Introduction: Acinetobacter baumannii is an opportunistic bacterial pathogen causing numerous deaths each year due to its ability to develop antibiotic resistance. It is urgent today to find new strategies to fight this bacterium.
Methods: The integration of already published data is an important source of information that can reveal the molecules interrelationships and derive clinical knowledge. Here, we re-analysed dozens of proteomic datasets across different strains of A. baumannii and different culture or stress conditions to predict novel co-regulated proteins and PTM cross-talks.
Results: Over a thousand LC-MS2 files were reanalysed to derive protein quantification across different experimental conditions. Using machine learning, we computed a co-regulation score for each protein pair and validated our prediction with a list of known protein interactions. Thousands of novel protein co-regulation links were predicted. For example, the ADC family cephalosporin-hydrolyzing class C beta-lactamase (AmpC), associated with antibiotic resistance, was co-regulated with a conserved exported protein of unknown function.
Several proteomic datasets also investigated the O-phosphorylation, K-succinylation, K-acetylation and K-trimethylation of proteins. For each modified protein, we used the multi-sequence alignment of all its orthologs to compute the co-conservation score of the modified residues. We identified 367 residues that were conserved and modified in A. baumannii. Upon testing for co-evolution, 80 pairs of modified residues were co-evolving thus suggesting a cross-talk. For example, the outer membrane protein Omp38 (OmpA) harbors two putative cross-talks involving succinylated and acetylated lysines.
Conclusions: Currently we focus on the bioinformatic validation of the co-evolving residues by verifying the modification status in other bacteria. This resource of protein co-regulation and putative PTM cross-talks is the first of its kind for A. baumannii.