Development and assessment of methods for membrane protein structure prediction

National Institute Of Neurological Disorders And Stroke

Investigators

Linked publications, trials & patents

Abstract

The ability to reliably align the amino acid sequences of proteins is a key requirement for confirming evolutionary relationships, identifying conserved elements, and enabling homology-based structural modeling. However, the more dissimilar the sequences, the more challenging it is to compute their alignments accurately. This can be particularly challenging for membrane proteins. This year, we continued to develop our software for alignment of membrane protein sequences with an eye to detecting template structures and constructing accurate homology models, even when the sequence relationships are faint. In collaboration with Dr. Staritzbichler (Leipzig), we implemented a method to allow experimental information to be incorporated into the alignments, using an approach called anchoring (Ref 1). This method was tested for proteins with different arrangements of transmembrane segments, such as the proteins BetP and LeuT, which share a common core architecture, but have different embellishments. Even state-of-the-art methods fail to align these sequences accurately. However, with a small amount of biochemical information - in this case, knowledge of the sodium-ion binding sites in both proteins - we showed that it is possible to anchor two of the transmembrane segments and thereby dramatically improve the overall alignment. The AlignMe server at http://www.bioinfo.mpg.de/AlignMe was updated accordingly and the code and tests were made available through GitHub and Zenodo.