Local Bilayer Hydrophobicity Modulates Membrane Protein Stability

Abstract

Through the insertion of nonpolar side chains into the bilayer, the hydrophobic effect has long been accepted as a driving force for membrane protein folding. However, how the changing chemical composition of the bilayer affects the magnitude of the side-chain transfer free energies ($\Delta G_{sc}^{°}$) has historically not been well understood. A particularly challenging region for experimental interrogation is the bilayer interfacial region that is characterized by a steep polarity gradient. In this study, we have determined the $\Delta G_{sc}^{°}$ for nonpolar side chains as a function of bilayer position using a combination of experiment and simulation. We discovered an empirical correlation between the surface area of the nonpolar side chain, the transfer free energies, and the local water concentration in the membrane that allows for $\Delta G_{sc}^{°}$ to be accurately estimated at any location in the bilayer. Using these water-to-bilayer $\Delta G_{sc}^{°}$ values, we calculated the interface-to-bilayer transfer free energy ($\Delta G_{i,b}^{°}$). We find that the $\Delta G_{i,b}^{°}$ are similar to the "biological", translocon-based transfer free energies, indicating that the translocon energetically mimics the bilayer interface. Together these findings can be applied to increase the accuracy of computational workflows used to identify and design membrane proteins as well as bring greater insight into our understanding of how disease-causing mutations affect membrane protein folding and function.

Figure 1.. Host-sites on OmpLA exist in different water concentrations in the bilayer.

(A) A snapshot of a molecular dynamics simulation of WT OmpLA in a DLPC bilayer is shown. Phosphate atoms of the DLPC bilayer are colored orange and six host sites on OmpLA used in this study are shown as colored spheres (black:210, yellow:164, red:120, purple:212, blue:223, brown:214). (B) The gradient of water concentrations inside the phosphate plane of the bilayer is plotted as a function of position in the bilayer. These values were obtained from previously published simulations of neat DLPC bilayers. This water gradient is aligned with the structure of OmpLA at the two phosphate planes. The sites on OmpLA were chosen because they sample a wide range of z-positions positions and chemical compositions.

Figure 2.. Nonpolar solvation parameter is not constant throughout the bilayer.

The nonpolar solvation parameter, ${\sigma }_{NP}$, was determined for each site on OmpLA by calculating the slope of the linear relationship between $\Delta \Delta G_{w,l}^{\circ }$ for each nonpolar side chain at that site with the change in buried surface area compared to alanine in a Gly-X-Gly peptide (Equation 1). Linear fits were weighted by the error in $\Delta \Delta G_{w,l}^{\circ }$ (standard deviations found in Table S2) for each site, with the slope of the line $\left({\sigma }_{NP}\right)$ and the Pearson correlation coefficient reported for each fit shown at the bottom of each panel. For sites 212 and 120, beta-branched side chains isoleucine and valine were omitted from ${\sigma }_{NP}$ determination as they had anomalously greater $\Delta \Delta G_{w,l}^{\circ }$ compared to the other nonpolar side chains. The ${\sigma }_{NP}$ determined from non-beta-branched side chains (solid line in 212 and 120 panels) fits the beta-branched side chains (dotted line), indicating that the increase in energy is derived from favorable local interactions that are restricted to beta-branched residues. By taking the difference in the y-intercept for the two lines, we estimate the energy gained due to local interactions for beta-branched residues at these sites to be 2.38 kcal mol⁻¹ for site 212 and 0.80 kcal mol⁻¹ for site 120. The ${\sigma }_{NP}$ for site 223 was determined only using previously collected data.

Figure 3.. Nonpolar solvation parameter is linearly correlated with water concentration in the bilayer.

The water-to-bilayer ${\sigma }_{NP}$ calculated at each host site in OmpLA (colored as in Figure 1) and PagP (gray point) plotted as a function of water concentration (Table S4) (A) and distance from the phosphate plane (B). The ${\sigma }_{NP}$ are colored according to the scheme in Figure 1, with the ${\sigma }_{NP}$ derived from PagP position 111 colored gray. The position of ${\sigma }_{NP}$ for a given site is the average of the positions of each side chain at that site and the error bars reflect the standard deviation (Table S1, NP row). ${\sigma }_{NP}$ and water concentration are linearly correlated with the equation shown in the bottom right hand corner of Panel A (black line: R² = 0.84; 95% confidence interval shaded in light blue). Using the derived relationship between bilayer position and water concentration shown in Figure S3, the direct relationship between ${\sigma }_{NP}$ and bilayer position can be determined (Panel B). We were also able to assign a previously measured “water-to-interface” solvation parameter (−12 cal mol⁻¹ Å⁻²) to an exact position in the bilayer (3.75 Å from the phosphate plane) using this function (shown as a black square in Panel B) and find that it is reporting on the position of the lipid carbonyl plane in the bilayer (dashed vertical gray line). The position-dependent function describing ${\sigma }_{NP}$ allows for the $\Delta G_{sc}^{\circ }$ of any nonpolar side chain to be determined for any position of the bilayer inside the phosphate plane (shown in Figure 4).

Figure 4.. Bilayer position dependence of nonpolar ΔGsc∘.

Reference-free side chain transfer free energies are plotted as a function of the average distance from the bilayer phosphate plane, as determined from molecular dynamics trajectories. $\Delta G_{sc}^{\circ }$ for each host site in OmpLA are colored as in Figure 1. Error bars for both $\Delta G_{sc}^{\circ }$ and bilayer position represent the standard deviations. The solid black line in each panel represents the simulated $\Delta G_{sc}^{\circ }\left(z\right)$ profile for each nonpolar side chain. This function is derived from ${\sigma }_{NP}\left(z\right)$ multiplied by the nonpolar surface area of each side chain (Equation 7). Vertical dotted lines represent the position of the lipid carbonyl groups.

Figure 5.. The translocon energetically mimics the bilayer interface.

(A) A cartoon schematic of the co-translational insertion of a helix (blue) via the translocon (ribosome colored dark gray, translocon colored red; PDB code: 5GAE) is shown on the left. The helix can either partition to an interfacial conformation, which is energetically described by $\Delta G_{t,i}^{\circ }$, or to a transmembrane conformation, described by $\Delta G_{t,b}^{\circ }$ (t = translocon, i = interfacial, b = transmembrane). The interfacial-to-transmembrane conformation, which can occur in the absence of the translocon, is energetically described by $\Delta G_{i,b}^{\circ }$. (B) The values for $\Delta G_{t,b}^{\circ }$ and $\Delta G_{i,b}^{\circ }$ are plotted for nine side chains, with the dotted line representing energetic equivalence between the two equilibria (black = nonpolar, gold = aromatic, blue = ionizable). For all side chains, except Trp, $\Delta G_{t,b}^{\circ }\approx \Delta G_{i,b}^{\circ }$, indicating that the translocon energetically mimics the interface $\left(\Delta G_{t,i}^{\circ }\approx 0\right)$. The deviation for Trp is hypothesized to be due to the energetic preference of Trp to exist in the bilayer interface ^,. The difference in $\Delta G_{t,b}^{\circ }$ and $\Delta G_{i,b}^{\circ }$ for Trp (and Tyr to a lesser degree) indicates that the translocon does not mimic all the chemical properties of the phospholipid-water interface.