Illuminating protein space with a programmable generative model.

Three billion years of evolution has produced a tremendous diversity of protein molecules1, but the full potential of proteins is likely to be much greater. Accessing this potential has been challenging for both computation and experiments because the space of possible protein molecules is much larger than the space of those likely to have functions. Here we introduce Chroma, a generative model for proteins and protein complexes that can directly sample novel protein structures and sequences, and that can be conditioned to steer the generative process towards desired properties and functions. To enable this, we introduce a diffusion process that respects the conformational statistics of polymer ensembles, an efficient neural architecture for molecular systems that enables long-range reasoning with sub-quadratic scaling, layers for efficiently synthesizing three-dimensional structures of proteins from predicted inter-residue geometries and a general low-temperature sampling algorithm for diffusion models. Chroma achieves protein design as Bayesian inference under external constraints, which can involve symmetries, substructure, shape, semantics and even natural-language prompts. The experimental characterization of 310 proteins shows that sampling from Chroma results in proteins that are highly expressed, fold and have favourable biophysical properties. The crystal structures of two designed proteins exhibit atomistic agreement with Chroma samples (a backbone root-mean-square deviation of around 1.0 Å). With this unified approach to protein design, we hope to accelerate the programming of protein matter to benefit human health, materials science and synthetic biology.

A Screen for Protein-Protein Interactions in Live Mycobacteria Reveals a Functional Link between the Virulence-Associated Lipid Transporter LprG and the Mycolyltransferase Antigen 85A.

Outer membrane lipids in pathogenic mycobacteria are important for virulence and survival. Although the biosynthesis of these lipids has been extensively studied, mechanisms responsible for their assembly in the outer membrane are not understood. In the study of Gram-negative outer membrane assembly, protein-protein interactions define transport mechanisms, but analogous interactions have not been explored in mycobacteria. Here we identified interactions with the lipid transport protein LprG. Using site-specific photo-cross-linking in live mycobacteria, we mapped three major interaction interfaces within LprG. We went on to identify proteins that cross-link at the entrance to the lipid binding pocket, an area likely relevant to LprG transport function. We verified LprG site-specific interactions with two hits, the conserved lipoproteins LppK and LppI. We further showed that LprG interacts physically and functionally with the mycolyltransferase Ag85A, as loss of either protein leads to similar defects in cell growth and mycolylation. Overall, our results support a model in which protein interactions coordinate multiple pathways in outer membrane biogenesis and connect lipid biosynthesis to transport.

Characterization of Engineered PreQ1 Riboswitches for Inducible Gene Regulation in Mycobacteria.

We report here the behavior of naturally occurring and rationally engineered preQ1 riboswitches and their application to inducible gene regulation in mycobacteria. Because mycobacteria lack preQ1 biosynthetic genes, we hypothesized that preQ1 could be used as an exogenous nonmetabolite ligand to control riboswitches in mycobacteria. Selected naturally occurring preQ1 riboswitches were assayed and successfully drove preQ1-dependent repression of a green fluorescent protein reporter in Mycobacterium smegmatis Using structure-based design, we engineered three preQ1 riboswitches from Thermoanaerobacter tencongensis, Bacillus subtilis, and Lactobacillus rhamnosus toward achieving higher response ratios and increased repression. Assuming a steady-state model, variants of the T. tencongensis riboswitch most closely followed the predicted trends. Unexpectedly, the preQ1 dose response was best described by a model with a second, independent preQ1 binding site. This behavior was general to the preQ1 riboswitch family, since the wild type and rationally designed mutants of riboswitches from all three bacteria behaved analogously. Across all variants, the response ratios, which describe expression in the absence versus the presence of preQ1, ranged from <2 to ∼10, but repression in all cases was incomplete up to 1 mM preQ1. By reducing the transcript expression level, we obtained a preQ1 riboswitch variant appropriate for inducible knockdown applications. We further showed that the preQ1 response is reversible, is titratable, and can be used to control protein expression in mycobacteria within infected macrophages. By engineering naturally occurring preQ1 riboswitches, we have not only extended the tools available for inducible gene regulation in mycobacteria but also uncovered new behavior of these riboswitches.IMPORTANCE Riboswitches are elements found in noncoding regions of mRNA that regulate gene expression, typically in response to an endogenous metabolite. Riboswitches have emerged as important tools for inducible gene expression in diverse organisms. We noted that mycobacteria lack the biosynthesis genes for preQ1, a ligand for riboswitches from diverse bacteria. Predicting that preQ1 is not present in mycobacteria, we showed that it controls optimized riboswitches appropriate for gene knockdown applications. Further, the riboswitch response is subject to a second independent preQ1 binding event that has not been previously documented. By engineering naturally occurring riboswitches, we have uncovered a new behavior, with implications for riboswitch function in its native context, and extended the tools available for inducible gene regulation in mycobacteria.

Using riboswitches to regulate gene expression and define gene function in mycobacteria.

Mycobacteria include both environmental species and many pathogenic species such as Mycobacterium tuberculosis, an intracellular pathogen that is the causative agent of tuberculosis in humans. Inducible gene expression is a powerful tool for examining gene function and essentiality, both in in vitro culture and in host cell infections. The theophylline-inducible artificial riboswitch has recently emerged as an alternative to protein repressor-based systems. The riboswitch is translationally regulated and is combined with a mycobacterial promoter that provides transcriptional control. We here provide methods used by our laboratory to characterize the riboswitch response to theophylline in reporter strains, recombinant organisms containing riboswitch-regulated endogenous genes, and in host cell infections. These protocols should facilitate the application of both existing and novel artificial riboswitches to the exploration of gene function in mycobacteria.