Clostridioides difficile strain 4118, obtained from American Type Culture Collection (www.atcc.org), was cultured anaerobically at 37°C on brain heart infusion agar (Oxoid Ltd., Basingstoke, UK) supplemented with 1.5% (w/v) glucose, 0.1% (w/v) L-cysteine, 1.5% (w/v) agar, and 0.1% (w/v) sodium taurocholate. Thereafter, the strain was inoculated into 500 mL of tryptone yeast broth containing 3% tryptone, 2% yeast extract, and 0.1% sodium thioglycolate and further cultured anaerobically at 37°C for 7 days, after which spores were harvested via centrifugation performed at 1,500 g and 4°C for 20 min, washed six times with ddH₂O, and resuspended in phosphate buffered saline (PBS). Next, heat-shocking was performed at 65°C for 30 min to remove bacteria. The spores were then 10-fold diluted and anaerobically cultured at 37°C for 48 h after, which the number of spores was calculated. C. difficile was isolated from stool samples from the control and infected rodents, and four toxin genes (tcdA, tcdB, cdtA, and cdtB) were identified as previously reported (Indra et al., 2008; Griffiths et al., 2010).

Male C57BL/6 J mice (age, 6–8 weeks) were purchased from Shanghai Laboratory Animal Co., Ltd. (Shanghai, China). Male golden hamsters (age, 6–8 weeks) were purchased from Beijing Weitong Lihua Laboratory Animal Technology Co., Ltd. (Beijing, China), and male Mongolian gerbils (age, 10–12 weeks) were purchased from the Experimental Animal Center of Hangzhou Medical College, China. The animals were housed at the Laboratory Animal Center at Hangzhou Medical College (23°C ± 2°C room temperature, 55% ± 5% relative humidity, and a 12 L:12D photoperiod), at a density of three animals per cage in a specific pathogen-free animal facility. The experimental protocols were approved by the Animal Care and Use Committee of the abovementioned institution (approval number SYXK 2017-0013). Furthermore, all the animal experiments were performed in accordance with Animal Research: Reporting of In Vivo Experiments guidelines (Percie Du Sert et al., 2020). The mice were euthanized 5 days after anesthesia administration via the intraperitoneal injection of chloral hydrate (375 mg/kg of body weight).

All the animals (mice, hamsters, and Mongolian gerbils) were randomized into groups with corresponding average body weights to establish the necessary durations of antibiotic pretreatment. The control and challenge groups for each species comprised five animals. Furthermore, the fresh antibiotic mixture, which comprised kanamycin (0.4 mg/mL), gentamicin (0.035 mg/mL), metronidazole (0.215 mg/mL), colistin (850 U/mL), and vancomycin (0.045 mg/mL), was purchased from Sigma–Aldrich (St. Louis, MO, United States). Once this mixture was prepared, it was added to the drinking water of the animals, which was replaced daily for 3 days. After this 3-day antibiotic pretreatment, the animals were further treated with clindamycin (10 mg/kg) for 1 day, as described previously (Chen et al., 2008) and then challenged with 5 × 10⁴ colony-forming units of C. difficile strain 4,118 via oral gavage. The animals were then monitored for signs of disease, such as diarrhea, wet tail, bow back, malaise, and weight loss. After 5 days, the animals were euthanized, and their stool samples and intestinal tissues were collected for analysis.

TcdB toxins in stool samples from the control and infected rodents were detected via ELISA. The collected stool samples were suspended in 10% (wt/vol) PBS (pH 7.4). Thereafter, the fecal suspensions were seeded into 96-well plates and incubated at 4°C overnight. This was followed by washing with PBST (PBS pH 7.4 + 0.05% tween-20) after which the plate was blocked with 0.5% BSA at 37°C for 2 h. TcdB polyclonal antibodies (List Biological Laboratories Campbell, CA, United States; Chicken IgY, 1:1000 dilution ratio) were used for the first hybridization process, while goat anti-chicken IgY secondary antibody and horse radish peroxidase (Invitrogen, Carlsbad, CA, USA; 1:5000 dilution ratio) were used for the second hybridization. The ELISA-TMB Chromogenic Reagent kit (Sangon, Shanghai, China) was used to determine the absorbance at 450 nm. Pure TcdB toxin (List Biological Laboratories) was used as the positive control, and standard curves were generated.

Resected ileal loops were fixed in 4% formaldehyde, buffered with PBS, and embedded in paraffin. Next, the tissue sections were deparaffinized, stained with hematoxylin and eosin, and subjected to histological analysis. Tissue injuries were scored by two blinded pathologists, who independently graded inflammation according to five grades based on the degrees of chronic inflammation and tissue damage, as described previously (Rugge et al., 2021).

Total genomic DNA was extracted from 0.25 g of intestinal contents using the Dneasy^® PowerSoil^® Pro Kit (QIAGEN, Hilden, Germany), according to the manufacturer’s protocol. DNA concentration and purity were verified via electrophoresis using 1% agarose gel. Furthermore, the genomic DNA was diluted to 1 ng/Μl using sterile water, and the 16S V3–V4 region of the 16S rRNA gene was amplified using specific primers (341F-5′-CCT AYG GGR BGC ASC AG-3′ and 805R-5′-GAC TAC HVG GGT ATC TAA TCC-3′). All the PCR assays were performed using 15 μL of PCR Master Mix (New England Biolabs, Ipswich, MA, United States), 2 μM forward and reverse primers, and approximately 15 ng of template DNA. The PCR program was as follows: denaturation at 94°C for 3 min, 5 cycles of denaturation at 94°C for 30 s, annealing at 45°C for 20 s, elongation at 65°C for 30 s, 20 cycles of denaturation at 94°C for 20 s, annealing at 55°C for 20 s, elongation at 72°C for 30 s, and 72°C for 5 min. The PCR products obtained were purified using the QIAGEN Gel Extraction Kit (QIAGEN).

The TruSeq DNA PCR Free sample preparation kit (Illumina, San Diego, CA, United States) was used to generate the sequencing library, and an index code was added to acquire operational taxonomic unit (OTU) clusters (sequences with ≥97% similarity were assigned to the same OTU). Taxonomic assignment in terms of 16S rRNA was performed based on the SILVA 132 reference database¹ using UCLUST ² at a 90% threshold. Furthermore, taxonomic assignment based on the ribosomal internal transcribed spacer region was performed using the UNITE reference database (https://unite.ut.ee/, version 01.12.2017) in BLAST at an e-value threshold of 0.001. Furthermore, the determination of phylogenetic relationships based on OTUs, differences in dominant species between different samples (groups), and multiple sequence alignments were carried out using PyNAST software version 1.2 (University of Colorado, Boulder, CO, United States).

Mothur ³ was used to determine parameters related to the complexity and diversity of the intestinal microbiota of the animals, such as abundance-based coverage estimators (ACE) and Chao 1 richness. Simpson, Shannon, and beta diversity indices on both weighted and unweighted UniFrac were also calculated using QIIME software. Cluster analysis was performed using principal component analysis, principal coordinate analysis, and non-metric multidimensional scaling (NMDS). Furthermore, linear discriminant analysis effect size (LefSe) was performed using LefSe software, and the filter value for the linear discriminant analysis (LDA) score was set to 2. Furthermore, to investigate the functions and pathways involved with C. difficile challenge in the test animals, Kyoto Encyclopedia of Genes and Genomes (KEGG) functions were predicted using Tax4Fun software, ⁴ and intergroup differences were analyzed in terms of functional abundance differences among the samples.

All data were presented as the mean ± SEM and were compared by performing t-test using GraphPad Prism software version 8 (San Diego, CA, United States). p < 0.05 was considered statistically significant. Furthermore, p-values relative to the control group (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001) were determined via one-way analysis of variance (ANOVA).

Three rodent models with C. difficile challenge were established using the classical intragastric administration method, as shown in Figure 1A. Before challenge, the C. difficile 4,118 strains with four toxin genes were not detected in stool samples of three rodents (Supplementary Figure S1A). On the second day after intragastric C. difficile administration, the stool samples were collected and cultured for C. difficile. The results showed that C. difficile 4,118 strains were obtained from stool samples of three rodents, and four toxin genes were positive (Supplementary Figure S1B). TcdB was also detected in all stool samples (Figure 1B). This observation indicated that C. difficile successfully existed in the intestines of the three rodents and produced the corresponding virulence factors. Based on changes in the body weights of the three rodents and their survival proportions, we found that both Mongolian gerbils and hamsters were susceptible to C. difficile, which showed weight loss (Figure 1C). Specifically, all the Mongolian gerbils challenged with C. difficile exhibited diarrhea (Supplementary Figure S2D). Furthermore, the survival rate of the hamsters was significantly lower than those of the gerbils and mice (Figure 1C). Our results also indicated that no diarrhea in mice challenged with C. difficile (Supplementary Figure S2B), and their weights did not change significantly relative to the weights of the controls (Figure 1C).

Intestinal histopathology changes were detected in all the rodents (Figure 2). The intestinal mucosal cells of the animals in the three control groups were closely packed and were clearly structured, without hyperemia, edema, and inflammatory infiltration. However, after infection, the hamsters exhibited pathological changes, including explosive intestinal bleeding, and quickly died owing to severe enterocolitis. C57BL/6J mice did not manifest diarrhea. Furthermore, the pathological changes in their intestine were not significant; infection only resulted in the irregular arrangement of myometrium cells, with loose cell arrangement and limited inflammatory cell infiltration. Mongolian gerbils exhibited characteristics that increased infiltration of inflammatory cells into intestinal tissues, including neutrophils, increased mucus secretion and significant epithelial cell shedding. This observation suggested that histopathology in Mongolian gerbil is a typical histologic feature of CDI.

Thirty intestinal contents were collected from six rodent groups (five CDI-treated rodents and five control rodents for each species) and subjected to microbiota community analysis using 16S rRNA amplicon sequencing. After quality control, 2,020,374 sequences (40,146 reads per sample) were included for downstream analysis. The overall beta-diversity of gut microbiome signatures in the three challenged and three control rodent groups was analyzed via NMDS. The results revealed different community structures for the gut microbiome in each of the six groups. Samples with high community structure similarities were clustered together, whereas samples with large community structure differences were separated. The NMDS results further showed a separation between the three rodent CDI-treated and control groups and, notably, revealed that the least distance existed between the gut microbiomes of the challenged mouse and gerbil (Figure 3A). This finding suggests that the gut microbiome of mice and Mongolian gerbils possibly became similar following CDI. Furthermore, alpha-diversity indices showed higher phylogenetic diversity in the challenged groups than their corresponding control groups. Notably, hamster gut microbiome showed the greatest changes following C. difficile challenge, Mongolian gerbils, and lastly mice (Figure 3B). The Chao1 and Shannon indices obtained were consistent with the phylogenetic diversity results (Figures 3C–E).

To further analyze the relationship between intestinal flora and CDI, we comparatively explored the dynamic alterations in the composition of the intestinal microbiome in the C57BL/6 J mouse, hamster, and Mongolian gerbil models before and after challenge with C. difficile. At the phylum level, we observed seven bacterial phyla, of which the four most abundant were Firmicutes, Bacteroidetes, Proteobacteria, and Verrucomicrobia, which jointly contributed to an average sequence ratio of approximately 90%. Actinobacteria, Epsilonbacteraeota, Patescibacteria, and Tenericutes only accounted for approximately 6% of the sequences (Figure 4). The gut microbiomes in the three rodent models appeared to be dominated by Firmicutes, followed by Bacteriodetes, while the relative abundances of Proteobacteria, Verrucomicrobia, and Actinobacteria were very low, ranging from 0.03 to 7.8%. Furthermore, after challenge with C. difficile, the relative abundance of Firmicutes in C57BL/6 J mice decreased significantly but was almost unchanged in Mongolian gerbils and hamsters. We also noted that after challenge with C. difficile, the relative abundance of Bacteriodetes in the C57BL/6J mice increased but decreased in the two other species, especially in hamsters, in which Bacteroidetes were hardly detected. The Firmicutes to Bacteriodetes ratio also decreased in C57BL/6 J mice but increased in the other two rodent models, especially in hamsters. Furthermore, the three rodent models, especially hamsters, showed an increase in the relative abundance of Proteobacteria, while the relative abundance of Verrucomicrobia showed a significant increase in C57BL/6 J mice and Mongolian gerbils but was very low in hamsters (undetectable). The relative abundance of Actinobacteria in C57BL/6 J mice and gerbils increased but basically remained unchanged in hamsters.

Consistent with phylum-level data, analyses at the family and genus levels showed that the relative abundances of family and genus varied greatly between the different host species and their controls following CDI. For instance, prior to challenge with C. difficile, Firmicutes in C57BL/6 J mice appeared to be dominated by the family Lactobacillaceae (and the genus Lactobacillus) followed by Lachnospiraceae (and the genus NK4A136_group) (Figures 5A,B). Furthermore, in Mongolian gerbils, Firmicutes predominantly comprised Lachnospiraceae (and the genus NK4A136_group) and Lactobacillaceae (and the genus Lactobacillus), followed by Ruminococcaceae and Erysipelotrichaceae, whereas in hamsters, Firmicutes predominantly comprised Erysipelotrichaceae (and the genera Ileibacterium and Allobaculum) followed by Lachnospiraceae (and the genus NK4A136_group), Lactobacillaceae (and the genus Lactobacillus), and Ruminococcaceae (and the genus Blautia). In the three rodent control groups, Bacteriodetes predominantly comprised the family Muribaculaceae. However, the detection rates for families or genera belonging to Proteobacteria or Verrucomicrobia in the three control groups were very low. After challenge with C. difficile, Firmicutes in C57BL/6J mice predominantly comprised OTUs belonging to the family Lachnospiraceae, followed by Erysipelotrichaceae. Notably, compared with the control C57BL/6 J mice, the relative abundances of the genera Blautia and Lachnoclostridium were higher in challenged mice, while those of the genera Lactobacillus and NK4A136_group were significantly lower. In Mongolian gerbils, the family Lactobacillaceae was the most abundant among Firmicutes. Compared with the corresponding control gerbil group, the relative abundances of genera such as Lactobacillus, Blautia, and Lachnoclostridium were higher, while that of the genus NK4A136_group was very low (undetectable). In hamsters, Firmicutes predominantly comprised OTUs belonging to the family Lactobacillaceae, followed by Erysipelotrichaceae and Lachnospiraceae. Similar to the results obtained for gerbils, compared with the corresponding controls, the genera Lactobacillus and Blautia were more abundant in challenged Mongolian gerbils, while the genus NK4Ahamsters136_group was undetectable. In contrast, the family Erysipelotrichaceae and genera Ileibacterium and Allobaculum were highly abundant in challenged gerbils. Bacteroidetes in C57BL/6J mice and Mongolian gerbils after C. difficile challenge predominantly comprised the family Bacteroidaceae (and the genus Bacteroides), followed by Tannerellaceae. The family Muribaculaceae was undetectable in all three models after C. difficile challenge. Notably, the relative abundances of the families Bacteroidaceae (and the genus Bacteroides) and Tannerellaceae were higher in C57BL/6 J mice than in Mongolian gerbils.

After challenge with C. difficile, Akkermansia (a major member of the family Verrucomicrobiaceae) showed a high relative abundance in C57BL/6 J mice and Mongolian gerbils but was rarely detected in hamsters (Figures 5A,B). Additionally, the three models challenged with C. difficile predominantly comprised the families Burkholderiaceae (and the genus Parasutterella) and Enterobacteriaceae (and the genus Escherichia−Shigella), and the relative abundances of Parasutterella and Escherichia−Shigella were higher in hamsters than in C57BL/6 J mice and Mongolian gerbils, respectively.

The analysis of the top 35 OTUs detected in the three models via heat map clustering before and after challenge with C. difficile showed similar patterns as reported above. Notably, after challenge with C. difficile, C57BL/6 J, clusters predominantly comprising Blautia, Erysipelatoclostridium, Anaerostipes, Robinsoniella, Lachnoclostridium, Bacteroides, Akkermansia, Parabacteroides, and Parasutterella were observed in C57BL/6J mice (Figure 5C). In Mongolian gerbils, the clusters predominantly comprised Blautia, Erysipelatoclostridium, Anaerostipes, Lachnoclostridium, Bacteroides, Akkermansia, Lactobacillus, Clostridioides, Klebsiella, Escherichia−Shigella, Enterococcus, and Proteus, and in hamsters, they predominantly comprised Lactobacillus, Ileibacterium, Allobaculum, Blautia, Parasutterella, Escherichia−Shigella, Enterococcus, and Proteus. These features were further confirmed via LEfSe analysis of the major OTUs detected in these groups (Figure 5D).

KEGG pathway enrichment analysis showed the involvement of multiple metabolic pathways in C. difficile challenge in the three rodent models. Statistically significant differences were also observed among the three models. Difference analysis at the second classification level performed under the conditions: relative abundance >0.1% of the total observed pathways and threshold p-value = 0.05. The results showed 36, 24, and 31 differential pathways in C57BL/6 J mice, Mongolian gerbils, and hamsters, respectively (Figure 6). Among the three groups, the terms “amino acid metabolism,” “membrane transport,” “metabolism of cofactors and vitamins,” “signal transduction,” “nucleotide metabolism,” “energy metabolism,” and “glycan biosynthesis and metabolism” dominated gut microbiome functions. Notably, upregulation of “glycan biosynthesis and metabolism” was observed in the three groups, while the other pathways were only consistent for Mongolian gerbils and hamsters but not for C57BL/6 J mice. Additionally, in C57BL/6 J mice, “amino acid metabolism,” “metabolism of cofactors and vitamins,” and “signal transduction and energy metabolism” were upregulated, while “membrane transport” and “nucleotide metabolism” were downregulated. The identified pathways in the three groups and their functions at the third classification level are shown in detail in Supplementary Figures S3–S5. We also observed that in C57BL/6 J mice, “cysteine and methionine metabolism,” “two-component system,” “oxidative phosphorylation,” “nitrogen metabolism,” “carbon fixation pathways,” and “sulfur metabolism” were upregulated, while “phosphotransferase system (PTS)” was downregulated, but in Mongolian gerbils and hamsters, “phosphotransferase system (PTS)” remained unchanged or showed trends opposite to those observed in C57BL/6 J mice.