C57/BL6 mice were purchased from the Shanghai SLAC Laboratory Animal Company (Shanghai, China). Trpm5-lacZ, ChAT-Ires-Cre and Rosa26-tdTomato (also known as Ai9) mouse lines (Jax stock numbers 005848, 006410 and 007909, respectively) were obtained from the Jackson Laboratory (22, 37, 38). In the Trpm5-lacZ mice, part of the Trpm5-coding region was replaced by the lacZ sequence to express β-galactosidase. ChAT-Ires-Cre mice carrying the internal ribosome entry site and Cre recombinase gene after the choline acetyltransferase (ChAT) coding sequence were crossed with Ai9 to generate ChAT-Ai9 mice in which ChAT-expressing tuft cells are labeled by tdTomato. The Gng13^flox/flox mice were generated previously (39), which were bred with ChAT-Ires-Cre mice to generate conditional Gng13 knockout mice (Gng13-cKO), in which the expression of the Gng13-encoded G protein subunit Gγ13 is abolished in the ChAT-expressing intestinal tuft cells. All mice were bred and maintained in the Laboratory Animal Center of Zhejiang University. All experiments with animals were conducted in accordance with the National Institutes of Health (NIH) guidelines for the care and use of laboratory animals and approved by the Animal Care and Use Committee of Zhejiang University.

Ruminococcus gnavus (R.gnav) (ATCC 29149) was obtained from American Type Culture Collection and cultured in ATCC 158 RGCA medium under anaerobic conditions (80% N₂, 10% H₂, 10% CO₂) at 37°C in an anaerobic chamber for ~18 hours. Bacterial growth curves were determined by counting the number of bacteria at given time points using Hungate’s roll-tube method (40). Cells at the end of the exponential growth phase were collected by three consecutive centrifugations at 4000x g for 40 min: after the first centrifugation, the pellet was resuspended in 1x oxygen-free PBS while the supernatant was subject to a second centrifugation, which was repeated once more. Part of the resuspended R. gnavus cells was used to inoculate mice by oral gavage or to prepare its lysate via ultrasonic crushing. The protein concentration of the lysate was determined using BCA assay and adjusted to 1.3 mg/ml, which was further diluted to a series of working concentrations in the subsequent experiments.

Trichinella spiralis (T. spiralis) strain (isolate code ISS634) was maintained in rodents and collected as previously described (22). About 5,000 muscle larvae in 0.5 ml 1xPBS were ultrasonically treated to prepare the T. spiralis lysate with a protein concentration of ~2.4 mg/ml, which was diluted to 15% (volume/volume) as a working concentration used in the subsequent assays.

R. gnavus bacteria were administered orally to mice. To increase the colonization efficiency of the bacteria, 8-12 weeks-old mice were deprived of food and water for 6 hours before being orally gavaged with 3 x 10⁷ CFUs of R. gnavus. Seven days post the first infection, the mice were administered again with the same amount of R. gnavus. At 21 days post the 1^st infection, colonization of R. gnavus was confirmed by qPCR with the primers for the 16S rRNA genes on the DNA samples extracted from fecal samples. The successfully infected mice were used for experiments.

To stimulate colonic epithelial cells, a mixture of 0.1 ml bitter tasting compounds containing 5 mM salicin and 5 mM quinine was anally administered twice a day for 7 days while control mice were administered with an equal volume of 1xPBS. The mice were sacrificed on day 8 for intestinal tissue analysis.

The processing and immunostaining of intestinal tissue sections were performed as previously described (22). Briefly, the intestines of R. gnavus-infected and uninfected control mice were cut into six segments: duodenum, jejunum, ileum, and proximal, mid and distal colons, and fixed on ice in 4% paraformaldehyde (PFA) for 3 h. The tissues were cryoprotected in 30% (w/v) sucrose overnight and embedded in the O.C.T. Compound embedding medium (Tissue-Tek) before being sliced into 14 μm-thick sections on a cryostat (Thermo scientific, HM525). The sections were then washed in 1x PBS and blocked in the blocking buffer (3% BSA, 0.3% Triton X-100, 2% donkey serum and 0.1% sodium azide in 1x PBS) for 2 h at room temperature. All primary antibodies used in this study were incubated at 4°C overnight and the secondary antibodies were applied at room temperature for 1 h. Primary antibodies used were as follows: anti-Dclk1 (Abcam, ab31704), Alexa Fluor^® 488 Rabbit monoclonal to DCAMKL1 (Abcam,ab202754), anti-PLCß2 (Santa Cruz Biotechnology, sc-206), anti-GSDMC2-GSDMC3 (Abcam, ab229896), anti-CD45 (Abcam, ab10558), anti-Muc2 (Abcam, ab272692), anti-cleaved caspase3 (Cell Signaling Technology, #9664), anti-cleaved caspase8 (Cell Signaling Technology, #8592), and anti-cleaved caspase9 (Cell Signaling Technology, #9509). The specificities of these primary antibodies were validated in the previous studies, respectively (22, 27, 41–47). Secondary antibodies used were as follows: goat anti-rabbit IgG H&L (Alexa Fluor^® 488) (Abcam. ab150077), goat anti-rabbit IgG H&L (Alexa Fluor^® 568) (Abcam, ab175471). DAPI was used for nuclear staining. Fluorescent images were captured using an Olympus FV3000 confocal laser-scanning microscope.

For some tissue sections from the Trpm5-lacZ mice, X-gal staining was used to colocalize Trpm5 with another protein to tuft cells (22). After being labeled with a primary antibody followed by a secondary antibody, sections were washed once in the permeabilization buffer (2 mM MgCl₂, 0.02% NP40 in 1x PBS) and twice in the washing buffer (2 mM MgCl₂ in 1x PBS) for 5 min at room temperature and then rinsed in the detergent solution (2 mM MgCl₂, 0.02% NP40, 0.01% deoxycholate in 1x PBS) for 10 min. The sections were incubated in the β-galactosidase substrate (2 mM MgCl₂, 0.02% NP40, 0.01% deoxycholate, 5 mM K₃Fe (CN)₆, 5 mM K₄[Fe (CN)₆]·3H₂O, 1mg/ml X-gal in 1x PBS) in the dark for 5 h at room temperature. Fluorescent and bright-field images were acquired with a fluorescence microscope (Nikon AZ100).

To detect apoptosis in tissues, we employed TUNEL staining to visualize nuclear DNA fragmentation in the nuclei (48). The intestinal tissue sections were prepared in the same way as described above for immunohistochemistry. The sections were washed twice for 10 min with 1xPBS, and then washed once for 5 min with PBS-T containing 0.5% Triton X-100 at room temperature. One Step TUNEL Apoptosis Assay Kit (Beyotime Biotechnology C1086) was used for TUNEL staining according to the manufacturer’s instructions.

Proximal colons from R.gnavus-infected and -uninfected control mice were dissected out and cut open longitudinally. After the luminal contents were washed out with ice-cold 1xPBS, the proximal colons were cut into small pieces and homogenized for total RNA isolation using TaKaRa MiniBEST Universal RNA Extraction Kit (TaKaRa, 9767). About 0.5–1 µg total RNA was reverse transcribed into first strand cDNA using PrimeScript™ 1st Strand cDNA Synthesis Kit (TaKaRa, 6110A) according to the manufacturer’s instructions. qPCR was set up using primer pairs listed in Supplementary Table 1 and iQ™ SYBR Green Supermix (Bio-rad, 170-8884), and performed on the CFX Connect™ Real-Time System. GAPDH was used as an endogenous “housekeeping” control gene for normalization between different samples, and relative expression levels of target genes were determined using the 2^-ΔΔCt method (49).

Intestinal tissue sections were processed and sectioned as described above for immunohistochemistry. An AB-PAS kit (BBI Life Science, E670107) was used to carry out alnin blue and periodic acid Schiff (PAS) staining according to the manufacturer’s instructions. Briefly, the tissue sections were washed with ultra-pure water, then stained with 1% alnin blue, followed by oxidization with the periodic acid solution and stained with Schiff reagent. Nuclei were stained with hematoxylin, followed by the Scott’s blue solution. Finally, the slides were mounted with neutral balsam mounting medium (Sangon Biotech, E675007) and imaged.

We further optimized a previously reported ELISA method to more sensitively measure stimuli-elicited IL-25 released from the intestinal samples (22). For the small intestinal samples, mouse small intestines were cut open longitudinally, washed with sterilized ice-cold 1xPBS for over 20 minutes. The intestinal villi were scraped with a glass slide, collected, and washed by centrifugation, and resuspended in 1x DPBS containing the P/S antibiotics (100 U/ml penicillin and 100 µg/ml streptomycin). The cell suspensions were washed two more times with the antibiotics-containing 1x DPBS, and then passed through a 70-µm cell strainer. The filtrates were centrifuged and the cell pellets were resuspended in Advanced DMEM/F12 supplemented with 10 mM HEPES, the antibiotics and 1 x GlutaMAX (Gibco, 35050061). The cell suspension then was aliquoted to a 24-well plate with each well containing 1x10⁷ cells, and cultured at 37°C and 5% CO₂ for 10 h for the IL-25 ELISA assays.

The proximal colonic segments were washed with ice-cold 1xPBS, cut into 3-5 mm² pieces, washed in the antibiotics-containing 1xPBS for at least 15 times and then dissociated in 1x DPBS containing 2.5 mM EDTA at 37 °C for 40 minutes, which was terminated by adding DMEM medium to the culture mixtures. The volumes of the mixtures were increased by adding additional antibiotics-containing 1x DPBS, which then were vigorously shaken a few times to release crypt cells from the epithelium before being filtered through a 70-µm cell strainer. The filtrates were centrifuged and resuspended in Advanced DMEM/F12 supplemented with 10 mM HEPES, the antibiotics and 1 x GlutaMAX as described above for the small intestinal samples. About 3 x 10⁶ cells were seeded in each well of a 24-well plate and cultured at 37°C and 5% CO₂ for 10 h for IL-25 ELISA assays.

During the culture of the cell mixtures, stimuli such as the R. gnavus or T. spiralis lysates were added to the cells in the wells of 24-well plates. And to determine the inhibitory effect, the bitter taste receptor inhibitor allyl isothiocyanate (AITC) (Sigma-Aldrich, 57-06-7) at 3 mM, G protein subunit inhibitor Gallein (APExBIO, B7271) at 100 µM, and PLCβ2 inhibitor U73122 (Sigma-Aldrich, U6756) at 30 µM were added to the wells 1 h before the corresponding stimuli were added to the same wells. Ten hours later, cell culture in each well was spun down and the supernatant was transferred to a fresh tube, to which 1x protease inhibitor (cOmplete™, EDTA-free Protease Inhibitor Cocktail, 11873580001, Roche) was added. And the samples were used for ELISA assays or stored at -80°C until use. The amount of the released IL-25 was determined using Mouse IL-17E DuoSet ELISA kits (R&D Systems, DY1399) according to the manufacturer’s instructions. If IL-25 concentration in the sample was below this kit’s low limit of 62.5 pg/ml, another Mouse IL-17E kit with a range from 3.9 to 250 pg/ml (Thermo Fisher Scientific, BMS6046), was used.

The numbers of tuft cells per villus in the small intestinal segments of R. gnavus-infected or uninfected mice were counted while the numbers of tuft cells and goblet cells per crypt in the proximal, mid and distal colons were quantified as well following the previously described methods (26, 50, 51). The size of goblet cells and the fluorescence intensities of immunostaining images were determined using the ImageJ software. The percentages of TUNEL-positive apoptotic cells and the cleaved caspase 3, 8 or 9- positive cells in the proximal colon of R. gnavus-infected or uninfected mice were counted, and data from these experiments as well as from the qPCR and ELISA assays were obtained from at least three independent experiments, each experiments including at least three biological replicates. Data from at least three independent experiments were averaged and presented as means ± SD. Shapiro-Wilk normal distribution was tested. F test was performed for the homogeneity of variance before pairwise comparisons were carried out. Unpaired two-tailed Student’s t-tests were performed. For data of three or more groups, one-way ANOVA was tested using GraphPad Prism 8 software. p values <0.05 were considered statistically significant.

R. gnavus has been reported to be particularly abundant in the gastrointestinal tract of some Covid-19 patients, and possibly associated with diarrhea, asthma, and other symptoms (5) ( Supplementary Figure 1 ). To understand how R. gnavus interacts with host cells in the intestine, we set out to determine whether R. gnavus infection can regulate the number of tuft cells, a rare type of cells that have recently been discovered to detect a number of pathogens, activating innate immune responses and triggering its own hyperplasia (33).

Following two doses of R. gnavus oral gavage on days 0 and 7, the fecal samples of both infected and uninfected control mice were collected on day 21 for DNA extraction and qPCR assays to determine the abundancy of R. gnavus using the PCR primers for the 16S rRNA genes, and then the mice were sacrificed, and their intestines were fixed, sectioned and immunostained with an antibody against a tuft cell marker protein Dclk1. Results showed that the abundancy of R. gnavus was significantly increased in the fecal samples following the oral gavage, indicating the successful inoculation of the microbe ( Supplementary Figure 2A ), and further, the number of tuft cells in the villi of the small intestine: duodenum, jejunum and ileum, did not change following R. gnavus inoculation in comparison with the uninfected control ( Supplementary Figures 2B, C ). In contrast, the number of tuft cells was significantly augmented in the proximal colon of R. gnavus-infected mice comparing with that of the uninfected wild-type control mice whereas the number was not significantly altered in the mid colon, and somewhat decreased in the distal colon ( Figures 1A, B , Supplementary Figure 2D ). Thereafter, this study would be mostly focused on the proximal colon although the effect of R. gnavus infection on the distal colonic tuft cells also warranted further studies.

Previous research has categorized tuft cells into two subtypes: I and II (28, 36). To determine which subtype undergoes expansion in response to R. gnavus infection, we used an antibody against CD45, a biomarker for subtype II tuft cells. Quantitative analyses showed that the numbers of both subtypes of tuft cells increased proportionally following R. gnavus infection, and the percentages of these two cell subtypes did not change ( Figures 1C, D ), indicating that R. gnavus infection can equally effectively evoke proliferation of these two tuft cell subtypes.

Activation of tuft cells from the small intestine and other tissues often results in the release of the cytokine interleukin 25 (IL-25) (22, 25, 27). To determine whether R. gnavus engenders the same effect on the intestinal tuft cells, we performed ELISA assays to evaluate the release of IL-25 from the intestinal cells ( Supplementary Figure 3A ). A series of R. gnavus lysate concentrations were used to stimulate small intestinal tuft cells, and results showed that even the highest concentration tested, i.e., 25% of R. gnavus lysate, did not evoke any detectable increase in IL-25 release whereas a positive control, i.e., the lysate of the parasitic helminth T. spiralis, indeed induced the release of IL-25 from the small intestinal epithelial cells ( Figure 2A ). On the contrary, ELISA data showed that both 15% and 20% R. gnavus lysate were able to stimulate the proximal colonic epithelial cells to release significantly more IL-25 than control while 10% R. gnavus lysate seemed to be able to evoke a perceivable amount of IL-25 as well ( Figure 2B ). Given that microbes synthesize abundant succinic acid, which can activate small intestinal tuft cells to release IL-25 (21), we also performed ELISA assays with succinic acid acting on the proximal colonic samples. The results showed that succinic acid was unable to induce the proximal colonic cells to release IL-25 ( Supplementary Figure 3B ), implying that R. gnavus lysate-elicited IL-25 was not mediated by the succinate receptor (Sucnr1), which is corroborated by recent reports indicating that Sucnr1 is not expressed in the proximal colonic epithelial cells (52, 53).

To determine whether Tas2r receptors are involved in the R. gnavus lysate-induced IL-25 release, we applied the Tas2r inhibitor allyl isothiocyanate (AITC) to the ELISA assay. And the results showed that AITC was able to partially inhibit the IL-25 release ( Figure 2C ), suggesting that R. gnavus lysate may activate Tas2r receptors present on the proximal colonic tuft cells. Since AITC can also act on the Trp A1 ion channel (54), our reanalysis of available single cell RNAseq data (55) showed that Trp A1 was hardly expressed in few transit amplifying cells in the proximal colon, but not in tuft cells that producing IL-25 ( Supplementary Figure 4 ), suggesting that the inhibitory effect of AITC on IL-25 release is likely through inhibiting the activity of Tas2rs expressed on tuft cells.

In an attempt to determine which Tas2rs may respond to R. gnavus lysate, quantitative reverse transcription-PCR was performed to identify Tas2rs of which the expression was upregulated along with tuft cell expansion following R. gnavus infection; and the results indicated that while the expression of a number of Tas2rs were detectable, Tas2r126 and Tas2r129 were significantly upregulated in the R. gnavus-infected proximal colon ( Figure 2D ), supporting the notion that Tas2r receptors may contribute to sensing R.gnavus infection. Since Tas2r126 can be activated by salicin while Tas2r129 is still an orphan receptor (56, 57), we prepared a bitter tastant mixture of salicin with another promiscuous bitter compound, quinine, which can stimulate multiple Tas2rs (58). Anal administration of this mixture to the mice was indeed able to increase the number of tuft cells in the proximal colon ( Figures 2E, F ), supporting Tas2rs’ possible roles in the response of the proximal colonic tuft cells to R. gnavus infection.

Taste signaling proteins have been shown to play important roles in tuft cells’ detection and response to many pathogens. To determine whether these proteins also mediate colonic tuft cells’ interactions with R. gnavus, we set out to characterize the possible roles of three key signaling proteins: heterotrimeric G protein subunit Gγ13, phospholipase Cβ2 (PLCβ2) and transient receptor potential ion channel Trpm5.

Bioinformatical reanalysis of the proximal colonic single-cell RNAseq data (52) indicated that both choline acetyltransferase (ChAT) and Gng13 were co-expressed in the proximal colonic tuft cells ( Supplementary Figures 5A, B ). Immunostaining with the Dclk1 antibody on the ChAT-Ai9 proximal colonic tissue sections indicated that about 83.6% of Dclk1-positive tuft cells were also tdTomato-positive ( Supplementary Figures 5C, D ) while immunostaining with the CD45 antibody showed that about 58.2% and 41.8% of ChAT-Ai9+ cells were CD45- and CD45+ subtypes I and II tuft cells, respectively ( Supplementary Figures 5E, F ), suggesting that ChAT is expressed in both subtypes of tuft cells. The Gng13 conditional knockout (Gng13-cKO) mice in the proximal colonic tuft cells were generated by crossing ChAT-Cre mice with Gng13^Flox/Flox mice and quantitative RT-PCR data showed a significantly reduced expression of Gng13 transcripts in the Gng13-cKO proximal colon ( Supplementary Figure 5G ). To test whether the Gng13-encoded Gγ13 subunit contributes to the R. gnavus-induced tuft cell expansion, qRT-PCR and immunostaining were performed on the Gng13-cKO mice following the R. gnavus inoculation. Results showed that Gng13-cKO abolished the R. gnavus-induced increase in Dclk1 expression ( Supplementary Figure 5H ) while the increase in the number of tuft cells in the R. gnavus-infected proximal colon was insignificant comparing with the uninfected control ( Figures 3A, B , Supplementary Figure 6A ). To assess the G protein βγ subunits’ role in IL-25 release from tuft cells, we carried out IL-25 ELISA assay on WT colonic cells in the presence of G protein βγ subunits’ inhibitor Gallein, or on Gng13-cKO colonic cells. The results indicated that Gallein was able to partially inhibit IL-25 release from WT colonic cells whereas Gng13 conditional knockout also resulted in a significant reduction in IL-25 release ( Figure 3C ).

PLCβ2 is known to be an effector enzyme that generates the second messengers upon activation by the Gβγ moiety in taste receptor cells as well as tuft cells in some tissues (22, 59). Reanalysis of the single proximal colonic cell RNAseq data indicated the expression of PLCβ2 in tuft cells ( Supplementary Figure 6B ) while double immunostaining showed that PLCβ2 is indeed colocalized with Dclk1 ( Figure 3D ). And IL-25 ELISA assays indicated that the PLCβ2 inhibitor U73122 significantly suppressed the IL-25 release from the proximal colonic tuft cells ( Figure 3E ).

The transient receptor potential ion channel Trpm5 is a key signaling protein in bitter, sweet and umami taste transduction in taste buds, and also critical to the detection and response of small intestinal tuft cells to pathogens (22, 25). Analyses of single colonic cell RNAseq data from the showed the expression of Trpm5 in tuft cells, and immunostaining with Dclk1 antibody on Trpm5^+/- heterozygous tissue sections also showed the colocalization of Dclk1 with lacZ staining to tuft cells ( Supplementary Figures 6C, D ). Further analyses indicated that R. gnavus infection did not increase Dclk1 transcripts or the number of Trpm5-expressing tuft cells in the proximal colon ( Figures 3F, G , Supplementary Figure 5H ), and that R. gnavus lysate was unable to induce significantly more IL-25 release from the Trpm5-KO proximal colonic cells comparing with the negative control sample ( Figure 3H ).

To determine whether R. gnavus infection also affects goblet cells’ gene expression, cell turnover or physiological activity, we carried out quantitative RT-PCR, immunohistochemistry and alnin blue and periodic acid Schiff (AB-PAS) staining ( Figure 4 , Supplementary Figure 7 ). qRT-PCR results indicated that R. gnavus infection did not significantly increase the expression of some goblet cells’ marker genes: Spink4, Muc2, Txndc5, Fcgbp and Spdef ( Supplementary Figure 7A ) whereas immunostaining with an anti-Muc2 antibody showed no significant increase in the number of goblet cells either following R.gnavus infection ( Figures 4A, B , Supplementary Figure 7B ). Neither did AB-PAS staining show any changes in goblet cell size ( Figures 4C, D ).

Infection of pathogens can lead to intestinal cell death (42). Analysis of the single-cell RNAseq data showed abundant expression of pyroptotic genes, including Gasdermin C2, C3, and C4 (Gsdmc2, Gsdmc3, Gsdmc4) in a number of colonic cell types ( Supplementary Figure 8 ) (52). To determine whether R. gnavus infection, Gγ13 or Trpm5 contributes to pyroptotic gene expression regulation, we first performed qRT-PCR and found that Gsdmc was indeed not expressed in the proximal colon ( Figure 5A ), which is consistent with the scRNAseq data ( Supplementary Figure 8A ), while the expression of Gsdmc2, Gsdmc3 and Gsdmc4 was strong in WT proximal colons ( Figure 5A ). R. gnavus infection, however, significantly reduced Gsdmc2, Gsdmc3 and Gsdmc4 expression ( Figure 5A ). Conditional knockout of Gng13 or nullification of Trpm5 greatly reduced the expression of all Gsdmc2, Gsdmc3 and Gsdmc4 genes comparing with WT control whereas R. gnavus infection did not further reduce these genes’ expression comparing with their corresponding uninfected controls.

To assess the protein expression levels of these gasdermins in the proximal colon, we performed immunohistochemistry with an antibody to both Gsdmc2 and Gsdmc3. And the results showed that fluorescent signal was strongest on the proximal colonic epithelium from the uninfected WT control mice while R. gnavus infection significantly decreased Gsdmc2/3 expression ( Figures 5B, C , Supplementary Figure 9A ). Both Gng13-cKO and Trpm5-KO uninfected colonic epithelia had much reduced Gsdmc2/3 signal intensity comparing with that of the uninfected WT, and R. gnavus infection did not further reduce the signal intensity in these mutant mice comparing with their respective uninfected knockout controls.

Given that R. gnavus infection and taste signaling proteins altered gasdermin expression, we then set out to determine how the infection affects cell death in the proximal colon. TUNEL assays were performed, which revealed that R. gnavus infection significantly increased the number of TUNEL-positive cells in WT proximal colon, whereas the R. gnavus-uninfected Gng13-cKO and Trpm5-KO mice showed significantly more TUNEL-positive cells than R. gnavus-uninfected WT ( Figures 5D, E , Supplementary Figure 9B ), and these TUNEL-positive cells were located at both the top and bottom of the crypts, suggesting that they include both mature colonocytes and stem cells, which is consistent with the spatial transcriptome analysis of the single-cell RNAseq data ( Supplementary Figure 9C ) (52, 60, 61). Comparing with R. gnavus-infected WT mice, R. gnavus-infected Gng13-cKO and Trpm5-KO mice displayed significantly more TUNEL-positive cells ( Figures 5D, E ).

Apoptosis is known to be activated via two signaling pathways: extrinsic and intrinsic (62). The former is triggered by extracellular death factors, leading to the cleavage of procaspase 8 into active caspase 8 while the latter is by intracellular factors such as genotoxic agents, leading to the modification of procaspase 9 into active caspase 9. Both active caspases 8 and 9 converge on cleaving procaspase 3 into active caspase 3, which degrade cellular proteins, resulting in eventual apoptosis (62). To determine whether both extrinsic and intrinsic signaling pathways contribute to R. gnavus-induced apoptosis, we performed immunohistochemistry with antibodies against each of these three activated caspases: 3, 8 and 9. Results showed that R. gnavus infection indeed activated significantly more caspases 3 and 8 but not 9 in the proximal colon ( Figure 6 , Supplementary Figure 10 ). The uninfected Gng13-cKO and Trpm5-KO mice displayed significantly more activated caspase 3, 8 or 9-positive cells than uninfected WT mice whereas the infection did not significantly alter the number of activated caspases 3, 8 or 9-positive cells in either the Gng13-cKO or Trpm5-KO proximal colon. However, only in the infected Gng13-cKO, but not in the infected Trpm5-KO, more proximal colonic cells with activated caspases 3, 8 and 9 were found than in the infected WT control mice ( Figure 6 ).