Prediction of membrane protein structures with complex topologies using limited constraints

Area of Science:

• Structural biology
• Computational biology
• Biophysics

Background:

• Experimental determination of high-resolution membrane protein structures is difficult.
• Membrane proteins, often exceeding 200 amino acids, pose significant challenges for computational structure prediction.
• Accurate structural models are crucial for understanding membrane protein function.

Purpose of the Study:

• To develop and test a novel method for predicting the structures of large membrane proteins.
• To improve the accuracy of membrane protein structure prediction by incorporating constraints.
• To assess the feasibility of achieving near-native models for complex membrane proteins.

Main Methods:

• Developed a structure prediction method incorporating constraints on helix-helix packing.
• Applied sequence-predicted or experimentally identified constraints during folding simulations.
• Tested the method on 12 diverse membrane proteins (190-300 residues).
• Evaluated near-native model selection using metrics like C-alpha root-mean-square deviation (RMSD).

Main Results:

• Enrichment of near-native models for 9 out of 12 tested membrane proteins with a single constraint.
• Achieved models superimposable within 4 Å of the native structure for 4 proteins using sequence-predicted constraints.
• Successfully selected near-native structures for specific domains (heme-binding, pore-forming) using experimental or conserved residue information.

Conclusions:

• Constrained simulations significantly improve the prediction of large membrane protein structures.
• Limited information on residue-residue interactions can lead to highly accurate models (within 4 Å) for complex membrane proteins.
• This approach offers a viable strategy for advancing membrane protein structure prediction.