A follow up to my previous post about preprints from my disseration work: the final chapter of my dissertation was split up into two papers focusing on two different aspects of the project. Both of these are now available in final peer-reviewed form. Citations and links below.
Here, we used a combination of high-throughput phage display binding screens and ASR to quantitatively examine shifts in the intrinsic binding specificity of S100 proteins for their peptide targets. We found that this historical pattern does not conform the to hypothesis that specificity of proteins generally increases over the course of evolution, instead supporting a more nuanced set of shifts that differ between protein lineages.
Wheeler LC, Perkins A, Wong CE, Harms MJ (2020). Learning Peptide Recognition Rules for a Low-Specificity Protein. Protein Science; doi.org/10.1002/pro.3958. (A version of this article is also available at bioRxiv; doi: 10.1101/2020.06.02.131086v1.)
Here, we used a machine-learning approach, in combination with docking simulations and x-ray crystallography to identify the key biochemical properties that predict peptide binding by the protein human S100A5. S100s are notorious for having “low specificity” that is essentially impossible to summarize using traditional motif-based concepts. We find that a set of biochemical features of peptides, such as hydrophobicity, shape complementarity, and secondary structure propensity can be used to accurately predict phage-display enrichment. These results are further explained by solving a crystal structure of S100A5 and modeling peptide binding to the protein surface.