BioAutoMATED: An end-to-end automated machine learning tool for explanation and design of biological sequences.

The design choices underlying machine-learning (ML) models present important barriers to entry for many biologists who aim to incorporate ML in their research. Automated machine-learning (AutoML) algorithms can address many challenges that come with applying ML to the life sciences. However, these algorithms are rarely used in systems and synthetic biology studies because they typically do not explicitly handle biological sequences (e.g., nucleotide, amino acid, or glycan sequences) and cannot be easily compared with other AutoML algorithms. Here, we present BioAutoMATED, an AutoML platform for biological sequence analysis that integrates multiple AutoML methods into a unified framework. Users are automatically provided with relevant techniques for analyzing, interpreting, and designing biological sequences. BioAutoMATED predicts gene regulation, peptide-drug interactions, and glycan annotation, and designs optimized synthetic biology components, revealing salient sequence characteristics. By automating sequence modeling, BioAutoMATED allows life scientists to incorporate ML more readily into their work.

Enhancers of Host Immune Tolerance to Bacterial Infection Discovered Using Linked Computational and Experimental Approaches.

Current therapeutic strategies against bacterial infections focus on reduction of pathogen load using antibiotics; however, stimulation of host tolerance to infection in the presence of pathogens might offer an alternative approach. Computational transcriptomics and Xenopus laevis embryos are used to discover infection response pathways, identify potential tolerance inducer drugs, and validate their ability to induce broad tolerance. Xenopus exhibits natural tolerance to Acinetobacter baumanii, Klebsiella pneumoniae, Staphylococcus aureus, and Streptococcus pneumoniae bacteria, whereas Aeromonas hydrophila and Pseudomonas aeruginosa produce lethal infections. Transcriptional profiling leads to definition of a 20-gene signature that discriminates between tolerant and susceptible states, as well as identification of a more active tolerance response to gram negative compared to gram positive bacteria. Gene pathways associated with active tolerance in Xenopus, including some involved in metal ion binding and hypoxia, are found to be conserved across species, including mammals, and administration of a metal chelator (deferoxamine) or a HIF-1α agonist (1,4-DPCA) in embryos infected with lethal A. hydrophila increased survival despite high pathogen load. These data demonstrate the value of combining the Xenopus embryo infection model with computational multiomics analyses for mechanistic discovery and drug repurposing to induce host tolerance to bacterial infections.

Nutritional deficiency in an intestine-on-a-chip recapitulates injury hallmarks associated with environmental enteric dysfunction.

Environmental enteric dysfunction (EED)-a chronic inflammatory condition of the intestine-is characterized by villus blunting, compromised intestinal barrier function and reduced nutrient absorption. Here we show that essential genotypic and phenotypic features of EED-associated intestinal injury can be reconstituted in a human intestine-on-a-chip lined by organoid-derived intestinal epithelial cells from patients with EED and cultured in nutrient-deficient medium lacking niacinamide and tryptophan. Exposure of the organ chip to such nutritional deficiencies resulted in congruent changes in six of the top ten upregulated genes that were comparable to changes seen in samples from patients with EED. Chips lined with healthy epithelium or with EED epithelium exposed to nutritional deficiencies resulted in severe villus blunting and barrier dysfunction, and in the impairment of fatty acid uptake and amino acid transport; and the chips with EED epithelium exhibited heightened secretion of inflammatory cytokines. The organ-chip model of EED-associated intestinal injury may facilitate the analysis of the molecular, genetic and nutritional bases of the disease and the testing of candidate therapeutics for it.

Harnessing Colon Chip Technology to Identify Commensal Bacteria That Promote Host Tolerance to Infection.

Commensal bacteria within the gut microbiome contribute to development of host tolerance to infection, however, identifying specific microbes responsible for this response is difficult. Here we describe methods for developing microfluidic organ-on-a-chip models of small and large intestine lined with epithelial cells isolated from duodenal, jejunal, ileal, or colon organoids derived from wild type or transgenic mice. To focus on host-microbiome interactions, we carried out studies with the mouse Colon Chip and demonstrated that it can support co-culture with living gut microbiome and enable assessment of effects on epithelial adhesion, tight junctions, barrier function, mucus production, and cytokine release. Moreover, infection of the Colon Chips with the pathogenic bacterium, Salmonella typhimurium, resulted in epithelial detachment, decreased tight junction staining, and increased release of chemokines (CXCL1, CXCL2, and CCL20) that closely mimicked changes previously seen in mice. Symbiosis between microbiome bacteria and the intestinal epithelium was also recapitulated by populating Colon Chips with complex living mouse or human microbiome. By taking advantage of differences in the composition between complex microbiome samples cultured on each chip using 16s sequencing, we were able to identify Enterococcus faecium as a positive contributor to host tolerance, confirming past findings obtained in mouse experiments. Thus, mouse Intestine Chips may represent new experimental in vitro platforms for identifying particular bacterial strains that modulate host response to pathogens, as well as for investigating the cellular and molecular basis of host-microbe interactions.

Deep-Learning Resources for Studying Glycan-Mediated Host-Microbe Interactions.

Glycans, the most diverse biopolymer, are shaped by evolutionary pressures stemming from host-microbe interactions. Here, we present machine learning and bioinformatics methods to leverage the evolutionary information present in glycans to gain insights into how pathogens and commensals interact with hosts. By using techniques from natural language processing, we develop deep-learning models for glycans that are trained on a curated dataset of 19,299 unique glycans and can be used to study and predict glycan functions. We show that these models can be utilized to predict glycan immunogenicity and the pathogenicity of bacterial strains, as well as investigate glycan-mediated immune evasion via molecular mimicry. We also develop glycan-alignment methods and use these to analyze virulence-determining glycan motifs in the capsular polysaccharides of bacterial pathogens. These resources enable one to identify and study glycan motifs involved in immunogenicity, pathogenicity, molecular mimicry, and immune evasion, expanding our understanding of host-microbe interactions.

Sequence-to-function deep learning frameworks for engineered riboregulators.

While synthetic biology has revolutionized our approaches to medicine, agriculture, and energy, the design of completely novel biological circuit components beyond naturally-derived templates remains challenging due to poorly understood design rules. Toehold switches, which are programmable nucleic acid sensors, face an analogous design bottleneck; our limited understanding of how sequence impacts functionality often necessitates expensive, time-consuming screens to identify effective switches. Here, we introduce Sequence-based Toehold Optimization and Redesign Model (STORM) and Nucleic-Acid Speech (NuSpeak), two orthogonal and synergistic deep learning architectures to characterize and optimize toeholds. Applying techniques from computer vision and natural language processing, we 'un-box' our models using convolutional filters, attention maps, and in silico mutagenesis. Through transfer-learning, we redesign sub-optimal toehold sensors, even with sparse training data, experimentally validating their improved performance. This work provides sequence-to-function deep learning frameworks for toehold selection and design, augmenting our ability to construct potent biological circuit components and precision diagnostics.

Author Correction: A complex human gut microbiome cultured in an anaerobic intestine-on-a-chip.

In the version of this Article originally published, the authors mistakenly cited Fig. 5d in the sentence beginning 'Importantly, the microbiome cultured in these primary Intestine Chips...'; the correct citation is Supplementary Table 2. This has now been amended.

A complex human gut microbiome cultured in an anaerobic intestine-on-a-chip.

The diverse bacterial populations that comprise the commensal microbiome of the human intestine play a central role in health and disease. A method that sustains complex microbial communities in direct contact with living human intestinal cells and their overlying mucus layer in vitro would thus enable the investigation of host-microbiome interactions. Here, we show the extended coculture of living human intestinal epithelium with stable communities of aerobic and anaerobic human gut microbiota, using a microfluidic intestine-on-a-chip that permits the control and real-time assessment of physiologically relevant oxygen gradients. When compared to aerobic coculture conditions, the establishment of a transluminal hypoxia gradient in the chip increased intestinal barrier function and sustained a physiologically relevant level of microbial diversity, consisting of over 200 unique operational taxonomic units from 11 different genera and an abundance of obligate anaerobic bacteria, with ratios of Firmicutes and Bacteroidetes similar to those observed in human faeces. The intestine-on-a-chip may serve as a discovery tool for the development of microbiome-related therapeutics, probiotics and nutraceuticals.

Species-specific enhancement of enterohemorrhagic E. coli pathogenesis mediated by microbiome metabolites.

BACKGROUND: Species-specific differences in tolerance to infection are exemplified by the high susceptibility of humans to enterohemorrhagic Escherichia coli (EHEC) infection, whereas mice are relatively resistant to this pathogen. This intrinsic species-specific difference in EHEC infection limits the translation of murine research to human. Furthermore, studying the mechanisms underlying this differential susceptibility is a difficult problem due to complex in vivo interactions between the host, pathogen, and disparate commensal microbial communities. RESULTS: We utilize organ-on-a-chip (Organ Chip) microfluidic culture technology to model damage of the human colonic epithelium induced by EHEC infection, and show that epithelial injury is greater when exposed to metabolites derived from the human gut microbiome compared to mouse. Using a multi-omics approach, we discovered four human microbiome metabolites-4-methyl benzoic acid, 3,4-dimethylbenzoic acid, hexanoic acid, and heptanoic acid-that are sufficient to mediate this effect. The active human microbiome metabolites preferentially induce expression of flagellin, a bacterial protein associated with motility of EHEC and increased epithelial injury. Thus, the decreased tolerance to infection observed in humans versus other species may be due in part to the presence of compounds produced by the human intestinal microbiome that actively promote bacterial pathogenicity. CONCLUSION: Organ-on-chip technology allowed the identification of specific human microbiome metabolites modulating EHEC pathogenesis. These identified metabolites are sufficient to increase susceptibility to EHEC in our human Colon Chip model and they contribute to species-specific tolerance. This work suggests that higher concentrations of these metabolites could be the reason for higher susceptibility to EHEC infection in certain human populations, such as children. Furthermore, this research lays the foundation for therapeutic-modulation of microbe products in order to prevent and treat human bacterial infection.

Directed differentiation of human induced pluripotent stem cells into mature kidney podocytes and establishment of a Glomerulus Chip.

Protocols have been established to direct the differentiation of human induced pluripotent stem (iPS) cells into nephron progenitor cells and organoids containing many types of kidney cells, but it has been difficult to direct the differentiation of iPS cells to form specific types of mature human kidney cells with high yield. Here, we describe a detailed protocol for the directed differentiation of human iPS cells into mature, post-mitotic kidney glomerular podocytes with high (>90%) efficiency within 26 d and under chemically defined conditions, without genetic manipulations or subpopulation selection. We also describe how these iPS cell-derived podocytes may be induced to form within a microfluidic organ-on-a-chip (Organ Chip) culture device to build a human kidney Glomerulus Chip that mimics the structure and function of the kidney glomerular capillary wall in vitro within 35 d (starting with undifferentiated iPS cells). The podocyte differentiation protocol requires skills for culturing iPS cells, and the development of a Glomerulus Chip requires some experience with building and operating microfluidic cell culture systems. This method could be useful for applications in nephrotoxicity screening, therapeutic development, and regenerative medicine, as well as mechanistic study of kidney development and disease.

Next-Generation Machine Learning for Biological Networks.

Machine learning, a collection of data-analytical techniques aimed at building predictive models from multi-dimensional datasets, is becoming integral to modern biological research. By enabling one to generate models that learn from large datasets and make predictions on likely outcomes, machine learning can be used to study complex cellular systems such as biological networks. Here, we provide a primer on machine learning for life scientists, including an introduction to deep learning. We discuss opportunities and challenges at the intersection of machine learning and network biology, which could impact disease biology, drug discovery, microbiome research, and synthetic biology.

Central role for RNase YbeY in Hfq-dependent and Hfq-independent small-RNA regulation in bacteria.

BACKGROUND: Conceptual parallels exist between bacterial and eukaryotic small-RNA (sRNA) pathways, yet relatively little is known about which protein may recognize and recruit bacterial sRNAs to interact with targets. In eukaryotes, Argonaute (AGO) proteins discharge such functions. The highly conserved bacterial YbeY RNase has structural similarities to the MID domain of AGOs. A limited study had indicated that in Sinorhizobium meliloti the YbeY ortholog regulates the accumulation of sRNAs as well as the target mRNAs, raising the possibility that YbeY may play a previously unrecognized role in bacterial sRNA regulation. RESULTS: We have applied a multipronged approach of loss-of-function studies, genome-wide mRNA and sRNA expression profiling, pathway analysis, target prediction, literature mining and network analysis to unravel YbeY-dependent molecular responses of E. coli exposed to hydroxyurea (HU). Loss of ybeY function, which results in a marked resistance to HU, had global affects on sRNA-mediated gene expression. Of 54 detectable E. coli sRNAs in our microarray analysis, 30 sRNAs showed a differential expression upon HU stress, of which 28 sRNAs displayed a YbeY-dependent change in expression. These included 12 Hfq-dependent and 16 Hfq-independent sRNAs. We successfully identified at least 57 experimentally inferred sRNA-mRNA relationships. Further applying a 'context likelihood of relatedness' algorithm, we reverse engineered the YbeY-dependent Hfq-dependent sRNA-mRNA network as well as YbeY-dependent Hfq-independent sRNA-mRNA network. CONCLUSION: YbeY extensively modulates Hfq-dependent and independent sRNA-mRNA interactions. YbeY-dependent sRNAs have central roles in modulating cellular response to HU stress.

The Mycobacterium tuberculosis regulatory network and hypoxia.

We have taken the first steps towards a complete reconstruction of the Mycobacterium tuberculosis regulatory network based on ChIP-Seq and combined this reconstruction with system-wide profiling of messenger RNAs, proteins, metabolites and lipids during hypoxia and re-aeration. Adaptations to hypoxia are thought to have a prominent role in M. tuberculosis pathogenesis. Using ChIP-Seq combined with expression data from the induction of the same factors, we have reconstructed a draft regulatory network based on 50 transcription factors. This network model revealed a direct interconnection between the hypoxic response, lipid catabolism, lipid anabolism and the production of cell wall lipids. As a validation of this model, in response to oxygen availability we observe substantial alterations in lipid content and changes in gene expression and metabolites in corresponding metabolic pathways. The regulatory network reveals transcription factors underlying these changes, allows us to computationally predict expression changes, and indicates that Rv0081 is a regulatory hub.

Fungicidal drugs induce a common oxidative-damage cellular death pathway.

Amphotericin, miconazole, and ciclopirox are antifungal agents from three different drug classes that can effectively kill planktonic yeast, yet their complete fungicidal mechanisms are not fully understood. Here, we employ a systems biology approach to identify a common oxidative-damage cellular death pathway triggered by these representative fungicides in Candida albicans and Saccharomyces cerevisiae. This mechanism utilizes a signaling cascade involving the GTPases Ras1 and Ras2 and protein kinase A, and it culminates in death through the production of toxic reactive oxygen species in a tricarboxylic-acid-cycle- and respiratory-chain-dependent manner. We also show that the metabolome of C. albicans is altered by antifungal drug treatment, exhibiting a shift from fermentation to respiration, a jump in the AMP/ATP ratio, and elevated production of sugars; this coincides with elevated mitochondrial activity. Lastly, we demonstrate that DNA damage plays a critical role in antifungal-induced cellular death and that blocking DNA-repair mechanisms potentiates fungicidal activity.

Wisdom of crowds for robust gene network inference.

Reconstructing gene regulatory networks from high-throughput data is a long-standing challenge. Through the Dialogue on Reverse Engineering Assessment and Methods (DREAM) project, we performed a comprehensive blind assessment of over 30 network inference methods on Escherichia coli, Staphylococcus aureus, Saccharomyces cerevisiae and in silico microarray data. We characterize the performance, data requirements and inherent biases of different inference approaches, and we provide guidelines for algorithm application and development. We observed that no single inference method performs optimally across all data sets. In contrast, integration of predictions from multiple inference methods shows robust and high performance across diverse data sets. We thereby constructed high-confidence networks for E. coli and S. aureus, each comprising ~1,700 transcriptional interactions at a precision of ~50%. We experimentally tested 53 previously unobserved regulatory interactions in E. coli, of which 23 (43%) were supported. Our results establish community-based methods as a powerful and robust tool for the inference of transcriptional gene regulatory networks.

Antibiotic-induced bacterial cell death exhibits physiological and biochemical hallmarks of apoptosis.

Programmed cell death is a gene-directed process involved in the development and homeostasis of multicellular organisms. The most common mode of programmed cell death is apoptosis, which is characterized by a stereotypical set of biochemical and morphological hallmarks. Here we report that Escherichia coli also exhibit characteristic markers of apoptosis-including phosphatidylserine exposure, chromosome condensation, and DNA fragmentation-when faced with cell death-triggering stress, namely bactericidal antibiotic treatment. Notably, we also provide proteomic and genetic evidence for the ability of multifunctional RecA to bind peptide sequences that serve as substrates for eukaryotic caspases, and regulation of this phenotype by the protease, ClpXP, under conditions of cell death. Our findings illustrate that prokaryotic organisms possess mechanisms to dismantle and mark dying cells in response to diverse noxious stimuli and suggest that elaborate, multilayered proteolytic regulation of these features may have evolved in eukaryotes to harness and exploit their deadly potential.

Functional characterization of bacterial sRNAs using a network biology approach.

Small RNAs (sRNAs) are important components of posttranscriptional regulation. These molecules are prevalent in bacterial and eukaryotic organisms, and involved in a variety of responses to environmental stresses. The functional characterization of sRNAs is challenging and requires highly focused and extensive experimental procedures. Here, using a network biology approach and a compendium of gene expression profiles, we predict functional roles and regulatory interactions for sRNAs in Escherichia coli. We experimentally validate predictions for three sRNAs in our inferred network: IsrA, GlmZ, and GcvB. Specifically, we validate a predicted role for IsrA and GlmZ in the SOS response, and we expand on current knowledge of the GcvB sRNA, demonstrating its broad role in the regulation of amino acid metabolism and transport. We also show, using the inferred network coupled with experiments, that GcvB and Lrp, a transcription factor, repress each other in a mutually inhibitory network. This work shows that a network-based approach can be used to identify the cellular function of sRNAs and characterize the relationship between sRNAs and transcription factors.

Systems biology strikes gold.

Integrating synthetic biology and systems biology efforts can advance our understanding of biomolecular systems. This is illustrated in this issue by Cantone et al. (2009), who construct a synthetic gene network in yeast and use it to assess and benchmark systems biology approaches for reverse engineering endogenous gene networks.

Comparison of reverse-engineering methods using an in silico network.

The reverse engineering of biochemical networks is a central problem in systems biology. In recent years several methods have been developed for this purpose, using techniques from a variety of fields. A systematic comparison of the different methods is complicated by their widely varying data requirements, making benchmarking difficult. Also, because of the lack of detailed knowledge about most real networks, it is not easy to use experimental data for this purpose. This paper contains a comparison of four reverse-engineering methods using data from a simulated network. The network is sufficiently realistic and complex to include many of the challenges that data from real networks pose. Our results indicate that the two methods based on genetic perturbations of the network outperform the other methods, including dynamic Bayesian networks and a partial correlation method.