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

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.

Iron Deficiency Modulates Metabolic Landscape of Bacteroidetes Promoting Its Resilience during Inflammation

Bacteria have to persist under low iron conditions in order to adapt to the nutritional immunity of a host. Since the knowledge of iron stimulon of Bacteroidetes is sparse, we examined oral (Porphyromonas gingivalis and Prevotella intermedia) and gut (Bacteroides thataiotaomicron) representatives for their ability to adapt to iron deplete and iron replete conditions. Our transcriptomics and comparative genomics analysis show that many iron-regulated mechanisms are conserved within the phylum. They include genes upregulated in low iron, as follows: fldA (flavodoxin), hmu (hemin uptake operon), and loci encoding ABC transporters. Downregulated genes were frd (ferredoxin), rbr (rubrerythrin), sdh (succinate dehydrogenase/fumarate reductase), vor (oxoglutarate oxidoreductase/dehydrogenase), and pfor (pyruvate:ferredoxin/flavodoxin oxidoreductase). Some genus-specific mechanisms, such as the sus of B. thetaiotaomicron coding for carbohydrate metabolism and the xusABC coding for xenosiderophore utilization were also identified. While all bacteria tested in our study had the nrfAH operon coding for nitrite reduction and were able to reduce nitrite levels present in culture media, the expression of the operon was iron dependent only in B. thetaiotaomicron. It is noteworthy that we identified a significant overlap between regulated genes found in our study and the B. thetaiotaomicron colitis study (W. Zhu, M. G. Winter, L. Spiga, E. R. Hughes et al., Cell Host Microbe 27:376-388, 2020, http://dx.doi.org/10.1016/j.chom.2020.01.010). Many of those commonly regulated genes were also iron regulated in the oral bacterial genera. Overall, this work points to iron being the master regulator enabling bacterial persistence in the host and paves the way for a more generalized investigation of the molecular mechanisms of iron homeostasis in Bacteroidetes. IMPORTANCE Bacteroidetes are an important group of anaerobic bacteria abundant both in the oral and gut microbiomes. Although iron is a required nutrient for most living organisms, the molecular mechanisms of adaptation to the changing levels of iron are not well known in this group of bacteria. We defined the iron stimulon of Bacteroidetes by examination of the transcriptomic response of Porphyromonas gingivalis and Prevotella intermedia (both belong to the oral microbiome) and Bacteroidetes thetaiotaomicron (belongs to the gut microbiome). Our results indicate that many of the iron-regulated operons are shared among the three genera. Furthermore, using bioinformatics analysis, we identified a significant overlap between our in vitro studies and transcriptomic data derived from a colitis study, thus underscoring the biological significance of our work. Defining the iron-dependent stimulon of Bacteroidetes can help to identify the molecular mechanisms of iron-dependent regulation as well as better understand the persistence of the anaerobes in the human host.

BacterAI maps microbial metabolism without prior knowledge

Training artificial intelligence (AI) systems to perform autonomous experiments would vastly increase the throughput of microbiology; however, few microbes have large enough datasets for training such a system. In the present study, we introduce BacterAI, an automated science platform that maps microbial metabolism but requires no prior knowledge. BacterAI learns by converting scientific questions into simple games that it plays with laboratory robots. The agent then distils its findings into logical rules that can be interpreted by human scientists. We use BacterAI to learn the amino acid requirements for two oral streptococci: Streptococcus gordonii and Streptococcus sanguinis. We then show how transfer learning can accelerate BacterAI when investigating new environments or larger media with up to 39 ingredients. Scientific gameplay and BacterAI enable the unbiased, autonomous study of organisms for which no training data exist.

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.

Impact of Growth Rate on the Protein-mRNA Ratio in Pseudomonas aeruginosa

Our understanding of how bacterial pathogens colonize and persist during human infection has been hampered by the limited characterization of bacterial physiology during infection and a research bias toward in vitro, fast-growing bacteria. Recent research has begun to address these gaps in knowledge by directly quantifying bacterial mRNA levels during human infection, with the goal of assessing microbial community function at the infection site. However, mRNA levels are not always predictive of protein levels, which are the primary functional units of a cell. Here, we used carefully controlled chemostat experiments to examine the relationship between mRNA and protein levels across four growth rates in the bacterial pathogen Pseudomonas aeruginosa. We found a genome-wide positive correlation between mRNA and protein abundances across all growth rates, with genes required for P. aeruginosa viability having stronger correlations than nonessential genes. We developed a statistical method to identify genes whose mRNA abundances poorly predict protein abundances and calculated an RNA-to-protein (RTP) conversion factor to improve mRNA predictions of protein levels. The application of the RTP conversion factor to publicly available transcriptome data sets was highly robust, enabling the more accurate prediction of P. aeruginosa protein levels across strains and growth conditions. Finally, the RTP conversion factor was applied to P. aeruginosa human cystic fibrosis (CF) infection transcriptomes to provide greater insights into the functionality of this bacterium in the CF lung. This study addresses a critical problem in infection microbiology by providing a framework for enhancing the functional interpretation of bacterial human infection transcriptome data. IMPORTANCE Our understanding of bacterial physiology during human infection is limited by the difficulty in assessing bacterial function at the infection site. Recent studies have begun to address this question by quantifying bacterial mRNA levels in human-derived samples using transcriptomics. One challenge for these studies is the poor predictivity of mRNA for protein levels for some genes. Here, we addressed this challenge by measuring the transcriptomes and proteomes of P. aeruginosa grown at four growth rates. Our results revealed that the growth rate does not impact the genome-wide correlation of mRNA and protein levels. We used statistical methods to identify the genes for which mRNA and protein were poorly correlated and developed an RNA-to-protein (RTP) conversion factor that improved the predictivity of protein levels across strains and growth conditions. Our results provide new insights into mRNA-protein correlations and tools to enhance our understanding of bacterial physiology from transcriptome data.

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.

Polymicrobial Interactions of Oral Microbiota: a Historical Review and Current Perspective

The oral microbiota is enormously diverse, with over 700 microbial species identified across individuals that play a vital role in the health of our mouth and our overall well-being. In addition, as oral diseases such as caries (cavities) and periodontitis (gum disease) are mediated through interspecies microbial interactions, this community serves as an important model system to study the complexity and dynamics of polymicrobial interactions. Here, we review historical and recent progress in our understanding of the oral microbiome, highlighting how oral microbiome research has significantly contributed to our understanding of microbial communities, with broad implications in polymicrobial diseases and across microbial community ecology. Further, we explore innovations and challenges associated with analyzing polymicrobial systems and suggest future directions of study. Finally, we provide a conceptual framework to systematically study microbial interactions within complex communities, not limited to the oral microbiota.