A Pseudomonas aeruginosa small RNA regulates chronic and acute infection.

The ability to switch between different lifestyles allows bacterial pathogens to thrive in diverse ecological niches1,2. However, a molecular understanding of their lifestyle changes within the human host is lacking. Here, by directly examining bacterial gene expression in human-derived samples, we discover a gene that orchestrates the transition between chronic and acute infection in the opportunistic pathogen Pseudomonas aeruginosa. The expression level of this gene, here named sicX, is the highest of the P. aeruginosa genes expressed in human chronic wound and cystic fibrosis infections, but it is expressed at extremely low levels during standard laboratory growth. We show that sicX encodes a small RNA that is strongly induced by low-oxygen conditions and post-transcriptionally regulates anaerobic ubiquinone biosynthesis. Deletion of sicX causes P. aeruginosa to switch from a chronic to an acute lifestyle in multiple mammalian models of infection. Notably, sicX is also a biomarker for this chronic-to-acute transition, as it is the most downregulated gene when a chronic infection is dispersed to cause acute septicaemia. This work solves a decades-old question regarding the molecular basis underlying the chronic-to-acute switch in P. aeruginosa and suggests oxygen as a primary environmental driver of acute lethality.

Convergent evolution in toxin detection and resistance provides evidence for conserved bacterial-fungal interactions.

Microbes rarely exist in isolation and instead form complex polymicrobial communities. As a result, microbes have developed intricate offensive and defensive strategies that enhance their fitness in these complex communities. Thus, identifying and understanding the molecular mechanisms controlling polymicrobial interactions is critical for understanding the function of microbial communities. In this study, we show that the gram-negative opportunistic human pathogen Pseudomonas aeruginosa, which frequently causes infection alongside a plethora of other microbes including fungi, encodes a genetic network which can detect and defend against gliotoxin, a potent, disulfide-containing antimicrobial produced by the ubiquitous filamentous fungus Aspergillus fumigatus. We show that gliotoxin exposure disrupts P. aeruginosa zinc homeostasis, leading to transcriptional activation of a gene encoding a previously uncharacterized dithiol oxidase (herein named as DnoP), which detoxifies gliotoxin and structurally related toxins. Despite sharing little homology to the A. fumigatus gliotoxin resistance protein (GliT), the enzymatic mechanism of DnoP from P. aeruginosa appears to be identical that used by A. fumigatus. Thus, DnoP and its transcriptional induction by low zinc represent a rare example of both convergent evolution of toxin defense and environmental cue sensing across kingdoms. Collectively, these data provide compelling evidence that P. aeruginosa has evolved to survive exposure to an A. fumigatus disulfide-containing toxin in the natural environment.

Improvement of a mouse infection model to capture Pseudomonas aeruginosa chronic physiology in cystic fibrosis.

Laboratory models are central to microbiology research, advancing the understanding of bacterial physiology by mimicking natural environments, from soil to the human microbiome. When studying host-bacteria interactions, animal models enable investigators to examine bacterial dynamics associated with a host, and in the case of human infections, animal models are necessary to translate basic research into clinical treatments. Efforts toward improving animal infection models are typically based on reproducing host genotypes/phenotypes and disease manifestations, leaving a gap in how well the physiology of microbes reflects their behavior in a human host. Understanding bacterial physiology is vital because it dictates host response and bacterial interactions with antimicrobials. Thus, our goal was to develop an animal model that accurately recapitulates bacterial physiology in human infection. The system we chose to model was a chronic Pseudomonas aeruginosa respiratory infection in cystic fibrosis (CF). To accomplish this goal, we leveraged a framework that we recently developed to evaluate model accuracy by calculating the percentage of bacterial genes that are expressed similarly in a model to how they are expressed in their infection environment. We combined two complementary models of P. aeruginosa infection-an in vitro synthetic CF sputum model (SCFM2) and a mouse acute pneumonia model. This combined model captured the chronic physiology of P. aeruginosa in CF better than the standard mouse infection model, showing the power of a data-driven approach to refining animal models. In addition, the results of this work challenge the assumption that a chronic infection model requires long-term colonization.

Microbial communication superhighways.

Exchange of information is critical for bacterial social behaviors. Now Dubey and Ben-Yehuda (2011) provide evidence for bacterial "nanotube" conduits that allow microbes to directly exchange cytoplasmic factors. Protein and DNA transfer between distantly related species raises the prospect of a new, widely distributed mechanism of bacterial communication.

Application of a quantitative framework to improve the accuracy of a bacterial infection model.

Laboratory models are critical to basic and translational microbiology research. Models serve multiple purposes, from providing tractable systems to study cell biology to allowing the investigation of inaccessible clinical and environmental ecosystems. Although there is a recognized need for improved model systems, there is a gap in rational approaches to accomplish this goal. We recently developed a framework for assessing the accuracy of microbial models by quantifying how closely each gene is expressed in the natural environment and in various models. The accuracy of the model is defined as the percentage of genes that are similarly expressed in the natural environment and the model. Here, we leverage this framework to develop and validate two generalizable approaches for improving model accuracy, and as proof of concept, we apply these approaches to improve models of Pseudomonas aeruginosa infecting the cystic fibrosis (CF) lung. First, we identify two models, an in vitro synthetic CF sputum medium model (SCFM2) and an epithelial cell model, that accurately recapitulate different gene sets. By combining these models, we developed the epithelial cell-SCFM2 model which improves the accuracy of over 500 genes. Second, to improve the accuracy of specific genes, we mined publicly available transcriptome data, which identified zinc limitation as a cue present in the CF lung and absent in SCFM2. Induction of zinc limitation in SCFM2 resulted in accurate expression of 90% of P. aeruginosa genes. These approaches provide generalizable, quantitative frameworks for microbiological model improvement that can be applied to any system of interest.

Bacterial Quorum Sensing During Infection.

Bacteria are highly interactive and possess an extraordinary repertoire of intercellular communication and social behaviors, including quorum sensing (QS). QS has been studied in detail at the molecular level, so mechanistic details are well understood in many species and are often involved in virulence. The use of different animal host models has demonstrated QS-dependent control of virulence determinants and virulence in several human pathogenic bacteria. QS also controls virulence in several plant pathogenic species. Despite the role QS plays in virulence during animal and plant laboratory-engineered infections, QS mutants are frequently isolated from natural infections, demonstrating that the function of QS during infection and its role in pathogenesis remain poorly understood and are fruitful areas for future research. We discuss the role of QS during infection in various organisms and highlight approaches to better understand QS during human infection. This is an important consideration in an era of growing antimicrobial resistance, when we are looking for new ways to target bacterial infections.

A quantitative framework reveals traditional laboratory growth is a highly accurate model of human oral infection.

Bacterial behavior and virulence during human infection is difficult to study and largely unknown, as our vast knowledge of infection microbiology is primarily derived from studies using in vitro and animal models. Here, we characterize the physiology of Porphyromonas gingivalis, a periodontal pathogen, in its native environment using 93 published metatranscriptomic datasets from periodontally healthy and diseased individuals. P. gingivalis transcripts were more abundant in samples from periodontally diseased patients but only above 0.1% relative abundance in one-third of diseased samples. During human infection, P. gingivalis highly expressed genes encoding virulence factors such as fimbriae and gingipains (proteases) and genes involved in growth and metabolism, indicating that P. gingivalis is actively growing during disease. A quantitative framework for assessing the accuracy of model systems showed that 96% of P. gingivalis genes were expressed similarly in periodontitis and in vitro midlogarithmic growth, while significantly fewer genes were expressed similarly in periodontitis and in vitro stationary phase cultures (72%) or in a murine abscess infection model (85%). This high conservation in gene expression between periodontitis and logarithmic laboratory growth is driven by overall low variance in P. gingivalis gene expression, relative to other pathogens including Pseudomonas aeruginosa and Staphylococcus aureus Together, this study presents strong evidence for the use of simple test tube growth as the gold standard model for studying P. gingivalis biology, providing biological relevance for the thousands of laboratory experiments performed with logarithmic phase P. gingivalis Furthermore, this work highlights the need to quantitatively assess the accuracy of model systems.

Precise spatial structure impacts antimicrobial susceptibility of S. aureus in polymicrobial wound infections.

A hallmark of microbial ecology is that interactions between members of a community shape community function. This includes microbial communities in human infections, such as chronic wounds, where interactions can result in more severe diseases. Staphylococcus aureus is the most common organism isolated from human chronic wound infections and has been shown to have both cooperative and competitive interactions with Pseudomonas aeruginosa. Still, despite considerable study, most interactions between these microbes have been characterized using in vitro well-mixed systems, which do not recapitulate the infection environment. Here, we characterized interactions between S. aureus and P. aeruginosa in chronic murine wounds, focusing on the role that both macro- and micro-scale spatial structures play in disease. We discovered that S. aureus and P. aeruginosa coexist at high cell densities in murine wounds. High-resolution imaging revealed that these microbes establish a patchy distribution, only occupying 5 to 25% of the wound volume. Using a quantitative framework, we identified a precise spatial structure at both the macro (mm)- and micro (µm)-scales, which was largely mediated by P. aeruginosa production of the antimicrobial 2-heptyl-4-hydroxyquinoline N-oxide, while the antimicrobial pyocyanin had no impact. Finally, we discovered that this precise spatial structure enhances S. aureus tolerance to aminoglycoside antibiotics but not vancomycin. Our results provide mechanistic insights into the biogeography of S. aureus and P. aeruginosa coinfected wounds and implicate spatial structure as a key determinant of antimicrobial tolerance in wound infections.

Metabolic diversity of human macrophages: potential influence on Staphylococcus aureus intracellular survival.

Staphylococcus aureus is a leading cause of medical device-associated biofilm infections. This is influenced by the ability of S. aureus biofilm to evade the host immune response, which is partially driven by the anti-inflammatory cytokine interleukin-10 (IL-10). Here, we show that treatment of human monocyte-derived macrophages (HMDMs) with IL-10 enhanced biofilm formation, suggesting that macrophage anti-inflammatory programming likely plays an important role during the transition from planktonic to biofilm growth. To identify S. aureus genes that were important for intracellular survival in HMDMs and how this was affected by IL-10, transposon sequencing was performed. The size of the S. aureus essential genome was similar between unstimulated HMDMs and the outgrowth control (18.5% vs 18.4%, respectively, with 54.4% overlap) but increased to 22.5% in IL-10-treated macrophages, suggesting that macrophage polarization status exerts differential pressure on S. aureus. Essential genes for S. aureus survival within IL-10-polarized HMDMs were dominated by negative regulatory pathways, including nitrogen and RNA metabolism, whereas S. aureus essential genes within untreated HMDMs were enriched in biosynthetic pathways such as purine and pyrimidine biosynthesis. To explore how IL-10 altered the macrophage intracellular metabolome, targeted metabolomics was performed on HMDMs from six individual donors. IL-10 treatment led to conserved alterations in distinct metabolites that were increased (dihydroxyacetone phosphate, glyceraldehyde-3-phosphate, and acetyl-CoA) or reduced (fructose-6-phosphate, aspartic acid, and ornithine) across donors, whereas other metabolites were variable. Collectively, these findings highlight an important aspect of population-level heterogeneity in human macrophage responsiveness that should be considered when translating results to a patient population.IMPORTANCEOne mechanism that Staphylococcus aureus biofilm elicits in the host to facilitate infection persistence is the production of the anti-inflammatory cytokine interleukin-10 (IL-10). Here, we show that exposure of human monocyte-derived macrophages (HMDMs) to IL-10 promotes S. aureus biofilm formation and programs intracellular bacteria to favor catabolic pathways. Examination of intracellular metabolites in HMDMs revealed heterogeneity between donors that may explain the observed variability in essential genes for S. aureus survival based on nutrient availability for bacteria within the intracellular compartment. Collectively, these studies provide novel insights into how IL-10 polarization affects S. aureus intracellular survival in HMDMs and the importance of considering macrophage heterogeneity between human donors as a variable when examining effector mechanisms.

Corrigendum: Progress in and promise of bacterial quorum sensing research.

This corrects the article DOI: 10.1038/nature24624.

Emerging strategies to target virulence in Pseudomonas aeruginosa respiratory infections.

Pseudomonas aeruginosa is an opportunistic pathogen that is responsible for infections in people living with chronic respiratory conditions, such as cystic fibrosis (CF) and non-CF bronchiectasis (NCFB). Traditionally, in people with chronic respiratory disorders, P. aeruginosa infection has been managed with a combination of inhaled and intravenous antibiotic therapies. However, due in part to the prolonged use of antibiotics in these people, the emergence of multi-drug resistant P. aeruginosa strains is a growing concern. The development of anti-virulence therapeutics may provide a new means of treating P. aeruginosa lung infections whilst also combatting the AMR crisis, as these agents are presumed to exert reduced pressure for the emergence of drug resistance as compared to antibiotics. However, the pipeline for developing anti-virulence therapeutics is poorly defined, and it is currently unclear as to whether in vivo and in vitro models effectively replicate the complex pulmonary environment sufficiently to enable development and testing of such therapies for future clinical use. Here, we discuss potential targets for P. aeruginosa anti-virulence therapeutics and the effectiveness of the current models used to study them. Focus is given to the difficulty of replicating the virulence gene expression patterns of P. aeruginosa in the CF and NCFB lung under laboratory conditions and to the challenges this poses for anti-virulence therapeutic development.

Progress in and promise of bacterial quorum sensing research.

This Review highlights how we can build upon the relatively new and rapidly developing field of research into bacterial quorum sensing (QS). We now have a depth of knowledge about how bacteria use QS signals to communicate with each other and to coordinate their activities. In recent years there have been extraordinary advances in our understanding of the genetics, genomics, biochemistry, and signal diversity of QS. We are beginning to understand the connections between QS and bacterial sociality. This foundation places us at the beginning of a new era in which researchers will be able to work towards new medicines to treat devastating infectious diseases, and use bacteria to understand the biology of sociality.

Spatial mapping of polymicrobial communities reveals a precise biogeography associated with human dental caries.

Tooth decay (dental caries) is a widespread human disease caused by microbial biofilms. Streptococcus mutans, a biofilm-former, has been consistently associated with severe childhood caries; however, how this bacterium is spatially organized with other microorganisms in the oral cavity to promote disease remains unknown. Using intact biofilms formed on teeth of toddlers affected by caries, we discovered a unique 3D rotund-shaped architecture composed of multiple species precisely arranged in a corona-like structure with an inner core of S. mutans encompassed by outer layers of other bacteria. This architecture creates localized regions of acidic pH and acute enamel demineralization (caries) in a mixed-species biofilm model on human teeth, suggesting this highly ordered community as the causative agent. Notably, the construction of this architecture was found to be an active process initiated by production of an extracellular scaffold by S. mutans that assembles the corona cell arrangement, encapsulating the pathogen core. In addition, this spatial patterning creates a protective barrier against antimicrobials while increasing bacterial acid fitness associated with the disease-causing state. Our data reveal a precise biogeography in a polymicrobial community associated with human caries that can modulate the pathogen positioning and virulence potential in situ, indicating that micron-scale spatial structure of the microbiome may mediate the function and outcome of host-pathogen interactions.

Porphyromonas gingivalis Tyrosine Kinase Is a Fitness Determinant in Polymicrobial Infections.

Many pathogenic microbial ecosystems are polymicrobial, and community function can be shaped by interbacterial interactions. Little is known, however, regarding the genetic determinants required for fitness in heterotypic community environments. In periodontal diseases, Porphyromonas gingivalis is a primary pathogen, but only within polymicrobial communities. Here, we used a transposon sequencing (Tn-Seq) library of P. gingivalis to screen for genes that influence fitness of the organism in a coinfection murine abscess model with the oral partner species Streptococcus gordonii and Fusobacterium nucleatum. Genes impacting fitness with either organism were involved in diverse processes, including metabolism and energy production, along with cell wall and membrane biogenesis. Despite the overall similarity of function, the majority of identified genes were specific to the partner species, indicating that synergistic mechanisms of P. gingivalis vary to a large extent according to community composition. Only two genes were identified as essential for P. gingivalis fitness in abscess development with both S. gordonii and F. nucleatum: ptk1, encoding a tyrosine kinase, and inlJ, encoding an internalin family surface protein. Ptk1, but not InlJ, is required for community development with S. gordonii, and we found that the action of this kinase is similarly required for P. gingivalis to accumulate in a community with F. nucleatum. A limited number of P. gingivalis genes are therefore required for species-independent synergy, and the Ptk1 tyrosine kinase network may integrate and coordinate input from multiple organisms.

Bacterial biofilms predominate in both acute and chronic human lung infections.

BACKGROUND: A basic paradigm of human infection is that acute bacterial disease is caused by fast growing planktonic bacteria while chronic infections are caused by slow-growing, aggregated bacteria, a phenomenon known as a biofilm. For lung infections, this paradigm has been thought to be supported by observations of how bacteria proliferate in well-established growth media in the laboratory-the gold standard of microbiology. OBJECTIVE: To investigate the bacterial architecture in sputum from patients with acute and chronic lung infections. METHODS: Advanced imaging technology was used for quantification and direct comparison of infection types on fresh sputum samples, thereby directly testing the acute versus chronic paradigm. RESULTS: In this study, we compared the bacterial lifestyle (planktonic or biofilm), growth rate and inflammatory response of bacteria in freshly collected sputum (n=43) from patient groups presenting with acute or chronic lung infections. We found that both acute and chronic lung infections are dominated by biofilms (aggregates of bacteria within an extracellular matrix), although planktonic cells were observed in both sample types. Bacteria grew faster in sputum from acute infections, but these fast-growing bacteria were enriched in biofilms similar to the architecture thought to be reserved for chronic infections. Cellular inflammation in the lungs was also similar across patient groups, but systemic inflammatory markers were only elevated in acute infections. CONCLUSIONS: Our findings indicate that the current paradigm of equating planktonic with acute and biofilm with chronic infection needs to be revisited as the difference lies primarily in metabolic rates, not bacterial architecture.

Development of a Versatile, Low-Cost Electrochemical System to Study Biofilm Redox Activity at the Micron Scale.

Spatially resolving chemical landscapes surrounding microbial communities can provide insight into chemical interactions that dictate cellular physiology. Electrochemical techniques provide an attractive option for studying these interactions due to their robustness and high sensitivity. Unfortunately, commercial electrochemical platforms that are capable of measuring chemical activity on the micron scale are often expensive and do not easily perform multiple scanning techniques. Here, we report development of an inexpensive electrochemical system that features a combined micromanipulator and potentiostat component capable of scanning surfaces while measuring molecular concentrations or redox profiles. We validate this experimental platform for biological use with a two-species biofilm model composed of the oral bacterial pathogen Aggregatibacter actinomycetemcomitans and the oral commensal Streptococcus gordonii. We measure consumption of H2O2 by A. actinomycetemcomitans biofilms temporally and spatially, providing new insights into how A. actinomycetemcomitans responds to this S. gordonii-produced metabolite. We advance our platform to spatially measure redox activity above biofilms. Our analysis supports that redox activity surrounding biofilms is species specific, and the region immediately above an S. gordonii biofilm is highly oxidized compared to that above an A. actinomycetemcomitans biofilm. This work provides description and validation of a versatile, quantitative framework for studying bacterial redox-mediated physiology in an integrated and easily adaptable experimental platform. IMPORTANCE Scanning electrochemical probe microscopy methods can provide information of the chemical environment along a spatial surface with micron-scale resolution. These methods often require expensive instruments that perform optimized and highly sensitive niche techniques. Here, we describe a novel system that combines a micromanipulator that scans micron-sized electrodes across the surface of bacterial biofilms and a potentiostat, which performs various electrochemical techniques. This platform allows for spatial measurement of chemical gradients above live bacteria in real time, and as proof of concept, we utilize this setup to map H2O2 detoxification above an oral pathogen biofilm. We increased the versatility of this platform further by mapping redox potentials of biofilms in real time on the micron scale. Together, this system provides a technical framework for studying chemical interactions among microbes.

Large-scale identification of pathogen essential genes during coinfection with sympatric and allopatric microbes.

Recent evidence suggests that the genes an organism needs to survive in an environment drastically differ when alone or in a community. However, it is not known if there are universal functions that enable microbes to persist in a community and if there are functions specific to interactions between microbes native to the same (sympatric) or different (allopatric) environments. Here, we ask how the essential functions of the oral pathogen Aggregatibacter actinomycetemcomitans change during pairwise coinfection in a murine abscess with each of 15 microbes commonly found in the oral cavity and 10 microbes that are not. A. actinomycetemcomitans was more abundant when coinfected with allopatric than with sympatric microbes, and this increased fitness correlated with expanded metabolic capacity of the coinfecting microbes. Using transposon sequencing, we discovered that 33% of the A. actinomycetemcomitans genome is required for coinfection fitness. Fifty-nine "core" genes were required across all coinfections and included genes necessary for aerobic respiration. The core genes were also all required in monoinfection, indicating the essentiality of these genes cannot be alleviated by a coinfecting microbe. Furthermore, coinfection with some microbes, predominately sympatric species, induced the requirement for over 100 new community-dependent essential genes. In contrast, in other coinfections, predominately with nonoral species, A. actinomycetemcomitans required 50 fewer genes than in monoinfection, demonstrating that some allopatric microbes can drastically alleviate gene essentialities. These results expand our understanding of how diverse microbes alter growth and gene essentiality within polymicrobial infections.

Streptococcus gordonii programs epithelial cells to resist ZEB2 induction by Porphyromonas gingivalis.

The polymicrobial microbiome of the oral cavity is a direct precursor of periodontal diseases, and changes in microhabitat or shifts in microbial composition may also be linked to oral squamous cell carcinoma. Dysbiotic oral epithelial responses provoked by individual organisms, and which underlie these diseases, are widely studied. However, organisms may influence community partner species through manipulation of epithelial cell responses, an aspect of the host microbiome interaction that is poorly understood. We report here that Porphyromonas gingivalis, a keystone periodontal pathogen, can up-regulate expression of ZEB2, a transcription factor which controls epithelial-mesenchymal transition and inflammatory responses. ZEB2 regulation by P. gingivalis was mediated through pathways involving ß-catenin and FOXO1. Among the community partners of P. gingivalis, Streptococcus gordonii was capable of antagonizing ZEB2 expression. Mechanistically, S. gordonii suppressed FOXO1 by activating the TAK1-NLK negative regulatory pathway, even in the presence of P. gingivalis Collectively, these results establish S. gordonii as homeostatic commensal, capable of mitigating the activity of a more pathogenic organism through modulation of host signaling.

Spatial determinants of quorum signaling in a Pseudomonas aeruginosa infection model.

Quorum sensing (QS) is a bacterial communication system that involves production and sensing of extracellular signals. In laboratory models, QS allows bacteria to monitor and respond to their own cell density and is critical for fitness. However, how QS proceeds in natural, spatially structured bacterial communities is not well understood, which significantly hampers our understanding of the emergent properties of natural communities. To address this gap, we assessed QS signaling in the opportunistic pathogen Pseudomonas aeruginosa in a cystic fibrosis (CF) lung infection model that recapitulates the biogeographical aspects of the natural human infection. In this model, P. aeruginosa grows as spatially organized, highly dense aggregates similar to those observed in the human CF lung. By combining this natural aggregate system with a micro-3D-printing platform that allows for confinement and precise spatial positioning of P. aeruginosa aggregates, we assessed the impact of aggregate size and spatial positioning on both intra- and interaggregate signaling. We discovered that aggregates containing ∼2,000 signal-producing P. aeruginosa were unable to signal neighboring aggregates, while those containing ≥5,000 cells signaled aggregates as far away as 176 µm. Not all aggregates within this "calling distance" responded, indicating that aggregates have differential sensitivities to signal. Overexpression of the signal receptor increased aggregate sensitivity to signal, suggesting that the ability of aggregates to respond is defined in part by receptor levels. These studies provide quantitative benchmark data for the impact of spatial arrangement and phenotypic heterogeneity on P. aeruginosa signaling in vivo.

Pseudomonas aeruginosa transcriptome during human infection.

Laboratory experiments have uncovered many basic aspects of bacterial physiology and behavior. After the past century of mostly in vitro experiments, we now have detailed knowledge of bacterial behavior in standard laboratory conditions, but only a superficial understanding of bacterial functions and behaviors during human infection. It is well-known that the growth and behavior of bacteria are largely dictated by their environment, but how bacterial physiology differs in laboratory models compared with human infections is not known. To address this question, we compared the transcriptome of Pseudomonas aeruginosa during human infection to that of P. aeruginosa in a variety of laboratory conditions. Several pathways, including the bacterium's primary quorum sensing system, had significantly lower expression in human infections than in many laboratory conditions. On the other hand, multiple genes known to confer antibiotic resistance had substantially higher expression in human infection than in laboratory conditions, potentially explaining why antibiotic resistance assays in the clinical laboratory frequently underestimate resistance in patients. Using a standard machine learning technique known as support vector machines, we identified a set of genes whose expression reliably distinguished in vitro conditions from human infections. Finally, we used these support vector machines with binary classification to force P. aeruginosa mouse infection transcriptomes to be classified as human or in vitro. Determining what differentiates our current models from clinical infections is important to better understand bacterial infections and will be necessary to create model systems that more accurately capture the biology of infection.