Investigation of protonation state dependent conformational dynamics of the nucleotide binding domain of Hsp70 protein homolog DnaK via computational methods

Abstract

70 kDa heat shock proteins (Hsp70) are a ubiquitious and well-conserved protein family with chaperone functions. They supervise protein folding process and assist in proper protein folding and renaturation by preventing partially folded or misfolded proteins from forming amorphous aggregates and amyloid fibrils. In terms of structure, Hsp70s are composed of approximately 45 kDa N-terminal nucleotide binding domain (NBD) with ATPase activity, 25 kDa C-terminal Substrate Binding Domain (SBD) with substrate peptide binding pocket. Also, there is a conserved hydrophobic linker segment connecting the 2 domains. Hsp70 proteins do not work alone. Rather, they need aid of some other proteins called "cochaperones" to work optimally. The two auxillary cochaperones required by the Hsp70s are the 40 kDa Hsp40s(J domain proteins) and nucleotide exchange factors (NEFs). Together with the cochaperones, Hsp70 proteins form 3-membered Hsp70 chaperone machinery. For Hsp70 chaperone system to carry out all of its functions, the NBD and SBD domains must communicate with each other during the functional cycle. This mutual signal transfer and crosstalk between the domains is an intricate example of "allosteric regulation". Basically, nucleotide status of the NBD exerts an effect on substrate affinity. Reciprocally, binding of a substrate protein to the SBD stimulates ATPase activity of the NBD. In ATP bound state of the NBD, the two domains are in docked conformation. On the other hand, in ADP bound or nucleotide free state of the NBD, these two domains are undocked and become independent of each other. In this undocked conformation, the two domains are connected to each other with the hydrophobic linker. The ATP-bound state of the NBD with docked NBD-SBD conformation binds and releases substrates at much higher rates than nucleotide free or ADP-bound state; thus, ATP-bound state of the Hsp70 is known as "low-affinity state" state with substrate binding pocket open. In contrast, when the NBD lacks any nucleotide or is bound by ADP, rate of substrate binding and release occur at way slower rates. This state of the Hsp70 proteins is called as "high-affinity state" with closed substrate binding pocket. Hence, Hsp70s shuttle between open and closed states. The hydrophobic linker is an undispensable component of the Hsp70 chaperone system. By functioning as a dynamic signal switch, the linker conveys messages in either direction, from NBD to SBD and from SBD to NBD, keeping both domains in touch. In literature, the linker, particularly the hydrophobic 388 VLLL 392 sequence, have been demostrated to be vital for both interdomain communication and dynamics of the ATPase domain. Likewise, there are many residues of the NBD playing key roles in the Hsp70 cycle, particularly in the mechanism of ATP hydrolysis. Especially, D8, K70, E171, D194, T199, and D201 residues of E.coli Hsp70 homolog DnaK and correspondants of these residues in other Hsp70 homologs have been given attention. In this study it was aimed to elucidate the conformational dynamics of the nucleotide binding ATPase domain (NBD) of E.coli DnaK in nucleotide free state by means of computational simulation techniques. First of all, so as to understand if protonation states of critical residues of interest have any effect on the opening-closure dynamics of the NBD in nucleotide free state, initially open and closed structures with different protonation states were prepared. By changing the protonation states of residues D194, D201, and H226 in distinct combinations, charge states +12, +13, and +14 were attained. Four protonation states, namely D194 protonated (194pr), D201 protonated (201pr), both D194 and D201 protonated (2pr), and both D194 and D201 unprotonated (2dep) were tested for both initially open and initially closed conformations. In each case, another residue of interest H226 was taken in protonated form. To broaden conformational space scanned and explore contribution of linker position to the dynamics of the NBD, positions of the linker were changed manually multiple times for each protonation state. According to our standard molecular dynamics (MD) simulation results (each 500 ns) of DnaK 1-392 construct of initially open conformations with these 4 protonation states, it was seen that protonation state can dramatically influence the tendency of initially open conformation towards closure. The cases in which D194 were unprotonated (2dep and 201pr) exhibited a tendency to close. Especially, 2dep became totally closed after 200 ns simulation period. On the other hand, 194pr had no tendency towards closure at all. On the other hand, all of the initially closed conformations for each of the 4 protonation states of the DnaK 1-392 retained their closed forms. Based on these simulation results, it can be deduced that the energy barrier between open and closed conformations was highest for 194pr and lowest for 2dep. In order to both decide on relative abundance of open - closed structures and circumvent sampling problems encountered during classical MD simulations, we carried out temperature replica exchange molecular dynamics (T-REMD) simulations for each protonation state. In addition to the 4 protonation states with protonated H226 residue investigated during MD simulations, 2 additional states, namely D194 protonated-H226 deprotonated (194prHID226) and D201 protonated-H226 deprotonated (201prHID226), were prepared by only removing the epsilon H atom of H226 from 194pr and 201pr cases, respectively. To elucidate whether the linker favors closed conformations, apart from these 6 different protonation cases with residues 1-392 of NBD, 6 further constructs were prepared by stripping the last 4 residues 389VLLL392 of the linker away from each of 1-392 construct. In total, 12 T-REMD simulations, 6 with 1-392 and 6 with 1-388 residues, and each with 300 ns-long were performed. Based on our T-REMD results, the most essential point to be understood is that closed conformations are much more favorable compared to open ones. Considering the ratios of open conformations, even the highest fractions of open conformations obtained in the case of 194pr388 did not exceed 25%. Looking at other cases, the percentage of open conformations can be as low as 2-3% for 201pr cases. If we look at 194pr cases, irrespective of the presence of last 4 residues of linker and protonation state of H226, open structures are most abundant in the cases of 194pr. Among 194pr and thereby all other protonation states-structures, 194pr388 (226 protonated) promotes, by far, the open structures most. Additionally, the role of the length of linker in favoring closed structures over open ones becomes more prominent when H226 is protonated with nearly 7% higher closed frames in 194pr392 than 194pr388. As opposed to 194pr, regardless of linker or protonation state of H226, 201pr cases are out and away the protonation states that exhibit lowest tendency towards opening with fraction of open structures no more than 5%. Unlike 194pr, there is no evident impact of linker on opening-closure behavior. Protonation state of H226 seemed not that important in either protonation state. In agreement with the standard MD simulations, 2dep was monitored to tend to have high number of closed frames. Indeed, 2dep comes after 201pr in terms of closed structure fractions. The contribution of the linker to the closure of the NBD was minor and seen only the last 200ns. As to 2pr states, the 2nd highest fraction of open frames, both in the presence and absence of 4 terminal residues of the linker, after 194pr were obtained. Another point of interest in this study was to scrutinize the pH dependence of DnaK ATPase domain. The active site of the ATPase domain comprises multiple charged amino acid residues; therefore, it can be expected from Hsp70 proteins to be susceptible to changes in pH conditions. To the best of our knowledge, only two experimental studies in early 2000's underlining pH dependent behavior of isolated DnaK NBD have been conducted thus far. On the other hand, no computational study has paid attention to change in the dynamics and activity of either full-length protein or NBD alone with respect to pH. Hence, which of the candidate residues in the active site are in protonated or deprotonated form around physiological pH remained elusive. To solve this mystery, thermodynamic integration (TI) simulations were performed, again for both DnaK 1-392 and DnaK 1-388. For TI simulations, for each of 12 protonation states used in the T-REMD, 2 initially closed structures alongside one open structure were picked from cluster analysis performed on the T-REMD trajectories. Additionally, pKa values of E171 and epsilon position of H226 were calculated in various protonation/deprotonation scenarios of D194 and D201. Regarding pKa values of epsilon H226, we inferred that pKa values were more or less the same around 9, indicating the fact that neither protonation states of active site aspartate and glutamate residues nor the lenght of the linker altered pKa values of H226. Protonation state of E171 is likely to depend on the protonation states of D194 and D201. When these aspartates were both deprotonated, E171 was protonated. D194, D201, and H226, on the other hand, must be in protonated states around physiological pH conditions. Consequently, according to our pKa values, we got +12 charge state, not +13 or +14. All in all, combining all atom MD simulations with REMD simulations and free energy TI simulations, the following questions were tried to be clarified in this thesis study: 1) Effect of protonation state on opening-closure abundance and dynamics of NBD 2)Whether the linker induces a change in the ratio of open to closed conformations. 3) pH dependence of the NBD through demonstration of the possible protonation/deprotonation status of critical residues 4) Whether linker promotes change in pKa of any of critical active site residues. At the end, we came to these conclusions. Dynamics of the DnaK NBD opening-closure were protonation state dependent. Intriguingly, protonation of 2 active site aspartate residues have opposite effect in such a way that protonation of D194 increases the abundance of open conformations, whereas D201 promotes more closed form of the NBD. In addition, at different charge states, there must be multiple protonation states in equilibrium. Regarding the linker, there was no direct indication of the linker being involved in opening/closure or pH dependence of the isolated DnaK NBD. Likewise, no apparent role of the linker in shifting the pKa of any critical residue in the active site.