Dihydrofolate reductase (DHFR) maintains the intracellular pool of tetrahydrofolate through catalysis of hydrogen transfer from reduced nicotinamide adenine dinucleotide to 7,8-dihydrofolate. We report results for pre-steady-state kinetic studies of the temperature dependence of the rates and the hydrogen/deuterium-kinetic isotope effects for the reactions catalysed by the enzymes from the mesophilic Escherichia coli and the hyperthermophilic Thermatoga maritima. We propose an evolutionary pattern in which catalysis progressed from a relatively rigid active site structure in the ancient thermophilic DHFR to a more flexible and kinetically more efficient structure in E. coli that actively promotes hydrogen transfer at physiological pH by modulating the tunnelling distance. The E. coli enzyme appeared relatively robust, in that kinetically severely compromised mutants still actively propagated the reaction. The reduced hydrogen transfer rates of the extensively studied Gly121Val mutant of DHFR from E. coli were most likely due to sterically unfavourable long-range effects from the introduction of the bulky isopropyl group.
5,6,7,8-Tetrahydrofolate (H4F) is required for the biosynthesis of thymidylate, purines and several amino acids. Its intracellular levels are maintained through the nicotinamide adenine dinucleotide phosphate (NADPH)-dependent reduction of 7,8-dihydrofolate (H2F) catalysed by the ubiquitous enzyme dihydrofolate reductase (DHFR; figure 1). This enzyme has therefore been a long-standing pharmacological target and hence has been studied extensively by X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy and computation (Sawaya & Kraut 1997; Casarotto et al. 1999; Feeney 2000; Radkiewicz & Brooks 2000; Agarwal et al. 2002a,b; Shrimpton & Allemann 2002; Benkovic & Hammes-Schiffer 2003; Garcia-Viloca et al. 2003a,b; Shrimpton et al. 2003; Thorpe & Brooks 2003, 2004; Venkitakrishnan et al. 2004; McElheny et al. 2005; Pu et al. 2005). While structural studies of enzymes have significantly influenced our view of catalysis, the static three-dimensional pictures provided by these studies do not incorporate the wide range of dynamical motions that occur in proteins. DHFR has recently served as a paradigm for the development and testing of concepts that relate the enormous catalytic power of enzymes to their dynamic properties.
Several studies of DHFR have indicated the importance of its dynamic properties for catalysis. DHFR from Escherichia coli (EcDHFR) is a monomeric enzyme consisting of four α-helices, eight β-sheets and four mobile loops (figure 2). The enzyme is separated into two domains, the adenosine binding domain and the loop domain. Previous studies have indicated the central role of the M20, the FG (residues 116–132) and the GH (residues 142–150) loops for the catalytic activity and mechanism of DHFR. The M20 loop adopts the closed conformation in the reactive ternary complex when both H2F and NADPH are bound (Sawaya & Kraut 1997). This conformation is stabilized through hydrogen bonds between residues in the M20 and the FG loops. The backbone nitrogen atoms of these loops displayed high dynamic mobility in NMR relaxation experiments, which has been interpreted to suggest a connection between the dynamic properties of these loops and the catalytic behaviour of DHFR (Epstein et al. 1995; Osborne et al. 2001). The relationship between movements of these loops and catalysis has been probed by site-directed mutagenesis. Replacement of Gly121, a highly mobile residue located in the middle of the FG loop over 19 Å from the active site, with Val or Leu slowed the hydride transfer rate dramatically and weakened the binding of NADPH (Cameron & Benkovic 1997). While being strictly conserved in all prokaryotic DHFRs, Gly121 does not appear to form any interactions with other residues. Molecular dynamic (MD) simulations of EcDHFR revealed a strong correlation between the movement of the catalytically important M20 and FG loops (Radkiewicz & Brooks 2000). These correlated motions were observed only in reactive complexes of the enzyme and absent in the product complex. Mixed quantum mechanical molecular mechanic simulations and genomic sequence analysis have identified a network of hydrogen bonds and van der Waals contacts from Asp122 on the surface of the protein to the active site (Agarwal et al. 2002b). This network may facilitate hydride transfer, suggesting a direct link between the motion of the FG loop and the catalytic events in the active site. In good agreement with the kinetic measurements, computation revealed a significant increase in the energy barrier for the hydride transfer of the Gly121 to Val mutant (EcDHFR-G121V) relative to the wild-type enzyme (DHFR; Watney et al. 2003).
2. Results and discussion
(a) Hydride transfer in DHFR from T. maritima and E. coli
The measurement of the temperature dependence of the kinetic isotope effects (KIEs) of the hydride transfer catalysed by DHFR is a sensitive method to probe the influence of protein dynamics on the chemistry of the reaction. Here, we report the results of such measurements for the monomeric DHFR from E. coli, its catalytically compromised G121V-mutant and DHFR from Thermatoga maritima (TmDHFR). Unlike its mesophilic homologue, this thermophilic enzyme, which unfolds at ca 80 °C—almost 30 °C above the melting temperature of EcDHFR (Maglia et al. 2003)—is the only dimeric DHFR known at present (figure 2; Dams et al. 2000). Mutants that increase the proportion of the monomer of TmDHFR display significantly reduced thermal stability (Rodriguez & Allemann 2006, unpublished data). The dimerization interface involves the catalytically important M20 and the FG loops and the steady state turnover rate of TmDHFR is reduced by approximately 1 order of magnitude relative to the enzyme from E. coli at their respective physiological temperatures (Maglia et al. 2003).
The hydride transfer rates for the reduction of H2F during Tm- and EcDHFR catalysis were measured as a function of temperature by fluorescence resonance energy transfer from the protein to the reduced nicotinamide moiety of NADPH. For TmDHFR, a biphasic temperature dependence was observed at pH 7.0 with a break point at approximately 25 °C (figure 3; Maglia & Allemann 2003; Maglia 2004). Below 25 °C the KIE increased with decreasing temperature, while above 25 °C the KIE was temperature-independent, at least within the relatively narrow range of experimentally accessible temperatures. The ratio of the Arrhenius pre-exponential factors, which was obtained from the extrapolation of the temperature dependence of the KIE, was inverse for the lower temperature range (AH/AD=0.002) and close to unity above 25 °C (AH/AD=1.56). The activation energies of hydride and deutride transfer were obtained from the temperature dependence of the reaction rates to the empirical Arrhenius equation. For the temperature range from 25 to 65 °C and for hydride and deuteride transfer were similar (; ), while below 25 °C the difference of the activation energies increased owing to an increased activation energy for deuteride transfer.
Several theoretical approaches to hydrogen transfer have been proposed that treat the hydrogen coordinate completely quantum mechanically and incorporate various degrees of heavy atom motions that modulate the tunnelling barrier (Bruno & Bialek 1992; Borgis & Hynes 1996; Agarwal et al. 2002b). In one such model, a distinction was made between motions that actively modulate the tunnelling barrier and passive motions that merely help bring the substrates into the reactive configurations (Kuznetsov & Ulstrup 1999; Knapp & Klinman 2002; Klinman 2003). This model predicts that when active dynamic motions are dominant and generate a temperature-dependent tunnelling distance, the KIEs become highly temperature-dependent and the ratio of the pre-exponential factors becomes inverse. Within the framework of this model, the experimental results observed for TmDHFR catalysis suggest that at low temperature, TmDHFR actively modulates the tunnelling distance and hence the reaction rates. However, at higher temperatures, the rigidity of the enzyme necessary to ensure its thermal stability appears to prevent such active tunnelling, and only passive motions that generate reactive configurations of substrate and cofactor promote the reaction. It is perhaps not surprising that low-frequency protein modes become more important for the reaction at lower temperatures as they may be excited at temperatures well below the C–H stretching mode, the excitation of which is much more temperature-dependent. A molecular dynamical simulation of the TmDHFR catalysed reaction at 5, 25 and 65 °C using ensemble-averaged variational transition state theory with multidimensional tunnelling confirmed significant contributions from H-tunnelling at all temperatures, with the relative contributions to the overall reaction rate decreasing with increasing temperatures (Pang et al. 2006).
H and D transfer catalysed by the mesophilic EcDHFR at pH 7.0 revealed a monophasic and relatively strong dependence of the reaction rates on the temperature resulting in activation energies of and leading to an inverse pre-exponential factor AH/AD=0.108 and temperature-dependent primary KIEs (figure 4; Swanwick et al. 2006). Within the dynamic model proposed by Knapp & Klinman (2002), the inverse AH/AD and the temperature dependence of the KIEs observed here suggest a dynamic active site for EcDHFR at pH 7.0 that, unlike the active site of TmDHFR, actively modulates the tunnelling distance and hence the reaction rates in the physiological temperature range.
EcDHFR appears to have evolved an active site structure that is organized to support hydrogen tunnelling and to use active dynamics to promote hydrogen transfer at physiological pH. Relatively small increases in the pH of the solution, however, lead to a stiffening of the active site that no longer allows DHFR to actively promote hydrogen transfer, resulting in a reduction of the reaction rate by more than 1 order of magnitude. Above pH 9, EcDHFR displayed a temperature dependence of the reaction rates similar to that of the structurally more rigid DHFR from the thermophilic bacterium T. maritima (Knapp & Klinman 2002; Sikorski et al. 2004), suggesting that they both relied on passive motions only to generate active site configurations conducive to hydrogen transfer. These observations may suggest an evolutionary pattern in which catalysis progressed from a relatively rigid active site structure of DHFR from the ancient thermophile T. maritima to a more flexible and kinetically more efficient structure in E. coli that actively promotes hydrogen transfer at physiological pH.
(b) Basis of reduced activity of EcDHFR-G121V
Gly121 in the FG loop of DHFR is on the exterior of the protein and approximately 19 Å from the centre of the enzyme. Several studies have shown that replacement of this residue can lead to significant reductions of the hydride transfer rates. Together with NMR and computational studies, these results have been interpreted as evidence for the existence of a network of coupled dynamic motions that include residues in exterior loops of DHFR and promote hydride transfer. Replacement of Gly121 has been proposed to lead to alterations in the network of coupled motions and altered dynamic properties of the enzyme resulting in a reduction of its catalytic efficiency (Watney et al. 2003).
Similar to the wild-type EcDHFR, a relatively strong temperature dependence of the hydride transfer rates and the KIEs was observed between 5 and 35 °C for catalysis by the mutant (Swanwick et al. 2006). The rates increased almost threefold in the temperature range leading to activation energies of and for H- and D-transfer, respectively, and temperature-dependent KIEs (figure 4). Together with the strongly inverse pre-exponential factors (AH/AD=0.0025), these results suggested that the reaction occurred with a significant contribution from the quantum mechanical tunnelling coupled to active dynamic motions.
Our observations indicated that, while replacing residue 121 in DHFR with Val leads to a significant reduction in the rate of hydrogen transfer, the general mechanism by which the reaction is coupled to the environment is unaltered, at least in a qualitative sense. This was in good agreement with MD simulations and CD and fluorescence experiments which had suggested that conformational changes that have been transmitted to the active sites as a consequence of the geometric constraints imposed by the bulky isopropyl group in EcDHFR-G121V may be the reason for the reduced catalytic efficiency of the mutant (Swanwick et al. 2004; Thorpe & Brooks 2004).
The pre-steady-state measurements of the hydride transfer event in EcDHFR-G121V catalysis revealed an isotope insensitive step before hydride transfer that occurred with a rate constant of approximately 3.5 s−1 at ambient temperature and could be interpreted as a conformational change (Cameron & Benkovic 1997; Swanwick et al. 2006). Thorpe and Brooks suggested previously that substrate and cofactor and NADPH populate preferentially a region of configuration space of EcDHFR from which the reaction can take place (Thorpe & Brooks 2004). In EcDHFR-G121V, however, NADPH and H2F are postulated to become trapped in unproductive configurations (Swanwick et al. 2004). The temperature dependence of the hydride transfer indicates that once NADPH and H2F are bound in the reactive configuration, both mutant and wild-type actively promote hydride transfer.
In summary, analysis of homologous DHFRs and their mutants has provided evidence that hydrogen transfer occurs to a significant amount by quantum mechanical tunnelling promoted by the environment. Structural changes as a consequence of mutations of EcDHFR appear to give rise to changes in these promoting motions leading to reduced hydrogen transfer efficiencies as a consequence of greatly changed ratios of pre-exponential factors. On the other hand, in the hyperthermophilic DHFR from T. maritima, the necessary stability against thermal denaturation appears to result in a rigid and hence less efficient enzyme.
The invaluable support of Donald Truhlar and Jiali Gao and stimulating discussions with Amnon Kohen are gratefully acknowledged. This work was generously supported by the BBSRC and Cardiff University.
One contribution of 16 to a Discussion Meeting Issue ‘Quantum catalysis in enzymes—beyond the transition state theory paradigm’.
- © 2006 The Royal Society