Stability and conformational memory of electrosprayed and rehydrated bacteriophage MS2 virus coat proteins

Abstract

Proteins are innately dynamic, which is important for their functions, but which also poses significant challenges when studying their structures. Gas-phase techniques can utilise separation and a range of sample manipulations to transcend some of the limitations of conventional techniques for structural biology in crystalline or solution phase, and isolate different states for separate interrogation. However, the transfer from solution to the gas phase risks affecting the structures, and it is unclear to what extent different conformations remain distinct in the gas phase, and if resolution in silico can recover the native conformations and their differences. Here, we use extensive molecular dynamics simulations to study the two distinct conformations of dimeric capsid protein of the MS2 bacteriophage. The protein undergoes notable restructuring of its peripheral parts in the gas phase, but subsequent simulation in solvent largely recovers the native structure. Our results suggest that despite some structural loss due to the experimental conditions, gas-phase structural biology techniques provide meaningful data that inform not only about the structures but also conformational dynamics of proteins.

Keywords: Bacteriophage; Electrospray ionization; Gas-phase structure; Molecular dynamics simulations; Protein structure; Solvation.

Graphical abstract

Fig. 1

The bMS2 capsid. The capsid of the bMS2 virus holds an interesting feature, as it is comprised of two sequence-identical dimers: the asymmetric A/B and symmetric C/C dimer. Predominantly similar in their structure, the main divergence is the well-defined, extended FG loop in A and C protein chains, which is disordered in B chains.

Fig. 2

Adjusting protein protonation for vacuum simulations. Focus was put on lysine (Lys), arginine (Arg), histidine (His), glutamine (Gln), aspartate (Asp) and glutamate (Glu) as potential protonation sites according to Marchese et al. (2010). At first, the solvent accessible surface area of the selected residues was calculated. Afterwards, the potential protonation sites were evaluated according to their gas-phase basicity, their surface area in solution, and their location within the protein chain, where candidates were ruled out, if a protonation wasn't possible due to Coulomb repulsion. At last, the net charge of the proteins were adjusted to reflect the vacuum charge state reported in literature, +10 e (Knapman et al., 2010).

Fig. 3

The RMSF of the individual protein chains. Comparison of the average RMSF in vacuum and during rehydration, where regions of interest within the protein chains are highlighted. RMSFs are displayed on the crystal structures as B-factor putty representation. Notably, the FG loops differ considerably between the symmetric and asymmetric bMS2 dimers, whilst other β-turns possess similar fluctuations. This can further be seen for each individual chain, where the RMSF was plotted onto the protein structures.

Fig. 4

Comparison of the average CCS data of both bMS2 dimers during the 500 ns of vacuum and rehydration simulations. The cross-section areas of the proteins were averaged and plotted as dashed baseline in order to estimate the evolution of the CCS over time. Upon vacuum exposure, the proteins compact rapidly within the first few nanoseconds, decreasing their cross-section down to 1950 ​Ų. Interestingly, during the rehydration process, the bMS2 dimers show an increase in their CCS data, yet seem to not completely recover the initial solution structures.

Fig. 5

Contact maps of the last 50 ns in vacuum and during rehydration in comparison with the initial bulk simulation data. Maps depicted in the upper row reflect the comparison of the vacuum contacts for the A/B and C/C bMS2 dimer (left to right, respectively shown in the lower triangular matrix) versus the contacts present during the initial bulk simulations (upper triangular matrix). Average normalized contacts during the rehydration simulations are illustrated in the lower row, with the data for the A/B and C/C dimers shown in the lower triangular matrix. Contacts were defined as existing, if the two evaluated residues had a distance within 3.5 ​Å of each other, and further plotted to account for the 200 simulated replicas, where a value of 0 on the scale reflects that no contact existed throughout all 200 simulations. The data suggests that upon vacuum exposure, new contacts form in both A/B and C/C dimers, as a result of a compaction of the protein structures due to the hydrophobic environment. Furthermore, during rehydration, most of the contacts seem to revert partially back, resembling those during the bulk. Nevertheless, as can be further seen from the rehydration plots, the bulk structures are not totally recovered, and few vacuum-derived contacts still remain present.