A missense mutation in Muc2 promotes gut microbiome and metabolome-dependent colitis-associated tumorigenesis

Abstract

Colitis-associated cancer (CAC) arises from a complex interplay between host and environmental factors. In this report, we investigated the role of the gut microbiome using Winnie mice, an ulcerative colitis-like (UC-like) model with a missense mutation in the Muc2 gene. Upon rederivation from a conventional (CONV) to a specific pathogen-free (SPF) facility, Winnie mice developed severe colitis and, notably, spontaneous CAC that progressively worsened over time. In contrast, CONV Winnie mice showed only mild colitis but no tumorigenesis. By comparison, when re-derived into germ-free (GF) conditions, SPF Winnie mice were protected from colitis and colon tumors, indicating an essential role for the gut microbiome in the development of CAC in these mice. Using shotgun metagenomics, metabolomics, and lipidomics, we identified a distinct proinflammatory microbial and metabolic signature that potentially drives the transition from colitis to CAC. Using either SPF Winnie or WT (Bl/6) donors, fecal microbiota transplantation (FMT) into GF Winnie recipients demonstrated that, while colitis developed regardless of the donor, only FM from SPF Winnie donors resulted in CAC in recipient mice. Our studies present a relevant model of CAC, providing strong evidence that the microbiome plays a key role in its pathogenesis, thus challenging the concept of colon cancer as a strictly nontransmissible disease.

Keywords: Colorectal cancer; Gastroenterology; Inflammation; Microbiome.

Figure 1. Increased disease severity and emergence of colonic tumors in SPF- versus CONV-raised Winnie mice.

(A) Breeding strategy used to generate Winnie mice and their littermate controls (Bl/6) in SPF and CONV facilities, showing the specific phenotype of SPF Winnie, Winnie/+, and WT (WT Bl/6) mice. (B) Weight curves of SPF vs. CONV Winnie mice measured every 4 weeks until sacrifice at 20 weeks. (C–G) Necroscopic and histological evaluations of colonic inflammation and tumorigenesis comparing SPF versus CONV Winnie colons. (C) Weight and (D) colon weight/BW ratio percentage of experimental mice measured at the time of sacrifice. Progression of (E) disease severity over time as evaluated by a gastrointestinal (GI) pathologist blinded to the H&E-stained slides of colon samples, using an established scoring system. (F) Tumor area over time was calculated by Image J using a necropsy image of the colon taken at the time of sacrifice. (G) Representative H&E-stained images of colonic tissue from CONV (upper panel) and SPF (lower panel) Winnie mice at the time of sacrifice. Scale bars: 50 μm, 100 μm, and 200 μm. Original magnification, ×10 and ×20 (inset highlights dysplastic crypt [arrow]). Data presented in the dot plots are expressed as the mean ± SEM. *P < 0.05, **P < 0.005, ***P < 0005, and ****P < 0.0001, by unpaired 2-tailed t test or Mann-Whitney U test. n =12 for each experimental group.

Figure 2. Development of colon tumors in SPF Winnie mice is dependent on the presence of gut microbiota.

(A) Representative light box images (left panels) and stereomicroscopic images (right panels) of fresh and fixed, respectively, 20-week-old SPF and GF WT (Bl/6) and Winnie colons. (B) Representative endoscopic images of colons of 20-week-old SPF (left panels) and GF (right panels) Bl/6 and Winnie mice using white light (upper panels) and narrow band imaging (NBI) (lower panels). The white-dotted outline highlights the representative tumor area.

Figure 3. Microbiome composition differs in tumor-bearing SPF versus CONV Winnie mice, with differential regulation of Gastranaerophilales, Rhodospirillales, Lachnospiraceae AC2044 group, and Alloprevotella.

(A–E) 16S rRNA-Seq of fecal total microbiota from 20-week-old SPF (n = 12) and CONV (n = 8) Winnie mice. (A) Observed operational taxonomic unit (OTU), (B) Shannon diversity index, and (C) PCA were assessed in the 2 experimental groups. (D) Pie charts for SPF (right) and CONV (left) Winnie stool, indicating percentages of the most abundant genera (>1%). (E) Relative abundance (expressed as a percentage) of significantly represented genera in the stools of 20-week-old SPF (n = 12) and CONV (n = 8) Winnie mice. (F–H) 16S rRNA-Seq of nontumor and tumor-associated microbiota from 20-week-old SPF Winnie mice. (F) PCA with (G) pie charts depicting nontumor mucosa (upper) and tumor-associated mucosa (lower), showing percentages of the most abundant genera (>1%) and the (H) relative abundance (expressed as a percentage) of significantly represented genera in SPF Winnie. Data presented in the dot plots are expressed as the mean ± SEM. n = 12 for each experimental group. *P < 0.05, ***P < 0.0005, and ****P < 0.0001, by unpaired 2-tailed t test or Mann-Whitney U test.

Figure 4. Metabolipidomics profiling of stool samples from 20-week-old SPF and CONV Winnie mice reveals distinct metabolomic and lipidomic alterations.

Metabolipidomics of stools from 20-week-old CONV and SPF Winnie mice shown by (A) score plot of PCA analysis relative to metabolipidomics and (B) volcano plot indicating significantly increased (blue) and decreased (red) metabolites and lipids in SPF Winnie mice. (C) Quantitative enrichment analysis overview showing the top 25 related diseases, (D) metabolic pathways ranked according to P value (0.0 equals < 0.05) and fold enrichment, and (E) heatmap for the top 50 significant metabolites and lipids of SPF relative to CONV Winnie mice.

Figure 5. Shotgun analysis identifies bacterial strains and related pathways driving a protumorigenic phenotype in SPF versus CONV Winnie mice.

Fecal samples from 20-week-old SPF versus CONV Winnie mice were analyzed and are presented by (A) PCA, (B) heatmap of the most abundant bacterial species (>1%), (C) circular dendrogram highlighting significantly different bacterial species, and (D) dot plots showing the relative abundance (expressed as a percentage) of the most abundant species. (E) Enrichment fold change of specific pathways relevant in colitis and colitis-associated cancer, comparing fecal material from SPF and CONV Winnie mice. (F) Spearman’s correlation between bacterial species and relevant pathways enriched in SPF versus CONV Winnie mice. Data presented in the dot plots are expressed as the mean ± SEM. n = 8 for each experimental group. *P < 0.05, **P < 0.005, ***P < 0.0005, and ****P < 0.0001 by unpaired 2-tailed t test or Mann-Whitney U test.

Figure 6. Integration of multiomics analyses shows a strong correlation between metabolomics, lipidomics, and bacterial genera.

(A) Heatmap represents the correlation structure extracted from the 4 omics datasets. The correlation of each original feature pair is determined by each of their correlations with the components from the integrative method. (B) Arrow plots depict the similarities and discrepancies between a given sample across the 4 datasets, which can be seen (refer to the legend) . Short arrows indicate strong agreement between datasets, while long arrows highlight significant disagreement. Each sample is represented by a centroid and associated with 4 data types. (C) The circos plot depicts the correlations between each feature of each dataset. The top selected features of each dataset are shown. Lines are only drawn for correlations above 0.8 (cutoff = 0.8) to reduce visual clutter. (D) The loading generated from the PLS-DA applied to the multiomics data highlights the variables that discriminate between treatment conditions. Blue indicates the SPF group, and red represents the CONV group. n = 8 for each experimental group.

Figure 7. FMT using SPF Winnie and Bl/6 donors indicates that tumor development depends on Winnie, not Bl/6, recipient genotype.

(A and B) FMT experimental design (A) and timeline (B) involving FMT from SPF Winnie and Bl/6 donor mice into GF Winnie and Bl/6 recipient mice. (C) Changes in weight of experimental mice over time and (D) colon weight/BW ratio (expressed as a percentage) measured at sacrifice. (E) Representative white light endoscopic images (left panels) of GF Winnie recipient mice receiving donor fecal microbiota from SPF Winnie (lower left panel) or SPF Bl/6 mice (upper left panel), with associated endoscopic scores (right panel). (F) Representative H&E-stained images of experimental groups (original magnification, ×10; scale bars: 100 μm and 200 μm), with associated histological assessment of (G) inflammation and (H) dysplasia, as evaluated by a GI pathologist blinded to the H&E-stained slides of colon samples, using an established scoring system. Data presented in the histograms are expressed as the mean ± SEM. *P < 0.05 and **P < 0.005, by ordinary 1-way ANOVA with Tukey’s multiple-comparison or Kruskal-Wallis test followed by Dunn’s test. Stool homogenates were pooled from n = 4 donor mice, with n = 3–8 recipient mice for each experimental group.