Machine learning analysis of microbial flow cytometry data from nanoparticles, antibiotics and carbon sources perturbed anaerobic microbiomes.
ABSTRACT: Flow cytometry, with its high throughput nature, combined with the ability to measure an increasing number of cell parameters at once can surpass the throughput of prevalent genomic and metagenomic approaches in the study of microbiomes. Novel computational approaches to analyze flow cytometry data will result in greater insights and actionability as compared to traditional tools used in the analysis of microbiomes. This paper is a demonstration of the fruitfulness of machine learning in analyzing microbial flow cytometry data generated in anaerobic microbiome perturbation experiments.Autoencoders were found to be powerful in detecting anomalies in flow cytometry data from nanoparticles and carbon sources perturbed anaerobic microbiomes but was marginal in predicting perturbations due to antibiotics. A comparison between different algorithms based on predictive capabilities suggested that gradient boosting (GB) and deep learning, i.e. feed forward artificial neural network with three hidden layers (DL) were marginally better under tested conditions at predicting overall community structure while distributed random forests (DRF) worked better for predicting the most important putative microbial group(s) in the anaerobic digesters viz. methanogens, and it can be optimized with better parameter tuning. Predictive classification patterns with DL (feed forward artificial neural network with three hidden layers) were found to be comparable to previously demonstrated multivariate analysis. The potential applications of this approach have been demonstrated for monitoring the syntrophic resilience of the anaerobic microbiomes perturbed by synthetic nanoparticles as well as antibiotics.Machine learning can benefit the microbial flow cytometry research community by providing rapid screening and characterization tools to discover patterns in the dynamic response of microbiomes to several stimuli.
Project description:Generating chemical energy carriers and bulk chemicals from solar energy by microbial metabolic capacities is a promising technology. In this long-term study of over 500 days, methane was produced by a microbial community that was fed by the mono-substrate glycolate, which was derived from engineered algae. The microbial community structure was measured on the single cell level using flow cytometry. Abiotic and operational reactor parameters were analyzed in parallel. The R-based tool flowCyBar facilitated visualization of community dynamics and indicated sub-communities involved in glycolate fermentation and methanogenesis. Cell sorting and amplicon sequencing of 16S rRNA and mcrA genes were used to identify the key organisms involved in the anaerobic conversion process. The microbial community allowed a constant fermentation, although it was sensitive to high glycolate concentrations in the feed. A linear correlation between glycolate loading rate and biogas amount was observed (R² = 0.99) for glycolate loading rates up to 1.81 g L-1 day-1 with a maximum in biogas amount of 3635 mL day-1 encompassing 45% methane. The cytometric diversity remained high during the whole cultivation period. The dominating bacterial genera were Syntrophobotulus, Clostridia genus B55_F, Aminobacterium, and Petrimonas. Methanogenesis was almost exclusively performed by the hydrogenotrophic genus Methanobacterium.
Project description:Over the past years, gut microbiota became a major field of interest with increasing reports suggesting its association with a large number of human diseases. In this context, there is a major interest to develop analysis tools allowing simple and cost-effective population pattern analysis of these complex ecosystems to follow changes over time. Whereas sequence-based metagenomics profiling is widely used for microbial ecosystems characterization, it still requires time and specific expertise for analysis. Flow cytometry overcomes these disadvantages, providing key information on communities within hours. In addition, it can potentially be used to select, isolate and cultivate specific bacteria of interest. In this study, we evaluated the culturability of strictly anaerobic bacteria that were stained with a classical Live/Dead staining, and then sorted using flow cytometry under anaerobic conditions. This sorting of "viable" fraction demonstrated that 10-80% of identified "viable" cells of pure cultures of strictly anaerobic bacteria were culturable. In addition, we tested the use of a combination of labeled vancomycin and Wheat Germ Agglutinin (WGA) lectin to discriminate Gram-positive from Gram-negative bacteria in complex ecosystems. After validation on both aerobic/anaerobic facultative and strictly anaerobic bacteria, the staining methods were applied on complex ecosystems, revealing differences between culture conditions and demonstrating that minor pH variations have strong impacts on microbial community structure, which was confirmed by 16S rRNA gene sequencing. This combination of staining methods makes it possible to follow-up evolutions of complex microbial communities, supporting its future use as a rapid analysis tool in various applications. The flow cytometry staining method that was developed has the potential to facilitate the analysis of complex ecosystems by highlighting changes in bacterial communities' dynamics. It is assumed to be applicable as an efficient and fast approach to improve the control of processes linked to a wide range of ecosystems or known communities of bacterial species in both research and industrial contexts.
Project description:Linking the genomic content of uncultivated microbes to their metabolic functions remains a critical challenge in microbial ecology. Resolving this challenge has implications for improving our management of key microbial interactions in biotechnologies such as anaerobic digestion, which relies on slow-growing syntrophic and methanogenic communities to produce renewable methane from organic waste. In this study, we combined DNA stable-isotope probing (SIP) with genome-centric metagenomics to recover the genomes of populations enriched in 13C after growing on [13C]butyrate. Differential abundance analysis of recovered genomic bins across the SIP metagenomes identified two metagenome-assembled genomes (MAGs) that were significantly enriched in heavy [13C]DNA. Phylogenomic analysis assigned one MAG to the genus Syntrophomonas and the other MAG to the genus Methanothrix. Metabolic reconstruction of the annotated genomes showed that the Syntrophomonas genome encoded all the enzymes for beta-oxidizing butyrate, as well as several mechanisms for interspecies electron transfer via electron transfer flavoproteins, hydrogenases, and formate dehydrogenases. The Syntrophomonas genome shared low average nucleotide identity (<95%) with any cultured representative species, indicating that it is a novel species that plays a significant role in syntrophic butyrate degradation within anaerobic digesters. The Methanothrix genome contained the complete pathway for acetoclastic methanogenesis, indicating that it was enriched in 13C from syntrophic acetate transfer. This study demonstrates the potential of stable-isotope-informed genome-resolved metagenomics to identify in situ interspecies metabolic cooperation within syntrophic consortia important to anaerobic waste treatment as well as global carbon cycling.IMPORTANCE Predicting the metabolic potential and ecophysiology of mixed microbial communities remains a major challenge, especially for slow-growing anaerobes that are difficult to isolate. Unraveling the in situ metabolic activities of uncultured species may enable a more descriptive framework to model substrate transformations by microbiomes, which has broad implications for advancing the fields of biotechnology, global biogeochemistry, and human health. Here, we investigated the in situ function of mixed microbiomes by combining stable-isotope probing with metagenomics to identify the genomes of active syntrophic populations converting butyrate, a C4 fatty acid, into methane within anaerobic digesters. This approach thus moves beyond the mere presence of metabolic genes to resolve "who is doing what" by obtaining confirmatory assimilation of the labeled substrate into the DNA signature. Our findings provide a framework to further link the genomic identities of uncultured microbes with their ecological function within microbiomes driving many important biotechnological and global processes.
Project description:Microbes are ubiquitous throughout the world's oceans, yet the manner and extent of their influence on the ecology and evolution of large, mobile fauna remains poorly understood. Here, we establish the intestinal microbiome as a hidden, and potentially important, 'functional trait' of tropical herbivorous fishes-a group of large consumers critical to coral reef resilience. Using field observations, we demonstrate that five common Caribbean fish species display marked differences in where they feed and what they feed on. However, in addition to space use and feeding behaviour-two commonly measured functional traits-we find that interspecific trait differences are even more pronounced when considering the herbivore intestinal microbiome. Microbiome composition was highly species specific. Phylogenetic comparison of the dominant microbiome members to all known microbial taxa suggest that microbiomes are comprised of putative environmental generalists, animal-associates and fish specialists (resident symbionts), the latter of which mapped onto host phylogeny. These putative symbionts are most similar to-among all known microbes-those that occupy the intestines of ecologically and evolutionarily related herbivorous fishes in more distant ocean basins. Our findings therefore suggest that the intestinal microbiome may be an important functional trait among these large-bodied consumers.
Project description:BACKGROUND:Quantification of viable microorganisms is an important step in microbiological research as well as in microbial product formulation to develop biological control products or probiotics. Often, the efficiency of the resulting product is dependent on the microbial cell density and their viability, which may decrease over time. Commonly, the number of viable cells is determined by serial dilution and plating techniques or flow cytometry. In 2017, we developed a mathematical model for isothermal microcalorimetry (IMC) data analysis and showed that the new method allows for a more rapid quantification of viable fresh and freeze-dried anaerobic Lactobacillus reuteri cells than traditional viable count methods. RESULTS:This study developed the new method further by applying it to well-known aerophilic plant-beneficial microbial species (Pseudomonas brassicacearum, Bacillus amyloliquefaciens subsp. plantarum and Clonostachys rosea) used in biological control products. We utilized IMC to quantify viable cells in microbial pure cultures as well as when coated onto wheat seeds. The results from this study confirmed that thermal viable count methods are more rapid and sensitive than traditional viable count techniques. Most interestingly, a thermal viable count method was able to quantify microbes coated on seeds despite the presence of the natural microbiota of the seeds. Our results also showed that, in contrast to plating techniques for which clustered cells skew the results, IMC does not require single cells for accurate viable counts. CONCLUSIONS:Thermal viable count methods are novel methods for the rapid quantification of divergent bacterial and fungal species and enhance the speed, sensitivity, and accuracy of routine viable counts of pure cultures and controlled microbiomes such as plant seed coatings.
Project description:BACKGROUND:Microorganisms are responsible for fermentation within the rumen and have been reported to contribute to the variation in feed efficiency of cattle. However, to what extent the breed affects the rumen microbiome and its association with host feed efficiency is unknown. Here, rumen microbiomes of beef cattle (n?=?48) from three breeds (Angus, Charolais, Kinsella composite hybrid) with high and low feed efficiency were explored using metagenomics and metatranscriptomics, aiming to identify differences between functional potentials and activities of same rumen microbiomes and to evaluate the effects of host breed and feed efficiency on the rumen microbiome. RESULTS:Rumen metagenomes were more closely clustered together and thus more conserved among individuals than metatranscriptomes, suggesting that inter-individual functional variations at the RNA level were higher than those at the DNA level. However, while mRNA enrichment significantly increased the sequencing depth of mRNA and generated similar functional profiles to total RNA-based metatranscriptomics, it led to biased abundance estimation of several transcripts. We observed divergent rumen microbial composition (metatranscriptomic level) and functional potentials (metagenomic level) among three breeds, but differences in functional activity (metatranscriptomic level) were less apparent. Differential rumen microbial features (e.g., taxa, diversity indices, functional categories, and genes) were detected between cattle with high and low feed efficiency, and most of them were breed-specific. CONCLUSIONS:Metatranscriptomes represent real-time functional activities of microbiomes and have the potential to better associate rumen microorganisms with host performances compared to metagenomics. As total RNA-based metatranscriptomics seem to avoid potential biases caused by mRNA enrichment and allow simultaneous use of rRNA for generation of compositional profiles, we suggest their use for linking the rumen microbiome with host phenotypes in future studies. However, if exploration of specific lowly expressed genes is desired, mRNA enrichment is recommended as it will enhance the resolution of mRNA. Finally, the differential microbial features observed between efficient and inefficient steers tended to be specific to breeds, suggesting that interactions between host breed genotype and the rumen microbiome contribute to the variations in feed efficiency observed. These breed-associated differences represent an opportunity to engineer specific rumen microbiomes through selective breeding of the hosts.
Project description:BACKGROUND:Alcohol use causes significant disruption of intestinal microbial communities, yet exactly how these dysbiotic communities interact with the host is unclear. We sought to understand the role of microbial products associated with alcohol dysbiosis in mice on intestinal permeability and immune activation in an in vitro model system. METHODS:Microbiota samples from binge-on-chronic alcohol-fed and pair-fed male and female mice were cultured in Gifu Anaerobic Broth for 24 hours under anaerobic conditions. Live/whole organisms were removed, and microbial products were collected and added to human peripheral blood mononuclear cells (PBMCs) or polarized C2BBe1 intestinal epithelial monolayers. Following stimulation, transepithelial electrical resistance (TEER) was measured using a volt/ohm meter and immune activation of PBMC was assessed via flow cytometry. RESULTS:Microbial products from male and female alcohol-fed mice significantly decreased TEER (mean percentage change from baseline alcohol-fed 0.86 ?/cm2 vs. pair-fed 1.10 ?/cm2 ) compared to microbial products from control mice. Following ex vivo stimulation, immune activation of PBMC was assessed via flow cytometry. We found that microbial products from alcohol-fed mice significantly increased the percentage of CD38+ CD4+ (mean alcohol-fed 17.32% ± 0.683% standard deviation (SD) vs. mean pair-fed 14.2% ± 1.21% SD, p < 0.05) and CD8+ (mean alcohol-fed 20.28% ± 0.88% SD vs. mean pair-fed 12.58% ± 3.59% SD, p < 0.05) T cells. CONCLUSIONS:Collectively, these data suggest that microbial products contribute to immune activation and intestinal permeability associated with alcohol dysbiosis. Further, utilization of these ex vivo microbial product assays will allow us to rapidly assess the impact of microbial products on intestinal permeability and immune activation and to identify probiotic therapies to ameliorate these defects.
Project description:This study investigated the operation mode of a step-feed anoxic/oxic (A/O) process with distribution of the carbon source from the anaerobic zone in terms of the treatment effects on sewage with low carbon and high nitrogen and phosphorus. After seven phases of operation, an optimal flow distribution ratio of 75%:25% was obtained from the anaerobic zone, and the concentrations of chemical oxygen demand, ammonia nitrogen, total nitrogen, and total phosphorous in the effluent were 20.8, 0.64, 14.2, and 0.89?mg/L, respectively. The presence of an internal reflux system in the deaeration zone improved the treatment. 16S rRNA high-throughput sequencing revealed that the microbial communities in aerobic zone I(O1) of the first-step A/O sludge were different from those in aerobic zone I (O2) of the second-step A/O sludge, whereas microbial communities of the seed sludge were similar to those in O2 of the second-step A/O sludge. The richness and diversity of microbial communities in O1 of the first-step A/O sludge samples were higher than those in O2 of the second-step A/O and seed sludge. At the optimal flow distribution ratio, the microbial abundance and treatment removal efficiency were the highest.
Project description:Much work has been dedicated to identifying members of the microbial gut community that have potential to augment the growth rate of agricultural animals including chickens. Here, we assessed any correlations between the fecal microbiome, a proxy for the gut microbiome, and feed efficiency or weight gain at the pedigree chicken level, the highest tier of the production process. Because selective breeding is conducted at the pedigree level, our aim was to determine if microbiome profiles could be used to predict feed conversion or weight gain in order to improve selective breeding. Using 16s rRNA amplicon sequencing, we profiled the microbiomes of high and low weight gain (WG) birds and good and poor feed efficient (FE) birds in two pedigree lineages of broiler chickens. We also aimed to understand the dynamics of the microbiome with respect to maturation. A time series experiment was conducted, where fecal samples of chickens were collected at 6 points of the rearing process and the microbiome of these samples profiled. We identified OTUs differences at different taxonomic levels in the fecal community between high and low performing birds within each genetic line, indicating a specificity of the microbial community profiles correlated to performance factors. Using machine-learning methods, we built a classification model that could predict feed conversion performance from the fecal microbial community. With respect to maturation, we found that the fecal microbiome is dynamic in early life but stabilizes after 3 weeks of age independent of lineage. Our results indicate that the fecal microbiome profile can be used to predict feed conversion, but not weight gain in these pedigree lines. From the time series experiments, it appears that these predictions can be evaluated as early as 20 days of age. Our data also indicates that there is a genetic factor for the microbiome profile.
Project description:Lysine acetylation (Kac), an abundant post-translational modification (PTM) in prokaryotes, regulates various microbial metabolic pathways. However, no studies have examined protein Kac at the microbiome level, and it remains unknown whether Kac level is altered in patient microbiomes. Herein, we use a peptide immuno-affinity enrichment strategy coupled with mass spectrometry to characterize protein Kac in the microbiome, which successfully identifies 35,200 Kac peptides from microbial or human proteins in gut microbiome samples. We demonstrate that Kac is widely distributed in gut microbial metabolic pathways, including anaerobic fermentation to generate short-chain fatty acids. Applying to the analyses of microbiomes of patients with Crohn's disease identifies 52 host and 136 microbial protein Kac sites that are differentially abundant in disease versus controls. This microbiome-wide acetylomic approach aids in advancing functional microbiome research.