In this chapter we report the backbone and side-chain dynamics of DHFR G121V in complex with NADPH and MTX. This represents the first effort to examine the solution dynamics of DHFR G121V in the closed conformation. We show that the G121V substitution does not dramatically alter the dynamics of DHFR. However, the residues that do exhibit significant changes are functionally important in DHFR. The active site of DHFR is dynamically perturbed, although not to the extent one would expect given the kinetic nature of DHFR G121V. These results support experimen- tal findings that G121V mainly alters the hydride donor–acceptor distance (Wang, Goodey, Benkovic and Kohen, 2006a; Wang, Tharp, Selzer, Benkovic and Kohen, 2006). However, subtle changes at functionally important residues suggest dynam- ics may be a mechanism for propagating long-range coupling. Additional studies of DHFR G121V M42W may serve to elucidate the dynamic mechanism of long-range communication.
Chapter 5
Perspective and Future Directions
The study presented here had two purposes: to begin to understand the dynamics of DHFR drug inhibition and to interrogate the dynamic effects of mutation. In Chapter 2, we examined the changes in dynamics due to drug binding. Chapters 3 and 4 inves- tigated the dynamics of functionally important mutants in DHFR. In essence, these studies can be viewed as efforts towards understanding how dynamic perturbations are propagated out of, and relayed into the active site of an enzyme. These stud- ies represent a fundamental step in understanding how or even if, motions in distal regions of a protein promote ligand binding, catalysis and functional regulation.
In light of reports that propose the rate of ligand binding and release are regulated byµs-ms dynamics, we hypothesized that tight binding inhibitors would quench these motions. Surprisingly, we found that, in the case of drug binding, the rate of µs-ms conformational exchange was independent of the ligand bound. However, the residues that move were modulated by the drug. This result leads to the possibility that it is not the rate per se that is important for ligand affinity but the nature of the dynamical event as a whole. If the protein cannot make the conformational transition it has evolved to make, then the ligand cannot be ejected. If this hypothesis is true, then the interactions between the protein and the ligand may be the most important
It is tempting to test this hypothesis by examining a set of chemically different ligands. However, the result may be difficult to interpret because a protein–ligand interaction within complex “A” is drastically different than compound “B”. To ad- dress this issue, a series of similar ligands should be synthesized with only slight differences, in order to interrupt individual hydrogen bonds or hydrophobic interac- tions. If this is not possible, conservative mutations on the protein may be used to disrupt these interactions. In the case of DHFR, most of the protein–ligand contacts are side-chain mediated so mutagenesis may be indispensable in understanding the molecular interactions that result in dynamical modification.
Finally, if conformational exchange is relevant to ligand release then mutations that change the rate of exchange should correlate with kof f. We show in Chapter
3 that M42W increases the rate of conformational exchange in the pABG binding cleft of DHFR in addition to drastically altering the ps-ns dynamics of the protein. This increase in dynamics is independent of ligand release in this complex because the experiments were performed under saturating MTX concentrations. Thus, these motions likely represent an intrinsic dynamical model resulting from the mutation. Studies of the product release complex (M42W:NADPH:THF) would address the issue of whether protein motion correlates with product (or ligand) release.
It may be difficult to definitively correlate internal dynamics on theµs-ms timescale with product release from the mutant or even the wild-type protein. As described in Chapters 2 and 3 and by others (Boehr et al., 2006) the rates of DHFR conformational exchange appear to be correlated with function. However, a definitive link has yet to be established. A series of relaxation dispersion and complementary kinetic experi- ments using different crowding agents may provide valuable insight into the dynamic contributions to product release. While the macroscopic rate may not change, one would anticipate that the populations of ground and excited state complexes would
change as a function of viscosity, thus changing k1. This type of analysis should be performed using two or more co-solutes at several different concentrations. This type of analysis would require a large investment of NMR spectrometer time. However, the results may provide a fundamental link between protein dynamics on the µs-ms timescale and function.
It is currently in vogue to attribute drastic changes in catalysis or ligand binding affinity to dynamic modulation. We show definitively in Chapter 4 that this assump- tion may be invalid in some cases. However, the lack of dynamical perturbation in DHFR G121V does not necessarily mean these motions are not important in every case. Actually, the lack of dynamical change on the ps-ns timescale is encouraging in light of kinetic and theoretical evidence that suggests these motions do not contribute to regulating catalysis in this mutant protein. In the same light, the wide spread ps- ns dynamical change in the M42W mutant are supported by previous studies. These results lend credence to the notion that the ps-ns dynamical changes measured in this study are functionally relevant.
Dynamic investigations of the DHFR M42W G121V mutant is a tempting future direction. The double mutant drastically alters the dynamics of hydride transfer (Wang, Goodey, Benkovic and Kohen, 2006a). In theory, the double mutant should show the largest change in backbone and side-chain dynamics if these motions are functionally relevant. The E:NADPH:MTX model system examined here is a prime candidate for investigating a complex that resembles the transition state complex. However, the prospect of obtaining high quality data from the M42W G121V double mutant is not good. The G121V substitution drastically decreases the stability of the closed complex. Excessive line broadening increased the error of backbone and side- chain dynamics measurements. The stability of the double mutant is only expected to be worse. In light of these facts, it may be worth examining the G121A M42W
DHFR double mutant. Given the smaller side-chain, the G121A mutant is expected to be more stable. In addition, spectroscopic analysis indicates G121A and M42W are non-additive (Rajagopalan et al., 2002). Presumably, the dynamic contribution to catalysis would be evident in the NMR relaxation experiments.
In each chapter, the ps-ns dynamical changes propagated to the adenosine binding domain. In particular, ligand binding and two different mutants significantly modify the dynamics of I61. This residues has been shown to be conserved and part of a proposed “allostery wiring network” (Chen, Dima and Thirumalai, 2007). This po- sition may serve to correlate motions required for DHFR function. Thus, a study of mutant DHFRs with substitutions at I61 may enhance our knowledge of how long- range interactions mediate enzyme chemistry, or even if change in ps-ns dynamics are functionally relevant. This study should be approached using both kinetic and dynamic experiments. Naturally, the kinetics of the mutant proteins should be eval- uated first due to the rapid pace by which these measurements can be made. Unlike the elegant work of Benkovic and coworkers, the complete kinetic scheme does not need to be analyzed for these purposes. At the very least, the rate of kcat should be
determined however it would also be wise to characterize the rate of hydride transfer. This approach would allow one to screen several mutants in a very short time and select those that display the largest changes for further analysis. Other positions that should be considered include W74 and G67.
Understanding the nature of protein dynamics and their role in ligand binding, enzyme function and allosteric regulation is certainly a worthwhile endeavour. Un- fortunately, the experiments performed here and the future directions suggested in this work do not, and cannot definitively, address this issue. Only measurements on multiple timescales using nonsteady-state techniques will truly address these issues. However, the results presented here strongly suggest that protein dynamics play an
important role for enzymatic function. Furthermore, they provide the basis for future experiments using this remarkably versatile model system.