement which becauses of inertia caused by solvent bulk has a finite lifetime. During this period mole- cules enter the same solvent shell the molecules then mix their electronic orbitals to give product and a similar process in reverse occurs to release the products to bulk solution. At all stages solvent is intimately involved with the reaction. The three- dimensional energy diagram (Figure 1) demonstrates the con- nection between concerted and stepwise mechanisms and refers to reaction within the solvent shell. 2 Bonding Indicators There are numbers of techniques which attempt to reveal the extent of bonding in the transition state in solution reactions. Partial completion in the transition state of bond-breaking and forming steps is consistent with concertedness but formation of an intermediate could involve changes in hybridization in a bond which only subsequently undergoes fission or formation. For example it is well established that attack of hydroxide ion on esters involves a tetrahedral intermediate although the nature of the bond to the leaving group changes on formation of the intermediate. The observation of changes in bonding to entering and leaving groups would therefore not be diagnostic of concer- tedness unless there were some way of comparing these with expected hybridization changes. A process involving passage through a structure correspond- ing to a species which is still an intermediate but has negligible barrier to decomposition is defined as an enforced concerted mechanism. An enforced concerted mechanism would result from change in the conditions in a stepwise process (I Figure 2) decreasing the stability of an intermediate to the point where its barrier to decomposition (11 Figure 2) has a lifetime commen- surate with a vibration period (ca. 10-l3 seconds). The ‘enforced concerted’ mechanism is on the borderline between stepwise and regular concerted mechanisms (111 Figure 2) and possesses no energy well at the structure corresponding to that of an intermediate. hP w 1 Reaction coordinate Figure 2 Two-dimensional energy diagram illustrating an enforced concerted mechanism (11) on the borderline between stepwise (I) and ‘regular’ concerted (I) paths. 3 Energy Considerations2 The energetic advantages of a concerted process are that (a) when the timing is substantially synchronous there is no large build up of charge in the transition state and (b) bond fission is aided by the energy released on bond formation. The energy of activation is usually less than the sum of the bond energies of the bonds undergoing major change. The disadvantage of a con- certed process is that it generally requires more nuclear motion than do the individual steps of the corresponding stepwise process. Represention of reaction mechanisms by a series of curved arrows to simulate electron flow and to account for electron movement too easily degenerates into the description of a mechanism by a concerted process without attention being paid to its experimental demonstration. 4 Techniques for Excluding Mechanisms 4.1 General Techniques for demonstrating intermediates and excluding con- certedness are manifestly important in studies of timing of bond changes. We shall be discussing methods for detecting inter- CHEMICAL SOCIETY REVIEWS 1994 mediates at concentrations too low for general instrumental techniques. 4.2 Stereochemistry The discovery by Walden in the 19th century that change in configuration takes place during the course of aliphatic substitu- tion reactions was followed in the first third of this century by a vigorous series of experiments which culminated in the formula- tion of the SN2concerted mechani~m.~ Despite its importance in formulating the concerted mechanism the phenomenon of inversion has long been recognized as not being able to exclude stepwise mechanisms. Racemization or partial racemization is diagnostic of a symmetrical intermediate and therefore excludes a concerted mechanism. The phenomenon of retention of configuration also diagnoses intermediates -for example many enzyme phosphotransfer reactions involve retention at the isotopomeric phosphorus atom by double inversion through the phospho-enzyme inter- mediate.s In summary (a) Retention is consistent with mechanisms which are adjacent concerted stepwise ion-pair or involve double inversion with an intermediate. (b) Inversion is consistent with mechanisms which are concerted or involve an ion-pair reacting within the encounter complex faster than rotation. (c) Racemization requires a mechanism involving an inter- mediate stable enough to rotate prior to reaction. Techniques of analysis are the same as those used for studies of structure and need not be repeated at length. Chirality due to gross change measured by means such as rotation of the plane of polarized light NMR experiments with chiral shift reagents or diastereoisomers or by chiral chromatography require some form of Phillips-Kenyon or Walden cycle to identify the stereo- chemistry of the material. Isotopes have been employed increas- ingly since the late 1960swhen isotopic chirality was introduced6 for the study of mechanism in enzyme reactions. The advanced analytical procedures now available have enabled the wide- spread application of isotopic chirality as a mechanistic tool. The technique is particularly valuable in studies of bio-organic systems because gross structural change (such as methyl to hydrogen or ethyl etc.)employed to introduce chirality can serve to render a substrate completely inactive towards an enzyme or make it bind in an orientation different from that in the natural case. Isotopic change does not suffer from this problem but the tax to be paid for this increased flexibility is the requirement for advanced and expensive analytical techniques. Recent advances in the study of phosphorus stereochemistry in reactions of [180,170,160]pho~phate esters involve 31P-NMR as the analytical method of ~hoice.~,~ This depends on the following two properties of oxygen isotopes bonded directly to phosphorus. The oxygen isotope 170broadens the 31Preso-nance which is not therefore observed in the NMR spectrum; the oxygen isotope attached to the phosphorus causes a signifi- cant shift in the 31Presonance to a field higher than with l6O. The magnitude of the shift increases with P-0 bond order and the position of the resonance therefore identifies whether the isotopic [ sO]oxygen is in a P-0 or in a P=O bond. In a typical procedure the stereochemistry of phosphorus in phenyl [I 70,160]phosphate may be determined by transferring the phosphate to S-propan- I ,2-diol with alkaline phosphatase as catalyst. The transfer occurs with retention of configuration. Cyclization of the phosphorylated diol (with inversion at the phosphorus) followed by methylation of the resulting phosphate ester gives syn and anti forms of the methyl ester. The stereoche- mica1 purity is obtained from a comparison of the peak areas in the 31P NMR spectra. The accuracy with which the peak areas can be estimated (associated with the repeatability of the spec- tra) give confidence limits to the stereochemistry of the phos- phate being analysed. Stereochemistry at carbon resulting from experiments with asymmetric isotopic labelling with deuterium may be analysed THE DIAGNOSIS OF CONCERTED ORGANIC MECHANISMS-A by optical rotation but direct NMR analysis of the intact methyl group is the method of choice Diastereotopic protons in -CH,D in a suitable chiral molecule have an observable chemical shift difference Normally the CH3-group is degraded to CH3COOH for chirality analysis Analysis of chirality at sulfur as a result of transfer of the sulfuryl group may be carried out by employing infra red spectral analysis using a Fourier transform instrument 9a The sulfuryl group is transferred to an optically active diol by heating in solution in carbon tetrachloride The sulfuryl diol is then deprotected and cyclized to yield a cyclic sulfate and the FTIR spectrum determined The shift in the S=O vibration (symmetric and antisymmetric stretching frequencies of the >SO bond at 1400 and 1200 cm-’) caused by isotopic substitution of oxygen in axial and equatorial positions forms the basis for the stereo- chemical analysis of the chirality of the sulfur 4.3 Positional Isotopic Exchange (PIX) The system under investigation can often be designed so that return of intermediate to reactants would incur an exchange of isotopes (Scheme 1) The composition of the reactant is exa- mined at increasing times for evidence of incorporation of the isotope The absence of positional isotope exchange in a system does not exclude a stepwise mechanism because return to exchanged reactant (k’-could be slower than the forward rate constant (k,)in the general scheme The presence of PIX could also result from intervention of solvent molecules to form a reactive intermediate (eg A-solvent) which then reacts with either X* or X Dissociative Mechanism ~A-X ’A ‘ * products kl E111 A-X* Associative Mechanism k2A-X * X-A-X‘ products4 kl z41 A-X‘ Scheme I General equations for positional isotope exchange in stepwise dissociative and associative reactions 5 Exclusion of Stepwise Mechanisms 5.1 Preliminaries The observation of results arising from a change in rate limiting step which are thus indicative of a stepwise pathway is a classical method excluding a concerted process The absence of evidence for a change in rate determining step is not immediate evidence for a concerted mechanism (excluhng a stepwise path) because the experiments could have been carried out under conditions where thechange in rate limiting step would not be expected If it can be established that a change in rate determining step would be expected for a putative stepwise process then the absence of such a change is diagnostic for a concerted mechanism 5.2 Primary Kinetic Isotope Effects The observation of a primary isotope effect indicates bond fission in the transition state of the rate limiting step in order to demonstrate concertedness primary isotope effects should be measured for all bonds undergoing a major bonding change Such a study is clearly not a minor undertaking and has only been achieved for a few systems notably in elimination reactions WILLIAMS of 2-arylethyl derivatives Heavy atom isotope effects as well as the hydrogen isotope effect must be studied and it is essential to know the expected primary isotope effects 96 Even if a primary isotope effect is demonstrated for the two bonding changes it is still necessary to show that these are not due to a ‘balanced’ stepwise process (I e where neither formation nor decompo- sition of the intermediate is rate limiting) 5.3 Double Isotope Fractionation Testlo One way of excluding a ‘balanced’ stepwise process 1s to study the effect of substitution of an isotope in one of the bonds on the isotope effect in the other bond undergoing a major change The influence of k on the overall rate constant (Scheme 2) would be altered by variation in the partition ratio (R = k-,/k2)and if k bears a second isotopic change (z e it is an isotopically sensitive step) the isotope effect on the overall rate constant should also be affected by the change in ‘R’ No change in isotope effects would be observed as a function of the change in isotope in the other bond if both the isotopically sensitive bonds are undergoing concerted change The principle is illustrated for a simple reaction when carbanion formation precedes bond formation between carbanion and electrophile in a stepwise process For this idealized case the possible observed isotope effects divide into four types (a) When the reaction is stepwise and k is rate limiting the substitution of 2H for 1H will have no effect on the I3C isotope effect (E+ is for example -CHO and the carbon is isotopically substituted) which should be unity (b) when k and k-are of the same order the substitution of 2H for 1H will reduce k-and hence the partitioning ratio will increase and the effect of k will decrease leading to a reduced 3Cisotope effect on the C-E bond (c) a concerted process yields primary deuterium and I3C isotope effects which are independent of isotopic substitution in the other bond (d) when k is rate limiting there will only be a secondary hydrogen isotope effect (equilibrium) and a primary isotope effect for the C-E bond kbs = k~ kz[Bl[E+l/(k-~+ k2[E+1when k is rate limiting kobs= k,[B] when k is rate limiting kobs = (k,k,/k-,)/[B][E+] Scheme 2 Schematic view of free energies of activation for the two consecutive steps in a putative stepwise process the energies of the reaction states are reduced to zero The example can be applied to an associative reaction A-B + D-D-A-B-A-D + B but in practice this would require two heavy atom isotope effects which would not have the substantial effect on ‘R’ that hydrogen isotopes have 5.4 Polar Substituent Effects The observation of Hammett or Brernsted relationships involv- ing two intersecting straight lines is classic evidence for a stepwise mechanism Linearity of a free energy relationship is construed as evidence for a single rate limiting step within the range of substituent parameters measured (such as sigma or pK,) and a break in the plot would thus indicate the presence of at least two transition states Change in substituent leading to an increase in rate constant over that predicted indicates a change in rate limiting step (Figure 3) i 6 X" en 4 C I I I I I 2 4 6 8 10 PKXPY Figure 3 A non-linear Brsnsted correlation indicating an intermediate in attack of pyridines on 2,4,6-trinitrophenyl acetate the figure is plotted from data in E A Castro,F Ibanez S Lagos M Schick,and J G Santos,J Org Chem 1992,57,2691 The arrow indicates the pK of the pyridine where k- = k The inset shows three types of non-linear free-energy plot demonstrating change in rate limiting step It is difficult to predict unambiguously the position of the breakpoint in a two-step mechanism except when the entering and leaving groups have similar structures Let us assume that a Brsnsted dependence is under investigation for the putative stepwise mechanism involving nucleophilic (Nu) displacement of a leaving group (Lg) (equation la) Each step has its own Brarnsted (or Hammett etc ) equation (log k, = fin pK + C,) and the measured rate constant will be governed by equation 1b A-Lg+ Nu F' Nu-A-Lg &A-Nu + Lg (la) where k is a constant ApK = pK, -pK1 and LIP = fI2 -/3-1 /3s refer to the exponents for Brsnsted-type equations for the individual rate constants in the scheme Equation 1 predicts a free energy relationship with two straight lines intersecting (when the change in rate limiting step occurs) at ApK = 0 (z e when k-= k,) It is immediately obvious that nucleophile and leaving group are required to have similar structures otherwise the individual Brsnsted relationships for k-and k will not be the same thus destroying the validity of the general equation and making it difficult to predict from simple considerations the value of pKnUc or pK when k-= k Kinetic data can normally be measured extraordinarily accu- rately with little effort and yet Brsnsted and Hammett plots all show varying degrees of scatter about a mean line The source of the scatter (when all obvious sources of error such as steric effects etc have been excluded) is ascribed to the small difference CHEMICAL SOCIETY REVIEWS 1994 in energy induced by the substituent on the microscopic solva- tion in the standard reaction (usually an ionization) compared with the reaction under investigation The presence of such effects (called microscopic medium effects) means that the Brernsted or Hammett plots must possess sufficient data points to enable a reasonable statistical estimate to be made of slope curvature etc The points must spread over a range of substituent parameters which substantially encompasses the pK or (T corres-ponding to the breakpoint The value of A/3 corresponds to the difference in effective charge on a reacting atom in the transition states corresponding to k-and k in the putative stepwise path so that d/3 = 0 (corresponding to a linear plot) indicates that the transition states have the same electronic charge (and by infer- ence the same structure) and that there is no intermediate intervening between them Experimental techniques can never provide exact data and the fit gives an error on LIP which is the extent to which we can be sure of the effective charge difference between the putative transition states In the examples to be shown where A/3 has a zero value a typical error is f0 2 Taken as a percentage of the overall change in effective charge on the atom from reactant to product (Beq) the error in d/3 gives an estimate of the maximal barrier height from putative inter- mediate to either forward or reverse transition states Thus an error of f0 2 with a typical Peq of 1 8 (for carbonyl transfer between oxyanions) gives an overall uncertainty of 11'/o over the total change in effective charge Since the intermediate must lie somewhere between the two transition states on the effective charge scale the change in effective charge from intermediate to one of the transition states must be less than 11%/2 of the overall change in effective charge It is considered unlikely that such a change would give rise to sufficient barrier to support a discrete intermediate Thus even a slightly curved free energy relation- ship is consistent with concertedness it is clearly of great interest to examine what ratio of A/3//3 might be expected to be at the borderline between concerted and stepwise mechanisms or indeed if there is a fixed ratio 5.5 Kinetics Free energy relationships may be employed to calculate rate constants for putative reactions comparison with the maxi- mal rate constant (limited by the period for a bond vibration) may then be employed to decide if the reaction has a con- certed mechanism within the range of substituent variation The decomposition of the carbanions of the aryl esters of phenylmethanesulfonic acid follows the Brmsted law log k = -2 OpK + 24 and extrapolation to the pK (Figure 4) of 2,4-dinitrophenol indicates that the rate constant should be 1OI6 s-l which exceeds that for a reaction running at the vibration limit Thus for the 2,4-dinitrophenyl ester it is reason- able to assume that attack of hydroxide ion yields a carbanion which has no significant lifetime and that the mechanism is probably enforced concerted in the region where the 'rate constants' are greater than 1013 s 6 Systems under Contemporary Scrutiny 6.1 Displacement Reactions 6 1 1 Phosphorus Acyl (Phosphyl) Group Transfer The most important biological phosphyl groups are the phos- phoryl group 0,P- the phosphodiester group RO-PO,- and the neutral phosphoryl group (R-O),PO-Considerable mechanistic work has centred around the transfer of these and related groups which possess mechanisms conforming to those displayed in Figure 1 The metaphosphate ion (PO ,analogous to a carbonyl acylium ion) formed by the dissociative pathway was for long postulated as an intermediate in phosphorylation reactions but good evidence was not available although analo- gous metaphosphate species (1)-(4) are intermediates in phos- phyl transfer The case for the metaphosphate intermediate is difficult to make because in aqueous solution the putative intermediate appears to be very reactive indeed THE DIAGNOSIS OF CONCERTED ORGANIC MECHANISMS-A WILLIAMS __ -'+-Vibration limit \'. I I I 6 8 10 PKA ~H Figure 4 Brnrnsted dependence of the decompositon of PhCH -SO,-0-Ar indicating a change to enforced concerted elimi- nation at pK,'s for leaving groups lower than 5 the figure is plotted from ddtd in M B Ddvy K TDouglas J S Loran A Steltner and A Williams J Am Chem SOC 1977,99 1196 The stereochemistry of transfer of the PO; group from Ar-O- PO;-to an alcohol involves an inversion of configuration within the limits of the analytical data (about f 2 5%) Inver-sion could arise from a dissociative path provided the inter- mediate reacted with the nucleophile in the cage in a 'pre- association' mechanism (equation 2) Reaction of substituted pyridines with pyridine-N-phosphonateslagives N-phosphoryl- substituted pyridines The rate constants obey a Brarnsted dependence with no evidence for a break at the pKa of the attacking pyridine corresponding to that of the leaving group indicating that there is no change in rate limiting step required by a stepwise process The linearity excludes the pre-association mechanism where the metaphosphate ion would be formed in the solvent cage and react with nucleophile within the cage before the ion could escape (equation 2) the data also exclude a mechanism involving formation of a discrete metaphosphate ion intermediate [lsq xpy-PO s;$4xpy-Po When fission of the N-P bond (step 2) is rate limiting (at pK, > pK,,,) the Brarnsted slope should approximate to zero because fission of the N-P bond should be independent of the substituent in the attacking pyridine The observation of positional isotope exchange is in favour of a PO; intermediate in the reaction of ADP in acetonitrile or acetonitrile-t-butyl alcohol solvents This is due to the non- intervention of solvent which is sufficiently weak as a nucleo- phile to allow the existence of the PO; intermediate The linearity of the Brernsted plot for the attack of aryloxy-anions on the 4-nitrophenyl diphenylphosphatel is consistent with a concerted mechanism shown in structure (5) 6 I 2 Sulfur Acyl (Sulfuryl) Group Transfer Sulfuryl group (-SO;) transfer is closely analogous to the transfer of the phosphoryl group and for many years SO (analogous to PO;) was postulated as a very reactive inter- mediate The observation of a linear Brsnsted dependence (Figure 5) for attack of substituted pyridines on isoquino1ine-N- sulfonate (isq-SO,) excludes SO as an intermediate in this reaction 0 21 X" 0 -I -1 -4 I I I I I 2 4 6 8 10 PKXPY Figure 5 Brnrnsted dependence of the reaction of substituted pyridines (xpy) with isoquinoline-N-sulfonate (isq-SO,) indicating a single transition step The figure is drawn from data in N Bourne A Hopkins and A Williams J Am Chem Soc ,1985,107,4327 The position of the breakpoint expected for a change in rate limiting step is indicated by the arrow Retention of configuration at sulfur in the product of hyd- roxyl attack on phenyl sulfatels is consistent with either a concerted mechanism (equation 3) or a mechanism involving formation of sulfur tnoxide in a cage which reacts with nucleo- phile faster than rotation can occur Analogues of sulfur trioxide have been demonstrated as intermediates [(6)-(S)] and some have even been isolated Ph-0-SO ROH Ph-O-;,$;-?-R-I ROS03-+ PhOH (3) It should be noted that although 'sulfur trioxide' may be purchased it is not as a monomer it is so reactive that in the condensed phase it exists only in a polymeric form Stabilized sulfur trioxide liquid has an amount of stabilizer (usually dioxane) present which bonds with the sulfur The monomer only exists in the dilute gas phase Displacement reactions of oxyanions with aryl sulfonate esters have been shownlU to involve a concerted mechanism 6 I 3 Carbonyl Group Transfer Mechanisms available for transfer of the carbonyl group (RCO-) bear a strong relationship with other group transfers such as alkyl substitution (Figure 1 where 'A' is the RCO-group)In direct opposition to the general opinion of the previous 20 or 30 years that carbonyl group transfer reactions always involve the intervention of tetrahedral intermediates the most dramatic conclusion of the recent results is that some acyl group transfer reactions can involve a concerted mechanism Incorporation of l80into the carbonyl oxygen of the ester during alkaline hydrolysis in 80-enriched water1 indicates the existence of a tetrahedral intermediate (Scheme 3) and that the proton transfer step is faster than decomposition of the tetrahed- ral intermediate The detection of 'tetrahedral' intermediates is difficult in general because of their great propensity to decom- pose to product or reactant and many laboratories have searched for stable examples Observable but reactive tetrahed-ral intermediates such as (9) have been thoroughly investigated ,?-R-C +6H c R-CcOH -R-COO-+ EtOH 0Et OEt It 4H ,yR-C c R-CcO-R-COO-+ EtOH 0El 0El Scheme 3 Positional isotope exchange for ester hydrolysis The acylium ion (RCO+) has long been accepted as an intermediate in the gas phase although it is likely to be very reactive in nucleophilic solvents The expression of the acylium ion in solution reactions depends on the provision of factors stabilizing it and on good leaving groups Some examples of stabilized acylium ions occurring as intermediates in various dissociative mechanisms are illustrated in structures (lo)-( 12) The existence of both extremes of timing in acyl group transfer (Figure 1) begs the question of the existence of a concerted mechanism on the border between the extremes The concerted mechanism for ester hydrolysis was formally expressed by Dewar in 1949l"but it was overshadowed by Myron Bender's seminal paper on the tetrahedral intermediate l6 There was no compelling evidence against a stepwise process until it was reported'" that the reaction of substituted pyridines (xpy) with the N-methoxycarbonylisoquinolinium ion (isq-COOMe) to CHEMICAL SOCIETY REVIEWS 1994 yield N-methoxycarbonylpyridiniumions had a linear Brmsted correlation over a range of pK values greater than and less than that of isoquinoline A stepwise process would require a change in rate limiting step to occur at the pK of isoquinoline and give rise to a breakpoint at that place Similar evidence indicates that transfer of the acetyl group between phenolate ions is concerted (Figure 6) lU I I I 5 7 9 11 PKAW Figure 6 Brernsted dependence of the reaction of substituted phenolate anions with 4-nitrophenyl acetate The figure is drawn from data in S A Ba-Saif,A K Luthra and A Williams,J Am Chem Soc 1987 109 6362 The arrow shows the breakpoint for the putative stepwise process The value of LIB of zero (see equation 1) indicates that there is no charge difference between the nitrogen on the substituted pyridine for both transition states of the putative stepwise process Reaction of substituted phenolate ions on 4-nitro- phenylacetate has likewise been shown to involve a concerted pathway la Thermodynamic arguments indicate that the adduct from methyl acetate and hydroxide ion decomposes with a half life of about 10-seconds The adduct between phenolate ions has a predicted half-life too short for the species to exist and the reaction thus becomes concerted Oxyanions with pK greater than about 11 were shown to give tetrahedral adducts which had sufficient stability to exist The transition state of the concerted mechanism probably has square planar stereochemistry (13) for weakly basic nucleophiles and leaving groups the transition state would possess a tetrahedral shape (14) for more basic nucleophiles and leaving groups * THE DIAGNOSIS OF CONCERTED ORGANIC MECHANISMS-A 6.2 Substitution at Aromatic and Olefinic Carbon Concerted mechanisms for nucleophilic vinyl and aromatic substitution have been considered from time to time Various spatial models have been considered (15) to (18) and in some cases the concerted mechanism has been discarded' because concerted displacements were thought to require 'in-line' bond formation and fission I * I-Such restrictions disappear when it is considered that the stereochemistry of the concerted acyl group transfer need not have square planar geometry (except when the entering and leaving bonds are weak) Observations of stereochemical reten- tion were believed to indicate concertedness but these results are now thought to be due to a barrier to rotation of the central bond in a carbanion intermediate significant compared with that for expulsion of the leaving group (equation 4) 2o The observation of generally stable Meisenheimer addition complexes has dominated the field of mechanism in nucleophilic aromatic substitution and requires that they must be positively excluded if a concerted mechanism is to be believed Many nucleophilic substitutions at the aromatic centre involve groups which activate the reaction by withdrawing electrons in particu- lar the nitro-group is a favourite activating species The nitro- group can stabilize Meisenheimer adducts by strongly localizing the negative charge on its oxygens so that it would probably be futile to look for concertedness in substitution in nitro-activated aromatic species The nucleophilic attack of phenolate ions on 4-nitrophenoxy- triazine ethers has been shown to have a single transition state (19) The second-order rate constant for the displacement has a linear Brmsted relationship over a large range of pKd values on either side of that (pKd of the leaving 4-nitrophenol) where a change in rate limiting step is predicted for the stepwise process 21 6.3 Substitution at Saturated Carbon Undoubtedly the most important reaction as an architype/ paradigm for mechanism is the nucleophilic aliphatic substitu- tion reaction The work of Hughes and Ingold4 culminated in WILLIAMS the description of the classic transition state It was discovered during the investigations of the reaction mechanism of nucleo- philic substitution that a carbenium ion mechanism could also occur Carbenium ion intermediates which are also so reactive that they cannot exist outside the reaction complex have been demonstrated in substitution reactions The inversion of confi- guration at the central carbon atom can sometimes amount to loo% simply because the carbenium ion intermediate reacts within the complex before it can rotate Although the relative timing involving addition of nucleophile prior to departure of leaving group has not been demonstrated (corresponding to the top left corner of the reaction map in Figure l) there seems no reason why a pentacoordinate intermediate (20) should not exist under favourable conditions Species such as (21)22 and other less obvious analogues (22) and (23) are well known Br-br -Br 1-7-1 Positional isotope exchange has also been employed extensi- vely in studies of ion-pair intermediates in nucleophilic aliphatic substitution and in electron-deficient rearrangements The etha- nolysis of norbornyl brosylate can be followed readily by 170-NMR23 and suffers positional isotopic exchange Although the SN2 mechanism was thought at one time to be firmly established new ideas concerning reactivity of caged intermediates4 cast substantial doubt on many of the conclusions based on stereo- chemistry It would now appear that the aliphatic substitution mechanism is presumed to involve reactive carbenium ions unless proved otherwise The concerted aliphatic mechanism should still be considered as possible especially as there are analogues of the pentacoordinate intermediate which suggest a continuum of mechanism from SN~through sN2 to stepwise associative Recent work has shown that concerted mechanisms hold in substitution at other saturated centres such as at silicon,24u and more positive evidence is appearing for the concerted process in aliphatic substitution For example some carbenium ions may be estimated to have lifetime less than the vibration limit in the presence of azide ions 246 6.4 Elimination Reactions Historically the first well-documented evidence for a concerted mechanism appears to be for base-catalysed elimination reac- tions the transition state of the reaction is familiar to all The evidence for this has been discussed in depth and we refer readers to Bordwell's critical discussion An interesting modern ex- ample is the mechanism of the aspartate deaminase reaction which has been studied by the double isotope fractionation method The equality of the I5Nisotope effect for both 'H and 2H substituents (1 0246 f0 0013) indicates that the reaction of (2S,3S)-3-methylaspartic acid has a concerted mechanism (equation 5) 26 6.5 Cyclical Reactions The evidence for the concertedness of certain rearrangements which form the basis of the data correlated by the famous Woodward-Hoffman rules is summarized by Rhoads. 27 Evi-dence unambiguously excluding stepwise mechanisms for these reactions is rare; the major stepwise competitor for these reac- tions involves radicals. Probably the most useful tool in studies of concertedness in these reactions is the isotope effect. A modern example centres on the Claisen-type rearrangement (equation 6) where the primary isotope effects have been mea- sured for 14C-2 14C-4 14C-6 and l8O,as well as for secondary deuterium isotope effects (where the C-H bonds are not broken).28 Comparison with a matrix of calculated effects for various extents of bonding indicates the best fit of the observed isotope effects to be C-4-C-6 at 10-30% formed and C-4-0 to be 50-70% broken in the rate limiting step (Figure 7). 1 5 Figure 7 Reaction map for the Claisen-type rearrangement of ally1 vinyl ether. The shaded portion indicates the area of uncertainty for the transition ‘structure’ derived from multiple kinetic isotope effects. The top left and bottom right corners represent dissociative and associat- ive stepwise paths respectively. CHEMICAL SOCIETY REVIEWS 1994 While there is excellent evidence that there are no free radical or diradical intermediates future work would be to exclude the intervention of radical pairs within an encounter complex in a balanced stepwise process. 7 Envoi It is generally assumed that due to their lack of inertia solvation changes and changes in other forms of weak bonding will be coupled with major bonding changes. It is possible to conceive of a solvation or hydrogen bond change which cannot keep up with a very fast major bonding change occurring within an encounter complex. Indeed there are a number of experimental observa- tions which are consistent with bonding changes being uncoupled from solvation change^,^,^^ and this region of study will surely come under scrutiny as the effect of solvents on reactions becomes better understood. 8 References 1 (a) A. Williams Ace. Chem. Res. 1989 22 387. (6) J. P. Lowe J. Chem. Educ. 1974,51,785. 2 F. G. Bordwell Ace. Chem. Res. 1970,3 281. 3 W. P. Jencks Chem. SOC.Rev. 1981,10,354. 4 D. J. McLennan Ace. Chem. Res. 1978,9,281. 5 G. Lowe Ace. Chem. Res. 1983,16,244. 6 J. W. Cornforth Quart. Rev. Chem. Soc. 1969,23 125. 7 P. A. Frey Tetrahedron 1982,38 1541. 8 F. A. L. Anet and M. Kopelewich J. Am. Chem. Soc. 1989 111 3429. 9 (a)G. Lowe Phil. Trans. R. SOC.(London),1991,332(B) 141. (6) W. H. Saunders Ace. Chem. Res. 1976,8 19. 10 J. G. Belasco W. J. Albery and J. R. Knowles J. Am. Chem. SOC. 1983,105,2475. 11 A. Williams and K. T. Douglas Chem. Rev. 1975,75,627. 12 S. L. Buchwald J. M. Friedman and J. R. Knowles J. Am. Chem. Soc. 1984 106,491 1. 13 P. M. Cullis and D. Nicholls J. Chem. SOC.,Chem. Commun. 1987 783. 14 S. A. Ba-Saif M. A. Waring and A. Williams J. Am. Chem. Soc. 1990,112,8115. 15 C. L. L. Chai W. A. Loughlin and G. Lowe Biochem.J. 1992,287 805. 16 M. L. Bender Chem. Rev. 1960,60,53. 17 (a) B. Capon A. K. Ghosh and D. Mc. L. A. Grieve Ace. Chem. Res. 1981 14 306. (6) R. A. McClelland and L. J. Santry Ace. Chem. Res. 1983 16 394. 18 J. P. Guthrie J. Am. Chem. SOC.,1991,113 3941. 19 J. F. Bunnett and R. E. Zahler Chem. Rev. 1951,51,273. 20 Z. Rappoport Acc. Chem. Res. 1992,25,474. 21 A. H. M. Renfrew J. A. Taylor J. M. J. Whitmore and A. Williams J. Chem. SOC.,Perkin Trans. 2 1993 1703. 22 J. C. Martin Science 1983 221 509. 23 S. Chang and W. J. LeNoble J. Am. Chem. SOC.,1983,105,3708. 24 (a)Y. Xu and P. E. Dietze J. Am. Chem. SOC.,1993,115,10722.(6) J. P. Richard and W. P. Jencks J. Am. Chem. SOC.,1984,106 1383. 25 F. G. Bordwell Ace. Chem. Res. 1972,5 374. 26 N. P Botting A. A. Jackson and D. Gani J. Chem. Soc. Chem. Commun. 1989 1583. 27 S. J. Rhoads in ‘Molecular Rearrangements’ Vol. 1 ed. P. B. DeMayo Interscience New York 1963 p.655. 28 L. Kupczyk-Subotkowska W. H. Saunders H. J. Shine and W. Subotkowska J. Am. Chem. SOC.,1993,115,5957. 29 T. E. Casamassina and W. P. Huskey J.Am. Chem. SOC.,1993,115 14.

ISSN:0306-0012    

DOI:10.1039/CS9942300093

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

MELDOLA LECTURE. The Role of Aromatic Interactions in Molecular Recog n ition Christopher A. Hunter Department of Chemistry, University of Sheffield, S3 7HF, U.K. -Offset =1 Introduction 3.5 A Interactions between aromatic molecules represent an import- ant class of intermolecular force in chemistry, biology, and materials science. They control a range of molecular recognition and self-assembly phenomena such as: (a) the packing of aromatic molecules in the crystalline state and hence the materials properties of these compounds;' (b) the base-stacking interactions which determine the sequence-dependent structure and properties of DNA as well as recognition of DNA by drugs and regulatory proteins;* (c) the three-dimensional structures of protein^;^ (d) molecular recognition of drugs by biological receptors or enzymes and of guests by synthetic host^;^^^ (e) template-directed synthesk6 I will use the term T-T interaction to describe non-covalent interactions between delocalized v-systems, including interac- tions between aromatic molecule^.^ 2 The Nature of w Interactions Working with Jeremy Sanders in Cambridge, I began to study the properties of T-I~interactions in covalently linked cofacial porphyrin dimers.These systems were originally designed as potential artificial enzymes, but strong TIT interactions between the two porphyrins forced the cavity to collapse and inhibited substrate binding. The geometry of the stacking interaction was determined from the ring current induced changes in the chemi- cal shifts of the porphyrin 'H NMR signals (Figure 1).8 This geometry was unaltered when the porphyrins were metallated with zinc, but we were able to use these metallated dimers to determine the magnitude of the porphyrin-porphyrin 7~-inter-action. Bifunctional ligands such as DABCO can force the porphyrins apart by coordinating to both metal centres simulta- neously.By comparing the thermodynamics of this binding interaction with model compounds, we determined the por- phyrin-porphyrin interaction to be 48 f10 kJ mol- in chloro- form (Figure 2).9This approach was subsequently extended to a number of other T-I~interactions in related porphyrin dimers. Armed with this experimental data, we set about trying to understand the nature of the T-I~interaction.Molecular mecha- nics calculations using a number of force-fields consistently predicted that the low energy conformation for two stacked Chris Hunter was born in 1965. He obtained a B.A. in chem- istry in 1986 from Churchill College, Cambridge, where he stayed to research with Dr. J. K. M. Sanders. He received a Ph.D. in 1989 and was then appointed to a lectureship in the University of Otago, Dune- din, New Zealand. In 1991, he moved to his current position in the Chemistry Department at the University of Shefield. His research interests are in mole- cular recognition and supramo- lecular chemistry. Figure 1 The geometry of the porphyrin-porphyrin stacking interaction observed in cofacially linked porphyrin dimers.porphyrins is a perfectly aligned arrangement with maximum v-overlap rather than the experimental geometry in Figure 1. Clearly, there was something going on which was not accounted for by the molecular mechanics force-fields. If we think about any non-covalent interaction between two molecules, it involves the interplay of several different effects which can be divided into five categories: (a) Van der Waals (VDW) interactions which are the sum of the dispersion and repulsion energies. These define the size and shape specificity of non-covalent interactions. (b) Electrostatic interactions between the static molecular charge distributions. These are particularly important in conferring specificity on molecular recognition events.(c) The induction energy which is the interaction between the static molecular charge distribution of one molecule and the proximity-induced change in the charge distribution of the other. (d) Charge transfer which is a stabilization due to mixing of the ground state (AB) with an excited charge-separated state (A+B-). (e) Desolvation. Two molecules which form a complex in solution must be desolvated before complexation can occur. The solvent may compete for recognition sites thereby desta- bilizing the complex. Alternatively in polar solvents, solvo- phobic effects can stabilize the complex. There is clear experimental evidence that VDW interactions, electrostatic interactions, and desolvation play an important role in molecular recognition.However, there is not yet any experimental evidence that induction effects are as important. Charge transfer (CT) and electron donor-acceptor (EDA) effects are commonly invoked to explain interactions between aromatic molecules, but there is no evidence of CT in our porphyrin systems. Indeed, there is an increasing body of evidence which shows that CT and EDA effects are negligible compared with electrostatics. lo In addition, ab initio calcula-tions suggest that induction and CT energies are much smaller than electrostatic and VDW energies in non-covalent interac- tions between aromatic molecules. If we are to understand TIT interactions, we require the simplest possible model which is sufficiently complex to account for the experimental obser- vations.We have therefore decided to ignore induction and charge-transfer effects in our analysis. Let us therefore consider the three remaining effects listed above, VDW interactions, electrostatics, and desolvation. Aro- 101 CHEMICAL SOCIETY REVIEWS, 1994 Kmonomer 1 Figure 2 (a) A bifunctional ligand prises open a covalently-linked cofacial metalloporphyrin dimer (b) The control experiment used to determine the intrinsic binding energy associated with the zinc-ligand interaction By companng AHmOnOmerwith AH,,,,,,, (and allowing for any cooperativity in binding both ends of the Iigand), we can calculate the magnitude of the n-interaction which is lost on opening the cavity in the porphyrin dimer, n--n interaction = 2AHmonomer -AHdimer matic molecules are planar so that VDW interactions are maximized when the molecules adopt a perfectly stacked arrangement The flat n-electron surfaces of the molecules are non-polar so that solvophobic effects also favour stacking In addition, the porphyrin experiments descnbed above were carried out in chloroform where solvophobic effects are small Our experiments show that porphyrins prefer an offset or U 8 staggered arrangement This implies that the electrostatics of the interaction must provide a large driving force which pushes the n-systems away from optimal n-overlap These electrostatic interactions are clearly not present in molecular mechanics simulations, which generally use atomic monopole charge distri- butions, and suggests that a more detailed description of the charge distribution around n-systems is required to calculate the electrostatic interactions accurately We decided to use a charge distribution which explicitly represents the out-of-plane n-electron density which is charac- teristic of n-systems In our model, a positively charged U-framework is sandwiched between two regions of negatively charged n-electron density (Figure 3) This model accounts well for the observed geometry and magnitude of the porphyrin- porphyrin stacking interaction (Figure 4) Encouraged by this success, we used this model to derive some general principles governing the properties of M interactions We examined interactions between simple three-centre charge distributions of the type shown in Figure 3 which we termed idealized n-atoms Figure 5 illustrates how the electrostatic interaction between two x-atoms varies as a function of orien- tation The face-to-face stacked orientation (the origin in Figure 5)maximizes n-electron repulsion For edge-to-face and offset stacked orientations, attractive interactions between the positively charged a-frameworks and the negatively charged 7~-Figure 3 An sp2-hybridized atom in a 7-system Figure 4 A contour plot showing the total interaction energy (in kJ mol -l) for two stacked porphyrins as a function of translations in the xy plane (0 to 8 A) The contour spacing is 5 kJ mo1-I Vertical separation of the porphyrins is 3 4A and there is no rotation of the porphyrins relative to one another Energy minima are labelled (Reproduced with permission from the Journal of the American Chemi- cal Society ) Offset0 0-0-0 0 0 0 -@--@-0 0 Figure 5 Electrostatic interaction between two idealized 7-atoms as a function of orientation two attractive geometries and the repulsive face-to-face geometry are illustrated (Reproduced with permission from the Journal of the American Chemr- cal Society ) MELDOLA LECTURE.THE ROLE OF AROMATIC INTERACTIONS IN MOLECULAR RECOGNITION-C. A. HUNTER electrons dominate. Thus =-stacking is associated with large repulsive electrostatic interactions and is therefore much less favourable than one might expect on the basis of VDW interac- tions and solvophobic effects.The attractive edge-to-face and offset stacked interactions give rise to the characteristic herr- ingbone packing of aromatic hydrocarbons in the crystalline state.' The ideas summarized in Figure 5 hold for non-polarized T-systems, i.e. hydrocarbons which contain no heteroatoms. Polarization of the =-system by heteroatoms alters this picture and will be discussed later. 3 FP Interactions in Proteins Experimental evidence for the picture presented in Figure 5 comes from an analysis of the orientations of aromatic interac- tions in proteins. Singh and Thornton have compiled a database of the geometry of all side-chain-side-chain interactions in high resolution X-ray crystal structures of proteins.14 We will con- centrate on the interactions observed between the aromatic rings of phenylalanine (Phe) side-chains.Figure 6(a) shows how the calculated electrostatic interaction between two benzene mole- cules varies as a function of orientation. This can be compared with the experimental distribution of Phe-Phe interaction geo- metries observed in the Singh-Thornton database [Figure 6(b)]. The experimental scatter maps out the region of conformational space which corresponds to an attractive electrostatic interac- tion. Orientations which would result in repulsive electrostatic interactions, such as face-to-face T-stacking, are not observed. A strong clustering of experimental observations, which would indicate a deep well-defined energy minimum, is not (a) Angle (degrees)* I 90 0 Offset (A) (b) Angle (degrees) 180 1 I,1 .Offset0 (A) Figure 6 (a) Electrostatic interaction between two benzene rings as a function of orientation. The geometries of interaction are illustrated. The shaded area represents attractive interactions and the unshaded area represents repulsive interactions. (b) A scatter plot of the experimentally observed geometries of interaction between phenyla- lanine aromatic rings in proteins. (Reproduced with permision from the Journal of Molecular Biology.) observed. This is consistent with our calculations which indicate that the maximum electrostatic interaction is 6 kJ mol- l.This is a rather weak interaction. However, aromatic side-chains often occur in clusters in protein^.^ These clusters contain several M interactions and so make a significant contribution to the stability of that particular structural motif. 4 m-Facial Hydrogen Bonds If we consider T-interactions in the wider context of electro- static interactions between neutral molecules, which play an important role in molecular recognition, we have a heirarchy of electrostatic interactions with hydrogen bonds at the top and interactions between benzene rings at the bottom (this will hold in a vacuum, but solvent can influence this ordering). Interme- diate between these two extremes, there are C-H**.O hydrogen bonds (or C-H X hydrogen bonds, where X is an electronega- tive heteroatom) which have been extensively investigated' and r-facial hydrogen bonds which have surprisingly not received much attention (Figure 7)." Strong v-y 3 Weak Figure 7 Heirarchy of electrostatic interaction between neutral molecules.(Reproduced with permission from the Journal of the Chemical Society, Chem ica 1 Comm un icat ions. ) Crystal structure databases such as that compiled by Singh and Thornton represent an excellent source of experimental data on the preferred geometries of non-covalent interactions. We therefore decided to search for =-facial hydrogen bonds in proteins using this database. The interaction which we chose was the arginine-phenylalanine side-chain interaction.Calculations using the electrostatic model discussed above predicted that the orientation with the edge of the guanidinium group hydrogen- bonded to the face of the phenylalanine =-system should be the most stable orientation, with a very well-defined energy mini- mum. However, the experimental data show that this geometry is extremely rare whereas almost every other orientation is highly populated. The reason for this discrepancy lies in the inhomogeneity of the protein environment. Our calculations were performed in a vacuum, but proteins contain many polar groups which can compete for the hydrogen-bonding sites on the arginine side-chain. A glance at Figure 7 shows that given the choice, a guanidinium proton will hydrogen bond to a proper hydrogen-bond acceptor [Figure 7(a)] rather than the face of an- system [Figure 7(c)].Proteins which contain an excess of hydro- gen-bond acceptors, in particular amides, are therefore unlikely to show any =-facial hydrogen-bonding. CHEMICAL SOCIETY REVIEWS, 1994 Figure8 Intermolecular interactions observed in the crystal structure of (1). Hydrogen bonds are shown as dotted lines. (Reproduced with permission from the Journal of the Chemical Society, --.:-:-.. ..Chemical Communications. ) H-Bond.,,,*.H-Bond H-Bond"' : ..-.,s-=.-'..-' n-5 h-n n-n (2) + pbenzoquinone Figure 9 (a) Interaction sites available for molecular recognition of p-benzoquinone. (b) Complexation of p-benzoquinone by a synthetic macrocyclic host.(Reproduced with permission from the Journal of the Chemical Society, Chemical Communications.) product was completely unexpected, especially as the reaction was carried out under high dilution conditions (lop4 to M). The three-dimensional structure of the catenane was deter- mined using 2-D NMR and provides a clue as to why this self- templating process is so efficient.*' There are a number of hydrogen bonds between the two interlocked macrocycles [Figure 1](a)] as well as several 7~-interactions [Figure 1l(b)]. The geometry of the T-?T interactions is consistent with the model presented above. Each i-phthaloyl group bound inside the central cavity of the other macrocycle makes two offset stacked contacts and an edge-to-face contact with the aromatic side-walls of the cavity.7 A Self-organizing Macrocycle Neither (3)nor (4)complexp-benzoquinone. The cyclic tetramer adopts an open conformation [Figure 12(a)] and so does not have a cavity with dimensions suitable for p-benzoquinone recognition, and the catenane of course does not possess a cavity. When Duncan Purvis replaced i-phthaloyl dichloride by 2,6-pyridine dicarbonyl dichloride in the final cyclization step, he produced a similar set of macrocyclic products, but we were We have studied a molecular assembly which does not contain any obvious hydrogen-bond acceptors, the crystal structure of an aniline derivative, (1). This system shows n-facial hydrogen- bonding involving an amine group directed into the centre of the face of an aromatic ring (Figure 8).5 Design of a Synthetic Receptor In order to test these ideas further, I decided to design a synthetic host which would complex a guest using the edge-to-face 72-interactions which are predicted to be attractive.'* I chose p-benzoquinone as the guest since I was interested in mimicking some of the properties of bacterial photosynthetic reaction centres where quinone cofactors play an essential role in the primary photoinduced charge-separation process. 9~20 p-Benzoquinone has a range of possible recognition sites which could be exploited in host design [Figure 9(a)]. In order to select the optimum 7~--7~interactions, we must consider the effects of the polarizing heteroatoms.The quinone carbonyl oxygens withdraw n-electron density from the ring. This will weaken an edge-to-face interaction with the face of the quinone n-system. In contrast, the polarization will enhance an interaction between the edge of the quinone and the face of another n-system. The host (2) which uses these edge-to-face interactions and maxi- mizes the hydrogen-bonding interactions is illustrated in Figure 9(b).Macrocycle (2) has been synthesized (see below), and it complexes p-benzoquinone with an association constant (K,) of lo3 M -in chloroform. Complexation-induced changes in the lH NMR chemical shifts of both the host and the guest indicate that the structure of the complex is as shown in Figure 9(b). These observations do not prove that there is an attractive TIT interaction in this system, but they are consistent with such an interaction.Moreover, (2) is selective for p-benzoquinone: sub- stitution of methyl groups, chlorines, fluorines, or fused aroma- tic rings for the quinone hydrogens completely inhibits complex- ation (K, < 5 M-l). These changes would block the edge-to- face minteractions. 6 Template-directed Catenane Synthesis The synthetic route to (2) produced the macrocycle, which will be referred to as the cyclic dimer, in 51% yield (Figure 10). However, there were two major side-products, a cyclic tetramer, (3), (10%) and a [2]-catenane, (4), (34%).'l The catenated MELDOLA LECTURE. THE ROLE OF AROMATIC INTERACTIONS IN MOLECULAR RECOGNITION-C. A.HUNTER o&o o&o CI CI \ NH HN I X=CH,N 1 Cyclic Dimer Cyclic TetramerX = CH (2) X = CH (3)X=N (5) X = N (6) [2]-Catenane (4) Figure 10 Synthetic route to amide macrocycles. The final macrocycliza- tion step produces a cyclic dimer, a cyclic tetramer, and a [2]-catenane. (Reproduced with permission from the Journal of the Americun Chemi- cal Society.) surprised to find that both the cyclic dimer (5) and cyclic tetramer (6) formed complexes with p-benzoquinone.22 The NMR structure of the tetramer revealed the reason for this. Macrocycle (6) adopts a folded conformation in solution [Figure 12(bj] and so has two cavities complementary to p-benzoquinone. A titration showed that (6) binds two quinones with weak cooperativity [Figure 12(bj].The interactions involved in recognition are again hydrogen bonds and edge-to- face X-T interactions. Macrocycle (6) has a covalent structure which differs only very subtly from that of (3), yet it adopts a completely different conformation and therefore has very differ- ent properties. The origin of this difference lies in the electro- static interactions which control the conformational properties of the diamide subunits. 2,6-Pyridine diamides prefer a confor- mation with the amide NHs cis,because this minimizes lone pair repulsion and allows attractive electrostatic interactions with the pyridine nitrogen lone pair [Figure 12(b)]. In contrast, i-phthaloyl diamides prefer a conformation with the amide NHs trans, because this optimizes the electrostatic interactions between the two amide groups [Figure 12(a)].22 8 The Effect of Polarization by Heteroatoms on-IT Interactions We have touched on the role of polarizing heteroatoms in modulating n-n interactions.The systems discussed in the first part of the paper were aromatic hydrocarbons which lacked heteroatoms. In these systems, face-to-face stacking is inhibited by n-electron repulsion. However, the introduction of polarizing substituents which perturb the molecular charge distribution changes this picture. Heteroatoms cause large partial atomic charges and lead to additional electrostatic interactions. We divide the electrostatic interactions in these systems into three categories:* co OC& Figure 11 Interactions which help to interlock the two macrocycles of the [2]-catenane (4).(Top) The NMR-derived structure of the [2]- catenane which was energy minimized using the CHARMm force- field subject to NOE constraints. One macrocycle is coloured blue, the other yellow. Hydrogen bonds between the two macrocycles are shown as white dotted lines. (Bottom) Inter-macrocycle T-interac-tions (two offset stacked and one edge-to-face interaction). (Reproduced with permission from the Journal of the American Chemi- cal Society.) CHEMICAL SOCIETY REVIEWS, 1994 No Interaction Figure 12 Preferred conformations and recognition properties of the tetrameric macrocycles (a) The tetra i-phthaloyl macrocycle (3) (b) The bis 2,6-pyridyl tetramer (6) (Reproduced with permission from the Angewandte Chemie Znter-national Edition in English ) (a) ~o-~T(Tinteractions These are the interactions associated with the out-of-plane n-electron density if no polarization is present, I e the kind of interaction we have already discussed (b) Atom-atom These are the interactions between the partial atomic charges (c) Atom70 This is the cross-term of the other two interac- tions, i e the interaction between the partial atomic charges on one molecule and the out-of-plane n-electron density on the other For molecules which are highly polarized such as the DNA bases, 7~0-o interactions are not very important It is usually the atom-atom term which is the largest electrostatic interaction, but the atom-0 term can also be significant Moreover, the atom-nu term is very sensitive to changes in geometry and can play an important role in determining the orientation of x-n interactions between polarized n-systems 9 Sequence-dependent DNA Structure The three-dimensional structures and properties of double helical DNA depend critically on the sequence of the aromatic bases Evidence comes from a wide variety of sources including, X-ray fibre diffraction studies of X-ray crystal Figure 13 (a) The chemical structures of the DNA base-pairs C' is the atom which connects the bases to the sugar (b) The molecular charge distribution of the base-pairs (Reproduced with permission from the Journal of Molecular Biology ) (a) H A-T pbenzoquinone structures of olig~mers,~~ gel running experiments,26 and recognition of DNA by small organic molecules and pro- teins 25 27 An examination of the chemical structure of DNA shows that the only difference between different sequences lies in the aromatic bases [Figure 13(a)] The sugar-phosphate back- bone is identical regardless of sequence It therefore follows that the cause of sequence-dependent variations in DNA structure is the interactions between the aromatic bases which are ~T-T stacked up the centre of the double helix We have applied the 7~-interaction model described above to elucidate the molecular basis for these structural variations The conceptual basis for tackling this problem was established by Calladine and Drew 28 They showed how the conformation of a single base-pair step can be related to the overall three- dimensional structure of the double helix The conformation of a base-pair step can be defined in terms of six degrees of freedom (Figure 14) 29* An analysis of DNA oligomer X-ray crystal structures shows that the parameters which are most important for defining sequence-dependent variations in structure are twist, slide, and roll 28 We have therefore analysed the 7~-n interactions for all ten possible base-pair steps as a function of these parameters The results correlate rather well with experi- ment and throw new light on many properties of nucleic acids, including the different conformational preferences of RNA and DNA, the formation of left-handed Z-DNA, and the role of TATA in originating replication Inspection of the covalent structures of the bases (Figure 13) reveals that all sequence-dependent effects must be caused by (a) steric interactions with the guanine amino group in the minor groove, * There are another six internal degrees of freedom for each base pair One of these parameters propeller twist (a rotation about the long axis of the base pair) plays an important role in defining sequence dependent changes in DNA structure but it will not be discussed here * H-NbN\0 C' H C-G b *''* H MELDOLA LECTURE.THE ROLE OF AROMATIC INTERACTIONS IN MOLECULAR RECOGNITION-C. A. HUNTER rise t4twist Figure 14 The conformation of a base-pair step in double-helical DNA can be defined by three rotations (twist, roll, and tilt) and three translations (rise, slide, and shift).The block edges which correspond to the minor groove are shaded. (b) steric interactions associated with the configuration of the step, whether the sequence of bases is purine-purine, purine-pyrimidine, or pyrimidine-purine; (c) steric interactions with the thymine methyl group in the major groove; (d) electrostatic interactions associated with the different mole-cular charge distributions across the A-T and C-G base-pairs [Figure 13(b)]. The first two effects have been discussed in detail by Calladine and Their observations are reproduced by our calculations, but in this paper we will concentrateon the last two effects.We will consider just two base-pair steps to illustrate the approach. A detailed analysis of the conformational properties of all ten base-pair steps can be found in reference 2. Two base-pair steps which have been thoroughly character-ized by experiment are AA/TT and CC/GG. These steps exem-plify the two most common conformational families which were first observed by fibre diffraction methods, the A and B poly-morphs. AA/TT has a very strong preference for the B-form (roll z 0" and slide =0 A).23,26,31CC/GG has a very strong preference for the A-form (roll = 12"and slide z -1.5 8,).23,32 Why is this? Contour plots of the magnitude of the T-T interaction between the base-pairs as a function of slide and roll are shown in Figures 15 and 17.The calculation for AA/TT shows a deep well-defined energy minimum at the origin (slide = 0 8, and roll = the region of conformational space which corres-OO), ponds to the B-form that is observed experimentally (Figure 15). Comparison of the VDW interaction energy with the total nT interaction energy shows that the conformational properties of this step are dominated by steric effects. Electrostatics are relatively unimportant because the AT molecular charge distri-bution has no large regions of high charge density. The major steric interaction which locks AA/TT into the B-form is a clash between the thymine methyl group and the 5'-Figure 16 An AXjXT step viewed along the slide/roll axis (X is any base).The direction of positive slide is towards the reader. The block edges which correspond to the minor groove are shaded. The thymine methyl group prevents positive roll. Increasing roll leads to a steric clash (asterisk) between the thymine methyl and the other base on the same strand. (Reproduced with permission from the Journal of Molecular Biology.) twist+ roll3 (a) roll -40 -40T -40 -40 slide -30 -30 -4b -5d roll T-20 -20 -20 -20 -30 -30 -slide -20 -30 -20 -30 -30 Figure 15 Contour plots showing the interaction energy for the AA/TT step (in kJ mol-l) as a function of slide and roll. The contour spacing is 2kJ mol-l. The slide axis runs from -3 8, to 2 8, and the roll axis runs from -15" to 25".Helical twist = 36", propeller twist = 15" and all other parameters are set to zero. Energy minima and confor-mations corresponding to A, B, and C-DNA are labelled. (a) VDW interaction energy; (b) total T-T interaction energy (VDW + electrostatic). (Reproduced with permission from the Journal of Molecular Biology.) neighbouring base which occurs in positive roll A-type confor-mations (Figure 16). This interaction also provides an explana-tion for the different conformational properties of double-helical DNA and RNA. RNA has uracil in place of thymine and so lacks the major groove methyl groups. Thus double-helical RNA has only been observed in the A-f~rm,~~,~~whereas the A-form of DNA is destabilized by the thymine methyl groups so that DNA is polymorphi~.~~ Electrostatic interactions play a more important role in the CC/GG step.C-G base-pairs have two regions of very high charge density, a positive charge over cytosine and a negative charge over guanine [Figure 13(b)]. The VDW interaction for CC/GG shows a broad flat energy minimum close to the origin, the B-form [Figure 17(a)].Addition of the electrostatic interac-tion moves the energy minimum to large negative slide, close to the A-type conformation, in agreement with the experimental results for this sequence. When the calculations were repeated omitting the out-of-(3') positive roll t roll -50 -50 - slide -50 -40 -40 -50 (b) -10 o io rolI io T 0 -20 -20 Figure 17 Contour plots showing the interaction energy for the CC/GG step (in kJ mol- l) as a function of slide and roll The contour spacing is 2 kJ mol- The slide axis runs from -3 8, to 2 8, and the roll axis runs from -15" to 25" Helical twist = 36", propeller twist = 15" and all other parameters are set to zero Energy minima and confor- mations corresponding to A, B and C-DNA are labelled (a) VDW interaction energy, (b) total n-n interaction energy (VDW + electrostatic) (Reproduced with permission from the Journal of Molecular Biology ) plane n-electron density and using only partial atomic charges to calculate the electrostatic interaction, a very broad flat energy minimum with no clearly defined conformational preference was obtained Clearly, it is essential to allow for out-of-plane n-electron density to model accurately the properties of ~TT--~T interactions even for these very highly polarized n-systems The important interaction which causes the negative slide A-type conformation for CC/GG is the atom-0 interaction Figure 18 shows that the A-form is associated with a movement of the T-electron density of the bottom base-pair away from the guanine negative charge and towards the cytosine positive charge Thus calculations using this model for TT-~Tinteractions repro- duce the experimental conformational preferences of different Figure 18 Atom-m interaction in a CXjXG step (X is any base) The thin lines map out the area covered by the n-electron density of the X-X base-pair The molecular charge distribution of the C-G base-pair is shown 0 indicates the position of the C', atom where the bases are attached to the sugar (Reproduced with permission from the Journal of Molecular Biology ) B-form slide negative CHEMICAL SOCIETY REVIEWS, 1994 DNA sequences and allow us to probe the molecular basis for these properties * 10 Conclusion These studies show that in order to understand or model non- covalent interactions between aromatic molecules, it is essential to consider electrostatic interactions involving the out-of-plane n-electron density The pictures presented in this paper show how the properties of TTT--~Tinteractions can be understood simply on the basis of the shapes and charge distributions of the individual molecules It is clear that T-interactions have quite strong geometrical requirements and are directional to a much greater extent than was previously thought They therefore play a very important role in controlling specificity in molecular recognition 11 References 1 G R Desiraju and A Gavezzotti, J Chem SOC Chem Commun , 1989,621 2 C A Hunter, J Mol Biol, 1993,230, 1025 3 S K Burley and G A Petsko, Science, 1985, 229, 23 4 F Diederich, Angew Chem Int Ed Engl, 1988,27, 362 5 A V Muehldorf, D Van Engen, J C Warner, and A D Hamilton, J Am Chem SOC,1988,110,6561 6 P L Anelli, P R Ashton, R Ballardini, V Balzani, M Delgado, M T Gandolfi, T T Goodnow, A E Kaifer, D Philp, M Pietraszkiewicz, L Prodi, M V Reddington, A M Z Slawin, N Spencer, J F Stoddart, C Vicent, and D J Williams, J Am Chem SOC , 1992,114, 193 7 C A Hunter and J K M Sanders, J Am Chem SOC , 1990, 112, 5525 8 P Leighton, J A Cowan,R J Abraham,andJ K M Sanders, J Org Chem , 1988,53 733 9 C A Hunter, M N Meah, and J K M Sanders, J Am Chem Soc , 1990,112, 5773 10 F COZZI,M Cinquinti, R Annuziata, and J S Siegel,J Am Chem SOC,1993, 115, 5330 11 W L Jorgensen and D L Severance, J Am Chem SOC ,1990,112, 4768 12 D B Smithrud and F Diederich, J Am Chem SOC , 1990,112,339 13 S L Price and A J Stone, J Chem Phys , 1987,86,2859 14 J Singh and J M Thornton, J Mol Biol ,1990,211, 595 15 C A Hunter, J Singh, and J M Thornton, J Mol Biol, 1991,218, 837 16 R Taylor and 0 Kennard, J Am Chem SOC , 1982,104,5063 17 L R Hanton,C A Hunter,andD H Purvis,J Chem SOC Chem Commun , 1992, 1 134 18 C A Hunter, J Chem SOC Chem Commun , 1991,749 19 H Michel, 0 Epp, and J Diesenhofer, EMBO J , 1986,5, 2445 20 J P Allen, G Feher, T 0 Yeates, H Komiya, and D C Rees, Proc Nut1 Acad Scr USA, 1988,85,8487 21 C A Hunter, J Am Chem SOC , 1992,114,5303 22 C A Hunter and D H Purvis, Angew Chem Znt Ed Engl, 1992, 31, 792 23 A G W Leslie, S Arnott, R Chandrasekaran, and R L Ratliff, J Mol Biol , 1980, 143,49 24 H R Drew, M J McCall, and C R Calladine, in 'DNA Topology and its Biological Effects', Cold Spring Harbour, 1990, p 1 25 0 Kennard and W N Hunter, Angew Chem Znt Ed Engl , 1991, 30, 1254 26 C R Calladine, H R Drew, and M J McCall, J Mol Bzol, 1988, 201, 127 27 A A Travers and A Klug, in 'DNA Topology and its Biological Effects', Cold Spring Harbour, 1990, p 57 A-fom -sg MELDOLA LECTURE THE ROLE OF AROMATIC INTERACTIONS IN MOLECULAR RECOGNITION-C A HUNTER 28 C R Calladine and H R Drew, J Mu1 Biol, 1984,178, 773 32 M McCall, T Brown, and 0 Kennard, J Mol Biol ,1985,183,385 29 S Diekmann, J Mol Bid, 1989,205,787 33 S Arnott, D W L Hukins, S D Dover, W Fuller, and A R 30 C R Calladine, J Mol Biol ,1982, 161, 343 Hodgson, J Mu1 Biol, 1973,81, 107 31 H C M Nelson, J T Finch, B F Luisi, and A Klug, Nature, 1987, 34 A C Dock-Bregon, B Chevrier, A Podjarny, D Moras, J S de 330,221 Bear, G R Gough, and P T Gilham, Nature, 1988,335,375

ISSN:0306-0012    

DOI:10.1039/CS9942300101

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Non-Bonding Molecular Orbitals and the Chemistry of Non-Classical Organic Molecules Christopher A. Ramsden Department of Chemistry Keele University Keele Staffordshire ST5 5BG U.K. 1 Introduction Valence bond structures depicting localized two-centre two- electron 0 and 7~ bonds have proved to be an extremely useful concept for discussing the structures and reactions of molecules. However there are molecules which cannot be satisfactorily represented by classical uncharged valence bond structures employing atoms in their normal valence states. These ‘non- classical’ molecules are usually represented either as resonance hybrids of several dipolar canonical forms or as hypervalent structures. Representative examples (1)-(6) of these mole- cules are shown in Figure 1. Ozone is a typical example of a molecule which is represented only by dipolar structures (la) t)(1b) the alternative hyperva- lent structure (lc) is not favoured. In contrast xenon difluoride is ‘satisfactorily’ represented by the hypervalent structure (2c). The description as a resonance hybrid of two dipolar forms (2a)++ (2b) is seldom used probably because distinction between these canonical forms and an ionized molecule (FXe+ + F-) is not clear. Examples of non-classical organic molecules include heterocyclic mesomeric betaines such as 1-arylpyridinium-3-olates (3a) -(3b) which like ozone partici- pate in 1,3-dipolar cycloaddition reactions and [hydroxy(tosyl- oxy)iodo]benzene (4c) a hypervalent iodine derivative which is a useful and versatile reagent. In spite of the segregation of these non-classical molecules into two camps the ‘1,3-dipoles’ and the ‘hypervalents’ all these molecules have fundamental similarities in their bonding and modes of reaction. The purpose of this article is to emphasize the similarities of this large and diverse group of molecules in terms of the three-centre four-electron bonding which was developed by Pimentel Hach and Rundle and Musher,1-3 and to explore applications of this molecular orbital description by reviewing some areas of current interest. 2 Three-centre Four-electron Bonds The classical description of molecules in terms of localized two- electron bonds is derived from the interaction of a pair of atomic orbitals on adjacent atoms to form two molecular orbitals of Chris Ramsden obtained his B.Sc. (1967) Ph.D. (1970) and D.Sc. (1990) degrees from Shefield University. He was a Robert A. Welch Post-doctoral Fellow (1971-3) at the University of Texas at Austin and an I.C.I. Post-doctoral Research Fellow (19734) at the University of East Anglia. After working in the pharmaceutical industry for a number of years he was appointed Professor of Or-ganic Chemistry at Keele Uni- versity in 1992. Currently his research interests include the synthesis of novel imidazoles and imidazolines of biological interest and investigations of new reactions of hypervalent reagents and their application to biological problems. DiDolar HvDervalent ReDresentations Remesentations +F-Xe-F -F-Xe+ F-I -F-Xe-F Ph Ph -OH I Ph-I+ -OH I I OTS OTS OTS (44 (4b) (44 Me\+ -Me\7-O -Me7=OMe Figure 1 Examples of non-classical molecules represented by dipolar or hypervalent structures. which one is bonding and the other is antibonding. Two electrons are fed into the bonding orbital leading to a localized bond (0or T)which can aptly be described as a two-centre two- electron [2c-2e] bond. This description of the m-bonding in ethylene is summarized in Figure 2a. The bonding is rather different when three atomic orbitals are placed in linear conjugation -a situation exemplified by the interaction of three carbon 2p atomic orbitals to give the T-molecular orbitals of the allyl anion. In this case the atomic orbitals combine to form three molecular orbitals of which one is bonding one is antibonding but the third is non-bonding (Figure 2b). Two electrons can be placed in the bonding orbital and two in the non-bonding molecular orbital (NBMO) to give an energetically favourable configuration. However the struc- ture cannot be represented by two-centre bonds. In contrast to ethylene which can be represented by a single localized bond structure the structure of the allyl anion can only be represented 111 Figure 2 The interaction of (a) two and (b) three 2pz atomic orbitals to form bonding non-bonding and anti-bonding molecular orbitals. as a resonance hybrid of two structures. The bonding is asso- ciated with all three atoms it is fundamentally different from the bonding in ethylene and can be described as a three-centre four-electron [3c-4e] bond.2 Molecules which are associated with three-centre four-electron bonds will usually have a high energy occupied molecular orbital which has the topology of a NBMO and the symmetry and energy of this orbital frequently has a profound influence on the chemistry of these molecules. This type of bonding is not limited to the conjugation of three 2pz orbitals. Any three atomic orbitals in linear conjugation and with positive overlap will be isoconjugate with the allyl anion and the bonding will be equivalent in terms of molecular orbital (MO) theory. Quantitative differences will arise as a result of variations in the efficiency of overlap and the energies of the orbitals participating but the molecular orbitals will be topolo- gically equivalent. Figure 3 shows three arrangements (8)-(lo) of atomic orbitals or their hybrids which are isoconjugate with the 7~ system of the allyl anion (7). In each case interaction of the atomic orbitals leads to molecular orbitals having similar nodal properties each system has an orbital associated with the characteristics of a NBMO which vanishes on the central atom. Four electrons can be fed into the two lowest energy orbitals (bonding and non-bonding) to give an energetically favourable configuration. It is instructive to consider molecules and ions which are isoelectronic with these isoconjugate four-electron bonding systems (7)-(lo) and to focus in particular on molecules in which the central atom formally donates two electrons and the CHEMICAL SOCIETY REVIEWS 1994 two peripheral atoms each donate one electron to the electron quartet. Examples of such molecules and ions (1 1)-( 14) are shown at the bottom of Figure 3. The ozone molecule (1 1) is clearly isoconjugate with the allyl anion and its highest occupied molecular orbital (HOMO) has the symmetry of a NBM0.4 For a similar reason xenon difluoride (12) also has a high energy occupied orbital which is related to a NBMO. The strength of the xenon-fluorine bonds is mainly derived from the bonding MO which has high electron density between the xenon atom and each fluorine at~m.~?~ In effect electron pairs on the central atoms of ozone and xenon difluoride are participating in bond- ing interactions with the adjacent atoms while the two electrons orginating on each of these peripheral atoms are effectively retained by them because they are located in a NBMO which vanishes on the central atom. Viewed in this way (Figure 3) the similarity between the ‘1,3-dipole’ ozone (1 a*) and the ‘hyper- valent’ xenon difluoride (2a-c) is easy to recognize. In molecu- lar orbital terms they are close relatives and the nature of the bonding in the two molecules is essentially the same. Whether we represent them by hypervalent or dipolar structures is purely a matter of convention. The third example in Figure 3 is the hydrogen difluoride ion (13) and this differs from the previous examples (1 1) and (12) in that the central participant is an s-orbital rather than ap-orbital. Nevertheless the orbital overlap (9) is isoconjugate with the others (7) and (S) and this bonding scheme accounts for the strong symmetrical hydrogen bonding in (1 3) and related specie^.^ The final example (1 0) differs from the others in that two of the orbitals are on the same atom.’ This provides an interesting view of the bonding in for example sulfoxides (14) (and ylides) for which both dipolar (6a) and hypervalent (6b) structures are in common usage. In these cases (10) the bonding is correctly described as a two-centre four-electron [2c-4e] bond and the NBMO is effectively a lone pair of electrons in a pl-orbital. In Figure 3 this p,-orbital is represented as the combination of a pair of degeneratep-orbitals in order to emphasize the relation- ship with the other NBMOs in the figure (these combinations should not be confused with 3d-orbitals). In this article we wish to emphasize similarities between all the types of non-classical Figure 3 The isoconjugate relationship of sets of three atomic orbitals (7)-( lo) and the resulting bonding (i) non-bonding (ii) and anti- bonding (iii) molecular orbitals together with molecular examples (1 I)-( 14). NON-BONDING MOLECULAR ORBITALS AND NON-CLASSICAL ORGANIC MOLECULES-C A RAMSDEN molecule exemplified in Figure 2 and therefore we stress here and the species (25) can collapse to a fluoride ion and the radical the relationship between the [2c-4e] bonding system (10) and the (26) which then undergoes further reaction Some reactions of [3c-4e] bonding systems (7)-(9) while recognizing that the xenon difluoride (12) can be rationalized in a similar way MOs resulting from overlap (10) can be expressed in other (XeF -+ XeF;-+XeF’ + F-)(see Section 6) forms \-3 The Stability and Characteristic Reactions of 0 Non-classical lsoconjugates According to the analysis summarized in Figure 3 three large and discrete families of isoconjugate non-classical molecules having the general formulae (15)-(17) can be recognized and for the purposes of this discussion these can be labelled as 13 dipoles (1 5),hypervalents (1 6) and ylzdes (17) A simple analysis of the electron distribution in species (15)-(17) based on approximate LCAO coefficients of the molecular orbitals given in Figure 3 shows that the central atoms ‘b’ (the electron pair donors) are associated with significant positive charge and the peripheral atoms ‘a’ and ‘c’ are associated with corresponding negative charge The structures (1 8)-(20) are probably the most realistic general representations of these molecules + + - &b\ - a-b-c b-a a c t t +b -+ a’ \c a b-c b=a 26t 6 26t 6 6t6 &apb\c5-a- -b- -c b-a Two electrons are accommodated in a high energy NBMO (Figure 3) and the stability of species (15)-(17) increases as the energy of this orbital is lowered Because the NBMO concen- trates the electrons on the terminal atoms (a and c) these molecules are most stable if these atoms have high electronegati- vity This is well illustrated by the fact that iodine forms hypervalent iodine(u1) compounds with fluorine and chlorine but not with bromine Stable oxygen and carbon derivatives can also be obtained when the ligand can stabilize an anion (e g MeCO 0 and NC-) For the same reason the 1,3-dipoles (15) and the ylides (1 7) are most stable when the peripheral atoms are oxygen the nitro- and sulfoxide functional groups are stable and well characterized whereas the carbon analogues (1 5)and (1 7) (a and c = CH,) are much more elusive The factors which result in a lowering of the bonding and non- bonding molecular orbitals in the species (1 5)-( 17)(z e the electronegativity of the atoms a-c) will also lower the energy of the empty antibonding orbital Consequently those species which are particularly stable are found to function as one- electron oxidizing agents For example the nitro-group can be reduced to a radical anion both chemically and electrochemi- cally [Scheme 1 (21) .+ (22)] In fact the magnitude of the one- electron reduction potential determines the antibacterial activity of nitroimidazoles such as metronidazole which are important anaerobic antibacterial agents Introduction of the extra elec- tron into the antibonding orbital (Figure 3) results in weakening of the N-0 bonds and further reduction leads to formation of ni troso-derivatives (23) via N-0 bond cleavage Analogous behaviour is seen with many hypervalent species which often function as powerful oxidizing agents For example the reaction of iodoarene difluorides (24) with electron-rich dienes can be interpreted in terms of a single electron transfer (SET) to give the radical anion (25)(Scheme 1) The I-F bonds are now weaker .[&-!I-+ ArIF -F ArI The isoconjugate bonding in the species (15)-(17) is also reflected in similarities in their chemical reactions In particular two general modes of reaction are common to all three species The first can be termed ligand coupIzng and is generalized in Figure 4a with representative examples shown in Figure 4b In the forward reactions the three-centre four-electron bond is disrupted to give a lonepair on atom b and the second pair of electrons is used to form a a bond between atoms a and c (the ligand coupling) or in the case of the ylides formation of a nitrene or carbene (which can be regarded as internally coupled ligands) Equation 1 shows an example of the well known ring- opening of aziridines to azomethine ylides Similarities with the thermal decomposition of iodobenzene dichloride to chlorine and iodobenzene (equation 2) can be recognized the reverse process can be achieved by passing chlorine through a chloro- form solution of iodobenzene The formation of sulfur ylides by addition of carbenes to sulfides (equation 3) is well known and the reverse reaction can be achieved both thermally and photochemically lo It must be emphasized that it is the overall reactions which are being compared in Figures 4and 5 a similarity in the reaction mechanisms is not implied The second general mode of reaction can best be described as syn-addition This type of reaction is generalized in Figure 5a with representative examples given in Figure 5b Again the driving-force of the forward reaction is isolation of the lone-pair on atom b and at the same time the other pair of electrons of the three-centre four-electron bond forms bonds with the atoms of a x bond In the case of the 1,3-dipoles (15) this mode of reaction is the well known 1,3-dipolar cycloaddition which is exemplified by the regiospecific addition of the N-phenylnitrone shown in equation 4 In equation 5 the outcome of treatment of cyclohexene with fluoroxenonium triflate is exclusively the syn-2-fluorocyclohexy1 triflate Equation 6 shows a character- istic transformation of sulfimides and sulfur ylides which results in three-membered ring formation in the case of reaction 6 the divalent ‘ligand’ which can be regarded as undergoing syn-addition is HN l3 4 Three-centre Bonding and SN~Transition States Three-centre four-electron bonding is not restricted to neutral molecules We have already encountered the hydrogen difluor- ide ion (13) Another anionic system in which this type of bonding makes an important contribution is the transition state of an SN2 reaction which is generalized in equation 7 and which has orbital overlap of the type (8) (Figure 3) The distribution of negative charge in these transition states (27) is closely related to the distribution of the NBMO which for simple reactions is localized on the atoms X and Y (equation 7) It is interesting to consider how the nature of these atoms (X and Y) influences the energy of the transition state This is an exercise which is left to CHEMICAL SOCIETY REVIEWS 1994 CHC02Me C02Me-Ph-N+// -t Ph-Nd \-CHC02Me ‘C02Me a c1 c +-b-a b + a Figure 4 Generalized (a) and exemplified (b) ‘ligandcoupllng’ reactions of non-classical molecules a F I FDI Xe (51I OSO2CF3 CF3S020 +-b-a+ 11 -b + ad Ph2g-kH + COPh phi + HNq ,,COPh (6)-Ph Ph Figure 5 Generalized (a) and exemplified (b) ‘syn-addition’reactions of non-classical molecules the reader who should note that in the gas phase and in aprotic solvents fluoride ion is the most nucleophilic halide ion (F-> C1-> Br->I-) which is the reverse of the generally recognized order based on studies in protic solvents HH F 5-CI-+ CH3Br C’-rBrH Recent studiesI4 of the gas phase sN2 reaction shown in Scheme 2 have shown that the transition state (29)is lower in energy than the reactants (Cl-+ CH,Br) clearly demonstrat-ing the positive contribution of three-centre four-electron bonding in sN2 transition states However the structure (29)is still a transition state and does not lie in an energy minimum Br-(CH3CI)since ion-molecule complexes (28)and (30) held together by ion-dipole forces are more stable The potential energy surface along the reaction coordinate for this gas-phase reaction (Scheme 2) is shown in Figure 6 Because the negative charge in the transition state (29)is distributed between the entering and Figure 6 The schematic reaction coordinate diagram for the SN2 leaving groups the solvation energy of this charge-delocalized reaction C1-+ CH,Br +Br-+ CH,C1 in the gas phase species is less than that for the charge-localized reactant and (Reproduced by permission from J Am Chem Soc 1991,113,9696 ) product anions Hence in solution the reactants and products have lower energy than the transition state and in contrast to C1-+ CH3Br ClCH + Br-Figure 6 the reaction profile in solution has the conventional shape This differential solvation of reactants and transition state is greatest in protic solvents and accounts for the significant t * It rate enhancement of SN2reactions in polar aprotic solvents * r 1-In contrast to pentavalent derivatives of carbon,’ pentacoor-dinated silicon anions are often stable enough to be isolated X-Ray crystallographic studies have shown a number of anionic fluorosilicates to have a trigonal-bipyramidal geometry analo-gous to an sN2 transition state l7 l8 Typical examples which have been characterized as quaternary ammonium salts are the Scheme 2 115NON-BONDING MOLECULAR ORBITALS AND NON-CLASSICAL ORGANIC MOLECULES-C A RAMSDEN pentafluorosilicate (3 1) and the diphenyltrifluorosilicate (32) In 0-these anions two distinct Si-F bond lengths are observed those I associated with the weaker hypervalent three-centre four- electron bonds are longer (ca 1 65-1 69 A) than those asso- ciated with the stron er classical two-centre two-electron bonds (ca 1 59-1 65 1)It is notable that it is electronegative fluorine that forms the most stable pentacoordinate silicon species and that in the organic anion (32) fluorine occupies the hypervalent positions (31) (32) In many modern synthetic reactions nucleophilic activation of silicon to form a hypervalent species is a key mechanistic step The organic chemistry of silicon is extensive and only a single example can be included here Allylation of benzaldehyde with (0-crotyltrifluorosilane (33) activated by caesium fluoride gives a 92% yield of the alcohol (34) with high diastereoselectivity (Scheme 3) l9 This allylic rearrangement is believed to occur by initial nucleophilic attack by fluoride ion to give the hypervalent anion (35) which then coordinates with the aldehyde to form a hexacoordinate complex (36) Reaction vzu a six-membered transition state in which the phenyl group is equatorial then leads to the observed product (34) Activation of the Si-C bond in the trifluorosilane (33) by nucleophilic attack of the fluoride ion to give the more reactive pentacoordinate species with a hypervalent Si-C bond illustrates an important general prin- ciple of modern organosilicon chemistry - F3SI 1CsF/’I€lF fl PhCHO iii MeOH/HCI (33) (34) MeOH/HClt PhCHO - F (35) (36) Scheme 3 5 Heterocyclic Mesomeric Betaines and 1,3-Dipolar Cycloadditions Heterocyclic mesomeric betaines are a large and diverse group of 1,3-dipolar molecules and a number of discrete families of these heterocycles can be recognized One large family is composed of betaines which are isoconjugate with odd alternant hydrocar- bon anions the essential relationship between betaine and anion can be summarized by the following rule 2o Neutral heterocycles isoconjugate with odd alternant hydrocarbon anions can be represented only by dipolar or hypervalent structures if the heteroatom which donates two n-electrons to the conjugated system is located at an unstarred position Examples of heterocyclic betaines defined by this rule are the systems (37)-(39) which are isoconjugate with the odd alter- nant hydrocarbons anions (40)-(42) The number of possible systems is very large (38) (39) * * -* *a** ** (43) (44) (45) Because odd alternant hydrocarbon anions are associated with a NBMO [e g (43)-(45)] which vanishes on the unstarred atoms (the smaller set) then these betaines can also be expected to be associated with a HOMO which has the characteristics of a NBMO and there is ample theoretical and experimental evi- dence to support this view 2o We should not be surprised by the association of these betaines with a NBMO This is characteris- tic of a system which is associated with three-centre four- electron bonding and these molecules are indeed systems in which a three-centre four-electron bond is part of a larger conjugated ring system This is why they can only be represented by dipolar or hypervalent structures Other types of hydrocarbon anion which are associated with NBMOs give rise to different families of neutral heterocyclic betaine when heteroatoms donating a pair of n-electrons are introduced at positions where the NBMO vanishes Consider for example the cyclooctatetraene dianion (47) which is asso- ciated with a pair of NBMOs (48) and (49) The heterocyclic betaine (46),2i in which the two sulfur atoms are at positions in which one of the NBMOs (48) vanishes is isoconjugate with the dianion (47) Similarly the pentalene dianion (5 1) is associated with a NBMO (52) and the three types of heteropentalene betaine (50A-C) are isoconjugate Again these ‘non-classi- cal’ heterocycles are examples in which a three-centre four- electron bond is integrated into a larger n-system Using the special properties of odd alternant hydrocarbons Dewar has developed a perturbation molecular orbital (PMO) theory of organic chemistry which provides deep insight into the structure and reactivity of molecules 23 The close relationship between the family of betaines exemplified by (37)-(39) and odd alternant hydrocarbon anions encouraged us to employ the PMO approach to develop a model of the structure bonding and reactivity of this class of molecule This model which is described elsewhere,*O has been successful in generalizing many properties including ionization potentials absorption spectra solvatochromism relative stability pK values and cycloaddi- tion reactions In considering the betaines [eg (37)-(39)] as perturbations of the anions [e g (40)-(42)] the largest pertur- bation is caused by the heteroatom which donates two T-electrons to the conjugated system (ze -RN+=,-O+= or -S+=) Since this perturbation by the very nature of the betaines occurs at a position at which the NBMO vanishes then we might expect the PMO approximation to be particularly good for analysing properties which are closely associated with CHEMICAL SOCIETY REVIEWS. 1994 the betaine HOMO. In fact the nature of the HOMO has a profound effect on their properties and this is particularly notable in their cycloaddition reactions and the effect of substi- tuents and aza substitution on these reactions.20 The recognition of the general rule described above provided the opportunity of systematically considering the known and unknown members of the large family of heterocylic betaines which it defines. Our interest in the relationship between struc- ture and properties and in the use of heterocyclic betaines in synthesis led us to study some simple but novel betaines including molecules which are isoconjugate with the phenalen- 1- ide anion (42). There are two distinct types of unstarred carbon atom in the phenalen-1-ide anion (cf. positions 2 and lo). Little was known about isoconjugate betaines in which a nitrogen atom is located at a peripheral bridgehead position and we embarked on a study of the simple tricyclic betaine (54) (Scheme 4).The species (54) is too reactive to be isolated but it can be trapped in situ by treatment of a solution of the salt (53) with base in the presence of a 1,3-dipolar0phile.~~ In accord with frontier molecular orbital (FMO) theory,25 this betaine with a high energy HOMO (a perturbed NBMO) can be expected to be particularly reactive towards electron-deficient 1,3-dipolaro- philes (low energy LUMO) and this proved to be the case. Reaction with ethyl acrylate produced the adduct (55) which after oxidation hydrolysis and decarboxylation gave the novel cyclazine (56).24 The regioselectivity of the cycloaddition(54)-+(55) is also notable. The asymmetry of the betaine HOMO means that one mode of addition of ethyl acrylate in which the larger coefficient of the HOMO interacts with the larger coefficient of the LUMO (Scheme 4),is energetically preferred and leads to the formation of a single regioisomer. Both the energy and topology of the betaine HOMO have a decisive influence on the reactivity. 6 Do NBMOs Control the Reactions of Hypervalent Molecules? In the preceding discussion we have emphasized the view that the NBMO-like molecular orbitals of ‘1,3-dipoles,’ including heterocyclic mesomeric betaines have an important influence on their reactivity. This observation has prompted us to wonder if the NBMO-like orbital of ‘hypervalent’ species (Figure 3) might similarly influence their rate and mode of reaction. The chemistry of hypervalent iodine compounds is an area of (53) iii liv v,vi.vii@ ci;c;H H \I HE COzEt (551 (56) Reagents 1 ClCH,CHO; ii HClO,; iii Et,N; iv CH,=CH.CO,Et; v oxidation; vi hydrolysis; vii heat. Scheme 4 considerable current research intere~t,~ 6~2 and the reactions of hypervalent xenon compounds are also attracting increasing attention. The mechanisms by which these reagents react is not clear. Plausible proposals involving carbocation intermediates have been described’ 2~26but as far as we are aware the possibi- lity that the NBMO of a hypervalent three-centre four-electron bond might play an essential role in stabilizing cationic interme- diates has not been discussed. We conclude this review by considering this interesting possibility. For the purposes of discussion we consider a mechanism (Scheme 5) for the syn-addition (Figure 5) of a generalized hypervalent iodine reagent (57) to a simple double bond. The same mechanistic principles can be applied to the rationalization of reactions of hypervalent reagents with a wide variety of substrates. There are two possibilities for the initial step (Scheme 5). The hypervalent reagent (57) may oxidize the n-system by single electron transfer (SET) to form a radical cation and a radical [e.g. (SS)] which then combine to give a cationic intermediate (60). Alternatively the iodine reagent may dissociate to an iodonium ion (59) which then undergoes electrophilic addition to the olefin yielding the same intermediate (60). The cation (60) which contains a hypervalent I-C bond may then react with the counter ion (X-) to give molecule (62).Subsequent dissociation to the iodonium salt (63) followed by SN2displacement of aryliodide gives the syn-adduct (64). When the substituent R1 is H or SiMe an alternative reaction of the cation (60) may lead to the vinyl iodonium salt (61). This sequence of events (Scheme 5) provides an interesting example of umpolung in which the iodine species function first as an electrophile and then as a leaving group. A crucial intermediate in this mechanistic interpretation (Scheme 5) is the hypervalent cation (60) which may well be stabilized by an energetically and topologically favourable interaction between the hypervalent NBMO and the empty p-orbital of the cationic centre (65). This orbital interaction is isoconjugate with the interaction between an ally1 anion and a methyl cation (66) when they are combined by intellectual ‘union’ (tu +)to form butadiene (equation 8).23Since the total n-electron energy of butadiene is demonstrably lower than that NON-BONDING MOLECULAR ORBITALS AND NON-CLASSICAL ORGANIC MOLECULES-C A RAMSDEN (57) sE$ x-X-x J Y \ Y (63) (64) Scheme 5 of the separate ions we might expect the hypervalent cation (65) to be similarly stabilized Note that during the union (67) one of the ally1 x-bonds becomes stronger (I e becomes the 1,2-bond in butadiene) whereas the other is weakened (I e becomes the 2,3- bond) The same effect might also be expected in the hypervalent cation (60) with the I-Y bond strengthened but the I-C bond weakened When the cation (60) reacts with an anion (60) +(62)(Scheme 5) this stabilization of the hypervalent system is lost Simultaneously the C-I bond should strengthen but the I-Y bond weakens making it easier to dissociate to the salt (63) thus transforming the iodine function into a good leaving group and achieving the umpolung H H (67) This possibility that the NBMO of hypervalent bonds may stabilize strategic cationic intermediates encouraged us to investigate a potential new method for introducing fluoro- substituents onto aromatic rings using xenon difluoride In particular we considered that xenon difluoride might react with aryltrimethylsilanes (68) to form an intermediate of the type (69) in which the cation is stabilized not only by the 8-effect of the trimethylsilyl substituent but also by the hypervalent xenon bonds Subsequent fluorodesilylation by F-could then give the arylxenon intermediate (70) [cf Scheme 5 (60) +(61)] which could be expected to undergo rapid ligand coupling (cf Figure 4) to form an arylfluoride When we embarked on this study the idea of generating an arylxenon intermediate [eg (70)] of even transient stability was considered rather 'speculative 'However within a few months a paper appeared describing the prep- aration isolation (80% yield) and X-ray structure determi- nation of the stable xenon derivative (71) 28 Furthermore in accord with the general properties summarized in Figure 4 this compound (7 1) was shown to undergo thermal ligand coupling to form the ester C,F,O CO C,F (68) (69) FF When 4-t-butylphenyltrimethylsilane(72) was reacted with xenofdihuoride in hexafluorobenzene solution at room temper- ature a clean reaction took place giving 1-t-butyl-4-fluoroben-zene (73) in high yield together with trimethylsilylfluoride (Scheme 6) 29 The actual mechanism of this transformation (72)-+(73) has not yet been elucidated but at present we consider that it may proceed via either the hypervalent xenon intermediates (77) and (78) or the hypervalent silicon interme- diate (75) Since xenon difluoride is a powerful oxidizing agent the first stage is probably a SET to form the radical cation (74) This species may then react to give the cationic intermediate [(77) cf (69)] or alternatively may undergo immediate fluorode- silylation via the pentavalent silicon radical (75)(cf Section 4) Both pathways lead to the aryl radical (76) which may then react with F. or XeF to form the observed product (73) Although we cannot discriminate between these two pathways (Scheme 6) at present results using CHCI or CFCl as solvent do suggest that the reaction proceeds via an aryl radical For example when the reaction of compound (72) with xenon difluoride was repeated using chloroform as solvent the yield of the fluoroderivative (73) was much lower (40%) and it was accompanied by the chloro (79)(30%) proto(80)( lo%) and trichloromethyl(8 I)( 10%) der- ivatives (Scheme 7) suggesting that an intermediate radical (76) had reacted with the solvent 29 7 Concluding Remarks The electrons in two-centre (T and x bonds are not localized between pairs of atoms they occupy molecular orbitals which are delocalized throughout the molecule Nevertheless for the rationalization of many properties particularly those which depend on the total number of electrons in a molecule (collective properties) the localized (T and 7~ bond model is very successful Similarly three-centre four-electron bonds are not localized but provided the same limitations which apply to two-centre bonds are recognized the concept can successfully rationalize many properties of the large and diverse group of molecules which are described as dipolar or hypervalent In some mole- cules d-orbitals have the correct symmetry to mix with the CHEMICAL SOCIETY REVIEWS 1994 SMe3 8 References + Me,SiFQxeFT&.;80LQ F or XeF (72) F-(731xey EZSiMe \ SNe,Q 2-8-QMe,SiF (74) (75) (76) I I (77) (78) 1 J 1 Musher Angew Chem Int Ed Engl 1969,8 54 2 R E Rundle J Am Chem Soc ,1963,85 112 R E Rundle Sun 3 G C Chem 1963 RProg Pimentel and 1,81 D Spratley ‘Chemical Bonding Clarified Through Quantum Mechanics’ Holden-Day San Francisco 1969 4 R L Kuczkowski Chem SOCRev 1992,21 79 5 G F Koser ‘Hypervalent Halogen Compounds’ in ‘The Chemistry of Functional Groups Supplement D’ ed S Patai and Z Rappo-port Wiley Chichester 1983 6 ‘Nitroimidazoles Chemistry Pharmacology and Clinical Appli- cations’ ed G E Adams A Breccia and B Cavalleri Nato Advanced Study Institutes Series Series A Life Sciences Plenum Press New York 1982 vol 42 7 J J Edmunds and W B Motherwell J Chem Soc Chem Commun 1989,881 8 A Padwa and A D Woolhouse in ‘Comprehensive Heterocyclic Chemistry’ Vol 7 ed W Lwowski Pergamon Press Oxford 1984 P 53 9 D F Bank Chem Rev 1966,66,243 10 TL Gilchrist and C W Rees ‘Carbenes Nitrenes and Arynes’ Nelson London 1969 p 14 11 R D Little in ‘Comprehensive Organic Synthesis,’ Vol 5 ed L A Scheme 6 Paquette Pergamon Press Oxford 1991 p 255 12 N S Zefirov A A Gakh V V Zhdankin and P J Stang J Org Chem 1991,56 1416 13 C R Johnson in ‘Comprehensive Organic Chemistry’ Vol 3 ed N Jones Pergamon Press Oxford 1979 p 219 14 (a)S TGraul and M T Bowers J Am Chem Soc ,1991,113,9696,(72) XeF2.CHC13.1SoC \ \ \ (b)J * (73)(4Ooh)+ Q+j+Q9699 L Wilbur and J I Brauman J Am Chem Soc 1991 113 15 W N Olmstead and J I Brauman J Am Chem Soc 1977 99 (79)(300/b) (80)(100/b) (81)(100/0) Scheme 7 NBMO to form a weakly bonding MO Present evidence suggests that this contribution of d-orbitals is small and does not appear to affect the qualitative conclusions of the [3c-4e]bond-ing model significantly A critical quantitative analysis of the role of d-orbitals in hypervalent molecules has recently been published 30 We are interested in using the ideas reviewed in this article as an aid to the discovery of reactions which may be of general value in synthetic organic chemistry new reactions of hyperva- lent derivatives of tellurium iodine and xenon are of particular interest The chemistry of hypervalent compounds is deservedly attracting increasing attention from synthetic organic chemists3 but the mechanisms of many of their transformations remain uncertain We hope that a combination of theoretical and experimental studies will lead to both a better understand- ing and useful applications of this interesting class of molecule 4219 16 J C Martin Science 1983 221 509 17 C Chuit R J P Corriu C Reye and J C Young Chem Rev 1993,93 1371 18 R R Holmes Chem Rev 1990,90 17 19 M Kira THino and H Sakurai Tetrahedron Lett ,1989,30 1099 20 C A Ramsden Adv Heterocycl Chem 1980,26 1 21 I Ernest W Holick G Rihs D Shomburg G Shoham D Wenkert and R B Woodward J Am Chem Soc 198 1,103,1540 22 C A Ramsden Tetrahedron 1977,33 3203 23 M J S Dewar and R C Dougherty ‘The PMO Theory of Organic Chemistry’ Plenum Press New York 1975 24 W D Ollis S P Stanforth and C A Ramsden J Chem Soc Perkin Trans 1 1989 945 25 I Fleming ‘Frontier Orbitals and Organic Chemical Reactions’ Wiley London 1976 26 R M Moriarty R K Vaid and G F Koser Synlett 1990 365 27 P J Stang Angew Chem Int Ed Engl 1992,31,274 28 H J Frohn,A Klose,andG Henkel Angew Chem Znt Ed Engl 1993,32,99 29 A P Lothian and C A Ramsden Synlett 1993,753 30 A E Reed and P v R Schleyer J Am Chem Soc ,1990,112,1434 3 I (a)P Magnus J Lacour and P A Evans Janssen Chzm Acta 1993 11,3 (b)M Kaqan D Koyuncu and A McKillop J Chem Soc Perkin Trans 1,1993 1771 (c) D H R Barton J Cs Jaszberenyi K LeBmann and T Timar Tetrahedron 1992,48,8881

ISSN:0306-0012    

DOI:10.1039/CS9942300111

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

TILDEN LECTURE. Studies on Thymidylate Synthase and Dihydrofolate Reductase -Two Enzymes Involved in the Synthesis of Thymidine Douglas W. Young School of Chemistry and Molecular Sciences, University of Sussex, Falmer, Brighton, BN 1 9QJ, U.K. 1 Introduction Compounds which interfere with the biosynthesis of the nucleo- side thymidine are among the most effective clinical anti-cancer drugs. The nucleic acids, DNA and RNA, use two purine bases and two pyrimidine bases in the three letter codons which ensure integrity of the genetic message on cell division and which translate to the twenty letter ‘message’ found in proteins. Three of these bases are common to both DNA and RNA but the pyrimidine base thymine (2, R’ = H) is unique to DNA, its equivalent in RNA being the base uracil ( 1, R1= H).Thymidine 5‘-phosphate, (2, R1 = 2-deoxyribose-5-phosphate), must be available before cell division will occur and so inhibition of the biosynthesis of thymidine will prevent cell division. Since uncon- trolled cell division is a feature of many forms ofcancer, we can see how studies of the biosynthesis of the base thymidine are important in the development of new anti-cancer drugs. The final process in the biosynthesis of thymidine monophos- phate, (2, R’ = 2-deoxyribose-5-phosphate), is methylation of deoxyuridine monophosphate, (1, R1= 2-deoxyribose-5-phos-phate), shown in Scheme 1 and, like many biological reactions involving the transfer of a one-carbon unit, this involves the coenzyme 5,6,7,8-tetrahydrofolic acid (5).This coenzyme is a 1,2-diamine and the nitrogen atoms N-5 and N-10 are able to form the aminal (6) as in Scheme 2. This compound is a one- carbon adduct of the coenzyme and it is able to transfer this carbon to a substrate in an enzyme-catalysed reaction. Thus the amino acid glycine (9) will react with the aminal (6) in the presence of the enzyme serine hydroxymethyltransferase (EC 2.1.2.1) and pyridoxal phosphate to yield the amino acid serine (lo) as in Scheme 3. This biological aldol condensation may be regarded as a one-carbon transfer at the formaldehyde oxi- dation level. The aminal(6) may be oxidized by a dehydrogenase to 5,lO- rnethenyl-5,6,7,8-tetrahydrofolicacid (7) which effects one-car- bon transfers at the formic acid oxidation level such as the formylation of the imidazole (1 1) which is involved in synthesis of the purine (12).Enzymatic reduction of aminal (6) to 5-Douglas Young is Professor of Chemistry at the University of Susse-u. He completed his Ph.D. with Professor A. I. Scott and post-doctoral studies with Professor R. El. Woodward. His research covers stereochemistry and mechanism of enzyme-cata-Iq’sed reactions; heterocyclic synthesis and photochemistry. In addition to the work reported here, his group were Jirst to discover the stereochemistry of ,&lactam ring-closure in peni- cillin biosynthesis, and have in- vestigated the stereochemistry of a variety of other biological processes. They have developed novel syntheses of heterocyclic systems, new p-Iactam systems and glutamate antagonists.Professor Young is Chairman of the Bio-organic Group of the Royal Society of Chemistry and a member of Perkin Council. methyl-5,6,7,8-tetrahydrofolicacid (8) allows one-carbon transfers at the methanol oxidation level to take place. Thus methylation of the amino acid homocysteine (13) to methionine (14) involves the coenzyme ‘adduct’ (8). In all of the one-carbon transfers in Scheme 3, the coenzyme (5) is regenerated and, since biosynthesis of thymidine (2) is a methylation reaction, we might expect 5-methyl-5,6,7,8-tetra- hydrofolic acid (8) to be involved in the process. Nature, however, effects this methylation in a more complex manner as shown in Scheme 4.Here the aminal (6a) is used as the one- carbon source and, since it is at ‘wrong’ oxidation level for the methylation reaction (1) +(2a), it is oxidized in the process.The hydrogen Hc in the aminal is transferred to become the third hydrogen of the methyl group and the coenzyme is oxidized to 7,8-dihydrofolic acid (1 5) in the process. The overall reaction is catalysed by the enzyme thymidylate synthase (EC 2.1.1.45) and, since the coenzyme (5) is consumed in the process, a mechanism for its regeneration is required. This is achieved by the enzyme dihydrofolate reductase (EC 1.5.I .3) which reduces dihydrofolate (1 5) to the coenzyme (5)which is in turn converted into the aminal (6a) by the retro-aldol reaction of serine (10a) using serine hydroxymethyltransferase.The methylation of the base uracil (1) to the base thymine (2a) therefore involves a cyclic process, and two of the enzymes in it are targets for clinically used anti-cancer drugs. Thus 5-fluorouracil (16) is an anti- cancer drug since it is converted into a suicide substrate for thymidylate synthase, and the anti-cancer drug methotrexate (23) acts by inhibiting the enzyme dihydrofolate reductase. The unusual nature of the one-carbon transfer involved in the methylation ofuracil (1) attracted our attention some years ago and we have worked on the enzymes involved in the process since then. Because of this interest, we also became interested in the catabolism of the two bases uracil (1, R1= H) and thymine (2, R1 = H).These are degraded in nature to p-alanine (3) and 3-amino-2-methylpropanoic acid (4), respectively, by reduction and hydrolysis. We have sho~n~,~ that, in each case, the reduction involves anti-addition of hydrogen at the si-face of C- 5 and the si-face of C-6 of the pyrimidine as indicated in Scheme 1. We have also shown3 that the anti-cancer drug 5-fluorouracil (16, R’= H) is catabolized with the same stereospecificity. 2 Studies on the Enzyme Thymidylate Synthase 2.1 A Model for the Methylation of Thymine The question of a valid chemical model for the methylation of deoxyuridinemonophosphate ( 1, R = 2-deoxyri bose-5-phos- phate) has excited much speculation and Friedkin revised an earlier4 model by suggesting5 that 2’-deoxyuridine-5’-phosphate (1, R’= 2-deoxyribose-5-phosphate) might act as an enamine with electrophile (6b)+(6c) to give the intermediate (19) as in Scheme 5.This would then rearrange in some way, with transfer of the hydrogen Hc from C-6 of the pteridine to the methyl group of the thymine moiety. Evidence in favour of the intermediacy of the Friedkin intermediate includes the discovery of a ternary complex (18) when 5-fluorodeoxyuridinemonophosphate(1 6, R’ = 2-deoxy-uridine-5-phosphate) was used as a substrate for the Here the intermediate (1 7) cannot rearrange to the enamine (1 9) when X = F. It is extremely nucleophilic and reacts with the thiol of an active-site cysteine residue. 119 CHEMICAL SOCIETY REVIEWS, 1994 H.5.".+ OAN Hc HC R' R' HC Scheme 1 H (6) Scheme 2 (7) oH3Y HN-R + H 02C H2N*sH -H02CH2NFscH3 H2NAN NHNkN$H H Scheme 3 An attractive explanation for the conversion of the interme- loss of deuterium and any thymine formed might be expected to diate (19) into thymidine (2b) and 7,8-dihydrofolic acid (15) be deuteriated at C-6 as shown in (2c). seemed to us to be the retro-ene process shown in (1 9) +(20) of Synthesis of the desired compound (19, R1 = H) had been Scheme 5. This concerted reaction would lead to 7,8-dihydrofo- reported by Huennekens'O to be achieved by simple alkylation late (15) and the quinone-methide type intermediate (20) in of 5,6,7,8-tetrahydrofolic acid (5) with 5-chloromethyluracil which the hydrogen Hc had been transferred to C-6 of the (21) as shown in Scheme 6.When we repeated this reaction, pyrimidine. A stereospecific enzyme-catalysed prototropic shift however, we found that bis-alkylation to the product (22) had involving a single active-site base would then transfer Hc fortuitously the same calculated combustion data as the mono- entirely to the methyl group as is founds>g in nature and shown in adduct (19, R1 = H), and our product had the reported'* ultra- (2b). violet spectrum. The 'H-NMR and mass spectral data, however, As a means of testing this hypothesis, we decided to synthesize indicated that it was, in fact, the bis-adduct (22) which was the compound (19, R1= H) in which Hc was deuteriated and obtained in this reaction." then attempt its chemical conversion into thymine by pyrolysis.In order to obtain a suitable model we therefore needed to If the retro-ene process were to give (20, R' = H, Hc = 2H)then protect N-10 of 5,6,7,8-tetrahydrofolic acid against alkylation. there would be no reason for the purely chemical reaction to be We were able to achieve this" as outlined in Scheme 7 by stereospecific. Indeed an isotope effect would mitigate against hydrolysing the anti-cancer drug methotrexate (23) to the cor- TlLDEN LECTURE: STUDIES ON THYMIDYLATE SYNTHASE AND DIHYDROFOLATE REDUCTASE-D W. YOUNG H02C HD HBH2NxHE 3 R = aCO-gf,C02H METHOTREXATE3 C02H Scheme 4 H -0 N4 ov>R'l, X=H X/ 16,X=F X=F Enz-SH ____t H -- enzymatic i<Ood N-R' R= I I.I. H L02H Scheme 5 CHEMICAL SOCIETY REVIEWS, 1994 H H (5) (19, R' = H) \\ Scheme 6 H2N ANANd . .. ANANJ 0 \CO2H Scheme 7 responding amide (24). This could be reduced to N-lO-methyl- 7,8-dihydrofolic acid (25) using aqueous dithionite. Reduction with either NaBH, or NaB2H, then gave the unlabelled N-10-methyl-coenzyme (26) or its deuteriated analogue (26, Hc = 2H). These were separately alkylated to the protected putative intermediates (27) and (27,Hc = 2H) using 6-chloro- methyluracil (21).l Pyrolysis of the model compound (27) in the solid state gave thymine (2b, R1= H), identical in all respects with an authentic sample. When the deuteriated compound (27, Hc = 2H) was used, a sample of thymine was obtained in 47% yield.This had a 2H-NMR spectrum indicating that all of the deuterium was in the methyl group, and a mass spectrum which indicated 25% monodeuteriation. The 2H: H ratio was extremely high, consi- dering the ratio of protium to deuterium in the starting com- pound (27,Hc = 2H) and the fact that an isotope effect would favour transfer of protium rather than deuterium. It was evident from the fact that (2b, R1 = H, Hc = 2H) was obtained rather than (2c,R1 = H, Hc = 2H), that the retro-ene mechanism sug- gested in Scheme 5 could not account for the results. The results of the model reaction parallel closely those found in the overall enzymatic reaction and so, if the intermediate (19) is involved in the process, the chemical results may be relevant to the enzymatic mechanism, One possible explanation of the results is suggested in Scheme 8 where the concerted process shown in (19b) may account for cleavage of the carbon-nitrogen bond.The hydrogen Hc may be sufficiently close in the crystal (or at the active site) to allow the redox process shown in (28) to occur and the labelled hydrogen would end up in the methyl group of the resultant thymine (2b,R1 = H,Hc = 2H). 2.2 Studies on the Enzyme-catalysed Reaction In order to understand the nature of the reaction catalysed by thymidylate synthase further, we determined to investigate the overall stereochemistry of the process. This would involve TILDEN LECTURE: STUDIES ON THYMIDYLATE SYNTHASE AND DIHYDROFOLATE REDUCTASE-D.W. YOUNG 0 (19,R = H) (2b,R' = H) (36) (35) (34) Scheme 9 synthesis of the intermediate (6a) stereospecifically labelled at C- 1 1 with deuterium and at C-6 with tritium (see Scheme 4). Use of this in the enzymic reaction and degradation of the resultant thymidine (2a) to acetic acid would allow us to assess the stereochemistry of the chiral methyl group and hence the overall stereochemistry of the reaction. However, when we began this work the absolute stereochemistry of the coenzyme 5,6,7,8- tetrahydrofolic acid (5) at C-6 had not been determined and so we first investigated this. This work is reported in Section 3.1 of this review. As a first step to investigating the stereochemistry of the methylation reaction, we determined to prepare samples of L-serine, stereospecifically labelled at C-3 with deuterium and to use the enzymes serine hydroxymethyltransferase and thymidy- late synthase in tandem to follow the stereochemistry from (1Oa) through (6a) to (2a) in the cycle represented in Scheme 4.This required synthesis of samples of L-serine (1Oa) stereospe- cifically labelled at C-3. We had previously prepared samples of L-cystine stereospecifically labelled', at C-3 and had used these to study the stereochemistry of the cyclization which gives rise to the p-lactam ring in penicillins' and cephalosporins. l3 This work involved total synthesis and was unduly laborious and so, for other labelled amino acids, we devised a chemico-enzymatic method which we were able to use to prepare samples of L-glutamic acid stereospecifically labelled at C-3 with deuter- ium.I4 This method was adapted to the synthesis of the labelled samples of serine,' as shown in Scheme 9.The enzyme aspartase is commercially available and was known to add the elements of ammonia with anti-stereo- chemistry across the double bond of fumaric acid (29) to give aspartic acid. Thus by using (a) fumaric acid (29) and N2H402H and (b) [2,3-2H,]fumaric acid (29, HA = ,H) and NH,OH we were able to prepare large quantities of (2S, 3R)-[3-2Hl]- and OS, 3S)-[2,3-2H,]aspartic acids, (30, HB= 2H) and (30, HA= *H), respectively. Conversion of these into the trifluoro- acetyl anhydrides (32) and ring-opening with methanol gave primarily the westers (33) which were purified as the acid chlorides (34).These were converted into the diazoketones (35) which could be reduced to the methyl ketones (36) without loss of label. Although Baeyer-Villiger rearrangement was not as regiospecific as we had hoped, it proceeded with retention of stereochemistry at the migrating primary chiral centre so that hydrolysis of the products gave (2S,3R)-[3-2Hl]- and (2S, 33- [2,3-2H,]serines(10b, HB= ,H) and (lob, HA = ,H), respect- ively. 's In a recent synthesis of labelled D-amino acids, we have found that the quality and yield of the product from the first step in this synthesis can be improved by using immobilized cells of Escherichia coli.' At this point in our studies, Benkovic and Floss' showed that the overall stereochemical outcome in the conversion of serine (1Oa) into thymine (2a) was as shown in Scheme 4. Benkovic'* also prepared samples of 5,10-methylene-5,6,7,8-tetrahydro-folic acid (6a) stereospecifically deuteriated at C-1 1 and used the NOE between H-7R and H-11s shown in (6d) to define the absolute stereochemistry of the label. This allowed him to determine the stereospecificity of the two individual reactions (loa) -+(6a) and (6a) -+ (2a) in Scheme 4, the stereochemistry being as indicated in the Scheme. Although our original goal had been so elegantly achieved, we were able to use our stereospecifically labelled samples of L-serine, with the biosyn- thesis of tuberin (37) as a 'black box', to verify that the dehydrogenase which converts (6)into (7) does so by removal of the 11-pro-R-hydrogen.lg We have also converted our samples of labelled L-serine into stereospecifically labelled samples of ethanolamine (38) and have shown that the stereochemistry of the reaction catalysed by the enzyme ethanolamine ammonia- lyase (EC 4.3.1.7) and mediated by coenzyme B,, is as shown in Scheme 10.2o CHEMICAL SOCIETY REVIEWS, 1994 HA \OH HB (38) (39) Scheme 10 H LO*HH Scheme 11 3 Studies on the Enzyme Dihydrofolate Reductase 3.1 Stereochemistry of Reduction of the Substrate Our initial interest in this enzyme was sparked by our realization that, when we wished to assess the overall stereochemistry of the reaction catalysed by the enzyme thymidylate synthase, the absolute stereochemistry at C-6 of the coenzyme (5) was un- known.This centre is introduced during the reduction of 7,8- dihydrofolate (15) by dihydrofolate reductase, and indeed the enzyme will also reduce the vitamin folic acid (41) to the coenzyme (5)as in Scheme 1 1. We therefore determined to assess the stereochemistry of the reduction at both C-6 and C-7 of folic acid (41), originally intending to achieve this using comparison with unlabelled and stereospecifically labelled samples prepared by total synthesis. We had gone some way towards this goa121 when two events changed our direction. The first was the opportunity to collaborate with Dr.J. Feeney and his colleagues .and the second was a report by Fontecilla-Camps et ~1who had separated the diastereoisomers of the 'cancer rescue' agent folinic acid (42) and converted them separately into the bromide hydrobromides (7) as in Scheme 11. The X-ray crystal structures of these allowed their absolute stereochemistry to be assigned. We therefore prepared 7,8-dihydrofolic acid (1 5)by dithionite reduction of folic acid (41) and reduced this, as in Scheme 12, to the biologically active coenzyme (5)using dihydrofolate reduc- ta~e.~Clean monoformylaton of the biologically produced compound proved difficult but we were able to bis-formylate using formic acetic anhydride. The product (43) could be specifically deformylated at N-10 using alkali to yield a sample of folinic acid (42) which was compared with the samples whose stereochemistry had been assessed by X-ray structure analysis of the cyclization product (7).This showed that the stereo-chemistry at C-6 of the active coenzyme could formally be defined as (S).23 To complete our study of the stereochemistry of reduction of folic acid (41) by the enzyme dihydrofolate reductase we needed to discover the stereochemistry of reduction at C-7 and also the stereochemical origin of the hydrogen at C-4 of NADPH (46) which was transferred to C-6 and C-7. This was achieved by the particularly economical experiment shown in Scheme 13. (4R)- [4-2H1]NADPH (46) was first synthesized by reduction of [4-2H]NAD+ (45) (itself prepared using the method of San Pietr~~~)using the pros-specific enzyme glucose-6-phosphate ~ dehydrogenase and glu~ose-6-phosphate.~ This was then used ~ in an 'NMR tube experiment'.Figure 1A shows the 'H-NMR spectrum of the coenzyme (5) obtained using unlabelled NADPH to reduce folic acid (41). The two protons H-7 of the coenzyme (5) are part of an ABX system and, by careful comparison with model we were able to assign the proton with the larger vicinal coupling to H-7R, anti to H-6 and that with the smaller vicinal coupling to H-7S, syn to H-6. Figure 1B shows the spectrum when (4R)-[4-2-H,]NADPH was TILDEN LECTURE: STUDIES ON THYMIDYLATE SYNTHASE AND DIHYDROFOLATE REDUCTASE-D. W. YOUNG R = eCO-Mf,CO2H CHO 0 H‘ AH2N N H (7) (42) Scheme 12 I ’@ -&oNH2 ‘;3&CON..I0 ?PONH2____t R R’ R’ Scheme 13 used in the experiment. Only H-7R remains, indicating that H-6 A 9,Q’-Hand H-7S are deuteriated and that reduction has occurred at both centres using the 4-pro-R hydrogen of NADPH.23 An X-ray structure of a ternary complex of the anti-cancer drug methotrexate, the coenzyme NADPH, and dihydrofolate 7SH reductase has shown,26 that, whilst H-4R of NADPH is close to C-6 and C-7 of the pteridine ring of methotrexate (23), it is the opposite face of the pteridine system which is close to this reducing hydrogen from that face of the pteridine ring which is actually reduced when folic acid (41) is the substrate. Put in terms of the Fischer ‘lock/key’ analogy for enzymes, one key is entering the lock ‘upside down’ compared to the other. Figure 2 is an attempt to illustrate this in cartoon form.B 3.2 Stereochemical Aspects of the Protein After these initial studies on substrate binding, Feeney and his colleagues went on to showz7 that dihydrofolate reductase exists in three conformations whose proportions are pH-dependent and that both folate (41) and methotrexate (23) bind in the same way in two of these conformations. Only folic acid (41) binds in the third conformation where the opposite face of the pteridine I 1 I ring is exposed to the coenzyme NADPH from that exposed in 3.5 3.3 3.1 the other two conformations. It is this third conformation which p.p.m. is catalytically competent.Figure 1 Part of the 270 MHz ‘H-NMR spectrum in ‘H,O of 5,6,7,8- Feeney et showed in this work that protons tetrahydrofolic acid (5) prepared by reduction using dihydrofolate from Leu-19 and Leu-27 Of the enzyme came within reductase and (A) NADPH; (B) (4R)-[4-’H1]NADPH. The chemical distance of H-7 Of bound meth~trexate.~ At this time the shift scale is referenced to sodium 2,2-dimethyl-2-silapentane-5-prochiral methyl groups of leucine and valine residues in pro- sulfonate. CHEMICAL SOCIETY REVIEWS, 1994 \H A B Figure 2 Representation of the binding of (A) methotrexate (23) and (B) folic acid (41) at the active site of dihydrofolate reductase. H n-COiBu (474 (48) r Scheme 14 teins could not be distinguished by NMR spectroscopy, although such side chains were important in hydrophobic binding.We therefore resolved to prepare a sample of dihydro- folate reductase in which the leucine residues had their terminal methyl groups distinguishable by stereospecific isotopic label- ling, and to use these to assign the prochiral methyls of these residues in the protein. Our first task was to synthesize stereospecifically labelled samples of leucine and to this end we were able to adapt some work which we had undertaken to assess the stereochemistry of the reaction catalysed by the enzyme glutamate r-carboxylase. The plan was to use a protected pyroglutamic acid derivative (49) as a chiral template as shown in Scheme 14. Here the bulky group at C-2 would have a large 1,3-interaction at C-4 so that either cis or trans products, (50a) or (48) respectivelv, might be prepared stereospecifically.Further, the NOEs shown in (50a) and (48) could be used unambiguosly to define the stereo- chemistry. Ring-opening and functionalization would then give the labelled samples of leucine, (47b) or (47a) respectively. The protected pyroglutamic acid derivative (49) proved sur- prisingly difficult to obtain, but we were eventually able to prepare it in good yield and in large amounts as shown in Scheme 15 by esterification of pyroglutamic acid (51) using t-butyl acetate and perchloric acid followed by preparation of the urethane (49), using di-t-butyldicarbonate and DMAP in aceto- nitrile. The urethane (49) was converted into the enaminone (53) using Bredereck's reagent as shown in Scheme 15.29 This enaminone has served as a useful synthetic intermediate in our studies on glutamate antagonist^,^^ glutamate-y-carboxylase,28 and amino acid synthesis31 and our intention was to convert it into the enone (54) so that we might investigate asymmetric induction in hydrogenation of this compound.The enamine (53) was readily converted into the enone (54) in excellent yield using DIBAL.29 A small amount of a mixture of the diastereoisomers of the Mannich base (55) was also obtained as a by-product. Hydrogenation of the enone (54) specifically gave the cis-methylated product (50b), the catalyst directing attack from the less-hindered face. The stereochemistry of the product was verified by observation of the NOEs indicated in (50a) in Scheme 14.When the diastereoisomeric mixture (55) was subjected to catalytic hydrogenation, the cis-product (50b) was again obtained in excellent yield, suggesting an elimination/ addition mechanism with stereochemical control in the addition step. This encouraged us to investigate direct catalytic hydroge- nation of the enaminone (53) which gave the cis-4-methyl- substituted pyroglutamate derivative (50b) in good yield.29 We had now succeeded in the first part of our synthetic plan, outlined in Scheme 14. All that remained was to effect ring- opening and to convert the carbonyl group of the pyroglutamate to C2H3. The ring-opening reaction was achieved without epimerization using aqueous LiOH in the THF, yielding the optically pure acid (56) as shown in Scheme 16.This was converted into the deuteriated alcohol (57) by formation of the mixed anhydride and reduction with NaB2H,. The alcohol was converted into the iodide (58) using methyltriphenoxyphospho-nium iodide in HMPA, and this was reduced in situ at 70 "Cto give the protected labelled leucine derivative (59). Deprotection was achieved by hydrolysis in 6N aqueous HCl at room temper- ature giving (2S,4R)-[5,5,5,-2H3]leucinehydrochloride (47b), the I3C-NMRspectrum of which is shown in Figure 3, allowing 127TILDEN LECTURE STUDIES ON THYMIDYLATE SYNTHASE AND DIHYDROFOLATE REDUCTASE-D W YOUNG + C0;Bu (55) Scheme 15 'BUO2C' 'Bu02C' H ,.CH3 'BUO~C-~~J~~~,H H ,CH3 ,- 'BUO~C-~~@~~3 H02C 'Bu02C 'Bu02C (59) Scheme 16 ~,____,__._,__..,.__.24.5 23.5 G(PPm) Figure 3 Part of the 'H-decoupled 13C-NMR spectrum in 20% 2HCl/ 2H,0 at 125 8 MHz of (a) (2S)-leucine hydrochloride, and (b) (2S,4R)-[5,5,5-2H3]leucinehydrochloride (47b) assignment of the methyl absorptions. Because of overlap in the H-NMR spectrum, a two-dimensional 3C-1H shift correla- tion, shown in Figure 4, was required to assign the methyl proton absorptions In both 'H- and 13C-NMR spectra, the signals due to the 4-pro-R methyl group were to lower field The It I''"~~'"l"~"~~'~I"'~~~~''I~~~25 24 23 22 G(PPm1 Figure 4 Two-dimensional IH-l 3C shift correlation of leucine hydro- chloride in 20% zHC1/2H20 remainder of the 13C-NMR spectrum of the labelled sample of leucine was as expected, except that there was a second reso- nance in the region of the absorption due to C-4.This could be explained by the presence of a small amount of (2S, 4R)-[5,5-2H,]leucine which would arise due to the NaB2H3CN having been less than 100% isotopically pure. The p-isotope shift would cause the C2H3-Cand C2H2H-C resonances to absorb differ- ently This interpretation was confirmed by a DEPT experiment and the ‘H-NMR spectrum indicated that there was ca 8% protium in the 4-pro R-methyl group The stereospecifically labelled sample of leucine (47b) was used as the sole source of this amino acid in growing an auxotrophic strain of LactobaczZZus case2 The resultant dihydro- folate reductase was isolated and purified and a binary complex with methotrexate (23) was prepared 32 Comparison of the leucine methyl absorptions in the ‘H-NMR spectrum of this sample with those of an unlabelled sample, as shown in the Table, allowed twelve of the thirteen methyl resonances to be assigned 32 Table 1 Chemical shifts (measured from dioxan and then referenced to DSS) of the leucine methyl resonances in the complex L casez dihydrofolate reductase- metho trexate-NADPH High field Low field Leu 4 117* 0 52t Leu 12 0 93* 0 SO? Leu 19 0 66* 0 46t Leu 23 0 65* 0 06t Leu 27 0 64* -0 Olt Leu 54 0 367 -0 07* Leu 62 0 66t 0 43* Leu 94 0 78 0 78 Leu 113 0 30t -0 96* Leu 114 107t 0 98* Leu 1 18 -0 33* -0 52t Leu I31 0 44t -0 02* Leu 151 115* 0 85t * Absent In enzyme from (2S,4R)-[5,5,5-*H3]1euc~net Present In enzyme from (2S,4R)-[5,5,5-ZH,]leucine This method is applicable to other proteins although, since we commenced this work, the elegant use of ‘biosynthetic fractional 13Clabelling’ has allowed the methyl resonances in valine and leucine residues in some proteins to be assigned 33 4 Conclusions and Prospects In this article I have concentrated mainly on our own experi- ments on two of the enzymes involved in the synthesis of the DNA base thymine in nature Where appropriate, I have referred briefly to other areas of study that this work has taken us into The importance of these enzymes as targets for anti- cancer drugs means that work will continue and our knowledge of them will continue to be enhanced The details of the chemical mechanism of the methylation of uracil will continue to present an intellectual challenge for some time to come It is perhaps appropriate that, in this year of the centenary of Emil Fischer’s lock/key analogy for enzymes,34 I have reviewed our work on a case which required a less rigid view of this analogy Our recent work on examination of protein interac- tions IS currently being applied to proteins other than dihydrofo- late reductase, and we are looking at details of hydrophobic binding in leucine zipper proteins and ubiquitin CHEMICAL SOCIETY REVIEWS, 1994 5 References 1 For a review on one-carbon transfer mediated by tetrahydrofolic acid see D W Young in ‘Chemistry and Biology ofPteridines’, ed J A Blair, Walter de Gruyter, Berlin, 1983, p 321 2 D Gani and D W Young, J Chem Soc Chem Commun , 1983, 576, ibid, J Chem SOC Perkin Trans I, 1985, 1355 3 D Gani, P B Hitchcock, and D W Young, J Chem Soc Chem Commun , 1983, 898, ibid, J Chem SOC Perkin Trans I, 1985, 1363 4 M Friedkin and A Kornberg In ‘The Chemical Basis of Heredity’, ed W D McElroy and B Glass, Johns Hopkins Press, Baltimore, 1957,p 609 5 M Friedkin in ‘The Kinetics of Cellular Proliferation’, ed F Stohlman, Grune and Stratton, New York and London, 1959, p 97 6 C A Lewis, P D Ellis, and R B Dunlap, Biochemistry, 1981, 20, 2275 7 P C Plese and R B Dunlap J Biol Chem , 1977, 252, 6139 and references cited therein 8 E J Pastore and M Friedkin, J Biol Chem , 1962,237, 3802 9 M Y Lorenson, G F Maley, and F Maley, J Biol Chem, 1967, 242,3332 10 V S Gupta and F M Huennekens, Biochemistry, 1967,6,2168 11 P A Charlton and D W Young, J Chem SOC Chem Commun , 1980,614, ibid, J Chem Soc Perkin Trans 1, 1982, 1363 12 D J MorecombeandD W Young, J Chem Soc Chem Commun , 1975, 198, D W Young, D J Morecombe, and P K Sen, Eur J Biochem, 1977,75, 133 13 J A Huddleston,E P Abraham,D W Young,D J Morecombe, and P K Sen, Biochem J, 1978,169,705 14 S J Field and D W Young, J Chem Soc Chem Commun ,1979, 1163, ibid ,J Chem Soc Perkin Trans 1, 1983,2387 15 D Gani and D W Young, J Chem Soc Chem Commun, 1982, 867, ibid, J Chem Soc Perkin Trans I, 1983,2393 16 B S Axelsson, K J O’Toole, P A Spencer, and D W Young, J Chem Soc Chem Commun ,199 1,1085, ibid ,J Chem Soc Perkin Trans 1, 1994,807 17 C Tatum, J Vederas, E Schleicher, S J Benkovic, and H Floss, J Chem SOC Chem Commun ,1977,218 18 L J Slieker and S J Benkovic, J Am Chem Soc ,1984,106,1833 19 K M Cable, R B Herbert, V Bertram, and D W Young, Tetrahedron Lett, 1987,28,4101 20 D Gani, 0 C Wallis, and D W Young, Eur J Biochem ,1983,136, 303 21 D W Young, Chem Ind (London), 1981,556 22 J C Fontecilla-Camps, C E Bugg, C Temple, J D Rose, J A Montgomery, and R L Kisliuk, J Am Chem Soc ,1979,101,6114 23 P A Charlton, D W Young, B Birdsall, J Feeney, and G C K Roberts, J Chem Soc Chem Commun , 1979,922, ibid, J Chem SOC Perkin Trans I, 1985, 1349 24 A San Pietro, J Biol Chem , 1955,217, 579 25 D W Young in ‘Isotopes in Organic Synthesis’ ed E Buncel and C C Lee, Elsevier, Amsterdam, 1978, Vol 4, pp 184-188 26 D A Matthews, R A Alden, J T Bolin, D J Filman, S T Freer, R Hamlin, W G J Ho1,R L Kisliuk,E J Pastore,L T Plante,N Xuong, and J Kraut, J Biol Chem ,1978,253,6946 27 B Birdsall, J Feeney, S J B Tendler, S J Hammond, and G C K Roberts, Biochemistry, 1989,28, 2297 28 R A August, A N Bowler, P M Doyle, X J Durand, P Hudhomme, J A Khan, C M Moody , and D W Young in ‘Molecular Recognition Chemical and Biochemical Problems 11’, ed S M Roberts, Special Publication No 1 I 1, The Royal Society of Chemistry, Cambridge, 1992, p 121 29 R A August, J A Khan, C M Moody, and D W Young, Tetrahedron Lett, 1992,33,4617 30 A N Bowler,P M Doyle,andD W Young,J Chem Soc Chem Commun, 1991,314 31 C M Moody and D W Young, Tetrahedron Lett, 1993,34,4667 32 G Ostler, A Soteriou, C M Moody, J A Khan, B Birdsall, M D Carr, D W Young, and J Feeney, FEBS Lett, 1993,318, 177 33 D Neri, G Otting, and K Wuthrich, Tetrahedron, 1990, 46, 3287 and references cited therein 34 E Fischer, Ber Dtsch Chem Ges , 1894, 27,2985

ISSN:0306-0012    

DOI:10.1039/CS9942300119

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

Structure and Dynamics of Electrolyte Solutions. A NMR Relaxation Approach Antonio Sacco Dipartimento di Chimica, Universita di Bari, Campus Universitario, Via Orabona 4, 70126 Bari, Italy f Introduction To understand the properties of electrolytes in solvent media it is important to examine the solute-solvent (solvation) interactions or, in the case of solvent mixtures, the enrichment of one of the components of the mixture in the solvation sphere of the solute (preferential or selective solvation). Numerous theoretical approaches have been applied to the study of solvationl and a large number of experimental techniques have also been used.2 Of the latter, the best results have been obtained using spectro- scopic methods. In recent years NMR spectroscopy has been very successful.In fact, it is possible to monitor the behaviour of the individual components of the solution by observing separa- tely the properties of either the nuclei of dissolved electrolyte or the nuclei residing in solvent molecules. However, NMR relaxa- tion methods provide information on both the structure and dynamics of solutions. 2 Theoretical Principles of Nuclear Magnetic Relaxation Around 1960 it was realized that nuclear magnetic relaxation allowed microscopic properties of electrolyte solutions to be studied. Until that time NMR studies had been based almost exclusively on the determination of chemical shifts. NMR relaxation experiments observe the decay of the longitudinal or transversal components of the nuclear magnetization towards equilibrium values following the application of a perturbating radiofrequency pulse.The approach towards the equilibrium is different for the components parallel and perpendicular to the external magnetic field, and is characterized by relaxation times, T, and T,, respectively. Here we are interested in the longitudi- nal (spin-lattice) relaxation, as characterized by Tl. The advantage of relaxation studies stems from the fact that generally the relaxation rate, l/Tl, obeys the relation where E, and T~ represent the energy of interaction of the particular relaxation process and a molecular correlation time, Antonio Sacco was born in Foggia in 1939. He received his laurea degree in Chemistry from the University of Bari.He began his academic career as an assistant in Physical Chem- istry and lecturer in Spectro- scopy at the University of Bari, where since 1980 he has been Associate Professor of Spec-troscopy. His principal re-search interests lie in the investigation of the structure and microdynamics of pure li- quids, binary mixtures, and electrolyte solutions by NMR spectroscopy. 129 respectively; h is Planck's constant divided by 2~.The energy E, is linked to the structural properties of the solution, while T~ is linked to the dynamics. Therefore by determining the relaxation time, and with the knowledge of E,, we obtain information on the dynamics of the solutions and, vice versa, with the knowledge of T~ we obtain information on the structure of the solutions.Another advantage derives from the use of relaxation pheno- menon: it is possible to separate inter- and intramolecular contributions. The most important relaxation mechanisms involve the magnetic dipole-dipole (DD) and nuclear electric quadrupole-electric field gradients (QF) interactions. In both cases, the relaxation rate is the sum of intra- and intermolecular contributions: A knowledge of the two contributions leads not only to the identification of the interactions between ion-solvent, ion-ion, and solvent-solvent but also probes the interactions within the solvent molecules. In the case of dipolar relaxation, the sepa- ration is directly obtained using the method of isotopic dilution which singles out interactions between nuclei at specific molecu- lar sites.3 In the case of quadrupolar interaction the relaxation is either purely intramolecular, if the nucleus resides in a molecule, or purely intermolecular, if the nucleus resides in a monoatomic ion or noble gas species.2.1 Dipole-Dipole Relaxation (DD) Two types of dipole-dipole magnetic interactions are effective in NMR relaxation in solution: (i) electronic dipole-nuclear dipole and (ii) nuclear dipole-nuclear dipole. The former are of major importance where unpaired electron spins are present in solu- tion: e.g., in the presence of dissolved oxygen or paramagnetic transition metal ions. In the cases considered here, simple diamagnetic ions are involved with complete electronic structure and therefore this effect is absent.Therefore, in cases where the relaxation in solution is dominated by nuclear dipole-nuclear dipole interaction, the following equations apply for the spin- lattice relaxation rate 1/T, for two like spins (e.g. protons): -1 = -3 y4h2Z(Z + l)[J(w) + J(2W)l (3)TI 2 with w = ~XVwhere v is the resonance frequency of the particular nucleus, 7 its magnetogyric ratio, Ithe nuclear spin, and J(w)are the spectral densities at frequencies o and 2~.The spectral density is the Fourier transform of the autocorrelation function which describes motion of the internuclear vector r. Let us now consider an intramolecular relaxation process, where both nuc- lei belong to the same molecule.Then, the correlation function is often assumed to decay exponentially with a single correlation time T~,which describes the molecular reorientation. With this assumption the familiar equation for interacting spins of the same type is obtained: Moreover, for fluids of not too high viscosity, the correlation time is of the order of picoseconds, while for protons v is I30 typically of the order of 100 MHz, so that WT, << I. In this case, denoted as an 'extreme-narrowing' regime, equation 4simplifies to I(I + 1)-2y4h2-.TC -r6 On the other hand, if the interacting nuclei belong to different molecules, the situation is more complicated, because the inter- nuclear distance also changes with time due to self-diffusion of the molecule^.^ In this case, assuming again the extreme-narrowing assumption to hold, the intermolecular relaxation rate between two protons is written in a proper approximation as3 where cH is the proton concentration and gHH is the radial distribution function, which is the probability of finding a proton at a distance Y from another proton selected at random, and 'a' is the closest distance of approach.In this case the correlation time 7, is to a good approximation given by a2 7c= -(7)30 where D is the self-diffusion coefficient of the fluid (or in mixtures the mean self-diffusion coefficient of the spin-carrying species). As D is often known, or can be measured by the NMR spin-echo technique (see below), information can be obtained on gHH,i.e.on the structure of the s~lution.~ This method is particularly suitable when combined with the so-called isotopic dilution technique. The constant in equation 6 contains the product of the squares of the magneticogyric ratios of the interacting spins Iand S,yfy;. If we replace the interacting proton S by a deuteron with a substantially smaller magneto- gyric ratio, we can 'switch off' the respective interaction. In this way, by site-specific isotopic substitutions, one obtains separate informations on site-site distribution functions, the combi- nation of which then yields information on mutual molecular orientations. Evidently, such experiments are not limited to protons as interacting nuclei, but can e.g. be performed with heteronuclear interactions such as I9F-'H or 7Li-1H as well.2.2 Quadrupole Relaxation of Ionic Nuclei in Pure Solvents A nucleus endowed with an electric quadrupole moment eQ may interact with external electric field gradients, causing the nucleus 1)to relax. In the limit of extreme narrowing (~7~<< the relaxation rate is given by: -I --2z j," (VJO). VZz(t))dt+ re).TI 812(2Z+ 1) h where (Vz,(0).VZZ(t))is the time correlation function of the fluctuation of the electric field gradient at the nucleus, Vz,being a diagonal component of the electric field gradient tensor in the laboratory frame. The brackets (. ..) denote the ensemble average. No experimental method is, however, capable of yield- ing (. ...) directly in the isotropic liquid phase.If we assume that the field gradient time correlation function decays exponentially with a characteristic single value, we have: (9) where A is a constant for a given nucleus. We limit discussion to the quadrupolar relaxation of nuclei in monoatomic ions, where the field gradients causing relaxation are of purely intermolecular origin. Examples are alkali metal nuclei (as 23Na+ or or halide nuclei (as 35Cl- or 81Br-). Various theories have been formulated in order to interpret CHEMICAL SOCIETY REVIEWS. 1994 experimental relaxation data. The best results have been obtained by the 'electrostatic the~ry'.~ According to this theory, the electric field gradients necessary for the relaxation of a quadrupolar nucleus are generated principally by dipoles of the solvent molecules and by monopoles of the counter ions sur- rounding the relaxing ionic nucleus. Experimentally, the sepa- ration of the ion-solvent interaction is achieved by measuring the ionic relaxation rates as a function of the salt concentration, followed by extrapolation to infinite dilution of all ionic specie^.^ The field gradient acting at the ionic nucleus results from the following factors: (i) The dipole moments m of the solvent molecules cause relaxation. The resulting field gradient is pro- portional to rn/r& where ro is the closest distance of approach between the ion and the solvent molecule. Hence, the resulting contribution to the relaxation rate depends on r; 8, which makes quadrupolar relaxation a very short-range probe for local structure and dynamics of ions.(ii) There is a distortion of the spherical electronic cloud around the ionic nucleus caused by polarization due to the perturbing field of the dipoles. This distortion can be appropriately taken into account by the so-called Sternheimer anti-shielding factor, which enters into the constant A in equation 9. (iii) There is a further 'electronic' contribution due to the distortion of the spherical electronic cloud by collision with the solvent molecules which appears however to play an insignificant and is thus neglected in the formula given below. Neglecting minor effects, the final formula of the relaxation rate of a quadrupolar nucleus accord- ing to Hertz's theory is given by: where n, is the solvation number of the ion. Moreover, the relaxation rate is very sensitive to the arrange- ment of the solvent molecules around the nucleus.In fact, the geometry of the solvation shell plays a major role in nuclear relaxation: e.g.,in a perfect arrangement of tetrahedral or octahedral symmetry of the solvent molecules the mean-square field gradient would vanish, and hence, the nuclear relaxation would be infinitely slow. This effect is called 'field gradient quenching', and is taken into account in Hertz's theory by the parameter A, which assumes values ranging between 0 and 1. A = 1 means the total absence of quenching (complete asym- metry), while perfect symmetry is described by A = 0.Obviously, for intermediate situations of symmetry we shall have all the values included between 0 and 1. 2.3 Quadrupole Relaxation of Ionic Nuclei in Mixed Solvents For mixed solvents further complications emerge. In fact, analogous to what was said in the case of ions in pure solvents, we ought now to take into account the dipolar moments of the two solvents (m,and m,),two distances (rol and rO2),and two reorientational correlation times (7~1and T,~).Then, the genera- lization of equation 10 leads to the following expression for the relaxation rates of ions dissolved in mixed solvents as a function of the molar fractions x,and x, = 1 -x,of the two components of the mixture:s In order to use this expression it is necessary to know the reorientational correlation times of solvent molecules in the mixtures TJx,) and in the corresponding pure solvents 7{, (i = 1,2).These may be obtained by means of subsidiary meas- urements of QF intramolecular relaxation rates of nuclei such ,H and 14N which may be present in the solvent molecules.6 The STRUCTURE AND DYNAMICS OF ELECTROLYTE SOLUTIONS A NMR RELAXATION APPROACH-A SACCO k coefficients which represent the ratio between the reorienta- tional correlation times of the solvent molecules in the first coordination sphere of the ions and those in the bulk, may be obtained by means of the so-called B' NMR coefficient^,^ as described below Of fundamental importance in equation 1 1 are the two quantities X,~and A* The former represents the molar fraction of solvent 1 in the ionic solvation sphere so that if x,~= x, (molar fraction of solvent 1 in the bulk) there will be no preferential solvation, while for x,, # x1 there is preferential solvation A*(xl)takes account of possible non-additivity in total symmetry quenching effects in simple terms it describes asymmetry effects due to the presence of two different components 2.4 Self-diffusion Coefficients So far we have only considered relaxation processes However, a further fundamental quantity to be considered is the self- diffusion coefficient Self-diffusion coefficients can be measured very accurately by using NMR techniques If a spin-system is perturbed by a radiofrequency (r f ) pulse, perpendicular components of the magnetization are produced These decay either with the time constant T2or due to magnetic field inhomogenities, whichever process is faster The latter decay is however reversible, so that the relevant component of the magnetization can be restored This gives rise to the so-called spin-echo,* which may be generated by a second restoring pulse at time T after the first perturbing pulse If such echoes are produced, their amplitude A decays as a function of the time T between the r f pulses with the spin-spin relaxation time T2 In the presence of large magnetic field inhomogenities, a second irreversible process occurs due to the self-diffusion of the molecules The latter effect can be magnified by producing a well-defined artificial linear magnetic field gradient, G,superim-posed on the static magnetic field This gradient can be applied at certain times between the r f pulses, as suggested by Stejskal and Tanner The pulse sequence used by Stejskal and Tanner is shown in Figure 1 In the presence of the two pulsed field gradients, the attenuation of the echo amplitude is given by the following ratio A(G)= A(0)exp -(y G 8)' (A -8/3)0 (12) In equation 12 A(0) is the amplitude of the echo observed without the field gradient, d and 6 are the parameters of the experiment, as defined in Figure 1 Thus, if G, A, and 6 are known, a measurement of the echo amplitude with and without application of the gradient pulses enables calculation of D Figure 1 The time sequence of pulses for pulsed gradient spin-echo diffusion measurements 3 Electrolytes in Pure Solvents The following discussion will be devoted to a method for obtaining information on microdynamical processes in solution, and adapted on a broad basis in our recent work At the end of this section we briefly consider some other recent work dealing with pure electrolytes solutions Hertz et a1 'published a fundamental paper concerning the study of microdynamics of aqueous electrolyte solutions using nuclear magnetic relaxation and self-diffusion data They intro- duced, in analogy to viscosity B-coefficients, NMR B'-coeffi- cients and showed that these coefficients contain information about the reorientational behaviour of solvent molecules in the first coordination sphere of ions Self-diffusion data were used in a similar way to study the influence of the dissolved ions on the translational behaviour of solvent molecules The viscosity B-coefficient obtained by means of the Jones- Dole equation 2=I +Bc+Cc'+ 70 provides information on the properties of electrolytes in solu- tion In analogy, Hertz et a1 introduced the following equation for the relative relaxation rate of solvent nuclei ($)I($)' = 1 + B'c + C'c2+ The NMR B'-coefficients were used to yield a microdynamical model of the electrolyte solution characterized by certain corre- lation times of solvent molecules We note that, as in the case of viscosities, the measured B'-coefficients may be separated into single-ion contributions Then, Hertz and co-workers show that from the B'-coefficients it is possible to estimate the factor k = .,IT," which represents the ratio between the reorientational or internal correlation times in the solvation sphere and the correlation time in the bulk solvent where yt, is the hydration number of the ion The results obtained for aqueous solutions indicated that typical structure-breaking ions show correlation times in their hydration spheres which are smaller than those of pure water In other words, the liquid within the hydration spheres of these ions is more fluid than in pure water Subsequently, Engel and HertzlO extended those measure- ments and data analysis to non-aqueous solutions, including methanol, ethanol, acetone, dimethyl sulfoxide (DMSO), for-mamide (FA), N-methyl formamide (NMF), ethylene glycol, and glycerol Except for solvents with two or more OH-groups, giving rise to extensive hydrogen-bonding, only positive values for B'-coefficients were found Again, there is the problem of subdividing these experimental quantities into single-ion contri- butions This can only be done using somewhat arbitrary assumptions based on plausible arguments rather than rigorous methods The resulting anion sequence is I-<Br-<C1-, while that for thecationsisCs+ <Rb+<K+<Na+<Li+ Nearly all single-ion values are positive Exceptions are the largest ions considered, Cs+ and I- in glycol and additionally +Rb and Br -in glycerol, where B' <0 are found So far, the measurements have been concerned with the total proton relaxation rate in organic solvents, which according to equation 2 consists of an intra- as well as an inter-molecular contribution Considering solutions in DMSO as representative examples, we have shown that the separate determination of these two contributions is possible In detail, we first measured the total proton relaxation rates After subdivision into single ion values a first important result was the observation that B'(Na+) M B'(K+) This is surprising, because in other dipolar aprotic solvents the B'-coefficients (eg the viscosity B-coeffi- cients) for alkali metal ions decrease as the ionic radius increases Therefore B'(Na+) > B'(K+) is expected On the other hand, these results confirm the viscosity measurements in DMSO by Feakins et a1 ,12 where an even slightly higher B+(K+)com-pared with B+(Na+) was found DMSO is not the first solvent where the relative position of Na+ and K+ has reversed, this pattern has also been observed for the corresponding B-coeffi- cients in NMF l3 Now let us consider the application of equation 2 in more detail If the solvent nuclei relax by dipolar interaction, as in the case of the DMSO protons, the total relaxation rate (l/Tl) has an intra- and an inter-molecular contribution.The correlation time for the intramolecular contribution is the reorientational correlation time of the vector connecting the interacting protons in the solvent molecule.On the other hand, the correlation time for the intermolecular contribution is proportional to D-',the inverse self-diffusion coefficient of the solvent m01ecules.~~'~ CHEMICAL SOCIETY REVIEWS, 1994 Hitherto, we were interested only in dipole4ipole relaxation of nuclei residing in solvent molecules. It is, however, worth- while to mention the possibility of monitoring dipole relaxation of solute nuclei, and of discussing the information gained from studies of this type. As is seen from equation 6, the relevant dipole-dipole energy depends on the distance between the interacting spins. Thus, it is possible to obtain precise infor- mation on the structure of solvation around an ion by measuring the interactions of the solute nucleus with well-defined interac- tion partners, for example at specific sites in solvent molecules.For this purpose it is convenient to apply the so-called isotopic substitution technique, which we use to single out interaction between specific molecule sites by deuteration of the remaining Thus (l/T,),,,,, is connected with the rotational and (l/T,),,,,, with the translational motion of the solvent molecules. If we are able to separate l/Tl into the two contributions, we can define in analogy to equation 14 new coefficients by: = 1 + Bl'ntrac+ .. . . (k)mtra/($)intra.c = 0 and These coefficients and Binterreflect the influence of the dissolved salt on the rotational and translational motion of the solvent respectively. However, there is another measurable quantity, namely the self-diffusion coefficient of the solvent molecule, which leads to the definition of BDcoefficients of the reciprocal self-diffusion coefficients by: = 1 + BDC +....Clearly, the two coefficients BD and B:nter,both representing the influence that the dissolved salts exert on the translational properties of the DMSO molecules, must be the same, i.e. BInter= BD. Then the intermolecular contribution at any salt concentration can be calculated: D, = 0 In our calculations for DMSO we use the value (I/Tl)intra,c=o = 0.17 s-obtained by the isotopic dilution method. Hence we obtain (l/Tl)inter,c= = 0.15 s-l. With the measured D values and equation 19 we obtain the results shown in Table 1. Table 1 BD and B;,,,, coefficients in DMSO at 25 "C Salt BD(= B:nter) BAra NaBr 0.178( f0.006) 0.178 f(0.009) KBr 0.174( * 0.003) 0.180 f(0.005) NaI 0.156( k0.003) 0.150 f(0.005) KI 0.153( f0.003) 0.155 f (0.005) RbI 0.140( f0.004) 0.140 f(0.009) CSI 0.134( f0.005) 0.130 f(0.008) Values in parentheses are standard errors.For DMSO, B' = B:,,,, = BDholds for all the salts investigated, which means that rotational and translational motion of the solvent DMSO is affected to the same extent by addition of a salt. Moreover, in molecules with internal flexibility, the intra- molecular relaxation rate does not only reflect molecular overall reorientation, but also internal rotations in the solvent mole- cules. This fact has been used by Ansari and Hertz16 to monitor the influence of ion-solvent interactions on the internal dyna- mics of solvent molecules, using ethanol as the representative example.For these purposes, they determined the B' coefficients of all non-equivalent protons in the ethanol molecule as a function of the concentration of added lithium halides. Apply- ing this procedure they found in the case of ethanol: B'(CH,) >B'(0H) >B'(CH,). parts of the molecule. In other words we 'switch on and off' the DD (dipole-dipole) interactions of the solute nuclei with the interacting partners. Hertz and Miiller' obtained information on the orientation of water molecules in the first hydration sphere of fluoride ions by comparing the 19F-'H interactions with the 19F-l 7Ointerac-tions, obtained from 9Frelaxation-time measurements in HzO, D,O, and D, 70.Surprisingly they found that the most favour- able configuration of the two possibilities in Figure 2 corres-ponds to the 'symmetrical' arrangement shown in (a).This result differs from what has been found in quantum mechanical, Monte Carlo, and molecular dynamics computations. On the other hand, the same technique applied to 7Li relaxation in LiCl aqueous solutions' gives results in complete agreement with theoretical predictions. Hence, there is no obvious reason why the experimental results for F-should be incorrect. Figure 2 Two possible orientations of the water molecules in the hydration sphere of the F-ion. (a) Symmetrical arrangement; (b) 'H-bonded' arrangement of water molecules.(Reproduced with permission from reference 17.) Finally, we draw attention to other important studies of electrolyte solutions in pure solvents, and above all in water, for illustrating the type of information that can be gained from magnetic relaxation experiments. Van der Maarel et al. studied proton, deuteron, and oxygen-17 relaxation rates of water molecules in solutions of alkali metal, magnesium, and zincZo salts. The nuclear quadrupolar coupling constant (NQCC) of zHand the 0-H distance in water were calculated. When salts are added, these quantities are remarkably changed relative to their values in pure water. Above all, there are indications from 2H and 7Orelaxation for anisotropic reorien- tation of water molecules in the solvation spheres of cations.With respect to the relaxation of nuclei residing in ions, more extensive and systematic investigations have been done. Most of this work has dealt with ionic nuclei of the alkali metal group. We consider here some illustrative examples using 7Li relaxation. All alkali metal nuclei posses quadrupole moments and therefore they relax predominantly by the quadrupolar mechan- ism. A notable exception is 7Li, where the quadrupole moment is low and the DD mechanism can compare with the quadrupole interaction. Different authorsi 8~z1have attempted to separate these contributions by investigating 'Li relaxation as a function of the HzO/DzOratio. While early work was limited to high salt concentrations in the molar range, we have extended these studies towards low concentrations, close to the Debye-Huckel STRUCTURE AND DYNAMICS OF ELECTROLYTE SOLUTIONS.A NMR RELAXATION APPROACH-A. SACCO 133 regime.,, Indeed we found that at very dilute concentration of LiCl salt the relaxation rate exhibits a +?law, which can be interpreted in terms of Debye-Hiickel ion-ion distribution. As already mentioned above, Mazitov et al.,Is have determined the orientation of water molecules in the hydration sphere of LiCl in 'Li by means of 'Li relaxation. 4 Electrolytes in Binary Mixtures Perhaps the most important problem in studying ion-solvent interactions in binary mixtures is the preferential solvation of ions; various techniques have been used to look at this.In the present context we make use of D,O-H,O dynamical isotopic effects on the magnetic relaxation by quadrupole interaction of ionic nuclei.23 The method until recently was used for studying only mixtures containing water as one of the components, but has now been generalized to treat phenomena in binary organic mixtures. As we have seen in Section 2.3, the general equation for the quadrupolar relaxation rate of the ionic nucleus in binary mixtures, i.e. equation 11, contains two unknown quantities xil and A*. A first approach assumes xil = x1and puts A*(x,) = 0. This implies that preferential solvation and any field gradient quenching effects are absent. Surprisingly, as can be observed in Figures 3 and 4, this model yields good agreement between the experimental and predicted relaxation rates for Na+ in metha- nol-water5 and Na+ in FA-and NMF-~ater,~~ which means that preferential solvation effects are largely absent in these systems. (1/TI) /s-' 60 50 40 30 20 10 I I I I I * 20 40 60 80 100 mol% MeOH Figure 3 23Na+ relaxation rates extrapolated to zero salt concentration in water-methanol(Me0H) mixtures (0)as a function of mol% MeOH.Dashed line, values calculated from equation 11 assuming xil = x1and A* = 0. (Reproduced with permission from reference 5.) When the investigations were extended to Na+ and Cs+ in water-acetonitrile (ACN),' and Na in water-hexamethyl- + phosphorous triamide (HMPT) mixtures,6 we observed dramatic differences between measured and predicted relaxation rates (Figures 5 and 6).It is intriguing to attribute these discrepancies to selective solvation and associated field gradient quenching. To obtain independent information on xil and A*(x,)we could apply, as mentioned before, measurements of the H,O-D20 isotopic effect on the relaxation rates of ionic FA lot. 1.0 0.8 0.6 0.4 0.2 0.0 Xl Figure 4 23Na relaxation rates extrapolated to zero salt concentration + in water-FA and water-NMF mixtures as a function of x,, the mole fraction of water. A, experimental values. 0,values calculated from equation 11 assuming x,,= x, and A*(x,)= 0. (Reproduced with permission from reference 24.) lot111111-0.2 0.4 0.6 0.8 1.0 XACN Figure 5 23Na+ relaxation rates for 1 m NaI in D,O-ACN (A) and H,O-ACN (0)mixtures as a function of the mole fraction of ACN, xACN.The dashed line represents the 'expected' relaxation rate curve according to equation 11 with xi, = x1 and A*@,) = 0.(Reproduced with permission from reference 25.) nuclei in the binary aqueous mixtures. In detail, separate measurements of 1/T, of the ionic nucleus were made in mixtures of organic solvent with H20,and subsequently with D20,which produced two sets of relaxation rates which deter- mined the two unknown quantities xi, and A*(xil)as a function of the macroscopic mole fraction x,. The results obtained in this + +way are displayed in Figure 7 for the systems Na and Cs in water-ACN and Na+ in water-HMPT. As is seen, in the case of ACN-H,O mixtures both ions are preferentially solvated by water.In fact the local mole fraction of water, xi],in the solvation sphere of ions is higher than the mole fraction x1 in the bulk. In the case of Na+ dissolved in I34 CHEMICAL SOCIETY REVIEWS, 1994 HMPT-H20, our result indicates that in the HMPT-rich region selective solvation by HMPT occurs The preference for HMPT is as expected from its high donor number and low acceptor number 26 Also, the Gibbs energy of transfer from water to HMPT2' dGP,(Na+)= -39 kJ mol indicates preferential solvation by HMPT Therefore, it was surprising that in the range 0 5 <xi <O 7 no preferential solvation or even small preference for water was found It is, however, not the first time that a change in the selectivity has been observed For DMF- water and DMSO-water such a change has been e~tablished~~ and ascribed to structural peculiarities in the binary solvent system An analogous investigation has been performed by us for Na+ and Cs+ in water-dimethylacetamide (DMA) mixtures 28 As seen in Figure 8, the microscopic mole fractions xI1for Na+ and Cs+ given as a function of the macroscopic mole fractions clearly reveal preferential solvation of these cations in water- DMA mixtures In the range 1 >x, >O 7 we observed weak 01 preferential hydration Below these concentrations a change in 0 02 04 06 08 10 the selectivity occurs, and for 0 7 >xi >0 we observed marked preferential solvation by DMA Interestingly, this change in XHMPT selectivity occurs near the mole fraction, where the reorientatio- Figure 6 23Na+ relaxation rates of 0 4 m NaBr in D,O-HMPT (A)and nal correlation time of water has its maximum, as also found for H,O-HMPT (0)mixtures as a function of the mole fraction of cations in DMF-water, DMSO-water, and HMPT-water HMPT, xHM~T All other details as in Figure 5 These maxima in reorientational correlation times appear to be (From reference 6) characteristic for organic co-solvents with more than one hydro- phobic methyl group 0 05 1002 04 06 08 10 Xl Xl 4 Figure 8 Local mole fraction x,,of water in the solvation sphere of Na + water in water DMA mixtures The straight line corresponds to the case of non-preferential solvation (From reference 28 ) Until recently the application of this technique had been limited to systems containing water as one of the components, because it had been assumed that the relatively large H,0-D20 dynamic isotopic effect of 23% at 25 "Cwas the only effect which could be exploited However, in recent separate studies on the isotope effects of non-aqueous solvents we found that in several cases there were surprisingly high effects as well, e g about 12% in methanol/methanol-d, and DMF/DMF-d, 29 30 We thought, therefore, that with a high measuring accuracy it would also be possible to use the method of the dynamic solvent isotope 02 04 06 08 10 effect to study electrolytes in binary non-aqueous mixtures Moreover, another important advantage turns out to result Xl from the fact that in suitable cases both components of the Figure 7 Local mole fraction x,, of water in the solvation sphere of Na+ solvent mixture may show dynamic isotope effects sufficiently (0)and Rb+ (A)plotted against x,, the macroscopic mole fraction of high for direct observation Then it is possible to check the water in water-ACN (a) and water-HMPT (b) mixtures The straight reliability of the method immediately by comparing the two line corresponds to the case of non-preferential solvation independent sets of results obtained (Reproduced with permission from references 6 and 25 ) In order to confirm this possibility we have studied the system NaI in methanol + DMF 31 The relaxation experiments for obtaining the local mole fractions (x,,) of methanol in the STRUCTURE AND DYNAMICS OF ELECTROLYTE SOLUTIONS A NMR RELAXATION APPROACH-A SACCO solvation sphere of Na+ have covered the following solvent mixtures CH,OH + DMF, CD,OD + DMF, and CH30H + DMF-d, The results for xll of CH,OH obtained from the data for CH,OH + DMF and CD,OD + DMF are reported in Figure 9 The local mole fraction of methanol around Na is always lower than the macroscopic value, which + leads to the conclusion that in the whole composition range Na + is preferentially solvated by DMF Subsequently, we performed ‘inverse’ experiments in mixtures of DMF + CH,OH and DMF-d, + CH,OH These yielded the local mole fractions x12 of DMF As can be seen from Figure 10, in the whole compo- sition range, x,, is higher than its macroscopic analogue x2, once more confirming preferential solvation by DMF Two entirely independent sets of experiments have thus led to the same conclusions 10.08 -06 -04 -02 -I I I I I) 02 04 06 08 10 Figure 9 Local mole fraction xtlof CH,OH in the solvation sphere of Na+ as a function of x,,the macroscopic mole fraction of CH,OH (From reference 31 ) 10-08-06-04-02-I 1 I I I 0 02 04 06 08 10 x2 (DM F) Figure 10 Local mole fraction xI2of DMF in the solvation sphere of Na as a function of x2,the macroscopic mole fraction of DMF + (From reference 31 ) In more quantitative terms, the values for xll and x,, can be used to check the consistency of the results (and the quality of the method) by noting that x,, + x,, = 1 must be valid In Table 2 we report the resulting numerical values of x,, and x12as a function of composition, taken from the smoothed curves in Figures 9 and 10 In a separate column of Table 2 the sum Table 2 Local mole fraction values in function of mole fraction of methanol in methanol-DMF mixtures x,(CH,OH) X,l(CH,OH) xt,(DMF) (XI1 + X/t) 1 09 08 07 06 05 04 03 02 01 0 1 0 1 0 83 0 12 0 95 0 71 0 23 0 94 0 60 0 35 0 95 0 52 0 47 0 99 0 42 0 58 1 00 0 34 0 69 103 0 25 0 79 1 04 0 16 0 89 1 05 0 08 0 96 1 04 0 1 1 (xI1+ x,,) is shown the experimental values of (xll + x,,) agree within 5% with the theoretical value of unity The remaining small deviations are explained by the experimental error and by certain approximations of minor importance entering in the equations The results show convincingly that this method provides a powerful tool for investigating preferential solvation of relatively weakly solvated monovalent ions like Na+ In fact, comparatively strong dynamic isotope effects now found in various organic solvents open the possibility of quantitative investigations in a great number of non-aqueous binary mixtures Finally, in the context of selective solvation studies, we briefly review other relaxation methods Finter and Hertz3 studied preferential ion solvation in aqueous mixtures of FA and NMF Their method was based upon a separation of the dipole-dipole contributions of ionic nuclei to the relaxation rate of the formyl or amide protons of FA, and of formyl, amide, and methyl protons of NMF, respectively They showed that over the complete composition range of NaI in aqueous mixtures of FA, preferential solvation of Na + by FA occurs, while I -is preferen- tially hydrated 32 The observed effects were termed ‘hetero- selectivity’, because the cation and anion prefer different compo- nents of the binary mixture According to these authors, hetero- selectivity occurs more often than homo-selectivity In the case of NaI dissolved in water + NMF,l3 analogous results were obtained, but were subject to larger uncertainty Acknowledgements I should like to thank H Weingartner for many helpful comments I am also grateful to the Ministry of University and of Scientific and Technological Research (MURST), Italy, for financial support 5 References 1 W E Waghorne, Chem Soc Rev, 1993,22,285 2 S Taniewska-Osinska, Chem Soc Rev, 1993,21,205 3 See e g H G Hertz, A Kratochwill, and H Weingartner, ‘Nuclear Magnetic Relaxation and Intermolecular Interactions’, in ‘Studies of Physical and Theoretical Chemistry’, Vol 37, ed W J Orville-Thomas, H Ratajczak, and C N R Rao, Elsevier, Amsterdam, 1985 4 H G Hertz, Ber Bunsenges Phys Chem , 1973,77,531 and 688 5 M Holz, H Weingartner, and H G Hertz, J Chem Soc Faraday Trans I, 1977,73, 71 6 M Holz and A Sacco, Mof Phys , 1985,54, 149 7 L Endom, H G Hertz, B Thul, and M D Zeidler, Ber Bunsenges Phys Chem , 1967,71, 1008 8 T C Farrar and E D Becker, in ‘Pulse and Fourier Transform NMR’, Academic Press, New York, 1971 9 E 0 Stejskal and J E Tanner, J Chem Phys , 1965,42,288 10 G Engel and H G Hertz, Ber Bunsenges Phys Chem , 1968,72, 808 11 A Sacco, M Carbonara, and M Holz, J Chem Soc Faraday Trans 1, 1989,85, 1257 12 R T M Bicknell, K G Lawrence,andD Feakins, J Chem Soc Faraday Trans I, 1980,76,637 13 C.K. Finter and H. G. Hertz, J. Chem. Soc., Faraday Trans. I, 1988, 84,2735. 14 H. G. Hertz, in ‘The Chemical Physics of Solvation’, ed. R. R. Dogonadze, E. Kalman, A. A. Kornyshev, and J. Ulstrup, Part B, Elsevier, Amsterdam, 1986, p. 31 1. 15 M. D. Zeidler, Ber. Bunsenges. Phys. Chem., 1965,69,659. 16 M. S. Ansari and H. G. Hertz, J. Solution Chem., 1984,13, 877. 17 K. J. Muller and H. G. Hertz, 2.Phys. Chem. N. F., 1984, 140, 31. 18 R. Mazitov, K. J. Muller, and H. G. Hertz, Z. Phys. Chem. N. F., 1984,140, 55. 19 J. R. C. Van der Maarel, D. Lankhorst, J. de Bleijser, and J. C. Leyte, J. Phys. Chem., 1986,90, 1470. 20 J. R. C. Van der Maarel, J. Magn. Reson., 1989,81, 92. 21 H. G. Hertz, R. Tustsch, and H. Versmold, Ber. Bunsenges. Phys. Chem., 1971,75, 1177. 22 M. Holz and A. Sacco, Z. Phys. Chem. N. F., 1987,155, 133. CHEMICAL SOCIETY REVIEWS, 1994 23 M. Holz, J. Chem. Soc., Faraday Trans. I, 1978,74,644. 24 M. Holz and C. K. Rau, J. Chem. Soc., Faraday Trans. I, 1982,78, 1899. 25 B. M. Braun and M. Holz, J. Solution Chem., 1983, 12, 685. 26 U. Mayer, in ‘Ions and Molecules in Solution’, ed. N. Tanaka and H. Ohtaki, 1983. 27 A. J. Parker, Pure Appl. Chem., 1981,53, 1437. 28 M. Holz, A. Sacco, and M. Trotta, J. Solution Chem., 1990,19, 193. 29 H. Weingartner, M. Holz, A. Sacco, and M. Trotta, J. Chem. Phys., 1989,91,2568. 30 M. Holz, H. Weingartner, and A. Sacco, Ber. Bunsenges. Phys. Chem., 1990,94,332. 31 A. Sacco, M. C.Piccinni, and M. Holz, J. Solution Chem., 1992,21, 109. 32 C. K. Finter and H. G. Hertz, 2.Phys. Chem. N. F., 1986,148,75.

ISSN:0306-0012    

DOI:10.1039/CS9942300129

出版商:RSC    

年代:1994

数据来源: RSC