Accurate de novo structure prediction of large transmembrane protein domains using fragment-assembly and correlated mutation analysis

Abstract

A new de novo protein structure prediction method for transmembrane proteins (FILM3) is described that is able to accurately predict the structures of large membrane proteins domains using an ensemble of two secondary structure prediction methods to guide fragment selection in combination with a scoring function based solely on correlated mutations detected in multiple sequence alignments. This approach has been validated by generating models for 28 membrane proteins with a diverse range of complex topologies and an average length of over 300 residues with results showing that TM-scores > 0.5 can be achieved in almost every case following refinement using MODELLER. In one of the most impressive results, a model of mitochondrial cytochrome c oxidase polypeptide I was obtained with a TM-score > 0.75 and an rmsd of only 5.7 Å over all 514 residues. These results suggest that FILM3 could be applicable to a wide range of transmembrane proteins of as-yet-unknown 3D structure given sufficient homologous sequences.

Fig. 1.

Prediction of TMP structures. Superposition between native (red) and models (green) of (A) Rhodopsin (1gzmA), (B) Cytochrome c oxidase (2occA), (C) Ammonium transporter Amt-1(2b2fA), (D) Aquaporin-4 (2d57A), (E) Ammonia channel (1xqfA), (F) MDR efflux pump (3mktA). The two black lines indicate the approximate position of the membrane.

Fig. 2.

Re-entrant helices in Aquaporin-4 (2d57A, Left) and Glycerol uptake facilitator (1ldiA, Right). Superposition between native (red) and models (green).

Fig. 3.

Observed TM-score of the final refined model plotted against mean pairwise TM-score of all 28 target structures and an additional 4 structures where poor quality contact predictions were used. A mean pairwise TM-score > 0.32 is likely to yield a final model with TM-score > 0.5.

Fig. P1.

Top: A model of the water-conducting channel Aquaporin Z (green) superposed with the native crystal structure (red) embedded within the plasma membrane. Bottom: Contact map for Aquaporin Z representing the distance between all possible residue pairs in the 3D protein structure using a 2D matrix. For two residues i and j, element ij of the matrix is filled if the two residues are closer than 8 Å. Contacts predicted using PSICOV are shown in green, whereas those derived from the native crystal structure are shown in red. Folding of the model is guided solely by the predicted contacts.