Virome drift in ulcerative colitis patients: faecal microbiota transplantation results in minimal phage engraftment dominated by microviruses

Abstract

Ulcerative colitis (UC) is an inflammatory bowel disease characterized by recurrent colonic inflammation. Standard treatments focus on controlling inflammation but remain ineffective for one-third of patients. This underscores the need for alternative approaches, such as fecal microbiota transplantation (FMT), which transfers healthy donor microbiota to patients. The role of viruses in this process, however, remains underexplored. To address this, we analyzed the gut virome using metagenomic sequencing of enriched viral particles from 320 longitudinal fecal samples of 44 patients enrolled in the RESTORE-UC FMT trial. Patients were treated with FMTs from healthy donors (allogenic, treatment) or themselves (autologous, control). We found that colonic inflammation, both its presence and location, had a greater impact on the gut virome than FMT itself. In autologous FMT patients, the virome was unstable and showed rapid divergence over time, a phenomenon we termed virome drift. In allogenic FMT patients, the virome temporarily shifted toward the healthy donor, lasting up to 5 weeks and primarily driven by microviruses. Notably, two distinct virome configurations were identified and linked to either healthy donors or patients. In conclusion, inflammation strongly affects the gut virome in UC patients, which may lead to instability and obstruct the engraftment of allogeneic FMT.

Keywords: Ulcerative colitis; fecal microbiota transplants; virome; virome community types; virome drift.

Figure 1.

Virome covariates of the UC cohort undergoing fecal microbiota transplantation. (a), Principal coordinate analysis of inter-individual differences in the relative virome profiles (genus-like level, Bray-Curtis dissimilarity) of the complete UC cohort (n = 301, biologically dependent samples, coloured by FMT treatment profile). The arrows on the plot represent effect sizes of significant virome covariates in the multivariate model in panel B. (b,c), Clinical metadata correlating to the relative virome profiles in the complete (n = 301, Patients+Donors) and FMT treated (n = 139, patients) UC cohort (dbRDA, genus-like level, Bray-Curtis dissimilarity), respectively. The effect sizes of correlating variables are calculated using either univariate (coloured in black) or multivariate (coloured in grey) analysis. The multivariate contribution in the latter analysis is indicated by a vertically dashed line and is 7.56%. Correction for multiple testing was performed, when appropriate, with the Benjamini-Hochberg method and significant associations were identified based on AdjP < 0.05. The aforementioned analyses include patient identifier as a grouping variable to account for the biological dependencies among multiple measurements within each patient. Abbreviations: ulcerative colitis (UC) and fecal microbiota transplant (FMT).

Figure 2.

Donor virome engraftments in the complete UC cohort. (a), Boxplot of differences in the relative virome profiles (genus-like level, Hellinger transformation) between patients and their corresponding autologous or allogenic donor sample (LMM, AdjP < 0.5) at different timepoints. (b), Boxplot of the percentage of shared donor taxa between patients and their corresponding donor sample (LMM, AdjP < 0.5) at different timepoints. Shared donor taxa were defined as those genera present in the corresponding donor batch that were simultaneously identified in individual patient samples. Significant associations were determined after multiple testing (Benjamini-Hochberg method) by comparing each post-FMT to baseline timepoint and are represented by an asterisk (*). The aforementioned analyses include patient identifier as a grouping variable to account for the biological dependencies among multiple measurements within each patient. Abbreviations: ulcerative colitis (UC) and linear mixed-effect model (LMM).

Figure 3.

Summary of individual UC patients undergoing allogenic fecal microbiota transplants. Barplot showing the relative abundance of major (≥1% of reads) in silico predicted hosts, phage genera (≥15% prevalent), phage classes and viral community types, along a longitudinal axis for each UC patient undergoing healthy donor FMT. Each patient received a sequence of four FMTs, collectively referred to as “Allogenic donor x” (in bold), and were consistently derived from the same donor. Patients are shown who had available samples from the healthy donor, baseline (week 0) and week 8. The endoscopic outcomes at week 8 are represented as either remission (green) or non-remission (red). Patients who exhibit transplanted donor genera at both week 4 or week 8 are indicated by the letter “T”. Abbreviation: fecal microbiota transplantation (FMT) and ulcerative colitis (UC).

Figure 4.

Transplanted donor phages in the complete UC cohort. (a), Barplot of successfully engrafted donor genera (abundance increase ≥ 100%) at week 4 (1 week post-FMT). (b), Barplot of successfully engrafted donor genera (abundance increase ≥ 100%) at week 8 (5 weeks post-FMT). The shaded green section on the plot denotes viral genera transplanted in more than 20% of the patients. (c), Barplot of successfully engrafted donor contigs present within each highly transplanted genus (abundance increase ≥ 100%) at week 4 (1 week post-FMT). (d), Barplot of successfully engrafted donor contigs present within each highly transplanted genus (abundance increase ≥ 100%) at week 8 (5 weeks post-FMT). Abbreviations: ulcerative colitis (UC) and fecal microbiota transplant (FMT).

Figure 5.

Viral community types of the complete UC cohort. (a), Principal coordinate analysis of inter-individual difference in the relative virome profiles (genus-like level, Bray-Curtis dissimilarity) of the complete UC cohort (n = 301, biologically dependent samples, coloured by VCT). (b), Boxplot showing relative abundance of key phage classes categorized according to VCT (n = 301, LMM, AdjP < 0.5). (c), Boxplot showing relative abundance of lysogenic potential of all discovered phages (left) and unclassified Caudoviricetes phages (right) categorized according to their VCT (n = 301, LMM, AdjP < 0.5). (d), Boxplot showing relative abundance of major predicted bacterial hosts (≥1% of reads) categorized according to their VCT (n = 301, LMM, AdjP < 0.5). (e), left, Barplot showing VCT prevalence in complete UC cohort categorized according to FMT treatment profile (n = 301, GLMM, AdjP < 0.5). (e), right, Barplot showing the longitudinal prevalence of VCT in the allogenic FMT-treated cohort (n = 99, GLMM, AdjP < 0.5). Multiple testing adjustment (Benjamini-Hochberg method) was performed and significant associations (AdjP <0.05) are represented by an asterisk (*). The aforementioned analyses include patient identifier as a grouping variable to account for the biological dependencies among multiple measurements within each patient. Abbreviations: ulcerative colitis (UC), viral community types (VCT), linear mixed-effect model (LMM) and generalized linear mixed-effect model (GLMM) and non-significant (ns).