Computational Modeling of Protein-DNA Complexes for Artificial Regulation of Shiga Toxin Expression

Abstract

Selective DNA-Protein interactions precisely control gene expression.  The CI repressor protein binds phage DNA in E. coli and prevents expression of the Shiga toxin. Disrupting this interaction can lead to life-threatening hemolytic uremic syndrome in humans. CI repressor mimetics should, in principle, prevent Shiga toxin production, even when cellular events disfavor native CI binding. De Novo proteins were previously generated to target the DNA encoding the Shiga toxin.  These proteins avoid the additional biological regulatory structures associated with the CI protein. By generating model structures with AlphaFold and models of DNA-Protein complexes with Haddock and ChimeraX, this study explored the possibility of correlation between binding mode and affinity. Further, we discuss the implications of comparisons between models of these proteins and their dimeric counterparts, both in terms of affinity prediction and consistency of the predicted binding modes.  This work lays a firm foundation for rapid exploration of alternate de novo designs intended to repress the Shiga toxin gene.