Mapping the determinants of substrate selectivity in a bacterial metal transporter with model-guided deep mutagenesis.

Abstract

Membrane transporter proteins control what molecules are able to enter and exit the cell based on their ability to discriminate between substrates. The model bacterial metal transporter Deinococcus radiodurans (Dra)Nramp is responsible for importing rare divalent transition metals such as manganese and iron while excluding common alkaline earth metals such as magnesium and calcium. While recent high-resolution structures of DraNramp have mapped the conformational cycle underlying transport and identified key residues involved in metal binding and conformational gating, exactly which features of DraNramp control selectivity remains unclear, especially given the recent discovery of homologous proteins with altered substrate selectivity. We are simultaneously profiling the effect of thousands of DraNramp variants on import of manganese and magnesium, with the hypothesis that specificity may involve epistatic interactions between key active site positions and residues distal to the active site. To probe beyond single mutants, we have designed a strategy to bias our variant library using evolutionary sequence modeling to test a set of single- to octuple-mutants of DraNramp at positions likely to be involved in controlling substrate selectivity and unlikely to dramatically destabilize the protein. We have also designed novel assays to couple manganese and magnesium import to fluorescence in E. coli, and are testing our biased library in high-throughput with these assays to map the mutational landscape of substrate selectivity in DraNramp. This study will advance our understanding of how proteins bind metals as well as broadly what features of membrane transporter proteins are key to selecting between substrates. The methods for developing and assessing model-guided deep mutagenesis strategies also provide a framework bridging sequencing modeling and experimental biochemistry for future researchers.