Host immunity modulates the efficacy of microbiota transplantation for treatment of Clostridioides difficile infection

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. Transfer of a host’s microbiota by faecal microbiota transplantation has shown benefit in the context of recurrent Clostridioides difficle infection. Here the authors shows the inflammatory status of the recipient can impact on engraftment and the efficacy of the introduced microbiota in a model of C.difficile infection.

T he 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 infection [1][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 disease 5,6 . This theory is supported by animal studies demonstrating alleviation of disease through transplantation of bacteria into the diseased host's intestinal tract 7-12 . Significant advancements have been made in defining the community of microbial species that are important in supporting human health [11][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 dysbiosis 19 . One notable exception is the use of fecal microbiota transplantation (FMT) to treat recurrent Clostridioides difficile infection 20 . FMT treatment of C. difficile infection is the first microbiota-based therapy clinically proven to ameliorate disease 21 . 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 alone [22][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 microbiota 25 . Recurrence of C. difficile infection due to repeated failed antibiotic treatments is estimated as high as 25-35% and can cause potentially lethal fulminant colitis 22,26,27 . 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 infection 21 . 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 USA [28][29][30] .
Intestinal dysbiosis in C. difficile-infected patients is characterized by a loss of bacterial diversity and altered production of microbiota-derived intestinal metabolites 31,32 . Successful FMT restores homeostasis both in the intestinal microbial community and metabolome 33,34 . 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 individuals 35 . The presence of 2°bile acid converting bacteria promotes colonization resistance against C. difficile 13 and restoration of the 2°b ile acid pools is associated with successful FMT in recurrent C. difficile-infected patients [34][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 (T reg ) 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.

Results
FMT mediated clearance of C. difficile infection resolves intestinal inflammation. Fecal microbiota transplantation is a proven treatment for recurrent C. difficile disease 37 , 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 humans 7,21,38 , 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. difficile infection 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 (Rag1 HET ) were compared using our C. difficile infection followed by FMT experimental system (Fig. 1a). In agreement with previous reports [39][40][41] , Rag1 HET 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, Rag1 HET 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. Rag1 HET mice resolved C. difficile infection 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 Rag1 HET and Rag1 −/− mice demonstrated pronounced submucosa edema, cellular infiltration, and crypt elongation in Rag1 −/− mice compared to Rag1 HET mice (Fig. 1e). Following FMT, intestinal inflammation was resolved in Rag1 HET 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 strains 42 . Despite cohabitation and continuous microbial exposure from FMTtreated 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 Rag1 HET and Rag1 −/− mice by bacterial 16S rRNA marker gene profiling on fecal samples from Rag1 HET 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 Rag1 HET 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 Rag1 HET 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 Rag1 HET and Rag1 −/− mice by relative bacterial genera abundance ( Fig. 2b and Supplementary Fig. 3c, d), Uni-Frac distances ( Fig. 2c and Supplementary Fig. 3E-G) and unsupervised hierarchical clustering ( Fig. 2d and Supplementary   Fig. 1 Immunodeficient Rag1 −/− mice exhibit impaired resolution of C. difficile infection following FMT. a Schematic of antibiotic (ABX) treatment, C. difficile infection, and FMT. b Weight loss following infection in littermate, antibiotic-treated Rag HET (n = 6) and Rag1 −/− (n = 5) mice. Data is representative of three independent experiments. Data are presented as mean values ± SEM. c C. difficile burden and d toxin B levels in the fecal pellets of Rag1 −/− (n = 10) and Rag1 HET (n = 14) mice following FMT. Data is a combination of three independent experiments. Statistical significance was calculated by a two-sided unpaired t-test. ***p < 0.0001. Data are presented as mean values ± SEM. e H&E stained cecal tissue sections and f mean crypt length from naïve, C. difficile-infected PBS treated, or C. difficile-infected FMT-treated Rag1 HET and Rag1 −/− mice. n = 3 naïve Rag1 HET and Rag1 −/− mice; n = 3 C. difficileinfected PBS treated Rag1 HET mice; n = 4 C. difficile-infected PBS treated Rag1 −/− mice; n = 6 C. difficile-infected FMT-treated Rag1 HET mice; n = 4 C. difficile-   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 Rag1 HET and Rag1 −/− mice prior to antibiotics as has been previously reported 43 , but community differences were not statistically significant at later timepoints (Supplementary Table 1 Table 2).
16S rRNA marker gene profiling does not yield consistent species-level resolution of microbial communities 44 , and therefore i. Bar plot is displayed at the genus level except for orange bars that represent an ASV aligning to C. difficile. c Unweighted UniFrac distance comparing the microbiota beta diversity dissimilarity within and between C. difficile-infected Rag1 −/− and Rag1 HET groups at day 36 p.i. One-way ANOVA conducted for statistical comparison. Boxes represent median, first and third quartile. Whiskers extend to the highest and lowest data point. d Dendrogram representation of intestinal microbial communities of C. difficile-infected of Rag1 −/− and Rag1 HET mice at day 36 p.i. using unsupervised hierarchical clustering of unweighted UniFrac distances to identify similarities between samples. e mRNA expression of proinflammatory immune response genes in whole colon tissue as assessed by qRT-PCR. Gene expression relative to ABX-treated Rag1 HET mice and normalized to Hprt. Statistical significance was calculated by a two-sided unpaired t-test. *p < 0.05, **p < 0.01. Data are presented as mean values ± SEM. f Proinflammatory cytokine and chemokine protein levels in cecal tissue homogenates. Statistical significance was calculated by a two-sided unpaired t-test. *p < 0.05, **p < 0.01, ***p < 0.001. Data are presented as mean values ± SEM. g Relative abundance of Proteobacteria and gamma-Proteobacteria ASVs in fecal pellets of Rag1 −/− and Rag1 HET mice at day 36 p.i. Statistical significance was calculated by a two-sided unpaired t test. Data are presented as mean values ± SEM. Data presented in figure are representative of three independent experiments. Prior to ABX treatment (n = 10 Rag1 −/− ; n = 12 Rag1 HET mice), day 0 of infection (n = 6 Rag1 −/− ; n = 5 Rag1 HET mice), day 36 post-C. difficile (n = 5 Rag1 −/− and Rag1 HET mice) or ABX-treatment alone (n = 2 Rag1 −/− and Rag1 HET mice). ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-20793-x may not reveal microbial differences that drive differential outcomes following FMT in C. difficile-infected Rag1 HET 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 Rag1 HET 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). Rag1 HET 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 Rag1 HET 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 Rag1 HET 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 Rag1 HET 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 Rag1 HET 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 Rag1 HET mice (Fig. 2f). Gastrointestinal infection-induced inflammation is associated with bystander γ-Proteobacteria bloom in the intestine 45,46 . Indeed 16S rRNA marker gene profiling revealed a trend, though not statistically significant, toward more γ-Proteobacteria in C. difficile infected 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 (β 2 M −/− 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 (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 (T reg ) and T helper 17 (T H 17) cells responding and interacting with cues from the intestinal microenvironment 48,49 . 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 T H 17 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 T H 17 cells in supporting FMT we genetically ablated the signature effector molecules for this T cell subset, IL-17a or IL-22, and assess FMTmediated resolution following C. difficile infection. 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 T H 1 cell differentiation were also capable of resolving C. difficile infection following FMT ( Supplementary Fig. 7C). These data suggest FMT-mediated resolution of C. difficile infection can occur independent of the T H 17 and T H 1 CD4 + T cell lineage.
Depletion of CD4 + T-regulatory cells impairs FMT-mediated resolution of C. difficile infection. In parallel with T H 17 cell expansion, the Foxp3 + and Foxp3 + Rorγt + T reg 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 T reg cells in FMT-mediated resolution of C. difficile, the Foxp3 DTR knockin mice system 50 was used to selectively deplete T reg 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 mice 50 , 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 T reg cells. 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 T reg cells in combination with persistent C. difficile infection reactivated intestinal inflammation. DT treatment in Foxp3 DTR mice did not alter C. difficile burden prior to FMT compared to PBS-treated Foxp3 DTR mice (Fig. 4c). Following FMT, however, DT-treated Foxp3 DTR mice exhibited a significantly delayed resolution of C. difficile infection compared to PBS-treated Foxp3 DTR mice (Fig. 4c). Transient ablation of the large intestinal T reg cells by DT administration was confirmed at day 12 p.i. (Fig. 4d-f), and, in agreement with a previous report 51 , the T reg cell population began to return by day 21 p.i. (day of FMT; Fig. 4d, g). Despite comparable total numbers, the relative proportion of T reg cells compared to T H 17 and T H 1 CD4 + T cell subsets remained diminished in DT-treated Foxp3 DTR mice (Fig. 4h). The repopulated T reg cell compartment was not sufficient to limit intestinal inflammation as absolute numbers of T H 17, T H 1 CD4 + T cells, infiltrating inflammatory monocytes and neutrophils was increased in C. difficile-infected DT-treated Foxp3 DTR mice compared to C. difficile-infected PBS-treated Foxp3 DTR mice (Fig. 4i). Further, C. difficile-infected DT-treated Foxp3 DTR 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 Foxp3 DTR 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 T reg 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 T reg 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 + T reg 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 Rag HET and C57BL/6 mice respectively. The microbiota of C. difficile infected FMT-treated Rag1 HET (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 Rag1 HET 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 Rag1 HET 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 Rag1 HET or Rag1 −/− mice before and after FMT was analyzed for amplicon sequence variants (ASVs) that met the following criteria: (1) present in the FMT  Supplementary Fig. 10).
FMT fails to restore intestinal metabolites in C. difficileinfected, FMT non-responsive mice. Successful FMT treatment in recurrent C.difficile infected patients correlates with restoration of the intestinal metabolites to pre-infection concentrations 34,35 . Therefore, targeted metabolite profiling was conducted on cecal content from naïve, C. difficile-infected, and C. difficile-infected FMT-treated Rag1 Het and Rag1 −/− mice to assess whether FMT bacterial engraftment failure functionally impaired restoration of the intestinal metabolites. Amino acid 52,53 , short chain fatty acid (SCFA) 54,55 , primary and secondary bile acid 13,56-60 metabolites were selected based on reports demonstrating C. difficile colonization is influenced by the availability of these metabolites in the intestine. The metabolite composition in the cecum distinctly shifted between naïve and C. difficile infected mice (Fig. 6a, b), however, no significant difference was observed between Rag1 HET and Rag1 −/− mice prior to FMT (Fig. 6b and Supplementary  Fig. 12A, B). Following FMT, the amino acids, SCFAs, 1°and 2°b ile acids profile of Rag1 HET 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 Rag1 HET 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 Rag1 HET 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 Rag1 HET 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.

Discussion
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 + T reg cells fail to resolved C. difficile infection following FMT. Persistent C. difficile infection in Rag1 −/− mice or Foxp3 + T reg celldepleted 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. Weight loss of C. difficile infected Foxp3 DTR mice following DT (n = 7) or PBS (n = 7) administration. Data are presented as mean values ± SEM. c C. difficile burden in the feces of Foxp3 DTR mice following DT or PBS administration and FMT. n = 9 mice per group. Data is a combination of three independent experiments. Statistical significance was calculated by a two-sided unpaired t-test. *p = 0.015, **p = 0.001, ***p < 0.0001. d Frequency of Foxp3 + CD4 + T cells in the large intestine lamina propria of Foxp3 DTR mice at day 12 p.i. (day 4 post-DT treatment) or day 21 p.i. (day of FMT). n = 4-6 mice per group. FACS plots gated on live, CD45 + , CD3ε + , CD4 + cells. Gating strategy shown in Supplemental Fig. 13. e Total number of Foxp3 + CD4 + T cells and f relative proportion of CD4 + T cell subsets in the large intestine lamina propria at day 12 p.i. (day 4 post-DT treatment). n = 4 PBS-treated Foxp3 DTR mice; n = 5 DT-treated Foxp3 DTR mice. Statistical significance was calculated by a two-sided Mann-Whitney test. ***p < 0.0001. Data are presented as mean values ± SEM. g Total number of Foxp3 + CD4 + T cells and h relative proportion of CD4 + T cell subsets in the large intestine lamina propria at day 21 p.i. (day of FMT). n = 6 PBS & DT-treated Foxp3 DTR mice. Statistical significance was calculated by a two-sided unpaired Mann-Whitney test. Data are presented as mean values ± SEM. i Total number of T H 17 cells, T H 1 cells, inflammatory monocytes, and neutrophils in the large intestine lamina propria at day 21 p.i. (day of FMT). n = 10 uninfected PBS-treated Foxp3 DTR mice; n = 8 uninfected DT-treated Foxp3 DTR mice; n = 10 C. difficile infected PBS-treated Foxp3 DTR mice; n = 10 C. difficile infected DT-treated Foxp3 DTR mice examined over three independent experiments. Statistical significance was calculated by two-sided unpaired t-test. *p = 0.013, **p = 0.002 Data are presented as mean values ± SEM. j mRNA expression of proinflammatory immune response genes in whole colon tissue as assessed by qRT-PCR. Gene expression relative to ABX-treated, uninfected Foxp3 DTR mice and normalized to Hprt. n = 10 uninfected PBS-treated Foxp3 DTR mice; n = 8 uninfected DT-treated Foxp3 DTR mice; n = 10 C. difficile infected PBStreated Foxp3 DTR mice; n = 9 C. difficile infected DT-treated Foxp3 DTR mice examined over three independent experiments. Statistical significance was calculated by two-sided unpaired t-test. *p < 0.05, **p < 0.01. Data are presented as mean values ± SEM.
Foxp3 + T reg 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. Hostderived inflammatory byproducts can be used by pathogens and bystander commensal species alike to promote their growth 61 . 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 species 62,63 while limiting the growth of commensal Clostridia species 64 . Further studies have demonstrated that ethanolamine, nitrates, formate, and lactate metabolites induced by inflammation can support selective bacterial growth in the intestine [65][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 T reg 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 T reg cell population ultimately depriving Enterobacteriaceae species of an oxygen electron acceptor 69 . 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 Rag1 HET 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. difficile colonization in the intestine 13,52-60 . 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 susceptibility 70,71 . Hostderived primary bile acids are germinants for C. difficile spores 72,73 while select commensal bacteria species capable of converting 1°bile acids to 2°bile acids can inhibit C. difficile growth 13,56-60 . 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 T reg cell development in the colon by directly signaling on T reg cells or indirectly signaling on dendritic cells that in turn promote T reg cell development [74][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 T reg cell development at the time FMT. Following FMT, however, restoration of 2°bile acids could promote T reg 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 model 77 . The timing and context of this potential 2°bile acid -T reg cells 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 success 78 . Bacteriophages may directly act on C. difficile itself through lytic processes 79,80 or stimulate immune cell activation in the intestine 81 . The relative contribution of bacteria and the bacteriophages in interacting with the immune system and supporting C. difficile resolution following FMT will require future exploration.
Use of FMT therapy in immunocompromised patient cohorts remains limited due to concerns about adverse effects 82 . Encouragingly, several case reports have described resolution of recurrent C. difficile in HIV + , transplant-recipient, and cancer patients 83 . A multicenter, retrospective study found the FMT success rate in treating recurrent C. difficile in 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 assessed 84 . A subsequent study found the microbiota of immunocompromised hematopoietic stem cell recipients drifted away from the FMT inoculum over time 85 while recipients without overt immune defects exhibit long-term engraftment following FMT 86 . Combined, these reports suggest microbiotabased 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, T H 17 and T H 1 cell compartments did not impact FMT success while a defect in the T reg 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.

Methods
Mice. C57BL/6, Rag1 −/− , μMT −/− , β 2 M −/− , C-II −/− (H2 dlAb1-Ea ), Il17a −/− , Tbx21 −/− , and Foxp3 DTR mice were purchased from the Jackson Laboratory. Il22 −/− mice were provided by R. Flavell (Yale University). All mouse strains were derived on a C57BL/6 background. All mice were bred and maintained in sterile autoclaved cages under specific pathogen-free conditions and kept on a grain-based diet (Labdiet 5053) at the Memorial Sloan Kettering Research Animal Resource Center or the University of Pennsylvania. Mice were provided autoclaved water ad libitum from water bottles. Sex and age-matched controls were used in all difficile-infected Rag1 −/− and C-II −/− mice exhibit impaired FMT engraftment. C. difficile-infected (a-d) Rag1 −/− and Rag1 HET mice or (e-h) C57BL/6 and C-II −/− mice were administered a FMT and microbial composition was analyzed. a Unweighted UniFrac principal coordinate analysis plot of 16S bacterial rRNA ASVs from the FMT inoculum, fecal pellets of C. difficile infected Rag1 −/− and Rag1 HET mice at the time of FMT, and cecal content of Rag1 −/− and Rag1 HET mice at day 21 post-FMT. b Relative abundance of top 15 bacterial ASVs in the microbiota of C. difficile-infected FMT-treated Rag1 −/− and Rag1 HET mice (day 21 post-FMT) Bar plot is displayed at the genus level except for orange bars that represent an ASV aligning to C. difficile. c Unweighted UniFrac distance comparing the microbiota beta diversity dissimilarity of Rag1 −/− and Rag1 HET groups to FMT inoculum. n = 9 C. diff. Rag1 HET ; n = 8 C. diff. Rag1 −/− ; n = 6 FMT Rag1 HET ; n = 4 FMT Rag1 −/− mice. Statistical significance was calculated by one-way ANOVA test using Dunnett method for multiple comparison adjustments. ***p < 0.0001. Boxes represent median, first and third quartile. Whiskers extend to the highest and lowest data point. d Dendrogram representation of intestinal microbial communities of FMT inoculum, Rag1 −/− and Rag1 HET groups using unsupervised hierarchical clustering of unweighted UniFrac distances to identify similarities between samples. e Unweighted UniFrac principal coordinate analysis plot of 16S bacterial rRNA ASVs from the FMT inoculum and fecal pellets from C. difficile infected C57BL/6 and C-II −/− mice at time of FMT and day 15 post-FMT. f Relative abundance of top 15 bacterial ASVs in the microbiota of C. difficile-infected FMT-treated C57BL/6 and C-II −/− mice (day 15 post-FMT). Bar plot is displayed at the genus level except for orange bars that represent an ASV aligning to C. difficile. g Unweighted UniFrac distance comparing the microbiota beta diversity dissimilarity of C57BL/6 and C-II −/− groups to FMT inoculum. n = 4 mice per group . Statistical significance was calculated by a one-way ANOVA test using Dunnett method for multiple comparison adjustments. ***p < 0.0001. Boxes represent median, first and third quartile. Whiskers extend to the highest and lowest data point. h Dendrogram representation of intestinal microbial communities of FMT inoculum, C57BL/6 and C-II −/− groups using unsupervised hierarchical clustering of unweighted UniFrac distances to identify similarities between samples. Data are presented as mean values ± SEM. ***p < 0.001. experiments according to institutional guidelines for animal care. All animal procedures were approved by the Institutional Animal Care and Use Committee of the Memorial Sloan Kettering Cancer Center and University of Pennsylvania.
Antibiotic pretreatment, C. difficile infection, and T reg cell depletion. Two to four month old mice were administered drinking water supplemented with neomycin (Sigma; 0.25 g/L), metronidazole (0.25 g/L), and vancomycin (Novaplus) (0.25 g/L) for 72 h, and then replaced with normal water for the duration of the experiment. For experiments without littermates, mice were cohoused for three weeks prior to antibiotic treatment to allow for equilibration of the microbiota. Forty-eight hours following cessation of antibiotic water, mice received 200 μg of clindamycin (Sigma) by intraperitoneal injection. Twenty-four hours later, mice received~1000 C. difficile spores (CD196, ribotype 027 strain 87 , or VPI10463, ribotype 087 strain ATCC #43255) via oral gavage. After infection, mice were monitored for C. difficile burden in the feces and weight loss. Foxp3 DTR mice were administered diphtheria toxin (20 μg/kg of body weight) or PBS (i.p.) on consecutive days at timepoints indicated.
Quantification of C. difficile burden and toxin. Fecal pellets or cecal content were resuspended in deoxygenated PBS, and ten-fold serial dilutions were plated on BHI agar supplemented with yeast extract, taurocholate, L-cysteine, D-cycloserine, and cefoxitin at 37°C in an anaerobic chamber (Coy Labs) overnight 88 . The presence of C. difficile toxins was determined using a cell-based cytotoxicity assay as previously described 89 . Briefly, human embryonic lung fibroblast WI-38 cells (ATCC# CCL-75) were incubated in a 96-well plate overnight at 37°C. Ten-fold serial dilutions of supernatant from resuspended cecal content was added to WI-38 cells, incubated overnight at 37°C and the presence of cell rounding was observed the next day. The presence of C. difficile toxin B was confirmed by neutralization by antitoxin antisera (Techlab, Blacksburg, VA). The data are expressed as the log 10 reciprocal value of the last dilution where cell rounding was observed.
FMT treatment. The donor source for FMT was fecal pellets from naïve, C57BL/6 mice purchased from Charles River Laboratory and housed under specific pathogen-free conditions. Fecal pellets were collected fresh for each FMT and resuspended at a concentration of 0.2 g/mL of deoxygenated PBS under anaerobic conditions to preserve obligate anaerobe bacteria. FMT inoculum was flash spun to settle food debris and administered to mice via oral gavage (200 μl) and intrarectal instillation (100 μl) into chronically C. difficile-infected mice. Twenty-fours hours following initial FMT, mice receive a second dose of freshly prepared FMT via oral gavage (200 μl). To normalize FMT from experiment to experiment, fecal pellets were collected from the same colony of C57BL/6 mice. Following FMT, unless specified, mice were separated into individually housed cages.
Bacterial DNA extraction, amplification, and analysis. DNA was extracted from fecal pellet using a Qiagen Powersoil kit (Qiagen) according to manufacturer's instructions and 16S rRNA genes were amplified by PCR. Amplicons of the V4-V5 16S rRNA region were amplified and sequenced using an Illumina MiSeq platform as described in Buffie et al. 13 . Raw 16S sequence reads from experiments prior to FMT and with C-II −/− mice (Figs. 2 and 5e-h, and Supplementary Fig. 3 and 8E-H) were quality filtered and clustered into 97% ASVs using UPARSE 90 (USEARCH v11.0667) and ASV representatives were classified using BLAST against the NCBI refseq database as previously described 91 . Unweighted UniFrac distances were calculated using the phyloseq (v. 1.32.0) 92 R package using R version 3.2.1 93 . Relative abundance and PCoA plots were made using ggplot2 (v. 3.2.0) 94 . Weighted UniFrac distances were hierarchically clustered using the R function hclust from the stats package 93 . 16S rRNA sequencing data from experiments post-FMT and the validation cohorts ( Fig. 5a-d, Supplementary Fig. 4, 8A-D, 9, and 11) was processed and analyzed using the QIIME2 pipeline 95 . DADA2 96 , implemented as a QIIME2 plug-in, was used for sequence quality filtering. Taxonomic analysis was done using a Naïve Bayes classifier trained on the Greengenes 13_8 99% ASVs. For diversity metrics including UniFrac distances, a rooted phylogenetic tree was generated: first, a multiple sequence alignment was performed using MAFFT 97 and high variable positions were masked to reduce noise in a resulting phylogenetic tree. A mid-point rooted tree was then generated using FastTree 98 .
Tissue processing, RNA isolation, cDNA preparation, and qRT-PCR. Cecal tissues were fixed with 4% paraformaldehyde, embedded in paraffin and 5 µm sections were cut and stained with hematoxylin and eosin. Crypt length was measured using ImageJ software (1.52q). Three distinct crypt lengths were measure for each sample and averaged. RNA was isolated from proximal colon tissue using mechanical homogenization and TRIzol isolation (Invitrogen) according to the manufacturer's instructions. cDNA was generated using QuantiTect reverse transcriptase (Qiagen). Quantitative RT-PCR was performed on cDNA using Taqman primers and probes in combination with Taqman PCR Master Mix (ABI). Genes of interest were displayed in arbitrary units relative to the expression of Hprt.
Cytokine and chemokine quantification. Cecal tissue was homogenized in 1 mL DMEM media for one minute using a stainless steel bead beater. Homogenates were centrifuged at 10,000 × g for 5 min and supernatant collected. Supernatants were assessed for cytokine & chemokine concentrations using a Luminex bead array (Invitrogen) and concentrations displayed as μg of analyte per gram of cecal tissue.
Intestinal metabolite quantification. Bile acids were quantified using a Waters Acquity uPLC System with a QDa single quadrupole mass detector 99,100 . Briefly, fecal samples were suspended in methanol (5 μL/mg stool), vortexed for 1 min, and centrifuged twice at 13,000 × g for 5 min. The supernatants were analyzed on an Acquity uPLC with a Cortecs UPLC C-18 + 1.6 mm 2.1 × 50 mm column. The flow rate was 0.8 mL/min, the injection volume was 4 μL, the column temperature was 30°C, the sample temperature was 4°C, and the run time was 4 min per sample. Eluent A was 0.1% formic acid in water; eluent B was 0.1% formic acid in acetonitrile; the weak needle wash was 0.1% formic acid in water; the strong needle wash was 0.1% formic acid in acetonitrile. The seal wash was 10% acetonitrile in water. The gradient was 70% eluent A for 2.5 min, gradient to 100% eluent B for 0.6 min, and then 70% eluent A for 0.9 min. The mass detection channels were: +357.35 for chenodeoxycholic acid and deoxycholic acid; +359.25 for lithocholic acid; −407.5 for cholic, alphamuricholic, betamuricholic, gamma muricholic, and omegamuricholic acids; −432.5 for glycolithocholic acid; −448.5 for glycochenodeoxycholic and glycodeoxycholic acids; −464.5 for glycocholic acid; −482.5 for taurolithocholic acid; −498.5 for taurochenodeoxycholic and taurodeoxycholic acids; and −514.4 for taurocholic acid. Samples were quantified against standard curves of at least five points run in triplicate. Standard curves were run at the beginning and end of each metabolomics run. Quality control checks (blanks and standards) are run every eight samples.
Amino acids were quantified using a Waters Acquity uPLC System with an AccQ-Tag Ultra C18 1.7 μm 2.1 × 100 mm column and a Photodiode Detector Array 100,101 . Fecal samples were homogenized in methanol (5 μL/mg stool) and centrifuged twice at 13,000 × g for 5 min. Amino acids in the supernatant were derivatized using the Waters AccQ-Tag Ultra Amino Acid Derivatization Kit (Waters Corporation, Milford, MA) and analyzed using the UPLC AAA H-Class Application Kit (Waters Corporation, Milford, MA) according to manufacturer's instructions. Quality control checks (blanks and standards) were run every eight samples. All chemicals and reagents used were mass spectrometry grade.
Short chain fatty acids were quantified using a Waters Acquity uPLC System with a HSS T3 1.8 μm 2.1 × 150 mm column and a Photodiode Detector Array 99,100 . Fecal samples were homogenized in volatile free fatty acid mix (5 μL/mg stool, AccuStandard, New Haven, CT) and centrifuged twice (13,000 × g for 5 min). The supernatant was filtered through 1.2, 0.65, and 0.22 μm filter plates (Millipore, Billerica, Massachusetts), and the filtrate was loaded into a total recovery vial (Waters, Milford, MA) for analysis. The flow rate was 0.25 mL/min, the injection volume was 5 μL, the column temperature was 40°C, the sample temperature was 4°C, and the run time was 25 min per sample. Eluent A was 100 mM sodium phosphate monobasic, pH 2.5; eluent B was methanol; the weak needle wash was 0.1% formic acid in water; the strong needle wash was 0.1% formic acid in acetonitrile, and; the seal wash was 10% acetonitrile in water. The gradient was 100% eluent A for 5 min, gradient to 70% eluent B from 5 to 22 min, and then 100% eluent A for 3 min. The photodiode array was set to read absorbance at 215 nm with 4.8 nm resolution. Samples were quantified against standard curves of at least five points run in triplicate. Standard curves were run at the beginning and end of each metabolomics run. Quality control checks (blanks and standards) were run every eight samples.
Statistical analysis. Results represent means ± SEM from distinct biological samples. For comparison of C. difficile burden, values were converted into log 10 value and then statistical analysis was performed. Statistical significance was determined using a two-sided unpaired t-test for comparison between two groups. One-way ANOVA used for comparison between multiple groups. Statistical significance was determined using a Mann-Whitney test for small samples size (n < 5 per group) and with a non normal distribution. Statistical analyses were performed using Prism GraphPad software v6.0. (*p < 0.05; **p < 0.01; ***p < 0.001). For 16S sequencing data, in order to test the null hypothesis of no differences in the study group centroids, a PER-MANOVA test was implemented by the function adonis() in the vegan package 2.5-5 [https://CRAN.R-project.org/package=vegan] was used in R 3.6.0. Metabolite profile was analyzed using Metabolanalyst. Log2 transformation and auto scaling of metabolite concentrations were performed to normalize the data. Significant metabolites were identified that met both a two-fold absolute value change threshold and 0.05 p-value using a t-test and adjusted for false discovery rate. Fold changes and adjusted p-values were log transformed and plotted on a volcano plot.