Identification of trypsin-degrading commensals in the large intestine

Abstract

Increased levels of proteases, such as trypsin, in the distal intestine have been implicated in intestinal pathological conditions^1-3. However, the players and mechanisms that underlie protease regulation in the intestinal lumen have remained unclear. Here we show that Paraprevotella strains isolated from the faecal microbiome of healthy human donors are potent trypsin-degrading commensals. Mechanistically, Paraprevotella recruit trypsin to the bacterial surface through type IX secretion system-dependent polysaccharide-anchoring proteins to promote trypsin autolysis. Paraprevotella colonization protects IgA from trypsin degradation and enhances the effectiveness of oral vaccines against Citrobacter rodentium. Moreover, Paraprevotella colonization inhibits lethal infection with murine hepatitis virus-2, a mouse coronavirus that is dependent on trypsin and trypsin-like proteases for entry into host cells^4,5. Consistently, carriage of putative genes involved in trypsin degradation in the gut microbiome was associated with reduced severity of diarrhoea in patients with SARS-CoV-2 infection. Thus, trypsin-degrading commensal colonization may contribute to the maintenance of intestinal homeostasis and protection from pathogen infection.

Fig. 1. Microbiota-mediated regulation of trypsin in the large intestine.

a, Proteins with reduced levels in the caecum of SPF mice compared with in the caecum of GF mice, as determined by proteome analysis. b, Faecal trypsin activity in SPF mice compared with in GF mice. c, Western blot analysis of trypsin (PRSS2) in the faeces of SPF and GF mice. d, Immunostaining of colon sections of SPF and GF mice. Blue, DAPI; green, PRSS2; red, UEA1 (mucus). e,f, Prss2 expression levels in the pancreas of SPF or GF mice measured using quantitative PCR with reverse transcription (RT–qPCR) (e) and western blotting (f). Heat-shock protein 90 (HSP90) was the loading control. g, Trypsin activity of intestinal contents at the indicated locations. h, Faecal trypsin activity of GF mice or GF mice inoculated with faecal samples from the indicated healthy donors (A–F). i, Trypsin activity in faeces of GF mice after inoculation with the caecal contents of mouse C5 and concomitant treatment with antibiotics (Abx) or vehicle control. For b, e and g–i, data are mean ± s.d. Each dot represents one mouse (b, e, g and h). Statistical analysis was performed using two-sided Mann–Whitney U-tests with Welch’s correction (nonparametric) (b, e and g) and one-way analysis of variance (ANOVA) with Tukey’s test (h and i); ****P < 0.0001, ***P < 0.001, *P < 0.05; NS, not significant. For d, scale bar, 500 μm. For c, a representative image from two independent experiments with similar results is shown. For f, images from one experiment including all of the mice used in e are shown. Blot source data are provided in Supplementary Fig. 1. Source data

Fig. 2. Identification of Paraprevotella as trypsin-degrading species.

a, The caecal microbiota composition of individual mice determined by 16S rRNA gene sequencing. Operational taxonomic units (OTUs) significantly negatively correlated (ρ ≤ −0.5, P < 0.05), negatively but not significantly correlated, and positively correlated with trypsin activity are marked in red, grey and blue, respectively. OTUs corresponding to the 35 strains isolated from the caecal contents of mouse C5-Amp#5 are marked in yellow and their closest species and percentage similarity in the NCBI-RefSeq 16S rRNA gene database are shown. b–e,g,h, The faecal trypsin activity of mice colonized with the indicated bacterial mixtures. f,j, Recombinant mouse trypsin (rmPRSS2) was in vitro incubated with individual strains of the 9-mix (f) or the indicated Paraprevotella or Prevotella strains (j), and degradation of rmPRSS2 was analysed using western blotting. The asterisk indicates the cleaved fragment of rmPRSS2. i, Recombinant human trypsin isoforms PRSS1, PRSS2 and PRSS3 (rhPRSS1–3) were incubated with P. clara 1C4 and degradation of human trypsin was analysed using western blotting. For b–e, g and h, data are mean ± s.d. Each dot represents one mouse. Statistical analysis was performed using two-sided Mann–Whitney U-tests with Welch’s correction (nonparametric) (h) and one-way ANOVA with Tukey’s test (b–e and g); ****P < 0.0001, ***P < 0.001, **P < 0.01. For f, i and j, representative images from two independent experiments with similar results are shown. Blot source data are provided in Supplementary Fig. 1. Source data

Fig. 3. Identification of effector molecules responsible for Paraprevotella-mediated trypsin degradation.

a, Recombinant mouse trypsin (rmPRSS2) pretreated with the indicated protease inhibitors was incubated with P. clara 1C4, and degradation of rmPRSS2 was analysed using western blotting. b, Alexa Fluor 488-labelled rmPRSS2 (green) was incubated with the indicated species, and association of rmPRSS2 with the bacterial surface was examined using confocal microscopy. The black square indicates the region magnified in the top right, showing P. clara cells. c,d, rmPRSS2 degradation (c) and association with the bacterial surface (d) after incubation with P. clara 1C4 pretreated with tunicamycin  or vehicle control. e, rmPRSS2 degradation mediated by WT or PorU-mutant P. clara JCM14859. f, P. clara proteins with elevated levels in the culture supernatants after tunicamycin treatment, as determined by proteome analysis. g, rmPRSS2 degradation mediated by WT or the indicated mutants of P. clara JCM14859. h, Association of rmPRSS2 with the surface of WT or the indicated deletion mutants of P. clara JCM14859. i, Transmission electron microscopy images of WT or Δ00502 strains incubated with rmPRSS2. The green arrowheads indicate immunogold-labelled rmPRSS2. j, rmPRSS2 degradation after incubation with microbead-coupled or free-form recombinant 00502 and/or 00509. k, Association of rmPRSS2 with microbead-coupled recombinant 00502 and/or 00509 or albumin control (BSA). For f, data are mean ± s.d. Each dot represents one technical replicate. Statistical analysis was performed using two-sided multiple unpaired t-tests (not corrected for multiple comparisons); ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05. Scale bars, 5 μm (b, d, h and k) and 200 nm (i). For a–e and g–k, representative images from two independent experiments with similar results (a–e, g, h, j and k) or images from one experiment (i) are shown. Blot source data are provided in Supplementary Fig. 1. Source data

Fig. 4. Paraprevotella-mediated degradation of trypsin modulates colonic homeostasis.

a–c, GF mice were colonized with the indicated P. clara strains together with the 2-mix (a,b; n = 5 mice per group) or the 34-mix (c; n = 6 mice per group) for 14 days. Faecal trypsin activity (a,c) and the amount of indicated proteins (b; determined by western blotting) are shown. d–f, The viral RNA levels in the faeces or the indicated tissues (d), survival curve (e) and representative images of haematoxylin and eosin (H&E) staining of liver sections (f) of GF+2-mix+WT or GF+2-mix+Δ00502 mice infected with MHV-2 (intragastric inoculation). Among the 32 (GF+2-mix+WT group) and 33 (GF+2-mix+Δ00502 group) infected mice, 16 mice from each group were euthanized on day 5 for tissue viral RNA analysis (d) and the rest of the mice were followed for survival analysis (e). g,h, Viral RNA levels (g) and survival curve (h) of GF+34-mix+WT or GF+34-mix+Δ00502 mice after intragastric inoculation with MHV-2. n = 15 mice per group (10 mice were euthanized on day 5 for tissue viral RNA analysis and the rest of the mice were followed for survival analysis). i, Survival curve of GF+2-mix+WT or GF+2-mix+Δ00502 mice intraperitoneally injected with MHV-2. n = 5 mice per group. j, Genome neighbourhood of the homologues of the P. clara 00502-00509 locus in human and mouse (P. rodentium and P. muris) gut microorganisms. The percentage amino acid identity with P. clara 00502 and 00509 is shown. k,l, The frequency of patients with COVID-19 experiencing more than 1 day with more than 2 diarrhoeal episodes per day (k) or requiring oxygen inhalation therapy (l), stratified by the presence (00502 (+)) or absence (00502 (−)) of 00502 homologue genes in the faecal metagenome. For a, c, d and g, data are mean ± s.d. Each dot represents one mouse. Statistical analysis was performed using one-way ANOVA with Tukey’s test (a), two-sided Mann–Whitney U-tests with Welch’s correction (nonparametric) (c, d and g), log-rank (Mantel–Cox) tests (e, h and i) and one-sided Fisher’s tests (k and l); ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05. For f, scale bar, 500 μm. For b, images from one experiment, including all of the mice used in a, are shown. Blot source data are provided in Supplementary Fig. 1. Source data

Extended Data Fig. 1. Elevated trypsin levels in germ free (GF) mice, and in humans and mice with intestinal inflammation.

Distribution of all host-derived proteins detected in the proteome analysis of the caecal contents from specific-pathogen-free (SPF) or GF mice, with protein relative abundance plotted against p value. Anionic trypsin 2 (PRSS2) is highlighted in red. See Supplementary Table 1 for the complete list of proteins detected. b, Faecal trypsin activity of healthy controls, ulcerative colitis (UC) and Crohn’s disease (CD) patients. c, Faecal trypsin activity of Il10^+/− and Il10^−/− mice. b, c, Data shown as mean ± s.d. Each dot represents one human subject or one mouse. ** p < 0.01; * p < 0.05. One-way ANOVA with Tukey’s test (b) and two-sided Mann-Whitney test with Welch’s correction (nonparametric) (c). Source data

Extended Data Fig. 2. Reduced faecal trypsin levels in gnotobiotic mice colonized with bacterial mixtures containing P. clara.

a, Schematic representation of the strategy for isolating trypsin-reducing bacteria from the healthy human gut microbiota. The caecal contents from a GF mouse colonized with the donor C microbiota and receiving ampicillin treatment were cultured anaerobically on various types of agar plates containing different growth media including EG, ES, M10, NBGT, VS, TS, BL, BBE, Oxoid CM0619, CM0619-supplemented SR0107, CM0619-supplemented SR0108, mGAM and Schaedler. 432 bacterial colonies were picked and sequenced. The 35 strains identified were subjected to further rounds of gnotobiotic and in vitro screening until identification of P. clara as the effector strain. b, Spearman’s correlation coefficient quantifying the association between relative abundance and trypsin activity for individual bacterial OTUs detected in mice in Fig. 2a. Operational taxonomic units (OTUs) significantly negatively correlated (ρ ≤ −0.5, p < 0.05), negatively but not significantly correlated, and positively correlated with trypsin activity are marked in red, grey and blue, respectively. c, Western Blot analysis of mouse trypsin (PRSS2) in the faeces from GF mice colonized with the indicated bacterial mixtures. c, Images from one experiment are shown. See Supplementary Figure 1 for blot source data. Source data

Extended Data Fig. 3. Initial mechanistic studies of Paraprevotella-mediated trypsin degradation.

a, GF mouse caecal contents were incubated with P. clara 1C4 [P. clara (+)] or medium control [P. clara (−)]. Supernatant samples were collected at the indicated time points and subjected to peptidome analysis. Changes in levels of peptides derived from representative mouse proteins are shown. See Supplementary Table 2 for the complete list of peptides detected. b, c, GF mouse caecal contents (contain high levels of trypsin) were incubated with P. clara 1C4 [P. clara (+)] or medium control [P. clara (−)]. Trypsin (PRSS2) levels were analysed by Western Blot (b) or by trypsin activity assay (c) at the indicated time points. d, His-tagged recombinant mouse trypsin (rmPRSS2) was incubated with P. clara 1C4 cultured in a low-calcium medium (mGAM) or in mGAM supplemented with 1mM Ca²⁺ and degradation of rmPRSS2 was analysed by Western Blot with anti-His-tag antibody. e, rmPRSS2 was incubated with P. clara 1C4 or filtered P. clara 1C4 supernatant, and degradation of rmPRSS2 was analysed by Western Blot with anti-His-tag antibody. f, Protease activity of overnight live P. clara 1C4 culture or filtered P. clara 1C4 supernatant as determined by cleavage of FITC-labelled casein. Trypsin (1 ng μl⁻¹) was used as the positive control. Protease activity was expressed as change in relative fluorescence units (RFU). g, P. clara 1C4 was incubated with rmPRSS2 and then treated with disuccinimidyl sulfoxide (DSSO) cross-linker. The cross-linked interaction complex between rmPRSS2 and P. clara-derived molecules was analysed by Western blot with anti-His tag antibody. P. clara 1C4 without incubation with rmPRSS2 (P. clara 1C4 only) was used as the negative control. b, d, e, g, Representative images from two (d, e) or three (g) independent experiments with similar results, or an image from one experiment (b) are shown. See Supplementary Figure 1 for blot source data. Source data

Extended Data Fig. 4. Shedding of Paraprevotella proteins into the supernatant following treatment with tunicamycin.

a, b, P. clara 1C4 was pre-treated with 2F-Fuc [2F-Fuc (+)] or vehicle control [2F-Fuc (−)] followed by incubation with rmPRSS2. Association of rmPRSS2 with P. clara 1C4 was examined by confocal microscopy (a) and degradation of rmPRSS2 was analysed by Western Blot with anti-His-tag antibody (b). Scale bar: 5 μm (a). c, P. clara 1C4 was treated with tunicamycin or vehicle control, and whole cell lysates were analysed for protein (left) and glycan (right) contents with Colloidal Coomassie Blue staining and Pro-Q Emerald 300 staining, respectively. d, Supernatant proteins from samples in (c) were analysed with Colloidal Coomassie Blue staining. Arrowheads indicate the bands that were decreased (c) or increased (d) after tunicamycin treatment. a–d, Representative images from two independent experiments with similar results are shown. See Supplementary Figure 1 for gel and blot source data.

Extended Data Fig. 5. Type IX secretion system (T9SS) components in Paraprevotella genomes and generation of insertional mutants for P. clara JCM14859.

a, Alignment of T9SS gene components in the genomes of P. clara JCM14859, P. xylaniphila JCM14860 and P. gingivalis ATCC33277. b, Schematic illustration of insertional mutagenesis by plasmid integration. PCR validation results for the indicated mutants are shown. Primers used for mutagenesis and PCR validation are listed in Supplementary Table 5. pLGB30: suicide vector used for cloning and integrating sequences into P. clara JCM14859. b, Images from one experiment are shown. See Supplementary Figure 1 for gel source data.

Extended Data Fig. 6. Generation of gene deletion mutants of P. clara JCM14859, growth of mutants deficient in trypsin degradation and analysis of genes located between 00502 and 00509.

a, Schematic illustration (upper panel) and PCR validation (lower panels) of 03048-03053, 00502 and 00509 gene deletion. Primers used for mutagenesis and PCR validation are listed in Supplementary Table 5. pLGB30: suicide vector used for cloning and integrating sequences into P. clara JCM14859. b, Wild type (WT), Δ00502 or Δ00509 P. clara JCM14589 strains were incubated with recombinant mouse PRSS2 (rmPRSS2) and degradation of rmPRSS2 was analysed by Western Blot with anti-His-tag antibody. c, In vitro growth rate of mutants deficient in trypsin degradation as determined by OD₆₀₀. : d, Alignment of the 00502-00509 gene cluster in Paraprevotella genomes and annotation of each protein with Prokka 1.14.6. e, Wild type (WT) or the indicated mutant strains of P. clara JCM14589 were incubated with rmPRSS2, and degradation of rmPRSS2 was analysed by Western Blot. c, Data shown as mean ± s.d. n = 5 wells of individual bacterial cultures per group. a, b, e, Images from one experiment are shown. See Supplementary Figure 1 for gel and blot source data. Source data

Extended Data Fig. 7. Generation of recombinant PROKKA_00502 (r00502) and PROKKA_00509 (r00509), and assessment of their trypsin-binding and -degrading properties.

a, b, E. coli hosts carrying expression vectors for r00502 or r00509 were treated with IPTG to induce recombinant protein expression (a), and the expressed r00502 or r00509 were purified from cell lysates (b). Protein contents of the whole cell lysates (‘Input’ and ‘Flow through’) or purified recombinants (‘Eluted’) were analysed with Coomassie Blue staining. Arrows indicate protein bands of r00502 or r00509 with the predicted molecular weights. c, Protease activity of r00502 or r00509 as determined by cleavage of FITC-labelled casein. Trypsin was used as the positive control. (-): no protein added. Protease activity was expressed as change in relative fluorescence units (RFU). d, Caecal contents from germ-free (GF) mice were incubated with medium control (-) or beads coupled with recombinant 00502 [00502 (beads)], and ex vivo degradation of trypsin was analysed by Western Blot at the indicated time points with anti-mouse PRSS2 antibody. * Cleaved fragments of PRSS2. e, SDS-PAGE (left) and Native PAGE (right) analysis of the purified r00502. Arrows indicate the monomer (1) and the possible oligomer form (2) of r00502 on a native PAGE gel. f, r00502 was incubated with recombinant human trypsin (hPRSS2, pretreated with trypsin inhibitor AEBSF) at the indicated concentrations at room temperature for 20 min, the reaction mix was analysed by a native PAGE and then subject to Coomassie Blue staining (left) or Western Blot analysis using antibodies against r00502 (anti-His-tag, middle) and hPRSS2 (right). Arrows indicate the bands corresponding to r00502 monomer (1), r00502 oligomer (2), r00502 monomer complexed with hPRSS2 (3) and r00502 oligomer complexed with hPRSS2 complex (4) that were excised for proteomic analysis (Supplementary Table 3). The marker used here is designed for SDS-PAGE-based chemiluminescent Western blot and does not reflect the actual molecular weight on a Native PAGE gel. It was used only for the purpose of alignment of the individual bands between the gel and the blots. g, Native PAGE analysis and Coomassie Blue staining of the recombinant proteins incubated alone or as mixtures at room temperature for 20 min. hPRSS2 was pre-treated with AEBSF to inhibit the trypsin activity. Arrows indicate the migration shifts of the r00502 bands when hPRSS2 was present. * degraded fragment of r00509 by hPRSS2. a, b, d–g, Representative images from two independent experiments with similar results are shown. See Supplementary Figure 1 for gel and blot source data. Source data

Extended Data Fig. 8. Model of Paraprevotella-mediated trypsin degradation and structural predictions of 00502.

a, Model of Paraprevotella-mediated trypsin degradation: 00502 and 00509 proteins are transported across the outer membrane of Paraprevotella via the Type IX secretion system (T9SS). PorU is an essential T9SS component that cleaves the C-terminal domain (CTD) of T9SS-dependent proteins and anchors the proteins to Paraprevotella LPS molecules. WecA mediates the initial step of LPS O-glycan synthesis, and disruption of WecA function (e.g., with tunicamycin treatment) causes release of T9SS-dependent proteins. 00502 acts as a core effector component, facilitating trypsin association and auto-degradation possibly mediated by 00502 oligomerization, whereas 00509 may play a supporting and dispensable role in facilitating trypsin recruitment. Sec: Sec system that exports proteins across the cytoplasmic membrane. b, AlphaFold2-based structural prediction of P. clara 00502 protein with individual domains highlighted. Four out of the five Ig-like domains are shown; the last Ig-like domain that serves as the T9SS C-terminal target domain therefore does not form part of the Ig-like domain clustering zone is omitted here (shown separately in panel g). c–f, AlphaFold2-based structural prediction of P. xylaniphila, P. rara, P. rodentium and P. muris 00502 homologues with the conserved Ig-like domain clustering zone highlighted. g, Alignment of the C-terminal domain (CTD) of P. clara 00502 with that of Porphyromonas gingivalis RgpB protein. The “KXXXK” motif is a signature of T9SS C-terminal target domain-containing proteins.

Extended Data Fig. 9. Trypsin degradation confers P. clara a fitness advantage under competitive conditions.

a–d, Germ-free (GF) mice were colonized with wild type (WT), Δ00502 or Δ00509 P. clara strains together with the 2-mix (B. uniformis 3H3 and P. merdae 1D4) (a & b, left panels, c), or colonized with WT or Δ00502 P. clara together with the 34-mix (a & b, right panels, d) for 14 days. n = 5 and 6 mice per group, respectively. Faecal P. clara DNA levels were determined by qPCR from a standard curve generated from serial dilutions of P. clara genome DNA (a). Fold change of total faecal bacterial DNA (relative to the average of the 2-mix+WT group and that of the 34-mix+WT group, respectively) was determined by a universal bacterial 16S rRNA gene primer pair (b). Faecal DNA of the 3 individual species was quantified by qPCR and their relative abundance was shown as percentage values (DNA of individual strain/total DNA of the 3 strains) (c). Relative abundance of the 35 individual bacterial species was analysed by 16s rRNA sequencing (d). e, Validation of the primers specifically amplifying genomic fragments from WT or Δ00502 P. clara strains for quantifying their abundance in (f, g). f, g, WT and Δ00502 P. clara strains were co-administered together with the 2-mix to GF mice. n = 7 mice. At the indicated days faecal DNA from each P. clara strain was quantified by qPCR. Both the absolute quantities (f) and the relative abundance (percentage of total P. clara DNA) (g) are shown. a, b, f, Data shown as mean ± s.d. **** p < 0.0001; ** p < 0.01; n.s., not significant. One-way ANOVA with Tukey’s test (a & b, left panels), two-sided Mann-Whitney test with Welch’s correction (nonparametric) (a & b, right panels), and two-sided multiple unpaired t tests (not corrected for multiple comparisons) (f). Each dot represents one mouse (a, b). d, Two-sided multiple unpaired t tests (corrected for multiple comparisons using the Sidak-Bonferroni method); *** adjusted p value < 0.001. All primers used for faecal bacterial DNA quantification are listed in Supplementary Table 5. e, An image from one experiment is shown. See Supplementary Figure 1 for gel source data. Source data

Extended Data Fig. 10. Low trypsin levels enhanced the effectiveness of oral vaccines against Citrobacter rodentium in vivo and reduced MHV-2 infection in mouse intestinal organoids.

a, Ex vivo degradation of IgA heavy chain by trypsin: faeces from the 2-mix+WT P. clara-colonized mice (2-mix+WT) and germ-free mice (GF) were diluted and filtered, incubated alone, mixed together (Mixture), or mixed in the presence of trypsin-specific inhibitor TLCK (Mixture+TLCK). Alternatively, faeces from the 2-mix+WT P. clara-colonized mice (2-mix+WT) were incubated with the indicated concentrations of recombinant mouse trypsin (rmPRSS2). After incubation at 37 °C for 24 h the indicated proteins were analysed by Western Blot (left panel, anti-mouse PRSS2 antibody was used to detect both faecal and recombinant mouse PRSS2). Right panel: trypsin activity of the loaded samples (left panel). b, Schematic of the experimental setup for C. rodentium vaccination and infection (c–f). GF mice were inoculated with WT or Δ00502 P. clara JCM14859 strains (together with the 2-mix), orally vaccinated with peracetic acid-inactivated C. rodentium once per week for three weeks, followed by C. rodentium infection via oral gavage. c, Changes in body weight of mice following C. rodentium infection. d, Caecal patches and luminal contents were collected on day 14 post infection and analysed for C. rodentium CFU. e, Western Blot analysis for the indicated proteins in the caecal luminal contents following C. rodentium vaccination and infection. * non-specific band. See Materials & Methods for detection of total and C. rodentium-specific IgA. f, Agglutination effect of the filtered caecal suspension from 2-mix+WT P. clara- and 2-mix+Δ00502 P. clara-colonized mice (following C. rodentium vaccination and infection), as demonstrated by incubation with an in vitro culture of live C. rodentium. g, Relative expression of transmembrane protease, serine 2 (TMPRSS2) and CEA cell adhesion molecule 1 (CEACAM1) in the organoids derived from mouse small intestine and colon was determined by RT-qPCR using β-Actin (ACTB) as the reference gene. h, Colon organoids were infected with MHV-2 at MOI (multiplicity of infection) = 1 in the presence or absence of bovine trypsin for 2 h and washed with DMEM/F12 medium to remove uninfected virus. The viral RNA was quantified by RT-qPCR at 24 hrs post infection. MDCK cell line expressing the canine CEACAM1 with low homology to rodent CEACAM1 was used as the negative control. c, d, n = 7 mice per group. Data shown as mean ± s.d. (c) and geometric mean ± geometric s.d. (d); *** p < 0.001; * p < 0.05; n.s., not significant. Two-sided multiple unpaired t-tests (not corrected for multiple comparisons) (c) and two-sided Mann-Whitney test with Welch’s correction (nonparametric) (d). Each dot represents one mouse (d). g, h, n = 3 wells of cells per group. Each dot represents one well of cells. Data shown as mean ± s.d. * p < 0.05; n.s., not significant. Two-sided unpaired t test (parametric). N. D., not detected (h). Scale bar: 10 μm (f). a, f, Representative images from two experiments with similar results (f), or images from one experiment (a) are shown. e, Images from one experiment including all the mice used in panel c are shown. See Supplementary Figure 1 for blot source data. Source data

Extended Data Fig. 11. Detection of 00502-carrying species in human and mouse microbiome.

a, Computational mining for genes homologous to P. clara 00502-00509 and encoding species. Results of homology search with USEARCH ublast (protein level) against a non-redundant gut microbiome gene catalogue with 5,929,528 genes, constrained to hits with minimum e-value of 0.1. Two metagenomic species (MSPs) annotated to Paraprevotella genus (MSP 0303 and MSP 0335) encoded all or almost all homologues to P. clara genes 00502-00509. 5 MSPs annotated to Bacteroidetes (MSP 0081, MSP 0224, MSP 0288, MSP 0410 and MSP 0435) encoded homologues to genes 00502 and 00509 but lacked homologues to genes 00503-00508. To arrive at these additional MSPs, we interrogated homology hits that showed levels of amino acid identity and coverage similar to that between P. clara and P. xylanphila homologues (see Methods) and were encoded by the same MSP. b, c, Relative abundance (b) and prevalence (c) of the 9 identified human 00502-carrying species across 3372 de novo assembled human gut metagenomes from USA [PRISM (n = 152), HMP2 (n = 1462), FHS (n = 618)], Netherlands [500FG (n = 468), CVON (n = 288)] and China [Jie (n = 384)]. In b, thick horizontal lines indicate the median; box boundaries indicate interquartile range (IQR); whiskers represent values within 1.5 x IQR of the first and third quartiles. Source data

Extended Data Fig. 12. Trypsin degradation by 00502-carrying species.

a, Degradation of recombinant mouse trypsin (rmPRSS2) following in vitro incubation with the indicated bacterial strains. b, Quantification of faecal DNA from Prevotella rodentium and Prevotella muris in SPF mice reared at RIKEN’s facility by RT-qPCR. c, Alexa Fluor 488-labelled rmPRSS2 (green) was incubated with the indicated strains, and association of rmPRSS2 with the bacterial surface was examined by confocal microscopy. Scale bar: 5 μm (c). b, Data shown as geometric mean ± geometric s.d. n = 5 mice. DL, detection limit. N. D., not detected. a, c, A representative image from two independent experiments with similar results (a), or images from one experiment (c) are shown. See Supplementary Figure 1 for blot source data. Source data