Framework for exploring the sensory repertoire of the human gut microbiota

Abstract

Bacteria sense changes in their environment and transduce signals to adjust their cellular functions accordingly. For this purpose, bacteria employ various sensors feeding into multiple signal transduction pathways. Signal recognition by bacterial sensors is studied mainly in a few model organisms, but advances in genome sequencing and analysis offer new ways of exploring the sensory repertoire of many understudied organisms. The human gut is a natural target of this line of study: it is a nutrient-rich and dynamic environment and is home to thousands of bacterial species whose activities impact human health. Many gut commensals are also poorly studied compared to model organisms and are mainly known through their genome sequences. To begin exploring the signals human gut commensals sense and respond to, we have designed a framework that enables the identification of sensory domains, prediction of signals that they recognize, and experimental verification of these predictions. We validate this framework's functionality by systematically identifying amino acid sensors in selected bacterial genomes and metagenomes, characterizing their amino acid binding properties, and demonstrating their signal transduction potential.IMPORTANCESignal transduction is a central process governing how bacteria sense and respond to their environment. The human gut is a complex environment with many living organisms and fluctuating streams of nutrients. One gut inhabitant, Escherichia coli, is a model organism for studying signal transduction. However, E. coli is not representative of most gut microbes, and signaling pathways in the thousands of other organisms comprising the human gut microbiota remain poorly understood. This work provides a foundation for how to explore signals recognized by these organisms.

Keywords: amino acids; gut microbiome; receptors; signal transduction.

Fig 1

Workflow for identifying sensory domains in the human gut microbiome. Sources of information are shown in gray. Computational steps are shown in blue and experimental steps are shown in pink. FRET, Förster resonance energy transfer; HMM, hidden Markov model; ITC, isothermal titration calorimetry; MAG, metagenome assembled genome; MiST, Microbial Signal Transduction Database. Protein Data Bank (PDB).

Fig 2

Extracytoplasmic sensory domains identified in the human gut microbiome. (A) Distribution of domains identified in genomes of cultivated human gut commensal bacteria. The underlying data are available as Data Set S2. (B) Distribution of domains identified in the UHGP catalog. The underlying data are available as Data Set S3. Families of the Cache superfamily (clan CL0165) are shown in blue: double-module Cache domain families in dark blue and single-module Cache domain families in light blue. A family of the 4HB_MCP superfamily (clan CL0457) is shown in yellow. Domain nomenclature and accession numbers are from the Pfam database.

Fig 3

Representative predicted dCache_1AA domains and their putative ligands. (A) Multiple sequence alignment of the known dCache_1AA domain from Enterobacter cloacae chemoreceptor with six putative dCache_1AA domains from human gut commensals. (B) The putative ligands for six dCache_1AA domains. Thermal shift assays for six purified dCache_1AA domains (AA1–AA6) were performed in the presence of bacterial nitrogen sources from the Biolog screen plate PM3B. Except for the dCache_1AA domain AA4, all compounds that induced a shift in the midpoint of protein unfolding transition (Tm) of at least 2°C are presented. The complete list of positive ligands for the dCache_1AA domain AA4 is shown in Fig. S1. The data shown are the means and standard errors from three biological replicates conducted in triplicate.

Fig 4

Computational docking of putative ligands to predicted dCache_1AA domains from human gut commensal bacteria. (A) The AlphaFold structural model of the dCache_1 domain and flanking transmembrane helices from Enterobacter hormaechei chemoreceptor (AA4). Docking of putative ligands to the distal module of selected predicted dCache_1 AA domains: (B) AA4 from Enterobacter hormaechei with L-Phe, (C) AA5 from Agathobacter faecis with L-Ile, and (D) AA6 from Ligilactobacillus ruminis with L-Arg. Amino acid residues highlighted in blue are making contacts with the amino group of the ligand, and amino acid residues highlighted in red are making contacts with the carboxyl group of the ligand. Structural models with docked ligands are available at https://zenodo.org/records/10806258.

Fig 5

Microcalorimetric titration of selected recombinant dCache_1AA domains with amino acids. The upper panel upper shows raw titration data, and the lower panel shows integrated corrected peak areas of the titration data fit using the single-site model. Further experimental details are provided in Table S3. (A) AA4 from Enterobacter hormaechei, (B) AA5 from Agathobacter faecis, and (C) AA6 from Ligilactobacillus ruminis.

Fig 6

FRET measurements for E. coli cells expressing AA6-Tar hybrid. (A) Schematic representation of the hybrid receptor construction, with domains of tar shown in different shades of gray and domains of AA6 in different shades of red. Transmembrane (TM) sequences are indicated. The hybrid is composed of TM1 and ligand-binding domain of AA6 receptor, fused within TM2 to the cytoplasmic region of tar using the indicated linker. (B) FRET measurement of E. coli chemotaxis pathway response in cells expressing AA6-tar hybrid as a sole receptor to the indicated concentrations of L-Arg. Glucose was used as a positive control, since it elicits a chemotactic response in a receptor-unspecific manner. (C) Corresponding dose response mediated by AA6-tar hybrid with stimulation by L-Arg. Data were fitted using Hill equation, with the EC50 fit value being indicated. (D) FRET measurement of the wild-type tar, with the known ligand α-methyl-D,L-aspartate used as a positive control.