Protein sequence design with a learned potential.

The task of protein sequence design is central to nearly all rational protein engineering problems, and enormous effort has gone into the development of energy functions to guide design. Here, we investigate the capability of a deep neural network model to automate design of sequences onto protein backbones, having learned directly from crystal structure data and without any human-specified priors. The model generalizes to native topologies not seen during training, producing experimentally stable designs. We evaluate the generalizability of our method to a de novo TIM-barrel scaffold. The model produces novel sequences, and high-resolution crystal structures of two designs show excellent agreement with in silico models. Our findings demonstrate the tractability of an entirely learned method for protein sequence design.

Computationally Modeling Electrostatic Binding Energetics in a Crowded, Dynamic Environment: Physical Insights from a Peptide-DNA System.

The cell is a crowded place, and it may be crucial at times to account for the local environment when studying determinants of molecular recognition. In this work, we use continuum electrostatics calculations on snapshots extracted from molecular dynamics simulations to understand how various aspects of a crowded environment affect electrostatic binding energies between the antimicrobial peptide buforin II and DNA. By comparing multiple models for representing crowding, sequentially introducing layers of model complexity for maximum control, we explore how electrostatic binding energetics depend on crowder physical properties, the sampling of the binding partners and crowder molecules, and the treatment of bulk solvent. We show that physical characteristics can combine to create an interplay of competing effects in this highly charged system. For example, increased ionic strength screening due to crowding partially cancels out the reduced solvent screening due to water depletion. We also quantify the effect of crowders' charge distributions on binding energetics. While we focus on electrostatic effects of crowding on binding, we begin to consider nonpolar components as well, and we implement a thermodynamic cycle accounting for both bound and unbound states to show the necessity of adequate crowder sampling in future studies. The insights developed here provide a rich starting point for experiments to further explore these competing effects and, ultimately, to rationally modulate molecular recognition in the complex cellular environment.