How physical forces drive the process of helical membrane protein folding

Abstract

Protein folding is a fundamental process of life with important implications throughout biology. Indeed, tens of thousands of mutations have been associated with diseases, and most of these mutations are believed to affect protein folding rather than function. Correct folding is also a key element of design. These factors have motivated decades of research on protein folding. Unfortunately, knowledge of membrane protein folding lags that of soluble proteins. This gap is partly caused by the greater technical challenges associated with membrane protein studies, but also because of additional complexities. While soluble proteins fold in a homogenous water environment, membrane proteins fold in a setting that ranges from bulk water to highly charged to apolar. Thus, the forces that drive folding vary in different regions of the protein, and this complexity needs to be incorporated into our understanding of the folding process. Here, we review our understanding of membrane protein folding biophysics. Despite the greater challenge, better model systems and new experimental techniques are starting to unravel the forces and pathways in membrane protein folding.

Keywords: co-translational folding; energetics; folding pathways; helix insertion; membrane protein topology.

Figure 1. The original and updated models of membrane protein folding

(A) The original two‐stage model of membrane protein folding. In the first step, the transmembrane helices, facilitated by cellular machinery, are inserted into the membrane bilayer. In the second step, the helices assemble into their final fold. (B) A co‐translational model of membrane protein folding. Initial transmembrane helix insertion is governed by a helices’ thermodynamic preference for insertion, which is defined by its amino acid composition and can be quantified by biological hydrophobicity scales. Subsequent helix insertion is influenced by the orientation of, and interactions with, previously inserted helices, as well as interactions with other proteins in the translocon complex. As helices emerge from the translocon, folding can begin and helical hairpin folding units may be preferred. In some cases, proteins may adopt temporary folds that must later be corrected through topological adjustments. The final fold of the protein is defined by interactions within the protein, the general electromechanical properties of the lipid bilayer, and interactions between the protein and lipids. The white arrows show the orientation of the helices and illustrate how they can flip to achieve their final conformation. The gray arrows represent the pressure on the protein within the bilayer.

Figure 2. Transmembrane helix amino acid composition and free energies of insertion into the membrane bilayer

The propensity of each amino acid to be located at a specific depth through the bilayer (orange) is overlayed with the free energy of insertion ΔG for each amino acid as a function of distance from the center of a 19‐residue transmembrane helix as defined by the biological hydrophobicity scales (blue). The biological hydrophobicity scale of each amino acid correlates well with its observed distribution. The data for each plot was obtained from (Hessa et al, ; Schramm et al, 2012a), and the bilayer depth and test‐residue length were normalized to their respective maximum lengths.

Figure 3. Some key contributing forces in membrane protein folding

(A) Hydrogen bonds. Interhelical side chain hydrogen bonds must often compete with hydrogen bonds to water and hydrogen bonds of the side chain back to its own helix. As a result, competition exists between inter‐ and intra‐helix hydrogen bonds, as well as with water. (B) Van der Waals forces and packing. These interactions play a crucial role in the folding of the transmembrane regions. The plot shows the fraction of side chain area buried as a function of protein size for different protein structures. The transmembrane regions bury more surface area than soluble proteins, while the extramembrane regions bury less area than soluble proteins. This finding may explain the lower mutation rates in transmembrane regions compared to soluble proteins, and the higher mutation rates in extramembrane regions compared to soluble proteins. Adapted from Oberai et al (2009). (C) Protein entropy. A new class of fast motions dubbed J’ is observed in the helical membrane protein pSRII, and the rigid ω class is nearly absent. The older class of fast motions J is more prevalent in the membrane protein compared to the soluble protein. The ω‐class represents motions around a single rotomeric conformation. The α‐class represents larger motions around a single conformation, as well as occasional rotomeric interconversions. The J and J’ classes represent progressively increasing motion and sampling between different rotomeric states. Reprinted from O’Brien et al (2020). (D) Solvent entropy. Solvation entropy likely contributes toward membrane protein folding and stability. In the disassociated/unbound state, the lipids immediately surrounding transmembrane helices have reduced entropy. As the helices or domains associate, some of these lipids (light grey) are released back to the bulk bilayer, thereby increasing the entropy of the system (Helms, 2002). (E) Lipid bilayer forces. The lipid bilayer is composed of many kinds of lipids, including cylindrically shaped bilayer‐forming lipids and cone‐shaped non‐bilayer lipids. The latter typically increase the lateral pressure in the hydrocarbon chain region of the bilayer by causing overpacking, while the interfacial region becomes underpacked.

Figure 4. Examples of post‐translational topological changes

For Aquaporin, several transmembrane helices are initially placed outside of the membrane and must later be integrated. For Lac Permease, helices 1 through 6 invert when the concentration of PE in the membrane is too low. The topology is mostly corrected upon addition of PE. EmrE, a dual topology protein, may initially be inserted in an ensemble of configurations. To achieve the final fold, individual loops likely flip in the membrane, with the rate of achieving the final topology limited by the slowest‐flipping loop (shown in red).

Figure 5. The magnetic tweezer technique for forced unfolding of membrane proteins yields new insights into membrane protein folding

(A) The magnetic tweezer setup. The target protein is inserted into a bicelle or liposome composed of detergents or lipids such as CHAPSO and DMPC (blue) and attached to DNA handles (gray), which are then bound to a surface and a magnetic bead. By applying a controlled magnetic force to the bead, the protein can be unfolded by the transmitted force. (B) Folding pathway of GlpG. At high forces (> 20 pN), the protein is extracted from the bilayer in an extended worm‐like chain conformation. At moderate forces (~3–8 pN), the protein remains ensconced in the bilayer in an apparent zig‐zag state. This zig‐zag state appears to consist of inserted and weakly interacting helices, corresponding quite well to the unfolded state envisioned in the second stage of the two‐stage folding model. From the zig‐zag state, folding occurs from the N terminus in units of helical hairpins. (Choi et al, 2019). (C) Topology and unfolding of the ClC chloride channel. The topology of the ClC transporter is shown in the top panel. Many of the transmembrane helices only insert part way before looping back out. The interface between the N and C domains contains many polar residues. The bottom panel shows mechanical force applied to the ClC chloride channel. Under mechanical force, the N and C domains can separate but remain folded, exposing polar residues to the bilayer. How this polar exposure is tolerated remains an open question.

Figure 6. Assembly of membrane protein oligomers

(A) The interface between monomeric subunits of an oligomer may determine how cooperative the assembly process is. The oligomerization pathway likely attempts to bury as much surface area as quickly as possible, and the final fold is probably only achieved upon complete oligomerization. (B) Membrane protein complexes like the multimeric cytochrome bo₃ complex typically assemble in an ordered manner. For the cytochrome bo₃ complex, 10 assembly intermediates are theoretically possible, but only 2 are observed. In Step 2, only an intermediate with subunits III‐IV is observed, but complexes with I‐II, I‐III, I‐IV, II‐III, and II‐IV are theoretically possible. In Step 3, only an intermediate with subunits I‐III‐IV is observed, but complexes with I‐II‐III, I‐II‐IV, and II‐III‐IV are theoretically possible. These results suggest that multimeric protein assembly is an ordered process(Daley, 2008).