Pairwise Evolution in Functionally Related Proteins

This project will use an inter-disciplinary approach to study the co-evolution of a pair of functionally related proteins. Our hypothesis is that residue pairs in functionally co-dependent proteins have a high joint conservation probability across species and have co-evolved. To test this hypothesis, the student will develop a mathematical model to predict pairs of co-evolved residues using the bacterial KefFC system and test this in the laboratory. Little information currently exists concerning the co-evolution of functionally-related protein pairs and this project will reveal novel structure-function relationships between such pairs.

The KefFC systems are found throughout Gram negative species and are an essential element for survival of electrophilic chemical stress. In many bacteria the system comprises two proteins, an integral membrane protein (KefC) and a cytoplasmic ancillary protein (KefF). This system is most understood in Escherichia coli, where we have extensive molecular, biochemical and structural data. We already possess many of the tools (strains, clones) that will be used in the project.

The KefC membrane protein possesses an amino-terminal inner membrane domain which is the site of potassium translocation. The remainder of the protein comprises a soluble C-terminal domain fused to the membrane domain via a Q-linker. The C-terminal domain includes a KTN (potassium transport nucleotide-binding) subdomain and the terminal peripheral domain (PD). The KTN domain is the site of ligand-binding regulation of potassium transport via KefC. The tripeptide glutathione (GSH) is a negative regulator of the system, thus when it is bound no transport occurs. In the presence of specific electrophilic compounds, GSH adducts (GSX) are formed which bind to the KTN domain and activate transport. Potassium transport out of the cell is accompanied by proton influx (via Kef or as yet unidentified system). The resulting internal pH drop is crucial for cell survival of the electrophile. KefF is not required for ligand binding but is essential for full activity. We obtained crystal structures of the KefF-KefCKTN-PD domains which revealed a dimer, thus two ligand-binding sites per system. We have structures in the apo-, GSH-, and GSX-bound conformations providing models of the molecular rearrangements that occur during gating, which we have tested using site-directed mutants and has also allowed us to investigate ligand-binding requirements. Thus we have extensive biochemical, molecular and structural data which make this an ideal system of co-evolved and functionally-related proteins to use for this project.

This project will determine residues of the KefC transmembrane protein and the KefF ancillary protein which have co-evolved by determining the joint probability of change (from one homologue to another) of pairs of residues, using sequence data of Kef homologues from publicly available databases and bioinformatics methods inspired by techniques from statistical physics. We will establish whether residues in the KefC and KefF proteins have evolved by using a set of sequences obtained from a suitably filtered BLAST search, where proteins homologous to KefC and KefF are found. Co-evolution of two residues (one in KefF and one in KefC) will result in a correlation between those two locations in the BLAST sequences corresponding to KefC and KefF. This correlation is manifested as a significant deviation of the residue pair frequencies from those expected in the absence of correlations. We shall quantify this correlation using information-theoretical measures, such as mutual information. We will build on a similar method for single protein correlations we have developed recently, and generalise it for multiple proteins.

We shall also develop Bayesian statistical evolutionary models to test ideas about how these correlations have evolved and how robust they are in the evolutionary time scale, and to tackle issues such as sampling bias, duplications and annotation errors in the BLAST databases. Co-evolved pairs discovered using our method will be analysed further using what is known about the biochemistry, genetics and 3D structure of the proteins, with the goal of understanding the mechanistic reason for the co-evolution. We shall apply complex network theory to tease out global patterns in the co-evolution network. The co-evolution analysis we will develop will be applicable not only to the exemplar system, but to any pair of proteins.

This project would suit biologists interested in collaborating with theoreticians, and theoreticians interested in biology.