Machine learning modeling of family wide enzyme-substrate specificity screens

Biocatalysis is a promising approach to sustainably synthesize pharmaceuticals, complex natural products, and commodity chemicals at scale. However, the adoption of biocatalysis is limited by our ability to select enzymes that will catalyze their natural chemical transformation on non-natural substrates. While machine learning and in silico directed evolution are well-posed for this predictive modeling challenge, efforts to date have primarily aimed to increase activity against a single known substrate, rather than to identify enzymes capable of acting on new substrates of interest. To address this need, we curate 6 different high-quality enzyme family screens from the literature that each measure multiple enzymes against multiple substrates. We compare machine learning-based compound-protein interaction (CPI) modeling approaches from the literature used for predicting drug-target interactions. Surprisingly, comparing these interaction-based models against collections of independent (single task) enzyme-only or substrate-only models reveals that current CPI approaches are incapable of learning interactions between compounds and proteins in the current family level data regime. We further validate this observation by demonstrating that our no-interaction baseline can outperform CPI-based models from the literature used to guide the discovery of kinase inhibitors. Given the high performance of non-interaction based models, we introduce a new structure-based strategy for pooling residue representations across a protein sequence. Altogether, this work motivates a principled path forward in order to build and evaluate meaningful predictive models for biocatalysis and other drug discovery applications.     

17O]oxygen hyperfine structure for the hydroxyl and superoxide radical adducts of the spin traps DMPO, PBN and 4-POBN

[17O]oxygen hyperfine coupling constants are reported for the superoxide and hydroxyl radical adducts with the spin traps 5,5-dimethyl-1-pyrroline N-oxide, N-t-butyl-alpha-phenylnitrone and alpha-(4-pyridyl 1-oxide)-N-t-butylnitrone. These couplings provide spectroscopic evidence that the spin adducts have been correctly identified.     

A cellular trafficking signal in the SIV envelope protein cytoplasmic domain is strongly selected for in pathogenic infection

The HIV/SIV envelope glycoprotein (Env) cytoplasmic domain contains a highly conserved Tyr-based trafficking signal that mediates both clathrin-dependent endocytosis and polarized sorting. Despite extensive analysis, the role of these functions in viral infection and pathogenesis is unclear. An SIV molecular clone (SIVmac239) in which this signal is inactivated by deletion of Gly-720 and Tyr-721 (SIVmac239ΔGY), replicates acutely to high levels in pigtail macaques (PTM) but is rapidly controlled. However, we previously reported that rhesus macaques and PTM can progress to AIDS following SIVmac239ΔGY infection in association with novel amino acid changes in the Env cytoplasmic domain. These included an R722G flanking the ΔGY deletion and a nine nucleotide deletion encoding amino acids 734–736 (ΔQTH) that overlaps the rev and tat open reading frames. We show that molecular clones containing these mutations reconstitute signals for both endocytosis and polarized sorting. In one PTM, a novel genotype was selected that generated a new signal for polarized sorting but not endocytosis. This genotype, together with the ΔGY mutation, was conserved in association with high viral loads for several months when introduced into naïve PTMs. For the first time, our findings reveal strong selection pressure for Env endocytosis and particularly for polarized sorting during pathogenic SIV infection in vivo.