DeepBL: a deep learning-based approach for in silico discovery of beta-lactamases.

Beta-lactamases (BLs) are enzymes localized in the periplasmic space of bacterial pathogens, where they confer resistance to beta-lactam antibiotics. Experimental identification of BLs is costly yet crucial to understand beta-lactam resistance mechanisms. To address this issue, we present DeepBL, a deep learning-based approach by incorporating sequence-derived features to enable high-throughput prediction of BLs. Specifically, DeepBL is implemented based on the Small VGGNet architecture and the TensorFlow deep learning library. Furthermore, the performance of DeepBL models is investigated in relation to the sequence redundancy level and negative sample selection in the benchmark dataset. The models are trained on datasets of varying sequence redundancy thresholds, and the model performance is evaluated by extensive benchmarking tests. Using the optimized DeepBL model, we perform proteome-wide screening for all reviewed bacterium protein sequences available from the UniProt database. These results are freely accessible at the DeepBL webserver at http://deepbl.erc.monash.edu.au/.

The evolutionary mechanism of non-carbapenemase carbapenem-resistant phenotypes in Klebsiella spp.

Antibiotic resistance is driven by selection, but the degree to which a bacterial strain's evolutionary history shapes the mechanism and strength of resistance remains an open question. Here, we reconstruct the genetic and evolutionary mechanisms of carbapenem resistance in a clinical isolate of Klebsiella quasipneumoniae. A combination of short- and long-read sequencing, machine learning, and genetic and enzymatic analyses established that this carbapenem-resistant strain carries no carbapenemase-encoding genes. Genetic reconstruction of the resistance phenotype confirmed that two distinct genetic loci are necessary in order for the strain to acquire carbapenem resistance. Experimental evolution of the carbapenem-resistant strains in growth conditions without the antibiotic revealed that both loci confer a significant cost and are readily lost by de novo mutations resulting in the rapid evolution of a carbapenem-sensitive phenotype. To explain how carbapenem resistance evolves via multiple, low-fitness single-locus intermediates, we hypothesised that one of these loci had previously conferred adaptation to another antibiotic. Fitness assays in a range of drug concentrations show how selection in the antibiotic ceftazidime can select for one gene (blaDHA-1) potentiating the evolution of carbapenem resistance by a single mutation in a second gene (ompK36). These results show how a patient's treatment history might shape the evolution of antibiotic resistance and could explain the genetic basis of carbapenem-resistance found in many enteric-pathogens.

Targeting bacterial outer-membrane remodelling to impact antimicrobial drug resistance.

The cell envelope is essential for survival and adaptation of bacteria. Bacterial membrane proteins include the major porins that mediate the influx of nutrients and several classes of antimicrobial drugs. Consequently, membrane remodelling is closely linked to antimicrobial resistance (AMR). Knowledge of bacterial membrane protein biogenesis and turnover underpins our understanding of bacterial membrane remodelling and the consequences that this process have in the evolution of AMR phenotypes. At the population level, the evolution of phenotypes is a reversible process, and we can use these insights to deploy evolutionary principles to resensitize bacteria to existing antimicrobial drugs. In our opinion, fundamental knowledge is opening a new way of thinking towards sustainable solutions to the mounting crisis in AMR. Here we discuss what is known about outer-membrane remodelling in bacteria and how the process could be targeted as a means to restore sensitivity to antimicrobial drugs. Bacteriophages are highlighted as a powerful means to exert this control over membrane remodelling but they require careful selection so as to reverse, and not exacerbate, AMR phenotypes.

Novel VIM metallo-beta-lactamase variant, VIM-24, from a Klebsiella pneumoniae isolate from Colombia.

We report the emergence of a novel VIM variant (VIM-24) in a Klebsiella pneumoniae isolate in Colombia. The isolate displays MICs for carbapenems below the resistance breakpoints, posing a real challenge for its detection. The blaVIM-24 gene was located within a class 1 integron carried on a large plasmid. Further studies are needed to clarify its epidemiological and clinical impact.