Abstract

Introduction

For decades, fecal microbiota transplantation (FMT) from heterologous donors has been successfully used to eradicate the causative toxin-producing bacterium in patients with intestinal Clostridioides difficile infections (CDI)(Quraishi et al., 2017; Tvede and Rask-Madsen, 1989; van Nood et al., 2013). This success has sparked interest in extending its application to other diseases affecting the gastrointestinal tract, such as inflammatory bowel disease (IBD)(Allegretti et al., 2019; Paramsothy et al., 2017; Weingarden and Vaughn, 2017), graft versus host disease (GVHD)(Kakihana et al., 2016), metabolic syndrome(Vrieze et al., 2012), and immune checkpoint inhibitor associated colitis (ICIAC)(Wang et al., 2018). In contrast to CDI, where FMT successfully treats >90% of cases (Quraishi et al., 2017; van Nood et al., 2013), FMT application in IBD faces significant challenges due to complex disease etiology and the absence of a clear microbial target. We and others have shown that FMT promotes symptoms resolution (also termed clinical remission) and mucosal healing in ulcerative colitis (UC), a major form of IBD(Costello et al., 2019; Fuentes et al., 2017; Imdad et al., 2018; Moayyedi et al., 2015; Paramsothy et al., 2017). However, the benefits were limited to a subset of patients (Paramsothy et al., 2017). This heterogeneity highlights the urgent need to identify mechanisms and define predictive markers to forecast FMT treatment response prior to the initiation of therapy.

In a large multi-center, double-blind, randomized, placebo-controlled FMT trial in UC (the FOCUS study), steroid-free clinical remission was achieved in 44% (18 of 41) of patients receiving multidonor intensive heterologous FMT versus 20% (8 of 40) of those receiving placebo (Paramsothy et al., 2017). FMT induced bacterial changes that related to the treatment outcome(Paramsothy et al., 2017; Paramsothy et al., 2019), such as increased bacterial diversity when compared to pre-FMT baseline and better engraftment of donor bacteria when compared to non-responders, consistent with other clinical studies(Costello et al., 2019; Fuentes et al., 2017; Imdad et al., 2018; Moayyedi et al., 2015). However, as in other trials for FMT in UC, we could not identify specific factors that influenced the observed bacterial engraftment and treatment response.

In addition to bacteria, FMT transfers all constituents of the intestinal microbiota including viruses and fungi(Zuo et al., 2018). Likewise, FMT treatment does not selectively target intestinal bacteria but it likely affects and is affected by the various members of the microbiota. In CDI, where the intestinal niche is severely affected by C. difficile domination, an overgrowth of Candida species has been associated with a negative outcome of FMT(Zuo et al., 2018), a finding that has been replicated in mouse models (Li et al., 2019; Panpetch et al., 2019; Zuo et al., 2018). Experimental studies, however, suggest that gut colonization with Candida can exert contrasting effects on intestinal inflammation and that this modulation is strongly dependent on the host’s genetic predisposition and gut microbiota composition (Jawhara et al., 2008; Leonardi et al., 2018; Li et al., 2019; Sovran et al., 2018). In hosts with an intact microbiota, targeting the healthy gut fungal communities can worsen the outcome of experimental colitis and lung disease(Li et al., 2018; Wheeler et al., 2016), suggesting that specific members of the gut mycobiota can play a protective function in the intestine. These data suggest that members of the mycobiota might be important players in the response to FMT.

Results

To investigate the dynamics of the intestinal mycobiota in UC patients undergoing FMT we used an Internal Transcribed Spacer (ITS)-1 based barcoding approach to deep sequence the ITS-1 regions of fungal rDNA in fecal samples from the FOCUS study(Paramsothy et al., 2017). Specifically, ITS sequencing was performed on 129 fecal samples collected from FMT donors and UC patients pre-FMT and following 8 weeks of intensive FMT or placebo treatment (post-FMT). ITS sequencing resulted in a uniform library size distribution across the analyzed samples (Extended Data Fig. 1A, Table 1). The mycobiota was generally dominated by the Ascomycota phylum in all groups (Extended Data Fig. 1 D–E). In line with previously published cohorts(Chehoud et al., 2015a; Hoarau et al., 2016; Sokol et al., 2017), we observed a trend towards lower fungal alpha-diversity and increased Ascomycota/Basidiomycota ratio in feces from UC patients pre-FMT when compared to healthy donors (Fig 1A, ​,BB and Extended Data Fig. 1C, ​,2A,2A, ​,B).B). However, these differences did not reach statistical significance, possibly due the mild-moderate nature of UC in our patients’ cohort.

Open in a separate window

Figure 1.

Mycobiota composition in the FOCUS cohort.

A. Alpha diversity measured by observed OTU and Shannon Index for donor, fecal transplantation recipient (FMT), and placebo recipient (CTRL) pre- and post-FMT. B. Ratio between Ascomycota and Basidiomycota are shown for donor (grey), fecal transplantation recipient (FMT, red), and placebo recipient (CTRL. black) at pre- and post-FMT. C. Stack plot of relative abundances at the genus level for the 15 genera with the highest average abundance are shown. Dots represent individual patients, line and bars represent mean +/− SEM.

Although the bacterial composition changed post-FMT by both increased observed OTUs and diversity relative to baseline(Paramsothy et al., 2019), the gut mycobiota diversity, the abundance of the two major fungal phyla and observed OTUs all remained unaltered, suggesting that FMT did not affect the overall structure of gut fungal communities in our UC cohort. A deeper analysis of the mycobiota taxonomic composition revealed 15 major fungal genera with Candida and Saccharomyces being the most abundant in both healthy donors and UC patients (Fig 1C, Extended Data Fig. 2 C, ​,D).D). Within the Candida genus, Candida albicans was the most abundant and prevalent species (Extended Data Fig. 1F). Both Candida and Saccharomyces showed a similar relative abundance between donors and recipients pre-FMT (Extended Data Fig. 2C, ​,D).D). In contrast to previously reported post-FMT mycobiota outcomes in a CDI cohort (Zuo et al., 2018), multi-donor intensive FMT did not affect the relative abundance of Candida and Saccharomyces post-FMT in the UC recipients (Fig 2 A, ​,B),B), consistent with major differences between FMT outcomes in these intestinal diseases of different etiology.

Open in a separate window

Figure 2.

A. Relative abundance of Candida in fecal transplantation recipient (FMT, red), and placebo recipient (CTRL. black) at pre- and post-FMT. The paired mean difference between FMT pre- and post-FMT is 0.0234 [95.0%CI −0.203, 0.266]; P value>0.1 (Wilcoxon test). The paired mean difference between CTRL pre- and post-FMT is 0.111 [95.0%CI −0.198, 0.371]; pValue>0.1 (Wilcoxon test). B. Relative abundance of Saccharomyces in the FMT and CTRL group at pre- and post-FMT. The paired mean difference between FMT pre- and post-FMT is −0.048 [95.0%CI −0.264, 0.167]; pValue>0.1 (Wilcoxon test). The paired mean difference between CTRL pre- and post-FMT is 0.0136 [95.0%CI −0.203, 0.28]; pValue>0.1 (Wilcoxon test). C. Non metric multidimensional scaling (NMDS) based on Bray-Curtis dissimilarities of the mycobiota community at different time points in the donor, Endoscopic non-responders (NR) and Endoscopic responders (Res). D. Pre-FMT Relative abundance of Candida among FMT recipient experiencing clinical response (Resp) and non-responders (NR). Two-sided P value, Mann-Whitney T test, * pValue= 0.0185. E. Correlation between the pre-FMT relative abundance of Candida and the total Mayo Index post-FMT in placebo (CTRL) and FMT recipient (FMT). Lines represent linear regression for the placebo and FMT groups. Lower and upper hinges represent the 90% confidence bands. A,B,E. Cumming estimation plot of the paired mean difference, raw data is plotted on the upper panel; each set of paired observations is connected by a line. Lower panel: each paired mean difference is plotted as a bootstrap sampling distribution. Mean differences are depicted as dots; 95% confidence intervals are indicated by the ends of the vertical error bars. Effect size [CI width lower bound; upper bound]

As the pre-existing intestinal landscape can dramatically affect the outcome of FMT(Ericsson et al., 2017; Zuo et al., 2018), we next assessed the composition of the gut mycobiota pre-FMT and analyzed correlations with the clinical outcomes post-FMT therapy. While FMT responders and non-responders did not cluster separately based on overall taxonomic composition (Fig 2C), clinical FMT response was associated with significantly higher levels of Candida in pre-FMT samples compared to samples from FMT non-responder recipients (Fig 2D).

These findings were corroborated by a significant negative correlation between the pre-FMT relative abundance of Candida and the total Mayo Score post-FMT in the FMT recipients but not in the placebo controls (Fig 2 E). Similarly, higher levels of Candida were observed in endoscopic responders with respect to non-responders (Extended Data Fig. 4A, ​,B)B) and in patients achieving the primary endpoint (Extended Data Fig. 4A), although the difference did not reach statistical significance.

The pre-FMT difference between responders and non-responders was specific to Candida species as Saccharomyces and other less abundant genera were similar between the two groups (Extended Data Fig. 3, & 4C). Altogether, these findings suggest that preexisting intestinal Candida colonization is associated with positive clinical outcome of FMT.

Candida species presence and expansion has been consistently observed in IBD patients (Chehoud et al., 2015b; Davidson et al., 2019; Hoarau et al., 2016; Limon et al., 2019; Sokol et al., 2017) and can negatively affect disease outcome in genetically susceptible host during experimental colitis(Jawhara et al., 2008; Leonardi et al., 2018). On the other hand, environmental competition and bacteria-dependent induction of a host response against intestinal Candida (Fan et al., 2015) have both been shown to eradicate this opportunistic fungus in mice. We thus hypothesized that in clinical responders, FMT might reduce the intestinal burden of Candida, thus attenuating its pro-inflammatory effects. Alternatively, Candida might directly affect bacterial expansion or engraftment and thus influence the outcome of FMT, as previously shown for specific bacteria (Paramsothy et al., 2019). To investigate the possible interaction between Candida species and the bacterial microbiota, we analyzed the bacterial composition of the trial participants stratifying pre-FMT subjects by Candida species abundance (Fig 3A, ​,B,B, Extended Data Fig. 4D, ​,4E).4E). Notably, UC patients with high levels of Candida pre-FMT, had significantly higher bacterial alpha diversity that persisted at week 8 post-treatment irrespective of FMT treatment (Fig 3A,​,B),B), suggesting that Candida might contribute to or associate with the establishment of a niche more permissive to FMT engraftment.

Open in a separate window

Figure 3.

Changes in bacterial communities and association with fecal Candida levels

A. Simpson diversity index of 16s rRNA batches in patients at pre- and post-FMT. Patients were grouped based on the relative abundance of Candida (Ca) reads among fungal ITS reads pre-FMT. Low and high samples were defined as samples with a Candida relative abundance smaller and higher than the average Candida relative abundance (see also Figure S 4D). Two way ANOVA. Lower and upper hinges correspond to the 25th and 75th percentiles, dots represent individual samples. B. Upper panel: Relative abundance of bacterial phyla among 16s rRNA reads in FMT recipients pre-FMT. Lower panel: Relative Candida abundance in FMT recipients pre-FMT (showed as % of total ITS reads). C. Pearson correlation between bacterial genera in FMT recipients post-FMT and Candida abundance pre-FMT; grey band represent 95% confidence interval. Dots represent individual samples.

We next assessed whether bacteria found to be associated with FMT outcome would correlate with the initial abundance of Candida. We have previously reported that, Haemophilus, Fusobacterium, Prevotella, and Sutterella genera correlated with a negative FMT outcome (Paramsothy et al., 2019). Among these genera, both Haemophilus and Sutterella negatively correlated with the abundance of Candida (Pearson correlation, Fig. 3C). In contrast, among the genera previously reported to positively correlate with FMT outcome, we found that Roseburia negatively correlated with Candida abundance, whereas Ruminococcus displayed a significant positive correlation with Candida abundance. Our results suggest the existence of complex trans-kingdom relationship where Candida might be associated with or directly affect the engraftment of particular bacterial strains.

Candida spp. have been associated with intestinal inflammation and experimental studies have shown that several Candida species can worsen the outcome of intestinal inflammation in mouse models (Hoarau et al., 2016; Iliev et al., 2012; Jawhara et al., 2008; Leonardi et al., 2018; Sovran et al., 2018). We thus assessed the changes in Candida abundance following treatment in FMT and placebo recipients. In FMT recipients the reduction in Candida abundance correlated positively with a reduction of endoscopic and clinical disease severity as indicated by change in Mayo score (Fig 4A). This correlation was only observed in FMT recipients and not in placebo controls (Fig 4B), suggesting that a reduction in disease severity does not cause the reduction of Candida abundance. These result support the hypothesis that FMT might act at least partially by reducing the abundance of Candida in the gut. Candida species have the potential to induce strong pro-inflammatory immune responses during intestinal disease (Hoarau et al., 2016; Iliev et al., 2012; Jawhara et al., 2008; Leonardi et al., 2018; Sovran et al., 2018). In our cohort, Candida albicans was the most prevalent and abundant Candida species (Extended Data Fig. 1F). We thus assessed the effect of FMT on the induction of immune response against Candida by assessing the levels anti-C. albicans IgG antibodies in the patents’ serum. Over 8 weeks, placebo recipients experienced a significant increase in anti-Candida IgG responses (Fig 4C–D). In contrast, anti-C. albicans IgG titers were stable in FMT recipients (Fig 4C–D) during the same period. Collectively, these data suggest that FMT might act by reducing Candida abundance and containing pro-inflammatory immune responses induced by gut-fungi during intestinal inflammation.

Open in a separate window

Figure 4.

A reduction in Candida abundance and associated immune responses correlate with clinical and endoscopic response in FMT patients.

A. Correlation between the change in relative abundance of Candida following FMT and the change in total Mayo Index following FMT in FMT recipient (FMT). Lines represent linear regression for the placebo and FMT groups. Lower and upper hinges represent the 90% confidence bands. B. Correlation between the change in relative abundance of Candida following FMT and the change in total Mayo Index following FMT in placebo (CTRL) patients. Lines represent linear regression for the placebo and FMT groups. Lower and upper hinges represent the 90% confidence bands. C. The paired mean difference for the anti C. albicans IgG in FMT and control patients before and after FMT shown as a Cumming estimation plot. The raw data is plotted on the upper axes; each paired set of observations is connected by a line. On the lower axes, each paired mean difference is plotted as a bootstrap sampling distribution. Mean differences are depicted as dots; 95% confidence intervals are indicated by the ends of the vertical error bars. The paired mean difference between pre- and post-FMT in FMT recipients is 0.0175 [95.0%CI −0.198, 0.181], pValue= 0.141 (Wilcoxon test). The paired mean difference between pre- and post-FMT in Placebo recipients (CTRL) is 0.315 [95.0%CI 0.0802, 0.516], pValue= 0.0186 (Wilcoxon test). D. Variation of anti C. albicans IgG in FMT and placebo (CTRL) recipients between pre- and post- treatment. The unpaired mean difference between FMT and Placebo is 0.41 [95.0%CI 0.109, 0.738], pValue= 0.0215 (Mann-Whitney test).

Discussion

The positive outcomes of several pilot- and larger-scale clinical studies highlight the increasing promise of FMT for the treatment of diseases with complex etiology such as IBD, IBS, immune checkpoint inhibitor-associated colitis, metabolic, liver and neurological diseases(Allegretti et al., 2019; Imdad et al., 2018; Kakihana et al., 2016; Kang et al., 2017; Paramsothy et al., 2017; Vrieze et al., 2012; Wang et al., 2018; Weingarden and Vaughn, 2017). Our mycobiota analysis of participants in the largest, double blind, placebo-controlled FMT trial in UC to date, revealed a key role of trans-kingdom interactions between intestinal fungi and bacteria that positively affected the clinical outcome of FMT. Among the patients of the FOCUS cohort, we observed considerable heterogeneity in Candida abundance, reflecting the heterogeneity of IBD patient populations reported by previous studies (Chehoud et al., 2015a; Hoarau et al., 2016; Limon et al., 2019; Sokol et al., 2017). Nonetheless, preexisting high Candida abundance in UC patients prior to FMT treatment was associated with increased bacterial diversity and a better therapy outcome. Although the determination of the mechanistic basis for this phenomenon requires further studies, our data suggest that a reduction of intestinal Candida colonization contributes to the positive outcome of FMT on intestinal inflammation. While the exploration of mycobiota in FMT is still in its infancy(Zuo et al., 2018), the concept that gut mycobiota can influence intestinal health and bacteria engraftment has been explored by several studies(Li et al., 2018; Mason et al., 2012; Sovran et al., 2018; Wheeler et al., 2016). Fungi might affect the outcome of FMT in a context-dependent manner. Indeed, in CDI, where C. difficile dominates the intestinal niche, presence of Candida has been associated with a negative outcome of FMT (Zuo et al., 2018).

A growing number of studies support the existence of complex trans-kingdom interactions between fungi and bacteria (Mason et al., 2012; Sokol et al., 2017). In agreement with this notion, our data suggest that fungal and bacterial counterparts of the microbiota are interdependent, and that intestinal Candida colonization correlates with an increased diversity of the bacterial microbiota. This observation supports previous in vivo results demonstrating an increased recovery of bacterial diversity following antibiotic treatment in the presence of intestinal Candida (Mason et al., 2012). Analysis of the bacterial composition in UC patients from the FOCUS cohort, suggests that the presence of Candida might indicate a niche permissive to better bacterial engraftment. In contrast to CDI where FMT is effective in over 90% of the cases, in other diseases, such as IBD, response to FMT treatment appears to be highly heterogeneous and it is currently not possible to predict the clinical efficacy of FMT. Consequently, substantial effort is needed to identify factors that contribute to FMT treatment response that can be used as predictors prior to the initiation. Our findings suggest that mycobiota, host response to these organisms, and specific fungal species might harbor potential predictive value as to the outcome of microbiome-based therapies and should be further explored in future FMT trials.

STAR Methods

Extended Data

Extended Data Fig. 1:

Open in a separate window

Mycobiota in the FOCUS cohort A, related to Figure 1.

Numbers of ITS reads recovered for the different patients’ samples at the indicated time-points. B. Non metric multidimensional scaling (NMDS) based on Bray-Curtis dissimilarities of the mycobiota community at different time points in the donor (green), fecal transplantation recipient (FMT, red), and placebo recipient (CTRL, black) pre- and post-FMT. C. Alpha diversity measured by observed OTU for the different patients’ samples at the indicated time-points. Relative abundance of Ascomycota (D), Basidiomycota (E), are shown for donor (green), fecal transplantation recipient (FMT, red), and placebo recipient (CTRL. black) pre- and post-FMT. (F) Relative abundance of C. albicans and non-C. albicans (other) Candida species across the analyzed samples. Dots represent individual patients, line and bars represent mean +/− SEM.

Extended Data Fig. 2:

Open in a separate window

Mycobiota composition in UC patients and healthy donors pre-FMT, related to Figure 1.

A. Shannon diversity index, B. Log10 (Ascomycota/Basidiomycota) Ratio. C. Relative abundance of Candida reads among total ITS fungal reads. D. Relative abundance of Saccharomyces reads among total ITS fungal reads. Data are shown as Gardner-Altman estimation plot of the mean differences between groups. Dot represent individual patients. Mean difference is plotted on floating axes on the right as a bootstrap sampling distribution. The mean difference is depicted as a dot; the 95% confidence interval is indicated by the ends of the vertical error bar.

Extended Data Fig. 3:

Open in a separate window

Relative abundance of major genera in the study cohort pre-FMT, Related to Figure 1.

Lower and upper hinges correspond to the 25th and 75th percentiles, dots represent individual samples. FMT: FMT recipients, CTRL: Placebo recipient. Colors denote the Blinded Clinical response post-FMT: NR: non responders, Res: Responders.

Extended Data Fig. 4:

Open in a separate window

Candida and Saccharomyces relative abundance, related to Figure 1 and ​and22.

A. Relative abundance of Candida reads among total ITS fungal reads. The unpaired mean difference between FMT patients who met the primary endpoint at week 8 (RES) and patients who failed to meet the criteria (NR) is 0.19 [95.0%CI −0.0871, 0.456]. B. Relative abundance of Saccharomyces reads among total ITS fungal reads. The unpaired mean difference between NR and R is 0.106 [95.0%CI −0.176, 0.371]. Data in A, B, are shown as Gardner-Altman estimation plot of the mean differences between groups. Dot represent individual patients. Mean difference is plotted on a floating axes on the right as a bootstrap sampling distribution. The mean difference is depicted as a dot; the 95% confidence interval is indicated by the ends of the vertical error bar. C. Relative abundance of Candida in clinical non-responders (NR) and responders (R) of the FMT (red) and placebo (black) group pre- and post-FMT. The paired mean difference between pre- and post-FMT is 0.202 [95.0%CI −0.1, 0.51] in the FMT_NR group; −0.194 [95.0%CI −0.475, 0.131] in the FMT_R group; 0.0524 [95.0%CI −0.262, 0.33] in the Ctrl_NR group. pValue>0.1 (Wilcoxon test). Two-sided P value, Mann-Whitney T test is shown. D. Density distribution of Candida relative abundance in the analyzed patient samples. In Fig 3. Samples with Candida relative abundance smaller than the geometric mean (< 0.4347454) were defined as “low”. The remaining samples were defined as “high”. E. Ratio of clinical responders versus non-responders among FMT recipients post-FMT among patients with low Candida abundance (<10%) and high Candida abundance (>10%).

Acknowledgements

Footnotes

References