Discovery of a structural class of antibiotics with explainable deep learning.

The discovery of novel structural classes of antibiotics is urgently needed to address the ongoing antibiotic resistance crisis1-9. Deep learning approaches have aided in exploring chemical spaces1,10-15; these typically use black box models and do not provide chemical insights. Here we reasoned that the chemical substructures associated with antibiotic activity learned by neural network models can be identified and used to predict structural classes of antibiotics. We tested this hypothesis by developing an explainable, substructure-based approach for the efficient, deep learning-guided exploration of chemical spaces. We determined the antibiotic activities and human cell cytotoxicity profiles of 39,312 compounds and applied ensembles of graph neural networks to predict antibiotic activity and cytotoxicity for 12,076,365 compounds. Using explainable graph algorithms, we identified substructure-based rationales for compounds with high predicted antibiotic activity and low predicted cytotoxicity. We empirically tested 283 compounds and found that compounds exhibiting antibiotic activity against Staphylococcus aureus were enriched in putative structural classes arising from rationales. Of these structural classes of compounds, one is selective against methicillin-resistant S. aureus (MRSA) and vancomycin-resistant enterococci, evades substantial resistance, and reduces bacterial titres in mouse models of MRSA skin and systemic thigh infection. Our approach enables the deep learning-guided discovery of structural classes of antibiotics and demonstrates that machine learning models in drug discovery can be explainable, providing insights into the chemical substructures that underlie selective antibiotic activity.

Draft genomes of three closely related low light-adapted Prochlorococcus.

OBJECTIVES: The marine cyanobacterium Prochlorococcus is a critical part of warm ocean ecosystems and a model for studying microbial evolution and ecology. To expand the representation of this organism's vast wild diversity in sequence collections, we performed a set of isolation efforts targeting low light-adapted Prochlorococcus. Three genomes resulting from this larger body of work are described here. DATA DESCRIPTION: We present draft-quality Prochlorococcus genomes from enrichment cultures P1344, P1361, and P1363, sampled in the North Pacific. The genomes were built from Illumina paired reads assembled de novo. Supporting datasets of raw reads, assessments, and sequences from co-enriched heterotrophic marine bacteria are also provided. These three genomes represent members of the low light-adapted LLIV Prochlorococcus clade that are closely related, with 99.9% average nucleotide identity between pairs, yet vary in gene content. Expanding the powerful toolkit of Prochlorococcus genomes, these sequences provide an opportunity to study fine-scale variation and microevolutionary processes.

An engineered live biotherapeutic for the prevention of antibiotic-induced dysbiosis.

Antibiotic-induced alterations in the gut microbiota are implicated in many metabolic and inflammatory diseases, increase the risk of secondary infections and contribute to the emergence of antimicrobial resistance. Here we report the design and in vivo performance of an engineered strain of Lactococcus lactis that altruistically degrades the widely used broad-spectrum antibiotics ß-lactams (which disrupt commensal bacteria in the gut) through the secretion and extracellular assembly of a heterodimeric ß-lactamase. The engineered ß-lactamase-expression system does not confer ß-lactam resistance to the producer cell, and is encoded via a genetically unlinked two-gene biosynthesis strategy that is not susceptible to dissemination by horizontal gene transfer. In a mouse model of parenteral ampicillin treatment, oral supplementation with the engineered live biotherapeutic minimized gut dysbiosis without affecting the ampicillin concentration in serum, precluded the enrichment of antimicrobial resistance genes in the gut microbiome and prevented the loss of colonization resistance against Clostridioides difficile. Engineered live biotherapeutics that safely degrade antibiotics in the gut may represent a suitable strategy for the prevention of dysbiosis and its associated pathologies.

Engineered bacteria recycle tumor metabolic waste to boost immunotherapy.

A recent study published in Nature by Canale et al. (2021) shows that engineered probiotic bacteria can be used to augment the availability of nutrients required for optimal immune cell function in tumors. This approach enhances anti-tumor immunity and improves the efficacy of immunotherapy in mouse models of cancer.

Engineering living therapeutics with synthetic biology.

The steadfast advance of the synthetic biology field has enabled scientists to use genetically engineered cells, instead of small molecules or biologics, as the basis for the development of novel therapeutics. Cells endowed with synthetic gene circuits can control the localization, timing and dosage of therapeutic activities in response to specific disease biomarkers and thus represent a powerful new weapon in the fight against disease. Here, we conceptualize how synthetic biology approaches can be applied to programme living cells with therapeutic functions and discuss the advantages that they offer over conventional therapies in terms of flexibility, specificity and predictability, as well as challenges for their development. We present notable advances in the creation of engineered cells that harbour synthetic gene circuits capable of biological sensing and computation of signals derived from intracellular or extracellular biomarkers. We categorize and describe these developments based on the cell scaffold (human or microbial) and the site at which the engineered cell exerts its therapeutic function within its human host. The design of cell-based therapeutics with synthetic biology is a rapidly growing strategy in medicine that holds great promise for the development of effective treatments for a wide variety of human diseases.

Toward a genetic system in the marine cyanobacterium Prochlorococcus.

As the smallest and most abundant primary producer in the oceans, the cyanobacterium Prochlorococcus is of interest to diverse branches of science. For the past 30 years, research on this minimal phototroph has led to a growing understanding of biological organization across multiple scales, from the genome to the global ocean ecosystem. Progress in understanding drivers of its diversity and ecology, as well as molecular mechanisms underpinning its streamlined simplicity, has been hampered by the inability to manipulate these cells genetically. Multiple attempts have been made to develop an efficient genetic transformation method for Prochlorococcus over the years; all have been unsuccessful to date, despite some success with their close relative, Synechococcus . To avoid the pursuit of unproductive paths, we report here what has not worked in our hands, as well as our progress developing a method to screen the most efficient electroporation parameters for optimal DNA delivery into Prochlorococcus cells. We also report a novel protocol for obtaining axenic colonies and a new method for differentiating live and dead cells. The electroporation method can be used to optimize DNA delivery into any bacterium, making it a useful tool for advancing transformation systems in other genetically recalcitrant microorganisms.

A Deep Learning Approach to Antibiotic Discovery.

Due to the rapid emergence of antibiotic-resistant bacteria, there is a growing need to discover new antibiotics. To address this challenge, we trained a deep neural network capable of predicting molecules with antibacterial activity. We performed predictions on multiple chemical libraries and discovered a molecule from the Drug Repurposing Hub-halicin-that is structurally divergent from conventional antibiotics and displays bactericidal activity against a wide phylogenetic spectrum of pathogens including Mycobacterium tuberculosis and carbapenem-resistant Enterobacteriaceae. Halicin also effectively treated Clostridioides difficile and pan-resistant Acinetobacter baumannii infections in murine models. Additionally, from a discrete set of 23 empirically tested predictions from >107 million molecules curated from the ZINC15 database, our model identified eight antibacterial compounds that are structurally distant from known antibiotics. This work highlights the utility of deep learning approaches to expand our antibiotic arsenal through the discovery of structurally distinct antibacterial molecules.

Probiotic strains detect and suppress cholera in mice.

Microbiota-modulating interventions are an emerging strategy to promote gastrointestinal homeostasis. Yet, their use in the detection, prevention, and treatment of acute infections remains underexplored. We report the basis of a probiotic-based strategy to promote colonization resistance and point-of-need diagnosis of cholera, an acute diarrheal disease caused by the pathogen Vibrio cholerae Oral administration of Lactococcus lactis, a common dietary fermentative bacterium, reduced intestinal V. cholerae burden and improved survival in infected infant mice through the production of lactic acid. Furthermore, we engineered an L. lactis strain that specifically detects quorum-sensing signals of V. cholerae in the gut and triggers expression of an enzymatic reporter that is readily detected in fecal samples. We postulate that preventive dietary interventions with fermented foods containing natural and engineered L. lactis strains may hinder cholera progression and improve disease surveillance in populations at risk of cholera outbreaks.

Evolutionary radiation of lanthipeptides in marine cyanobacteria.

Lanthipeptides are ribosomally derived peptide secondary metabolites that undergo extensive posttranslational modification. Prochlorosins are a group of lanthipeptides produced by certain strains of the ubiquitous marine picocyanobacteria Prochlorococcus and Synechococcus Unlike other lanthipeptide-producing bacteria, picocyanobacteria use an unprecedented mechanism of substrate promiscuity for the production of numerous and diverse lanthipeptides using a single lanthionine synthetase. Through a cross-scale analysis of prochlorosin biosynthesis genes-from genomes to oceanic populations-we show that marine picocyanobacteria have the collective capacity to encode thousands of different cyclic peptides, few of which would display similar ring topologies. To understand how this extensive structural diversity arises, we used deep sequencing of wild populations to reveal genetic variation patterns in prochlorosin genes. We present evidence that structural variability among prochlorosins is the result of a diversifying selection process that favors large, rather than small, sequence changes in the precursor peptide genes. This mode of molecular evolution disregards any conservation of the ancestral structure and enables the emergence of extensively different cyclic peptides through short mutational paths based on indels. Contrary to its fast-evolving peptide substrates, the prochlorosin lanthionine synthetase evolves under a strong purifying selection, indicating that the diversification of prochlorosins is not constrained by commensurate changes in the biosynthetic enzyme. This evolutionary interplay between the prochlorosin peptide substrates and the lanthionine synthetase suggests that structure diversification, rather than structure refinement, is the driving force behind the creation of new prochlorosin structures and represents an intriguing mechanism by which natural product diversity arises.

Nitrogen cost minimization is promoted by structural changes in the transcriptome of N-deprived Prochlorococcus cells.

Prochlorococcus is a globally abundant marine cyanobacterium with many adaptations that reduce cellular nutrient requirements, facilitating growth in its nutrient-poor environment. One such genomic adaptation is the preferential utilization of amino acids containing fewer N-atoms, which minimizes cellular nitrogen requirements. We predicted that transcriptional regulation might further reduce cellular N budgets during transient N limitation. To explore this, we compared transcription start sites (TSSs) in Prochlorococcus MED4 under N-deprived and N-replete conditions. Of 64 genes with primary and internal TSSs in both conditions, N-deprived cells initiated transcription downstream of primary TSSs more frequently than N-replete cells. Additionally, 117 genes with only an internal TSS demonstrated increased internal transcription under N-deprivation. These shortened transcripts encode predicted proteins with an average of 21% less N content compared to full-length transcripts. We hypothesized that low translation rates, which afford greater control over protein abundances, would be beneficial to relatively slow-growing organisms like Prochlorococcus. Consistent with this idea, we found that Prochlorococcus exhibits greater usage of glycine-glycine motifs, which causes translational pausing, when compared to faster growing microbes. Our findings indicate that structural changes occur within the Prochlorococcus MED4 transcriptome during N-deprivation, potentially altering the size and structure of proteins expressed under nutrient limitation.

Imaging secondary metabolism of Streptomyces sp. Mg1 during cellular lysis and colony degradation of competing Bacillus subtilis.

Soil streptomycetes are saprotrophic bacteria that secrete numerous secondary metabolites and enzymes for extracellular functions. Many streptomycetes produce antibiotics thought to protect vegetative mycelia from competing organisms. Here we report that an organism isolated from soil, Streptomyces sp. Mg1, actively degrades colonies and causes cellular lysis of Bacillus subtilis when the organisms are cultured together. We predicted that the inhibition and degradation of B. subtilis colonies in this competition depends upon a combination of secreted factors, including small molecule metabolites and enzymes. To begin to unravel this complex competitive phenomenon, we use a MALDI imaging mass spectrometry strategy to map the positions of metabolites secreted by both organisms. In this report, we show that Streptomyces sp. Mg1 produces the macrolide antibiotic chalcomycin A, which contributes to inhibition of B. subtilis growth in combination with other, as yet unidentified factors. We suggest that efforts to understand competitive and cooperative interactions between bacterial species benefit from assays that pair living organisms and probe the complexity of metabolic exchanges between them.

IS-seq: a novel high throughput survey of in vivo IS6110 transposition in multiple Mycobacterium tuberculosis genomes.

BACKGROUND: The insertion element IS6110 is one of the main sources of genomic variability in Mycobacterium tuberculosis, the etiological agent of human tuberculosis. Although IS 6110 has been used extensively as an epidemiological marker, the identification of the precise chromosomal insertion sites has been limited by technical challenges. Here, we present IS-seq, a novel method that combines high-throughput sequencing using Illumina technology with efficient combinatorial sample multiplexing to simultaneously probe 519 clinical isolates, identifying almost all the flanking regions of the element in a single experiment. RESULTS: We identified a total of 6,976 IS6110 flanking regions on the different isolates. When validated using reference strains, the method had 100% specificity and 98% positive predictive value. The insertions mapped to both coding and non-coding regions, and in some cases interrupted genes thought to be essential for virulence or in vitro growth. Strains were classified into families using insertion sites, and high agreement with previous studies was observed. CONCLUSIONS: This high-throughput IS-seq method, which can also be used to map insertions in other organisms, extends previous surveys of in vivo interrupted loci and provides a baseline for probing the consequences of disruptions in M. tuberculosis strains.

Genomic signatures of the haarlem lineage of Mycobacterium tuberculosis: implications of strain genetic variation in drug and vaccine development.

Tuberculosis is the world's leading cause of death due to a single infectious agent, and efforts aimed at its control require a better understanding of host, environmental, and bacterial factors that govern disease outcome. Growing evidence indicates that certain Mycobacterium tuberculosis strains of distinct phylogeographic lineages elicit unique immunopathological events. However, identifying the genetic basis of these phenotypic peculiarities has proven difficult. Here we report the presence of six large sequence polymorphisms which, together with two single-nucleotide changes previously described by our group, consistently differentiate Haarlem strains from the remaining M. tuberculosis lineages. The six newly found Haarlem-specific genetic events are four deletions, which altogether involve more than 13 kb, and two intragenic insertions of the element IS6110. The absence of the genes involved in these polymorphisms could have an important physiological impact on Haarlem strains, i.e., by affecting key genes, such as Rv1354c and cyp121, which have been recently proposed as plausible drug targets. These lineage-specific polymorphisms can serve as genetic markers for the rapid PCR identification of Haarlem strains, providing a useful tool for strain surveillance and molecular epidemiology studies. Strain variability such as that described here underscores the need for the definition of a core set of essential genes in M. tuberculosis that are ubiquitously present in all circulating lineages, as a requirement in the development of effective antituberculosis drugs and vaccines.

Analysis of the genetic variation in Mycobacterium tuberculosis strains by multiple genome alignments.

BACKGROUND: The recent determination of the complete nucleotide sequence of several Mycobacterium tuberculosis (MTB) genomes allows the use of comparative genomics as a tool for dissecting the nature and consequence of genetic variability within this species. The multiple alignment of the genomes of clinical strains (CDC1551, F11, Haarlem and C), along with the genomes of laboratory strains (H37Rv and H37Ra), provides new insights on the mechanisms of adaptation of this bacterium to the human host. FINDINGS: The genetic variation found in six M. tuberculosis strains does not involve significant genomic rearrangements. Most of the variation results from deletion and transposition events preferentially associated with insertion sequences and genes of the PE/PPE family but not with genes implicated in virulence. Using a Perl-based software islandsanalyser, which creates a representation of the genetic variation in the genome, we identified differences in the patterns of distribution and frequency of the polymorphisms across the genome. The identification of genes displaying strain-specific polymorphisms and the extrapolation of the number of strain-specific polymorphisms to an unlimited number of genomes indicates that the different strains contain a limited number of unique polymorphisms. CONCLUSION: The comparison of multiple genomes demonstrates that the M. tuberculosis genome is currently undergoing an active process of gene decay, analogous to the adaptation process of obligate bacterial symbionts. This observation opens new perspectives into the evolution and the understanding of the pathogenesis of this bacterium.

Nitrate reductase assay for the rapid detection of pyrazinamide resistance in Mycobacterium tuberculosis using nicotinamide.

OBJECTIVES: The purpose of this study was to develop the nitrate reductase assay (NRA) for the rapid detection of pyrazinamide resistance in Mycobacterium tuberculosis using nicotinamide resistance as a marker of pyrazinamide resistance in Löwenstein-Jensen (LJ) medium at neutral pH. METHODS: We tested 68 M. tuberculosis isolates using nicotinamide at three different concentrations (1000, 500 and 250 mg/L) by the NRA in LJ medium and compared the results with those obtained with the BACTEC 460-TB or the BACTEC MGIT 960 as reference standard methods. Mutations in the pncA gene were detected by DNA sequencing of the pyrazinamide-resistant isolates. RESULTS: Out of 34 M. tuberculosis pyrazinamide-resistant isolates, 31 were found to be resistant to 1000 and 500 mg/L nicotinamide giving sensitivity and specificity of 91% and 94%, respectively. At 250 mg/L nicotinamide, the sensitivity and specificity decreased to 91% and 71%, respectively. Results were obtained in an average of 10 days. Based on these results, a tentative breakpoint concentration of 500 mg/L nicotinamide was defined. DNA sequencing of the pncA gene detected mutations in 26 out of 34 M. tuberculosis isolates resistant to pyrazinamide. CONCLUSIONS: The NRA using nicotinamide to detect resistance to pyrazinamide in LJ medium is a rapid and accurate method that could be useful in limited-resource countries where the BACTEC 460-TB or the BACTEC MGIT 960 system is not available.