дНЙСЛЕМР БГЪР ХГ ЙЩЬЮ ОНХЯЙНБНИ ЛЮЬХМШ. юДПЕЯ НПХЦХМЮКЭМНЦН ДНЙСЛЕМРЮ : http://www.enzyme.chem.msu.ru/ryabov/pdf/CEJ_98.pdf
дЮРЮ ХГЛЕМЕМХЪ: Wed Jul 26 16:03:02 2000
дЮРЮ ХМДЕЙЯХПНБЮМХЪ: Mon Oct 1 21:37:15 2012
йНДХПНБЙЮ:

FULL PAPER

Enantioselectivity in Enzyme-Catalyzed Electron Transfer to and from Planar Chiral Organometallic Compounds
Alexander D. Ryabov,* Yulia N. Firsova, Vasily N. Goral, Ekaterina S. Ryabova, Angelina N. Shevelkova, Ludmila L. Troitskaya, Tatyana V. Demeschik, and Viatcheslav I. Sokolov
Abstract : Asymmetric cyclopalladation of dimethylaminomethylferrocene in the presence of N-acetyl-(R)- or (S)leucine afforded enantiomerically enriched palladacycles (S)- and (R)[Pd{C5H3(CH2NMe2)FeC5H5}(m-Cl) ]2 , respectively. Carbonylation of each enantiomer followed by iodomethylation and reduction by sodium amalgam gave (S)- and (R)-2-methylferrocene carboxylic acid (1) with an optical purity of 80 and 93 %, respectively. (S)- and (R)-1 readily undergo one-electron (1e) oxidation to form the corresponding ferricenium cations by hydrogen peroxide, catalyzed by horseradish peroxidase (HRP) and chloroperoxidase (CLP) from Caldariomyces fumago (25 8C, pH 5 ╠ 8 and 2.75, respectively). In the case of HRP, the reaction is strictly firstorder with respect to (S)- and (R)-1 (rate k[HRP][1] ), whereas Michaelis ╠ Menten kinetics are observed for CLP. The strongly pH-dependent kinetic enantioselectivity is, however, only observed in the case of HRP. HRP-generated cations (S)-1 and (R)-1 have been used to demonstrate that their enzymatic reduction by reduced glucose oxidase (GO) is also enantioselective ; the (S)-1 enantiomer is more reactive than (R)-1 by a factor of 1.54. The existence of the planar chiral enantioselectivity in the GO catalysis was also confirmed by the cyclic voltammetry study of (S)-1 and (R)-1 in the presence of GO and b-d glucose with glassy carbon and pyrolytic Keywords : chiral recognition Є electron transfer Є ferrocenes Є oxidoreductases Є planar chirality graphite electrodes. The corresponding enantioselectivity factors k(S)-1/k(R)1 are 1.7 and 1.6, respectively. Based on the known X-ray structural data for the active site of GO, it has been tentatively suggested that the enantioselectivity originates from the hydrophobic contact between the enzyme tyr-68 residue and the h5-C5H5 ring of 1, and a hydrogen bond network formed by his-516 and/or his-559 residues and the carboxylic group of the ferrocene derivative. The findings reported confirm the existence of enantioselective electron transfer between oxidoreductases and organometallic compounds with a planar chirality. The lack of kinetic enantioselectivity may be a result of i) the incorrect ratelimiting step, ii) unfavorable pH region, and iii) the deficit of charged groups attached to ferrocenes.

Introduction
Our recent investigations in the field of organometallic biochemistry[1, 2] have demonstrated that the ferrocene/ferricenium couple is unique for redox enzymes that are capable of efficient electron transfer to and from the ferricenium cation (Fc) and ferrocene (Fc) (Scheme 1). In particular, ferrocenes
[*] Prof. A. D. Ryabov, V. N. Goral, E. S. Ryabova, Dr. A. N. Shevelkova Department of Chemistry, Moscow State University, 119899 Moscow (Russia) Fax : ( 7) 095-939-2742 E-mail : ryabov@enzyme.chem.msu.su Y. N. Firsova Division of Chemistry, G. V. Plekhanov Russian Economic Academy Stremyanny per. 28, 113054, Moscow (Russia) Dr. L. L. Troitskaya, T. V. Demeschik, Prof. V. I. Sokolov A. N. Nesmeyanov Institute of Organoelement Compounds Russian Academy of Sciences Vavilov St. 28, 117813, Moscow (Russia)

Scheme 1. Electron transfer to and from ferrocene and the ferricenium cation.

are oxidized by hydrogen peroxide in the presence of horseradish peroxidase (HRP) and chloroperoxidase (CLP). In the former case, the enzymatic reaction is characterized by unexpected first-order kinetics with respect to the ferrocene derivative, however the reactivity under the steady-state[3] and stopped-flow[4] conditions is comparable with that for typical organic substrates associated with HRP. Since 1984 ferricenium ions have been known to be excellent oxidants of reduced glucose oxidase {GO(red)}, which is produced during
0947-6539/98/0405-0806 $ 17.50+.50/0 Chem. Eur. J. 1998, 4, No. 5

806

WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

806 ╠ 813 the oxidation of b-d -glucose to d -gluconolactone.[5] This was confirmed in a variety of electrochemical studies by many research groups[6╠9] including ourselves.[10, 11] Recently, a detailed analysis of the steady ╠ state kinetic data, which was obtained spectrophotometrically by monitoring the presence of ferricenium dyes RFcPF6Ч (R alkyl group), was carried out with respect to the interaction between GO(red) and RFc.[12] It was found that the reduction of RFc follows Michaelis ╠ Menten kinetics, and the intrinsic kinetic parameters for HFc fall in the same range that is typical for b-d glucose. Thus, there is evidence that ferrocene derivatives effectively mediate biocatalyzed reactions. This presents the possibility that the enzymatic transformations of ferrocenes might also proceed stereoselectively. Ferrocene derivatives and related organometallics are nowadays recognized substrates of proteases and oxidoreductases in synthetically relevant reactions that are aimed at modifying side-chain functional groups.[13╠16] Enantiomers of organometallics with planar chirality can display different reactivity or be accumulated with distinct rates that allows for kinetic resolution.[1, 2] These examples and our findings described above raise the question of whether it is possible to observe enantioselectivity in an electron-transfer process involving a redox enzyme and a planar chiral organometallic molecule, that is, unnatural substrate with unnatural chirality type ? Whilst this manuscript was in preparation, a preliminary communication was published in which planar ╠ chiral enantioselectivity was observed in oxidation catalyzed by wild-type and mutant forms of cytochrome c peroxidase.[17] However, the question is still intriguing when we take into consideration the controversial publications by Willner et А al.[18] and Saveant et al.[19] The former has claimed that ferrocene derivatives with a central carbon chirality, namely the R and S enantiomers of Me2NC*MeHFc, show different Abstract in Russian : reactivity in the oxidized state towards GO and glutathione reductase. The latter group has been unable to reproduce these results. In this work, we demonstrate that it is possible to achieve stereoselectivity for the planar chiral ferrocene molecules[20╠23] in the presence of several oxidoreductases under properly selected conditions. In particular, we describe i) the optimized synthetic approach to (S)-2-methylferrocene carboxylic acid, (S)-1, and (R)-2-methylferrocene carboxylic acid, (R)-1, the key step of which is asymmetric cyclopalladation of dimethylaminomethyferrocene in the presence of either N-acetyl-(R)- or (S)-leucine ; ii) the spectrophotometrically measured kinetic data of the HRP- and CLP-catalyzed oxidation of (S)-1 and (R)-1 by H2O2 revealing the pHdependent enantioselectivity for the former enzyme and its absence for the latter ; iii) the enantioselective reduction of ferricenium cations (S)-1 and (R)-1 by GO(red) in the presence of b-d -glucose evaluated by both conventional spectrophotometry and cyclic voltammetry with two different carbon electrodes.

Results
Synthesis of planar chiral ferrocene by cyclopalladation : A synthetic approach to a pair of planar chiral enantiomers is shown in Scheme 2. The key step is asymmetric cyclopalladation of dimethylaminomethylferrocene by Na2PdCl4 in the presence of sodium salts of N-Ac-(R)-Leu or N-Ac-(S)-Leu. The ability of enantiomerically pure N-acetyl amino acids to

Scheme 2. Reaction scheme for the preparation of (S)-1 and (R)-1.
Chem. Eur. J. 1998, 4, No. 5 WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998 0947-6539/98/0405-0807 $ 17.50+.25/0

807

FULL PAPER
induce the preferential formation of planar chiral enantiomers has already been established.[24] However, the whole reaction sequence was previously performed for only one enantiomeric series.[20, 22] Here we report on the substantially improved preparative synthesis of both series in order to obtain a pair of enantiomerically pure target molecules. The absolute configuration and the enantiomeric purity of the key organopalladium compounds 2 and their derivatives were determined as previously described.[21] The procedure in Scheme 2 is easier and more efficient than previously reported methods of asymmetric synthesis of planar chiral ferrocene derivatives,[25, 26] in which interest has markedly increased in recent years.[27] Its most advantageous feature is an easy access to both planar chiral enantiomers. Moreover, an alternative chemical approach to such molecules[26] is more laborious and hardly applicable to the preparation of 2-methylferrocene carboxylic acid. The successful realization of the transformations in Scheme 2 demonstrates once again the high potential of palladacycles in a variety of organic syntheses.[28]

A. D. Ryabov

Figure 2. a) Steady-state rate of HRP-catalyzed oxidation of (R,S)-1 as a function of pH : [1] 8 А 10Ч4 m , [H2O2] 2 А 10Ч4 m , [HRP] 10Ч7 m, 25 8C. b) Effects of pH on the HRP enantioselectivity in terms of kR/kS ratio for HRP-catalyzed oxidation of 1 versus solution pH : [H2O2] 2 А 10Ч4 m , [HRP] 10Ч7 m, 25 8C.

HPR-catalyzed oxidation of (R)- and (S)-1 by H2O2 : Compound 1 is soluble in water at pH b 6, that is, when the acid is deprotonated.[29] This means that the HRP-catalyzed oxidation of ferrocenes by H2O2 , which follows 2 :1 stoichiometry, can be carried out in the absence of surfactants that were previously necessary in order to increase the solubility of alkylferrocenes.[3] The steady-state rate of oxidation of (R)and (S)-1 by H2O2 at pH 7 in the presence of HRP as a function of [ (R)-1] and [ (S)-1] is shown in Figure 1. As in

vanishes at pH 5. Curiously, the activity of HRP is higher at lower pH, (Figure 2a). Therefore, the highest enzymatic activity is not required for achieving the highest enantioselectivity ; this is observed for the electron-transfer enzymatic process that occurs without kinetically meaningful enzyme ╠ substrate binding. CLP-catalyzed oxidation of (R)- and (S)-1 by H2O2 : Our recent study of the mechanism of CLP-catalyzed halogenation led to the conclusion that stereoselectivity cannot be achieved in this process.[30] However, selectivity is observed when CLP displays peroxidase activity, that is, when the CLP-catalyzed oxidation by H2O2 proceeds in the absence of halide ions.[31╠33] The pH optimum of CLP is around 3 and the enzyme is inactive toward ferrocenes at pH 5. Therefore, the oxidation kinetics of 1 were measured at pH 2.75 and 25 8C in the presence of Triton X-100, since the protonated acid is not soluble enough in acidic aqueous solution. The steady-state rate of generation of the ferricenium dye as a function of concentration of racemic and enantiomerically pure forms of 1 is shown in Figure 3. The figure shows a Michaelis ╠ Menten-type dependence that is not, however, accomFigure 3. Steady-state rate of panied by enantioselective CLP-catalyzed oxidation of (R)1 (&), (S)-1 (&), and (R,S)-1 (*) discrimination. The maxiby H2O2 (1.4 А 10Ч4 m ) as a funcmal rates Vm,obs and the Mition of concentration of 1: chaelis constants Km,obs with pH 2.75, 25 8C, [CLP] 10Ч7 m. different concentrations of Triton X-100 are summarized in Table 1. The different behavior of HRP and CLP with respect to (R)- and (S)-1 may be accounted for in terms of the pH effect as suggested by the pH-dependent enantioselectivity in the case of HRP for which the kR/kS ratio is almost unity at pH 5.
0947-6539/98/0405-0808 $ 17.50+.50/0 Chem. Eur. J. 1998, 4, No. 5

Figure 1. Steady-state rate of HRP-catalyzed oxidation of (R)-1 and (S)-1 by H2O2 (2 А 10Ч4 m) as a function of concentration of 1: pH 7, 25 8C, [HRP] 10Ч7 m.

previous work,[3] the first-order kinetics associated with 1 emphasize the different reactivities of the R and S enantiomers. Since the reaction is first-order in HRP, the corresponding observed second-order rate constants (k vo/[1][HRP] ) are (4.6 Т 0.4) А 104 and (2.5 Т 0.3) А 104 m Ч1 sЧ1 for (R)-1 and (S)-1, respectively, ( [H2O2] 2 А 10Ч4 m , pH 7, and 25 8C). As expected, the rate constant for the racemate (R,S)-1, (3.8 Т 0.6) А 104 m Ч1 sЧ1, lies between the two values. The effect appears to be strongly pH dependent and the resulting curve is bell-shaped (Figure 2b). The enantioselectivity almost 808

WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

Electron Transfer
Table 1. The values of Vm,obs and Km,obs for the CLP-catalyzed oxidation of 1 and RFc at different concentrations of Triton X-100. Conditions : pH 2.75, 25 8C, [CLP] 10Ч7 m . [Triton X-100] 106 А V [mm] [m s Ч 1 ] 1
[b] m,obs

806 ╠ 813 As for 1, the values Vm,obs for RFc are almost independent of both the nature of ferrocene and Triton X-100 concentration, Table 1. The former contrasts with the HRP case for which the rate decreases strongly with increaseing length of alkyl radical ; the rate decreases by a factor of 32 on going from HFc to BuFc.[3] The independence of Vm,obs on the nature of RFc supports the fact that for CLP the electron transfer from ferrocene is not the rate-limiting step, but rather the interaction of CLP with H2O2 to form the compound CLP ╠ I, Equation (1). Precedents for such a mechanism exist in the literature,[34] and an estimate for k1 is available.[35] Assuming that k1 is rate-limiting, it follows that Vm k1[CLP][H2O2] , and with concentrations of CLP and H2O2 used in this work one arrives at k1 % 105 m Ч1 sЧ1, which is comparable with the rate constant reported previously of 1.1 А 105 m Ч1 sЧ1 (pH 3.4).[35] This coincidence also suggests that k1 refers to rate-limiting steps and the simplest scheme to account for the kinetics observed is given by Equations (1) and (2).
CLP H2O23CLP ╠ I CLP ╠ I(II) RFc3CLP RFc
k
2

106 А V [m sЧ1]

m

[a]

103 А K [m ]

m,obs

50 67.5 85.3 100 50 67.5 100 116 3.25 8.14 16.4 33.1 85.3 16.4 33.1 50.1 67.5 85.3

1.2 Т 0.2 1.6 Т 0.4 1.2 Т 0.2 1.04 Т 0.07 0.4 Т 0.05 0.5 Т 0.05 0.54 Т 0.28 0.54 Т 0.31 1.1 Т 0.2 1.4 Т 0.1 0.93 Т 0.06 1.0 Т 0.1 0.9 Т 0.1 0.44 0.48 0.60 0.63 0.49 Т Т Т Т Т 0.04 0.07 0.09 0.17 0.14

1.3 Т 0.2

8.7 Т 1.18 11 Т 4 15.7 Т 3.0 16.4 Т 1.6 7.2 Т 1.3 13.3 Т 1.5 17 Т 11 18 Т 11 1.0 Т 2.3 Т 3.3 Т 5.2 Т 12 Т 3 3.4 5.4 7.6 14.6 16.5 Т Т Т Т Т 0.3 0.4 0.6 1.4 0.9 2.1 2.4 8.2 8.4

HFc

0.49 Т 0.07

EtFc

1.1 Т 0.2

BuFc

0.53 Т 0.08

k

1

(1)

[a] Mean value. [b] Since no enantioselectivity was found for CLP, each value was calculated from the data for (R)-, (S)-, and (R,S)-1.

(2)

Also, a neutral molecule of 1 is oxidized in the case of CLP. One may speculate that a charged substrate is also a key feature necessary for the enzymatic chiral recognition. It is also possible that the lack of enantioselectivity in the CLP catalysis results from the fact that the electron transfer from ferrocene is not the rate-limiting step under the steady-state conditions. To provide evidence for the latter, we have tested the reactivity of a series of monoalkyl-substituted ferrocenes. CLP-catalyzed oxidation of alkylferrocenes by H2O2 : We investigated the CLP-catalyzed oxidation of HFc, EtFc, and BuFc. The data for EtFc in Figure 4 show that the reaction

We do not specify here which oxidized form of CLP, namely CLP ╠ I or CLP ╠ II (two and one oxidation equivalents above the native state, respectively), contributes to the overall kinetics. What is important is that CLP is not the best enzyme for demonstrating kinetic enantioselectivity under the steadystate conditions. In fact, the resulting rate equation [Eq. (3) ] shows that the key step driven by k2 does not play a role when k1[H2O2] ( k2[RFc] .
dRFc dt k1 k2 E0 H2 O2 0 RFc k1 H2 O2 0 k2 RFc (3)

Here [E]0 and [H2O2]0 are the total concentrations of CLP and hydrogen peroxide ; [RFc] is the concentration of a ferrocene in the aqueous pseudophase. To relate this with the total concentration, one should take into account the fact that ferrocenes bind with micelles of Triton X-100, Equation (4). If we assume that [M]0 ) [RFc]0 (the condition holds in a limited concentration range, [M]0 is the total micelle concentration), we obtain [RFc] [RFc]0/(K4[M]0 1). Substitution into Equation (3) gives Equation (5), which shows that Km,obs should be a linear function of micelle concentration as seen from the data in Table 1 for all ferrocenes tested.
M RFc{M,RFc} dRFc k1 k2 E0 H2 O2 0 RFc0 k1 H2 O2 0 K 4 M0 1 RFc0 k2
K
4

(4) (5)

Figure 4. Steady-state rate of CLP-catalyzed oxidation of EtFc by H2O2 (1.4 А 10Ч4 m ) as a function of concentration of EtFc at different [Triton X100]: pH 2.75, 25 8C, [CLP] 10Ч7 m.

dt

rate levels off on increasing [EtFc] as opposed to the HRP catalysis where a clean first-order behavior with respect to RFc was observed.[3] The Michaelis ╠ Menten equation holds, and the values of Vm,obs and Km,obs with different [Triton X-100] are given in Table 1. The same rate law is valid for ferrocene and n-butylferrocene and the corresponding parameters are also in Table 1.
Chem. Eur. J. 1998, 4, No. 5

GO-catalyzed reduction of (R)- and (S)-1 by b-d -glucose : a spectral study. We have found that the HRP-catalyzed oxidation of 1 into 1 affords a catalytically active material that can then be reduced by GO(red). Thus, a direct spectrophotometric study of the kinetics of this process given by stoichiometric Equation (6) is possible.
GO(red) 2 RFc 3GO(ox) 2 RFc 2 H
0947-6539/98/0405-0809 $ 17.50+.25/0

(6)

WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

809

FULL PAPER
The kinetic data is presented in Figure 5. Evidently, the (S)-1 enantiomer is more reactive than (R)-1. The reactivity order in the reduction reaction is thus reversed compared with the oxidation reaction, that is, the following relative reactivity (R)-1 b (S)1 and (S)-1 b (R)-1 is observed in the HRP- and GOcatalyzed electron-transfer reFigure 5. Steady-state rate of GO-catalyzed reduction of actions, respectively. Another (R)-1 and (S)-1 by b-d -glufeature worth comment in cose (0.1m ) as a function of Figure 5 is that the straight concentration of 1: pH 7, 25 8C, lines do not go through origin. Ч7 [GO] 10 m . We ascribe this effect to the necessity of performing kinetic measurements in the presence of appreciable concentrations of (S)-1 and (R)-1 (ca. 3.6 А 10Ч4 m ), that is, 40 % of the highest concentration of 1 in both series. We have previously shown that ferrocene carboxylic acid is a competitive inhibitor of GO(red) in reactions with Fc.[12] Assuming that 1 can also suppress the enzymatic activity, the lower reaction rate under the condition [1] ) [1] does not seem very surprising. When the excess of 1 is not very large, the reaction follows first-order kinetics with respect to 1. The slopes of the linear portions of the curves are (9.6 Т 0.3) А 10Ч3 and (6.25 Т 0.52) А 10Ч3 sЧ1 for (S)-1 and (R)-1 ferricenium cations, respectively (total [GO] 2 А 10Ч7 m , [d -glucose] 0.05 m , 25 8C, and pH 7). The ratio of 1.54 is evidently the enantioselectivity factor in the GO-catalyzed reaction. It should, however, be pointed out that the ratio must be treated as a rough estimate, since the rate constants were obtained in the presence of (S)-1 and (R)1 in both experimental series, and the two enantiomers may have different ability to suppress the GO activity. In order to minimize the latter effect and to gain additional evidence for the existence of planar chiral enantioselectivity in the GO catalysis, we have estimated the reactivity of (R)- and (S)-1 towards GO(red) by means of cyclic voltammetry (CV).

A. D. Ryabov

Figure line) in and in Glassy

6. Cyclic voltammograms of (S)-1 (solid line) and (R)-1 (dotted aqueous solution of pH 7.0 (0.1m phosphate) in the absence (below) the presence (above) of GO (1.1 А 10Ч6 m) and b-d-glucose (0.1m ). carbon electrode, scan rate 2 mV sЧ1, 25 8C.

which most probably refers to the rate-limiting transfer of the first electron from GO(red) at 1, is shown in Figure 7. From the slopes of the linear plots passing through the origin, defined by 3.17 А (k3RT/F)1/2, values of k3 of (7.4 Т 0.4) А 104 and (4.4 Т 0.2) А 104 m Ч1 sЧ1 were calculated for (S)1 and (R)-1 ferricenium cations, respectively ( [d -glucose] 0.1m , pH 7, and 25 8C). The enantioselectivity factor k(S)-1/ k(R)-1 is 1.7. Experiments carried out with a pyrolytic

GO-catalyzed reduction of (R)- and (S)-1 by b-d -glucose : a CV study. This electrochemical method is very useful to gain information about the coupling between GO(red) and electrochemically generated ferricenium ions.[5] CV data reported in this work was obtained on glassy carbon and pyrolytic graphite electrodes. Cyclic voltammograms for both enantiomers of 1 in the absence and in the presence of GO and b-d -glucose at a glassy carbon electrode are shown in Figure 6. The voltammograms of the both enantiomers in the absence of the enzyme are very similar, whereas in the presence of GO the (S)-1 enantiomer gives rise to a larger peak current provided all other conditions except the enantiomeric nature of 1 are kept constant. The quantitative information on the efficacy of the electron transfer from GO(red) to 1 was obtained by the procedure developed by Bourdillon et al.[8] The typical plot for evaluation of the rate constant k3 (in terms of the formalism adopted in ref. [8] ), 810

Figure 7. Plot for evaluation of the rate constants k3 as described by Bourdillon et al. ,[8] by the example of (S)-1 (*) and (R)-1 (*). For conditions, see legend to Figure 6.

graphite working electrode led to the same conclusions and the values of k3 calculated were (9.3 Т 0.3) А 104 and (5.7 Т 0.4) А 104 m Ч1 sЧ1 for (S)-1 and (R)-1, respectively, with an enantioselectivity factor of 1.6. The increase of the enantioselectivity factors obtained by CV is insignificant compared with that obtained by the spectrophotometric technique.
0947-6539/98/0405-0810 $ 17.50+.50/0 Chem. Eur. J. 1998, 4, No. 5

WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

Electron Transfer However, the CV experiment appears to be more accurate, since on the one hand the reaction mixture does not contain HRP, and on the other the rate constants are evaluated at a low total concentration of 1, which is better for minimizing the inhibition. It should also be pointed out that in the present case, when the reaction is first-order with respect to 1 (see Figure 5), the rate constants evaluated spectrophotometrically and electrochemically show reasonable agreement. The slopes in Figure 5 were obtained at [GO] 10Ч7 m and the corresponding second-order rate constants (9.6 А 104 and 6.2 А 104 m Ч1 sЧ1) are close to those obtained by CV at both electrodes.

806 ╠ 813 enzyme ╠ substrate hydrogen-bond network and/or hydrophobic contacts. The hydrogen bonding within the enzyme active center should be regulated by pH and, therefore, the enantioselectivity appears to be strongly pH sensitive. The network seems to be more efficient in the case of charged ferrocene substrates. A possible detailed mechanistic picture can be provided by the example of GO for which the composition of the active site is known from the X-ray structural data.[36] Figure 8 shows the amino acid residues

Discussion
Three oxidoreductases, horseradish peroxidase (HRP) and chloroperoxidase (CLP) from C. fumago, and glucose oxidase (GO) from A. niger, were screened for the planar chiral enantioselectivity in a single electron transfer processes to and from a 1,2-disubstituted ferrocene derivative. Glucose oxidase was investigated by visible spectrophotometry and cyclic voltammetry. The principal results obtained are summarized in Table 2. The enantioselective electron transfer in
Table 2. Enantioselectivity factors observed for the three oxidoreductases. Enzyme HRP GO (spectral control) GO (CV, pyrolytic graphite) GO (CV, glassy carbon) CLP Enantioselectivity factor ( ( ( ( ( R R R R R )-1/(S)-1 )-1/(S)-1 )-1/(S)-1 )-1/(S)-1 )-1/(S)-1 1.84 0.65 0.625 0.58 1

Figure 8. Modeling of the active site architecture of GO to show the possible binding of (S)-1 in the vicinity of FAD. For details, see text.

the case planar chiral ferrocene derivatives was established for HRP and GO ; no kinetic preference was observed for CLP. It is interesting to note that the compounds that followed first-order kinetics or, in other words, that did not display kinetically meaningful enzyme ╠ substrate binding, showed different reactivity for R and S enantiomers. We have recently put forward the argument that the electron-transfer between HRP and alkylferrocenes has features typical of an outersphere electron-transfer process.[3] Nevertheless, enantioselectivity was observed. It is also interesting to note that the highest enantioselectivity factor of 1.84 observed in the case of HRP is very similar to that reported (1.81) for wild-type cytochrome c peroxidase-catalyzed oxidation of R- and Senantiomers of 1-hydroxymethyl-2-dimethylaminomethylferrocene.[17] Remarkably, the R enantiomers reacted faster in both the cases. The expectation to observe enantioselectivity in the CLP catalysis failed. However, this example might be representative for discussing the origin of enantioselectivity or the lack of it. The fact that enantioselectivity is not observed in this case may be a result of i) the incorrect rate-limiting step, ii) unfavorable pH region, and iii) the lack of charged groups attached to ferrocenes. The second and third issues are most probably interrelated. The charged ferrocene must be properly oriented in the vicinity of the enzyme active center as a result of weak interactions made, for example, by the
Chem. Eur. J. 1998, 4, No. 5

separated from the N5 atom of FAD 600 by about 8 and the (S)-1 molecule incorporated into the architecture of the active site of GO. Tyr-68 has been suggested as a residue involved in the substrate binding.[36] Its aromatic ring can be viewed as a platform to facilitate a hydrophobic or stacking contact with the cyclopentadienyl ring of ferrocene. Located in the active center, histidines 516 and 559 might be involved in the interactions with the carboxylic group of the substrate. Involvement of these amino acid residues is strongly suggested by the shape of the rate versus pH profile for the reduction of ferricenium ion by reduced GO with the maximum of activity in the pH range 7.5 ╠ 8.[12] Thus, the most reactive enantiomer may be fixed in the active site by at least two weak interactions that involve Tyr-68, His-559 and/or His-516. In the case of (R)-1 enantiomer, similar interactions do not appear to be very advantageous because the methyl group would be a natural barrier between the ferricenium ion and reduced FAD. It should also be mentioned that the results of the modeling shown in Figure 8 should be considered as a first level of approximation. The enantioselectivity factors reported are not that high and, alternatively, their origins could stem from as yet unspecified weak interactions of a substrate with surface chiral amino acid residues that are not necessarily very close to the enzyme active site. Another interesting feature, which naturally may just be a coincidence, is the reactivity order in the oxidation and reduction reactions (Table 2). In particular, (R)-1 is more reactive than (S)-1 in the HRP-catalyzed oxidation, whereas (S)-1 is reduced faster than (R)-1 in the presence of GO. However, this does not seem very surprising in light of the fact that the enantioselectivity factor inverts on going from a wildtype to a mutant form of cyctochrome c peroxidase.[17]
0947-6539/98/0405-0811 $ 17.50+.25/0

WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

811

FULL PAPER
Conclusions
In addition to the related previous recent[17, 18] and relevant older work[37] we have presented here several new examples of the enantioselective electron transfer to or from ferrocenes with the elements of planar chirality. HRP and GO proved to be the enzymes with different reactivity toward both enantiomers of 2-methylferrocene carboxylic acid and the corresponding ferricenium cations. CLP did not show kinetic preference for any enantiomer, most likely because of the incorrect rate-limiting step. Taking into account the results reported here and elsewhere it seems likely that the question mark at the end of the title of the paper, ╙Molecular Recognition of Artificial Single-Electron Acceptor CosubА strates by Glucose Oxidase ?╨ by Saveant et al.[19] can now be omitted. Perspectives of this study are clear. First, the results help us to understand the structural factors that bring about the highest stereoselectivity in the enzymatic electron transfer involving planar chiral organometallics and the elucidation of intimate mechanisms of the enantioselectivity. Second, the observation found suggests an approach to the kinetic resolution of planar chiral molecules based on electron transfer to or from oxidoreductases. Although the best enantioselectivity factor equals 1.84, high rates of enzymatic reactions and an easy recycling of partly resolved material makes this approach very challenging.

A. D. Ryabov
over MgSO4 , and the solvent was finally removed in vacuo. 1.25 g (79 %) of the ester was obtained. The product was however contaminated with traces of Me2NCH2Fc. ()-2-Methylferrocene carboxylic acid, (R)-1: 1-Methoxycarbonyl-2-dimethylaminomethylferrocene was converted into the corresponding iodomethylate.[38] A suspension of the iodomethylate (1.7 g) in water (120 mL) was added to sodium amalgam prepared from Hg (7 mL) and Na (2.8 g). The reaction mixture was refluxed for 9 h. The solution was separated from the amalgam and dimethylaminomethylferrocene was extracted with n-hexane. The aqueous layer was acidified with 50 % H3PO4 and the precipitate formed was filtered off. Additional amount of the acid was obtained after its extraction by ether from the acidic aqueous solution. Compound (R)-1 was purified by column chromatography on SiO2 eluting with hexane/ether (5 :1). Yield 0.5 g (52 %) ; [a]D 49.268 (c 2, EtOH), 92.9 % optical purity ; 1H NMR (CDCl3): d 2.3 (s, 3 H ; CH3), 4.18(s, 5 H ; C5H5), 4.30, 4.37, 4.78 (m, 3 H ; C5H3) ; C12H12FeO2 (244.073): calcd C 59.05, H 4.96 ; found C 59.08, H 5.07. (Ч)-2-Methylferrocene carboxylic acid, (S)-1: Compound (S)-1 was obtained in a similar way from ()-N-acetyl-(R)-leucine (provided by Dr. Yu. A. Davidovich) in the asymmetric cyclopalladation : [a]D Ч 42.278 (c 2.37, EtOH), 79.7 % optical purity. Peroxidase reaction : Solutions of (S)-1 and (R)-1 (0.001 m) were prepared by dissolving 0.0099 g (4 А 10Ч5 mol) 1 in phosphate buffer (40 mL, 0.013 m , pH 7). Ferrocene derivative 1 has an absorption maximum at 442 nm (e 182 m Ч1 cmЧ1). The spectral characteristics of its 1e oxidation product were determined by titrating 1 with hydrogen peroxide in the presence of HRP to give l(max) 647.5 nm and e 431 m Ч1 cmЧ1 as described previously.[39] Almost all kinetic data were obtained at [HRP] 1 А 10Ч7 m and [H2O2] 2 А 10Ч4 m. The oxidation reaction was initiated by addition of HRP solution (26 mL, 7.7 А 10Ч6 m ) to a 1 cm spectrophotometric cuvette containing a solution of 1 (1.94 mL) and H2O2 (38 mL, 0.01m ). An increase in absorbance was monitored at 647.5 nm and the data obtained were processed as described in the recent work.[3] The pH dependence of the HRP activity towards (R,S)-1 was studied at a fixed [H2O2] (2 А 10Ч4 m), [1] being in excess. This enabled us to estimate e of the product in the same run. The extinction coefficient appeared to be pH independent in this pH region. CLP oxidation : Owing to the insolubility of ferrocene, n-butylferrocene (Aldrich), ethylferrocene (Strem), and 1 in aqueous solution at pH 2.75, the compounds were dissolved in the phosphate buffer in the presence of Triton X-100 (3 ╠ 116 mm) . Kinetic data were obtained at [CLP] 1 А 10Ч7 m and [H2O2] 1.4 А 10Ч4 m. The kinetics of the CLP-catalyzed oxidation was measured and the data analyzed as described for alkylferrocenes[3] or above for 1. Reduction of (S)-1 and (R)-1 by reduced GO : spectral control. The ferricenium cations were generated in situ by HRP-catalyzed oxidation of 1 by H2O2 . In a typical experiment, solutions of hydrogen peroxide (0.28 mL, 7.4 А 10Ч3 m ) and HRP (0.129 mL, 8.1 А 10Ч6 m ) were added to 1 (10 mL, 0.001m) in phosphate buffer (0.01 m, pH 7). Such an optimal reagent ratio provided 60 % yield of 1. The oxidation was registered spectrophotometrically by following a decrease in absorbance at 647 nm. The GOcatalyzed reduction of (S)-1 and (R)-1 was monitored by following the protocol described below. The reaction mixture was composed by addition of solutions of d -glucose (0.18 mL, 0.7 m), GO (0.02 mL, 2 А 10Ч5 m ), and 1 (0.3 ╠ 1.8 mL) to a 1 cm spectrophotometric cuvette with phosphate buffer (0.01m, pH 7) to achieve the total volume of 2 mL. Since the solution of 1 always contained 40 % of 1, which is known to be a competitive inhibitor of GO,[12] the required amount of 1 was added, when necessary, to the reaction solution to achieve constant [1] 3.6 А 10Ч3 m in the whole series. The rate of 1 fading at 647 nm was registered every 10 s over a peroid of 1 min. It has been confirmed that HRP has no effect on the GO-catalyzed reaction under the conditions used. Reduction of (S)-1 and (R)-1 by reduced GO : a CV study. Stock solutions of 1 (1.1 А 10Ч3 m ) were prepared in phosphate buffer (0.1m, pH 7). An aqueous solution of b-d-glucose (1m) was prepared beforehand and kept overnight. For the measurement of i o the protocol was standp ardized as follows : a buffered solution of 1 (2.7 mL) and glucose aqueous solution (0.3 mL, 0.1m) were introduced into the thermostated electrochemical cell. The cyclic voltammograms were recorded with glassy carbon and pyrolytic graphite electrodes at seven to nine different scan rates from
0947-6539/98/0405-0812 $ 17.50+.50/0 Chem. Eur. J. 1998, 4, No. 5

Experimental Section
General : Enzymes HRP (R/Z 3), CLP (R/Z 1), and GO from Aspergillus niger (250 U per mg) were purchased from Dia-M, Sigma, and Serva, respectively, and all used as received. Spectrophotometric measurements were carried on a Shimadzu UV-160A spectrophotometer equipped with a CPS-240A cell positioner/temperature controller. CV measurements were performed on a PC-interfaced potentiostat-galvanostat IPC-3 (Institute of Physical Chemistry, RAS). A three-electrode scheme was used with working glassy carbon (Moscow State University (Russia), diameter 1.8 mm) and pyrolytic graphite electrodes (Tokaii (Japan), diameter 1.8 mm), saturated calomel reference electrode, and ancillary Pt electrode. An electrochemical cell was thermostated at 25 8C by circulating water. The opticatl rotation measurements were carried out at 22 8C. Preparation of (R)-1. Synthesis of ()-2 : A solution of (Ч)-N-acetyl-(S)-leucine (Reakhim, 3.41 g, 0.02 mol) and NaOH (0.8 g) in water (75 mL) was added to a solution of Na2PdCl4 (5.85 g, 0.02 mol) in MeOH (225 mL), and the pH of the resulting solution was adjusted to 7.85 with 50 % NaOH. A solution of dimethylaminomethylferrocene (4.9 g, 0.02 mol) in MeOH (75 mL) was then added to the stirred solution and a precipitate began to appear after 10 min. The mixture was allowed to stand overnight, the precipitate was then filtered off, washed with water, and dried over P2O5 . The solid was dissolved in benzene, and the insoluble admixtures were removed by filtration. n-Heptane was added to the filtrate until a slight clouding was evident, the mixture was concentrated in vacuo, n-heptane was again added, and the solution filtered. The filtrate was evaporated almost to dryness, the precipitate was separated and dried to obtain 3.34 g of ()-(R)2 [a]D 537.68, (c 1, CH2Cl2), 80.7 % optical purity. The same procedure was applied to the methanolic mother liquor to obtain 1.86 g of ()(R)-2 [a]D 4638, (c 1, CH2Cl2), 69.5 % optical purity. Total yield 67 %. 1-Methoxycarbonyl-2-dimethylaminomethylferrocene : Carbon monoxide was bubbled for 1 h through a suspension of ()-(R)-2 (2.0 g, 0.005 mol, [a]D 537.68) in methanol (50 mL). Palladium metal was filtered off, the filtrate treated with sodium bicarbonate, extracted with diethyl ether, dried

812

WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

Electron Transfer
0.6 to 50 mV sЧ1. For the measurement of ip , GO solution (0.01 mL, its final concentration in the cell was 1.1 А 10Ч6 m ) was added to the electrochemical cell and cyclic voltammograms were recorded as previously described.[10] Both carbon electrodes were polished by 0 ╠ 1 mm diamond paste before every series of measurements consisting of 7 ╠ 9 runs either with or without GO and d -glucose. It was confirmed that a higher frequency of polishing does not affect both the shape of voltammograms and peak currents. The rate constant k3 was calculated by the described procedure.[8] Acknowledgments : The research described in this publication was made possible in part by financial support from the Russian Foundation for Fundamental Research (Grant No 96-03-34328a). Received : August 12, 1997 [F 792]

806 ╠ 813
Ю [16] G. Jaouen, A. Vessieres, I. S. Butler, Acc. Chem. Res. 1993, 26, 361 ╠ 369. [17] S. J. Sadeghi, G. Gilardi, G. Nicolosi, A. E. G. Cass, Chem. Commun. 1997, 517 ╠ 518. [18] S. Marx-Tibbon, E. Katz, I. Willner, J. Am. Chem. Soc. 1995, 117, 9925 ╠ 9926. А [19] P. Alzari, N. Anicet, C. Bourdillon, J. Moiroux, J.-M. Saveant, J. Am. Chem. Soc. 1996, 118, 6788 ╠ 6789. [20] V. I. Sokolov, L. L. Troitskaya, O. A. Reutov, Dokl. Akad. Nauk SSSR 1979, 246, 124 ╠ 127. [21] V. I. Sokolov, L. L. Troitskaya, O. A. Reutov, J. Organomet. Chem. 1979, 182, 537 ╠ 546. [22] V. I. Sokolov, L. L. Troitskaya, L. A. Bulygina, P. V. Petrovski, Metalloorg. Khim. 1989, 2, 356 ╠ 359. [23] V. I. Sokolov, J. Organomet. Chem. 1995, 500, 299 ╠ 306. [24] V. I. Sokolov, L. L. Troitskaya, Chimia 1978, 32, 122 ╠ 123. [25] R. L. Halterman, Chem. Rev. 1992, 92, 965 ╠ 994. [26] O. Riant, O. Samuel, H. B. Kagan, J. Am. Chem. Soc. 1993, 115, 5835 ╠ 5836. [27] A. Togni, Angew. Chem. 1996, 108, 1581 ╠ 1583 ; Angew. Chem. Int. Ed. Engl. 1996, 35, 1475 ╠ 1477. [28] A. D. Ryabov, Synthesis 1985, 233 ╠ 252. [29] A. L. Pendin, P. K. Leontevskaya, E. I. Lvova, B. P. Nikolsky, Dokl. Akad. Nauk SSSR 1969, 189, 115 ╠ 118. [30] A. N. Shevelkova, Y. I. Salnikov, N. L. Kuzmina, A. D. Ryabov, FEBS. Lett. 1996, 383, 259 ╠ 263. [31] S. G. Allenmark, M. A. Andersson, Tetrahedron : Asymmetry 1996, 7, 1089 ╠ 1094. [32] A. Dexter, F. J. Lakner, R. A. Campbell, L. P. Hager, J. Am. Chem. Soc. 1995, 117, 6412 ╠ 6413. [33] A. Zaks, D. R. Dodds, J. Am. Chem. Soc. 1995, 117, 10 419 ╠ 10 424. [34] R. D. Libby, N. S. Rotberg, J. Biol. Chem. 1990, 265, 14 808 ╠ 14 811. [35] J. A. Thomas, D. R. Morris, L. P. Hager, J. Biol. Chem. 1970, 245, 3129 ╠ 3134. [36] H. J. Hecht, H. M. Kalisz, J. Hendle, R. D. Schmid, D. Schomburg, J. Mol. Biol. 1993, 229, 153 ╠ 161. [37] K. Bernauer, J.-J. Sauvain, J. Chem. Soc. Chem. Commun. 1988, 353 ╠ 354. [38] G. W. Gokel, D. Marquarding, I. K. Ugi, J. Org. Chem. 1972, 37, 3052 ╠ 3058. [39] V. N. Goral, M. I. Nelen, A. D. Ryabov, Anal. Lett. 1995, 28, 2139 ╠ 2149.

[1] A. D. Ryabov, Angew. Chem. 1991, 103, 945 ╠ 955 ; Angew. Chem. Int. Ed. Engl. 1991, 30, 931 ╠ 941. [2] A. D. Ryabov, Ross. Khim. Zh. 1995, 137 ╠ 144. [3] A. D. Ryabov, V. N. Goral, J. Biol. Inorg. Chem. 1997, 2, 182 ╠ 190. [4] V. N. Goral, A. D. Ryabov, Biochem. Mol. Biol. Int. 1998, in press. [5] A. E. G. Cass, G. Davis, G. D. Francis, H. A. O. Hill, W. J. Aston, I. J. Higgins, E. O. Plotkin, L. D. L. Scott, A. P. F. Turner, Anal. Chem. 1984, 56, 667 ╠ 671. [6] A. Heller, Acc. Chem. Res. 1990, 23, 128 ╠ 134. [7] A. Heller, J. Phys. Chem. 1992, 96, 3579 ╠ 3587. А [8] C. Bourdillon, C. Demaille, J. Moiroux, J.╠M. Saveant, J. Am. Chem. Soc. 1993, 115, 2 ╠ 10. [9] C. Deshaies, J. Chopineau, J. Moiroux, C. Bourdillon, J. Phys. Chem. 1996, 100, 5063 ╠ 5069. [10] A. D. Ryabov, A. Amon, R. K. Gorbatova, E. S. Ryabova, B. B. Gnedenko, J. Phys. Chem. 1995, 99, 14 072 ╠ 14 077. [11] A. D. Ryabov, A. M. Trushkin, L. I. Baksheeva, R. K. Gorbatova, I. V. Kubrakova, V. V. Mozhaev, B. B. Gnedenko, A. V. Levashov, Angew. Chem. 1992, 104, 788 ╠ 790 ; Angew. Chem. Int. Ed. Engl. 1992, 31, 789 ╠ 790. [12] A. D. Ryabov, Y. N. Firsova, M. I. Nelen, Appl. Biochem. Biotechnol. 1996, 61, 25 ╠ 37. [13] V. I. Sokolov, L. L. Troitskaya, O. A. Reutov, Dokl. Akad. Nauk SSSR 1977, 237, 1376 ╠ 1379. [14] V. I. Sokolov, L. L. Troitskaya, T. I. Rozhkova, Gazz. Chim. Ital. 1987, 117, 525 ╠ 527. [15] Y. Yamazaki, M. Uebayasi, V. Hosono, Eur. J. Biochem. 1989, 184, 671 ╠ 680.

Chem. Eur. J. 1998, 4, No. 5

WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998

0947-6539/98/0405-0813 $ 17.50+.25/0

813