Adenomatous Polyps Are Driven by Microbe-Instigated Focal Inflammation and Are Controlled by IL-10–Producing T Cells

Inflammation is a cardinal feature of polyposis in mice as well as colon cancer in humans. Our earlier studies show that reduced expression of interleukin (IL)-10 by regulatory T cells (Treg) contributes to the dysregulation of inflammation in polyposis and colon cancer (1, 2). Many different types of cells are known for producing IL-10, including activated macrophages, dendritic cells, B cells, mast cells, and intestinal epithelial cells (3–6). Expression of IL-10 by macrophages is thought to limit inflammatory responses in the mouse intestine by ensuring expression of Foxp3 by CD4⁺CD25⁺ Tregs (7), although Tregs have also been reported to develop in the absence of IL-10 (8), and Tregs have IL-10–dependent, but also IL-10–independent, suppressive effects (reviewed in ref. 9). IL-10–deficiency does not hamper T cell–suppressive properties of Tregs but does eliminate their ability to suppress intestinal inflammation (10–12). In addition to Tregs, IL-10–expressing regulatory type I (Tr1) cells are potent suppressors of immune effector cells (13). It is not known how expression of IL-10 by T cells contributes to tissue levels of IL-10 or to increases in IL-10 during colon cancer. Furthermore, it is not known to what extent the natural history of tumor growth in cancer is affected by IL-10 or by T cells expressing IL-10.

Human familial adenomatous polyposis syndrome is caused by the inheritance of a defective allele of the adenomatous polyposis coli (APC) gene (14). Random LOH of this locus with age initiates formation of numerous polyps in the colon and predisposes to colon cancer; the same defect is found in up to 90% of individuals with sporadic colorectal cancer (15). Microbial-induced inflammation is an inherent causative component of colon cancer (16) and inflammatory bowel diseases (IBD) in humans (17, 18). Earlier studies show that polyposis in the multiple intestinal adenoma APC^Min/+ and in APC^Δ468 mouse models is driven by inflammation (19–21), however the source of this inflammation is not clear. Although germ-free APC^Min/+ mice are not protected against polyposis (22), colonic polyposis seems to be microbial-dependent and we earlier documented that oral delivery of beneficial commensal bacteria prevents polyp growth in the colon (23). Furthermore, microbes are causatively associated with colitis and colitis-induced cancer in mice (18, 24).

To determine the role of T cell–derived IL-10 in polyposis and colon cancer, we generated APC^Δ468 mice that expressed IL-10 and also expressed a Treg reporter (IL-10^Thy1.1xFoxp3^GFP), or had a selective deficiency for IL-10 only in T cells (CD4^CreIL-10^fl/fl), or were completely devoid of IL-10 (IL-10^−/−). We compared T cell and innate immune responses of these animals with responses of IL-10–competent APC^Δ468 mice. We found that T cells and Tregs were the primary if not exclusive source of IL-10 in the colon in health and disease, and that T-cell deficiency in IL-10 increased inflammation and polyposis. We provide evidence that IL-10–producing T cells control microbial densities within colonic polyps and limit inflammation and growth of adenomatous polyps in the colon. Thus, IL-10–expressing T cells are a suitable therapeutic target for polyposis and potentially colon cancer.

B6 (C57BL/6J), CD4^Cre, and IL10^−/− mice were purchased from The Jackson Laboratory. The APC^Δ468 mouse model of polyposis has been reported earlier and extensively characterized for intrapolyp inflammatory reactions and in the role of inflammation in polyp growth (1, 2, 19–21, 25). IL-10^fl/fl mice were generated by Roers and colleagues (26). IL-10^Thy1.1xFoxp3^GFP reporter mice were generated by Maynard and colleagues (8). All animal work was approved and conducted according to the guidelines of the Animal Care and Use Committee of Northwestern University (Chicago, IL).

Paraffin sections (4 μm) were used throughout. Details are in the Supplementary Materials.

Of note, 4- to 6-week-old mice were lethally irradiated at 1,000 rad (split dose). Mice were retro-orbitally injected with bone marrow cells, which had been Lin-depleted. Details can be found in the Supplementary Materials.

Mice were treated with an antibiotic cocktail as described previously (27). Details are in the Supplementary Materials.

Microbial DNA was extracted from mouse colonic and fecal samples as described previously (28). Details are in the Supplementary Materials.

Microbial DNA was amplified as described previously (29). Sequencing was conducted by the Next Generation Sequencing Core at Argonne National Laboratory (Argonne, IL) using an Illumina MiSeq (29) and sequences were classified, analyzed, and uploaded in the MG-RAST system (http://metagenomics.anl.gov/; 4513044.3 and 4513045.3; project title: T cell IL-10 regulates inflammation and colorectal cancer-Dennis). Details are in the Supplementary Materials.

Box and whisker histogram depicts box with median flanked by upper and lower 25% quartile with whiskers showing the maximum and minimum data points. The statistical analyses were conducted with the use of the Prism 4 software. P values were determined with two-tailed unpaired t test unless where otherwise indicated; *, P < 0.01; **, P < 0.001; ***, P < 0.0001; ns, not statistically significant data. For Fig. 4B–D, significance was determined by conducting a paired, one sample t test with the theoretical mean set to a value of 0. Only values above 0.002 (0.2%) abundance were analyzed.

We monitored IL-10 expression in the colonic mucosa using transgenic IL-10^Thy1.1xFoxp3^GFP reporter mice (8). Lethally irradiated Thy1.2 B6 and APC^Δ468 mice were reconstituted with bone marrow from reporter IL-10^Thy1.1xFoxp3^GFP mice. Chimeric mice were sacrificed and colons were excised, frozen, and sectioned for immunofluorescence staining with Thy1.1 antibody and GFP fluorescence (compare Thy1.1 specific antibody with IgG1, κ isotype control in Supplementary Fig. 1). Polyp-ridden APC^Δ468 mice had elevated frequencies of IL-10–expressing cells in both the polyp and marginal tissue as compared with B6 (Fig. 1A). We found that total CD4⁺ cells were significantly elevated in polyps of APC^Δ468 mice as compared with healthy B6 colon (Fig. 1B and C). Furthermore, a large fraction of both T cells and Tregs did not express IL-10 in the polyp-ridden colons (Fig. 1B and D). Interestingly, CD4⁺ T cells and Foxp3⁺ Tregs remained the major cellular source of IL-10 in the colon in healthy mice and also during polyposis (Fig. 1E). We did not detect IL-10 expression in any cells other than CD4⁺ cells.

We examined how epithelial mitotic activity and polyposis in colon were altered by T-cell deficiency for IL-10 in CD4^CreIL-10^fl/flAPC^Δ468 mice or complete IL-10 deficiency in IL-10^−/−APC^Δ468 mice. As reported previously, CD4^CreIL-10^fl/fl and IL-10^−/− mice developed spontaneous colitis shown by increased epithelial mitotic activity and crypt elongation as compared with healthy B6 mice (Fig. 2A, top and B). Likewise, colons of CD4^CreIL-10^fl/fl and IL-10^−/− mice were elongated as compared with B6 colons (Fig. 2A, bottom). APC^Δ468 mice developed abundant polyposis by 4 months of age, however, colonic mitotic activity and overall colon length were comparable with that of B6 mice (Fig. 2A and B). At this age, APC^Δ468 mice with a T-cell deficiency for IL-10 or complete IL-10 deficiency produced comparable pathologies with one another. Epithelial mitotic activity was significantly increased (Fig. 2A), producing crypt elongation (Fig. 2B) and overall colon elongation (Fig. 2A, bottom) in both groups of mice as compared with IL-10–proficient APC^Δ468 mice.

APC^Δ468 mice with a T-cell or complete IL-10 deficiency increased colonic polyps by roughly 5-fold (Fig. 2C and D). Furthermore, the polyps had expanded stroma and were filled throughout with abnormal crypts containing large polyploid nuclei, whereas in IL-10–proficient APC^Δ468 mice, the polyp stroma was less expanded and the abnormal crypts typically also contained hyper-proliferative but otherwise normal appearing crypts at the basal edge of the lesions (Fig. 2E). Thus, T-cell IL-10 deficiency produced similar pathologies to complete IL-10 deficiency, in both instances aggravating epithelial mitosis and polyposis in the colon.

In earlier studies, we showed that mast cells are essential for polyp growth in the small intestine (20, 21). Surprisingly, IL-10 deficiency reduced mast cell density in the colon, as determined by in situ staining with chloracetate esterase (CAE) and antibody to murine mast cell protease-2 mMCP2 staining (Fig. 3A and B). In contrast, there were significant increases in the densities of eosinophils, stained with a specific antibody to major basic protein (MBP; Fig. 3C and D; ref. 30). MBP is contained in eosinophil granules, and its release causes inflammation, breakdown of epithelial barrier, and tissue damage (31). Interestingly, MBP⁺ eosinophils preferentially accumulated within the submucosa and colonic polyp stroma (Fig. 3C–F).

Depending on their composition and abundance, colon-resident microbiota either maintain a balanced healthy inflammation or provoke chronically elevated pathogenic inflammation (reviewed in ref. 32). To determine the complexity of the bacterial communities inside colonic polyps relative to the marginal tissue and also relative to healthy colon, and link microbiota, inflammation, and polyp growth, we analyzed the phylogenetic diversity of the sampled microbial assemblages by sequencing the V4 region of the bacterial 16S RNA gene, generating an average of 3,000 reads per sample. To control for diversity of microbial populations between animals of the different mouse strains, littermate B6, APC^Δ468, and CD4^CreIL-10^fl/flAPC^Δ468 mice were generated by crossing heterozygous CD4^Cre/+IL-10^fl/+APC^Δ468 and IL-10^fl/+ mice, yielding offspring from a single mother. Although we did not detect a major shift in the bacterial populations at the phylum level (Supplementary Fig. 2), this analysis revealed that the composition of resident microbiota at the genus level was disparate when comparing marginal and polyp tissues, as well as polyp versus healthy B6 tissues (Fig. 4A–D).

Two genera, Bacteroides and Porphyromonas of the phylum Bacterioidetes, were consistently and significantly elevated in abundance within colonic polyps as compared with marginal tissues (P values of 0.0300 and 0.0024, respectively; Fig. 4B). In addition, Bacteroides and Rikenella of the Bacterioidetes phylum were significantly elevated in abundance within polyps as compared with healthy B6 tissues (P values of 0.0143 and 0.0409, respectively; Fig. 4B). Prevotella of the Bacteroidetes phylum also tended toward increased abundance in the polyp as compared with marginal or B6 tissue (Fig. 4B). Elevated genera of the phylum Bacteroidetes have been implicated as a risk factor for IBD and colorectal cancer in humans (33). In addition, Bacteroides and Porphyromonas, as well as Prevotella, have been shown to adhere to the colon mucosa and tumor tissues are enriched for these bacteria in patients with CRC (34, 35). Polyp tissue was not enriched overall above marginal tissue or B6 tissue in other bacteria that fell under different phyla, including Proteobacteria, and Firmicutes (Fig. 4C and D).

Evidence from animal studies specifically implicates T cell–derived IL-10 as a requisite, nonredundant mediator of colonic immune homeostasis (11, 12, 36) essential for control of microbial-driven TH17 inflammation (37) and colitis (38, 39). Past reports have documented that colons of IL-10^−/− mice have a different microbial composition than colons of healthy wt mice (39). To investigate how T cell IL-10 deficiency in polyposis alters microbiota composition of the colon, we carried out sequencing of the bacterial DNA from colons of littermate wt, APC^Δ468, and CD4^CreIL-10^fl/flAPC^Δ468 mice, and conducted principle coordinate analysis (PCoA). We observed a genotype-specific clustering dependent on two components of variation (Fig. 5). PC1 (37.24% of variation) strongly separated B6 and APC^Δ468 groups from the CD4^CreIL-10^fl/flAPC^Δ468 group, whereas PC3 (10.77% of variation) strongly separated the B6 group from the APC^Δ468 and CD4^CreIL-10^fl/flAPC^Δ468 groups (Fig. 5). Altogether, these observations suggest that the microbial composition of polyps is distinct from that of the healthy surrounding tissue and that T cell IL-10 deficiency alters the composition of microbial communities in the colonic mucosa. Next, we examined the role of microbiota in polyposis.

Formation of polyps coincides with loss of epithelial barrier integrity and incursion of microbial products (40), and we reasoned that this may have been the cause of the observed intrapolyp eosinophilic response. Therefore, we treated littermate APC^Δ468 and CD4^CreIL-10^fl/flAPC^Δ468 mice with a cocktail of broad-spectrum antibiotics: kanamycin, gentamycin, colistin, metronidazole, and vancomycin by intraperitoneal (i.p.) injection. This cocktail produces profound shifts in the composition of the intestinal microbiota and ablates microbial-induced TLR4 signaling in the murine colon (41). Mice were administered antibiotics starting at the onset of polyposis (9–10 weeks of age) and continuing every other day for a total of 3 weeks (Fig. 6A). After a week of rest, the mice were analyzed for polyp load, inflammatory eosinophilia, and the abundance of specific polyp-associated microbial taxa.

Antibiotic treatment significantly decreased eosinophilia (Fig. 6C) and polyp load (Fig. 6D) in the colon of CD4^CreIL-10^fl/flAPC^Δ468 mice, however did not have any impact on mastocytosis (Fig. 6B). To analyze microbial complexity following antibiotic treatment, fecal pellets were collected from mice just before antibiotic treatment (pretreatment) and again at the conclusion of the treatment (posttreatment). Amplicon analysis of the 16S rRNA V4 region identified the two genera, Bacteroides and Porphyromonas, which were significantly elevated in colonic polyps as compared with healthy marginal tissues; these significantly decreased relative abundance in all antibiotic-treated APC^Δ468 and CD4^CreIL-10^fl/flAPC^Δ468 mice (Fig. 6E and F). By calculating the Shannon index for overall microbial diversity, we were able to show a drop in the overall microbial diversity in fecal content from animals treated with antibiotics (Fig. 6G). These observations indicate that microbiota critically contribute to eosinophilic infiltration of polyps and to polyposis in the colon.

We provide evidence that colonic polyposis is driven by intrapolyp inflammatory reactions, which are tuned by microbial communities within the polyps and by infiltrating T cells. Our findings reveal that T cells and Tregs are the major cellular source of IL-10 in the colon, and their expression of IL-10 is critical for the control of intrapolyp inflammation and polyp growth.

It is surprising but at the same time reassuring that the major source of IL-10 in the colon is T cells and Tregs. This explains why in the polyposis-prone APC^Δ468 mice T-cell/Treg ablation of IL-10 produced near identical pathologies as complete IL-10-deficiency. When we ablated IL-10 specifically in T cells, polyp frequency in the colon increased by nearly 5-fold, producing a pathology reminiscent of human familial adenomatous polyposis coli. Ubiquitous ablation of IL-10 produced a near identical result, emphasizing the critical role of IL-10–expressing T cells in limiting both inflammation and polyp growth in the colon.

T cell IL-10 deficiency in the colon resulted reduced mast-cell numbers. IL-10 dependence of mast cells has been reported before (42), and could explain the loss of mast cells from the colon of IL-10-deficient mice. There was a notable increase in the density of polyp-infiltrating eosinophils in colons of IL-10-deficient APC^Δ468 mice. The enrichment of colonic polyps for microbial communities suggested us that the eosinophils were responding to microbial intrusion. We validated this by showing that both eosinophils and microbes were eliminated by treatment with broad-range antibiotics. Loss of eosinophils coincided with reduction in the number of colonic polyps. These observations suggest that chronic inflammatory reactions to intrapolyp microbes help promote polyp growth.

Our findings extend an earlier report suggesting that adenomas are leaky and serve as ports of entry of microbial products (40) by showing that polyps are focally enriched in proinflammatory bacteria. Our finding agrees with recently reported penetration of bacteria into the inner colonic mucosa layers of both mice with colitis and patients with ulcerative colitis (18). Our findings are also in line with an earlier report that a mutant strain of Lactobacillus acidophilus lacking cell surface expression of the TLR5 ligand and lipoteichoic acid protects against colonic polyposis (23). In this earlier study we did not see any response by small intestine polyps.

Mucosal immune responses determine the quality and quantity of commensal bacteria (43). In the chronic situation, this leads to selection of pathogenic bacteria (17). Thus, both epithelial-barrier defects and growing tumors can selectively alter the microbial community causing expansion of pathogenic bacteria (16, 44) that promote vicious cycles of inflammation in colitis and colon cancer (45–47). Accordingly, specific microbiota were enriched within the colonic polyps and were of distinct taxa with significant abundance of Bacteroides and Porphyromonas as compared with those found in healthy gut regions. This finding is in agreement with previous studies that have shown biopsies from patients with colorectal adenomas to have elevated bacterial abundance as compared with healthy colon tissue, and suggests that analysis of specific taxa of human microbiota could be used to potentially identify patients at risk for IBD (48, 49) and colonic adenomas (50). Our observations show that IL-10 producing T cells and Tregs regulate bacterial-instigated polyp growth and therefore enhancing their activity is a promising strategy for immune therapy of polyposis and colon cancer.

No potential conflicts of interest were disclosed.

Conception and design: K.L. Dennis, E.B. Chang, K. Khazaie

Development of methodology: K.L. Dennis, Y. Wang, A. Roers

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.L. Dennis, Y. Wang, N.R. Blatner, S. Wang, A. Saadalla, E. Trudeau, J.J. Lee, E.B. Chang, K. Khazaie

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.L. Dennis, Y. Wang, N.R. Blatner, S. Wang, J.A. Gilbert

Writing, review, and/or revision of the manuscript: K.L. Dennis, Y. Wang, N.R. Blatner, J.J. Lee, J.A. Gilbert, K. Khazaie

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.L. Dennis, Y. Wang, E. Trudeau, C.T. Weaver, J.A. Gilbert

Study supervision: N.R. Blatner, K. Khazaie

The authors thank Nicoleta Carapanceanu for her assistance with mouse work and genotyping, and Elias Gounaris for his assistance with microscope.

Funding for this project included an NIH-1R01CA160436-01, a Zell Family Award and an anonymous foundation award of the Robert H. Lurie Comprehensive Cancer Center (K. Khazaie) and an NIH-DK097268 and University of Chicago Digestive Disease Research Core Center grant-DK42086 (E.B. Chang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.