A White-Box Machine Learning Approach for Revealing Antibiotic Mechanisms of Action

Bacterial death from treatment with fluoroquinolones and other lethal stressors

INTRODUCTION

Lethal stressors, including antimicrobials, kill bacteria in part through a metabolic response proposed to involve reactive oxygen species (ROS). The quinolone anti-bacterials have served as key experimental tools in developing this idea.

AREAS COVERED

Bacteriostatic and bactericidal action of quinolones are distinguished, with emphasis on the contribution of chromosome fragmentation and ROS accumulation to bacterial death. Action of non-quinolone antibacterials and non-antimicrobial stressors is described to provide a general framework for understanding stress-mediated, bacterial death.

EXPERT OPINION

Quinolones trap topoisomerases on DNA in reversible complexes that block DNA replication and bacterial growth. At elevated drug concentrations, DNA ends are released from topoisomerase-mediated constraint, leading to the idea that death arises from chromosome fragmentation. However, DNA ends also stimulate repair, which is energetically expensive. An incompletely understood metabolic shift occurs, and ROS accumulate. Even after quinolone removal, ROS continue to amplify, generating secondary and tertiary damage that overwhelms repair and causes death. Repair may also contribute to death directly via DNA breaks arising from incomplete base-excision repair of ROS-oxidized nucleotides. Remarkably, perturbations that interfere with ROS accumulation confer tolerance to many diverse lethal agents.

Predicting Drug Response and Synergy Using a Deep Learning Model of Human Cancer Cells

Most drugs entering clinical trials fail, often related to an incomplete understanding of the mechanisms governing drug response. Machine learning techniques hold immense promise for better drug response predictions, but most have not reached clinical practice due to their lack of interpretability and their focus on monotherapies. We address these challenges by developing DrugCell, an interpretable deep learning model of human cancer cells trained on the responses of 1,235 tumor cell lines to 684 drugs. Tumor genotypes induce states in cellular subsystems that are integrated with drug structure to predict response to therapy and, simultaneously, learn biological mechanisms underlying the drug response. DrugCell predictions are accurate in cell lines and also stratify clinical outcomes. Analysis of DrugCell mechanisms leads directly to the design of synergistic drug combinations, which we validate systematically by combinatorial CRISPR, drug-drug screening in vitro, and patient-derived xenografts. DrugCell provides a blueprint for constructing interpretable models for predictive medicine.

Network-based machine learning in colorectal and bladder organoid models predicts anti-cancer drug efficacy in patients

Cancer patient classification using predictive biomarkers for anti-cancer drug responses is essential for improving therapeutic outcomes. However, current machine-learning-based predictions of drug response often fail to identify robust translational biomarkers from preclinical models. Here, we present a machine-learning framework to identify robust drug biomarkers by taking advantage of network-based analyses using pharmacogenomic data derived from three-dimensional organoid culture models. The biomarkers identified by our approach accurately predict the drug responses of 114 colorectal cancer patients treated with 5-fluorouracil and 77 bladder cancer patients treated with cisplatin. We further confirm our biomarkers using external transcriptomic datasets of drug-sensitive and -resistant isogenic cancer cell lines. Finally, concordance analysis between the transcriptomic biomarkers and independent somatic mutation-based biomarkers further validate our method. This work presents a method to predict cancer patient drug responses using pharmacogenomic data derived from organoid models by combining the application of gene modules and network-based approaches.

Leveraging genome-scale metabolic models for human health applications

Genome-scale metabolic modeling is a scalable and extensible computational method for analyzing and predicting biological function. With the ongoing improvements in computational methods and experimental capabilities, genome-scale metabolic models (GEMs) are demonstrating utility in addressing human health applications. The initial areas of highest impact are likely to be health applications where disease states involve metabolic changes. In this review, we focus on recent application of GEMs to studying cancer and the human microbiome by describing the enabling methodologies and outcomes of these studies. We conclude with proposing some areas of research that are likely to arise as a result of recent methodological advances.

Combining structure and genomics to understand antimicrobial resistance

Antimicrobials against bacterial, viral and parasitic pathogens have transformed human and animal health. Nevertheless, their widespread use (and misuse) has led to the emergence of antimicrobial resistance (AMR) which poses a potentially catastrophic threat to public health and animal husbandry. There are several routes, both intrinsic and acquired, by which AMR can develop. One major route is through non-synonymous single nucleotide polymorphisms (nsSNPs) in coding regions. Large scale genomic studies using high-throughput sequencing data have provided powerful new ways to rapidly detect and respond to such genetic mutations linked to AMR. However, these studies are limited in their mechanistic insight. Computational tools can rapidly and inexpensively evaluate the effect of mutations on protein function and evolution. Subsequent insights can then inform experimental studies, and direct existing or new computational methods. Here we review a range of sequence and structure-based computational tools, focussing on tools successfully used to investigate mutational effect on drug targets in clinically important pathogens, particularly Mycobacterium tuberculosis. Combining genomic results with the biophysical effects of mutations can help reveal the molecular basis and consequences of resistance development. Furthermore, we summarise how the application of such a mechanistic understanding of drug resistance can be applied to limit the impact of AMR.

Combating Antibiotic Tolerance Through Activating Bacterial Metabolism

The emergence of antibiotic tolerance enables genetically susceptible bacteria to withstand the killing by clinically relevant antibiotics. As is reported, an increasing body of evidence sheds light on the critical and underappreciated role of antibiotic tolerance in the disease burden of bacterial infections. Considering this tense situation, new therapeutic strategies are urgently required for combating antibiotic tolerance. Herein, we provide an insightful illustration to distinguish between antibiotic resistance and tolerance, and highlight its clinical significance and complexities of drug-tolerant bacteria. Then, we discuss the close relationship between antibiotic tolerance and bacterial metabolism. As such, a bacterial metabolism-based approach was proposed to counter antibiotic tolerance. These exogenous metabolites including amino acids, tricarboxylic acid cycle (TCA cycle) metabolites, and nucleotides effectively activate bacterial metabolism and convert the tolerant cells to sensitive cells, and eventually restore antibiotic efficacy. A better understanding of molecular mechanisms of antibiotic tolerance particularly in vivo would substantially drive the development of novel strategies targeting bacterial metabolism.

The era of big data: Genome-scale modelling meets machine learning

With omics data being generated at an unprecedented rate, genome-scale modelling has become pivotal in its organisation and analysis. However, machine learning methods have been gaining ground in cases where knowledge is insufficient to represent the mechanisms underlying such data or as a means for data curation prior to attempting mechanistic modelling. We discuss the latest advances in genome-scale modelling and the development of optimisation algorithms for network and error reduction, intracellular constraining and applications to strain design. We further review applications of supervised and unsupervised machine learning methods to omics datasets from microbial and mammalian cell systems and present efforts to harness the potential of both modelling approaches through hybrid modelling.

The Biophysics of Regenerative Repair Suggests New Perspectives on Biological Causation

Evolution exploits the physics of non-neural bioelectricity to implement anatomical homeostasis: a process in which embryonic patterning, remodeling, and regeneration achieve invariant anatomical outcomes despite external interventions. Linear "developmental pathways" are often inadequate explanations for dynamic large-scale pattern regulation, even when they accurately capture relationships between molecular components. Biophysical and computational aspects of collective cell activity toward a target morphology reveal interesting aspects of causation in biology. This is critical not only for unraveling evolutionary and developmental events, but also for the design of effective strategies for biomedical intervention. Bioelectrical controls of growth and form, including stochastic behavior in such circuits, highlight the need for the formulation of nuanced views of pathways, drivers of system-level outcomes, and modularity, borrowing from concepts in related disciplines such as cybernetics, control theory, computational neuroscience, and information theory. This approach has numerous practical implications for basic research and for applications in regenerative medicine and synthetic bioengineering.

Deep learning meets metabolomics: a methodological perspective

Deep learning (DL), an emerging area of investigation in the fields of machine learning and artificial intelligence, has markedly advanced over the past years. DL techniques are being applied to assist medical professionals and researchers in improving clinical diagnosis, disease prediction and drug discovery. It is expected that DL will help to provide actionable knowledge from a variety of 'big data', including metabolomics data. In this review, we discuss the applicability of DL to metabolomics, while presenting and discussing several examples from recent research. We emphasize the use of DL in tackling bottlenecks in metabolomics data acquisition, processing, metabolite identification, as well as in metabolic phenotyping and biomarker discovery. Finally, we discuss how DL is used in genome-scale metabolic modelling and in interpretation of metabolomics data. The DL-based approaches discussed here may assist computational biologists with the integration, prediction and drawing of statistical inference about biological outcomes, based on metabolomics data.

Fast antimicrobial susceptibility testing on Escherichia coli by metabolic heat nanocalorimetry

Fast spreading of antimicrobial resistance is now considered a major global health threat. New technologies are required, enabling rapid diagnostics of bacterial infection combined with fast antimicrobial susceptibility testing (AST) for evaluating the efficiency and dosage of antimicrobial compounds in vitro. This work presents an integrated chip-based isothermal nanocalorimetry platform for direct microbial metabolic heat measurements and evaluates its potential for fast AST. Direct detection of the bacteria-generated heat allows monitoring of metabolic activity and antimicrobial action at subinhibitory concentrations in real-time. The high heat sensitivity of the platform enables bacterial growth detection within only a few hours of incubation, whereas growth inhibition upon administration of antibiotics is revealed by a decrease or the absence of the heat signal. Antimicrobial stress results in lag phase extension and metabolic energy spilling. Oxygen consumption and optical density measurements provide a more holistic insight of the metabolic state and the evolution of bacterial biomass. As a proof-of-concept, a metabolic heat-based AST study on Escherichia coli as model organism with 3 clinically relevant antibiotics is performed and the minimum inhibitory concentrations are determined.

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.

Evaluation of Nucleoside Analogs as Antimicrobials Targeting Unique Enzymes in Borrelia burgdorferi

The first line therapy for Lyme disease is treatment with doxycycline, amoxicillin, or cefuroxime. In endemic regions, the persistence of symptoms in many patients after completion of antibiotic treatment remains a major healthcare concern. The causative agent of Lyme disease is a spirochete, Borrelia burgdorferi, an extreme auxotroph that cannot exist under free-living conditions and depends upon the tick vector and mammalian hosts to fulfill its nutritional needs. Despite lacking all major biosynthetic pathways, B. burgdorferi uniquely possesses three homologous and functional methylthioadenosine/S-adenosylhomocysteine nucleosidases (MTANs: Bgp, MtnN, and Pfs) involved in methionine and purine salvage, underscoring the critical role these enzymes play in the life cycle of the spirochete. At least one MTAN, Bgp, is exceptional in its presence on the surface of Lyme spirochetes and its dual functionality in nutrient salvage and glycosaminoglycan binding involved in host-cell adherence. Thus, MTANs offer highly promising targets for discovery of new antimicrobials. Here we report on our studies to evaluate five nucleoside analogs for MTAN inhibitory activity, and cytotoxic or cytostatic effects on a bioluminescently engineered strain of B. burgdorferi. All five compounds were either alternate substrates and/or inhibitors of MTAN activity, and reduced B. burgdorferi growth. Two inhibitors: 5'-deoxy-5'-iodoadenosine (IADO) and 5'-deoxy-5'-ethyl-immucillin A (dEt-ImmA) showed bactericidal activity. Thus, these inhibitors exhibit high promise and form the foundation for development of novel and effective antimicrobials to treat Lyme disease.

Predictive Signatures of 19 Antibiotic-Induced Escherichia coli Proteomes

Identifying the mode of action (MOA) of antibacterial compounds is the fundamental basis for the development of new antibiotics, and the challenge increases with the emerging secondary and indirect effect from antibiotic stress. Although various omics-based system biology approaches are currently available, enhanced throughput, accuracy, and comprehensiveness are still desirable to better define antibiotic MOA. Using label-free quantitative proteomics, we present here a comprehensive reference map of proteomic signatures of Escherichia coli under challenge of 19 individual antibiotics. Applying several machine learning techniques, we derived a panel of 14 proteins that can be used to classify the antibiotics into different MOAs with nearly 100% accuracy. These proteins tend to mediate diverse bacterial cellular and metabolic processes. Transcriptomic level profiling correlates well with protein expression changes in discriminating different antibiotics. The reported expression signatures will aid future studies in identifying MOA of unknown compounds and facilitate the discovery of novel antibiotics.

Cysteine Potentiates Bactericidal Antibiotics Activity Against Gram-Negative Bacterial Persisters

Purpose

Bacterial metabolism regulators offer a novel productive strategy in the eradication of antibiotic refractory bacteria, particularly bacterial persisters. However, the potential of amino acids in the fight against Gram-negative bacterial persisters has not been fully explored. The aim of this study is to investigate the potentiation of amino acids to antibiotics in combating Gram-negative bacterial persisters and to reveal the underlying mechanisms of action.

Methods

Bactericidal activity of antibiotics in the absence or presence of amino acids was evaluated through detecting the reduction of bacterial CFUs. The ratio of NAD⁺/NADH in E. coli B2 persisters was determined using assay kit with WST-8. Bacterial respiration and ROS production were measured by the reduction of iodonitrotetrazolium chloride and fluorescent probe 2′,7′-dichlorodihydrofluorescein diacetate, respectively.

Results

In this study, we found that cysteine possesses excellent synergistic bactericidal activity with ciprofloxacin against multiple Gram-negative bacterial persisters. Furthermore, the potentiation of cysteine was evaluated in exponential and stationary-phase E. coli ATCC 25922 and E. coli B2. Interestingly, cysteine significantly improves three bactericidal antibiotics killing against stationary-phase bacteria, but not exponential-phase bacteria, implying that the effect of cysteine correlates with the metabolic state of bacteria. Mechanistic studies revealed that cysteine accelerates the bacterial TCA cycle and promotes bacterial respiration and ROS production. These metabolic regulation effects of cysteine re-sensitive bacterial persisters to antibiotic killing.

Conclusion

Collectively, our study highlights the synergistic bactericidal activity of bacterial metabolism regulators such as cysteine with commonly used antibiotics against Gram-negative bacterial persisters.

Rapid antibiotic susceptibility testing of bacteria from patients' blood via assaying bacterial metabolic response with surface-enhanced Raman spectroscopy

Blood stream infection is one of the major public health issues characterized with high cost and high mortality. Timely effective antibiotics usage to control infection is crucial for patients’ survival. The standard microbiological diagnosis of infection however can last days. The delay in accurate antibiotic therapy would lead to not only poor clinical outcomes, but also to a rise in antibiotic resistance due to widespread use of empirical broad-spectrum antibiotics. An important measure to tackle this problem is fast determination of bacterial antibiotic susceptibility to optimize antibiotic treatment. We show that a protocol based on surface-enhanced Raman spectroscopy can obtain consistent antibiotic susceptibility test results from clinical blood-culture samples within four hours. The characteristic spectral signatures of the obtained spectra of Staphylococcus aureus and Escherichia coli—prototypic Gram-positive and Gram-negative bacteria—became prominent after an effective pretreatment procedure removed strong interferences from blood constituents. Using them as the biomarkers of bacterial metabolic responses to antibiotics, the protocol reported the susceptibility profiles of tested drugs against these two bacteria acquired from patients’ blood with high specificity, sensitivity and speed.

Metabolic Modelling as a Framework for Metabolomics Data Integration and Analysis

Metabolic networks are regulated to ensure the dynamic adaptation of biochemical reaction fluxes to maintain cell homeostasis and optimal metabolic fitness in response to endogenous and exogenous perturbations. To this end, metabolism is tightly controlled by dynamic and intricate regulatory mechanisms involving allostery, enzyme abundance and post-translational modifications. The study of the molecular entities involved in these complex mechanisms has been boosted by the advent of high-throughput technologies. The so-called omics enable the quantification of the different molecular entities at different system layers, connecting the genotype with the phenotype. Therefore, the study of the overall behavior of a metabolic network and the omics data integration and analysis must be approached from a holistic perspective. Due to the close relationship between metabolism and cellular phenotype, metabolic modelling has emerged as a valuable tool to decipher the underlying mechanisms governing cell phenotype. Constraint-based modelling and kinetic modelling are among the most widely used methods to study cell metabolism at different scales, ranging from cells to tissues and organisms. These approaches enable integrating metabolomic data, among others, to enhance model predictive capabilities. In this review, we describe the current state of the art in metabolic modelling and discuss future perspectives and current challenges in the field.

Temperature-Induced Annual Variation in Microbial Community Changes and Resulting Metabolome Shifts in a Controlled Fermentation System

We used Chinese liquor fermentation as a model system to show that microbiome composition changes more dramatically across seasons than throughout the fermentation process within seasons. These changes translate to differences in the metabolome as the ultimate functional outcome of microbial activity, suggesting that temporal changes in microbiome composition are translating into functional changes. This result is striking as it suggests that microbial functioning, despite controlled conditions in the fermentors, fluctuates over season along with external temperature differences, which threatens a reproducible food taste. As such, we believe that our study provides a stepping-stone into novel taxonomy-functional studies that promote future work in other systems and that also is relevant in applied settings to better control surrounding conditions in food production.

We are rapidly increasing our understanding on the spatial distribution of microbial communities. However, microbial functioning, as well as temporal differences and mechanisms causing microbial community shifts, remains comparably little explored. Here, using Chinese liquor fermentation as a model system containing a low microbial diversity, we studied temporal changes in microbial community structure and functioning. For that, we used high-throughput sequencing to analyze the composition of bacteria and fungi and analyzed the microbially derived metabolome throughout the fermentation process in all four seasons in both 2018 and 2019. We show that microbial communities and the metabolome changed throughout the fermentation process in each of the four seasons, with metabolome diversity increasing throughout the fermentation process. Across seasons, bacterial and fungal communities as well as the metabolome driven by 10 indicator microorganisms and six metabolites varied even more. Daily average temperature in the external surroundings was the primary determinant of the observed temporal microbial community and metabolome changes. Collectively, our work reveals critical insights into patterns and processes determining temporal changes of microbial community composition and functioning. We highlight the importance of linking taxonomic to functional changes in microbial ecology to enable predictions of human-relevant applications.

IMPORTANCE We used Chinese liquor fermentation as a model system to show that microbiome composition changes more dramatically across seasons than throughout the fermentation process within seasons. These changes translate to differences in the metabolome as the ultimate functional outcome of microbial activity, suggesting that temporal changes in microbiome composition are translating into functional changes. This result is striking as it suggests that microbial functioning, despite controlled conditions in the fermentors, fluctuates over season along with external temperature differences, which threatens a reproducible food taste. As such, we believe that our study provides a stepping-stone into novel taxonomy-functional studies that promote future work in other systems and that also is relevant in applied settings to better control surrounding conditions in food production.

A mechanism-aware and multiomic machine-learning pipeline characterizes yeast cell growth

Metabolic modeling and machine learning are key components in the emerging next generation of systems and synthetic biology tools, targeting the genotype-phenotype-environment relationship. Rather than being used in isolation, it is becoming clear that their value is maximized when they are combined. However, the potential of integrating these two frameworks for omic data augmentation and integration is largely unexplored. We propose, rigorously assess, and compare machine-learning-based data integration techniques, combining gene expression profiles with computationally generated metabolic flux data to predict yeast cell growth. To this end, we create strain-specific metabolic models for 1,143 Saccharomyces cerevisiae mutants and we test 27 machine-learning methods, incorporating state-of-the-art feature selection and multiview learning approaches. We propose a multiview neural network using fluxomic and transcriptomic data, showing that the former increases the predictive accuracy of the latter and reveals functional patterns that are not directly deducible from gene expression alone. We test the proposed neural network on a further 86 strains generated in a different experiment, therefore verifying its robustness to an additional independent dataset. Finally, we show that introducing mechanistic flux features improves the predictions also for knockout strains whose genes were not modeled in the metabolic reconstruction. Our results thus demonstrate that fusing experimental cues with in silico models, based on known biochemistry, can contribute with disjoint information toward biologically informed and interpretable machine learning. Overall, this study provides tools for understanding and manipulating complex phenotypes, increasing both the prediction accuracy and the extent of discernible mechanistic biological insights.

Biology of antimicrobial resistance and approaches to combat it

Insufficient development of new antibiotics and the rising resistance of bacteria to those that we have are putting the world at risk of losing the most widely curative class of medicines currently available. Preventing deaths from antimicrobial resistance (AMR) will require exploiting emerging knowledge not only about genetic AMR conferred by horizontal gene transfer or de novo mutations but also about phenotypic AMR, which lacks a stably heritable basis. This Review summarizes recent advances and continuing limitations in our understanding of AMR and suggests approaches for combating its clinical consequences, including identification of previously unexploited bacterial targets, new antimicrobial compounds, and improved combination drug regimens.

A biochemically-interpretable machine learning classifier for microbial GWAS

Current machine learning classifiers have successfully been applied to whole-genome sequencing data to identify genetic determinants of antimicrobial resistance (AMR), but they lack causal interpretation. Here we present a metabolic model-based machine learning classifier, named Metabolic Allele Classifier (MAC), that uses flux balance analysis to estimate the biochemical effects of alleles. We apply the MAC to a dataset of 1595 drug-tested Mycobacterium tuberculosis strains and show that MACs predict AMR phenotypes with accuracy on par with mechanism-agnostic machine learning models (isoniazid AUC = 0.93) while enabling a biochemical interpretation of the genotype-phenotype map. Interpretation of MACs for three antibiotics (pyrazinamide, para-aminosalicylic acid, and isoniazid) recapitulates known AMR mechanisms and suggest a biochemical basis for how the identified alleles cause AMR. Extending flux balance analysis to identify accurate sequence classifiers thus contributes mechanistic insights to GWAS, a field thus far dominated by mechanism-agnostic results.

Current machine learning classifiers have been applied to whole-genome sequencing data to identify determinants of antimicrobial resistance, but they lack interpretability. Here the authors present a metabolic machine learning classifier that uses flux balance analysis to estimate the biochemical effects of alleles.

Biological Machine Learning Combined with Campylobacter Population Genomics Reveals Virulence Gene Allelic Variants Cause Disease

Highly dimensional data generated from bacterial whole-genome sequencing is providing an unprecedented scale of information that requires an appropriate statistical analysis framework to infer biological function from populations of genomes. The application of genome-wide association study (GWAS) methods is an appropriate framework for bacterial population genome analysis that yields a list of candidate genes associated with a phenotype, but it provides an unranked measure of importance. Here, we validated a novel framework to define infection mechanism using the combination of GWAS, machine learning, and bacterial population genomics that ranked allelic variants that accurately identified disease. This approach parsed a dataset of 1.2 million single nucleotide polymorphisms (SNPs) and indels that resulted in an importance ranked list of associated alleles of porA in Campylobacter jejuni using spatiotemporal analysis over 30 years. We validated this approach using previously proven laboratory experimental alleles from an in vivo guinea pig abortion model. This framework, termed µPathML, defined intestinal and extraintestinal groups that have differential allelic porA variants that cause abortion. Divergent variants containing indels that defeated automated annotation were rescued using biological context and knowledge that resulted in defining rare, divergent variants that were maintained in the population over two continents and 30 years. This study defines the capability of machine learning coupled with GWAS and population genomics to simultaneously identify and rank alleles to define their role in infectious disease mechanisms.

Systems metabolomics: from metabolomic snapshots to design principles

Metabolomics is a rapidly expanding technology that finds increasing application in a variety of fields, form metabolic disorders to cancer, from nutrition and wellness to design and optimization of cell factories. The integration of metabolic snapshots with metabolic fluxes, physiological readouts, metabolic models, and knowledge-informed Artificial Intelligence tools, is required to obtain a system-level understanding of metabolism. The emerging power of multi-omic approaches and the development of integrated experimental and computational tools, able to dissect metabolic features at cellular and subcellular resolution, provide unprecedented opportunities for understanding design principles of metabolic (dis)regulation and for the development of precision therapies in multifactorial diseases, such as cancer and neurodegenerative diseases.

Chromosome-free bacterial cells are safe and programmable platforms for synthetic biology

A type of chromosome-free cell called SimCells (simple cells) has been generated from Escherichia coli, Pseudomonas putida, and Ralstonia eutropha. The removal of the native chromosomes of these bacteria was achieved by double-stranded breaks made by heterologous I-CeuI endonuclease and the degradation activity of endogenous nucleases. We have shown that the cellular machinery remained functional in these chromosome-free SimCells and was able to process various genetic circuits. This includes the glycolysis pathway (composed of 10 genes) and inducible genetic circuits. It was found that the glycolysis pathway significantly extended longevity of SimCells due to its ability to regenerate ATP and NADH/NADPH. The SimCells were able to continuously express synthetic genetic circuits for 10 d after chromosome removal. As a proof of principle, we demonstrated that SimCells can be used as a safe agent (as they cannot replicate) for bacterial therapy. SimCells were used to synthesize catechol (a potent anticancer drug) from salicylic acid to inhibit lung, brain, and soft-tissue cancer cells. SimCells represent a simplified synthetic biology chassis that can be programmed to manufacture and deliver products safely without interference from the host genome.

Artificial intelligence in oncology

Artificial intelligence (AI) has contributed substantially to the resolution of a variety of biomedical problems, including cancer, over the past decade. Deep learning, a subfield of AI that is highly flexible and supports automatic feature extraction, is increasingly being applied in various areas of both basic and clinical cancer research. In this review, we describe numerous recent examples of the application of AI in oncology, including cases in which deep learning has efficiently solved problems that were previously thought to be unsolvable, and we address obstacles that must be overcome before such application can become more widespread. We also highlight resources and datasets that can help harness the power of AI for cancer research. The development of innovative approaches to and applications of AI will yield important insights in oncology in the coming decade.

Constructing and deconstructing the bacterial cell wall

The history of modern medicine cannot be written apart from the history of the antibiotics. Antibiotics are cytotoxic secondary metabolites that are isolated from Nature. The antibacterial antibiotics disproportionately target bacterial protein structure that is distinct from eukaryotic protein structure, notably within the ribosome and within the pathways for bacterial cell-wall biosynthesis (for which there is not a eukaryotic counterpart). This review focuses on a pre-eminent class of antibiotics-the β-lactams, exemplified by the penicillins and cephalosporins-from the perspective of the evolving mechanisms for bacterial resistance. The mechanism of action of the β-lactams is bacterial cell-wall destruction. In the monoderm (single membrane, Gram-positive staining) pathogen Staphylococcus aureus the dominant resistance mechanism is expression of a β-lactam-unreactive transpeptidase enzyme that functions in cell-wall construction. In the diderm (dual membrane, Gram-negative staining) pathogen Pseudomonas aeruginosa a dominant resistance mechanism (among several) is expression of a hydrolytic enzyme that destroys the critical β-lactam ring of the antibiotic. The key sensing mechanism used by P. aeruginosa is monitoring the molecular difference between cell-wall construction and cell-wall deconstruction. In both bacteria, the resistance pathways are manifested only when the bacteria detect the presence of β-lactams. This review summarizes how the β-lactams are sensed and how the resistance mechanisms are manifested, with the expectation that preventing these processes will be critical to future chemotherapeutic control of multidrug resistant bacteria.

Antibiotic Lethality Is Impacted by Nutrient Availabilities: New Insights from Machine Learning

In this issue of Cell, Yang, Wright et al. describe a machine learning approach that that can provide mechanistic insight from chemical screens. They use this approach to uncover how the nutritional availability for Escherichia coli impacts lethality toward three widely used antibiotics.

Bioinformatics Approaches to the Understanding of Molecular Mechanisms in Antimicrobial Resistance

Antimicrobial resistance (AMR) is a major health concern worldwide. A better understanding of the underlying molecular mechanisms is needed. Advances in whole genome sequencing and other high-throughput unbiased instrumental technologies to study the molecular pathogenicity of infectious diseases enable the accumulation of large amounts of data that are amenable to bioinformatic analysis and the discovery of new signatures of AMR. In this work, we review representative methods published in the past five years to define major approaches developed to-date in the understanding of AMR mechanisms. Advantages and limitations for applications of these methods in clinical laboratory testing and basic research are discussed.

Genome-scale transcriptional dynamics and environmental biosensing

Genome-scale technologies have enabled mapping of the complex molecular networks that govern cellular behavior. An emerging theme in the analyses of these networks is that cells use many layers of regulatory feedback to constantly assess and precisely react to their environment. The importance of complex feedback in controlling the real-time response to external stimuli has led to a need for the next generation of cell-based technologies that enable both the collection and analysis of high-throughput temporal data. Toward this end, we have developed a microfluidic platform capable of monitoring temporal gene expression from over 2,000 promoters. By coupling the "Dynomics" platform with deep neural network (DNN) and associated explainable artificial intelligence (XAI) algorithms, we show how machine learning can be harnessed to assess patterns in transcriptional data on a genome scale and identify which genes contribute to these patterns. Furthermore, we demonstrate the utility of the Dynomics platform as a field-deployable real-time biosensor through prediction of the presence of heavy metals in urban water and mine spill samples, based on the the dynamic transcription profiles of 1,807 unique Escherichia coli promoters.

The mechanisms and cell signaling pathways of programmed cell death in the bacterial world

While programmed cell death was once thought to be exclusive to eukaryotic cells, there are now abundant examples of well regulated cell death mechanisms in bacteria. The mechanisms by which bacteria undergo programmed cell death are diverse, and range from the use of toxin-antitoxin systems, to prophage-driven cell lysis. Moreover, some bacteria have learned how to coopt programmed cell death systems in competing bacteria. Interestingly, many of the potential reasons as to why bacteria undergo programmed cell death may parallel those observed in eukaryotic cells, and may be altruistic in nature. These include protection against infection, recycling of nutrients, to ensure correct morphological development, and in response to stressors. In the following chapter, we discuss the molecular and signaling mechanisms by which bacteria undergo programmed cell death. We conclude by discussing the current open questions in this expanding field.

Bayesian Proxy Modelling for Estimating Black Carbon Concentrations using White-Box and Black-Box Models

Black carbon (BC) is an important component of particulate matter (PM) in urban environments. BC is typically emitted from gas and diesel engines, coal-fired power plants, and other sources that burn fossil fuel. In contrast to PM, BC measurements are not always available on a large scale due to the operational cost and complexity of the instrumentation. Therefore, it is advantageous to develop a mathematical model for estimating the quantity of BC in the air, termed a BC proxy, to enable widening of spatial air pollution mapping. This article presents the development of BC proxies based on a Bayesian framework using measurements of PM concentrations and size distributions from 10 to 10,000 nm from a recent mobile air pollution study across several areas of Jordan. Bayesian methods using informative priors can naturally prevent over-fitting in the modelling process and the methods generate a confidence interval around the prediction, thus the estimated BC concentration can be directly quantified and assessed. In particular, two types of models are developed based on their transparency and interpretability, referred to as white-box and black-box models. The proposed methods are tested on extensive data sets obtained from the measurement campaign in Jordan. In this study, black-box models perform slightly better due to their model complexity. Nevertheless, the results demonstrate that the performance of both models does not differ significantly. In practice, white-box models are relatively more convenient to be deployed, the methods are well understood by scientists, and the models can be used to better understand key relationships.

Bacterial metabolic state more accurately predicts antibiotic lethality than growth rate

Growth rate and metabolic state of bacteria have been separately shown to affect antibiotic efficacy^1-3. However, the two are interrelated as bacterial growth inherently imposes a metabolic burden⁴; thus, determining individual contributions from each is challenging^5,6. Indeed, faster growth is often correlated with increased antibiotic efficacy^7,8; however, the concurrent role of metabolism in that relationship has not been well characterized. As a result, a clear understanding of the interdependence between growth and metabolism, and their implications for antibiotic efficacy, are lacking⁹. Here, we measured growth and metabolism in parallel across a broad range of coupled and uncoupled conditions to determine their relative contribution to antibiotic lethality. We show that when growth and metabolism are uncoupled, antibiotic lethality uniformly depends on the bacterial metabolic state at the time of treatment, rather than growth rate. We further reveal a critical metabolic threshold below which antibiotic lethality is negligible. These findings were general for a wide range of conditions, including nine representative bactericidal drugs and a diverse range of Gram-positive and Gram-negative species (Escherichia coli, Acinetobacter baumannii and Staphylococcus aureus). This study provides a cohesive metabolic-dependent basis for antibiotic-mediated cell death, with implications for current treatment strategies and future drug development.

Bacterial Metabolism and Antibiotic Efficacy

Antibiotics target energy-consuming processes. As such, perturbations to bacterial metabolic homeostasis are significant consequences of treatment. Here, we describe three postulates that collectively define antibiotic efficacy in the context of bacterial metabolism: (1) antibiotics alter the metabolic state of bacteria, which contributes to the resulting death or stasis; (2) the metabolic state of bacteria influences their susceptibility to antibiotics; and (3) antibiotic efficacy can be enhanced by altering the metabolic state of bacteria. Altogether, we aim to emphasize the close relationship between bacterial metabolism and antibiotic efficacy as well as propose areas of exploration to develop novel antibiotics that optimally exploit bacterial metabolic networks.

Bedaquiline Eliminates Bactericidal Activity of β-Lactams against Mycobacterium abscessus

The β-lactams imipenem and cefoxitin are used for the treatment of Mycobacterium abscessus lung infections. Here, we show that these cell wall synthesis inhibitors trigger a lethal bacterial ATP burst by increasing oxidative phosphorylation. Cotreatment of M. abscessus with the antimycobacterial ATP synthase inhibitor bedaquiline suppresses this ATP burst and eliminates the bactericidal activity of β-lactams. Thus, the addition of bedaquiline to β-lactam-containing regimes may negatively affect treatment outcome.