A large number of computational methods is available for the superposition and comparison of protein structures. However, it is not always clear which method is the best for a particular type of study. Several studies have compared structure alignment methods using data sets of water-soluble proteins. Membrane proteins, however, have unique structural characteristics, such as close helix packing and high propensity for secondary structure. We therefore tested a large array of structure alignment methods for their ability to accurately identify the correct evolutionary relationships between two membrane protein structures. This information is necessary for construction of high-quality reference datasets for development of membrane protein sequence alignment methodologies, such as our software and web server, AlignMe. Our analysis suggested that a number of structure alignment methods perform similarly well on average, but that no single method performs well for all pairs of membrane protein structures (ref. 1). We therefore developed a consensus-type approach that takes advantage of the strengths of the best four methods (FR-TMalign, DaliLite, MATT and FATCAT) to identify accurately aligned positions for a given pair of membrane protein structures (ref. 1). I also carried out a survey of known membrane protein structures, to identify symmetries within their architectures (ref. 2). My results suggested that membrane proteins apparently have a higher tendency to oligomerize than water-soluble proteins, as well as a higher proportion of internal repeats, leading to an apparently overall greater propensity to exhibit symmetry. We found that almost all possible rotational (cyclic) symmetries or pseudo-symmetries are observed, including screw axis symmetries, and symmetries in which the axis is parallel to the membrane plane, leading to repeated elements with inverted transmembrane topologies. Such inverted-topology repeat proteins are exclusively found in channel and transport proteins, and have been shown to be important both for creating pathways through the membrane, and for allowing the protein to adopt alternate conformations required for active transport (ref. 2). We previously developed a structure prediction methodology in which the structure of each inverted-topology repeat of a membrane transport protein is modeled using its corresponding repeat as a template. For protein structures in which the two repeats are distinctly asymmetric, this so-called repeat-swap modeling, results in a structural model of an alternate conformation for the same protein. Recently, we developed an improved, semi-automated protocol for this procedure, in which restraints are applied selectively in order to increase the accuracy of the model (ref. 3). We demonstrated the accuracy of this updated method for predicting the conformation of a glutamate transporter homolog called GltPh.
| Forrest, Lucy R (2015) Structural Symmetry in Membrane Proteins. Annu Rev Biophys 44:311-37 |
| Stamm, Marcus; Forrest, Lucy R (2015) Structure alignment of membrane proteins: Accuracy of available tools and a consensus strategy. Proteins 83:1720-32 |
| Stamm, Marcus; Staritzbichler, René; Khafizov, Kamil et al. (2014) AlignMe--a membrane protein sequence alignment web server. Nucleic Acids Res 42:W246-51 |
