20686
J. Phys. Chem. B 2006, 110, 20686-20692
Comparative Computational Analysis of Different Active Site Conformations and Substrates in a Chalcone Isomerase Catalyzed Reaction J. Javier Ruiz-Pernı´a, Estanislao Silla, and In˜ aki Tun˜ o´ n* Departamento de Quı´mica Fı´sica, UniVersitat de Vale` ncia, 46100 Burjassot, Spain ReceiVed: June 12, 2006; In Final Form: August 7, 2006
Chalcone isomerase catalyzes the transformation of chalcones to flavanones. We present a computational study of the rate-limiting chemical step, an intramolecular Michael addition of a 2′-oxyanion to the R,βdouble bound. By using quantum mechanical/molecular mechanical hybrid methods we traced the free-energy profiles associated with the reaction of two different substrates (chalcone and 6′-deoxychalcone) in two different conformations of the active site that are described in the different crystallographic structures available. We have obtained significant differences (about 4 kcal/mol) in the free-energy barriers calculated for the two active sites. According to our results, the active site conformation with larger catalytic power presents a positively charged lysine residue much closer to the substrate than the other. Complementary electronic and electrostatic analysis shows that the charge is transferred from the 2′-oxyanion to the β-carbon atom. Interactions of the environment with these two atoms are essential to understand the differences between both active sites and also the origin of catalysis in this enzyme.
1. Introduction Understanding the relationship between structure and catalytic power is fundamental to obtain a deeper insight into enzymatic activity. It has been recognized recently from experimental1-3 and theoretical4-7 studies that subtle structural variations may lead to very different rate constants: Conformational changes in the enzyme’s active site may lead to quite different freeenergy barriers for the catalyzed reaction. This can be a source of discrepancy among different theoretical approaches to a particular enzymatic reaction. In a very illustrative paper, Garcı´a-Viloca et al. have carefully discussed two examples where quite different results are obtained depending on the selected starting structure for the enzyme.7 The cases studied (xylose isomerase and acyl-CoAdehydrogenase) illustrate the sensitivity of catalytic efficiency to small changes in the protein conformation near the active site. This problem can have dramatic consequences on computational studies. In principle, estimation of activation and reaction free energies should be independent of the starting point, but in practice, molecular simulations explore a limited region of the potential energy hypersurface. The myriad of degrees of freedom of a protein make a complete exploration unaffordable, and thus the starting point can be decisive to determine the particular region explored in a given study. As stated in ref 7, “if the crystal structure used as a starting point for the simulation does not correspond to a typical productive structure of the enzyme, one may draw incorrect conclusions”. The relationship between enzyme structure and activity can also be a source of divergence in computational studies afforded by different groups. In other words, to what extent can different simulation protocols starting from the same structure lead to significantly different results? This problem was detected in the comparison among different studies of catechol O-methyl transferase (COMT).8-17 Three different computational strategies * Author to whom correspondence should be addressed. E-mail: tunon@ uv.es.
used as a starting point the same crystallographic structure corresponding to an inhibitor bound in the active site. With different simulation protocols, three different interaction patterns between the substrate and the enzyme were obtained, differing essentially in the coordination of catecholate oxygen atoms with a magnesium ion present in the active site.8-17 In this case, our strategy, based on the localization of the transition structure and the reaction path up to the reactant valley, was shown to lead to the reactive conformation, where the nucleophilic oxygen was already prepared to receive the methyl group from the donor agent.15-17 Chalcone isomerase (CHI) offers an excellent opportunity to analyze the dependence of free-energy profiles with the active site structure. CHI is a functional monomer of about 220 residues that has been isolated from a variety of higher plants.18,19 This enzyme plays a central role in flavonoid biosynthetic pathways, catalyzing the cyclization of chalcone (4,2′,4′,6′-tetrahydroxychalcone) and 6′-deoxychalcone (4,2′,4′trihydroxychalcone) into (2S)-naringerin (5,7,4′-trihydroxyflavanone) and (2S)-5-deoxyflavanone (7,4′-dihydroxyflavanone),20,21 as shown in Scheme 1. The pH dependence of the nonenzymatic and CHI-catalyzed processes implies that both chalcone and 6′-deoxychalcone are largely deprotonated at physiological conditions.21 Generation of the 2′-oxyanion is required to produce an intramolecular Michael addition to the R,β-double bound.21 This is followed by protonation of the resulting β-carbanion. Both chalcone and 6′-deoxychalcone spontaneously suffer cyclization in solution giving enantiomeric mixtures. CHI ensures the rapid formation of the biologically active (2S)-flavanones operating near the diffusion-controlled limit.21 There are up to six sets of X-ray coordinates of CHI in the Protein Data Bank (PDB codes 1EYQ,20 1EYP,20 1FM7,21 1FM8,21 1JEP,22 and 1JXO22), corresponding to the apoenzyme or to enzyme-flavanone complexes. CHI crystallizes as a dimeric structure, and then each set of coordinates contains two active sites. Analysis of these structures suggests that the CHI
10.1021/jp0636470 CCC: $33.50 © 2006 American Chemical Society Published on Web 09/19/2006
Analysis of CHI Conformations and Substrates
J. Phys. Chem. B, Vol. 110, No. 41, 2006 20687
SCHEME 1
active site is flexible and some residues adopt different conformations.20 In particular two different rotamers of Thr48 are found.23 In one of them the side-chain hydroxyl group is placed in such a way that it could establish a hydrogen-bond interaction with the carbonyl oxygen atom of the substrate while in the other the hydroxyl group interacts with one crystallographic water molecule. The two active sites described in 1EYP, 1FM7, and 1FM8 and one of the active sites of 1EYO and 1JEP correspond to the former, while the two active sites of 1JXO and one of the active sites of 1EYO and 1JEP correspond to the latter. Molecular dynamics (MD) simulations of the two active site conformations with a transition-like structure in the active site have shown that only in the first case the approach of a positively charged lysine (Lys97) to the substrate is allowed.23 It has been proposed that this Lys97 could act as a general acid catalyst stabilizing the developing negative charge on the carbonyl oxygen atom of the substrate.23,24 However, the difference between the catalytic efficiencies of the two conformations of the active site has not yet been estimated. In this paper we carry out a hybrid quantum mechanical/ molecular mechanical (QM/MM) study of the reaction catalyzed by CHI, for both chalcone and 6′-deoxychalcone. Free-energy profiles for both substrates using two different conformations of the active site were obtained as potentials of mean force (PMFs). A comparative analysis of the results obtained may be useful to highlight not only into the origin of catalysis in the case of CHI but also to understand the relationship between structure and catalysis and the dependence of computational results on the selected starting conformation. 2. Methodology The initial coordinates of the system were taken from the X-ray crystal structure 1EYQ of CHI with (2S)-naringenin, which is the product of the reaction for chalcone.20 In particular, we selected the monomer in which the side-chain hydroxyl group of Thr48 is found at a shorter distance from the carbonyl group of the reaction product. The product was manually changed to chalcone or 6′-deoxychalcone with an oxyanion in the 2′ position. These reactant molecules constitute the QM subsystems, 31 atoms for chalcone and 30 for 6′-deoxychalcone, that were described using the AM1 Hamiltonian.25 Hydrogen atoms were added to all of the system using DYNAMO facilities.26 The protonation states were determined assuming
Figure 1. Chalcone transition structure in CHI active site. The substrate is shown using an atom color code while the enzyme is uniformly golden brown.
pH ) 7 and standard pKa values in solution. Afterward, the system was placed inside a cubic box (79.5 Å on each side) of water molecules centered on the QM subsystem. The MM subsystem was then formed by 3231 enzyme atoms, 591 crystallization water atoms, and 45972 solvation water atoms, described using the OPLS-AA27 and TIP3P potentials.28 During the QM/MM minimizations and simulations we employed periodic boundary conditions and a cutoff radius of 13.5 Å for all kinds of interactions. All the atoms of the system were allowed to move. To locate and characterize the transition structure corresponding to the Michael addition we employed a combination of the GRACE4,29 and DYNAMO programs that enables the determination of transition structures for molecular systems with a large number of atoms by means of QM/MM methods. GRACE divides the total coordinate space into two subsets of atoms: a control space in which the Hessian matrix is calculated (in this case the QM subsystem) and a complementary space. At each Newton-Raphson step of a QM/MM search in the control space all geometrical coordinates belonging to the complementary space are minimized.30 Once a saddle point on the potential energy surface is located and characterized, GRACE is capable of tracing the intrinsic reaction coordinate (IRC) paths down to the reactant and product valleys. For both reactant molecules a transition structure was located with a O2′-CR distance of approximately 1.9 Å. Figure 1 shows the disposition of the chalcone transition structure in the active site of the monomer with one of the rings pointing outward from the enzyme. To perform molecular dynamics simulations in the transition states (TS’s) of chalcone and 6′-deoxychalcone we added a parabolic potential to the O2′-CR distance, keeping this parameter centered around 1.9 Å. After heating and equilibrating, we ran a 300 ps simulation for both transition states using the NVT ensemble at a reference temperature of 300 K. The configurations obtained for the CHI active sites with chalcone and 6′-deoxychalcone transition states are schematically shown in Figure 2A. From the last snapshot obtained in these simulations we prepared the other conformation observed for the active site. According to the descriptions given by Bruice and co-workers23 this new conformation was obtained by rotation of the Thr48 side chain around the Ph-CH2OH bond.
20688 J. Phys. Chem. B, Vol. 110, No. 41, 2006
Ruiz-Pernı´a et al.
Figure 2. Schematical views of the two transition state conformations (A and B) found for CHI active site.
Figure 3. Superposition of the TS structures of 6′-deoxychalcone in the A (red) and B (blue) conformations of the active site. Only the substrate and Lys97 are shown.
In addition, to favor the approach between Lys97 and the substrate we added a soft harmonic potential to reduce the distance between this residue and the carbonyl oxygen atom of the transition structures in a series of biased simulations. Finally, after equilibration, we also ran 300 ps of MD simulations for both chalcone and 6′-deoxychlacone with the new active site conformation. The configurations obtained for the new active sites are schematically shown in Figure 2B. Superimposition of substrate and Lys97 corresponding to both conformations of the active site in Figure 3 shows that the approach is accomplished by means of rotation of the lysine side chain and displacement of the substrate. To evaluate the catalytic ability of both active site conformations (hereafter called A and B) we obtained the PMFs corresponding to the Michael addition of the 2′-oxyanion to the R,β-double bound for chalcone and 6′-deoxychalcone. Four different PMFs, corresponding to chalcone in active sites A and B (chalcone-A and chalcone-B) and to 6′-deoxychalcone in active sites A and B (6′-deoxychalcone-A and 6′-deoxychalcone-
B), were obtained. The procedure for the PMF calculation requires a series of MD simulations in which the reaction coordinate was harmonically restrained around different values, covering the transformation from reactants to products. For this reaction, the O2′-CR distance was the natural choice. This variable represented very closely the minimum energy paths that were found in the enzyme and in the gas phase. The different values of the variable sampled during the simulations were then pieced together by means of the weighted histogram analysis method (WHAM)31 to construct the full distribution function from which the PMF was obtained. The value of the force constant used for the harmonic umbrella sampling (2500 kJ mol-1 Å-2 on the reaction coordinate) was determined to allow a full overlapping of the different windows traced in the PMF evaluation but without losing control over the selected coordinate. The windows were run in a consecutive way starting from the transition states toward reactants and products. The former were identified as shallow minima, especially in the case of chalcone. Each window was started from the final configuration of the precedent window and consisted in 2 ps of equilibration followed by 10 ps of production, using a time step of 1 fs. This was long enough to sample a wide range of structures at a reference temperature of 300 K. The total number of windows employed to cover the whole range of the reaction coordinate from reactants to products was 61, and the total simulation length was then of 732 ps for each PMF, which should be enough for the range of reaction coordinate values covered. The canonical ensemble (NVT) was used for all the simulations, thus yielding estimates of the Helmholtz free-energy changes, which for condensed-phase reactions can be considered equivalent to Gibbs free-energy variations. Finally, being aware that the AM1 method is not always accurate enough, we considered a correction to the QM energy. The potential energy barrier was calculated at the AM1 and MP2/6-31+G(d,p)32 levels using AM1/MM optimized structures corresponding to the maximum and minimum of the free-energy profile. These calculations were carried out with Gaussian 03.33 The difference between both estimations was then added to the AM1/MM computed free-energy barrier to obtain a corrected estimation. Average properties of the relevant reaction states were obtained through 100 ps simulations carried out using the same details as explained above.
Analysis of CHI Conformations and Substrates
Figure 4. PMFs obtained as a function of the O2′-CR distance for chalcone and 6′-deoxychalcone in the active site conformations presented in Figure 2A (black line) and Figure 2B (red line). Note that large values of the reaction coordinate correspond to reactants, and small values to products ((2S)-naringerin and (2S)-5-deoxyflavanone).
TABLE 1: Calculated Activation Free Energies (in kcal/ mol) for Chalcone and 6′-Deoxychalcone Michael Addition in the Two Conformations of the CHI Active Site substrate-active site
∆Gq (AM1/MM)
chalcone-A chalcone-B 6′-deoxychalcone-A 6′-deoxychalcone-B
24.8 20.2 24.2 20.6
3. Results and Discussion The PMFs obtained for the intramolecular Michael addition in chalcone and 6′-deoxychalcone in both conformations of the active site (A and B) are presented in Figure 4 as a function of the O2′-CR distance. The four PMFs describe an endothermic process, where the products (left side of the figure) have a carbon-oxygen distance of approximately 1.5 Å, the TS’s are located between 1.8 and 1.9 Å, and the Michaelis complexes (MCs, right side of the figure) are located between 2.8 and 3.0 Å. Simulations in the B conformation of the active site display the lower activation free energies and also the smaller reaction free energies. The free-energy barriers, summarized in Table 1, are 24.8 and 24.2 kcal/mol in the A active site for chalcone and 6′-deoxychalcone, respectively. In the B conformation the free-energy barriers found are 20.2 and 20.6 kcal/mol, respectively. This means that the free-energy barriers are reduced by approximately 4 kcal/mol when calculated in the B conformation of the active site. This can be translated to a catalytic effect on the rate constant of about 103 at 298 K. Moreover, when both substrates only are compared, the results obtained in the B active site are in agreement with the experimental ordering of the rate constants. Chalcone reacts faster in CHI, with the rate constant being about 5 times larger,21 which can be translated, using
J. Phys. Chem. B, Vol. 110, No. 41, 2006 20689 transition state theory, to an activation free-energy difference of about 1 kcal/mol (15.4 and 14.4 kcal/mol for 6′-deoxychalcone and chalcone, respectively).21 In summary, the results presented mean that the B conformation is more catalytic than the A conformation, and thus we will assume the former as the active form of the enzyme. The origin of the differences in the catalytic power of both active site conformations will be discussed below. To compare the free-energy barriers obtained in the catalytic active site conformation with the experimental estimation, one should take into account that the semiempirical AM1 method overestimates the barrier for the Michael addition. If we add a single-point correction, obtained as the difference between the AM1/MM and MP2/6-31+G(d,p)/MM energy barriers, calculated for AM1/ MM transition and reactant structures localized and characterized from snapshots of the corresponding simulation windows, then the free-energy barriers are reduced to 15.9 and 14.3 kcal/mol for 6′-deoxychalcone and chalcone, respectively. These values compare very well with the experimental estimations obtained from transition state theory application to the experimental kcat, which are 15.4 and 14.4 kcal/mol.21 The excellent agreement must be considered with caution because the reaction is known to operate near the diffusion-controlled limit,21 and these freeenergy values should be taken as upper limits for the chemical step. Averaged geometries of both active site conformations in the Michaelis complex and transition states of chalcone and 6′-deoxychalcone are provided in Table 2. As commented previously, the main difference between both active site conformations, besides internal rotation of Thr48, is the relative position of Lys97, much closer to the substrate in the B conformations. As a consequence of this approach a water molecule is released from the active site. In the A conformation up to five water molecules are found hydrogen-bonded to the reactive part of the substrate directly or indirectly. (One of the phenyl rings of the substrate points toward the solvent, as can be seen in Figure 1.) Four of these water molecules come from the crystallographic structure, and one of them (labeled as Wat1 in Figure 2A) comes from the added solvation box. Two water molecules (Wat3 and Wat4) are hydrogen-bonded to the nucleophilic oxygen atom, and the other three (Wat1, Wat2, and Wat5) are found forming a hydrogen-bond network among Lys97, Thr48, Tyr106 and the substrate. In the B conformation one of these last three molecules is lost because the substrate and Lys97 become closer (Figure 2B). As in the A conformation, we also found two water molecules hydrogen-bonded to the O2′ oxygen atom in the MC (Wat3 and Wat4) corresponding to the B active site. The other two water molecules are found forming a hydrogen-bond network between the substrate (carbonyl oxygen and Cβ atoms) and the residues of the active site (Lys97 and Tyr106). In this conformation one of the hydrogen bonds of water molecules with O2′ is lost in the TS. This water molecule (labeled as Wat4) is released to the solvent when passing from the MC to the TS in the B conformation of the active site. The average oxygen-oxygen distance (O2′‚‚‚Wat4O) is increased from 2.82/2.78 Å in the MCs to 5.66/6.11 Å in the TSs of chalcone and 6′-deoxychalcone, respectively. Then, the O2′ oxygen atom presents three hydrogen bonds in the MC (with Wat3, Wat4, and Asn133) and only two in the TS (Wat3 and Asn133). Otherwise, in the A conformation the O2′ atoms keep the three hydrogen bonds in the TS, with both chalcone and 6′-deoxychalcone. The carbonyl oxygen of the substrate (Oγ) can establish different hydrogen bonds depending on the particular substrate
20690 J. Phys. Chem. B, Vol. 110, No. 41, 2006
Ruiz-Pernı´a et al.
TABLE 2: Average Distances (in Å) Obtained for Chalcone and 6′-Deoxychalcone in the Transition State (TS) and Michaelis Complex (MC) Obtained in the A and B Conformations of the CHI Active Site chalcone-A O2′‚‚‚CR
Lys97- NH‚‚‚Oγ Lys97- NH‚‚‚Wat1-O Lys97- NH‚‚‚Wat5-O Lys97- NH‚‚‚Wat2-O Lys97-NH‚‚‚Thr48-OH Wat2-O‚‚‚Wat5-O Wat1-O‚‚‚Cβ Wat3-O‚‚‚O2′ Wat4-O‚‚‚O2′ Asn113-NH‚‚‚O2′ Wat1-O‚‚‚Oγ Wat2-O‚‚‚Oγ Oγ‚‚‚O6′ Tyr106-OH‚‚‚O6′ Wat2-O‚‚‚O6′ Lys97-NH‚‚‚O6′ Thr190-OH‚‚‚O4′
6′-deoxychalcone-A
6′-deoxychalcone-B
MC
TS
MC
chalcone-B TS
MC
TS
MC
TS
2.94 8.67 4.60 2.78 4.98 4.29 2.69 4.51 2.66 2.66 3.45 2.63 5.25 2.63 4.30 3.79 3.11 3.96
1.91 7.19 4.48 2.72 5.00 4.71 2.87 4.24 2.78 2.78 3.30 2.80 3.99 2.60 3.12 4.83 2.62 3.77
3.13 5.30 2.67
1.84 4.44 2.69
1.90 2.90 2.85
2.68 3.04
2.61 2.66
2.78 2.72
3.91 2.84 2.82 3.50 2.89 7.14 2.64 4.03 3.40 4.85 3.43
2.80 2.86 5.66 4.32 2.65 5.85 2.81 3.25 2.92 3.70 3.25
1.88 6.60 3.22 2.84 4.56 2.70 2.87 3.22 3.08 3.22 3.38 3.08 2.68
2.96 3.81 2.77
2.70 2.83
2.93 7.05 3.32 2.84 4.67 2.70 2.94 3.33 2.91 2.91 3.14 2.94 2.88
4.05 2.91 2.78 3.40 2.85 2.69
3.40 2.84 6.11 3.27 3.63 2.72
2.97
2.81
3.16
3.14
(chalcone or 6′-deoxychalcone) and the active site conformation. When the substrate is chalcone, a strong intramolecular hydrogen bond is observed between this atom and the 6′-hydroxyl group. When the substrate is 6′-deoxychalcone this interaction is obviously lost and substituted by a hydrogen bond with a crystallographic water molecule (Wat2). In addition the solvation water molecule Wat1 is also at hydrogen-bond distances from the Oγ atom, although in the TS this water molecule tends to establish a hydrogen bond with the Cβ carbon atom. This water molecule is the obvious candidate for the subsequent proton transfer to the carbon atom (Scheme 1). As we will show below this atom receives a large fraction of the electron formally donated by the O2′ atom, and then it carries a noticeable negative charge in the TS. The 6′-hydroxyl group can also establish a hydrogen bond with the hydroxyl group of Tyr106, but this interaction is significant only in the TS. The average oxygen-oxygen distances are longer than 4.0 Å in the MCs of chalcone in both active sites. This feature could contribute (see below) to the fact that the Michaelis constant (KM) of the chalcone-CHI complex is larger than that of the 6’deoxychalcone-CHI one.21 The binding process involves desolvation of the substrate and then the loss of interactions of water molecules with this hydroxyl group. When the substrate is placed in the active site the 6′-hydroxyl group does not strongly interact with the enzyme in the MC. However, these distances are reduced by about 1 Å in the TSs of chalcone, and this interaction could contribute to a larger catalytic rate constant (kcat) for chalcone than for 6′deoxychalcone, in agreement with the experimental observations.21 Finally, the 4′-hydroxyl group of the substrates establishes a hydrogen-bond interaction with the hydroxyl group of Thr190 in both active sites. This interaction could be responsible for the reduced Michaelis constant of 6′-deoxychalcone as compared to that of 4′,6′-deoxychalcone.21 To relate these geometrical features of the active sites with the catalytic power we must take into account the change in the charge distribution of the substrate along the reaction progress. Table 3 shows the variation in the averaged Mulliken charges34 on selected substrate atoms when passing from the MC to the TS. The charges of carbon atoms include those of the neighboring nonpolar hydrogen atoms. From these results, in electronic terms, the Michael reaction can be described as an electron transfer from O2′ and CR atoms to the Cβ one. It is interesting to note that the charge on the Oγ is always quite
TABLE 3: Variation of the Mulliken Charges on Selected Atoms of the Substrate from Michaelis Complexes to Transition States (Q(TS) - Q(MC))a chalcone-A chalcone-B 6′-deoxychalcone-A 6′-deoxychalcone-B O2′ CR Cβ Cγ Oγ
0.096 0.350 -0.375 -0.003 -0.076
0.115 0.354 -0.418 -0.005 -0.064
0.070 0.299 -0.334 0.004 -0.033
0.092 0.304 -0.356 0.015 -0.037
a Charges on hydrogen atoms have been added to the neighboring carbon atoms. Values are given in au.
large (about -0.5 au in all cases), but the change during the reaction is quite moderate. Then, the enolate character of this oxygen is not substantially augmented when passing from the MC to the TS. This means that interactions with the partially negatively charged carbonyl oxygen (Oγ) are not decisive enough to discriminate between the MC and the TS, and they do not lead to relative TS stabilization. Instead, these interactions can play an important role in substrate recognition and binding. Otherwise, interactions with the Cβ and O2′ atoms can be more important in differentiating between the MC and the TS, thus contributing to the free-energy barrier lowering. We have previously described these interactions in Table 2 and found important differences depending on the active site conformation. In particular, a solvation water molecule (Wat1) is hydrogenbonded to the Cβ atom, stabilizing the developing negative charge in the TS of chalcone-B. According to the electronic description of the reaction we can now understand the origin of the differences in the catalytic power of the two active site conformations analyzed. As said, the reaction proceeds with an electron density flow toward the Cβ atom. In the B active site, Lys97 is significantly closer to the substrate, and we have an important relative stabilization of the TS with respect to the MC due to the charge-charge interaction. Feedback of this electrostatic attraction occurs by the reduction of the substrate-Lys97 distance in the TS, and the substrate moves toward Lys97 (see the Lys97- NH‚‚‚Oγ distance in Table 1). Probably as a consequence of this displacement of the substrate and the loss of negative charge on the O2′-oxyanion, this atom is significantly desolvated in the B active site TSs, a feature not found in the A active site. In free-energy terms, the charge-charge interaction between the substrate and Lys97 in the B active site stabilizes the TS relative to the MC, although the price to be paid is the
Analysis of CHI Conformations and Substrates
J. Phys. Chem. B, Vol. 110, No. 41, 2006 20691
Figure 5. Time evolution of the interaction energies between chalcone and the MM environment in the A (black line) and B (red line) active site conformations.
desolvation of the O2′-oxyanion. Obviously the balance between both effects, electrostatic stabilization of a negative charge on Cβ or O2′, gives the relative ordering between the two conformations of the active site in terms of catalytic power. It is important to stress here that this comparison cannot be directly extrapolated to explain the origin of catalysis in CHI. For such analysis, one should compare the effect of the active site on the substrate (in this case, the B active site conformation) with that of a water solution.35 Our computational model provides a reasonable picture of CHI catalysis. This model does not consider the possible proton transfer from the environment to the enolate group of the substrate, and the general acid seems to be nonessential for catalysis.23 In fact, as discussed above, interactions with the carbonyl oxygen do not seem to play a decisive role stabilizing the TS but rather, most probably, in the binding process. In any case, general acid could play a role by proton transfer to the Cβ atom, although good theory-experiment agreement is obtained without considering this possibility. It also has been proposed that the release of three water molecules from the active site could contribute to diminish the free-energy difference between the MC and the TS.23 We did not find evidence for such a large change in the number of water molecules in the active site along the distinguished reaction coordinate during PMF calculation. Only in the B active site we found the release of one water molecule (Wat4) when passing from the MC to the TS. Of course, this does not mean that more water molecules could not be released in longer simulations. However, taking into account the agreement between our calculated free-energy barriers and experimental estimations, the release of additional water molecules seems not needed to explain catalysis. Moreover, that conclusion was reached by comparing the TS obtained in active site B with the MC characterized in the A site.23 Our PMFs are obtained by joining the TSs with their corresponding MCs. Both reactant conformations are stable within our simulation times, and our strategy seems to be more adequate. Interconversion among different reactant conformations could contribute to the free-energy barriers, and this is in fact a very stimulating subject in enzymatic catalysis.36 To solve this question in a definite way one would need to know the relative stabilities of the different possible Michaelis complexes and the rates of conversion among them as well as the relative rates with respect to the binding process if equilibrium is not assumed.
There is no easy computational solution to this question as the differences between the two reported active sites are quite important, and thus the transformation between them could involve a complicated pathway. As a possible indication of the relative stabilities of the chalcone and 6′-deoxychalcone Michaelis complexes in A and B active sites we computed the averaged QM/MM interaction energies as the energy difference between the QM/MM subsystem and the noninteracting QM and MM subsystems. The interaction energies were always more favorable for the B active site than those for the A one. The QM/ MM interaction energy between 6′-deoxychalcone and the environment in the MC was 12 kcal/mol larger (in absolute value) in the B active site conformation. In the case of chalcone, the difference was of 19 kcal/mol, always more favorable for the B active site. The time evolution of the interaction energies of chalcone in the two active site conformations are shown in Figure 5. It seems then that the B active site is not only more catalytic (lower activation free energies) but also displays larger interaction energies (in absolute value) with the substrate, and it could also be the preferred one to form the Michaelis complex. Obviously, other energy contributions should be considered to give a final answer. For example, the deformation energy of the enzymatic system could be quite different between both active sites, not to mention entropic contributions due to the solvent, the substrate, or the protein structure. Finally, it is also interesting to point out that the averaged interaction energy of 6′-deoxychalcone in the B active site is slightly more negative (about 3 kcal/mol) than that for chalcone, which agrees with the relative ordering of the Michaelis constants of these two compounds with CHI.21 4. Conclusions Different protein conformations can lead to significantly different catalytic rate constants. CHI offers an excellent opportunity to calibrate the influence of the active site conformation on the barrier free energies, due to the fact that two different active site conformations are observed in the crystallographic structures reported in the literature. These conformations, here named as A and B, essentially differ in the rotational state of Thr48 side chain, the closer approach between the substrate and Lys97 in the latter and the larger number of hydrogen-bonded water molecules in the former.
20692 J. Phys. Chem. B, Vol. 110, No. 41, 2006 We have computed the PMFs associated with the chalcone to flavanone transformation for two different substrates in the two active site structures. The reduced distance between the substrate and a positively charged lysine residue (Lys97) in one of the two active sites provides the necessary additional driving force for the reaction. According to our computational QM/ MM analysis, the reaction proceeds with a charge flow from the O2′-oxyanion to the Cβ atom, and then the charge-charge interaction between the substrate and Lys97 becomes stronger in the TS than that in the MC, especially in the B active site. In this last case, the reinforcement of this interaction produces an important desolvation of the O2′-oxyanion in the TS, a feature not observed in the A active site. Comparison between both substrates (chalcone and 6′deoxychalcone) has also allowed rationalizing some experimental findings, such as the fact that both the Michaelis and the catalytic rate constants are larger in the former substrate. In this sense, it seems that enzyme-substrate interactions through the 6′-hydroxyl group are relevant only in the transition state. Thus, the TS of chalcone can be additionally stabilized through these interactions. Otherwise, as enzyme-6′-hydroxyl interactions are not significant in the MCs and the binding process involves desolvation of this hydroxyl group, the binding free energy is less negative for chalcone than that for 6’-deoxychalcone. Acknowledgment. We are indebted to DGI for Project No. BQU2003-04168-CO3-01 and Generalitat Valenciana for Projects Nos. GV04B-131 and GV06-021, which supported this research. J.J.R.-P. thanks the Spanish Ministerio de Educacio´n y Ciencia for a FPU doctoral fellowship. References and Notes (1) Min, W.; English, B. P.; Luo, G. B.; Cherayil, B. J.; Kou, S. C.; Xie, X. S. Acc. Chem. Res. 2005, 38, 923-931. (2) Xue, Q. F.; Yeung, E. S. Nature 1995, 373, 681-683. (3) Fenimore, P. W.; Frauenfelder, H.; McMahon, B. H.; Young, R. D. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 14408-14413. (4) Turner, A. J.; Moliner, V.; Williams, I. H. Phys. Chem. Chem. Phys. 1999, 1, 1323-1331. (5) Zhang, Y. K.; Kua, J.; McCammon, J. A. J. Phys. Chem. B 2003, 107, 4459-4463. (6) Thorpe, I. F.; Brooks, C. L., III. J. Am. Chem. Soc. 2005, 127, 12997-13006. (7) Garcia-Viloca, M.; Poulsen, T. D.; Truhlar, D. G.; Gao, J. L. Protein Sci. 2004, 13, 2341-2354. (8) Bruice, T. C. Acc. Chem. Res. 2002, 35, 139-148. (9) Lau, E. Y.; Bruice, T. C. J. Am. Chem. Soc. 2000, 122, 71657171. (10) Lau, E. Y.; Bruice, T. C. J. Am. Chem. Soc. 1998, 120, 1238712394. (11) Kahn, K.; Bruice, T. C. J. Am. Chem. Soc 2000, 122, 46-51.
Ruiz-Pernı´a et al. (12) Zheng, Y. J.; Bruice, T. C. J. Am. Chem. Soc. 1997, 119, 81378145. (13) Kollman, P. A.; Kuhn, B.; Donini, O.; Perakyla, M.; Stanton, R.; Bakowies, D. Acc. Chem. Res. 2001, 34, 72-79. (14) Kuhn, B.; Kollman, P. A. J. Am. Chem. Soc. 2000, 122, 25862596. (15) Roca, M.; Andre´s, J.; Moliner, V.; Tun˜o´n, I.; Bertra´n, J. J. Am. Chem. Soc. 2005, 127, 10648-10655. (16) Roca, M.; Martı´, S.; Andre´s, J.; Moliner, V.; Tun˜o´n, M.; Bertra´n, J.; Williams, A. H. J. Am. Chem. Soc. 2003, 125, 7726-7737. (17) Roca, M.; Moliner, V.; Ruiz-Pernı´a, J. J.; Silla, E.; Tun˜o´n, I. J. Phys. Chem. A 2006, 110, 503-509. (18) Bednar, R. A.; Hadcock, J. R. J. Biol. Chem. 1988, 263, 95829588. (19) Dixon, R. A.; Blyden, E. R.; Robbins, M. P.; Vantunen, A. J.; Mol, J. N. Phytochemistry 1988, 27, 2801-2808. (20) Jez, J. M.; Bowman, M. E.; Dixon, R. A.; Noel, J. P. Nature Struct. Biol. 2000, 7, 786-791. (21) Jez, J. M.; Noel, J. P. J. Biol. Chem. 2002, 277, 1361-1369. (22) Jez, J. M.; Bowman, M. E.; Noel, J. P. Biochemistry 2002, 41, 5168-5176. (23) Hur, S.; Newby, Z. E. R.; Bruice, T. C. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 2730-2735. (24) Hur, S.; Bruice, T. C. J. Am. Chem. Soc. 2003, 125, 1472-1473. (25) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J. Am. Chem. Soc. 1985, 107, 3902-3909. (26) Field, M. J.; Albe, M.; Bret, C.; Proust-De Martı´n, F.; Thomas, A. J. Comput. Chem. 2000, 21, 1088-1100. (27) Kaminski, G. A.; Friesner, R. A.; Tirado-Rives, J.; Jorgensen, W. L. J. Phys. Chem. B 2001, 105, 6474-6487. (28) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926-935. (29) Moliner, V.; Turner, A. J.; Williams, I. H. Chem. Commun. 1997, 1271-1272. (30) Martı´, S.; Moliner, V.; Tun˜o´n, I. J. Chem. Theory Comput. 2005, 1, 1008-1016. (31) Torrie, G. M.; Valleau, J. P. J. Comput. Phys. 1977, 23, 187-199. (32) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett. 1988, 153, 503-506. (33) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (34) Mulliken, R. S. J. Chem. Phys. 1955, 23, 2343-2346. (35) Martı´, S.; Roca, M.; Andre´s, J.; Moliner, V.; Silla, E.; Tun˜o´n, I.; Bertra´n, J. Chem. Soc. ReV. 2004, 33, 98-107. (36) Antikainen, N. M.; Smiley, R. D.; Benkovic, S. J.; Hammes, G. G. Biochemistry 2005, 44, 16835-16843.
