Predicting direct physical interactions in multimeric proteins with deep learning

Accurate descriptions of protein-protein interactions are essential for understanding biological systems. Very recently, AlphaFold2 has been shown to be remarkably accurate for predicting the atomic structures of individual proteins. Here, we demonstrate that the same neural network models developed for AlphaFold2 can be adapted to predict the structures of multimeric protein complexes without retraining. In contrast to common approaches that require paired multiple sequence alignments, our method, AF2Complex, works without using such paired alignments. It achieves higher accuracy than complex strategies that combine AlphaFold2 and protein-protein docking. New metrics are then introduced for predicting direct protein-protein interactions between arbitrary protein pairs. The approach is successfully validated on some challenging CASP14 multimeric targets, a small but appropriate benchmark set, and the E. coli proteome. Lastly, using the cytochrome c biogenesis system as an example, we present high-confidence models of three sought-after assemblies formed by eight members of this system.

In vitro reconstitution reveals major differences between human and bacterial cytochrome c synthases

Cytochromes c are ubiquitous heme proteins in mitochondria and bacteria, all possessing a CXXCH (CysXxxXxxCysHis) motif with covalently attached heme. We describe the first in vitro reconstitution of cytochrome c biogenesis using purified mitochondrial (HCCS) and bacterial (CcsBA) cytochrome c synthases. We employ apocytochrome c and peptide analogs containing CXXCH as substrates, examining recognition determinants, thioether attachment, and subsequent release and folding of cytochrome c. Peptide analogs reveal very different recognition requirements between HCCS and CcsBA. For HCCS, a minimal 16-mer peptide is required, comprised of CXXCH and adjacent alpha helix 1, yet neither thiol is critical for recognition. For bacterial CcsBA, both thiols and histidine are required, but not alpha helix 1. Heme attached peptide analogs are not released from the HCCS active site; thus, folding is important in the release mechanism. Peptide analogs behave as inhibitors of cytochrome c biogenesis, paving the way for targeted control.

Modified base-binding EVE and DCD domains: striking diversity of genomic contexts in prokaryotes and predicted involvement in a variety of cellular processes

Background DNA and RNA of all cellular life forms and many viruses contain an expansive repertoire of modified bases. The modified bases play diverse biological roles that include both regulation of transcription and translation, and protection against restriction endonucleases and antibiotics. Modified bases are often recognized by dedicated protein domains. However, the elaborate networks of interactions and processes mediated by modified bases are far from being completely understood. Results We present a comprehensive census and classification of EVE domains that belong to the PUA/ASCH domain superfamily and bind various modified bases in DNA and RNA. We employ the “guilt by association” approach to make functional inferences from comparative analysis of bacterial and archaeal genomes, based on the distribution and associations of EVE domains in (predicted) operons and functional networks of genes. Prokaryotes encode two classes of EVE domain proteins, slow-evolving and fast-evolving ones. Slow-evolving EVE domains in α-proteobacteria are embedded in conserved operons, potentially involved in coupling between translation and respiration, cytochrome c biogenesis in particular, via binding 5-methylcytosine in tRNAs. In β- and γ-proteobacteria, the conserved associations implicate the EVE domains in the coordination of cell division, biofilm formation, and global transcriptional regulation by non-coding 6S small RNAs, which are potentially modified and bound by the EVE domains. In eukaryotes, the EVE domain-containing THYN1-like proteins have been reported to inhibit PCD and regulate the cell cycle, potentially, via binding 5-methylcytosine and its derivatives in DNA and/or RNA. We hypothesize that the link between PCD and cytochrome c was inherited from the α-proteobacterial and proto-mitochondrial endosymbiont and, unexpectedly, could involve modified base recognition by EVE domains. Fast-evolving EVE domains are typically embedded in defense contexts, including toxin-antitoxin modules and type IV restriction systems, suggesting roles in the recognition of modified bases in invading DNA molecules and targeting them for restriction. We additionally identified EVE-like prokaryotic Development and Cell Death (DCD) domains that are also implicated in defense functions including PCD. This function was inherited by eukaryotes, but in animals, the DCD proteins apparently were displaced by the extended Tudor family proteins, whose partnership with Piwi-related Argonautes became the centerpiece of the Piwi-interacting RNA (piRNA) system. Conclusions Recognition of modified bases in DNA and RNA by EVE-like domains appears to be an important, but until now, under-appreciated, common denominator in a variety of processes including PCD, cell cycle control, antivirus immunity, stress response, and germline development in animals.

Deciphering the Role of Multiple Thioredoxin Fold Proteins of Leptospirillum sp. in Oxidative Stress Tolerance

Thioredoxin fold proteins (TFPs) form a family of diverse proteins involved in thiol/disulfide exchange in cells from all domains of life. Leptospirillum spp. are bioleaching bacteria naturally exposed to extreme conditions like acidic pH and high concentrations of metals that can contribute to the generation of reactive oxygen species (ROS) and consequently the induction of thiol oxidative damage. Bioinformatic studies have predicted 13 genes that encode for TFP proteins in Leptospirillum spp. We analyzed the participation of individual tfp genes from Leptospirillum sp. CF-1 in the response to oxidative conditions. Genomic context analysis predicted the involvement of these genes in the general thiol-reducing system, cofactor biosynthesis, carbon fixation, cytochrome c biogenesis, signal transduction, and pilus and fimbria assembly. All tfp genes identified were transcriptionally active, although they responded differentially to ferric sulfate and diamide stress. Some of these genes confer oxidative protection to a thioredoxin-deficient Escherichia coli strain by restoring the wild-type phenotype under oxidative stress conditions. These findings contribute to our understanding of the diversity and complexity of thiol/disulfide systems, and of adaptations that emerge in acidophilic microorganisms that allow them to thrive in highly oxidative environments. These findings also give new insights into the physiology of these microorganisms during industrial bioleaching operations.