Toward high-resolution prediction and design of transmembrane helical protein structures

Abstract

The prediction and design at the atomic level of membrane protein structures and interactions is a critical but unsolved challenge. To address this problem, we have developed an all-atom physical model that describes intraprotein and protein-solvent interactions in the membrane environment. We evaluated the ability of the model to recapitulate the energetics and structural specificities of polytopic membrane proteins by using a battery of in silico prediction and design tests. First, in side-chain packing and design tests, the model successfully predicts the side-chain conformations at 73% of nonexposed positions and the native amino acid identities at 34% of positions in naturally occurring membrane proteins. Second, the model predicts significant energy gaps between native and nonnative structures of transmembrane helical interfaces and polytopic membrane proteins. Third, distortions in transmembrane helices are successfully recapitulated in docking experiments by using fragments of ideal helices judiciously defined around helical kinks. Finally, de novo structure prediction reaches near-atomic accuracy (<2.5 A) for several small membrane protein domains (<150 residues). The success of the model highlights the critical role of van der Waals and hydrogen-bonding interactions in the stability and structural specificity of membrane protein structures and sets the stage for the high-resolution prediction and design of complex membrane protein architectures.

Fig. 1.

Modeling of TM distorted helices with ideal helix fragments defined around hinges predicted from sequence. Single distorted helices were cut away from membrane protein structures, and the cut-away region was docked back onto the remainder of the structures by using flexible side-chain rigid backbone docking with one of several representations for the backbone. In the leftmost plots, a single ideal helix was docked; in the two center plots (N term and C term), the regions N- and C-terminal to the kink were represented by short ideal helices and docked independently; the rightmost plots (combined) show compatible N- and C-terminal ideal helix pairs with distances and angles within the ranges predicted based on the sequence of the kink (Materials and Methods and SI Fig. 7). Each plot shows the energy (y axis) versus rmsd to the native structure (x axis) for structures generated in independent Monte Carlo docking calculations; the lowest-energy structure generated is indicated by a black box. Also shown are superpositions between the lowest-energy structure (magenta) from the “combined” plots with the native structure (blue). (Top) Bovine rhodopsin TMH kinked at Pro-53 with a kink angle of 17.1°. (Middle) Halorhodopsin kinked at Pro-94 with a kink angle of 30.4°. (Bottom) Halorhodopsin with a Pro-like kink at Thr-92 and a kink angle of 38.9°.

Fig. 2.

Structure prediction. (A) fd-coat protein. (Left) Backbone superposition of the experimental structure determined by solid-state NMR (19) (blue) and the lowest-energy decoy generated by ROSETTA (pink) starting from an extended chain. The rmsd over 30 C^α atoms is 2.4 Å. (Right) All-atom representation of the lowest-energy decoy generated by ROSETTA. The boundaries predicted by ROSETTA between the hydrophobic core and the interface regions of the membrane are represented with a black solid line. (B) Glycophorin A. Isolated monomers were docked with the ROSETTA protein–protein docking protocol and the all-atom membrane force field. The superposition between the native (blue) and the lowest-energy predicted structure (pink) is represented. The rmsd over 45 C^α atoms is 0.65 Å. (C and D) Ab initio structure prediction of polytopic membrane proteins. Native polytopic membrane protein conformations define a narrow energy basin in the all-atom conformational energy landscape. When near-native topologies are generated at the coarse-grained level, all-atom relaxed decoys define a funnel toward the native basin and the lowest-energy predicted structures have near-atomic resolution structures. In the energy versus rmsd plots, nonnative (red points) (generated from sequence by the ROSETTA coarse-grained structure prediction mode) and native conformations (green points) were relaxed by sampling the conformational degrees of freedom of all backbone and side-chain atoms. Cartoons show superposition between the native (blue) and the lowest-energy predicted structure (pink). Boxed areas show regions where close to native side-chain packing arrangements were obtained. The rmsd values are 2.1 Å over 111 C^α atoms for BRD4 (C) and 2.4 Å over 139 C^α atoms for VATP (D).