Significance

Abstract

Results

pL Dependence of WT ecDHFR Hydride Transfer Rate.

The experimental pH ranges in previous reports were limited to pH 9.5 or lower (1, 6, 15), and upon expanding the experimental conditions to higher pHs, a second plateau was observed (Fig. 2A). This finding is consistent with parallel reaction pathways separated by one ionization event (Fig. 3). The hydride transfer rate constant (k_hyd) in the low-pH domain (k₁) is computed to be 830 ± 70 s⁻¹ (SI Appendix, Table S3), which agrees with the previously determined value of ~950 s⁻¹ for the same process (1). The second k_hyd rate constant in the high-pH plateau, k₂, is 0.13 ± 0.01 s⁻¹, which is ~6,000-fold lower than k₁. The two pathways are separated by a kinetic pK_a of 6.70 ± 0.05, which is consistent with previously determined values of ~6.5 (1, 5).

Open in a separate window

Fig. 2.

Log k_hyd vs. pH/pD for the WT ecDHFR at 25 °C with NADPH in H₂O (blue □) and NADPH in D₂O (red ●) (A) and NADPD in H₂O (blue □) and NADPD in D₂O (red ●) (B). The data points were fit to SI Appendix, Eq. S1, and the rate constants are summarized in SI Appendix, Table S2.

Open in a separate window

Fig. 3.

Proposed reaction scheme for the ecDHFR-promoted hydride transfer reaction with parallel pathways. EP, enzyme/products complex; ES, ternary enzyme/substrate Michaelis–Menten complex.

A similar sigmoidal relationship was also found for the WT enzyme-catalyzed reaction in D₂O (Fig. 2A). The rate/pD profile (pD = −log[D₃O⁺]) crosses over once with the rate/pH profile at approximately pL = 7 (pL = −log[H₃O⁺] or −log[D₃O⁺]). As a result, a normal solvent kinetic isotope effect (SKIE) value was found at low pH (k₁) and an inverse SKIE was found at high pH (k₂) (Table 1). It should be noted that linear relationships were observed in the proton inventory studies in both plateau regions (SI Appendix, Fig. S2). This observation is consistent with the involvement of only one proton in the normal SKIEs observed at low pH, but with multiple protons leading to inverse SKIE at high pH (22). Although the SKIE at low pH can be attributed to the protonation of N5 of the DHF, the SKIE at high pH is indicative of the involvement of a whole network of protons in the protonation of N5-DHF. This explanation accords well with the remote effect of protonation of Y100 on D27 and the protonation of N5-DHF discussed below. By that interpretation, many exchangeable protons involved in the network of H-bonds related to the protonation of Y100, D27, and N5-DHF are more tightly bound in the transition state than at the ground state. Based on the fact that the observed 1° KIE was unchanged across the pH range measured, we can rule out the alternative explanation of inverse SKIEs on the DHF binding to the enzyme–NADPH complex (and the associated conformational/protonation states) before the protonation of N5. This mechanism would have led to an inflated commitment (C_f on single-turnover rates) and deflated 1° KIE at higher pH. The observed pH-independent 1° KIE also agrees well with the previously reported values of 2.7–3.0 for the WT ecDHFR (1, 13, 15). The data from the multiple isotope effect (23, 24) study (Table 1) show that both SKIE and 1° KIE are suppressed by deuterium substitution at low pH (k₁) but are orthogonal to each other at high pH (k₂).

Table 1.

SKIE and 1° KIEs in the plateau regions of the WT ecDHFR hydride transfer pL/rate profiles

pH region	SKIE (NADPH)	SKIE (NADPD)	KIE (H2O)	KIE (D2O)
Low pH (k1)	2.0 ± 0.2	0.9 ± 0.1	2.6 ± 0.3	1.2 ± 0.1
High pH (k2)	0.57 ± 0.09	0.53 ± 0.08	2.7 ± 0.3	2.6 ± 0.4

Open in a separate window

pL Dependence of D27S and Y100F Mutations.

The pL dependence of the k_hyd in the D27S, Y100F, and D27S/Y100F mutants was investigated in detail. At pH 7 in water, the rate for the NADPH-dependent reduction of DHF dropped by ~3,400-fold due to the D27S modification (k_WT = 279 s⁻¹, k_D27S = 0.082 s⁻¹), by ~14-fold due to the Y100F mutation (k_Y100F = 19.8 s⁻¹), and more than 100,000-fold with both D27S and Y100F mutations (k_Y100F/D27S = 0.0027 s⁻¹). Thus, the rate reduction in the double mutant is greater than the sum of the individual D27S and Y100F effects (i.e., the changes in the free energy barriers of the mutants satisfy the following relation: ΔΔG^‡_Y100F/D27S > ΔΔG^‡_Y100F + ΔΔG^‡_D27S, where G^‡ stands for the activation free energy). All three ecDHFR variants (D27S, Y100F, and D27S/Y100F) exhibit linear pL/rate dependence, which differs from the sigmoidal behavior observed in the WT enzyme (Figs. 2 and ​and44 and SI Appendix, Fig. S3). This behavior is true regardless of the identity of the cofactor [β-Nicotinamide adenine dinucleotide phosphate reduced (NADPH) or [Carbonyl-¹⁴C]-NADPH, (R)-[4-³H]-NADPH (NADPT) and (R)-[4-²H]-NADPH (NADPD)] or the solvent (H₂O or D₂O). The slopes for the linear pL/rate relationship were found to be slightly below unity (−0.6 to −0.9). Such nonunity dependencies have been observed before, and they are likely the result of multiple concurrent reaction pathways, each with individual kinetic rate constants (4, 25, 26).

Open in a separate window

Fig. 4.

Log (k_hyd) vs. pH/pD for the D27S/Y100F ecDHFR mutant at 25 °C. The data points are presented as NADPH in H₂O (●) and NADPH in D₂O (○) and NADPD in H₂O (■) and NADPD in D₂O (□). The parameters used for the fitted linear lines are given in SI Appendix, Table S1.

Because there are no plateau regions in the pL/rate profiles for the ecDHFR mutants, multiple isotope effect analyses (23, 24) were conducted at specific pL values of 5.1, 6.0, 7.0, and 8.2 (SI Appendix, Table S2). The Y100F mutant exhibited an inverse SKIE at high pH, and the SKIE progressively increased to slightly above unity as the pH was lowered to 5.1. Conversely, the primary KIE value starts at ~3.2–3.3 at pH 8.2, but it progressively decreases in magnitude in going to lower pHs. For the D27S and D27S/Y100F mutants, inverse SKIE values were found between pH 5.1 and pH 8.2, and the magnitude of the SKIE stayed constant. Similarly, the 1° KIE values on the hydride transfer reaction in these two ecDHFR mutants are ~2.7 and insensitive to pL changes.

Discussion

DHFR is one of the most well-studied enzyme systems and is commonly used as a model system for biochemical/biophysical inquiries. Although much is known about its catalytic cycle, inhibition characteristics, and structural dynamics, many mechanistic details regarding the chemical step of catalysis (DHF→THF) are still unresolved. This situation is troubling because the numerous mutagenesis studies that examine how the chemical step is being affected by specific mutations remain incomplete. Without a comprehensive understanding of the reaction mechanism, it is difficult to attribute the kinetic effect of the mutation(s) definitively.

To gain further mechanistic understanding, we first extended the pH range of the experimental conditions and found sigmoidal-shaped pH/rate profiles (Fig. 2), which are consistent with two parallel reaction pathways consisting of the ternary enzyme/substrate (ES) Michaelis–Menten complex and the associated deprotonated form, ES⁻¹ (separated by one ionization step). Although multiple ecDHFR ternary/binary complexes in solution have been reported (28), two catalytically competent ternary complexes, separated by protein ionization, were unknown (SI Appendix, Fig. S4). Also, the origin of the activity change with a pK_a of 6.5 had drawn several interpretations. The data here show that there is a second catalytic complex that exists predominantly at higher pH and is capable of converting DHF into THF, albeit with a much lower rate constant. The kinetic pK_a determined from the pH/rate profiles is associated with an ionization event that is related to the interconversion between these two reactive complexes (Fig. 3 and SI Appendix, Eq. S1), and not necessarily related to a change in the order or timing of DHF-N5 protonation and hydride transfer. As a consequence, the system will contain different ratios of the two reactive species at the various pH values in Fig. 2. It should be noted that the presence of two reactive ternary complexes that occupy the high- and low-pH domains has been hinted at in a recent study (G51PEKN, N23PP/G51PEKN, and N23PP/L28F/G51PEKN ecDHFR mutants) (5), but the implications were unclear at that point.

Because the reduction of DHF involves both proton transfer to its N5 and hydride transfer to C6, examination of the cross-effects of 1° KIEs (probing the hydride transfer) and SKIEs (probing the protonation step) is useful for dissecting the timing of these two processes. If protonation and hydride transfer occur simultaneously in a concerted mechanism (middle mechanism in Fig. 5 and SI Appendix, Fig. S5, Bottom), the observed 1° KIE should increase in deuterated solvent because the deuterium would increase the energy barrier for the hydride transfer event, making it more rate-limiting (24). Thus, the data in Table 1 are inconsistent with a concerted mechanism. At low pH, both 1° KIE and SKIE are suppressed by deuterium in the other step. At high pH, on the other hand, the 1° KIEs and the SKIEs are unchanged by D₂O and NADPD, respectively, suggesting those observed KIEs are not masked by deuterium in the other step. These data are consistent with a stepwise mechanism illustrated in SI Appendix, Fig. S5 (Upper). Previously it was suggested, based on KIEs, that the reaction mechanism might shift to a concerted process at a higher pH of 9.5 (15). However, this assessment overlooked the fact (24) that in this case, the inverse SKIEs at high pH would suppress a 1° KIE for a concerted process. That behavior was not observed. More importantly, a shift to a concerted mechanism would mean that the rate of the reaction would continuously drop by 10-fold for every pH unit increase, which is inconsistent with the high-pH plateau in Fig. 2.

Open in a separate window

Fig. 5.

Potential mechanistic pathways for the WT ecDHFR proton and hydride transfer reactions.

There are two potential stepwise mechanisms, shown in Fig. 5 (Top and Bottom), and they differ in the chronological ordering of events. The kinetic observable, oxidation of NADPH, may capture both the isotope effects on the hydride transfer and proton transfer. If hydride transfer occurs before the protonation step in a stepwise mechanism (Fig. 5, Top), one should find a nonunity 1° KIE and no significant SKIE effect (SKIE = 1) because the solvent-assisted protonation step would be after the kinetic observable (loss of NADPH fluorescence energy transfer). Table 1 shows that this situation is not the case, suggesting that the actual mechanism involves N5-DHF protonation preceding the hydride transfer (Fig. 5, Bottom).

The second ternary complex that exists at high pH is much less reactive, with its free energy barrier for DHF reduction being ~5.2 kcal/mol higher than the free energy barrier for DHF reduction for the more reactive species. This change could cause the kinetic expression to shift from a steady-state approximation (low pH; SI Appendix, Eq. S7) to one that involves a preequilibrium approximation (high pH; SI Appendix, Eq. S8). In other words, when the k_hyd << k_prot′ (protonation rate), the steady-state expression would simplify into a preequilibrium approximation. This interpretation is consistent with the observed inverse SKIE at high pH, which is indicative of a preequilibrium step. The inverse SKIE can then be attributed to the acid/base equilibrium of DHF being affected by the higher acidity of D₂O compared with H₂O (29, 30). Under comparable conditions, a higher concentration of the DHF-H⁺/D⁺ intermediate will be formed in D₂O, relative to H₂O, resulting in a larger observed rate (k_hyd * [DHF-D⁺]). It should be pointed out that global effects of higher medium viscosity (D₂O) and increased protein rigidity (N − D = O hydrogen bonds) have been ruled out as the cause for the inverse SKIE data in ecDHFR (15).

The relationships between the observed KIEs on WT enzyme k_hyd and their intrinsic values further support the above interpretations. The temperature dependence of the intrinsic KIEs is pH-independent (zero at both high and low pH), which further supports our conclusion that the C-H→C hydride transfer step, per se, is not concerted with the protonation step in either pH range. The data clearly reveal that the differences between the temperature-independent deuteride transfer rate constant ^Dk_hyd at pH 9 (SI Appendix, Fig. S1A) and temperature-dependent ^Dk_hyd at pH 7 (SI Appendix, Fig. S1B) actually result from a pH-dependent commitment to catalysis on the k_hyd (C in SI Appendix, Eq. S6). The likely explanation for these findings is that the commitment in the DHF protonation step reflects the steady-state and equilibrium approximations for the fraction of protonated DHF in the reactive Michaelis–Menten complex committed to the C-H→C hydride transfer.

We also examined the roles of two active site residues, D27 and Y100, which might play a role in the protonation step and the positioning of the substrates and water molecules for the subsequent hydride transfer, based on crystal structures and computer simulations. D27 has been shown in previous studies to be important for facilitating a solvent-assisted proton delivery to the N5 position of DHF, thus generating the protonated substrate for the subsequent hydride transfer (4, 16–18, 31). The direct source of the proton is likely to be a water molecule that could enter the active site as the Met20 loop changes conformations (18), and crystal data supporting this hypothesis are provided in the accompanying work by Wan et al. (32). Y100 has been implicated to provide electrostatic stabilization of the developing positive charge on the nicotinamide moiety, as well as exhibiting strong electrostatic interactions with NADP⁺ and folate (20, 21).

Mutagenesis studies show that the D27S, Y100F, and D27S/Y100F mutants all exhibit linear pH/rate profiles with no indication of a plateau region in either the low- or high-pH regions (Fig. 4 and SI Appendix, Fig. S3). This disruption of the WT pH dependence observed suggests that these two residues are important for the proton transfer to DHF. Through appropriate electrostatic interactions with the substrate and cofactor, Y100 and D27 provide optimal conditions for the solvent-assisted proton transfer step of the reaction (4, 7, 18, 20). In all mutants, 1° KIE values above unity were found (SI Appendix, Table S2). Inverse SKIE values were found in all cases across the probed pH range, which indicates a preequilibrium protonation of DHF in a stepwise mechanism for the reduction of DHF. Thus, the mutations lead to a suboptimal catalytic environment but do not alter the sequence of protonation followed by hydride transfer. Whereas the D27S mutation induces a greater catalytic penalty than the Y100F mutation, the rate reduction observed in the D27S/Y100F double mutation is greater than the sum of the respective effects of the individual mutations (i.e., ΔΔG^‡_Y100F/D27S > ΔΔG^‡_Y100F + ΔΔG^‡_D27S; SI Appendix, Table S4), strongly suggesting a synergistic function of these two residues.

The role of Y100 and D27 in the hydride transfer step (just C-H→C) was further investigated through the temperature dependence of the intrinsic 1° KIEs, and our findings support the synergism of D27/Y100. As shown by several theoretical (33–36) and experimental (9) studies, the temperature dependence of intrinsic KIEs is sensitive to changes in the hydride donor-acceptor distance fluctuations. Fig. 6 and Table 2 indicate that each single mutant not only has a larger temperature dependence of intrinsic KIEs than the WT enzyme but also that the effect of the double mutant is more than additive (Table 2; ΔE_{a Y100F/D27S} > ΔE_{a Y100F} + ΔE_{a D27S}). Replacing either Y100 or D27 resulted in inflated and more temperature-dependent intrinsic KIEs on hydride transfer, suggesting that these residues play a role in the narrow donor-acceptor distance distribution of the WT.

Open in a separate window

Fig. 6.

Arrhenius plots of the intrinsic H/D KIEs of the hydride transfer reaction catalyzed by WT (black) (27), Y100F (blue), D27S (green), and D27S/Y100 (magenta) ecDHFRs.

An interesting question regarding the second catalytic competent E:NADPH:DHF complex in the high-pH domain is the protonation of N5 to generate the reactive species for the hydride transfer event (here E refers to the enzyme). If the pK_a of the N5 stayed constant at ~6.5 (1, 12, 16), there would not be sufficient concentration of the reactive species E:NADPH:DHF-H⁺ in the high-pH domain for the subsequent hydride transfer reaction to occur at the observed rate. In fact, under single-turnover conditions (3 μM total ternary species), the second-order rate constant to protonate and achieve the observed k₂ value (0.13 s⁻¹) in the high-pH plateau region of Fig. 2A (WT enzyme, NADPH, and H₂O) from E:NADPH:DHF would be above the diffusion-controlled limit for a biomolecular process in water (37) (~4× diffusion limit at pH 12). Because it is known that the N5 pK_a can fluctuate (up to ~9) depending on the active site conformation (19), it is reasonable to speculate that structural rearrangement will impose a change in the active site electrostatics, which might affect N5 acidity. We focused on the effect of the Y100 residue, a high-pK_a residue close to the reactive center, and found that the deprotonation of Y100 alone can greatly perturb the N5 pK_a by ~4.7 units (i.e., up to ~11.2; Table 3). Therefore, a more anionic active site environment (with Y100/other residues deprotonated/partially deprotonated) could increase the N5 pK_a to ensure sufficient DHF protonation under alkaline conditions.

Conclusion

The mechanistic details of both proton and hydride transfers in the reaction catalyzed by ecDHFR have been elucidated through multiple isotope effects (SKIEs on 1° KIEs) (23, 24) over a broader pH range than has been previously studied. Whereas the measured rates and KIEs at the pH range of 7–9 were the same as reported (1, 6, 15), the broader pH range revealed a previously unidentified catalytically active ternary complex that exists predominantly at high pH. The data at both low and high pH support a stepwise mechanism in which protonation of the N5 atom of DHF precedes the reduction of a DHF-H⁺ intermediate by NADPH and no mechanistic shift as a function of pH.

Also, the active site D27 and Y100 residues play an important and synergistic role in facilitating both the proton transfer and hydride transfer steps. Mutations of D27, Y100, or both disrupt the sigmoidal pH/rate dependence observed in the WT. The protonation of DHF is likely to occur through a water molecule, as shown by the accompanying crystallographic data by Wan et al. (32), whereas the D27 and Y100 residues may help to orient the substrates and the water molecule through various electrostatic interactions. Further evidence of the important role of Y100 was obtained from FEP calculations of the pK_a of DHF N5. These calculations indicate that the N5 pK_a might be significantly altered by the active site electrostatics via the ionization state of residues, such as Y100, to ensure that DHF can be efficiently protonated before the hydride transfer event.

These findings provide an important validation for computational simulations concluding that protonation of DHF occurs before the transfer of a hydride, and further justify approaches that assumed a protonated DHF for simulating the free energy barriers for the hydride transfer reactions of ecDHFR/ecDHFR mutants (10–13, 38). Furthermore, the results here provide additional insights on how ecDHFR catalysis might be affected by pH, which could induce large effects on active site ionization states and electrostatics. Because ecDHFR catalysis has emerged as a popular model system for enzyme-catalyzed reactions, a better understanding of its reaction mechanism may have a broad impact on enzymology in general.

Materials and Methods

Supplementary Material

Acknowledgments

Footnotes

References