Akkermansia muciniphila Muc^T (= ATCC BAA-835^T = CIP 107961^T) was grown anaerobically in a BHI medium at 37°C for 48 h (Derrien et al., 2004). The broth was centrifuged at 8,000 g at 4°C for 10 min, washed, and re-suspended in sterile phosphate-buffered saline (PBS). We used a non-hyper virulent strain of C. difficile (strain VPI 10463) known to produce a high-level of toxin (Merrigan et al., 2010). The C. difficile VPI 10463 (ATCC 43255, binary toxin-negative, toxin A-positive, and toxin B-positive) was cultured anaerobically at 37°C for 24 h in anaerobic conditions in Difco cooked meat medium (BD Diagnostic Systems, United States). The culture was centrifuged (3,200 g, 10 min, 4°C). The sediments were washed twice with sterile PBS and re-suspended for further use (Chen et al., 2008).

Female C57BL/6 mice (6–8 weeks old, SLAC Lab, Shanghai, China) were housed under controlled conditions (specific pathogen free). We used a uniform CDI mouse model (Chen et al., 2008). Briefly, mice were grouped randomly: 0.2 mL PBS [CDI group, n = 13; standard control [NC] group, n = 8] and 0.2 mL A. muciniphila suspension [3 × 10⁹ CFUs, A. muciniphila (AKK) group, n = 8] once daily by oral gavage from day-8 to day 5 (Figure 1A). For C. difficile in vivo infection, AKK group and CDI group received antibiotics cocktail (0.4 mg/mL kanamycin, 0.035 mg/mL gentamicin, 850 U/mL colistin, 0.215 mg/mL metronidazole, and 0.045 mg/mL vancomycin) in drinking water for 5 days followed by 2 days of normal water. Mice were intraperitoneally injected with clindamycin (10 mg/kg) 1 day before infection. On day 0, mice were given oral gavage of C. difficile (10⁸ CFUs, 0.2 mL). The diarrhea was graded as 0, normal stool; 1, loose stool; 2, liquid feces or soiled tail (Tam et al., 2018). According to the manufacturers’ instructions, toxins A and B were detected in feces using an enzyme immunoassay (VIDAS C. difficile Toxin A and B; bioMerieux SA, Marcy-l’Étoile, France). The experiment was approved by the Animal Experimental Ethical Inspection of The First Affiliated Hospital, Zhejiang University School of Medicine (Zhejiang, China) (1-7-2021).

Upon euthanasia, distal colon segments were immediately postfixed in 10% formalin. The tissue sample was embedded in paraffin and stained with hematoxylin and eosin. The degree of enteritis, which uses a scoring system as previously described (Chen et al., 2008), consists of epithelial cell damage (score: 0–3), congestion/edema (score: 0–3), and neutrophil infiltration (score: 0–3). Two independent pathology professors double-blindly evaluated these sections. Alcian blue and periodic acid-Schiff (AB-PAS) staining were performed with the instructions (Solarbio). The sections were stained with p-mTOR, beclin1, zonula occludens-1 (ZO-1), occludin, and claudin-1 antibodies as previously described (Chung et al., 2014). Images were managed using the Zeiss LSM T-PMT confocal microscope (Zeiss, Jena, Germany).

The proximal colon samples were collected and immediately transferred to glutaraldehyde (2.5%) at 4°C for 4 h, postfixed with 1% OsO₄ and dehydrated in graded alcohol concentrations, and embedded in Spurr resin (SPI-CHEM). The specimen was sectioned using the LEICA EM UC7 ultratome (Leica Microsystems GmbH, Wetzlar, Germany), and sections were stained with uranyl acetate (SPI-CHEM) and alkaline lead citrate. These sections were observed by transmission electron microscopy (TEM) (H-7650; Hitachi, Tokyo, Japan).

The limulus amebocyte lysate (LAL) assay detected and quantitated the endotoxins. Serum LAL levels were measured using the LAL Chromogenic Endpoint Assay (Hycult Biotechnology, Uden, Netherlands). Lipopolysaccharide (LPS)-binding protein (LBP) concentrations were detected using an enzyme-linked immunosorbent assay (ELISA) kit (Abcam, Cambridge, MA, United States). The serum and colon cytokines were measured with the Bio-Plex Pro Mouse Cytokine 23-Plex Assay Kit and Chemokine 31-Plex Assay Kit (Bio-Rad, Hercules, CA, United States). The 23-plex contains granulocyte colony-stimulating factor (G-CSF), monocyte chemoattractant protein-1 (MCP-1), interleukin 17A (IL-17A), IL-1α, IL-6, and tumor necrosis factor-alpha (TNF-α). The 31-plex contains: IL-1β, IL-6, IP-10/C-X-C motif chemokine ligand 10 (CXCL10), MCP-1/CCL2, MDC/CCL22, BCA-1/CXCL13, I-309/C-C motif chemokine ligand 1 (CCL1), MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, TARC/CCL17, and TNF-α.

Total RNA of colorectal tissue was reverse transcribed into cDNA using the RNeasy Plus Mini Kit (Qiagen, Valencia, CA, United States) and PrimeScript RT Master Kit (Takara Biomedicals, Kusatsu, Japan). The mRNA relative abundance was measured in triplicate using the Premix Ex Taq (Takara Biomedicals) using the VIIA7 Real-time PCR System (Applied Biosystems, Beverly, MA, United States). The relative mRNA expression was determined using the 2^–ΔΔCT method and normalized to the internal control β-actin. The primer sequences are listed in Supplementary Table 1.

Fecal samples and extracted DNA through the DNeasy PowerSoil Pro Kit (Qiagen). DNA from fecal samples was used in quantitative PCR. Previously reported primers were used to detect and quantify C. difficile (Pereira et al., 2020) and A. muciniphila (Everard et al., 2013). The standard curve used serial dilutions of isolating DNA to quantify the DNA. The samples were tested in duplicate.

DNA from fecal samples on day 6 was extracted by the DNeasy PowerSoil Pro Kit (Qiagen) (NC = 8, CDI = 8, AKK = 8). The 16S rRNA was amplified using primer pairs 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) corresponding to the V3–V4 region. Sequencing was performed using the Illumina MiSeq PE300 system and was analyzed on Majorbio Cloud Platform¹ as previously described (Gu et al., 2020). Briefly, operational taxonomic units (OTUs) were clustered using UPARSE version 7.1 (Edgar, 2013) with a 0.97 threshold. Representative reads of each OTU were selected, and taxonomic data were then assigned using the RDP classifier and display a 70% confidence threshold. Alpha diversity and beta diversity were calculated using QIIME software. Alpha diversity was estimated with the Chao 1 index and Shannon index. Using the Adonis function, we performed a permutational multivariate analysis of variance (PERMANOVA) based on Bray–Curtis distances. Beta diversity was assessed by the Bray–Curtis distance and presented by principal coordinate analysis (PCoA). The differences between groups in taxonomic composition taxa were analyzed using the linear discriminant analysis (LDA) effect size (LEfSe) analysis, and LDA scores greater >4 were defined as discriminative taxa. The sequencing data has been uploaded to the NCBI Sequence Read Archive (SRA) database (PRJNA 766049).

Bile acids (TCA, taurocholic acid; βMCA, β-muricholic acid; CA, cholic acid; αMCA, α-muricholic acid; UDCA, ursodeoxycholic acid; CDCA, chenodeoxycholic acid; DCA, deoxycholic acid; ωMCA, ω-muricholic acid; HDCA, hyodeoxycholic acid; muroCA, murocholic acid; βDCA, β-deoxycholic acid; and LCA, lithocholic acid) in colon contents were detected according to previously described protocol (Xie et al., 2015; Yang et al., 2021) and purchased from Steraloids, Inc. (Newport, RI, United States). Briefly, colon contents were mixed with 200 μL acetonitrile/methanol (volume 8:2), which contained internal standards (GCA-d4, TCA-d4, GDCA-d4, CA-d4, DCA-d4, and LCA-d4). After homogenization and centrifugation (13,500 rpm, 20 min, 4°C), 10 μL supernatant was diluted with 90 μL acetonitrile/methanol (80/20) and ultrapure water mixture solution (volume 1:1). Colon contents BAs were detected by ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) (ACQUITY UPLC-Xevo TQ-S, Waters Corp., Milford, MA, United States).

Short-chain fatty acids in feces were detected according to the previous method (Bian et al., 2020). Stool samples were mixed with 500 μl water (hexanoic acid-d3, internal standard, 10 μg/ml). After homogenization, samples were centrifugated at 15,000 rpm and 4°C for 5 min and mixed the supernatant with the equivalent volume mixture of ethyl acetate/sulfuric acid (v/v = 10:1). After centrifugation and derivatization, the supernatant was run and analyzed by GC/MS (Agilent Technologies, Santa Clara, CA, United States).

GraphPad Prism (GraphPad Software, Inc., CA, United States) and SPSS 20.0 (SPSS, Inc., Chicago, IL, United States) was used for statistical analyses. For statistical significance, either one-way ANOVA followed by the Tukey’s test or Kruskal–Wallis test was used. Kaplan–Meier survival curves used the log-rank (Mantel–Cox) test. Spearman’s rank correlation test analyzed correlations between relevant variables. P < 0.05 was considered statistically significant.

As previously reported, body weight loss during CDI-induced infection (Xu et al., 2018) can be a marker of disease progression. Therefore, we used weight loss during the disease course to measure CDI disease severity. The CDI group showed distinct weight loss after infection (D3, CDI group vs. NC group, P < 0.001; Figure 1C), and five mice died by day 4 (Figure 1B; P < 0.05). All mice in the NC group and AKK group survived. Oral supplementation of A. muciniphila reduced body weight loss after infection (Figure 1C), alleviated the diarrhea score (Figure 1D), relieved colon shortening (NC vs. CDI groups, P < 0.001; AKK group vs. NC group, P < 0.05; AKK group vs. CDI group, P < 0.05; Figures 1E,F). Post-infection fecal samples showed a reduction of C. difficile burden and its toxins with AKK treatment (P < 0.001; Figures 1G,H). Therefore, probiotics significantly reduced CDI-induced clinical symptoms. To establish the mechanism of protection induced by A. muciniphila in CDI- induced mice we studied histopathological analysis and immune response to treat such infection.

Immunohistological staining suggested that A. muciniphila reduced histological injury in the distal colon, such as destroying the histological structure and epithelial barrier (Figure 2A and Supplementary Figure 1). TEM showed that intestinal epithelial cell (IEC) microvilli were ruptured in the CDI group. This disruption across the brush border was reduced in the AKK group (Figure 2A). We speculated that A. muciniphila may play a role by enhancing the mucosal barrier. Immunofluorescence staining and quantitative real-time PCR (qPCR) were used to assess the expression of tight junction proteins (Tjps) and mRNA, respectively. As shown in Supplementary Figure 1A, the AKK group showed stable mucosal integrity of the colon tissue and increased fluorescence intensity of the ZO-1, occludin, and claudin-1 compared to the CDI group. The mRNA expression of Tjps in the colonic tissue of the AKK group was higher than that of the CDI group (occludin, ZO-1/Tjps, claudin-1, P < 0.05 for all; Supplementary Figure 1A). Bacteria and their metabolites, such as LPS, invade the intestines and blood when a leaky gut occurs. LBP is a carrier for LPS and helps its macrophages recognize. ELISA measured serum LBP and LAL to assess the severity of bacterial translocation and inflammation. The results showed that serum LBP and LAL levels in the CDI group were higher than those in the NC and AKK groups (P < 0.05; Figure 2C). In addition, intestinal permeability was evaluated by assessing MUC2 and cannabinoid receptor (CB) levels. CB1 and CB2 had significantly increased expression in the CDI group than the NC group (P < 0.05), meanwhile the expression was downregulated in the AKK group than the CDI group (P < 0.05; Figures 2B,C). We also examined the mucin layer covering the proximal colon by AB-PAS staining in Supplementary Figure 1B. CDI group damaged the mucous layer, while A. muciniphila protected the mucous layer. As noted above, C. difficile caused severe intestinal mucosal injury, while probiotics treatment improved symptoms.

Inflammatory cytokines are related to the pathogenesis mechanism of CDI colitis. We measured serum and colon tissue cytokine levels to explore the protective effects of A. muciniphila on intestinal inflammation. Infected mice were detected with a series of elevated inflammatory cytokines in previous study (Pawlowski et al., 2010), increased levels of serum cytokines G-CSF, MCP-1, IL-17A, IL-1α, IL-6, and TNF-α were observed in CDI colitis mice. While the administration of A. muciniphila significantly reduced the levels of these cytokines (P < 0.05; Figure 3A). In contrast with the NC group, IL-6, TNF-α, IL-1β, CCL1, CCL2, CCL3, CCL4, CCL5, CCL17, CCL22, CXCL10, and CXCL13 were elevated significantly in CDI mice, and are related to intestinal inflammation (P < 0.05; Figure 3B). These cytokine concentrations were downregulated considerably after the administration of A. muciniphila (P < 0.05; Figure 3B). Thus, probiotics reduce inflammation levels and exhibit anti-inflammatory properties.

Due to the mechanism of toxins A/B-induced IEC injury, we investigated the mammalian target of rapamycin (mTOR) signaling pathway to explore intestinal inflammation in CDI mice. The treatment decreased the autophagy-related protein expression (Figure 4A) and mRNA level (Figure 4B). There was reduced expression of phosphorylated mTOR (p-mTOR) and beclin1 in the colon epithelial cells (Figure 4A). The gene expression of autophagy-related genes such as mTOR, beclin1, light chain 3 (LC3)-II, autophagy-related gene 5 (Atg5), Atg7, Atg9a, and Atg12 were more elevated in the CDI group than in the AKK group (P < 0.05; Figure 4B). Similarly, the mRNA expression of immune markers such as Toll-like receptor 4 (TLR4), cluster of differentiation 14 (CD14, a co-receptor of TLR4), and myeloid differentiation 88 (MyD88) were downregulated after A. muciniphila administration (P < 0.001, P < 0.05, and P < 0.05, respectively; Figure 4C).

To study the microbiome composition, we used 16S rRNA gene sequencing and obtained a total of 1,132,433 sequences (47,184 reads per sample). The CDI group showed a decline in microbiome diversity (Shannon index, Figure 5A) and richness (Chao1 index, Figure 5B). Furthermore, the PCoA plot showed significantly different microbiota among the three groups (Figure 5C, PERMANOVA, R² = 0.8065, p_adjust = 0.001). These results suggested the reduced abundance of A. muciniphila in the CDI group (Figure 5D), with a concomitant increased load of A. muciniphila in the AKK group (Supplementary Figure 2A).

The relative taxon abundance at the phylum level showed that the CDI group was enriched with Proteobacteria and depleted of Bacteroidota (Figure 5D). While at the family level, the CDI group considerably lacked Lactobacillaceae, Lachnospiraceae, and Akkermansiaceae but was enriched in Enterobacteriaceae, Enterococcaceae, Clostridiaceae, Morganellaceae, and Peptostreptococcaceae. The oral administration of A. muciniphila significantly improved this enrichment and depletion.

Linear discriminant analysis effect size analysis was used to identify taxa, and the most diverse results (LDA score >4) were analyzed. Compared with the NC group, the phylum Bacteroidota, class Bacteroidia, order Bacteroidales, and families Rikenellaceae, Muribaculaceae, and Prevotellaceae in the CDI group were depleted (Supplementary Figure 2B). Meanwhile, the phylum Actinobacteriota and the affiliated class Actinobacteria, order Bifidobacteriales, family Bifidobacteriaceae, and genus Bifidobacterium were also depleted in the CDI group. At the family level, Lachnospiraceae, Lactobacillaceae, and Erysipelotrichaceae in the CDI group were also reduced compared with the NC group. At the same time, the enrichment of families Peptostreptococcaceae, Enterococcaceae, Enterobacteriaceae, and Morganellaceae in the CDI group led to an imbalance of intestinal flora.

Compared with the CDI group, the AKK group showed depletion of opportunistic pathogens in the Enterococcus (belonging to family Enterococcaceae and order Lactobacillales), Escherichia–Shigella (belonging to family Enterobacteriaceae, order Enterobacterales, class Gammaproteobacteria and phylum Proteobacteria), Clostridioides, and Clostridium_sensu_stricto_1 (Figure 5E). In addition, the AKK group was enriched for the genera Blautia (belonging to family Lachnospiraceae and order Lachnospirales), Parabacteroides (belonging to family Tannerellaceae, order Bacteroidales, class Bacteroidia, and phylum Bacteroidota), and Akkermansia compared with the CDI group, showing the role of A. muciniphila in restoring gut microbiota. Next, we studied the biochemical mechanisms associated with CDI’s A. muciniphila mediated protection.

We detected the BA levels in the colon by UPLC-MS/MS. Partial least squares-discriminant analysis (PLS-DA) showed distinct clustering of the BAs among the three groups (Figure 6A), which showed alteration of BA metabolism by CDI. The CDI group showed decreased secondary BAs, DCA, ωMCA, HDCA, muroCA, βDCA, LCA, and UDCA (Figure 6C). Compared to the NC group, the levels of primary BAs were significantly raised in the CDI group. However, the level of secondary BA was restored, and primary BAs such as CDCA and CA were decreased after A. muciniphila administration (Figure 6C). The AKK group had an increased secondary BA to primary BA ratio than the CDI group (Figure 6B).

As shown in Figure 6D, we determined six major SCFAs in the fecal samples. The AKK group showed a significant increase in SCFA production. The acetic acid, propionic acid, and butyric acid concentrations were significantly decreased in the CDI group (CDI vs. NC: P < 0.05; Figure 6D). Furthermore, A. muciniphila administration increased the concentrations of acetic acid, propionic acid, isobutyric acid, and butyric acid (CDI vs. AKK: P < 0.05; Figure 6D). Thus, probiotics alter metabolic disorders caused by disease.

We performed Spearman’s correlation analysis between discriminative bacteria abundance and metabolic and serum inflammatory indexes to explore the relationship between the changed gut microbiota structures and disease severity. Our results showed that the relative abundance of A. muciniphila-modified bacteria was related to inflammatory indexes and mucosal barrier indicators (Figure 7). Potentially beneficial bacteria such as Bifidobacterium, Bacteroides, and Lactobacillus were negatively correlated with cytokines (e.g., IL-1α, IL-6, and TNF-α) but were positively correlated with SCFAs and secondary BAs. Opportunistic pathogens such as Enterococcus and Escherichia–Shigella were negatively correlated with SCFAs and secondary BAs but positively correlated with inflammatory indexes and primary BAs.