Mice colonized with the defined microbial community OMM19.1 are susceptible to Clostridioides difficile infection without prior antibiotic treatment

ABSTRACT

Diverse gut microorganisms present in humans and mice are essential for the prevention of microbial pathogen colonization. However, antibiotic-induced dysbiosis of the gut microbiome reduces microbial diversity and allows Clostridioides difficile (C. difficile) to colonize the intestine. The Oligo-Mouse-Microbiota 19.1 (OMM19.1) is a synthetic community that consists of bacteria that are taxonomically and functionally designed to mimic the specific pathogen-free mouse gut microbiota. Here, we examined the susceptibility of OMM19.1 colonized mice to C. difficile infection (CDI) at a range of infectious doses (10³, 10⁵, and 10⁷ spores) without prior antibiotic treatment. We found that mice colonized with OMM19.1 were susceptible to CDI regardless of the dose. The clinical scores increased with increasing C. difficile dosage. Infection with C. difficile was correlated with a significant increase in Ligilactobacillus murinus and Escherichia coli, while the relative abundance of Bacteroides caecimuris, Akkermansia muciniphila, Extibacter muris, and Turicimonas muris was significantly decreased following CDI. Our results demonstrate that the OMM19.1 community requires additional bacteria to enable C. difficile colonization resistance.

IMPORTANCE

The human gut microbiota consists of a wide range of microorganisms whose composition and function vary according to their location and have a significant impact on health and disease. The ability to generate and test the defined microbiota within gnotobiotic animal models is essential for determining the mechanisms responsible for colonization resistance. The exact mechanism(s) by which healthy microbiota prevents Clostridioides difficile infection is unknown, although competition for nutrients, active antagonism, production of inhibitory metabolites (such as secondary bile acids), and microbial manipulation of the immune system are all thought to play a role. Here, we colonized germ-free C57BL/6 mice with a synthetic bacterial community (OMM19.1) that mimics the specific pathogen-free mouse microbiota. Following breeding, to enable immune system development, F1 mice were infected with three different doses of C. difficile. Our research suggests that there are additional essential microbial functions that are absent from the current OMM19.1 model.

OBSERVATION

The Oligo-Mouse-Microbiota 19.1 (OMM19.1) synthetic community builds upon the widely used Oligo-Mouse-Microbiota 12 (OMM12) (1), with the addition of nine species and exclusion of Acutalibacter muris and Bifidobacterium animalis (2) (Table 1). This new 19-member community more accurately reflects specific pathogen-free (SPF) mouse microbiota, accounting for the absence of important microbial functions, including 7α-dehydroxylation via the addition of Extibacter muris (2, 3) and compensation for phenotypic differences between OMM12 and SPF mice, including body composition and immune cells in the intestine and associated lymphoid tissues (1). The original OMM12 community provided partial colonization resistance against Salmonella enterica serovar Typhimurium infection (1), as well as partial resistance to Clostridioides difficile infection following the addition of Clostridium scindens (4). C. scindens can modify primary bile acids (a key C. difficile germination trigger) via 7α-dehydroxylation and produce other bile acid-independent mechanisms that inhibit C. difficile outgrowth (5). Here, we examined the resistance of OMM19.1 to C. difficile infection (CDI).

TABLE 1.

Strains in OMM19.1

Strain	DSM number
Akkermansia muciniphila	26127
Bacteroides caecimuris	26085
Blautia pseudococcoides	26115
Enterocloster clostridioformis	26114
Clostridium innocuum	26113
Enterococcus faecalis	32036
Flavonifractor plautii	26117
Limosilactobacillus reuteri	23035
Muribaculum intestinale	28989
Turicimonas muris	26109
Thomasclavelia ramosa	29357
Adlercreutzia mucosicola	19490
Escherichia coli	28618
E. muris	28560
Flintibacter butyricus	27579
Ligilactobacillus murinus	28683
Mucispirillum schaedleri	104751
Parabacteroides goldsteinii	29187
Xylanibacter rodentium	105243

Germ-free C57BL/6 mice were obtained from the Functional Microbiomics, Inflammation, and Pathogenicity Center (University of Louisville) and colonized with pre-assembled frozen stocks available from the German strain collection DSMZ. After breeding, the F1 generation of OMM19.1 mice was used for CDI experiments.

Mice received low (10³ cfu), medium (10⁵ cfu), or high (10⁷ cfu) doses of C. difficile CD2015 (a clinical RT027 isolate) spores via oral gavage, without prior antibiotic treatment. All mice were susceptible to CDI regardless of the dose. Mice receiving medium and high doses lost up to ~12% of their body weight by days 3 and 4, respectively, before recovering weight. A similar weight loss to that was observed in antibiotic-treated SPF mice with this strain (6). Low-dosage mice only lost up to ~5% of their weight on day 2 before recovering (Fig. 1A). Clinical scores correlated with the C. difficile inoculum, with mice receiving the high dose exhibiting significantly higher disease scores than the low-dose mice on day 1 (median [IQR], 1.5 [1] vs. 3 [0.5], P = 0.026). Mice that received the low inoculum recovered more quickly. Mice receiving a medium- or high-dose had significantly higher disease scores than low-dose mice on day 4 (median [IQR], 2 [2.5] vs. 6 [1], P = 0.011, and 2 [2.5] vs. 6 [1.5], P = 0.008, Fig. 1B). On day 1 post-infection (PI), C. difficile levels were significantly different between groups (P = 0.002), with 2.0 × 10⁵ (SD 2.7 × 10⁵), 7.2 × 10⁵ (SD 6.9 × 10⁵), and 7.9 × 10⁶ (SD 5.5 × 10⁶) CFU/g C. difficile in the low-, medium-, and high-dose groups, respectively. No significant difference in C. difficile burden was observed at later time points (Fig. 1C). No mice succumbed to the disease or met the euthanasia criteria.

Fig 1.

Effect of C. difficile inoculation dose on OMM19.1 mice. OMM19.1 mice were challenged with low (10³ cfu, n = 8), medium (10⁵ cfu, n = 7), or high (10⁷ cfu, n = 7) doses of C. difficile. (A) Weight loss: peak weight loss occurred on days 2, 3, and 4 in the low-, medium-, and high-dose groups, respectively, with average body weight losses of 5.1% (SD = 3.1), 12.3% (SD = 4.1), and 11.8% (SD = 4.3). Significant differences in weight were observed on day 3 PI between the low and medium groups (P = 0.028) and the low and high groups (P = 0.028), and on day 4 PI between the low and high groups (P = 0.042). Dashed lines represent mean percent weight, shaded regions indicate standard deviation, and open circles denote individual mice. (B) Disease scores: higher C. difficile doses resulted in persistently elevated disease scores. Significant differences in clinical scores were observed on day 1 PI between the low and high groups (P = 0.026) and on day 4 PI between the low and medium (P = 0.011) and low and high groups (P = 0.008). Filled circles and error bars represent the median and interquartile ranges. Filled circles and error bars represent median and interquartile ranges. (C) C. difficile burden: C. difficile burden initially correlates with inoculum size, but plateaus by day 4 PI. On day 1 PI, a significant difference in C. difficile burden was observed between the low and high (P = 0.004), and medium and high (P = 0.009) groups. The open circles represent individual mice.

To assess microbial composition, DNA was extracted from stool samples before and during CDI, and full-length 16S rRNA gene was sequenced using long-read Oxford Nanopore sequencing. Several OMM19.1 strains were not detected in the stool samples (Fig. 2). This is consistent with the original synthetic community construction by A. Afrizal et al. (2), who reported that M. intestinale was not detected after breeding. T. ramosa (or Clostridium ramosum) and L. reuteri were intermittently detected and observed at only one facility. F. butyricus was not detected in any of the OMM19.1 mice (2). Importantly, given that stool samples were sequenced, it is not possible to rule out the colonization of these strains further up the gastrointestinal tract or at low abundance not sampled by sequencing.

Fig 2.

CDI significantly alters the OMM19.1 community without prior antibiotics. Infection with C. difficile resulted in a significant increase in E. coli and L. murinus while a significant decrease in B. caecimuris, A. muciniphila, E. muris, and T. muris. The dotted line indicates the limit of detection (e.g., no sequences were observed). The mice are represented by individual points (n = 22). **, P-value ≤ 0.01; ***, P-value ≤ 0.001.

CDI significantly correlated with an increase in L. murinus (log2 fold change [log2 FC] of 6.4 compared to uninfected, P < 0.001) and E. coli (log2FC = 7.59, P < 0.001), and a decrease in B. caecimuris (log2FC = −2.83, P < 0.001), A. muciniphila (log2FC = −3.57, P < 0.001), E. muris (log2FC = −3.51, P < 0.001), and T. muris (log2FC = −4.70, P = 0.004; Fig. 2). While these changes are likely caused in part by an increase in intestinal inflammation, direct antagonistic or mutualistic interactions between C. difficile and microbiota cannot be ruled out. Therefore, this model may provide a method to dissect these interactions. Interestingly, the supernatant from L. murinus culture has been demonstrated to significantly enhance the growth of C. difficile (7), while Proteobacteria are consistently increased in humans with CDI (8). Conversely, Bacteroides, A. muciniphila, and E. muris have been inversely associated with C. difficile and may ameliorate CDI (9–14).

Conclusions

The OMM19.1 synthetic community builds upon OMM12 with additional strains covering a greater taxonomic range, including secondary bile acid producers and fiber degraders. The phenotype of OMM19.1 colonized mice has been shown to be more similar to SPF mice both physiologically and immunologically. Colonization resistance is one of the strongest barriers to CDI, although the exact mechanisms remain to be determined. Despite being colonized by a functionally diverse community, OMM19.1 mice were susceptible to CDI. This may be attributed to several factors. Recent studies have suggested that the protective effects of C. scindens against C. difficile may be due to the production of the antimicrobial compound 1-acetyl-β-carboline rather than inhibition by secondary bile acids (15). Furthermore, A. M. Aguirre et al. (16) showed that protection against CDI occurs without the formation of secondary bile acids. C. difficile is capable of Stickland metabolism where proline and glycine are reduced to 5-aminovalerate and acetyl phosphate, respectively. Thus, restoring the gut microbiome with bacteria that consume proline and/or glycine helps promote protection against CDI (16). The OMM19.1 community provides a good starting point for the elucidation of colonization resistance but requires additional microbes.

OMM19.1 mice

The OMM19.1 community consists of 19 strains (Table 1) and was purchased, pre-mixed, from the DSMZ-German Collection of Microorganisms and Cell Cultures. Germ-free C57BL/6 mice were obtained from the Functional Microbiomics, Inflammation, & Pathogenicity Center (University of Louisville) and directly inoculated by gavage (50 µL orally and 50 µL rectally) with the OMM19.1 mixture. To ensure the complete transfer of microbes, inoculation was repeated 72 h after the initial inoculation. OMM19.1 mice were housed under gnotobiotic conditions and mated with the resultant F1 generation used for subsequent infection experiments. To confirm the colonization of the strains, fresh fecal pellets were obtained and frozen at −70°C until needed.

Gnotobiotic housing

Mice were housed in the germ-free/gnotobiotic facility at the CTR vivarium, University of Louisville. Each mouse cage was equipped with its own HEPA filter as the sole air inlet. The lids of the cages were secured with clamps, ensuring an airtight seal. The cages were filled with bedding, food (Teklad 2019S), and empty water bottles, and then sterilized on special racks, allowing steam from the autoclave to penetrate via the HEPA filter. Prior to transferring mice to the sterile cages, water bottles were filled with autoclaved, non-acidified, reverse osmosis water.

To maintain a sterile environment when handling the mice, a BSL-2 biosafety cabinet (BSC) was used. The BSC was sterilized with a chlorine dioxide sterilant for 15–20 minutes before any work began and cages were submerged in a tank of chlorine dioxide for 5 minutes before being transferred into the BSC. Any items used in the BSC were either submerged in chlorine dioxide, placed in a sterilization pouch and autoclaved, or sterilized with ethylene oxide (EtO). Before passing any items into the BSC, the container was sprayed with the sterilant, emptied, and then passed back out.

Personnel working in the BSC wore sterile surgical gowns and gloves, and their hands and arms were sprayed with chlorine dioxide sterilant.

C. difficile infection and enumeration

F1 OMM19.1 mice of both sexes aged 6–8 weeks were challenged with a low (10³), medium (10⁵), or high (10⁷) dose of C. difficile 2015 (RT027) spores by oral gavage (IACUC 24362). C. difficile in stool was enumerated on C. difficile moxalactam norfloxacin agar supplemented with 0.1% sodium taurocholate (CDMNT) (17). The CDMNT plates were incubated anaerobically at 37°C for 48 h prior to enumeration. Mice were weighed and monitored daily for signs of disease throughout the infection period. Clinical scores were determined using the criteria described by R. D. Shelby et al. (18). Briefly, signs of sickness, including behavior, weight loss, and stool consistency, were scored from 0 to 4 with 0 indicating no signs of sickness and 4 indicating the maximum signs of sickness for each group. A total score ≥6 was considered consistent with C. difficile colitis.

16S rRNA gene sequencing

Stool from the mice were taken on days 0 and 3 post-infection. DNA was extracted from frozen fecal pellets using the DNeasy PowerLyzer PowerSoil Kit (Qiagen). DNA samples were sent to SeqCoast Genomics for sequencing using 16S full-length Nanopore reads (10 k reads). Sequencing reads were analyzed with the Oxford Nanopore EPI2ME running the 16S workflow (wf-16s, v1.2.0) with the following settings: minimum reference coverage, 92%; minimum length, 1000 nt; maximum length, 1650 nt; and utilizing pysam (v0.21.0), pandas (v2.0.3), fastcat (v0.15.1), minimap2 (v2.26-r1175), samtools (v1.18), taxonkit (v0.15.1), and Kraken (v2.1.3).

Statistical analysis

The non-parametric Kruskal–Wallis and Wilcoxon rank-sum tests were used for all statistical analyses, with Holm correction for multiple comparisons.

DATA AVAILABILITY

Raw sequencing fastq files for all samples can be accessed via NCBI Accession: PRJNA1165789.

ETHICS APPROVAL

This study was performed in strict accordance with the recommendations described in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. This protocol was approved by the Institutional Animal Care and Use Committee of the University of Louisville (IACUC 21945).