Much is known about the thermodynamically favored conformation of the nuclear vitamin D receptor (VDR) and its’ role as a nuclear transcription factor; however, little is understood about its’ intrinsic flexibility (Mizwicki et al2007) and how VDR molecular ensembles may influence a given ligands affinity and function. Herein, molecular dynamics simulations were used to assess a) the molecular events dictating whether MK functions as a VDR antagonist or superagonist ligand; b) whether x-ray water is required to accurately simulate the bimolecular interaction between the VDR and ligand; and c) the effect removal of ligand, the opening helix 12 and mutation of H305 and H397 have on the VDR backbone and side-chain flexibility.
Evidence that the VDR-MK MD results, performed using an explicit solvation model, correlate well with MK structure-function results included a) the side-chain of R274 is the most flexible residue in the LBP of VDR when bound to MK; b) the backbone of K246, one of the charge clamp residues (Figure 1), becomes significantly more flexible and the clamp opens (Figure 2A); and c) the backbone entropy of helix 11 and helix 12 increases (Figure 2C). Consistent with previous flexible docking results (Mizwicki et al2009a), the MD runs also show that the side-chain of MK preferentially moves closer to L227 (helix 3) and L404 (helix 11), and away from helix 12, i.e. L414 and V418 (Figure 3C). The MD results also show that the migration of MK away from helix 12 and towards the C-terminal end of helix 11 is facilitated by the pi-pi interaction made between the MK lactone ring and H305. Migration of the MK lactone side-chain places the exo-cyclic methylene as close as it could possibly get to C403 in the x-ray VDR conformation. This movement of the MK side-chain towards L404 could increase the probability for the formation of a Michael adduct C403 (Mizwicki et al2012; Kakuda et al2010).
The conversion of MK into a VDR superagonist in VDR_FF has been postulated to be triggered by the intermolecular interaction between the F305 and the lactone ring of MK (Mizwicki et al2012; Kakuda et al2010). In support of this hypothesis, the VDR_FF-MK and 1,25D3 MD results show they have similar average charge clamp distances (Figure 2A), intramolecular interactions between helix 11 (H397 and Y401) and helix 12 (V418 and F422) side-chains (Figure 2B) and backbone/side-chain entropy. Importantly, the MD simulations demonstrate that F397 is significantly further away from 1,25D3 when compared to VDRwt, but when bound to MK, F397 and F422 become closer to one another. Overall analysis of the MD frames indicates that movement of F397 (helix 11) away from 1,25D3 is caused by F397 gravitating toward F422 (helix 12). This motion provides an explanation for why VDR_FF-1,25D3 has an increased charge clamp distance and why the electrostatic interaction between 1,25D3 and the VDR_FF complex is significantly reduced, whereas it is not for MK (Table 1).
A novel finding from the MM/GBSA interaction energies was that no real change in total interaction energy is observed for 1,25D3 or MK throughout the various MD runs; however, dramatic changes in the relative contributions of each term to the total interaction energy are observed. For example, MK had a weaker hydrophobic interaction (▵Evdw), stronger electrostatic interaction (▵Eeel) and greater polar solvation free energy (▵Epol) with VDRwt, when compared to 1,25D3 (Table 1). This indicates that the reduced vdW interaction likely drives the reduced affinity of MK for the VDR in a steroid competition assay (Mizwicki et al2009a). Combining this physical characteristic with an increased ▵Epol may be a signature that can be used in the design of future VDR antagonists (Table 1).
While we were unable to homology model the apoRXR-like opened LBD conformation, peeling of helix 12 away from the body of the LBD enhances the side-chain entropy of H305 and fractures the displaced parallel pi-pi interaction between H397 and Y401. In the closed helix 12 VDR MD run, H397 maintains the intramolecular interaction with Y401, perhaps explaining why the H397 side-chain entropy is similar in the helix 12 closed apo and holo models. It is noted that a recent variant of VDR_FF, where H397 was mutated to a Tyr rather than a Phe, showed enhanced sensitivity for vitamin D3, which has no activity in VDRwt (Ousley et al2011).
Opening of helix 12 also causes the residues of helix 11 and 12 to become significantly more flexible and gravitate toward the LBP; however unlike the C-terminal portion of helix 11 in the apoRXR crystal structure (Bourguet et al1995), H11 residues did not enter the region of the LBP normally occupied by the seco-B and A-ring of the ligand. It has been proposed that the ability of the FF mutation to greatly protect the apoVDR against trypsin cleavage by increasing the rigidity of the closed helix 12 conformation by stabilizing residues proximal to the two C-terminal trypsin sites, R402 (helix 11) and K413 (helix 12) (Mizwicki et al2007). However the MD results demonstrate that the backbone entropy of residues proximal to R402 and K413 was reduced when helix 12 was opened, rather than the apo-helix 12 closed conformer (Figure 5C and D). This result suggests that the FF mutations ability to protect against helix 12 libation may be through its’ altering the stability of other VDR conformational ensemble members than the helix 12 closed conformer. The fact that both the apo-helix 12 opened and closed VDR and VDR_FF MD runs show oscillation about a an average RMSD (Additional file2: Figure S1) support the theory that the closed helix 12 conformation of VDR is sampled in the absence of 1,25D3 and therefore a conformational ensemble model should be applied in rationalization/translation of VDR structure-function results (Mizwicki and Norman2009b).
Perhaps the most intriguing and novel finding in this study was that the replacement of x-ray water molecules with explicit water did not alter the accuracy in simulating the MD that exist between VDR and 1,25D3/MK.; even though, removal of the x-ray water allows the R274 side-chain to sample multiple conformational states in the presence of ligand. Given the kinked, x-ray geometry of the R274 side-chain is induced by 1,25D3 during the VDR-1,25D3 MD run, it is plausible that shifting the equilibrium of the R274 side-chain to favor the kinked geometry is key to the molecular switch underpinning the activation of VDR transcription. Screening of additional ligands and extending simulation times will allow for future testing of this hypothesis. In closing our MD, dihedral entropy and interaction energy calculations confirm and build upon previous structure-function studies and provide novel findings whose true value will be defined as we generate a greater understanding of the molecular dynamics underpinning VDR ligand recognition and dissociation and the apoVDR ensemble model.