D bethell (ed ) advances in physical organic chemistry, vol 29 academic press (1994)

The Stabilization of Transition States by Cyclodextrins and other Catalysts 3 Transition state stabilization 9 The Kurz approach 9 Cyclodextrin mediated reactions 11 do all this at r

Advances in Physical Organic Chemistry

www.pdfgrip.com

ADVISORY BOARD

W J Albery, FRS

A L J Beckwith

R Breslow

L Eberson Chemical Center, Lund

H Iwamura University of Tokyo

University of Southern California, Los Angeles The Hebrew University of Jerusalem

Universitat Erlangen-Nurnberg University of Illinois at Urbana-Champaign

Harcourt Brace & Company, Publishers

London San Diego New York

Boston Sydney Tokyo Toronto

www.pdfgrip.com

N o part of this book may be reproduced in any form by photostat, microfilm,

or any other means, without written permission from the publishers

A catalogue record for this book is available from the British Library ISBN 0-12-033529-8

ISSN 0065-3160

Printed and bound in Great Britain by

Hartnolls Ltd, Bodmin, Cornwall

Contents

Preface

Contributors to Volume 29

The Stabilization of Transition States by Cyclodextrins

and other Catalysts

4 Extrapolation to transition state structures 173

Electron Transfer in the Thermal and Photochemical Activation

of Electron Donor-Acceptor Complexes in Organic

CONTENTS

Time-resolved picosecond spectroscopic studies of

charge-transfer complexes 190

Variable charge-transfer structures of nitrosonium-EDA

complexes leading to thermal and photo-induced electron

Cumulative Index of Authors

Cumulative Index of Titles

273

333

35 1

353

With ever increasing specialization among chemists, there is a continuing need to ensure that research in one area is not hampered by lack of awareness of developments in contiguous areas, expressed in language that

is understood by both groups of specialists Over the thirty years of its

existence, such bridge-building has been a consistent aim of Advances in Physical Organic Chemistry in relation to the physical and organic chemical communities, and a considerable debt is owed to the many contributors who have striven to present their material in an attractive and comprehensible way More recently the series has sought to reflect the relevance of the physical organic approach to developments in the field of new materials and,

in an as yet small but it is hoped increasing way, in the burgeoning realm of bio-organic research The Editor and his Advisory Board continue to encourage comments on the series, suggestions of topics that are worthy of coverage in future volumes, and, perhaps best of all, offers to contribute articles on any aspect of the quantitative study of organic compounds and their reactions

D BETHELL

vii

www.pdfgrip.com

The Stabilization of Transition States by

Cyclodextrins and other Catalysts

3 Transition state stabilization 9

The Kurz approach 9

Cyclodextrin mediated reactions 11

do all this at rates that are 106-10” times faster than the uncatalysed reaction The origins of these impressive feats almost certainly lie in supramolecular behaviour (Lehn, 1985, 1988) since enzymes invariably form

1

ADVANCES IN PHYSICAL ORGANIC CHEMISTRY Copvrrghr 0 IYY4 Academic Press I.imrwd

www.pdfgrip.com

2 0 S TEE

enzyme substrate complexes from which the catalysed reactions ensue Many static and dynamic studies of enzyme behaviour have provided ample evidence of such complexes and great progress has been made in elucidating many of the mechanisms by which enzymes transform their substrates into products (Walsh, 1979; Fersht, 1985; Page and Williams, 1987; Liebman and Greenberg, 1988; Dugas, 1989) At the same time, there have been significant advances in understanding the factors underlying the catalytic abilities of enzymes (Jencks, 1969, 1975; Bender, 1971; Lienhard, 1973; Gandour and Schowen, 1978; Page, 1984; Fersht, 1985), although at times it

has seemed as though there were too many theories of enzymatic catalysis,

based on the multiplicity of ideas about the efficiency of intramolecular processes (Page, 1984, 1987; Menger, 1985; Page and Jencks, 1987)! The underlying principle of enzyme catalysis was expounded many years ago by Haldane (1930) and Pauling (1946) According to them, catalysis results from stabilization by the enzyme of the reaction transition state, relative to that of the initial state This view was developed by Kurz (1963) into a quantitative approach to transition state binding, and hence of transition state stabilization, albeit in the context of catalysis by acids and bases (Kurz, 1963, 1972) His approach was taken up and used by enzymologists (Wolfenden, 1972; Lienhard, 1973; Jencks, 1975; Schowen, 1978; Fersht, 1985; Kraut, 1988), so much so that it is now implicit in many

modern studies of enzyme action (see, for example: Fersht et al., 1986, 1987; Leatherbarrow and Fersht, 1987) Of particular note, Kurz’s innovation helped to develop the use of “transition state analogues” (Jencks, 1969) as efficient enzyme inhibitors, either for the purposes of mechanistic studies or for possible pharmaceutical use (Wolfenden, 1972; Wolfenden and Frick, 1987; Wolfenden and Kati, 1991) In turn, the availability of transition state analogues as haptens has been critical to the recent development of

“catalytic antibodies” (Schultz, 1988, 1989a,b)

The fascination of chemists with enzymes has led, in recent years, to many attempts to model or mimic their action (e.g Bender, 1971, 1987; Breslow,

1982, 1986a,b; Page, 1984; Tagaki and Ogino, 1985; Kirby, 1987; Stoddart, 1987; Schultz, 1988, 1989a,b; Dugas, 1989; Chin, 1991) The object of such studies has been to understand enzyme action and, in a broader sense, catalysis better, and possibly to learn how to synthesize catalysts (“artificial enzymes”) for specific purposes (Breslow, 1982; Schultz, 1988) Many such studies have employed model systems based on the binding and catalytic

properties of cyclodextrins (CDs) or their derivatives (Bender and

Komiyama, 1978; Breslow, 1980, 1982, 1986a,b; Tabushi, 1982; Komiyama and Bender, 1984; Bender, 1987; D’Souza and Bender, 1987; Tee, 1989) At the same time, CDs have commanded another, more practical and populous audience due to their many potential applications in the food, pharmaceutic-

al, and cosmetic industries (Szejtli, 1982; Pagington, 1987) These differing interests in the chemistry of CDs have led to an explosion in the literature

TRANSITION STATE STABILIZATION 3

concerning these molecules in recent years, especially now that they are produced commercially and are available relatively cheaply

The present review deals with a particular aspect of the chemistry of cyclodextrins: the effects that they can have on organic reactions by virtue of their abilities to bind to many organic and inorganic species (Bender and Komiyama, 1978; Saenger, 1980; Szejtli, 1982) It is a considerable expansion of an earlier work (Tee, 1989) which first showed how the Kurz approach to transition state stabilization can be employed profitably in discussing reactions mediated by cyclodextrins Most of the large amount of data that are analysed is collected in tables in the Appendix so as to avoid breaking up the discussion in the main text too frequently

While the main emphasis of this review is on catalysis, since this is of greater interest, the Kurz method can also be applied to retardation In fact, for some of the systems discussed later, the smooth transition from retardation, through inactivity, to full catalysis can be quantified and analysed in relation to the structure of the species concerned

At the end of the review there are some examples involving catalysis by acids and bases, metal ions, micelles, amylose, catalytic antibodies, and enzymes to give the reader a feeling for how Kurz’s approach may be usefully applied to other catalysts Very few of these examples, or those involving cyclodextrins, were discussed in the original literature in the same terms It is hoped that the present treatment will stimulate further use and exploration of the Kurz approach to analysing transition state stabilization

2 Cyclodextrins

These water-soluble molecules are cyclic oligomers of a-D-glucose formed

by the action of certain bacterial amylases on starches (Bender and Komiyama, 1978; Saenger, 1980; Szejtli, 1982) a-Cyclodextrin (cyclohexa- amylose) has six glucose units joined a(1,4) in a torus [l], whereas p-cyclodextrin (cycloheptaamylose) and y-cyclodextrin (cyclooctaamylose) have seven and eight units, respectively

The form of cyclodextrins (CDs) is variously described as being “conical”,

“toroidal”, “bucket shaped”, or “doughnut shaped” [2] Regardless of the adjective used and the finer details of their structure, the most important feature of CDs is the cavity, because this enables them to form inclusion complexes in solution and in the solid state By virtue of their cavities, CDs possess the requisite amount of preorganization and the convergent surfaces (Cram, 1983, 1988) necessary for them to function as hosts for small molecular guests of an appropriate size, shape, and polarity The depths of

CD cavities are all the same (approximately 7.5 A), being determined by the width of a glucose molecule, but the sizes of their cavities differ in diameter (a-CD about 5.0, p-CD about 7.0 and y-CD about 9 0 A ) (Bender and

www.pdfgrip.com

1980; Szejtli, 1982; Atwood et af., 1984) Nevertheless, CDs have been

labelled “promiscuous” for their propensity to act as hosts to a wide variety

of small- to medium-sized guests (Stoddart and Zarzycki, 1988) It is the ability of CDs to form complexes that enables them to influence chemical reactions through supramolecular effects (Sirlin, 1984; Lehn, 1985, 1988) In

what follows, some of the basic aspects of C D binding, relevant to the reactions discussed later, are presented More detailed discussions of CD inclusion complexes can be found in the references already cited

Broadly speaking, the cavity sizes of a-, p-, and y-CD are appropriate for binding simple derivatives of benzene, naphthalene, and anthracene, respectively (Sanemasa and Akamine, 1987; Fujiki et al., 1988; Sanemasa et af., 1989) Many studies of the inclusion of aromatics, particularly of dyes

and other molecules with strong chromophores, have been reported, and these have been useful in delineating the main features of C D binding (Bender and Komiyama, 1978; Saenger, 1980; Szejtli, 1982; Atwood et al.,

1984; Stoddart and Zarzycki, 1988) In contrast, the affinity of small to medium aliphatic molecules for CDs have been less well studied, most

TRANSITION STATE STABILIZATION 5

probably for practical reasons Nevertheless, there have been studies with

various surfactants (On0 et al., 1979; Satake et al., 1985, 1986; Diaz et al.,

1988; Palepu and Reinsborough, 1988; Palepu et al., 1989), alkanes (Sanemasa et al., 1990), and a particularly interesting study of the binding of

many alcohols to both a- and p-CD (Matsui and Mochida, 1979; see also,

Matsui et al., 1985; Fujiwara et al., 1987)

For the most part, CDs form simple 1 : 1 host-guest complexes with suitable guests But it is important to note that 2 : 1 binding can be significant with longer aliphatics (Palepu and Reinsborough, 1988; Palepu et a l , 1989; Sanemasa et al., 1990), aromatics (Sanemasa and Akamine, 1987; Fujiki et

al., 1988), azo dyes (Bender and Komiyama, 1978; Szejtli, 1982), and

aryl-alkyl guests (Tee and Du, 1988, 1992), and this can influence reactivity Also, there is now evidence of 1 : 1 1 binding of C D + two guests (Hamai, 1989a,b) which has been implicated in some reactions (Ramamurthy, 1986; Tee and Bozzi, 1990)

The ability of a C D to form inclusion complexes in aqueous solution results from its cavity, the interior of which is less polar than water and hydrophobic The apparent polarity of the C D cavity seems to depend on the probe used Some studies have suggested a similarity to dioxane (Bender and Komiyama, 1978; Hamai, 1982), while others favour ethanol (Cox et

al., 1984; Heredia et al., 1985) No doubt the particular observations are affected by the presence or absence of specific interactions, such as

hydrogen bonding, between the guest and the CD host, as well as by the depth of penetration of the guest/probe Decarboxylation studies, to be discussed more fully later, suggest an environment like 50% aqueous 2-propanol (Straub and Bender, 1972a,b)

Various other factors have been cited (Bender and Komiyama, 1978; Szejtli, 1982) as contributing to the binding ability of CDs However, the principal factors seem to be the hydrophobicity of the guest and the appropriateness of its size and shape in relation to that of the C D cavity (Tabushi, 1982) These factors are evident in the binding of alcohols to CDs

(Matsui et al., 1985) and of other guests with alkyl groups (Tee, 1989; Tee et

al 1990b) For illustrative purposes, and because of its relevance to a later section, the binding of alcohols will be discussed in some detail

For linear, primary alcohols (n-alkanols) the strength of complexation with CDs, expressed by pKs = -logKs, where Ks is the dissociation constant of the complex, correlates strongly with their coefficients for

partition ( P , ) between diethyl ether and water (Matsui and Mochida, 1979; Matsui et al., 1985), with slopes close to 1 ( l a and l b ) It has also been

a-CD: pKs = 0.91 log P, + 1.25; r = 0.994 ( l a ) P-CD: pKs = 0.94 log P , + 0.58; r = 0.994 ( l b ) noted (Tee, 1989; Tee et al., 1990b) that for these alcohols, and other linear

www.pdfgrip.com

V 2-alkanols

+ tertiary

pK, (a-CD)

Fig 1 Correlation between the binding of aliphatic alcohols to p-CD and to a-CD:

(-) the least-squares line for n-alkanols; ( ) pKs (p-CD) = pK, (a-CD); above this line a given alcohol binds strongly to p-CD than to a-CD Data

from Matsui and Mochida (1979) and Matsui et al (1985)

aliphatics, pKs values vary linearly with N , the number of carbon atoms in the chain These observations are reasonable since, as remarked above, the binding of guests to CDs is largely governed by their size and hydrophobicity (Tabushi, 1982) Obviously, the sizes of extended n-alkyl chains increase linearly with N , but so also do various measures of hydrophobicity, such as the logarithms of partition coefficients, critical micelle concentrations,

solubilities (Hansch, 1971; Leo et al., 1971; Hansch and Leo, 1979; Tanford, 1980; Menger and Venkataram, 1986)

Equations (la) and (lb) represent two nearly parallel lines with a vertical difference of about 0.7, indicating that a given linear alcohol binds about five times more tightly to a-CD than to p-CD This makes sense in terms of

the sizes of the a- and p-CD cavities (about 5 and about 7 & respectively)

in relation to the cross-section of methylene chains (about 4.5 A) (Sanemasa

et al., 1990) With bulkier types of alcohols (secondary, tertiary, cyclic, and branched) there is a tendency towards stronger binding in the larger cavity

correlation ( r = 0.9991) with a slope of 1.10 Other, bulkier alcohols deviate

above this line, showing the tendency to a stronger affinity with p-CD

Points for the bulkiest alcohols (branched, tertiary, cyclic >C,) lie above the dashed line corresponding to pKs (p-CD) = pKs (a-CD), since such alcohols are bound more strongly by p-CD (Fig 1)

One other feature of CDs is relevant to later discussion: the acidity of their secondary hydroxyl groups, with pK, values about 12.2 (VanEtten et

TRANSITION STATE STABILIZATION 7

al., 1967b; Gelb et al., 1980, 1982) The conjugate anions may function as

nucleophiles or general bases and react with substrates included in the CD cavity (Bender and Komiyama, 1978; Komiyama and Inoue, 1980c; Daffe and Fastrez, 1983; Cheng et al., 1985; Tee, 1989; Tee et al., 1993a)

By virtue of their complexing ability, CDs may influence the course of chemical reactions in respect of rates and/or product selectivity In conse- quence, there is a large body of data in the literature on the effect of CDs

on many types of reactions (Fendler and Fendler, 1975; Bender and Komiyama, 1978; Szejtli, 1982; Tabushi, 1982; Sirlin, 1984; Ramamurthy, 1986; Ramamurthy and Eaton, 1988) The present review concentrates on reactions for which sufficient kinetic data are available to allow quantifica- tion of the effects of CDs on transition state stability, in an attempt to understand how cyclodextrins influence reactivity in either a positive or negative sense

EFFECTS ON REACTIVITY

The kinetics of reactions which are influenced in a simple way by CDs may

be viewed in the following manner (Bender and Komiyama, 1978; Szejtli,

1982; Tee and Takasaki, 1985) For a substrate S that undergoes an

“uncatalysed” reaction (2) in a given medium and a “catalysed” reaction

through a 1 : l substrate/CD complex (3), the expected variation of the

observed rate constant with [CD] is given by (4)

constant, k,, and the dissociation constant, K s (VanEtten et al., 1967a;

Bender and Komiyama, 1978; Szejtli, 1982; Sirlin, 1984; Tee and Takasaki,

1985) The rate constant ku is normally determined directly (at zero [CD]),

and sometimes Ks can be corroborated by other means (Connors, 1987) Traditionally, data corresponding to (4) are analysed by using a Lineweaver-Burk approach, but an Eadie-Hofstee treatment is preferable for statistical reasons (Dowd and Riggs, 1965; VanEtten et al., 1967a;

Bender and Komiyama, 1978) With the present, widespread availability of

www.pdfgrip.com

8 0 S TEE

cheap microcomputers and appropriate software, it is now feasible to analyse data more directly in terms of (4), using non-linear least-squares fitting techniques (Bevington, 1969; Leatherbarrow, 1990; Duggleby, 1991)

In our own work, we have settled on this last approach, usually keeping k , fixed at the measured value, and treating k, and Ks as the constants to be

fitted (Tee and Takasaki, 1985) Using such non-linear fitting gives a more consistent approach to data analysis, particularly when one has to use

expressions more complex than (4), because of additional processes such as

non-productive 2: 1 binding or reactions with a second CD molecule (Tee and Du, 1988, 1992)

Generally speaking, discussions of the effects of CDs on reaction rates are given in terms of k,lk,, K s , and, sometimes, k,lKs Most often, the ratio

k,lk, is emphasized since this quantity measures the maximal rate accelera- tion (or retardation) due to binding to the CD Obviously, Ks measures the strength of binding of S to CD, but it conveys no information whatsoever about the mediation of the reaction by the CD or the mode of binding in the transition state which may be very different from that of the substrate (Tee,

1989; Tee et al., 1990b) Sometimes use is made of the apparent second order rate constant for the reaction of the substrate with the CD ( 5 ) , where

k i

S+CD-P

k2 = k , / K s ( 3 ) , since this rate constant measures the selectivity of the CD for different substrates This usage is analogous to the use of kcat/KM for measuring the “specificity” of enzymes (Fersht, 1985) In cases of catalysis where saturation kinetics are not observed, because binding of the substrate

to the CD is weak and K s is relatively large, k2 may be obtainable from the

linear increase of kobsd with [CD]

Provided due attention is paid to the potential deprotonation of the

substrate, and of the cyclodextrins (VanEtten et af., 1967a,b; Gelb et af.,

1980, 1982; Tee and Takasaki, 1985), the value of Ks should not be pH dependent However, for many reactions, such as the widely studied ester cleavage, k , , k , , and k2 are all dependent on the pH of the medium This

makes direct comparisons between the observed constants for different CD-mediated reactions either difficult or problematical However, in general, the ratios k,lk, and k21k, are independent of pH and so are more useful for comparative purposes

As remarked already, k,lk, measures the maximal acceleration at levels of

the CD sufficient to saturate complexation of the substrate By looking carefully at the variations of this ratio with structure one may obtain insights into the mode of transition state binding (VanEtten et a l , 1967a,b; Bender

and Komiyama, 1978) More useful is the ratio k21k, ( = k c / K s k , ) because it takes into account the effect of substrate binding and it scales the reactivity

of S towards the CD to its intrinsic reactivity in the absence of CD

TRANSITION STATE STABILIZATION 9

Enzymologists have used the analogous ratio k,,,lKM k , in full realization of

its significance and usefulness (Wolfenden and Kati, 1991) However, k,lk,

has been used only occasionally by chemists (Sirlin, 1984; Tee and Takasaki, 1985) without realizing that the ratio, or rather its reciprocal (k,lkl = K r s ) ,

has another, much greater significance The utility of K,, is the main focus

of this review; its significance will be made apparent in the next section

3 Transition state stabilization

Following on from the early ideas of Haldane (1930) and Pauling (1946), it has become widely accepted that the principal factor in enzymic catalysis is stabilization of the reaction transition state by binding to the enzyme (Jencks, 1969, 1975; Lienhard, 1973; Schowen, 1978; Page, 1984; Fersht, 1985) Likewise, lowering of the free energy of the transition state must be crucial in catalysis by other agents Therefore, any method that can provide quantitative information about the strength of such stabilization has great potential for use in the study of catalysis, whether it be enzymic or non-enzymic Application of the method to different substrates and catalysts might furnish insight into the nature of the catalysis involved and, in particular, into the manner in which catalysts bind to transition states and thereby stabilize them

Thirty years ago, Kurz (1963) devised a very simple method, based on transition state theory, whereby the energy of stabilization of transition states by catalysts may be estimated He used the method to probe the transition states of acid- and base-catalysed reactions, and developed the idea of transition state pK, values (Kurz, 1972) The approach was taken up

by enzymologists (Wolfenden, 1972; Lienhard, 1973; Jencks, 1975; Schowen, 1978; Kraut, 1988), and it proved to be very influential in the formulation of the ideas about enzyme catalysis referred to in the previous paragraph and in the Introduction It is, therefore, surprising that the Kurz method has been ignored by most physical organic (and inorganic) chemists studying the mechanisms of catalysed reactions Very recently, however, essentially the same method has been applied to organic reactions catalysed

by metal ions (Dunn and Buncel, 1989; Pregel et al., 1990; Ercolani and

Mandolini, 1990; Cacciapaglia et af., 1989, 1992), and the present author has

shown how the Kurz approach can be used in discussions of reactions mediated by cyclodextrins (Tee, 1989; Tee et al., 1990b; Tee and Du, 1992)

THE KURZ APPROACH

Consider two reactions, one of which is “uncatalysed” (6a) and the other

of which (6b) is influenced by some “catalyst”, cat According to simple

www.pdfgrip.com

10 0 S TEE

transition state theory (Glasstone et al., 1941; Laidler, 1987), the rate constant for the uncatalysed reaction is given by (7a), and that for the catalysed reaction by (7b), where v = kBT/h, and the transition state of the catalysed process (6b) is considered for mathematical and thermodynamic purposes to be that of reaction (6a) bound to the catalyst (TSecat) It is assumed that the average frequency of passage over the barrier (v) is the

same for (7a) and (7b), and that the transmission coefficients are equal for the two processes Kraut (1988) considers the possible consequences when these assumptions are relaxed

simple definition (8) of an apparent constant for the dissociation of TS.cat

into TS and catalyst Obviously, KTS is a quasi-equilibrium constant, since

It is important to note that the derivation of KTS, given above, involves

no ussumptions about the mechanisms of either the catalysed or uncatalysed

reactions Therefore, it is possible to use values of K,rs (and pKPrs = -log KTs) and their variations with substrate or catalyst structure (or some other reaction parameter) as probes of transition state structure (Kurz, 1972; Tee, 1989) Clearly, complications may arise when the mechanisms of the catalysed and uncatalysed reactions are quite different, but under such circumstances one can reasonably hope that trends in KPrs and other kinetic parameters may be such as to point to the discrepancy and that they may even suggest a resolution

It is not the purpose of the present review to give a critical appraisal of the Kurz approach; that can be found in the review by Kraut (1988) Rather, it is to show how this simple method can be used in the study of reactions influenced by cyclodextrins Some examples involving catalysed

TRANSITION STATE STABILIZATION 11

reactions of other types which may be of interest to a wider audience of physical organic chemists are also presented

CYCLODEXTRIN MEDIATED REACTIONS

Application of the Kurz approach to CD-mediated reactions, whether they

be accelerated or retarded, is straightforward (Tee, 1989), provided

appropriate kinetic data are available From the rate constants k , for the

normal, “uncatalysed” reaction (2) and for the mediated (“catalysed”) reaction ( k 2 = k , / K s ) as in (3), application of simple transition state theory,

in the manner shown above, leads to (9), where now KTs is the apparent

dissociation constant of the transition state of the CD-mediated reaction

(symbolized here as T S - C D ) into the transition state of the normal reaction

(TS) and the CD This constant and its logarithm, which is proportional to a free energy difference, is a valuable probe of the kinetic effects of CDs on reactions

As outlined in Section 2, discussions of catalysis (or inhibition) by CDs

are generally in terms of k,lk,, K s , and, to a lesser extent, k2 = k , / K s This

last quantity has the same usefulness (and significance) as does k C a t / K M for enzymes (Fersht, 1985) in that it is a measure of the substrate selectivity of

the CD (VanEtten et al., 1967b; Tee and Takasaki, 1985) With proteolytic

enzymes such as a-chymotrypsin, there is no major problem with the use of

k,,,lKM since the peptide bonds formed between different amino acids have

fairly similar intrinsic reactivities ( k , ) (Berezin et al., 1971; Dorovska et a l ,

1972; Fersht, 1985), but comparisons between substrates having quite different reactivities require some kind of scaling, and this can be achieved

by looking at k21k, As remarked already, such ratios have occasionally

been used (Sirlin, 1984; Tee and Takasaki, 1985), but it was not recognized

at the time that k21k, is simply the reciprocal of K r s , as seen in (9)

While purists of thermodynamics may cavil that KTS is not a true equilibrium constant, it does correspond to an energy of great interest and importance: the free energy difference between the transition states of the uncatalysed and catalysed reactions [(2) and (3), respectively] under

standard conditions Alternatively, one may prefer to consider this differ- ence as the free energy of transfer of the transition state from aqueous solution to a 1 M solution of the catalyst, as has been done recently (Dunn

and Buncel, 1989; Pregel et al., 1990) Whatever the case, the significance of

Kvrs can most easily be appreciated by consideration of the Gibbs energy

diagram in Fig 2 As indicated there, the relative free energies of various

species involved in reactions (2) and (3) are directly accessible from the

www.pdfgrip.com

Fig 2 Relative Gibbs energies for the species involved in a reaction which is

uncatalysed (S -+ TS + P) and mediated by a catalyst (S + cat * TS cat + P) For a

specified [cat] the free energy differences can be directly calculated from the

measurable constants k , , k, and K s , and the derived values k2 and KTs, as indicated

pKTs = -logKTs is a measure of the stabilization of the transition state by the

catalyst

measurable quantities k , , k , , and K s (or k , and k 2 ) As long as these

constants are all measured under the same conditions, the apparent

“equilibrium constant” KTS (through its logarithm) gives a direct measure of the binding energy of the transition state to the catalyst for those conditions,

regardless of the mechanism (Schowen, 1978)

The diagram in Fig 2 also serves to emphasize that stabilization of the transition state by the catalyst is primarily responsible for any rate increase

To a considerable extent the binding of S is irrelevant, except that strong substrate binding necessarily detracts from catalysis In fact, according to (9), the change in rate is determined by the strength of binding of TS, relative to that of S (k,lk, = K s / K T s ) (Lienhard, 1973) This emphasis has been termed the “fundamentalist view” by Schowen (1978) A much more

agnostic view of the importance of transition state stabilization has recently been presented by Menger (1992)

Obviously, strong binding of the substrate to the catalyst may distort the

structure of S towards that of TS, thereby making reaction easier However,

such distortion simply reflects the complementarity of the catalyst and the transition state (Fersht, 1985) From a purely thermodynamic point of view, the formation of a strong S.catalyst complex lowers free energy by an

additional amount that must be overcome in the process of activation of the

k , process (3) (Fig 2) Living organisms and their enzymes have evolved so

TRANSITION STATE STABILIZATION 13

as largely to avoid this problem by having working levels of [S] close to K , ;

thus the free energy difference between S + enzyme and S.enzyme is quite small and the cost in free energy is minimal (Jencks, 1969; Lienhard, 1973; Fersht, 1985)

As pointed out above, values of KTS are obtainable from rate data

without making any assumptions about the reaction mechanism Therefore,

one may use Kr.7 and its variation with structure as a criterion of mechanism,

in the same way that physical organic chemists use variations in other kinetic parameters ( B r ~ n s t e d plots, Hammett plots, etc.) For present purposes, the value of KTS can be useful for differentiating between the modes of binding

in the S C D complex and the TS C D transition state, between different

modes of transition state binding, and hence between different types of catalysis (Tee, 1989)

According to Bender and Komiyama (1978), CDs may show two basic forms of catalysis: “non-covalent” and “covalent” In the former case the

C D binds to the substrate(s) and provides an environment for the reaction that is different from the bulk solvent, whereas in the latter case there are also distinct covalent interactions between the substrate(s) and some functional group(s) on the C D in the rate-limiting step of the reaction Therefore, it seems reasonable to expect that values of KTS for these two types of catalysis may show different sensitivities to structural change, since the partial bonding involved in covalent catalysis will normally lead to stronger interactions with the CD and possibly to more stringent geometric requirements than non-covalent catalysis

4 Non-covalent catalysis

In this form of catalysis, inclusion of the substrate in the C D cavity provides

an environment for the reaction that is different from that of the bulk, normally aqueous, medium In the traditional view, the catalytic effect arises from the less polar nature of the cavity (a microdielectric effect) and/or from the conformational restraints imposed on the substrate by the geometry of inclusion (Bender and Komiyama, 1978) However, catalysis may also arise as a result of differential solvation effects at the interface of the CD cavity with the exterior aqueous environment (Tee and Bennett, 1988a,b; Tee, 1989)

INTRAMOLECULAR REACTIONS

A simple example of non-covalent catalysis is the intramolecular acyl transfer [3] to [4] which is catalysed by a-CD but retarded by p-CD (Griffiths and Bender, 1973) As seen by the constants in Table 1, the

www.pdfgrip.com

“Based o n data from Griffiths and Bender (1973)

difference in behaviour of the two CDs lies in the substrate binding ( K s ) ,

and not in the transition state binding ( K T s ) The binding of the transition

state to each CD is very similar, but the stronger binding of the reactant to

p-CD in the initial state leads to rate retardation (k,lk, < 1) Presumably,

the substrate [3] (or as [3’]) sits deeper and more tightly in the larger cavity

of p-CD so that access to the transition state geometry is made more

difficult It is noteworthy that the “transition state analogue” [5] binds to

a-CD (inhibition constant, K I = 12 f 2 mM) with almost the same strength

as the actual reaction transition state which presumably resembles the

tetrahedral intermediate [6]

In another example of intramolecular participation, the attack of the

carboxylate ion group of mono-p-carboxyphenyl esters of substituted

glutaric acids, the rate of anhydride formation is sharply reduced by p-CD

(VanderJagt et al., 1970) Apparently, the substrates bind to p-CD in a

conformation that is unsuitable for reaction At the same time, the large

rate reductions must also mean that the transition state of the reaction

cannot be bound by p-CD in such a way as to be significantly stabilized

TRANSITION STATE STABILIZATION 15

Several other intramolecular reactions showed only slight rate accelerations

or retardations (VanderJagt et af., 1970) Of potential synthetic use, it has been found that both intramolecular and intermolecular Diels-Alder reac- tions can be catalysed by p-CD (Sternbach and Rossana, 1982; Breslow and Guo, 1988)

DECARBOXYLATION

The rate of decarboxylation of activated carboxylate anions [e.g (lO)], shows strong solvent dependence It is not surprising, therefore, that these reactions have been used to probe the microsolvent effects of micelles and CDs (Fendler and Fendler, 1975) In particular, it was anticipated that complexation with a C D might result in catalysis by providing an environ- ment for the reaction that is less polar than water

X-Ph(CN)CHCO; + X-Ph(CN)CH- + COZ ( 10)

In keeping with this expectation, Straub and Bender (1972a) found that the decarboxylation of phenylcyanoacetate anions (10) shows catalysis in the presence of p-CD, albeit modest [Appendix, Table A4.11 The rate accelerations show little variation (12-23, at 60.4”C) even though the reactivity of the anions spans two orders of magnitude and Ks varies with the position and size of the substituent Consequently, the values of pKTs vary in parallel with pKs (slope = 1.08k0.13; r = 0.957) which strongly suggests that the binding of the transition state in the CD cavity is very similar to that of the substrate, S

The magnitude of the rate accelerations caused by p-CD is comparable to that brought about by a change from water to 55% (w/w) aqueous 2-propano1, but significantly less than those in wholly organic media: 100% 2-propanol (2600); dioxane (2800) Also, the activation parameters for reaction in the mixed solvent and for the S - C D complex in water are very similar (Straub and Bender, 1972a) Presumably, these findings mean that

the aryl ring of S is situated largely in the C D cavity, with the anionic moiety

directed towards the exterior, so that the reaction centre is situated in a

“mixed” environment near the interface between the bulk aqueous medium and the less polar C D cavity

Data for the 4-chlorophenyl derivative were obtained at three tempera- tures (Table A4.1) At the lower temperatures, the rate acceleration is greater because the transition state binding is strengthened more than the substrate binding The data may be analysed to elicit the enthalpic and entropic contributions to the free energy of transition state stabilization, obtainable from the variation of AC&( =AH+, - TAS+s) with temperature (Table 2) If desired, the data may be further dissected since, from ( 9 ) ,

www.pdfgrip.com

16 0 S TEE

Table 2 Thermodynamic parameters for the P-cyclodextrin-catalysed decarboxyl-

ation of the 4-chlorophenylcyanoacetate anion."

AAS = 9.40 AY! 2 4.57 A& = 13.9

"From the data of Straub and Bender (1972a) (see Table A l ) Free energies and enthalpies are

in kcal mo1-l: entropies are in cal K - ' mol-'

-RTlnKTS = -RTln(k,/k,) -RTlnKs, and so AC;, is given by (ll),

where AAG' = (AC; - AC:) is the difference in activation free energies of the two kinetic steps The relationship (11) is evident in the diagram in Fig

2 Likewise, for the enthalpy and entropy, the separate contributions are

AH!;s = AAH$ + AH: and AS& = AASs + AS: (Table 2)

As seen in Table 2, AH;.s = 9.42 kcal mol-' and AS;, = 13.9 e.u., and so the free energy of transition state stabilization (approximately 5 kcal mol-')

results from a favourable enthalpy change, partly offset by an unfavourable

entropy change A similar situation pertains to binding of the substrate also

(Table 2) Thus, the similarity between transition state binding and substrate

binding, pointed out above from the correlation of pKTs with pKs, is

evident in thermodynamic parameters as well

The decarboxylation of benzoylacetic acids in acidic solution proceeds with intramolecular proton transfer [7] + (81 This feature of the reaction appears to limit charge separation in the transition state since the rates in water are very insensitive to the electronic nature of the substituents

( p = +0.03), unlike the reaction of their anions ( p = +1.42) (Straub and

Bender, 1972b) The reaction of the acids shows catalysis by p-CD, with

limiting accelerations of 2-8 (Table A4.1) Values of Ks and of KTs d o not

vary greatly with the aryl substituent, probably because the hydrophilic keto and carboxyl groups of [7] do not allow the benzoyl function to penetrate deeply into the C D cavity in either the initial state or the transition state The modest catalysis presumably arises because binding to the p-CD heIps

to bring the reactive groups together and to stabilize the cyclic transition state It is highly unlikely that catalysis results from a microsolvent effect since the decarboxylation reaction [7] + [8] is not particularly sensitive to the solvent (Straub and Bender, 1972b)

TRANSITION STATE STABILIZATION 17

BROMINATION-DEBROMIN ATION

Ionic reactions of neutral substrates can show large solvent dependence, due

to the differential solvent stabilization of the ionic intermediates and their associated dipolar transition states (Reichardt, 1988) This is the case for the

electrophilic addition of bromine to alkenes (Ruasse, 1990, 1992; Ruasse et

al., 1991) and the bromination of phenol (Tee and Bennett, 1988a), both of

which have Grunwald-Winstein rn values approximately equal to I so that

the reactions are very much slower in media less polar than water Such processes, therefore, would be expected to be retarded or even inhibited by CDs for two reasons: (a) the formation of complexes with the C D lowers the free concentrations of the reactants; and (b) slower reaction within the microenvironment of the less polar C D cavity (if it were sterically possible) Contrary to the above expectations, the bromination of anisole (Tee and Bennett, 1984) and of phenols (Tee and Bennett, 1988a) in the presence of a-CD is not strongly retarded, so that some form of catalysis must occur In some cases, actual rate increases are observed in spite of the several complexations that reduce the free reactant concentrations Analysis of the

effects of substituents on the kinetics leads to the conclusion that the

catalysis by a-CD most probably results from reaction of CD-bound bromine with free substrate (12a) and that the a - C D B r 2 complex is 3-31

times more reactive than free Br2 towards phenols and phenoxide ions (cf

Tee et al., 1989) For the kinetically equivalent reaction of the substrate C D complex with free bromine (12b), the rate constants (k:) for phenols do not correlate sensibly with the nature and position of the substituents, and for three of the phenoxide ions they have unrealistically high values, greater

18 0 S TEE

be calculated most easily from (14), the ratio of the second-order rate constant for the normal reaction (13a) and the third-order rate constant for the CD-catalysed reaction (13b) [see Section 3, (S)], where TS and T S - C D are the transition states corresponding to k2" and k3c, respectively Note that

k3C = k q / K B or k!lKs, from (12a) and (12b)

This approach has been applied (Tee, 1989) to kinetic data for the bromination of phenols and phenoxide ions catalysed by a - C D For 15 different substrates (nine phenols and six phenoxides) K,rs values vary only

between 0.07 and 0.8mM, with most being between 0.1 and 0 5 m M , indicating very similar transition state stabilization for substrates with a

range of reactivity of 40 million (Table A4.2) Moreover, the values of K,rs

show no clear correlation with K s This lack of dependence of KTS on the structure of the substrate is strong evidence that the transition state for the catalysed process is one in which the phenol moiety is basically outside the

CD cavity while the bromine is inside ([9] -+ [ 101) The same conclusion was

X

=OH

[91 arrived at in the original paper (Tee and Bennett, 1988a), but using slightly different arguments In particular, it was noted that the Hammett p €or the

catalysed and uncatalysed reactions (kf and k2") are virtually equal, suggesting that the organic substrate remains in a largely aqueous environ-

ment Also, as noted above, rate constants ( k ; ) for the alternative

mechanism (12b) vary less sensibly and some are physically unreasonable The debrominations of a series of 4-alkyl-4-bromo-2,5-cyclohexadienones

(ipso-dienones [ll]) were also studied and found to undergo strong catalysis

by a-CD (Tee and Bennett, 1988b) These reactions were chosen for scrutiny since they should serve as good models for the reverse of the

TRANS IT1 ON STATE STAB I LlZATlO N 19

brominations just discussed Values of K,, for the debrominations fall in the

narrow range of 6 x lo-' to 12 X M (Table A4.3) and are insensitive to the structure of the dienone If, in the transition state for debromination,

the @so-dienone were bound inside the cavity of a-CD, particularly through its alkyl group, one would expect a greater dependence of KTs on the size and shape of the alkyl group(s) Thus, for debromination, as for bromina- tion, the catalysis data suggest a transition state in which the organic moiety

is largely outside the C D cavity, and the two bromine atoms involved in the reaction are essentially inside ( [ l l ] -+ [12]) It is gratifying (and reassuring)

that the two separate studies of a-CD catalysed bromination and debro- mination arrived at the same description of the transition state that the two reactions have in common

The origin of the CD catalysis of bromination and debromination probably relates to solvation; yet it cannot be a simple microsolvent effect since brominations are much slower in media less polar than water, as remarked above Most probably the catalysis arises from a differential effect

of the aqueous exterior, where the organic moiety resides, and the less polar

CD cavity containing the bromines For bromination, solvent reorganization around the developing bromide ion is less necessary (than in the normal aqueous reaction) since it is being formed in the CD cavity (Tee and Bennett, 1988a) For debromination, nucleophilic attack can occur by a largely desolvated bromide ion which thus behaves as a stronger nucleophile (Tee and Bennett, 1988b)

Debromination of the @so-dienone [13] (+ [14]), formed during the

www.pdfgrip.com

20 0 S TEE

course of the bromination of 5-methylsalicylic acid, is subject to intra- molecular general acid catalysis by the carboxyl group (Tee and Iyengar,

1985; Tee et a l , 1986) The effect of (u-CD on this reaction was studied

(Takasaki and Tee, 1989) to see how the two very different types of catalysis interact with one another, since enzymes normally use several catalytic effects to achieve large rate accelerations (Jencks, 1975; Gandour and Schowen, 1978; Fersht, 1985; Page and Williams, 1987) Conceivably, the two forms of catalysis might interact with one another in three different ways: destructively, one interfering with the other (worst case); indepen-

dently, each contributing its individual acceleration (acceptable); or con-

structively, each amplifying the effect of the other (best possible result)

In the event, it was found that the two forms of catalysis act together on a

single transition state to give an impressive rate enhancement of 12 million

However, each form of catalysis operates more or less independently of the other (Takasaki and Tee, 1989), an effect termed “cocatalysis” Analysis of the kinetic data showed that the component of catalysis due to the a-CD (3400 times) is within the range of values found for other ipso-dienones (2400-4600), even though the anion of [13] is 3500 times more reactive than the analogous dienone lacking the 2-carboxylate group (Scheme 1) More- over, the KTs of 0 0 8 8 m M for [13] is in the middle of the range of the

values for other ips0 dienones (Table A4.3), indicating the same degree

of transition state stabilization by a-CD Therefore, the findings for the CD-catalysed debromination of [ 131 are also consistent with the transition

TRANSITION STATE STABILIZATION 21

state having the dienone moiety outside of the C D cavity (as for [11]-+ [12]) Furthermore, the fact that the two forms of catalysis d o not interfere with each other may be taken as evidence that they take place in two spatially distinct regions: internal general acid catalysis in an aqueous environment outside the C D cavity; nucleophilic bromide ion attack inside the C D cavity

The effects of a-CD on the bromination of other substrates have been studied recently (Javed, 1990; Tee et al., 1990a; Tee and Javed, 1993), the object being to see if the catalytic effects observed earlier with phenols (Tee and Bennett, 1988a) are peculiar to these substrates or more general Broadly speaking, various aromatic and heteroaromatic substrates (Table

A4.4) showed behaviour (k$lk2, = 1.7 to 10; KTS = 0.2 to 1.2 mM) very similar to that of phenols, and so the catalytic effect appears to be fairly general The oxidation of formic acid by bromine also shows catalysis by

a-CD (Han et af., 1989; Tee et al., 1990a)

The first finding was that the four p-halogenophenols (X = F, CI, Br, or

1) have remarkably similar transition state stabilization ( KTs = 0.40, 0.43, 0.46, and 0.29mM), even though these substrates have a wide range of ability to bind to a-CD ( K s = 120, 3.6, 1.4, and 0.47mM) (Table A4.4) This finding is inconsistent with inclusion of the phenol by the C D during the catalysed bromination and so affords yet further support for the view of the transition state implied in [9] -+ [lo]

Three salicylate (2-hydroxybenzoate) anions, which have unusual reactiv- ity towards bromine that has been attributed to intramolecular proton transfer assisting electrophilic attack (Tee and Iyengar, 1985, 1990), exhibit

modest catalysis (k;\lk2, = 3 to 10) and have KTS values similar to phenols Pyridones and their N-methyl derivatives, three heteroaromatic acid anions,

and four phenoxy derivatives show comparable catalysis (k$lk2, = 1.7 to 9.5) and KrrS values (Table A4.4)

To provide an example of a reaction that is very different to electrophilic aromatic substitution, the oxidation of formic acid by bromine was also studied This reaction, which involves electrophilic attack on the formate anion (15) (Cox and McTigue, 1964; Smith, 1972; Herbine et al., 1980;

Brusa and Colussi, 1980), is catalysed by a - C D (ktlk,, = 11) (Tee et al.,

1990a), and the degree of transition state stabilization (KTS = 0 1 8 m ~ ) is similar to that for phenols (Table A4.2) and most of the other substrates (Table A4.4)

Br2 + HCO; + 2Br- + H + + C 0 2 (15)

Combining the results for 34 different substrates (Tables A4.2 and A4.4),

there is a good correlation of logk,, with logkz,, covering 10 orders of

magnitude, with unit slope (1.01; r = 0.993) (Fig 3) Because k3= = k $ / K B

(12a), logkt also correlates with logk,, in the same way Apparently, then,

www.pdfgrip.com

22 0 S TEE

Fig 3 Correlation of the third-order rate constants for a-CD catalysis of bromine

attack with the second-order rate constants for the uncatalysed reaction Data from

Tables A4.2 and A4.4 (Tee and Javed, 1993)

the nature of the catalysis of bromine attack (discussed above) is much the same for all of these 34 substrates, with only very minor variations in the extent of catalysis for the different structural types In the same vein, the amount of transition state stabilization provided by a-CD is virtually constant for substrates with a 10” range of reactivity, further supporting the reaction scheme, expressed in (12a) and illustrated by [9]+ [lo], in which the substrate remains essentially outside the CD cavity

5 Covalent catalysis

The term “covalent catalysis” was chosen by Bender and Komiyama (1978)

to classify reactions in which there are covalent interactions between a functional group on the CD and the substrate during the rate-limiting step of the reaction The reaction in this category which has been most studied is

the cleavage of aryl esters (Bender and Komiyama, 1978; Matsui et al.,

1985; Tee, 1989)

ESTER CLEAVAGE

In most cases the esterolysis takes place by nucleophilic attack of an ionized

hydroxyl of the C D (VanEtten et a l , 1967a), leading to acyl transfer

(VanEtten et al., 1967b) Under the reaction conditions the acylated CD

TRANSITION STATE STABILIZATION 23

which is produced is normally fairly resistant to hydrolysis so that overall the ester hydrolysis is not formally catalysed Because of the partial covalent interaction between the ester substrate and the CD in the transition state for

acyl transfer quite low values of KTS can be found (Tee, 1989) Moreover, they show a strong dependence on the position and size of substituents, rather than on their electronic character (Komiyama and Bender, 1978;

Matsui ef al., 1985)

These features emerge from the data in the classic paper by Bender and coworkers (VanEtten et al., 1967a), much of which is presented in Table

A5.1 Broadly speaking, they found that meta-substituted phenyl acetates

are superior to their para isomers as substrates for cleavage by both a- and

p-CD, a finding supported by much subsequent work (e.g Matsui et al., 1985; Tee and Takasaki, 1985; Tee et al., 1990b) This difference in

behaviour is strongly correlated to differences in transition state binding, as shown below

The transition state for the cleavage of phenyl acetate by a-CD has

KTS = 0.81 m M (Table A5.1) Acetates with para substituents have larger

values (weaker transition state binding) whereas for meta groups the values are generally lower (stronger transition state binding) Thus, the values of

K-rs are consistent with the view that mefa substituents, regardless of their

electronic nature, position the phenyl group of the ester in the CD cavity in

a geometry which facilitates the attack of an ionized hydroxyl group and the

formation of the transition state for acyl transfer (Scheme 2A) In contrast,

para substituents position the ester in the CD cavity in such a way that nucleophilic attack is more difficult and they also tend to interfere with transition state binding (Scheme 2B)

Support for the above view comes from NMR studies of the binding of phenyl and nitrophenyl acetates to a-CD (Komiyama and Hirai, 1980) These indicate that the nitro groups are located in the CD cavity and that the acetoxyl groups of the esters are held outside, more or less close to the secondary hydroxyls of the CD It was calculated that the distance between the ester carbonyl carbon and the secondary hydroxyls decreases as

p-nitro > phenyl> m-nitrophenyl, consistent with the observed order of rate acceleration (Komiyama and Bender, 1984)

The cleavage of phenyl acetates by p-CD shows the same general features

as that by a-CD (Table A5.1), although there are quantitative differences that must arise from the larger cavity size of p-CD Generally, the mefa-substituted esters are not cleaved as well as by a-CD and the pura-substituted esters are cleaved better Thus, the distinction between the kinetic parameters for two series of esters is less dramatic for p-CD, presumably because of the looser fit of substituted phenyl groups in p-CD This trend is continued with the two entries for y C D (which has a still larger cavity) where the differences between the meta and para isomers

of t-butylphenyl acetate are quite small (Tables A5.1) Nevertheless, the

www.pdfgrip.com

depictions in Scheme 2 are still appropriate as the difference between metu-

and para-substituted isomers is generally substantial

This difference is clearly shown by the p K ~ s values plotted in Fig 4, which are calculated from the extensive data for ester cleavage by p-CD (Tables A5.2 and A5.3) accumulated by Fujita and coworkers (Matsui et al.,

1985; Fujita, 1988) For rn-alkyl and halogen substituents there is a good

TRANSITION STATE STABILIZATION

and A5.3

linear correlation ( r = 0.992) between the free energy of transition state binding (expressed by pKTs) and that of substrate binding (pK,), with a slope of 1.63 k 0.07, strongly supporting the view that for metu substituents the S C D complex and the T S - C D complex have similar geometries (Scheme 2A) In contrast, the correlation is poorer ( r = 0.788) for puru

substituents and the slope is closer to zero (0.38 k 0.1 1); only in the case of two long, flexible alkyl groups (n-butyl and n-pentyl) is transition state binding improved significantly (Fig 4), perhaps because they can accommo- date better to the cavity

Fujita and coworkers (Matsui ef ul., 1985; Fujita, 1988) have also collected a large body of data for the basic cleavage of metu-substituted esters by a-CD The observed accelerations (k,lk,) vary from 41 (X = H ) to

360 (X = CHO), with most being in the range 100-250 (Table A5.4) The strongest transition state stabilization is for the m-iodo substituent

(KTS = 2 8 p ~ ) , but since this also gives the strongest substrate binding (Ks = 0.48 mM), the acceleration of 170 is not exceptional The plot of pK,,,s versus pKs (Fig 5) shows a fair correlation ( r = 0.928) between transition state binding and substrate binding, with near unit slope (1.09), even though

it includes substituents of various structural and electronic types This correlation is also consistent with the mechanism outlined in Scheme 2A

The correlations presented in Figs 4 and 5 are in stark contrast to the

disorder shown in a plot of logk,lk,, against the Hammett u constants for

www.pdfgrip.com

26 0 S TEE

6.0

5.0 u)

Data from Table A5.4

meta and para substituents (VanEtten el al., 1967a), which the late Professor

Myron Bender often claimed was “the world’s worst Hammett plot” (e.g Bender, 1987) His point in doing so was to emphasize that it is the position

of a substituent, rather than its electronic nature, that largely determines its effect on the acceleration of CD-induced phenyl acetate cleavage (Komiyama and Bender, 1978) This view is supported by the linear correlations of logk, with various parameters found by using multiple regression analysis (Matsui et a f , 1985) The correlation equations show that the electronic contribution of a substituent is virtually the same as that of

the normal reaction ( k , ) so that it cancels out in the acceleration ratio

( k c / k u ) The correlations also reveal an unfavourable steric term for para

substituents, whereas bulky meta substituents improve the esterolytic ability, again consistent with the portrayals in Scheme 2

Unlike the phenyl acetates in Tables A5.1 to A5.4, basic cleavage of ethyl

benzoates, ethyl cinnamates, and Ph(CH2),,COOEt ( n = 1, 2, 3) is slower with p-CD, except in the case of some benzoates which exhibit quite modest rate enhancements; with a-CD the cinnamate esters mainly show inhibition

(Tanaka et a f , 1976) All of these substrates show saturation kinetics, with

K s in the millimolar range, and so their KTs values are all high (Table A5.5)

On the other hand, esterolysis of phenyl benzoates shows more enhance- ment (k,lk, 10) with a-CD (VanEtten et al., 1967b) Thus, as has been shown by Menger and Ladika (1987) for ferrocenylacrylate esters, a good leaving group (normally phenoxy) seems to be a requirement for large rate accelerations

TRANSITION STATE STABILIZATION 27

The “best” substrate found by Bender’s group was rn-t-butylphenyl

acetate, undergoing cleavage by p-CD (VanEtten et al., 1967a) For this

ester, k,lk, = 250 and K s = 0.13 m M , so that KTs = 0.52 p ~considerably ,

lower than for most other phenyl acetates studied (Tables A5.1 to A5.4) Thus, the binding of the t-butyl group in the p-CD cavity stabilizes the transition state much better than other rneta substituents However, the

acceleration is no larger than that for other groups because the substrate binding is equally improved by an rn-t-butyl group In 20.5% aqueous CH3CN the rate acceleration is raised to 940 because substrate binding is weakened ( K s = 2.3 m M ) (VanEtten et al., 1967b) somewhat more than the

transition state binding ( K T S = 3.3 p ~ ) In the same medium, replacing all the primary hydroxyls of p-CD with mesyloxy groups (CH3SO20-) further enhances cleavage due to even weaker substrate binding and stronger transition state binding (k,lk, = 1550; K s = 3.1 mM; KTs = 2 0 p ~ ) , im- plying that the t-butyl group of the ester penetrates deeply enough into p-CD cavity to interact significantly with the substituents on the primary side, perhaps because they can fold inwards closing off the bottom of the

CD cavity In strong contrast, methylating the secondary hydroxyls com- pletely destroys the rate acceleration because the nucleophilic sites on the

wide end of the CD cavity are all blocked (VanEtten et ul., 1967a,b)

The pioneering studies of Bender’s group were followed by many attempts to increase the efficiency of esterolysis by cyclodextrins and several approaches have been tried, most notably in Breslow’s laboratory One may

“optimize” the structure of the substrate (Trainor and Breslow, 1981; Breslow et al., 1983), modify the cyclodextrin (Emert and Breslow, 1975;

Breslow et al., 1980; Fujita et al., 1980), or alter the solvent (Siegel and

Breslow, 1975) The last of these is the easiest to achieve but detailed studies are made tedious by the necessity to redetermine all of the relevant equilibrium and rate constants, and the acidity dependence of the catalysed and uncatalysed processes, in the new medium

The study by Siegel and Breslow (1975) is one of few involving solvent variation and having sufficient kinetic data to allow calculation of K r s for

different media First, they showed that various organic species bind to p-CD in DMSO solution, though not as well as in water A medium change from 0% to 50% (v/v) aqueous DMSO to 100% DMSO weakens the binding of rn-t-butylphenyl acetate substantially: K s = 0.1 to 2.0 to 18 mM

For basic cleavage of the same ester, with and without p-CD, the change from 0% to 60% (v/v) aqueous DMSO increases k , by 25, k , by 48, and the acceleration ( k , / k , ) rises from 270 to 510 (Table 3 ) As the authors emphasize, the reaction at kinetic saturation ( k , ) is 13000 times faster in 60% aqueous DMSO than the background reaction ( k , ) in water containing

the same buffer To get at the origins of this acceleration i t is necessary to dig deeper and to look at the effect of solvent change on transition state stabilization

www.pdfgrip.com

'Assumed value, given that Ks = 2 m M in 50% aq DMSO Any other value in the millimole

range would not alter the arguments in the text Note that the assumed value is incorporated into both K , , and k Z

Assuming K s = 5 mM in 60% (v/v) aqueous DMSO (since it is 2 mM in 50% aqueous DMSO), K,rs = 9 8 p ~ , as compared to 0 3 7 ~ ~ in water: transition state binding is 26 times weaker in the mixed solvent More surprising, however, k2 (= k , / K s ) is the same in both media (Table 3) Thus,

the much faster cleavage of the ester by p-CD in 60% aqueous DMSO

originates from two factors: (i) the enhanced nucleophilicity and basicity of anions in the mixed medium (Reichardt, 1988); and (ii) substantially weaker

substrate binding (1/5O) in 60% aqueous DMSO while transition state binding is weakened less (1126) Of the two, the first factor is much more

important The virtual equality of k2 in the two media arises because the

48-fold increase in k , is matched by the 50-fold increase in K s (Table 3)

Obviously, the esterolytic ability of a CD can be improved by replacing one of its primary or secondary hydroxyl groups by a stronger nucleophilic group such as thiol, amino, or imidazolyl (Fendler and Fendler, 1975; Bender and Komiyama, 1978; Fikes et al., 1992) However, such replace- ments bring about gross changes in reactivity which obscure the effect of CD

binding on the reaction It is more informative in this respect to make more subtle changes to the CD to modify its ability to bind substrates and transition states

TRANSITION STATE STABILIZATION 29

With such considerations in mind, presumably, Breslow and coworkers

(Emert and Breslow, 1975; Breslow et ai., 1980) prepared modified p-CD

with seven pendant N-methyl (or ethyl) formamido groups, in place of the primary hydroxyl groups These groups may form a flexible floor to the p-CD cavity which might adjust itself to suit the binding of different substrates and transition states A capped p-CD derivative with a diphenyl- oxy moiety was also prepared and studied Accelerations of up to I million

were observed, corresponding to low KTS values, down to 7.5 x lo-’ M

(Table A5.6) However, these impressive values are intrinsic to the esters, which had been carefully designed using CPK (space filling) models for optimal transition state binding In actuality, the flexible capping has only small effects on the efficiency of ester cleavage by p-CD For various esters the values of k J k , were raised by 7- to 20-fold, due partly to slightly weaker

substrate binding and/or marginal improvements in transition state binding (Table A5.6)

In a related study, Fujita et al (1980) modified p-CD by replacing one of the primary hydroxyl groups by S-methyl or S-t-butyl; they also prepared a derivative capped on the primary side with a diphenylmethyl unit The efficacy of these derivatives in cleaving rn- and p-nitrophenyl acetates was

measured (Table A5.6) Similar to Breslow’s work, it was found that the presence of the S-methyl group has little effect on either K s or K , r s ,

suggesting that it does not intrude far into the C D cavity The larger S-t-butyl group presumably provides more of an intrusive floor since it

lowers K s and KTS to a lesser extent, resulting in lower accelerations With

the diphenylmethyl capped p-CD, binding of the m- and p-nitrophenyl

acetate substrates is strengthened considerably ( K s = 6.1 + 0.11 and 4.80+ 0.012 mM, respectively), and so is transition state binding to a lesser

extent ( K T s = 0.085- 0.017 m ~and 620- 3.2 ; FM), so that the accelera- tions are reduced (k,lk, = 72- 6.5 and 7.7- 3.9)

Covalent modification represents only one way to alter the binding properties of a CD Obviously, changing the solvent system is another way, but this will normally affect reactivity at the same time (VanEtten et a l ,

1967a,b; Siege1 and Breslow, 1975), as already discussed in relation to the data in Table 3 But, as presented later, there is an even more subtle way to modify the binding capacity of the C D cavity, by the addition of an inert spacer molecule or “spectator” Besides the expected inhibition observed in most cases, there are instances where the addition of a potential inhibitor brings about rate increases due to improved transition state binding (Tee and Hoeven, 1989; Tee and Bozzi, 1990; Tee et al., 1993b)

Several of the entries in Table A5.6 also represent many of the efforts by Breslow’s group to “improve” substrates for cleavage by p-CD The adamantylpropiolate ester [ 15a] exhibits a healthy acceleration of 2150, which is raised to 14000 by flexible capping and to 15000 by judicious

placement of a t-butyl group (Breslow et ul., 1980); KTS values for these

www.pdfgrip.com

30 0 S TEE

situations are about lo-’ M In contrast, cleavage of the adamantanecar- boxylate ester [15b] is retarded 28-fold (k,lk, = 0.036; K T S = 42mM) by p-CD (Komiyama and Inoue, 1980a), and that of the homologous adaman- tylacetate [15c] is raised only threefold (Komiyama and Inoue, 1980b) Clearly, the size, shape, and rigidity of the side chain on the adamantane skeleton (which is the primary binding site of the esters [15]) greatly affects access of the secondary alkoxide nucleophile to the ester carbonyl and hence the stabilization of the cleavage transition state Similar considerations apply

to esters binding in the CD cavity through a ferrocene group (Fc): the Fc-propiolate [ 16a] is accelerated by 1.4 x 10’ and the Fc-acrylate [ 16b] by

7.5 x 10’; for these esters KTs drops to 3.6X lo-’ and 9.3 x 1 0 - ” ~ ,

respectively Capping affects these values only slightly (Table A5.6) As

noted above, a good phenoxy leaving group on ferrocenylacrylate esters such as [16b] appears to be mandatory for large accelerations (Menger and Ladika, 1987)

Further developments of ferrocene based esters led to even faster acyl

transfers to p-CD (Trainor and Breslow, 1981; Breslow et al., 1983), the

most spectacular rate accelerations, up to 6 million, being with the derivatives [ 171 and (181 in which an acrylate moiety is conformationally

Fe

restricted by a ring (Table 4) Since K s values are in the normal millimolar

range, the accelerations are solely due to much improved transition state binding: in one case KTS is reduced to 9.7 x 1 0 - ’ ” ~

As impressive as these developments have been, chemists still have a way

to go to catch up with “Mother Nature” For enzymes KTs may be as low as 10-20 M , since K M is generally in the range lo-’ to 10- M and k J k , values

are up to 1014 or more (Lienhard, 1973; Kraut, 1988) (see Enzymes, Section 6) Further lowering of KTs for “artificial enzymes” below 1 0 - l ” ~ will no doubt require more covalent interactions in the transition state, with better catalytic groups Nevertheless, the transition state stabilization evident in

Table 4 is comparable to that which has been achieved so far with catalytic

antibodies (Section 6)

The esters in Table 4 also provide two excellent examples of enan- tioselectivity This behaviour was revealed when Breslow and coworkers

TRANSITION STATE STABILIZATION 31

Table 4 The "best" esters for acylation of P-cyclodextrin."

40

1.4

0.26 5.4

"Based on the data of Van Hooidonk and Breebart-Hansen (1970)

noted biphasic kinetics due to the different reactivities of the two enan- tiomeric forms of the esters The selectivities of 20 and 62 are substantial, and the values of K s and K,, show that they are almost solely due to differences in the stabilization of the two diastereomeric transition states, rather than to differential binding of the enantiomeric substrates

The enantioselectivity just discussed arises because CDs are inherently chiral due the asymmetry of their D-glucose units Many attempts have been made to exploit this attribute for chemical purposes and some success has been achieved in synthesis (Bender and Komiyama, 1978), and in the

physical separation of enantiomers (Szejtli, 1982; Armstrong et al., 1986),

the latter now being in general use in chromatographic resolution More limited success has been obtained in studies of kinetic resolution, comparing the reactivity of one enantiomer to the other (Bender and Komiyama, 1978; Szejtli, 1982) For the cleavage of various aryl esters by CDs and by

modified p-CD derivatives, Fornasier et al (1983, 1987a,b) found selectivi-

ties up to 19 Similarly, the cleavage of oxazolones by CDs shows values up

to 12 (Daffe and Fastrez, 1983) With two N-carbomethoxyphenylalanine

esters very low selectivities of only 1.2-2.3 have been observed (Ihara et al.,

1986)

More significant is the enantioselectivity shown by the cholinesterase inhibitor, Sarin [19] (Van Hooidonk and Breebart-Hansen, 1970) This nerve agent (Benschop and D e Jong, 1988) is cleaved by a-CD, with a 36-fold preference for the more potent R ( - ) enantiomer (Table 5 ) The

www.pdfgrip.com