Host immunity modulates the efficacy of microbiota transplantation for treatment of Clostridioides d

Fecal microbiota transplantation (FMT) is a successful therapeutic strategy for treating recurrent Clostridioides difficile infection. Despite remarkable efficacy, implementation of FMT therapy is limited and the mechanism of action remains poorly understood. Here, we demonstrate a critical role for the immune system in supporting FMT using a murine C. difficile infection system. Following FMT, Rag1 heterozygote mice resolve C. difficile while littermate Rag1−/− mice fail to clear the infection. Targeted ablation of adaptive immune cell subsets reveal a necessary role for CD4+ Foxp3+ T-regulatory cells, but not B cells or CD8+ T cells, in FMT-mediated resolution of C. difficile infection. FMT non-responsive mice exhibit exacerbated inflammation, impaired engraftment of the FMT bacterial community and failed restoration of commensal bacteria-derived secondary bile acid metabolites in the large intestine. These data demonstrate that the host’s inflammatory immune status can limit the efficacy of microbiota-based therapeutics to treat C. difficile infection.

The microbiota, consisting of the community of bacterial, viral, fungal, and protozoa organisms that inhabit a mammalian host, can impact susceptibility to a range of diseases including cancer, diabetes, allergy, obesity, and infection1,2,3,4. An altered or dysbiotic intestinal microbiota is observed in all of these disease states suggesting that resetting the intestinal microbiota with a composition of bacteria from healthy individuals could treat disease5,6. This theory is supported by animal studies demonstrating alleviation of disease through transplantation of bacteria into the diseased host’s intestinal tract7,8,9,10,11,12. Significant advancements have been made in defining the community of microbial species that are important in supporting human health11,12,13,14,15,16,17,18. Translating these findings into viable microbiota-based therapies has had limited success in the clinic, however, and is not yet considered a treatment option for most diseases impacted by microbial dysbiosis19. One notable exception is the use of fecal microbiota transplantation (FMT) to treat recurrent Clostridioides difficile infection20. FMT treatment of C. difficile infection is the first microbiota-based therapy clinically proven to ameliorate disease21. Before such treatment strategies can be broadly implemented, the underlying principles determining success for microbiota-based therapy need to be elucidated.

C. difficile, a gram positive, spore-forming, obligate anaerobe is currently the most common nosocomial infection encountered by hospitalized patients, with nearly a half million patients infected and an estimated 13,000–30,000 deaths annually in the USA alone22,23,24. This opportunistic pathogen infects the large intestine following perturbation of the intestinal microbiota. Vancomycin or fidaxomicin are first line treatment options that effectively target the vegetative form of C. difficile, but fail to resolve the underlying condition that promotes infection, a dysbiotic microbiota25. Recurrence of C. difficileinfection due to repeated failed antibiotic treatments is estimated as high as 25–35% and can cause potentially lethal fulminant colitis22,26,27. The high recurrence rate following standard antibiotic treatment has spurred the development of new treatment modalities such as FMT. In 2013, the first controlled, double-blinded study reported the superior efficacy of FMT compared to antibiotics in treatment of recurrent C. difficile infection21. Subsequent studies demonstrate FMT has an 80–90% cure rate in individuals and this treatment option is now incorporated as standard care for recurrent disease in both Europe and the USA28,29,30.

Intestinal dysbiosis in C. difficile-infected patients is characterized by a loss of bacterial diversity and altered production of microbiota-derived intestinal metabolites31,32. Successful FMT restores homeostasis both in the intestinal microbial community and metabolome33,34. For example, following FMT, commensal bacteria that convert host-derived primary (1°) bile acids into secondary (2°) bile acids repopulate the large intestine and 2° bile acid pools are restored to levels observed in healthy individuals35. The presence of 2° bile acid converting bacteria promotes colonization resistance against C. difficile13 and restoration of the 2° bile acid pools is associated with successful FMT in recurrent C. difficile-infected patients34,35,36. These reports provide evidence that efficacy of the FMT is dependent on the bacterial consortium successfully engrafting in the intestine of the infected host to restore the intestinal microenvironment to pre-infection conditions and create an inhospitable environment for C. difficile. The role of host immune factors in supporting FMT engraftment and subsequent clearance of C. difficile has not been explored and is the focus of this report.

In this study, we demonstrate that multiple strains of immunodeficient mice, all of which lack CD4+ Foxp3+ T-regulatory (Treg) cells, exhibit increased intestinal inflammation compared to immunocompetent mice when persistently infected with C. difficile and fail to resolve infection following FMT. The transplanted bacteria from the FMT inoculum do not completely engraft in the large intestine of FMT non-responsive mice and the intestinal metabolite profile is not restored to pre-infection levels. These data reveal an important role for the host immune system in supporting bacterial engraftment and subsequent resolution of C. difficile infection.

FMT mediated clearance of C. difficile infection resolves intestinal inflammation

Fecal microbiota transplantation is a proven treatment for recurrent C. difficile disease37, however, whether the host’s immune system contributes to FMT-mediated resolution of C. difficile is unknown. To address this question, we first established a murine model of C. difficile infection followed by FMT treatment (Fig. 1a). Similar to previous reports in both mice and humans7,21,38, FMT reduced C. difficile burden and toxin levels in the large intestine of persistently infected C57BL/6 mice to below the limit of detection while sham PBS treatment did not impact C. difficile burden (Supplementary Fig. 1A,B). By day 10 post-FMT, recipient mice had resolved C. difficile infection-driven intestinal inflammation characterized by immune cell infiltration, submucosa edema (Supplementary Fig. 1C), large intestine crypt elongation (Supplementary Fig. 1D), and elevated mRNA expression of proinflammatory genes in the proximal colon (Supplementary Fig. 1E). These data establish that FMT can indirectly shape intestinal immune homeostasis via resolution C. difficileinfection and provoke further investigation into the role of the immune system in supporting FMT-mediated C. difficile resolution.

Immunodeficient Rag1 −/− mice exhibit impaired resolution of C. difficile infection following FMT

If FMT resolves C. difficile infection independently of the host immune system, FMT therapy should be equally successful in immunodeficient hosts. To test this null hypothesis and to begin to dissect the role of the immune system in supporting a FMT, T and B cell deficient Rag1−/− mice and littermate control, Rag1 heterozygous mice (Rag1HET) were compared using our C. difficile infection followed by FMT experimental system (Fig. 1a). In agreement with previous reports39,40,41, Rag1HET and cohoused Rag1−/− mice exhibited similar weight recovery from acute C. difficile disease (Fig. 1b) and both groups established a comparable persistent C. difficile infection in the large intestine (Supplementary Fig. 2A). Following establishment of persistent infection, Rag1HET and Rag1−/− mice were administered a FMT, separated into individually housed cages after FMT to prevent microbial cross contamination, and C. difficile was monitored in the feces. Rag1HET mice resolved C. difficileinfection following FMT and had no detectable C. difficile toxin present in its cecal content (Fig. 1c, d). In contrast, C. difficile burden and toxin titers persisted in the large intestine of FMT-treated Rag1−/− mice thus refuting our null hypothesis (Fig. 1c, d). Histological examination of the cecum of C. difficile-infected Rag1HET and Rag1−/− mice demonstrated pronounced submucosa edema, cellular infiltration, and crypt elongation in Rag1−/− mice compared to Rag1HET mice (Fig. 1e). Following FMT, intestinal inflammation was resolved in Rag1HET mice but not in FMT-treated Rag1−/− mice (Fig. 1e, f). These observations suggest that the immune status of the FMT recipient is an important determinant in the ultimate success of FMT therapy.

Two potential confounding factors to this observation are distinct microbial exposure following FMT due to individually housing recipient mice following FMT and the virulence of the C. difficile strain used. Therefore, we assessed the efficacy of FMT in wild-type C57BL/6 and Rag1−/− mice that remained cohoused throughout the course of the experiment using both the CD196 and highly pathogenic VPI10463 C. difficile strains42. Despite cohabitation and continuous microbial exposure from FMT-treated C57BL/6 mice, FMT-treated Rag1−/−mice exhibited significantly higher C. difficile burden (Supplementary Fig. 2B, D) and toxin titers in the feces (Supplementary Fig. 2C, E) compared to C57BL/6 mice in the context of either a CD196 strain infection (Supplementary Fig. 2B, C) or VPI10463 strain infection (Supplementary Fig. 2D, E). These data demonstrate that the differential outcome following FMT in wild-type and Rag1−/− mice is independent of C. difficile strain or cohousing conditions.

C. difficile-infected Rag1 −/− mice exhibit exacerbated intestinal inflammation compared to Rag1 HET mice despite similar microbiota composition

We next analyzed the intestinal microbiota composition of Rag1HET and Rag1−/− mice by bacterial 16S rRNA marker gene profiling on fecal samples from Rag1HET and Rag1−/− mice prior to antibiotic treatment, at the day of infection, and following establishment of persistent infection (day 36 post-infection [p.i.]). The null hypothesis that there was no difference in the microbial community of Rag1HET and Rag1−/− mice was tested to determine if intrinsic differences in the microbiota could explain the differential outcome following FMT. Analysis of unweighted (Fig. 2a), weighted (Supplementary Fig. 3A) UniFrac or Bray-Curtis (Supplementary Fig. 3B) distances between samples did not delineate beta diversity differences in bacterial composition between Rag1HET and Rag1−/− at day 36 p.i., despite significant shifts in microbiota composition of both groups following antibiotic treatment and subsequent C. difficile infection. Comparison of 16S rRNA bacterial community profiles between Rag1HET and Rag1−/− mice by relative bacterial genera abundance (Fig. 2b and Supplementary Fig. 3c, d), UniFrac distances (Fig. 2c and Supplementary Fig. 3E–G) and unsupervised hierarchical clustering (Fig. 2d and Supplementary Fig. 3H) do not lead us to reject the null hypothesis that there was no microbial community level differences between groups. A PERMANOVA test of UniFrac distance indicated a statistically significant difference between the microbiota of Rag1HET and Rag1−/− mice prior to antibiotics as has been previously reported43, but community differences were not statistically significant at later timepoints (Supplementary Table 1). Analysis of the 16S rRNA bacterial gene profile from a second validation cohort of C. difficile infected Rag1HET and Rag1−/− mice at time of FMT (day 32 p.i.) also failed to reject the null hypothesis as determined by unweighted UniFrac distances (Supplementary Fig. 4A), weighted UniFrac distances (Supplementary Fig. 4B), the magnitude of distances between groups (Supplementary Fig. 4C, E), unsupervised hierarchical clustering, and PERMANOVA tests of UniFrac distance between groups (Supplementary Fig. 4D, F and Supplementary Table 2).

16S rRNA marker gene profiling does not yield consistent species-level resolution of microbial communities44, and therefore may not reveal microbial differences that drive differential outcomes following FMT in C. difficile-infected Rag1HET and Rag1−/− mice. To address whether species-level differences at the time of FMT could drive FMT failure in C. difficile infected mice independent of the host’s immune status, the cecal content of C. difficile-infected C57BL/6 or Rag1−/− mice was transferred into antibiotic-treated Rag1−/− or Rag1HET recipient mice, respectively (Supplementary Fig. 5A). Recipient mice established a C. difficile infection derived from the donor transplant, recovered from initial disease morbidity and then received a FMT from naïve C57BL/6 mice donors (Supplementary Fig. 5A). Rag1HET mice receiving the microbiota from a C. difficile infected Rag1−/− mouse were still capable of resolving C. difficile infection following FMT (Supplementary Fig. 5B–D). In contrast Rag1−/− mice receiving the microbiota from a C. difficile infected C57BL/6 mouse failed to clear C. difficile following FMT (Supplementary Fig. 5B–D). Next, germ-free (GF) C57BL/6 mice were cohoused with Rag1−/− or Rag1HET mice and subsequently treated with antibiotics, infected with C. difficile and then administered a FMT (Supplementary Fig. 5E). C. difficile infected ex-GF mice cohoused with either Rag1−/− or Rag1HET mice prior to infection were equally capable of resolving C. difficile infection (Supplementary Fig. 5F). This experimental result demonstrates that the microbiota from naive Rag1−/− mice does not inherently imprint FMT failure in immunocompetent mice. Combined, these data provide evidence that microbiota composition is not intrinsically sufficient to prevent Rag1−/− mice from resolving C. difficile infection following FMT.

Next, the intestinal inflammatory milieu of Rag1HET and Rag1−/− mice was assessed at time of FMT. C. difficile-infected Rag1−/− mice exhibited elevated expression of proinflammatory immune response genes in the colon (Il22, Cxcl1, Cxcl2, Tnfa, Nos2, Reg3g, and Reg3b) compared to C. difficile-infected Rag1HET mice (Fig. 2e). Lack of T and B cells in Rag1−/− mice can conflate interpretation of mRNA transcripts quantification from whole colon tissue. Therefore, total protein levels were measured to assess the in vivo concentration of these proinflammatory cytokines and chemokines regardless of contributing cellular composition. In agreement with gene transcriptional expression, C. difficile-infected Rag1−/−mice had increased protein levels of proinflammatory cytokines and chemokines in the large intestine at time of FMT compared to C. difficile-infected Rag1HET mice (Fig. 2f). Gastrointestinal infection-induced inflammation is associated with bystander γ-Proteobacteria bloom in the intestine45,46. Indeed 16S rRNA marker gene profiling revealed a trend, though not statistically significant, toward more γ-Proteobacteria in C. difficileinfected Rag1−/− mice (Fig. 2g). These data support the hypothesis that FMT failure in Rag1−/− mice is driven by host immunity shaping intestinal environment at the time of FMT engraftment.

CD4+ T cells support FMT-mediated resolution of C. difficile infection

The inability of Rag1−/− mice to resolve C. difficile infection following FMT suggests adaptive immunity is important for FMT-mediated resolution of C. difficile infection. To determine the cellular components of the adaptive immune system that supports FMT, mice deficient in B cells (μMT−/− mice), CD8+ T cells (β2M−/− mice) or CD4+ T cells (MHC Class-II−/− mice) were screened for FMT responsiveness following C. difficile infection in comparison to cohoused C57BL/6 mice. All four groups of mice exhibited similar morbidity and mortality following acute C. difficile infection and exhibited persistent C. difficile colonization (Supplementary Fig. 6A–C,47). Following FMT, C57BL/6, μMT−/−, β2M−/− mice resolved C. difficile infection (Fig. 3a) while MHC Class II−/− (C-II−/−) mice failed to clear C. difficile (Fig. 3b) and maintained high toxin levels in the cecum (Fig. 3c). These data indicate that CD4+ T cells not B cells or CD8+ T cells are necessary for supporting FMT-mediated resolution of C. difficile infection.

Intestinal immune homeostasis is regulated, in part, by the balance of CD4+ T regulatory cells (Treg) and T helper 17 (TH17) cells responding and interacting with cues from the intestinal microenvironment48,49. C. difficile infection elicited a significant expansion in frequency (Fig. 3d) and total number of IL-17a+ competent (Fig. 3e) and IL-22+ competent (Fig. 3f) CD4+ T cells in the large intestine lamina propria. Further, ten days following FMT the TH17 cell population remained significantly elevated relative to naïve mice despite resolution of C. difficile infection at this timepoint (Fig. 3d–f). To test the potential role of TH17 cells in supporting FMT we genetically ablated the signature effector molecules for this T cell subset, IL-17a or IL-22, and assess FMT-mediated resolution following C. difficileinfection. Both Il17a−/− (Supplementary Fig. 7A) and Il22−/− mice (Supplementary Fig. 7B) exhibited equivalent capacity to reduce C. difficile burden following FMT compared to littermate or cohoused wild-type mice. T-bet deficient (Tbx21−/−) mice that lack the capacity for TH1 cell differentiation were also capable of resolving C. difficile infection following FMT (Supplementary Fig. 7C). These data suggest FMT-mediated resolution of C. difficileinfection can occur independent of the TH17 and TH1 CD4+ T cell lineage.

Depletion of CD4+ T-regulatory cells impairs FMT-mediated resolution of C. difficile infection

In parallel with TH17 cell expansion, the Foxp3+ and Foxp3+ Rorγt+ Treg cell significantly increased in the large intestine lamina propria following C. difficile infection and remained elevated following FMT compared to naive mice (Fig. 3g–i). To directly test the role of Tregcells in FMT-mediated resolution of C. difficile, the Foxp3DTR knockin mice system50 was used to selectively deplete Treg cells in the C. difficile infected mice prior to FMT (Fig. 4a). Continued administration of diphtheria toxin (DT) drives systemic autoimmunity and mortality in these mice50, therefore DT or PBS was administered only at day 8 and 9 post-infection, during the recovery from the acute phase of infection, to temporarily delete Tregcells. Diphtheria toxin administration resulted in a temporary 10–15% loss in weight in C. difficile infected mice (Fig. 4b), but minimal weight loss in antibiotic treated, uninfected mice (Supplementary Fig. 7D), suggesting the loss of Treg cells in combination with persistent C. difficile infection reactivated intestinal inflammation. DT treatment in Foxp3DTRmice did not alter C. difficile burden prior to FMT compared to PBS-treated Foxp3DTR mice (Fig. 4c). Following FMT, however, DT-treated Foxp3DTR mice exhibited a significantly delayed resolution of C. difficile infection compared to PBS-treated Foxp3DTR mice (Fig. 4c). Transient ablation of the large intestinal Treg cells by DT administration was confirmed at day 12 p.i. (Fig. 4d–f), and, in agreement with a previous report51, the Treg cell population began to return by day 21 p.i. (day of FMT; Fig. 4d, g). Despite comparable total numbers, the relative proportion of Treg cells compared to TH17 and TH1 CD4+ T cell subsets remained diminished in DT-treated Foxp3DTR mice (Fig. 4h). The repopulated Treg cell compartment was not sufficient to limit intestinal inflammation as absolute numbers of TH17, TH1 CD4+ T cells, infiltrating inflammatory monocytes and neutrophils was increased in C. difficile-infected DT-treated Foxp3DTR mice compared to C. difficile-infected PBS-treated Foxp3DTRmice (Fig. 4i). Further, C. difficile-infected DT-treated Foxp3DTR mice had significantly elevated expression of proinflammatory immune defense genes in the colon at day 21 p.i. compared to C. difficile-infected PBS-treated Foxp3DTR mice (Fig. 4j), similar to the enhanced proinflammatory signature observed in C. difficile infected Rag1−/− mice prior to FMT (Fig. 2e). These results demonstrate that depletion of the Treg cells enhances expression of proinflammatory immune defense genes in the context of C. difficile infection and is sufficient to impair FMT-mediated resolution.

Continued administration of DT administration up through FMT also impaired resolution of C. difficile infection (Supplementary Fig. 7E), while DT treatment initiated immediately prior to FMT did not alter C. difficile resolution (Suppl. Fig 7F). These results indicate immune activation downstream of Treg cell depletion is needed to re-shape the intestinal environment into a state that is refractory to FMT. Combined, the data presented support a role for CD4+ Treg cells in promoting the success of FMT therapy.

FMT fails to engraft in C. difficile infected non-responsive mice

To better understand the etiology of FMT failure, intestinal microbial community profiling was conducted following FMT in C. difficile infected Rag1−/− and C-II−/− mice and compared to RagHET and C57BL/6 mice respectively. The microbiota of C. difficile infected FMT-treated Rag1HET (Fig. 5a–d) or C. difficile infected FMT-treated C57BL/6 mice (Fig. 5e–h) shifted to phenotypically resemble the composition of the FMT inoculum, therby demonstrating successful FMT engraftment. In contrast, the microbiota from C. difficile infected FMT-treated Rag1−/− (Fig. 5a–d) or C. difficile infected FMT-treated C-II−/− mice (Fig. 5e–h) remained distinct from the FMT inoculum. Analyses of weighted UniFrac (Supplementary Fig. 8a, e), or Bray-Curtis (Supplementary Fig. 8B, F) distances confirmed FMT engraftment in Rag1HET and C57BL/6 mice. In contrast, C. difficile infected FMT-treated Rag1−/− and C-II−/−mice failed to engraft the FMT inoculum. Analyses of the microbial community profile by relative bacterial genera abundance (Fig. 5b, f), unweighted or weighted UniFrac distance relative to the FMT (Fig. 5c, g and Supplementary Fig. 8C, G), unsupervised hierarchical clustering (Fig. 5D, H and Supplementary Fig. 8d, h) and PERMANOVA test (Supplementary Table 3) all demonstrated the microbiota of FMT-responsive mice more closely resembled the FMT inoculum than the microbiota of FMT non-responsive mice. C. difficile infected Rag1HET or Rag1−/− mice receiving PBS instead of FMT did not exhibit significant microbiota composition changes (Supplementary Fig. 9A, B) and remained colonized with C. difficile(Supplementary Fig. 9C).

To identify the microbial species that differential engraft in FMT responsive vs. non-responsive mice, the 16S rRNA sequence dataset from C. difficile infected Rag1HET or Rag1−/−mice before and after FMT was analyzed for amplicon sequence variants (ASVs) that met the following criteria: (1) present in the FMT inoculum, (2) absent in C. difficile infected Rag1HETor C. difficile infected Rag1−/− mice prior to FMT, (3) absent in C. difficile infected PBS-treated Rag1HET or Rag1−/− mice, (4) absent in C. difficile infected FMT-treated Rag1−/− mice, and (5) present in C. difficile infected FMT-treated Rag1HET mice. Despite these criteria, 279 ASVs met all five conditions with the majority of ASVs identified not distinguishable at the species levels (Supplementary Fig. 10).

FMT rejection required C. difficile-driven inflammation as ABX-treated, uninfected Rag1−/−mice (Supplementary Fig. 11A) successfully engrafted the FMT inoculum as determined by relative bacterial genera abundance (Supplementary Fig. 11B) unweighted (Fig. 11c), weighted (Supplementary Fig. 11D) UniFrac distances, beta diversity dissimilarity relative to the FMT (Supplementary Fig. 11E, G), and unsupervised hierarchical clustering (Supplementary Fig. 11F, H). Combined, these data demonstrate that the host’s immune activation status can determine which transplanted bacteria successfully engraft in a C. difficile-infected host.

FMT fails to restore intestinal metabolites in C. difficile-infected, FMT non-responsive mice

Successful FMT treatment in recurrent C.difficile infected patients correlates with restoration of the intestinal metabolites to pre-infection concentrations34,35. Therefore, targeted metabolite profiling was conducted on cecal content from naïve, C. difficile-infected, and C. difficile-infected FMT-treated Rag1Het and Rag1−/− mice to assess whether FMT bacterial engraftment failure functionally impaired restoration of the intestinal metabolites. Amino acid52,53, short chain fatty acid (SCFA)54,55, primary and secondary bile acid13,56,57,58,59,60 metabolites were selected based on reports demonstrating C. difficilecolonization is influenced by the availability of these metabolites in the intestine. The metabolite composition in the cecum distinctly shifted between naïve and C. difficileinfected mice (Fig. 6a, b), however, no significant difference was observed between Rag1HETand Rag1−/− mice prior to FMT (Fig. 6b and Supplementary Fig. 12A, B). Following FMT, the amino acids, SCFAs, 1° and 2° bile acids profile of Rag1HET mice but not Rag1−/− mice was restored to resemble the composition of naïve mice (Fig. 6a, b). These data demonstrate that failed engraftment of FMT bacterial species led to functionally impairment at the metabolite level. Following FMT, 2° bile acid derivatives were the most significantly enriched in C. difficile infected FMT-treated Rag1HET mice compared to C. difficile infected FMT-treated Rag1−/− mice (Fig. 6c and Supplementary Fig. 12C). Specifically, deoxycholic acid, lithocholic acid, taurodeoxycholic acid, and omega-muricholic acid concentrations in the cecum of C. difficile infected FMT-treated Rag1HET mice were all restored to resemble the profile of naïve mice (Fig. 6d and Supplementary Fig. 8C). In contrast the 2° bile acid pool remained nearly undetectable in C. difficile infected FMT-treated Rag1−/− mice (Fig. 6d and Supplementary Fig. 12C) while the 1° bile acid pool remained elevated compared to naïve mice (Supplementary Fig. 12D, E). These data are in agreement with the observation that a canonical 2° bile acid converter bacterial specie, Clostridium scindens, engrafted in C. difficile infected FMT-treated Rag1HET mice but not C. difficile infected FMT-treated Rag1−/−mice (Supplementary Figs. 10 and 12F). Similarly, C. difficile infected FMT-treated C-II−/−mice exhibited significantly reduced cecal levels of deoxycholic acid and lithocholic acid compared to FMT-treated C57BL/6 mice (Supplementary Fig. 12G). These data identify intestinal metabolites composition, specifically 2° bile acids, as biomarkers for FMT engraftment failure in mice that exhibit elevated expression of immune defense genes. Combined, this report demonstrates that the immune status of the host is a critical factor in determining FMT therapy success in the context of C. difficile infection.

The clinical success of fecal transplants to treat recurrent C. difficile is an encouraging development for the treatment of this disease as well as other diseases associated with microbiota dysbiosis. The transplant’s mechanism of action in C. difficile infection is postulated to be transplanted bacteria directly altering the intestinal environment to render it inhospitable for C. difficile. In this report, we find that successful engraftment of transplanted bacteria is dependent on the immune status of the recipient, specifically, immunodeficient hosts that lack CD4+ Foxp3+ Treg cells fail to resolved C. difficile infection following FMT. Persistent C. difficile infection in Rag1−/− mice or Foxp3+ Treg cell-depleted mice led to exacerbated induction of inflammatory mediators in the large intestine compared to immunocompetent littermates. These FMT non-responsive mice exhibited impaired engraftment of the donor bacterial populations. Engraftment failure resulted in an inability to restore intestinal metabolite to pre-infection concentrations and ultimately failure to resolve C. difficile infection.

Foxp3+ Treg cells serve an important balance in limiting host inflammation in the intestine. Unchecked, inflammatory mediators produced by the host may inhibit specific transplanted bacteria critical for the resolution of C. difficile infection. Host-derived inflammatory byproducts can be used by pathogens and bystander commensal species alike to promote their growth61. For example, Baumler and colleagues observed that host production of reactive oxygen and reactive nitrogen species induced by Salmonella infection could support expansion of both the pathogen and commensal Enterococcus species62,63 while limiting the growth of commensal Clostridia species64. Further studies have demonstrated that ethanolamine, nitrates, formate, and lactate metabolites induced by inflammation can support selective bacterial growth in the intestine65,66,67,68. Thus, a potential positive feedback loop exist by which the inflammatory environment induced by infection concurrently promotes the survival of select inflammation-tolerant bacterial species and inhibits engraftment of transplanted inflammation-sensitive bacteria. This feedback loop may be potentiated in the absence of functional immunoregulatory mechanisms, such as Treg cells that normally act as checkpoints that limit expansion of inflammation tolerant bacteria. For example, butyrate drives PPAR-γ signaling in epithelial cells and augments the intestinal Treg cell population ultimately depriving Enterobacteriaceae species of an oxygen electron acceptor69. The inflammatory byproducts produced in the intestine of FMT non-responsive hosts with defective immunoregulatory mechanisms may establish an environment that prevents the engraftment of inflammation-sensitive bacterial species. Indeed, in this report we observed 279 unique bacterial ASVs that successfully engrafted in Rag1HET mice but could not engraft in Rag1−/− mice, supporting this concept.

This report conducted a limited intestinal metabolite screen of amino acids, SCFA, 1° and 2° bile acids targeted to metabolites that have known roles in modulating C. difficilecolonization in the intestine13,52,53,54,55,56,57,58,59,60. FMT non-responsive mice failed to restore 2° bile acid pools in the cecum. Primary and secondary bile acid levels are closely linked with C. difficile susceptibility70,71. Host-derived primary bile acids are germinants for C. difficile spores72,73 while select commensal bacteria species capable of converting 1° bile acids to 2° bile acids can inhibit C. difficile growth13,56,57,58,59,60. Our results indicate 2° bile acids may serve as a useful biomarker for successful FMT engraftment and support a role for 2° bile acid restoration as one of the possible mechanisms that contribute to resolution of C. difficile. The metabolite screen performed in this study was limited in scope. A more extensive characterization of intestinal metabolome is needed to understand the relative contribution of bacterial-derived metabolites in FMT success and how they are shaped by the host’s immune system.

Three recent reports observed that secondary bile acid derivatives can promote Treg cell development in the colon by directly signaling on Treg cells or indirectly signaling on dendritic cells that in turn promote Treg cell development74,75,76. Our data demonstrate low to absent 2° bile acid pools in C. difficile infected mice prior to FMT indicating that 2° bile acids are likely not promoting peripheral Treg cell development at the time FMT. Following FMT, however, restoration of 2° bile acids could promote Treg cell expansion and drive a feed forward loop that further reduces intestinal inflammation and enables repopulation of inflammation-sensitive commensal bacteria. FMT has previously been shown to promote the induction of anti-inflammatory properties via boosting IL-10 production in the context of a DSS-colitis mouse model77. The timing and context of this potential 2° bile acid - Tregcells feed forward mechanism in promoting bacterial engraftment will require further study.

This report assessed the ability of bacterial components of a FMT to engraft and resolve C. difficile infection in hosts with different immune statuses. The relative contribution of bacteriophages transplanted in the FMT in resolving C. difficile infection was not assessed. A recent study found donor-derived bacteriophage populations to be associated with FMT success78. Bacteriophages may directly act on C. difficile itself through lytic processes79,80 or stimulate immune cell activation in the intestine81. The relative contribution of bacteria and the bacteriophages in interacting with the immune system and supporting C. difficileresolution following FMT will require future exploration.

Use of FMT therapy in immunocompromised patient cohorts remains limited due to concerns about adverse effects82. Encouragingly, several case reports have described resolution of recurrent C. difficile in HIV+, transplant-recipient, and cancer patients83. A multicenter, retrospective study found the FMT success rate in treating recurrent C. difficilein IBD patients on immunosuppressive therapy was comparable to the FMT success rates reported in IBD-free C. difficile patients though sustained engraftment of the transplanted bacteria was not assessed84. A subsequent study found the microbiota of immunocompromised hematopoietic stem cell recipients drifted away from the FMT inoculum over time85 while recipients without overt immune defects exhibit long-term engraftment following FMT86. Combined, these reports suggest microbiota-based therapeutics for immunocompromised patients is possible, however, which immune pathways are dispensable will require further investigation and may vary depending on the disease being treated. For example, this report found that the immune deficiencies in the B-cell, CD8+ T cell, TH17 and TH1 cell compartments did not impact FMT success while a defect in the Treg cell compartment impaired FMT-mediated resolution of C. difficile infection. Similar to immunosuppressive intervention strategies that reduce the risk of organ transplant rejection, identifying and targeting specific immunological pathways that dictate microbiota transplantation success or failure will be important for the development of more informed clinical approaches that can harness the microbiota to shape host physiology and alleviate disease.