7700
J. Phys. Chem. B 2007, 111, 7700-7710
Theoretical Study of N-Demethylation of Substituted N,N-Dimethylanilines by Cytochrome P450: The Mechanistic Significance of Kinetic Isotope Effect Profiles Yong Wang,† Devesh Kumar,‡ Chuanlu Yang,†,§ Keli Han,*,†,§ and Sason Shaik*,‡ State Key Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, People’s Republic of China, Institute of Chemistry and the Lise Meitner-MinerVa Center for Computational Quantum Chemistry, The Hebrew UniVersity of Jerusalem, 91904 Jerusalem, Israel, and Department of Physics, Yantai Normal UniVersity, Yantai 264025, People’s Republic of China ReceiVed: March 25, 2007
The mechanism of N-demethylation of N,N-dimethylanilines (DMAs) by cytochrome P450, a highly debated topic in mechanistic bioinorganic chemistry (Karki, S. B.; Dinnocenczo, J. P.; Jones, J. P.; Korzekwa, K. R. J. Am. Chem. Soc. 1995, 117, 3657), is studied here using DFT calculations of the reactions of the active species of the enzyme, Compound I (Cpd I), with four para-(H, Cl, CN, NO2) substituted DMAs. The calculations resolve mechanistic controversies, offer a consistent mechanistic view, and reveal the following features: (a) the reaction pathways involve C-H hydroxylation by Cpd I followed by a nonenzymatic carbinolamine decomposition. (b) C-H hydroxylation is initiated by a hydrogen atom transfer (HAT) step that possesses a “polar” character. As such, the HAT energy barriers correlate with the energy level of the HOMO of the DMAs. (c) The series exhibits a switch from spin-selective reactivity for DMA and p-ClDMA to two-state reactivity, with low- and high-spin states, for p-CN-DMA and p-NO2-DMA. (d) The computed kinetic isotope effect profiles (KIEPs) for these scenarios match the experimentally determined KIEPs. Theory further shows that the KIEs and TS structures vary in a manner predicted by the MelanderWestheimer postulate: as the substituent becomes more electron withdrawing, the TS is shifted to a later position along the H-transfer coordinate and the corresponding KIEs increases. (e) The generated carbinolaniline can readily dissociate from the heme and decomposes in a nonenzymatic environment, which involves water assisted proton shift.
1. Introduction The ubiquitous cytochrome P450 (CYP450) enzymes, are considered to be among “the most versatile biological catalysts known”,1 owing to their vital roles in the metabolism of xenobiotics, such as drugs, and in the biosynthesis of endogenous compounds, such as steroids. A wide variety of chemical transformations including hydroxylation of inert C-H bonds, epoxidation of CdC bonds, dealkylations, etc., are all catalyzed by these monooxygenase enzymes. Despite the extensive mechanistic explorations of these processes by experimental means, there remain many intriguing questions on which theory can offer the missing insight and reveal new features. This is precisely the case for the mechanism of the amine N-dealkylation mediated by cytochrome P450,2 which is targeted in the present theoretical study. The overall reaction of CYP450-catalyzed N-dealkylation of amines involves two processes as outlined in Scheme 1. The initial process, catalyzed by CYP450, involves the conversion of the amine to the corresponding carbinolamine (i, iv, or iiiv in Scheme 1), and the second one, thought to be nonenzymatic, involves decomposition of the carbinolamine to a secondary amine and formaldehyde (process v in Scheme 1). The mechanistic debates concern the initial process, with two * To whom correspondence should be addressed.
[email protected] (K.H.);
[email protected] (S.S.). † Chinese Academy of Sciences. ‡ The Hebrew University of Jerusalem. § Yantai Normal University.
E-mail:
SCHEME 1: Alternative Mechanistic Hypotheses for N-Demethylation of N,N-Dimethylanilines
alternative mechanisms that have survived more than 30 years of discourse and are shown in i vs ii-iii in Scheme 1. One is the single electron transfer (SET) mechanism,3 involving an initial one-electron oxidation of the amine, followed by deprotonation of the cation radical to produce an R-amino radical (ii-iii in Scheme 1), which rebounds to produce the ferriccarbinolamine complex that follows onward with decomposition of the carbinolamine. The second mechanism (i in Scheme 1) begins with direct hydrogen atom transfer (HAT)4 and then proceeds via the same series of steps.
10.1021/jp072347v CCC: $37.00 © 2007 American Chemical Society Published on Web 06/09/2007
N-Demethylation of Substituted DMAs The evidence for SET has been mainly based on the following experimental observations: (a) the oxidative dealkylations generally exhibit low deuterium kinetic isotope effect (KIE);5 (b) the free energy relationship of N-demethylation for a series of para-substituted N,N-dimethylanilines by CYP450 and its model analogs features large negative Hammett F value, and the reaction rates correlate with substrate redox potentials E1/2;6-7 and (c) 1,4-dihydropyridine Hantzsch esters8 and cyclopropylamines9 are suicide inactivators of CYP450. The HAT mechanism is supported by the following facts: (a) the KIE profile (KIEP), found by Dinnocenzo and coworkers for a series of DMA oxidation reactions by cytochrome P450s, correlates linearly with the KIEP measured in a genuine HAT by the t-butoxyl radical reacting with the same DMA series;4 (b) Model electrochemical-mass spectrometric studies by Jurva et al.10 showed that amino radical cations were unlikely to be obligatory intermediates in the P450-catalyzed oxidations of N-cyclopropyl cyclic tertiary allylamines; (c) Bhakta and Wimalasena11 reported that the CR-H’s of both cyclopropyl and alkyl substituents were removed at isotopically sensitive steps without producing detectable amounts of cyclopropyl ring opened products; (d) Hanzlik et al.12 demonstrated that the CYP450-catalyzed oxidation of N-benzyl-N-cyclopropylamine fitted only a HAT mechanism; (e) DFT calculations by Li et al.13 on the C-H hydroxylation of DMA oxidation supported a HAT mechanism. However, the DFT calculations showed also significant differences between the C-H hydroxylation of DMA vis-a`-vis alkanes. It is apparent from the above introduction that the mechanism of DMA dealkylations poses some challenging puzzles. On the one hand, the facts that the reaction rates are related to the redox potentials of the amines2a,6d and that a large negative Hammett F parameter is observed in free energy relationship studies6a,c could be taken to indicate a SET mechanism. On the other hand, the extended correlation of the KIEPs for the oxidations catalyzed by CYP450 and the HAT reactions with the t-BuO• radical, favors a HAT mechanism for the CYP450-catalyzed DMA oxidations. Therefore, one obvious question to answer is: HAT or SET? Second, since theory was found13 to favor a HAT mechanism for DMA oxidation by P450, why does this HAT proceed in a spin-selective manner (SSM), whereas the HAT in the alkane hydroxylation proceeds by two-state reactivity (TSR)?14a-b If DMA oxidation involves indeed initial HAT, why then is there a good correlation of the reaction rates with the substrate redox potential, and what is the cause for the large negative Hammett F parameters observed in free energy relationship studies? Can theory reconstruct the KIEP found by Dinnocenzo and co-workers4b,d in the series of p-substituted DMAs? Since Li et al.13 have reported that higher energy was required for a SET process than a HAT process, for the best electron donor in the series, DMA, thus the SET mechanism was excluded. The present work tries to answer the aforementioned questions to resolve mechanistic controversies and to offer a consistent mechanistic view, by use of density functional theoretic (DFT) calculations of the P450 oxidation reactions of the series of p-X-C6H4N(CH3)2 (X ) H, Cl, CN, and NO2) in Scheme 2. SCHEME 2: Net Reaction of Oxidation of N,N-Dimethylanilines by P450
J. Phys. Chem. B, Vol. 111, No. 26, 2007 7701 2. Computational Methodology A recent study by Dowers et al. provided compelling evidence for the reactivity of Compound I (Cpd I).15 Therefore, to model Cpd I of CYP450, we employed a six-coordinate oxo-ferryl species, Fe4+O2-(C20N4H12)-(SH)- 14b,c,16 and used it with the series of para (H, Cl, CN, NO2) substituted DMAs in Scheme 2. The DFT calculations were carried out with the Gaussian 03 suit of programs.17 The spin-unrestricted B3LYP18 was employed using two basis sets: (a) LACVP(Fe)/6-31G(H, C, N, O, S, Cl), henceforth B1, was used to optimize transition states and minima without symmetry constraint. (b) A singlepoint energy was performed with a higher basis set, LACV3P+*(Fe)/6-311+G*(the rest), B2 in brief.19 The model and computational method were tested extensively and proved to be reliable.2c,d,13 Transition states were ascertained by vibrational frequency analysis to possess only one mode with an imaginary frequency. Bulk polarity effects of the active site in the protein environment were evaluated with the PCM solvation model using a nonpolar solvent, chlorobenzene ( ) 5.62). The carbinolamine decomposition in the polar nonenzymatic environment was studied using aqueous medium (water, ) 78.39). The effect of hydrogen bonding to the sulfur ligand was mimicked by adding two ammonia molecules that point toward the sulfur of the proximal ligand at fixed distances of rS-HN ) 2.660 Å.13,14c All of the energy corrections due to solvent and NH-S hydrogen bonds were evaluated with B2. All of the data are collected in the Supporting Information (SI) document, whereas the text discusses the highest-level results, UB3LYP/B2//B1. Thus, the single-point energy B2//B1 values will refer to hereafter as E1, whereas E2 will include in addition also the zero-point energy (ZPE) correction, the NH-S hydrogen bonding and the bulk polarity effects ( ) 5.6), whereas E3 will correspond to the case with ) 78.39.13,14c The kinetic isotope effect (KIE) of the HAT process was determined using the Gaussian frequency data based on two expressions.20 The first one is the semi-classical Eyring equation, where the KIE is given as
()
[
]
kH (G#H - GRH) - (G#D - GRD) ) exp kD s RT
(1)
The second expression uses a simple “quantum correction” by multiplying the semiclassical (kH/kD)s with the Wigner quantum correction Qcorr factor
() kH kD
w
) Qcorr w
() kH kD
(2)
s
2 2 Qcorr w ) (1 + ut /24)/(1 + u′t /24)
(3)
with ut ) hνH/kBT and ut′ ) hνD′/kBT, where ν is the imaginary frequency of the transition state. Detailed formulation and the corresponding Fortran codes are presented in the Supporting Information (SI). 3. Results and Discussion 3.1. Substituent Effects on Reaction Mechanisms and the Origins of Spin-Selective Reactivity. Figure 1 shows the calculated energy profiles for the formation of carbinolanilines for the pristine N,N-dimethylaniline substrate (Figure 1a) and its three para substituted derivatives (Figure 1b-d), whereas the geometric information of the corresponding 2,4TSH species
7702 J. Phys. Chem. B, Vol. 111, No. 26, 2007
Wang et al.
Figure 1. Energy profiles (in kcal/mol) for C-H hydroxylation of p-(H, Cl, CN, NO2)-DMAs by Cpd I on the quartet and doublet states. Relative energies are given at the E1 (UB3LYP/B2//B1) level out of parentheses and the E2 (UB3LYP/B2//B1; ZPE, NH-S, ) 5.6) level in parentheses.
is displayed in Figure 2. As usual for P450 reactions, the mechanism involves two states nascent from the degenerate states of Cpd I. In both states, the rate-determining step involves C-H bond activation, namely a HAT process followed by the usual O-rebound21 to form the carbinolaniline complexes, 2,4PC. The barrier data in the figure correspond to UB3LYP/B2//B1 values (E1 and E2), whereas the corresponding UB3LYP/B1// B1 data are given in the SI document (Tables S1-S4). Figure 1a shows the energy profile for C-H hydroxylation of the pristine DMA. At the E1 level, the barrier difference is within 1 kcal/mol for the two states, 7.7/8.6 kcal/mol for the LS/HS states, respectively. When the ZPE correction, the bulk polarity effects, and the NH‚‚‚SH hydrogen bond interactions are incorporated (E2 level), the energy difference between 2TSH and 4TSH increases to about 3.0 kcal/mol, which means that reaction should proceed in a spin-selective manner (SSM), predominantly via the LS reaction route, as found before.13 The TSH geometries (Figure 2a) reveal that the LS species is earlier than the HS on the H-transfer coordinate. Large spin density accumulation along with nearly zero charge on the generated organic moiety in the HS intermediate,4IM (see Table S1 in the SI), provide further support for a HAT process. The LS route is concerted and follows to the ferric-carbinolaniline complex without a distinct rebound step. The high-spin 4IM species behave like a shoulder and not a genuine minimum and fall in a barrier-free rebound to the corresponding ferric-carbinolaniline complex. Thus, on the two states, the C-H hydroxylation is an effectively concerted HAT process. These results are in agreement with a recent work by Li et al.13 and follow earlier predictions.14b For the remaining substrates, p-Cl-DMA, p-CN-DMA, and p-NO2-DMA (Figure 1b-d), the reactions proceed similarly, with an initial HAT followed by a barrier-free rebound to form the corresponding ferric-carbinolaniline complexes. The para substituent induces differences in the transition state structures (Figure 2b-d) and in the energy barriers (Figure 1b-d), but it does not change the “chemical mechanism” as such. From Figure 1a-d, we can see that the activation energies (both E1
and E2) for H-abstraction keep increasing as the electronwithdrawing power of the substituent becomes stronger. At the E1 level, the LS/HS activation energies are 7.7/8.6 kcal/mol for DMA, 8.9/10.2 kcal/mol for p-Cl-DMA, 12.0/13.2 kcal/mol for p-CN-DMA, and 13.5/14.4 kcal/mol for p-NO2-DMA. At the E2 level, the activation energies are 3.6/6.1 kcal/mol for DMA, 3.8/6.3 kcal/mol for p-Cl-DMA, 7.5/8.4 kcal/mol for p-CN-DMA, and 8.8/9.2 kcal/mol for p-NO2-DMA. The energy gaps between 2TSH and 4TSH are high (ca. 3 kcal/mol) for DMA and p-Cl-DMA and as low as 0.1-0.8 for the other two substrates. For these four reactions, the ratios of the reaction rates on the LS:HS route are DMA, 57.8:1; p-Cl-DMA, 57.8:1; p-CN-DMA, 4.3:1; p-NO2-DMA, 0.5:1. Therefore, with the large gaps for DMA and p-Cl-DMA, the mechanism is SSM; however, with the small gaps for p-CN-DMA and p-NO2-DMA, the mechanism is the traditional TSR.14,16 As a whole, there is a mechanistic switch from SSM to TSR in these four oxidation processes of the substituted DMAs. This trend also exists even when entropic contributions are taken into account. Table 1 shows that the corrected energy gaps for DMA and p-Cl-DMA are 1.9∼2.2 kcal/mol, whereas those for p-CN-DMA and p-NO2DMA are as small as 0.1∼0.4 kcal/mol. Table 2 presents further information on the roots of this mechanistic switch. At the UB3LYP/LACVP level, the energy gap decreases gradually, starting with 1.7 and down to 0.9 kcal/ mol, when the substrate changes in the series DMA, p-Cl-DMA, p-CN-DMA, and p-NO2-DMA. ZPE corrections decrease the energy gaps for all these four substrates. At the higher E1 level, the energy differences between 2TSH and 4TSH are 1.0∼1.3 kcal/ mol. However, taking into account the protein bulk polarity effects produces a quantitative difference between, on the one hand, DMA and p-Cl-DMA and, on the other, p-CN-DMA and p-NO2-DMA oxidations. Now the barrier difference between the spin states is different for the two subgroups, 2.6-3.3 kcal/ mol for the DMA and p-Cl-DMA and half as much, 1.1-1.9 kcal/mol, for p-CN-DMA and p-NO2-DMA. Further addition of the ZPE correction and the effect of NH‚‚‚S H-bond interaction sharpen the division between the two pairs of
N-Demethylation of Substituted DMAs
J. Phys. Chem. B, Vol. 111, No. 26, 2007 7703 TABLE 2: Breakdown of the Contributions of the Activation Energies (in kcal/mol) for the H-Abstraction Reactions of p-(H, Cl, CN, NO2)-DMAs with Cpd I DMA
p-Cl-DMA
p-CN-DMA
p-NO2-DMA
9.8 8.2
(a) UB3LYP/B1 11.8 14.2 10.4 13.0
16.9 16.0
4
TSH 2 TSH
6.8 5.8
(b) UB3LYP/B1+ZPE 8.7 11.1 7.9 10.3
13.3 12.7
4
TSH 2TS H
8.6 7.6
(c) UB3LYP/B2//B1 (E1) 10.2 13.2 8.9 12.1
14.4 13.4
4TS
7.3 4.7
(d) UB3LYP/B2//B1 + Bulk Polarity 7.8 10.8 4.5 8.9
11.9 10.8
4TS
H
2
TSH
H
2
TSH
4
TSH
2TS
4TS
H
H
2
TSH
(e) UB3LYP/B2//B1 + Bulk Polarity + NH-S H-Bonding 9.1 9.4 11.5 5.6 5.8 10.2
12.4 12.3
(f) UB3LYP/B2//B1 + Bulk Polarity + NH-S H-Bonding + ZPE (E2) 6.1 6.3 8.4 3.1 3.3 7.6
8.8 8.9
TABLE 3: Dipole Moments (in Debye) of the 2,4TSH Species for the H-Abstraction Reactions of p-(H, Cl, CN, NO2)-DMAs with Cpd I 2TS
H
4TS
H
∆µ (D)
Figure 2. Key geometric features (lengths in Å and angles in degree) of UB3LYP/LACVP optimized 2,4TSH species for C-H hydroxylation of p-(H, Cl, CN, NO2)-DMAs with Cpd I.
TABLE 1: Relative Free Energies (in kcal/mol) for the H-Abstraction Reactions of p-(H, Cl, CN, NO2)-DMAs with Cpd I 4RC 2RC 4TS
H
2TS
H
∆EHS ∆ELS
DMA
p-Cl-DMA
p-CN-DMA
p-NO2-DMA
0.0 -0.8 9.1 6.4 9.1 7.2
0.0 -0.4 10.6 8.0 10.6 8.4
0.0 -0.9 11.5 11.0 11.5 11.9
0.0 -0.3 10.6 10.7 10.6 11.0
substrates. Now, in the first two reactions, the barrier on the LS surface is significantly smaller than the one on the high spin surface, by ca. 3 kcal/mol. By contrast, in the latter two cases, the difference is small, ca. 0.1∼0.8 kcal/mol.
DMA
Cl-DMA
CN-DMA
NO2-DMA
7.5613 6.4658 1.0955
6.9605 6.1014 0.8591
7.6924 7.2511 0.4413
8.1441 8.0968 0.0473
These barrier differences that are caused by the external effects (bulk polarity and NH-S hydrogen bonding) are not mere artifacts and arise from the electric properties of the TSs. Table 3 makes this point by showing the dipole moments for 2,4TS . It is seen that dipole moments of 2TS are larger than H H those of 4TSH for all four DMAs, which means that 2TSH will be more stabilized than 4TSH by bulk polarity. However, as the para substituent becomes more electron-withdrawing, the dipole moment difference is reduced and for p-NO2-DMA, the difference is just 0.0473 Debye. These substituent-dependent dipole moments account at least in part for the finding that at the E2 level, the energy gap between 2TSH and 4TSH for the DMA and p-Cl-DMA oxidations is as large as ca. 3.0 kcal/mol, whereas for p-CN-DMA and p-NO2-DMA oxidations, the difference is small, 0.1∼0.8 kcal/mol. The structures of the 2,4TSH displayed in Figure 2 show some interesting variations along the H-transfer coordinate, C-HO. The 2TSH species of the pristine DMA is “early” with a rather short C-H distance and a long H-O distance. As the para substituent becomes more electron withdrawing, the corresponding 2TSH species progresses along the H-transfer coordinate and becomes less and less early; in the entire series, the C-H distances change from 1.222 to 1.249 Å, whereas the H-O distances change from 1.375 to 1.310 Å. However, the 4TSH species of the pristine DMA is more “central” with a long C-H of 1.318 Å, and as the para substituent on DMA becomes more electron withdrawing the transition state lies increasingly “later” along the H-transfer coordinate; in the entire series the C-H distances change from 1.318 to 1.341 Å, and the H-O distances change from 1.247 to 1.214 Å. Interestingly, the barriers for both the LS and the HS processes increase as the C-H bond undergoes more stretching in the TS. In short, the transition states of P450-catalyzed oxidation of substituted DMAs are all
7704 J. Phys. Chem. B, Vol. 111, No. 26, 2007
Wang et al.
Figure 3. Plot of Mulliken spin density on the N atom of 2,4TSH vs Hammett substituent constants σpara (from ref 22) of the p-X (X ) H, Cl, CN, NO2) substituents.
asymmetric, which results in relativly low KIE values, following the Melander-Westheimer postulate and as originally reasoned by Dinnocenzo and co-workers.4 To obtain some insights into the changes in geometries and barriers, we plotted in Figure 3 the Mulliken spin density on the N atom of the TSH species against the Hammett σ parameter of the four para substituents.22 It is seen that as the substituent becomes more electron-withdrawing, the spin density on N decreases. This in turn means that, as electron-withdrawing power of the substituent increases, the lone pair on the nitrogen is “sucked” into the phenyl ring and becomes less accessible for delocalization toward the developing radial center on the carbon atom of the C-H moiety. This impaired delocalization causes in turn destabilization of the TSs for the cases with the more electron-withdrawing substituents and requires a higher degree of C-H bond cleavage. Consequently, along the series, as the substituent becomes more electron withdrawing, the barriers increase and the corresponding TSH species lie later on the H-transfer coordinate. Thus, the various computational sets of the data are consistent and are based on clear physical effects. As we already mentioned, the free energy relationship of a series of para-substituted N,N-dimethylaniline oxidations by CYP450 and its model analogs feature a large negative Hammett F value,6a,c,7 which indicates that the rate-determining step is facilitated by electron-donating substituents,6a as expected if positive charge accumulates on the substrate during the reaction.2a This fact was originally used as evidence for the operation of a SET mechanism. However, our calculated results demonstrate
Figure 4. Some high lying occupied orbitals showing electronic delocalization in the 2,4TSH species, in the H-abstraction reactions of p-(H, Cl, CN, NO2)-DMAs with Cpd I.
the mechanism is HAT, and one must therefore find an alternative explanation for the observed free energy relationships. A few molecular orbitals shown in Figure 4 provide some insight into the nature of the transition state. The C-H hybrid orbital is coupled with the lone pair p orbital of N so as to form a πC-N orbital, which is in turn conjugated with the π orbital on the phenyl group and forms a conjugated πph-πC-N orbital; this is the reason why spin density was found above on the nitrogen (other sets of delocalized orbitals, which reveal this conjugation are given in the SI document). By virtue of this conjugative interaction, the electronic features of the
N-Demethylation of Substituted DMAs
J. Phys. Chem. B, Vol. 111, No. 26, 2007 7705
Figure 5. Plot of -Ea/(ln 10RT) against Hammett substituent constants σpara (from ref 22) of p-X (X ) H, Cl, CN, NO2) substituents.
substituents exert influence on the nature of the CR-H bond. Electron-withdrawing substituents will weaken this electron delocalization and will decrease the charge transfer from the substrate to Cpd I. The charges on the p-X-Ph-N-(CH3)(CH2) moieties are listed in Tables S1-S4 in the SI, which illustrate that the accumulated positive charges on TSs are 0.10-0.13 for Ph-N-(CH3)(CH2) and p-Cl-Ph-N-(CH3)(CH2), 0.070.11 for p-CN-Ph-N-(CH3)(CH2), and 0.03-0.06 for p-NO2Ph-N-(CH3)(CH2). These diminishing charges are precisely in line with the conclusions reached above on the role of the substituent on the ability of the N-lone pair to stabilize the developing radical-center (Figure 3). Let us further elaborate this point by reproducing the negative Hammett slope observed in the experimental free energy relationship of log Vmax against the Hammett parameter σpara. According to the classic Arrhenius formula, the rate constant for the catalysts is given by
( )
kcat ) A exp -
Ea RT
(4)
Since -Ea/RT ) ln kcat - ln A, and since the reaction rate kcat approximates to Vmax (of the Michaelis-Menten equation), the following relationship holds: log Vmax approximates to -Ea/ln 10RT. Figure 5 shows the plot of the -Ea/ln 10RT variable against the Hammett σ constants of the substituents. One can see a linear correlation with a negative slope (quartet, -5.4;
Figure 6. Plots of (a) the substrate redox potentials E1/2 (from ref 6d) against the HOMO energy levels of p-X-DMAs (X ) H, Cl, CN, NO2); (b and c) the activation energies, Ea, for the HS and LS mechanisms, against the HOMO energy levels of p-X-DMAs (X ) H, Cl, CN, NO2).
double,: -2.8) that corresponds to a large negative F value, as reported experimentally.6a,6c However, since our calculations show a HAT process with a development of a radical on the substrate, the observed large negative Hammett F parameter cannot be used as evidence for a SET mechanism. Furthermore, the conjugated πph-πC-N orbital in the TS evolves from the highest occupied molecular orbital (HOMO) of p-X-DMA substrates (see Figure S1 in the SI), and the energy level of this HOMO is related, in turn, to the one-electronic redox potential, E1/2, of the substrate. Figure 6a shows indeed a linear correlation of E1/2 vs E(HOMO). Thus, the observed correlation between the reaction rates and the substrate redox potentials reveals that the reactivity is correlated with the energy level of the substrate’s HOMO. From Figure 6b,c we can see that indeed there is a very good correlation between the HAT barriers and the HOMO energies (HS, R ) 0.999; LS, R ) 0.997). Clearly therefore, the experimentally observed correlation is not associated with SET and reflects simply the polar character of the TS in an otherwise H-abstraction process.
7706 J. Phys. Chem. B, Vol. 111, No. 26, 2007
Wang et al.
TABLE 4: Comparison of the Calculated Kinetic Isotope Effects (KIEs) with Experimental Values for the N-Demethylation of p-(H, Cl, CN, NO2)-DMA with Cpd Ia DMA p-Cl-DMA p-CN-DMA p-NO2-DMA (KIEexp ) 2.6)a (KIEexp ) 2.8)a (KIEexp ) 3.6)a (KIEexp ) 4.0)a KIELSb KIEHSb
3.6 (3.7) 5.5 (7.3)
3.8 (4.1) 5.4 (7.3)
4.2 (4.7) 5.5 (7.6)
4.4 (5.1) 5.6 (7.8)
a Experimental KIE values from refs 4b and 4d. b Theoretical KIEs. Figures outside of the parenthesis are semiclassical Eyring KIEs, and those inside the parenthesis are Wigner-corrected ones.
3.2. Substituent Effect on Kinetic Isotope Effect. Table 4 shows the calculated KIEs (KIELS and KIEHS) for both the HS and LS reaction routes, alongside the experimental KIEs determined by Dinnocenzo, Jones et al., using intramolecular KIE determination that is free of masking.4 Figure 7 shows various plots of the computed and experimental KIE values. The computed values, in Table 4, are consistently higher than experimental ones. However, the trends in the computational KIELS and experimental KIE data are similar, namely, there is a gradual increase in the KIE value as the substitutent becomes more electron-withdrawing. This relationship is shown pictorially in Figure 7a,b, where the KIE values are plotted against the Hammett σ parameters of the substituents. By contrast to the well-behaved KIELS, the KIEHS values are almost constant, irrespective of the substituent variation. Furthermore, as shown in Figure 7c the KIEHS values have no correlation with the Hammett σ parameters, and they also do not match the experimental trends. Although KIELS correlates nicely with the experimental values, the theory-experimental match can be further improved by considering the SSM and TSR scenarios based on the computed barrier differences, discussed above (see, e.g., E2 values in Table 2 and Figure 1). Thus, in accord with the SSM nature of the reactions of DMA and p-Cl-DMA, only the KIELS values match the experimental data both in the variation trends and in the incremental increase of 0.2 in the values, as the substituent changes from H to Cl. The corresponding KIEHS values are much too high and virtually constant. On the other hand, in accord with the TSR nature of the reactions of p-CNDMA and p-NO2-DMA, the experimental KIE increments can be matched well only if a combination of KIELS and KIEHS is assumed in line with the TSR scenario revealed by the computations. Indeed, the incremental differences of the KIELS values for p-CN-DMA and p-NO2-DMA relative to DMA are too small compared with the experimental values. Thus, taking the DMA oxidations as the reference reaction, and assuming that the percentage of KIELS in KIEobs is x, we can then derive the following relationships by comparing the experimental KIEs and the calculated KIELS and KIEHS for the entire series: i DMA KIEiexp - KIEDMA exp ) (KIEdoublet - KIEdoublet)x +
(KIEiquartet - KIEDMA doublet)(1 - x) where i ) p-Cl-DMA, p-CN-DMA, p-NO2-DMA (5) Substituting the relative values of Table 4 into this formula, we can obtain mechanistically averaged KIE values. The equation shows that p-Cl-DMA oxidation is also SSM, in accord with the computed HS and LS barriers in Figure 1, whereas for p-CN-DMA and p-NO2-DMA, the reactions involve TSR; for p-CN-DMA, 70% LS + 30% HS; for p-NO2-DMA, 50% LS + 50% HS. These percentages match reasonably well the ratio of reaction rates calculated with the activation energy differences between the HS and LS processes. Figure 7d shows a plot of
Figure 7. Plots of 2,4KIEs and KIE(exp) against Hammett substituent parameters σpara of para substituents: (a) KIE(experimental) vs the Hammett parameter, (b) KIELS vs the Hammett parameter, (c) KIEHS vs the Hammett parameter, and (d) mixed HS-LS KIE vs KIEexperimental.
these mechanistically averaged KIEs against corresponding experimental values, and the match is very good. Thus, all in all, the kinetic isotope effect profile study demonstrates that there is a mechanistic switch from SSM to TSR, the latter occurring for the p-CN-DMA and p-NO2-DMA oxidations. To sum up, our calculated results on the geometric structure of the transition states demonstrate that the low KIE values just reflect the asymmetry of HAT transition states (Figure 2), and do not emerge from a SET mechanism. Furthermore, the trends in the KIEP reveal a mechanistic switch from spin-selective
N-Demethylation of Substituted DMAs
J. Phys. Chem. B, Vol. 111, No. 26, 2007 7707
Figure 8. Energy profiles for the carbinolaniline decomposition in the enzymatic (bottom) and nonenzymatic (top) environments. Relative energies for the enzymatic process are given at the E1 (UB3LYP/B2//B1) level out of parentheses and the E2 (UB3LYP/B2//B1; ZPE, NH-S, ) 5.6) level in parentheses. For the nonenzymatic processes, the relative energies are at the E3 (UB3LYP/B2//B1; ZPE, ) 78.39) level. Energies are in kcal/mol, bond length in Å, and angles in degree.
reactivity to two-state reactivity, and reinforce the validity of the results obtained from energy profile studies (Figure 1). 3.3. Carbinolaniline Decomposition Process in the Enzymatic and Nonenzymatic Environments. Before summarizing the mechanistic picture, we must ascertain the nature of the carbinolaniline decomposition after C-H hydroxylation. The conventional concept about this process is that carbinolaniline decomposition proceeds in the nonenzymatic environment; therefore, our first study aim is that which factor causes this process proceeding in the nonenzymatic environment. Figure 8 shows the energy profiles for the carbinolaniline decomposition for the pristine DMA in the enzymatic and nonenzymatic environments. The process starts from the ferric complex that is ca. 50 kcal/mol lower than the reactants channel in Figure 1a. In the enzymatic environment, the activation energy for carbinolaniline decomposition is about 31.5 (28.5) kcal/mol for the quartet (doublet) route. No significant spin density change occurs, so this process involves proton transfer with simultaneous bond cleavage. In the gas phase, the excess energy of the carbinolamine would have been sufficient to lead to decomposition. However, in the protein environment, this excess energy is distributed very fast, and therefore, the process will encounter a very large barrier, > 30 kcal/mol. Since the binding energy of carbinolaniline to the heme is quite small (quartet, 2.6 kcal/mol; doublet, 8.5 kcal/mol), the carbinolaniline can readily dissociate from the heme and be released outside of the enzyme’s pocket. In such a case, the carbinolaniline will decompose in a nonenzymatic route, which is shown at the top left side of Figure 8. Since the process involves a proton shift from the O-H moiety to the nitrogen, we considered the assistance by a single water molecule. Thus as can be seen,
when the effects of water assistance, bulk solvation (water), and ZPE correction are taken into account, the activation energy for carbinolaniline decomposition in the nonenzymatic environment drops to 14.8 kcal/mol. Figure 9 shows the energetic effect brought about by one water molecule on the carbinolaniline decomposition in the enzymatic and nonenzymatic environments, whereas Figure 10 depicts the corresponding TSs. As seen from Figure 9, in both enzymatic and nonenzymatic processes, the water molecule lowers the barrier. However, the barriers in the enzymatic environment are consistently larger than in the nonenzymatic processes by 0.6-11.3 kcal/mol. Comparison of these data with the low binding energy (2.6-8.5 kcal/mol) of carbinolaniline to the heme provides further support for the hypothesis that the carbinolaniline decomposition proceeds in the nonenzymatic environment. The energetic effect induced by one water molecule is based on the fact that the carbinolaniline decomposition is a pericyclic reaction, as shown in Figure 10. Thus, without the water molecule, the transition state is a strained rhomboidal structure (∠N...C...O ) 93.9° ∼ 94.6°). When one water molecule is added, the transition state becomes hexagonal and relatively strain free (∠N...C...O ) 108.4° ∼ 109.5°). This reduces the energy barrier significantly. We also calculate KIE values for the proton shift in the p-(H, Cl, CN, NO2)-DMA decomposition processes in the nonenzymatic environment. The KIEs for cabinolaniline decompositions are as follow: for carbinolaniline, 3.0 (3.4); for p-Cl-carbinolaniline, 2.2 (2.5); for p-CN-carbinolaniline, 2.2 (2.5); for p-NO2carbinolaniline, 2.1 (2.4) (figures outside of the parenthesis are values of the semiclassical Eyring KIEs and those in the
7708 J. Phys. Chem. B, Vol. 111, No. 26, 2007
Wang et al.
Figure 9. Effect of one water molecule on the carbinolaniline decomposition barriers in the nonenzymatic (a and b) and enzymatic environments (c and d). Relative energies (in kcal/mol) in parts a and b are given at the E1 level out of parentheseses and the E3 (UB3LYP/B2//B1; ZPE, ) 78.39) level in parentheses, whereas in parts c and d at the E1 (UB3LYP/B2//B1) level out of parentheses at the E2 (UB3LYP/B2//B1; ZPE, NH-S, ) 5.6) level in parentheses.
Figure 10. Geometric properties (length in Å, and angle in degree) of the transition states of the carbinolaniline decomposition process in the enzymatic/nonenzymatic environment with/without one water molecular assistance.
parenthesis are values of the Wigner-corrected ones). These values are not consistent with the observed KIEs for the N-demethylation reactions. We may therefore conclude that the carbinolaniline decomposition step does not affect the KIE determination of the C-H hydroxylation process in the DMA N-demethylation by cytochrome P450. 3.4. Mechanistic Overview of the N-Demethylation Process and the Significance of the Kinetic Isotope Effect Profiles. As we mentioned in the introduction, one of the arguments in favor of a SET mechanism (Scheme 1) was the low KIE value (1.78) obtained for the demethylation reactions of N-methylN-(trideuteriomethyl)anilines.5a In their response, Dinnocenzo and co-workers4 showed that the KIEs determined with Nmethyl-N-(trideuteriomethyl)anilines were masked. They further pointed out that the proposed HAT and SET mechanisms could
not be distinguished by the magnitude of the isotope effect as such, and instead they proposed that kinetic isotope effect profiles (KIEP) correlations can be more diagnostic. In so doing these authors demonstrated that the KIEP obtained from P450 oxidation of the DMAs was virtually identical to the KIEP obtained for a model HAT reaction with t-BuO• radical. This correlation provided compelling evidence in support of the HAT mechanism. Furthermore, Donnocenzo and Jones showed also that different P450 isoforms, covering four different P450 families, gave the same KIEPs; a fact which means that all of the P450 isoforms metabolize the anilines via a common reaction mechanism. The large negative Hammett F value obtained from the free energy relationship of a series of para-substituted N,N-dimethylanilines by CYP450 and model analogs,6a,c,7 served as additional
N-Demethylation of Substituted DMAs evidence for a SET mechanism. However, as argued by Donnocenzo et al.,4c the F value of -0.6 was remarkably similar to the value of -0.4, which was obtained for the HAT reactions from DMAs by the OBut radical,23 thus reflecting a polar effect in an otherwise HAT process. Following the calculations by Li et al.,13 the present study provides systemic calculations of the mechanisms and the KIEs of the entire series of reactions between Cpd I and para substituted N,N-dimethylanilines (DMAs). All of the mechanisms are found to be exclusively HAT types. Moreover, the computed energy profiles reveal that the HAT process proceeds in a spin-selective manner (SSM) for the pair DMA and p-ClDMA but switches to two-state reactivity (TSR) for the pair p-CN-DMA and p-NO2-DMA. There is a very good match between the calculated KIE values and the experimental ones. The calculated geometries of the transition states reveal significantly asymmetric structures, which are expected to yield low isotope effects. Our calculations show that as the substituent becomes more electron-withdrawing, the transition state moves to a less early position along the H-transfer coordinate, and the KIE as well as the barrier increase. The observed6a,ac large negative Hammett F parameter in free energy relationships, and the correlation of the reaction rates with the substrate redox potentials, are reproduced by theory and shown to reveal the evolution of the delocalized πph-πC-N orbitals of the transition states, from the corresponding HOMOs of the DMA substrate. Thus, the more electron-withdrawing substituents suck the lone pair orbital on N into the aromatic moiety and inhibit conjugation with the developing radical in the TS. As such, the HAT process has some polar character that is substituent-dependent and is revealed by all of the free energy correlations. However, these correlations are not causal, and we found no evidence for a SET mechanism. Finally, the calculations of the carbinolaniline decomposition process demonstrated that the generated carbinolaniline can readily dissociate from the heme and decomposes in a nonenzymatic environment, which involves water assisted proton shift. 4. Conclusions This study which addressed the mechanism of N-demethylation of four N,N-dimethylanilines (DMAs) by Cpd I of P450, vis-a`-vis the experimental data,2,4 led to the following conclusions: (a) Both mechanistic and kinetic isotope effect profiles (KIEPs) reveal that all of the reactions proceed in a HAT mechanism, with a mechanistic switch from spin-selective reactivity for the DMA and p-Cl-DMA substrates to two-state reactivity for the p-CN-DMA and p-NO2-DMA substrates. (b) The computed KIEPs for these scenarios match the experimentally determined KIEPs. Theory shows that the low-spin KIEs are sensitive to the nature of the substrates and vary in a manner predicted by the Melander-Westheimer postulate: the TSs are “early” along the H-transfer coordinate for DMA and become progressively less early as the electron-withdrawing ability of the para substituent increases. By contrast, the KIE values for the high-spin route remain almost constant (∼5.5) and the corresponding TSs are all “central” or “late” along the H-transfer coordinate. (c) The electronic features of the para substituents exert influence on the nature of the CR-H bond via a conjugated πph-πC-N orbital that stretches to the aromatic moiety and is sensitive to the substituent; this is the route cause for the observation of a negative Hammett F parameter. (d) The observed correlation between reaction rates and the substrate redox potentials just reflects the polar character of the HAT process and the sensitivity of HAT energy barriers to the energy
J. Phys. Chem. B, Vol. 111, No. 26, 2007 7709 level of the highest occupied molecular orbital of the substituted DMAs. (e) The generated carbinolaniline can readily dissociate from the heme and decomposes in a nonenzymatic environment, which involves water assisted proton shift. Acknowledgment. The research in the Dalian Insititute is supported by NSFC (Grants 20373071, 20333050) to K.H. Y.W. acknowledges the advice of Dr. Sam de Visser on polarizability calculations. The research in the HU is supported by an ISF grant to S.S. Supporting Information Available: KIE expressions based on the theory of absolute reaction rates and the corresponding Fortran source code, 4 figures and 27 tables for detailed electronic and geometric features of various reaction species, reference 17 in full, and x, y, z, coordinates. This material is available free of charge Via the Internet at http://pubs.acs.org. References and Notes (1) (a) Porter, T. D.; Coon, M. J. J. Biol. Chem. 1991, 266, 13469. (b) Coon, M. J. Annu. ReV. Pharmacol. Toxicol. 2005, 45, 1. (2) (a) de Montellano, P. R. O. in Cytochrome P450: Structure, Mechanism and Biochemistry, 2nd ed.; Ortiz de Montellano, P. R., Ed.; Plenum Press: New York, 1995; Chapter 8. (b) Ortiz de Montellano, P. R.; De Voss, J. J. Nat. Prod. Rep. 2002, 19, 477. (c) Meunier, B.; de Visser, S. P.; Shaik, S. Chem. ReV. 2004, 104, 3947. (d) Shaik, S.; Kumar, D.; de Visser, S. P.; Altun, A.; Thiel, W. Chem. ReV. 2005, 105, 2279. (e) Sono, M.; Roach, M. P.; Coulter, E. D.; Dawson, J. H. Chem. ReV. 1996, 96, 2841. (f) Guengerich, F. P. Chem. Res. Toxic. 2001, 14, 611. (3) (a) Guengerich, F. P.; Macdonald, T. L. FASEB J. 1990, 4, 2453. (b) Guengerich, F. P.; Macdonald, T. L. AdV. Electron Transfer Chem. 1993, 3, 191. (4) (a) Dinnocenzo, J. P.: Karki, S. B.: Jones, J. P. J. Am. Chem. Soc. 1993, 115, 7111. (b) Karki, S. B.; Dinnocenczo, J. P.; Jones, J. P.; Korzekwa, K. R. J. Am. Chem. Soc. 1995, 117, 3657. (c) Karki, S. B.; Dinnocenzo, J. P. Xenobiotica 1995, 25, 711. (d) Manchester, J. I.; Dinnocenzo, J. P.; Higgins, L. A.; Jones, J. P. J. Am. Chem. Soc. 1997, 119, 5069. (5) (a) Miwa, G. T.; Walsh, J. S.; Kedderis, G. L.; Hollenberg, P. F. J. Biol. Chem. 1983, 258, 14445. (b) Miwa, G. T.; Garland, W. A.; Hodshon, B. J.; Lu, A. Y. H.; Northrop, D. B. J. Biol. Chem. 1980, 255, 6049. (c) Abdel-Monem, M. M. J. Med. Chem. 1975, 18, 427. (d) Hollenberg, P. F.; Miwa, G. T.; Walsh, J. S.; Dwyer, L. A.; Kedderis, G. L. Drug Metab. Dispos. 1985, 13, 272. (e) Guengerich, F. P.; Yun, C.-H.; Macdonald, T. L. Biochemistry 1996, 271, 27321. (6) (a) Burka, L. T.; Guengerich, F. P.; Willard, R. J.; Macdonald, T. L. J. Am. Chem. Soc. 1985, 107, 2549. (b) Galliani, G.; Rindone, B.; Dagnino, G.; Salmona, M. Eur. J. Drug Metab. Pharmacokin. 1984, 9, 289. (c) Galliani, G.; Nali, M.; Rindone, B.; Tollari, S.; Rocchetti, M.; Salmona, M. Xenobiotica 1986, 16, 511. (d) Macdonald, T. L.; Gutheim, W. G.; Martin, R. B.; Guengerich, F. P. Biochemistry 1989, 28, 2071. (e) Shono, T.; Toda, T.; Oshino, N. J. Am. Chem. Soc. 1982, 104, 2639. (f) Parker, V. D.; Tilset, M. J. Am. Chem. Soc. 1991, 113, 8778. (g) Goto, Y.; Watanabe, Y.; Fukuzumi, S.; Jones, J. P.; Dinnocenzo, J. P. J. Am. Chem. Soc. 1998, 120, 10762. (7) (a) Baciocchi, E.; Lanzalunga, O.; Lapi, A.; Manduchi, L. J. Am. Chem. Soc. 1998, 120, 5783. (b) Baciocchi, E.; Gerini, M. F.; Lanzalunga, O.; Lapi, A.; Mancinelli, S.; Mencarelli, P. Chem. Commun. 2000, 393. (c) Baciocchi, E.; Gerini, M. F.; Lanzalunga, O.; Lapi, A.; Piparo, M. G. L.; Mancinelli, S. Eur. J. Org. Chem. 2001, 2305. (d) Baciocchi, E.; Bietti, M.; Gerini, M. F.; Lanzalunga, O. J. Org. Chem. 2005, 70, 5144. (8) (a) Augusto, O.; Beilan, H. S.; Ortiz de Montellano, P. R. J. Biol. Chem. 1982, 257, 11288. (b) Lee, J. S.; Jacobsen, N. E.; Ortiz de Montellano, P. R. Biochemistry 1988, 27, 7703. (c) Guengerich, F. P.; Brian, W. R.; Iwasaki, M.; Sari, M.-A.; Ba¨a¨rnhielm, C.; Berntsson, P. J. Med. Chem. 1991, 34, 1838. (d) Okazaki, O.; Guengerich, F. P. J. Biol. Chem. 1993, 268, 1546. (e) Cuppoletti, A.; Dagostin, C.; Florea, C.; Galli, C.; Gentili, P.; Lanzalunga, O.; Petride, A.; Petride, H. Chem. Eur. J. 1999, 5, 2993. (f) Bietti, M.; Cuppoletti, A.; Dagostin, C.; Florea, C.; Galli, C.; Gentili, P.; Petride, H.; Caia, C. R. Eur. J. Org. Chem. 1998, 2425. (9) (a) Hanzlik, R. P.; Kishore, V.; Tullman, R. J. Med. Chem. 1979, 22, 759. (b) Hanzlik, R. P.; Tullman, R. J. Am. Chem. Soc. 1982, 104, 2048. (c) Macdonald, T. L.; Zirvi, K.; Burka, L. T.; Peyman, P.; Guengerich, F. P. J. Am. Chem. Soc. 1982, 104, 2050. (d) Bondon, A.; Macdonald, T. L.; Harris, T. M.; Guengerich, F. P. J. Biol. Chem. 1989, 64, 1988. (e) Chen, H.; deGroot, M. J.; Vermeulen, N. P. E.; Hanzlik, R. P. J. Org. Chem. 1997, 62, 8227.
7710 J. Phys. Chem. B, Vol. 111, No. 26, 2007 (10) Jurva, U.; Bissel, P.; Isin, E. M.; Igarashi, K.; Kuttab, S.; Castagnoli, N. J. Am. Chem. Soc. 2005, 127, 12368. (11) (a) Bhakta, M. N.; Wimalasena, K. J. Am. Chem. Soc. 2002, 124, 1844. (b) Bhakta, M. N.; Hollenberg, P. F.; Wimalasena, K. Chem. Commun. 2005, 265. (c) Bhakta, M. N.; Wimalasena, K. Eur. J. Org. Chem. 2005, 4801. (d) Bhakta, M. N.; Hollenberg, P. F.; Wimalasena, K. J. Am. Chem. Soc. 2005, 127, 1376. (12) (a) Cerny, M. A.; Hanzlik, R. P. J. Am. Chem. Soc. 2006, 128, 3346. (b) Cerny, M. A.; Hanzlik, R. P. Arch. Biochem. Biophys. 2005, 436, 265. (c) Shaffer, C. L.; Harriman, S.; Koen, Y. M.; Hanzlik, R. P. J. Am. Chem. Soc. 2002, 124, 8268. (d) Shaffer, C. L.; Morton, M. D.; Hanzlik, R. P. J. Am. Chem. Soc. 2001, 123, 8502. (e) Shaffer, C. L.; Morton, M. D.; Hanzlik, R. P. J. Am. Chem. Soc. 2001, 123, 349. (13) Li, C.; Wu, W.; Kumar, D.; Shaik, S. J. Am. Chem. Soc. 2006, 128, 394. (14) (a) Shaik, S.; Filatov, M.; Schroder, D.; Schwarz, H. Chem. Eur. J. 1998, 4, 193. (b) Ogliaro, F.; Harris, N.; Cohen, S.; Filatov, M.; de Visser, S. P.; Shaik, S. J. Am. Chem. Soc. 2000, 122, 8977. (c) de Visser, S. P.; Ogliaro, F.; Sharma, P. K.; Shaik, S. J. Am. Chem. Soc. 2002, 124, 11809. (15) (a) Dowers, T. S.; Rock, D. A.; Jones, J. P. J. Am. Chem. Soc. 2004, 126, 8868. (b) Volz, T. J.; Rock, D. A.; Jones, J. P. J. Am. Chem. Soc. 2002, 124, 9724.
Wang et al. (16) (a) Filatov, M.; Harris, N.; Shaik, S. Angew. Chem. Int. Ed. 1999, 38, 3510. (b) Harris, N.; Cohen, S.; Filatov, M.; Ogliaro, F.; Shaik, S. Angew. Chem. Int. Ed. 2000, 39, 2003. (c) Ogliaro, F.; Cohen, S.; de Visser, S. P.; Shaik, S. J. Am. Chem. Soc. 2000, 122, 12892. (d) de Visser, S. P.; Ogliaro, F.; Harris, N.; Shaik, S. J. Am. Chem. Soc. 2001, 123, 3037. (e) Kumar, D.; de Visser, S. P.; Shaik, S. J. Am. Chem. Soc. 2004, 126, 5072. (f) Schoneboom, J. C.; Cohen, S.; Lin, H.; Shaik, S.; Thiel, W. J. Am. Chem. Soc. 2004, 126, 4017. (g) Wang, Y.; Wang, H.; Wang, Y.; Yang, C.; Yang, L.; Han, K. J. Phys. Chem. B 2006, 110, 6154. (17) Frisch, M. J.; et al. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (18) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (b) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (19) (a) Hay, J. P.; Wadt, W. R. J. Chem. Phys. 1985, 82, 99. (b) Friesner, R. A.; Murphy, R. B.; Beachy, M. D.; Ringlanda, M. N.; Pollard, W. T.; Dunietz, B. D.; Cao, Y. X. J. Phys. Chem. A 1999, 103, 1913. (20) Melander, L.; Saunders, W. H., Jr. In Reaction rates of isotopic molecules.; Robert E. Krieger Publishing Company: Malabar, FL, 1987; Chapter 2. (21) (a) Groves, J. T.; McClusky, G. A. J. Am. Chem. Soc. 1976, 98, 859. (b) Groves, J. T. J. Chem. Educ. 1985, 62, 928. (22) Jaffe, H. H. Chem. ReV. 1953, 53, 191. (23) Uneyama, K.; Namba, H.; Oae, S. Bull. Chem. Soc. Jpn. 1968, 41, 1928.
