Proteome allocation is linked to transcriptional regulation through a modularized transcriptome.

It has proved challenging to quantitatively relate the proteome to the transcriptome on a per-gene basis. Recent advances in data analytics have enabled a biologically meaningful modularization of the bacterial transcriptome. We thus investigate whether matched datasets of transcriptomes and proteomes from bacteria under diverse conditions can be modularized in the same way to reveal novel relationships between their compositions. We find that; (1) the modules of the proteome and the transcriptome are comprised of a similar list of gene products, (2) the modules in the proteome often represent combinations of modules from the transcriptome, (3) known transcriptional and post-translational regulation is reflected in differences between two sets of modules, allowing for knowledge-mapping when interpreting module functions, and (4) through statistical modeling, absolute proteome allocation can be inferred from the transcriptome alone. Quantitative and knowledge-based relationships can thus be found at the genome-scale between the proteome and transcriptome in bacteria.

Independent component analysis reveals 49 independently modulated gene sets within the global transcriptional regulatory architecture of multidrug-resistant Acinetobacter baumannii.

Acinetobacter baumannii causes severe infections in humans, resists multiple antibiotics, and survives in stressful environmental conditions due to modulations of its complex transcriptional regulatory network (TRN). Unfortunately, our global understanding of the TRN in this emerging opportunistic pathogen is limited. Here, we apply independent component analysis, an unsupervised machine learning method, to a compendium of 139 RNA-seq data sets of three multidrug-resistant A. baumannii international clonal complex I strains (AB5075, AYE, and AB0057). This analysis allows us to define 49 independently modulated gene sets, which we call iModulons. Analysis of the identified A. baumannii iModulons reveals validating parallels to previously defined biological operons/regulons and provides a framework for defining unknown regulons. By utilizing the iModulons, we uncover potential mechanisms for a RpoS-independent general stress response, define global stress-virulence trade-offs, and identify conditions that may induce plasmid-borne multidrug resistance. The iModulons provide a model of the TRN that emphasizes the importance of transcriptional regulation of virulence phenotypes in A. baumannii. Furthermore, they suggest the possibility of future interventions to guide gene expression toward diminished pathogenic potential.IMPORTANCEThe rise in hospital outbreaks of multidrug-resistant Acinetobacter baumannii infections underscores the urgent need for alternatives to traditional broad-spectrum antibiotic therapies. The success of A. baumannii as a significant nosocomial pathogen is largely attributed to its ability to resist antibiotics and survive environmental stressors. However, there is limited literature available on the global, complex regulatory circuitry that shapes these phenotypes. Computational tools that can assist in the elucidation of A. baumannii's transcriptional regulatory network architecture can provide much-needed context for a comprehensive understanding of pathogenesis and virulence, as well as for the development of targeted therapies that modulate these pathways.

Laboratory evolution, transcriptomics, and modeling reveal mechanisms of paraquat tolerance.

Relationships between the genome, transcriptome, and metabolome underlie all evolved phenotypes. However, it has proved difficult to elucidate these relationships because of the high number of variables measured. A recently developed data analytic method for characterizing the transcriptome can simplify interpretation by grouping genes into independently modulated sets (iModulons). Here, we demonstrate how iModulons reveal deep understanding of the effects of causal mutations and metabolic rewiring. We use adaptive laboratory evolution to generate E. coli strains that tolerate high levels of the redox cycling compound paraquat, which produces reactive oxygen species (ROS). We combine resequencing, iModulons, and metabolic models to elucidate six interacting stress-tolerance mechanisms: (1) modification of transport, (2) activation of ROS stress responses, (3) use of ROS-sensitive iron regulation, (4) motility, (5) broad transcriptional reallocation toward growth, and (6) metabolic rewiring to decrease NADH production. This work thus demonstrates the power of iModulon knowledge mapping for evolution analysis.

A multi-scale expression and regulation knowledge base for Escherichia coli.

Transcriptomic data is accumulating rapidly; thus, scalable methods for extracting knowledge from this data are critical. Here, we assembled a top-down expression and regulation knowledge base for Escherichia coli. The expression component is a 1035-sample, high-quality RNA-seq compendium consisting of data generated in our lab using a single experimental protocol. The compendium contains diverse growth conditions, including: 9 media; 39 supplements, including antibiotics; 42 heterologous proteins; and 76 gene knockouts. Using this resource, we elucidated global expression patterns. We used machine learning to extract 201 modules that account for 86% of known regulatory interactions, creating the regulatory component. With these modules, we identified two novel regulons and quantified systems-level regulatory responses. We also integrated 1675 curated, publicly-available transcriptomes into the resource. We demonstrated workflows for analyzing new data against this knowledge base via deconstruction of regulation during aerobic transition. This resource illuminates the E. coli transcriptome at scale and provides a blueprint for top-down transcriptomic analysis of non-model organisms.

Machine learning uncovers the Pseudomonas syringae transcriptome in microbial communities and during infection.

IMPORTANCE: Pseudomonas syringae pv. tomato DC3000 is a model plant pathogen that infects tomatoes and Arabidopsis thaliana. The current understanding of global transcriptional regulation in the pathogen is limited. Here, we applied iModulon analysis to a compendium of RNA-seq data to unravel its transcriptional regulatory network. We characterize each co-regulated gene set, revealing the activity of major regulators across diverse conditions. We provide new insights on the transcriptional dynamics in interactions with the plant immune system and with other bacterial species, such as AlgU-dependent regulation of flagellar genes during plant infection and downregulation of siderophore production in the presence of a siderophore cheater. This study demonstrates the novel application of iModulons in studying temporal dynamics during host-pathogen and microbe-microbe interactions, and reveals specific insights of interest.

Elucidation of independently modulated genes in Streptococcus pyogenes reveals carbon sources that control its expression of hemolytic toxins.

Streptococcus pyogenes can cause a wide variety of acute infections throughout the body of its human host. An underlying transcriptional regulatory network (TRN) is responsible for altering the physiological state of the bacterium to adapt to each unique host environment. Consequently, an in-depth understanding of the comprehensive dynamics of the S. pyogenes TRN could inform new therapeutic strategies. Here, we compiled 116 existing high-quality RNA sequencing data sets of invasive S. pyogenes serotype M1 and estimated the TRN structure in a top-down fashion by performing independent component analysis (ICA). The algorithm computed 42 independently modulated sets of genes (iModulons). Four iModulons contained the nga-ifs-slo virulence-related operon, which allowed us to identify carbon sources that control its expression. In particular, dextrin utilization upregulated the nga-ifs-slo operon by activation of two-component regulatory system CovRS-related iModulons, altering bacterial hemolytic activity compared to glucose or maltose utilization. Finally, we show that the iModulon-based TRN structure can be used to simplify the interpretation of noisy bacterial transcriptome data at the infection site. IMPORTANCE S. pyogenes is a pre-eminent human bacterial pathogen that causes a wide variety of acute infections throughout the body of its host. Understanding the comprehensive dynamics of its TRN could inform new therapeutic strategies. Since at least 43 S. pyogenes transcriptional regulators are known, it is often difficult to interpret transcriptomic data from regulon annotations. This study shows the novel ICA-based framework to elucidate the underlying regulatory structure of S. pyogenes allows us to interpret the transcriptome profile using data-driven regulons (iModulons). Additionally, the observations of the iModulon architecture lead us to identify the multiple regulatory inputs governing the expression of a virulence-related operon. The iModulons identified in this study serve as a powerful guidepost to further our understanding of S. pyogenes TRN structure and dynamics.

Proteome allocation is linked to transcriptional regulation through a modularized transcriptome.

It has proved challenging to quantitatively relate the proteome to the transcriptome on a per-gene basis. Recent advances in data analytics have enabled a biologically meaningful modularization of the bacterial transcriptome. We thus investigated whether matched datasets of transcriptomes and proteomes from bacteria under diverse conditions could be modularized in the same way to reveal novel relationships between their compositions. We found that; 1) the modules of the proteome and the transcriptome are comprised of a similar list of gene products, 2) the modules in the proteome often represent combinations of modules from the transcriptome, 3) known transcriptional and post-translational regulation is reflected in differences between two sets of modules, allowing for knowledge-mapping when interpreting module functions, and 4) through statistical modeling, absolute proteome allocation can be inferred from the transcriptome alone. Quantitative and knowledge-based relationships can thus be found at the genome-scale between the proteome and transcriptome in bacteria.

Pan-Genome Analysis of Transcriptional Regulation in Six Salmonella enterica Serovar Typhimurium Strains Reveals Their Different Regulatory Structures.

Establishing transcriptional regulatory networks (TRNs) in bacteria has been limited to well-characterized model strains. Using machine learning methods, we established the transcriptional regulatory networks of six Salmonella enterica serovar Typhimurium strains from their transcriptomes. By decomposing a compendia of RNA sequencing (RNA-seq) data with independent component analysis, we obtained 400 independently modulated sets of genes, called iModulons. We (i) performed pan-genome analysis of the phylogroup structure of S. Typhimurium and analyzed the iModulons against this background, (ii) revealed different genetic signatures in pathogenicity islands that explained phenotypes, (iii) discovered three transport iModulons linked to antibiotic resistance, (iv) described concerted responses to cationic antimicrobial peptides, and (v) uncovered new regulons. Thus, by combining pan-genome and transcriptomic analytics, we revealed variations in TRNs across six strains of serovar Typhimurium. IMPORTANCE Salmonella enterica serovar Typhimurium is a pathogen involved in human nontyphoidal infections. Treating S. Typhimurium infections is difficult due to the species's dynamic adaptation to its environment, which is dictated by a complex transcriptional regulatory network (TRN) that is different across strains. In this study, we describe the use of independent component analysis to characterize the differential TRNs across the S. Typhimurium pan-genome using a compendium of high-quality RNA-seq data. This approach provided unprecedented insights into the differences between regulation of key cellular functions and pathogenicity in the different strains. The study provides an impetus to initiate a large-scale effort to reveal the TRN differences between the major phylogroups of the pathogenic bacteria, which could fundamentally impact personalizing treatments of bacterial pathogens.

Advanced transcriptomic analysis reveals the role of efflux pumps and media composition in antibiotic responses of Pseudomonas aeruginosa.

Pseudomonas aeruginosa is an opportunistic pathogen and major cause of hospital-acquired infections. The virulence of P. aeruginosa is largely determined by its transcriptional regulatory network (TRN). We used 411 transcription profiles of P. aeruginosa from diverse growth conditions to construct a quantitative TRN by identifying independently modulated sets of genes (called iModulons) and their condition-specific activity levels. The current study focused on the use of iModulons to analyze the biofilm production and antibiotic resistance of P. aeruginosa. Our analysis revealed: (i) 116 iModulons, 81 of which show strong association with known regulators; (ii) novel roles of regulators in modulating antibiotics efflux pumps; (iii) substrate-efflux pump associations; (iv) differential iModulon activity in response to beta-lactam antibiotics in bacteriological and physiological media; (v) differential activation of 'Cell Division' iModulon resulting from exposure to different beta-lactam antibiotics and (vi) a role of the PprB iModulon in the stress-induced transition from planktonic to biofilm lifestyle. In light of these results, the construction of an iModulon-based TRN provides a transcriptional regulatory basis for key aspects of P. aeruginosa infection, such as antibiotic stress responses and biofilm formation. Taken together, our results offer a novel mechanistic understanding of P. aeruginosa virulence.

Membrane transporter identification and modulation via adaptive laboratory evolution.

Membrane transport proteins are potential targets for medical and biotechnological applications. However, more than 30% of reported membrane transporter families are either poorly characterized or lack adequate functional annotation. Here, adaptive laboratory evolution was leveraged to identify membrane transporters for a set of four amino acids as well as specific mutations that modulate the activities of these transporters. Specifically, Escherichia coli was adaptively evolved under increasing concentrations of L-histidine, L-phenylalanine, L-threonine, and L-methionine separately with multiple replicate evolutions. Evolved populations and isolated clones displayed growth rates comparable to the unstressed ancestral strain at elevated concentrations (four-to six-fold increases) of the targeted amino acids. Whole genome sequencing of the evolved strains revealed a diverse number of key mutations, including SNPs, small deletions, and copy number variants targeting the transporters leuE for histidine, yddG for phenylalanine, yedA for methionine, and brnQ and rhtC for threonine. Reverse engineering of the mutations in the ancestral strain established mutation causality of the specific mutations for the tolerant phenotypes. The functional roles of yedA and brnQ in the transport of methionine and threonine, respectively, are novel assignments and their functional roles were validated using a flow cytometry cellular accumulation assay. To demonstrate how the identified transporters can be leveraged for production, an L-phenylalanine overproduction strain was shown to be a superior producer when the identified yddG exporter was overexpressed. Overall, the results revealed the striking efficiency of laboratory evolution to identify transporters and specific mutational mechanisms to modulate their activities, thereby demonstrating promising applicability in transporter discovery efforts and strain engineering.

Machine-learning from Pseudomonas putida KT2440 transcriptomes reveals its transcriptional regulatory network.

Bacterial gene expression is orchestrated by numerous transcription factors (TFs). Elucidating how gene expression is regulated is fundamental to understanding bacterial physiology and engineering it for practical use. In this study, a machine-learning approach was applied to uncover the genome-scale transcriptional regulatory network (TRN) in Pseudomonas putida KT2440, an important organism for bioproduction. We performed independent component analysis of a compendium of 321 high-quality gene expression profiles, which were previously published or newly generated in this study. We identified 84 groups of independently modulated genes (iModulons) that explain 75.7% of the total variance in the compendium. With these iModulons, we (i) expand our understanding of the regulatory functions of 39 iModulon associated TFs (e.g., HexR, Zur) by systematic comparison with 1993 previously reported TF-gene interactions; (ii) outline transcriptional changes after the transition from the exponential growth to stationary phases; (iii) capture group of genes required for utilizing diverse carbon sources and increased stationary response with slower growth rates; (iv) unveil multiple evolutionary strategies of transcriptome reallocation to achieve fast growth rates; and (v) define an osmotic stimulon, which includes the Type VI secretion system, as coordination of multiple iModulon activity changes. Taken together, this study provides the first quantitative genome-scale TRN for P. putida KT2440 and a basis for a comprehensive understanding of its complex transcriptome changes in a variety of physiological states.

Mathematical models to study the biology of pathogens and the infectious diseases they cause.

Mathematical models have many applications in infectious diseases: epidemiologists use them to forecast outbreaks and design containment strategies; systems biologists use them to study complex processes sustaining pathogens, from the metabolic networks empowering microbial cells to ecological networks in the microbiome that protects its host. Here, we (1) review important models relevant to infectious diseases, (2) draw parallels among models ranging widely in scale. We end by discussing a minimal set of information for a model to promote its use by others and to enable predictions that help us better fight pathogens and the diseases they cause.

Machine learning from Pseudomonas aeruginosa transcriptomes identifies independently modulated sets of genes associated with known transcriptional regulators.

The transcriptional regulatory network (TRN) of Pseudomonas aeruginosa coordinates cellular processes in response to stimuli. We used 364 transcriptomes (281 publicly available + 83 in-house generated) to reconstruct the TRN of P. aeruginosa using independent component analysis. We identified 104 independently modulated sets of genes (iModulons) among which 81 reflect the effects of known transcriptional regulators. We identified iModulons that (i) play an important role in defining the genomic boundaries of biosynthetic gene clusters (BGCs), (ii) show increased expression of the BGCs and associated secretion systems in nutrient conditions that are important in cystic fibrosis, (iii) show the presence of a novel ribosomally synthesized and post-translationally modified peptide (RiPP) BGC which might have a role in P. aeruginosa virulence, (iv) exhibit interplay of amino acid metabolism regulation and central metabolism across different carbon sources and (v) clustered according to their activity changes to define iron and sulfur stimulons. Finally, we compared the identified iModulons of P. aeruginosa with those previously described in Escherichia coli to observe conserved regulons across two Gram-negative species. This comprehensive TRN framework encompasses the majority of the transcriptional regulatory machinery in P. aeruginosa, and thus should prove foundational for future research into its physiological functions.

Optimal dimensionality selection for independent component analysis of transcriptomic data.

BACKGROUND: Independent component analysis is an unsupervised machine learning algorithm that separates a set of mixed signals into a set of statistically independent source signals. Applied to high-quality gene expression datasets, independent component analysis effectively reveals both the source signals of the transcriptome as co-regulated gene sets, and the activity levels of the underlying regulators across diverse experimental conditions. Two major variables that affect the final gene sets are the diversity of the expression profiles contained in the underlying data, and the user-defined number of independent components, or dimensionality, to compute. Availability of high-quality transcriptomic datasets has grown exponentially as high-throughput technologies have advanced; however, optimal dimensionality selection remains an open question. METHODS: We computed independent components across a range of dimensionalities for four gene expression datasets with varying dimensions (both in terms of number of genes and number of samples). We computed the correlation between independent components across different dimensionalities to understand how the overall structure evolves as the number of user-defined components increases. We then measured how well the resulting gene clusters reflected known regulatory mechanisms, and developed a set of metrics to assess the accuracy of the decomposition at a given dimension. RESULTS: We found that over-decomposition results in many independent components dominated by a single gene, whereas under-decomposition results in independent components that poorly capture the known regulatory structure. From these results, we developed a new method, called OptICA, for finding the optimal dimensionality that controls for both over- and under-decomposition. Specifically, OptICA selects the highest dimension that produces a low number of components that are dominated by a single gene. We show that OptICA outperforms two previously proposed methods for selecting the number of independent components across four transcriptomic databases of varying sizes. CONCLUSIONS: OptICA avoids both over-decomposition and under-decomposition of transcriptomic datasets resulting in the best representation of the organism's underlying transcriptional regulatory network.

Machine Learning Uncovers a Data-Driven Transcriptional Regulatory Network for the Crenarchaeal Thermoacidophile Sulfolobus acidocaldarius.

Dynamic cellular responses to environmental constraints are coordinated by the transcriptional regulatory network (TRN), which modulates gene expression. This network controls most fundamental cellular responses, including metabolism, motility, and stress responses. Here, we apply independent component analysis, an unsupervised machine learning approach, to 95 high-quality Sulfolobus acidocaldarius RNA-seq datasets and extract 45 independently modulated gene sets, or iModulons. Together, these iModulons contain 755 genes (32% of the genes identified on the genome) and explain over 70% of the variance in the expression compendium. We show that five modules represent the effects of known transcriptional regulators, and hypothesize that most of the remaining modules represent the effects of uncharacterized regulators. Further analysis of these gene sets results in: (1) the prediction of a DNA export system composed of five uncharacterized genes, (2) expansion of the LysM regulon, and (3) evidence for an as-yet-undiscovered global regulon. Our approach allows for a mechanistic, systems-level elucidation of an extremophile's responses to biological perturbations, which could inform research on gene-regulator interactions and facilitate regulator discovery in S. acidocaldarius. We also provide the first global TRN for S. acidocaldarius. Collectively, these results provide a roadmap toward regulatory network discovery in archaea.

Machine Learning of Bacterial Transcriptomes Reveals Responses Underlying Differential Antibiotic Susceptibility.

In vitro antibiotic susceptibility testing often fails to accurately predict in vivo drug efficacies, in part due to differences in the molecular composition between standardized bacteriologic media and physiological environments within the body. Here, we investigate the interrelationship between antibiotic susceptibility and medium composition in Escherichia coli K-12 MG1655 as contextualized through machine learning of transcriptomics data. Application of independent component analysis, a signal separation algorithm, shows that complex phenotypic changes induced by environmental conditions or antibiotic treatment are directly traced to the action of a few key transcriptional regulators, including RpoS, Fur, and Fnr. Integrating machine learning results with biochemical knowledge of transcription factor activation reveals medium-dependent shifts in respiration and iron availability that drive differential antibiotic susceptibility. By extension, the data generation and data analytics workflow used here can interrogate the regulatory state of a pathogen under any measured condition and can be applied to any strain or organism for which sufficient transcriptomics data are available. IMPORTANCE Antibiotic resistance is an imminent threat to global health. Patient treatment regimens are often selected based on results from standardized antibiotic susceptibility testing (AST) in the clinical microbiology lab, but these in vitro tests frequently misclassify drug effectiveness due to their poor resemblance to actual host conditions. Prior attempts to understand the combined effects of drugs and media on antibiotic efficacy have focused on physiological measurements but have not linked treatment outcomes to transcriptional responses on a systems level. Here, application of machine learning to transcriptomics data identified medium-dependent responses in key regulators of bacterial iron uptake and respiratory activity. The analytical workflow presented here is scalable to additional organisms and conditions and could be used to improve clinical AST by identifying the key regulatory factors dictating antibiotic susceptibility.

Restoration of fitness lost due to dysregulation of the pyruvate dehydrogenase complex is triggered by ribosomal binding site modifications.

Pyruvate dehydrogenase complex (PDC) functions as the main determinant of the respiro-fermentative balance because it converts pyruvate to acetyl-coenzyme A (CoA), which then enters the TCA (tricarboxylic acid cycle). PDC is repressed by the pyruvate dehydrogenase complex regulator (PdhR) in Escherichia coli. The deletion of the pdhR gene compromises fitness in aerobic environments. We evolve the E. coli pdhR deletion strain to examine its achievable growth rate and the underlying adaptive strategies. We find that (1) optimal proteome allocation to PDC is critical in achieving optimal growth rate; (2) expression of PDC in evolved strains is reduced through mutations in the Shine-Dalgarno sequence; (3) rewiring of the TCA flux and increased reactive oxygen species (ROS) defense occur in the evolved strains; and (4) the evolved strains adapt to an efficient biomass yield. Together, these results show how adaptation can find alternative regulatory mechanisms for a key cellular process if the primary regulatory mode fails.

Independent component analysis recovers consistent regulatory signals from disparate datasets.

The availability of bacterial transcriptomes has dramatically increased in recent years. This data deluge could result in detailed inference of underlying regulatory networks, but the diversity of experimental platforms and protocols introduces critical biases that could hinder scalable analysis of existing data. Here, we show that the underlying structure of the E. coli transcriptome, as determined by Independent Component Analysis (ICA), is conserved across multiple independent datasets, including both RNA-seq and microarray datasets. We subsequently combined five transcriptomics datasets into a large compendium containing over 800 expression profiles and discovered that its underlying ICA-based structure was still comparable to that of the individual datasets. With this understanding, we expanded our analysis to over 3,000 E. coli expression profiles and predicted three high-impact regulons that respond to oxidative stress, anaerobiosis, and antibiotic treatment. ICA thus enables deep analysis of disparate data to uncover new insights that were not visible in the individual datasets.

iModulonDB: a knowledgebase of microbial transcriptional regulation derived from machine learning.

Independent component analysis (ICA) of bacterial transcriptomes has emerged as a powerful tool for obtaining co-regulated, independently-modulated gene sets (iModulons), inferring their activities across a range of conditions, and enabling their association to known genetic regulators. By grouping and analyzing genes based on observations from big data alone, iModulons can provide a novel perspective into how the composition of the transcriptome adapts to environmental conditions. Here, we present iModulonDB (imodulondb.org), a knowledgebase of prokaryotic transcriptional regulation computed from high-quality transcriptomic datasets using ICA. Users select an organism from the home page and then search or browse the curated iModulons that make up its transcriptome. Each iModulon and gene has its own interactive dashboard, featuring plots and tables with clickable, hoverable, and downloadable features. This site enhances research by presenting scientists of all backgrounds with co-expressed gene sets and their activity levels, which lead to improved understanding of regulator-gene relationships, discovery of transcription factors, and the elucidation of unexpected relationships between conditions and genetic regulatory activity. The current release of iModulonDB covers three organisms (Escherichia coli, Staphylococcus aureus and Bacillus subtilis) with 204 iModulons, and can be expanded to cover many additional organisms.

Machine learning uncovers independently regulated modules in the Bacillus subtilis transcriptome.

The transcriptional regulatory network (TRN) of Bacillus subtilis coordinates cellular functions of fundamental interest, including metabolism, biofilm formation, and sporulation. Here, we use unsupervised machine learning to modularize the transcriptome and quantitatively describe regulatory activity under diverse conditions, creating an unbiased summary of gene expression. We obtain 83 independently modulated gene sets that explain most of the variance in expression and demonstrate that 76% of them represent the effects of known regulators. The TRN structure and its condition-dependent activity uncover putative or recently discovered roles for at least five regulons, such as a relationship between histidine utilization and quorum sensing. The TRN also facilitates quantification of population-level sporulation states. As this TRN covers the majority of the transcriptome and concisely characterizes the global expression state, it could inform research on nearly every aspect of transcriptional regulation in B. subtilis.