Redundant Innate and Adaptive Sources of IL17 Production Drive Colon Tumorigenesis

The immune system plays a major role in cancer development and growth. It can promote or inhibit cancer depending on the specific type of response mediated by both innate and adaptive elements, particularly T cells (1). While Th1 CD4 T cells producing IFNγ and CD8 CTL mediate antitumor immunity, recent evidence supports the notion that the alternative Th17 response is procarcinogenic, certainly at the level of cancer formation in the setting of chronic and/or recurrent inflammation.

We have studied the role of endogenous IL17 in de novo colon carcinogenesis after colonization with a human colonic bacterium, enterotoxigenic Bacteroides fragilis (ETBF; ref. 2) that is a bacterial candidate for induction of colon oncogenesis (3). ETBF is an anaerobe detected in up to 50% of individuals, without clinical symptoms (4, 5). However, ETBF colonization can also cause inflammatory diarrhea in children and adults (6) and is detected in the colon mucosa of approximately 90% of human colon cancer patients (4). ETBF induces acute followed by chronic colonic inflammation in mice and promotes tumorigenesis in Min^Apc+/− (Min) mice (2, 7). ETBF produces an oncogenic enterotoxin, B. fragilis toxin (BFT), which triggers the cleavage of E-cadherin in the colonic epithelium and leads to the activation of Wnt/β-catenin (targeting c-Myc and epithelial proliferation) and NF-κB pathways (targeting proinflammatory mediators), all of which are features of human colorectal cancer (3). ETBF-induced colitis is characterized by a STAT3/IL17–driven inflammatory response, associated with rapid DNA damage in epithelial cells, and hyperplasia, resulting in colonic tumors in Min mice that otherwise typically develops tumors predominantly in the small intestine (2, 8, 9). While our initial findings demonstrated the contribution of an adaptive Th17-driven mucosal immune response to ETBF-promoted colon carcinogenesis, the carcinogenic activity of Th17 cells versus other sources of IL17 remains to be elucidated.

Th17 cells do not represent an exclusive source of IL17. γδ T cells and a variety of immune cells also produce IL17 during autoimmune inflammation and infection, including invariant NK T (iNKT) cells, mucosal-associated invariant T (MAIT) cells, CD3⁻Thy1^hiSCA⁺ group 3 innate lymphoid cells (ILC3), as well as myeloid cells including mast cells (MC), macrophages (Mφ), and neutrophils (PMN; ref.10). Some of these cells have been reported in tumor tissues and their production of IL17 has been associated with tumor outcome, but with very mixed results, suggesting that the nature of the cells producing IL17 may play a critical role in oncogenesis (11). In humans, detection of IL17-producing MCs in esophageal squamous cell carcinoma (ESCC) was associated with a better outcome, whereas IL17-producing Mφ were associated with the production of the proangiogenic factor, VEGF, and decreased survival in colorectal cancers (12, 13). Non-Th17–derived IL17 is thought to be mainly provided early by innate cells at mucosal and skin surfaces in response to infection and serves to determine the amplitude of the ensuing adaptive Th17 immune response (10, 14). Whereas uncommitted CD4⁺ Th cells require TCR ligation in the presence of the combined action of IL6, IL1β, and TGFβ to differentiate into Th17 cells, innate IL17-producing cells seem to essentially rely on the action of IL23, IL1β, and/or TLR signaling (14–18). While each of these cell types has been reported to reside within tumors to various degrees, little is known about their role relative to Th17 cells in cancer development.

The role of IL17-producing γδ T cells (γδT17 cells) in early cancer initiation has not been studied and their role in cancer progression is controversial. γδT17 cells were recently suggested to improve the response to antitumor chemotherapy by recruiting immune effector cells (16) or alternatively to be the sole source of IL17 in promoting tumor growth (19–21). In human colon cancer, γδT17 cells were reported to be the only source of IL17 that was suggested to promote angiogenesis and tumor growth via the recruitment of myeloid-derived suppressor cells (MDSC; ref. 22).

Strikingly, although delayed, the ultimate ETBF-induced colon tumorigenesis in Min mice was not diminished in mice lacking Th17 (Th17^null CD4CRExSTAT3FLOX Min mice name hereafter Min-CD4^Stat3−/−), compared with parental Min mice. This later tumor development was nevertheless IL17-dependent. ETBF colonization induced initially less robust IL17 production in the colons of Th17-deficient Min-CD4^Stat3−/− mice and the only other significant source of IL17 in Min-CD4^Stat3−/− mice contributing to tumorigenesis was the γδ T-cell subset. While γδ T cells were not essential for tumorigenesis in Th17-competent mice, their ablation in Th17-deficient mice completely eliminated tumorigenesis. Whereas IL17 was recently proposed to protect Min mice from malignant progression of small intestine adenomas (23), paradoxically, our findings definitively identify ETBF-induced IL17 in the colon as procarcinogenic with mucosal γδ T cells as an innate source of IL17-complementing Th17 cells.

Min^{ApcΔ716+/−} (Min) mice were provided by Drs. David Huso and Bert Vogelstein (Johns Hopkins University School of Medicine, Baltimore, MD). Villin^Cre/Cre, TCRδ KO (γδ⁻) and RAG1 KO mice were purchased from The Jackson Laboratory. CD4^Cre/Cre and Stat3^flox/flox mice were provided by Dr. Charles Drake (Johns Hopkins University School of Medicine). IL17A knockout (KO) mice were obtained from Dr. Yoichiro Iwakura (University of Tokyo, Tokyo, Japan). Bone marrow chimera mice were established by tail injection of 10⁷ bone marrow cells from donor mice into lethally irradiated (900 rad) recipient Min mice. Animal protocols were approved by the Johns Hopkins University Animal Care and Use Committee. Mice were inoculated with ETBF strain 86-5443-2-2 (ETBF 086) as described previously (2). Orogastric inoculation with antibiotic-resistant ETBF 086 (∼0.1–0.2 mL/bacterial strain or ∼10⁸ CFU/mouse) or the buffer control (Sham) was performed in mice treated for 5 to 7 days prior inoculation with 5 g/L of streptomycin (Sigma) and 100 mg/L of clindamycin (Pharmacia) in their drinking water. Fecal samples were cultured to quantify strain colonization.

Macroscopic tumors were counted in formalin-fixed, methylene blue-stained colons blindly by two investigators (S. Wu and C.L. Sears). For histologic studies including microadenoma quantification, formalin-fixed tissue was paraffin-embedded (FFPE), sectioned (5 μm), and stained with hematoxylin and eosin (H&E staining). Inflammation scoring was performed blinded to identifiers by a board-certified veterinary pathologist (D. Huso), using a previously described scale (2, 9).

Colons were collected, flushed, and enzymatically processed as described previously (2). Briefly, minced distal colons were washed 3 times for 20 minutes at 37°C in 2 mmol/L EDTA, 10% FCS, 25 mmol/L Hepes, Hank balanced salt solution buffer and subsequently digested using a Liberase/DNAse 1 (Sigma-Aldrich). Mononuclear cells were isolated by Percoll gradient separation (GE Healthcare Life Science). Splenocytes were isolated from enzymatically dissociated spleen using Lymphoprep density gradient (Accurate Chemical & Scientific Corporation). For cytokine intracellular staining (ICS), cells were stimulated for 4.5 hours with phorbol 12-myristate 13-acetate (PMA; 30 nmol/L), ionomycin (1 μmol/L) in presence of monensin (GolgiStop, BD Biosciences). Antibodies used for staining are listed in Supplementary Table S1. A live/dead (L/D) dye (Life Technologies) was used to exclude dead cells. FACS data were acquired using a LSRII cytometer (BD Biosciences) and analyzed with DIVA software (BD Biosciences). In some experiments, populations were cell sorted using Aria II (BD Biosciences). To block IL17, mice were injected intraperitoneally with 150 μg of anti-IL17a mAb (rat IgG1κ, clone TC11-18H10.1, Biolegend) or isotype control mAb twice weekly as described in each figure.

Sorted cells were collected in lysis buffer of RNeasy purification kit, to perform RNA extraction following the manufacturer's instructions (Qiagen). RNA was converted to cDNA (High Capacity RNA-to-cDNA Kit, Life Technologies). A preamplification step was performed using preamplification Master Mix Kit (Life Technologies) and a pool of the primers used for the TaqMan PCR. The preamplified material is tested for the expression of target genes, and normalized by Gapdh expression using TaqMan-based technology (Life Technologies). Results were expressed as 2^−ΔC_t or 2^−ΔΔC_t.

Tumor tissues were collected at Johns Hopkins Hospital and University of Malaya Medical Center (UMMC, Kuala Lumpur, Malaysia) from patients with primary sporadic colorectal cancer and free of prior chemotherapy. Demographic and pathologic statuses are detailed in Supplementary Tables S2-S3. This study was approved by the JHU Institutional Review Board and UMMC Medical Ethics Committee. All samples were obtained in accordance with the Health Insurance and Accountability Act. Tumor specimens were collected and dissociated using an enzymatic cocktail as described previously (24). Antibodies for staining and FACS analysis are listed in Supplementary Table S1.

Comparison of means was done by unpaired, two-tailed Mann–Whitney U testing. A P value of < 0.05 designates a significant difference. For analysis of the tumor size distribution between ETBF-infected mice strains, we used χ² test. A P value of < 0.05 designates a significant difference.

ETBF colonization of C57BL/6 mice leads to rapid and persistent activation of Stat3 selectively in the colonic mucosa (2, 9). Stat3 phosphorylation in myeloid and lymphoid cells is first detected 6 hours after ETBF colonization, followed by Stat3 activation in colonic epithelial cells (CEC). We previously reported that ETBF colonization is long lasting, generally persisting for up to one year (9). As Stat3 is a key transcription factor in Th17-driven inflammation-associated cancer and Th17 cells were critical to ETBF-triggered IL17 colitis and tumorigenesis as we previously reported (2), we sought to further assess the contribution of Stat3 signaling selectively in the CD4 compartment. To do so, we generated CD4-CreXStat3-Flox (CD4^Stat3−/−) mice, and compared colonic inflammation with wild type (WT) C57BL/6 mice after ETBF colonization. Because Stat3 is a critical transcription factor for Il17a gene expression, Th17 development in CD4^Stat3−/− mice is completely inhibited following ETBF colonization (Fig. 1A; ref. 25). Obliteration of IL17 production from CD4⁺ T cells was found in all tissues tested including small intestine and colon lamina propria, spleen, and lymph nodes (25). In ETBF-colonized WT (CD4^Stat3+/+) mice, IL17-producing γδT cells and CD4⁺ T cells represent 12.5% ± 3.9 % and 55% ± 6.5% of the total IL17-producing cells in lamina propria, respectively (Supplementary Fig. S1). Upon colonization with ETBF, CD4^Stat3−/− mice develop colitis that is histologically indistinguishable from ETBF colitis in WT mice (Supplementary Table S3). γδT17 cells, which do not express CD4, were conserved during ETBF-triggered colitis in CD4^Stat3−/− mice and their numbers remained unchanged relative to WT mice (Fig. 1B). We did not detect a significant difference in the total number of IL17-producing cells in the colons of CD4^Stat3−/− mice developing ETBF-triggered acute colitis (day 7) compared with WT mice (Fig. 1C), likely, in part, due to higher numbers of T cells recovered from inflamed colons in these mice compared with WT mice (Supplementary Fig. S2). Total numbers of IFN-γ⁺ cells including IFN-γ⁺ CD4⁺ T (Th1) cells (P = 0.01), as well as IFN-γ⁺ γδT cells (P = 0.07) were increased in CD4^Stat3−/− mice (Fig. 1A and B) underscoring the inhibitory effect of Stat3 signaling on Th1 differentiation (25). Western blot analysis, performed on CD4-enriched splenocytes of ETBF-colonized WT and CD4^Stat3−/− mice, confirmed the absence of phosphorylated Stat3 (pStat3) in CD4⁺ T cells, proving the efficiency of the CD4-Cre system (Fig. 1D). IHC demonstrated that pStat3 was detected in the colonic epithelium and stroma of ETBF-colonized CD4^Stat3−/− mice, although modestly decreased (Fig. 1E and Supplementary Fig. S3). This result suggests selective Stat3 inactivation in the CD4 compartment may diminish, at early time points, CEC Stat3 activation in ETBF-colonized WT mice. This result is consistent with our previous report that, in ETBF-colonized WT mice, colon immune cell Stat3 activation precedes CEC Stat3 activation and that BFT, the key virulence protein of ETBF, does not directly activate CEC Stat3 (9).

Although the number of IL17-producing cells was not significantly decreased (Fig. 1C), we found that expression of Th17-associated genes including Il17a (P = 0.03; Fig. 1F), Il17f and Il21 (P = 0.05 and 0.0085, respectively; Fig. 1G and Supplementary Fig. S4) as well as Il23r (P = 0.18; Fig. 1G and Supplementary Fig. S4) were lower in colon tissues of CD4^Stat3−/− mice compared with WT mice. Of note, the abrogation of Stat3 signaling in CD4⁺ cells also led to the decreased expression of Treg-associated genes (Fig. 1G) including Foxp3, Il10, and Ctla4 (P = 0.02, 0.04, and 0.01, respectively; Supplementary Fig. S4). The ablation of Stat3 signaling in CD4 cells likely hinders Treg functions (i.e., decreased IL10 production) and therefore may account for the higher cellularity in the inflamed colons of ETBF-colonized CD4^Stat3−/− compared with WT mice (Supplementary Fig. S2). The decreased Il17a expression 7 days after ETBF colonization emphasizes that the absence of Th17 differentiation in CD4^Stat3−/− mice globally diminishes the ETBF-induced initial IL17 burst. Furthermore, the difference between IL17⁺ cell counts and RNA expression (Fig. 1C and F) suggests that, overall, alternate sources of IL17 that accumulate in the colon of ETBF-colonized CD4^Stat3−/− mice produce less IL17 per cell compared with Th17 cells, resulting overall in lower IL17 in the distal colon. However, γδT cells, whose numbers in the mucosa did not increase in ETBF-colonized CD4^Stat3−/− compared with WT mice (Fig. 1B), demonstrate higher production of IL17 than Th17 cells [mean fluorescence intensity (MFI) = 59,000 vs. 34,000, respectively; Supplementary Fig. S5). In contrast, the remaining non-γδ non-CD4⁺ T cells exhibit only a low MFI (17,000) for IL17 ICS (Supplementary Fig. S5).

While C57Bl/6 CD4^Stat3−/− mice exhibited an acute colitis similar to that observed in WT mouse upon ETBF colonization (Supplementary Table S4), Min-CD4^Stat3−/− are characterized by significantly decreased macroscopic colon tumor numbers 8 weeks after ETBF colonization compared with parental Min mice, validating our earlier report of the importance of Th17 responses for colon tumorigenesis within this time period after ETBF colonization (2). However, unexpectedly, tumor numbers in Min-CD4^Stat3−/−surged between 2 and 3 months following ETBF colonization. The tumor numbers were significantly higher at 3 months compared with 2 months after ETBF colonization of Min-CD4^Stat3−/− mice (23.4 ± 3.6 tumors per colon at 12 weeks vs. 6.8 ± 1.8 at 8 weeks postcolonization, mean ± SEM, P < 0.004) demonstrating a late acceleration of tumor development and/or growth in ETBF-colonized Min-CD4^Stat3−/− mice. In contrast, tumor numbers in parental ETBF-colonized Min mice stabilized between 2- and 3-month colonization (15.8 ± 1.9 tumors per colon at 3 month vs. 14.9 ± 2.7 at 2 months, mean ± SEM, P = 0.867). These findings show that, although delayed, ETBF-triggered tumorigenesis persisted in Min-CD4^Stat3−/− mice. When comparing microadenoma numbers in Min versus Min-CD4^Stat3−/− mice 4, 8, and 12 weeks after ETBF colonization, both strains of mice displayed a similar number of microadenomas (Fig. 2B and C), suggesting that, in Min-CD4^Stat3−/− mice, the diminished early IL17 burst (Fig. 1F) normally provided by Th17 cells in parental Min mice, impacted tumor growth rate rather than the ETBF-dependent initiation events. Consistent with this hypothesis, comparison of tumor size in 3-month ETBF-colonized mice revealed significantly higher numbers of smaller tumors in Min-CD4^Stat3−/− mice compared with Min-CD4^Stat3+/+mice (Supplementary Fig. S6). These findings underline a likely combined impact of IL17 on ETBF-mediated tumor initiation and tumor growth (2). It also follows that IL17-secreting non-Th17 cells can contribute to the initiation of the carcinogenesis.

To determine the nature of the protumoral factors in Min-CD4^Stat3−/− mice, we compared the inflammatory environment in colon tumors from 3-month ETBF-colonized parental Min and Min-CD4^Stat3−/− mice. We assessed expression of inflammation-related genes using qRT-PCR on whole tumor tissue. Surprisingly, we found that in CD4^Stat3−/−, similar to parental Min mice, Il17a gene expression was highly increased in ETBF tumors compared with surrounding normal colonic tissue (Fig. 3A). In two independent experiments, CD3⁺CD4⁺ T cells sorted from the Min-CD4^Stat3−/− colon tumors consistently expressed a higher level of Th1-(increased Tbx21, Stat1), Th2-(Gata3 and Il4) associated genes as well as T-cell effector function (Pdcd1, 4-1BB, Ox40, Klrg1, Ccr7, Itgae, and Il10ra) compared with those sorted from tumors from parental Min mice but were characterized by lower Th17-associated genes including Rorc, Il17a, IL17f, Il22, Il21, Il23r, and Ccr6 (Fig. 3B). This result confirms the absence of Th17 differentiation in Min-CD4^Stat3−/− 3 months after ETBF colonization. When testing the ability of tumor-infiltrating lymphocytes (TIL) to produce IL17 using ICS and flow cytometry analysis, we detected residual IL17⁺ cells including substantial numbers of γδ⁺ T cells in colon tumors derived from Min-CD4^Stat3−/− mice (Fig. 3C). Functional blockade of IL17 using anti-IL17 mAb injection twice weekly between 6 and 12 weeks after ETBF colonization significantly mitigated ETBF colon tumorigenesis in Min-CD4^Stat3−/− mice compared with mice injected with the isotype control mAb (Fig. 3D), indicating that sole blockade of IL17 is sufficient to impede ETBF-triggered tumorigenesis and importantly reinforces the direct contribution of IL17 to colon tumorigenesis. Interestingly, blockade of IL17 in ETBF-colonized parental Min mice between 6 and 12 weeks did not modify ETBF colon tumorigenesis (Supplementary Fig. S7), suggesting the critical role of IL17 in promoting tumor initiation and growth early during ETBF tumorigenesis in the presence of Th17 as we previously reported (2). Altogether, these findings demonstrate that IL17 but not Th17 cells per se is a critical requirement for ETBF tumorigenesis. In the absence of an initial Th17 response to ETBF colonization, the initial lower (Fig. 1F) and then seemingly slower accumulation of IL17 hinders but does not abrogate ETBF carcinogenesis.

We next sought to identify the IL17-producing cells promoting ETBF-mediated tumorigenesis in the absence of Th17 cells. Flow cytometry analysis of enzymatically digested tumor tissue from 3-month ETBF-colonized Min-CD4^Stat3−/− mice confirmed that only CD3⁺γδ⁺ T cells as well as undefined CD4⁻γδ⁻ CD3⁺ T cells secreted IL17 (Fig. 3C). We cell-sorted multiple leukocyte populations from colon tumors in ETBF-colonized Min-CD4^Stat3−/− mice and tested each of them for Il17a expression (Fig. 4A). These populations included granulocytic MDSC (PMN-MDSC; CD11b^hiGR1^hi cells), monocytic MDSC (MO-MDSC; CD11b^hiGR1^lo cells), and mature myeloid cells (CD11b⁺GR1^neg, including Mφ and dendritic cells (DC)], together with CD11b^negCD3^neg cells (including ILC3) and CD3⁺CD4⁺γδ⁻. Il17a expression was not detectable in CD11b⁺Gr1^neg or CD11b^negCD3^neg populations (Fig. 4A). Whereas ICS did not identify MDSC as IL17 producers (data not shown), qRT-PCR identified low level expression of the Il17a gene only in the MO-MDSC subset (Fig. 4A). Although MDSCs represented 80% to 90% of the infiltrating leukocytes in the tumor microenvironment (TME) and γδT17 cells only 0.1% (Supplementary Fig. S8), we found that γδ T cells expressed on average nearly 2,000 times more IL17 per cell than MO-MDSC (Fig. 4A). These findings established that, whereas MO-MDSC and γδT cells produce IL17 in the TME, γδT cells represent overwhelmingly the primary provider of IL17 in 3-month ETBF tumors when Th17 are missing. To confirm the role of γδ T cells as the principal alternate source of IL17 in the absence of Th17, we first studied colitis in ETBF-colonized γδ⁻-CD4^Stat3−/− mice, lacking γδ and Th17 cells. We found that γδ⁻ CD4^Stat3−/− mice expressed much less Il17a than WT and CD4^Stat3−/− C57BL/6 mice (Fig. 4B). γδ⁻-CD4^Stat3−/− mice also exhibited a reduced expression of MDSC-associated genes including Arg1 and Nos2 (70 and 9 times less, respectively, in Fig. 4C; P = 0.11 and 0.13, respectively, in Supplementary Fig. S4B), an immune profile that reflects overall diminished protumoral inflammation in γδ⁻-CD4^Stat3−/− mice compared with Th17^null-CD4^Stat3−/− mice 7 days after ETBF colonization, and supports the role of IL17 in recruiting MDSC in the TME (22).

Irradiated Min mice reconstituted with a RAG1^−/− BM [(Rag1^−/−→Il17a^−/−-Min)] and colonized with ETBF 6 weeks after engraftment developed no tumors 12 weeks after ETBF colonization (Fig. 4D), indicating that any IL17 provided de novo by myeloid, NK or ILC3 (Fig. 4A; Supplementary Fig. S9) cells was insufficient to sustain colon tumorigenesis. Colon tumorigenesis in ETBF-colonized [γδ⁻-CD4^Stat3−/− → IL17^−/−-Min] bone marrow chimera mice (no Th17 and γδ T cells) was markedly reduced compared with [WT → Il17a^−/−-Min] bone marrow chimeras, a result validating the tumorigenic potential of γδT17 cells (Fig. 4D and E). In contrast, ETBF-colonized Min-γδ⁻ mice showed no significant difference in tumor numbers compared with parental Min mice (Fig. 4F), demonstrating that, when Th17 are present, γδT17 cells are dispensable for tumor formation.

In contrast to murine transplantable tumor models and colonic epithelial barrier disruptive models of colonic inflammatory tumorigenesis (e.g., AOM/DSS), the ETBF model is induced by a human colonic bacterium linked to colitis and colon cancer (2). We sought to determine the distribution of IL17-producing cells in human colon cancer. Figure 5 and Tables 1 and 2 suggest that human colon cancer (JHU cohort, n = 12; Malaysian cohort, n = 11) has a similar distribution of Th17 and γδT17 cells as the ETBF model, that is, both Th17 and γδT17 cells are found in the majority of colon cancers (which have a wide range of overall IL17-producing TIL), with Th17 representing the majority of IL17 producers. This result contrasts with a recent publication proposing that γδT17 cells are overwhelmingly the primary source of IL17 in human colon cancer. This discrepancy does not appear to be due to the source of tumors (Asian vs. American populations), as we observe similar Th17 predominance in a cohort of patients from Malaysia (Table 2).

Our ETBF Min mouse model mirrors features of human colorectal cancer including altered APC/β-catenin signaling, a predominant distal location of colon tumors with validation of the proposed pathogenic role of protracted asymptomatic mucosal Th17 responses in human colorectal cancer (26, 27). We tackle herein the role and cellular origin of endogenous IL17 (induced by a single human colonic bacterium introduced to the microbiota as opposed to nonphysiologic chemical-induced models) in promoting tumor development and to distinguish the procarcinogenic role of adaptive Th17 cell-derived IL17 from alternative non-Th17 cell sources (primarily innate IL17). Inactivation of Stat3 in CD4 cells using CD4^Stat3−/− mice (28) led to ablation of Il17a transcription in CD4 cells as well as decreased expression of Th17-associated genes (Il17a, Il17f, Il21, Il22, and Il23r) during acute ETBF colitis (Day 7). However, though macroscopic, colon tumors were decreased 2 months after ETBF colonization, colon tumorigenesis rebounded and Il17a was highly expressed in tumor tissues, suggesting that non-CD4⁺ T cells (therefore not Th17, Treg or LTi) infiltrating colon tumors were producing IL17 (29). The unaltered early microadenoma numbers combined with the reduced tumor size in Th17^null Min-CD4^Stat3−/− compared with parental Min mice suggest that the lower levels of IL17 did not interfere with the ETBF-initiation step but likely impeded tumor growth. These findings emphasize that adaptive (Th17) and innate (non-Th17) IL17 sources differ in their kinetics but are redundant in their capacity to drive colonic tumor initiation and growth. Il17a transcription in the colon of ETBF-colonized CD4^Stat3−/− mice was associated with increased Arg1 and Nos2 expression (Fig. 4C) that play an essential role in the protumoral function of MDSC, a feature that emphasizes the persistence of a protumoral microenvironment in the colon of ETBF-colonized Min-CD4^Stat3−/− mice, although Th17 cells are absent.

Although access of microbiota to immune cells during acute ETBF colitis likely impacts the mucosal immune response to ETBF colonization, ETBF colon tumorigenesis requires molecular events triggered by the action of BFT on CECs that, in part, account for IL17-dependent tumor initiation and growth (3, 30). Indeed, in this model, Th17 detection in mice persists long after resolution of acute colitis and healing of colon barrier function (9). Importantly, these features reinforce the clear contrast in Min mice between spontaneous polyposis (mainly in small intestine), for which IL17 was reported in one publication to protect from cancer progression (23), and ETBF-induced tumorigenesis (strictly in the colon), for which IL17 is required to promote colon tumor (2). In another study, Min mouse small intestine tumorigenesis was also reported to be promoted by IL17 (31). A multitude of alternative IL17-producing cell subsets including γδ-T cells, NK cells, ILC3, and myeloid-derived cells such as PMN, MФ, and MDSCs exist (19, 20, 32–36). These innate cells are rapidly recruited to inflamed colon tissues, do not require antigen recognition but respond to stimulation of pattern recognition receptors (PRR) or the action of inflammatory mediators such as IL23 and IL1β (10). Herein, the ablation of colon tumorigenesis in bone marrow chimera Min mice reconstituted with RAG1^−/− or γδ⁻-CD4^Stat3−/− bone marrow provided confirmation of the protumoral role of γδT17 cells in the absence of Th17 cells while the other innate cells (ILC3, NK, and myeloid cells) although present in bone marrow chimera mice are not sufficient to provide IL17 in amounts able to sustain ETBF tumorigenesis.

γδT cells are detected at virtually all mucosal surfaces and skin where they participate in tumor immunosurveillance (16, 19, 22, 37). In microbial infection, γδT17 cells are critical to orchestrate the recruitment of monocytes and neutrophils (19, 38) and to amplify the ensuing adaptive Th17 response (17). That MO- and PMN-MDSC were highly recruited to colon tumors of Min-CD4^Stat3−/− mice in response to the BFT-mediated oncogenic alteration of CEC and IL17 production is compatible with the notion that γδT17 cells contribute to the production of protumoral IL17 as well as shape the TME, which includes the accumulation of MDSC (19, 21, 22). As a matter of fact, Wu and colleagues recently reported that virtually all the IL17-producing cells in human colon cancers were γδT cells and their detection was associated with the recruitment and survival of protumoral PMN-MDSC (22). Notably, we find a very similar distribution of Th17 and γδT17 cells subsets between our ETBF tumorigenesis model and human colon cancers that have high IL17 in their TME; namely, γδT17 cells represent a smaller proportion of the IL17-producing cell population than do Th17 cells (Fig. 5; Tables 1 and 2). It is unlikely that this marked discrepancy is due to fundamental differences in the colorectal cancer TME in Chinese versus American patients as we identified a similar distribution of Th17 and γδT17 cells between a cohort of U.S. colorectal cancer (Table 1) and Malaysian (that included Chinese patients; Table 2) samples. Our finding of Th17 and γδT17 redundancy in our model of de novo tumorigenesis driven by a human colonic bacterium also contrasts with the findings of Ma and colleagues in a transplantable murine hepatocellular carcinoma model where conventional Th17 cells played little role in promoting tumor growth while γδT17 cells (expressing Vγ4) were the predominant source of tumor growth-promoting IL17 (19). Coffelt and colleagues found in a KEP-based model of spontaneous breast cancer metastasis, that although both γδ and CD4⁺ T cells produced IL17 only depletion of γδ T cells led to a decreased level of IL17 in serum, and number of pulmonary metastasis (21). Therefore, whether γδT17 cells are protumorigenic or not appears to be highly context and model-dependent. However, the contribution of γδT17 cells to ETBF tumorigenesis when Th17 are present remains uncertain. ETBF infection of Min-γδ⁻ mice resulted in similar tumorigenesis as in parental Min mice, suggesting limited impact of γδT17 cells in the presence of Th17 cells. Early γδT17 cells have been shown to be important for triggering and amplifying the ensuing adaptive Th17 immune response (17). Combined, these pathways are novel and likely highly relevant to the pathogenesis of human colorectal cancer. Our model of microbial-induced de novo carcinogenesis has direct translational implications as it suggests that therapeutic approaches to limit pathogenic Th17 will likely be insufficient to impact colon carcinogenesis if alternative sources of IL17 (i.e., γδT cells and myeloid-derived cells) do not respond to the same regulatory mechanisms. Finally, we propose that our results herein, in combination with our previous demonstration that early inhibition of IL17, using anti-IL17 mAb injection (2) or Treg depletion (39) in ETBF-colonized Min mice impedes microadenoma formation, support a novel paradigm that IL17, independent of its origin, has carcinogenic potential, promoting not solely tumor growth but also tumor initiation.

No potential conflicts of interest were disclosed.

Conception and design: F. Housseau, C.L. Sears

Development of methodology: F. Housseau, S. Wu, X. Wu, J.S. Vadivelu, S.V. Meerbeke, C.L. Sears

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Housseau, S. Wu, E.C. Wick, H. Fan, X. Wu, N.J. Llosa, K.N. Smith, A. Tam, S. Ganguly, J.W. Wanyiri, A.A. Malik, A.C. Roslani, J.S. Vadivelu, S.V. Meerbeke, D.L. Huso

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Housseau, S. Wu, N.J. Llosa, K.N. Smith, A. Tam, A.C. Roslani, D.L. Huso, C.L. Sears

Writing, review, and/or revision of the manuscript: F. Housseau, S. Wu, A.C. Roslani, J.S. Vadivelu, D.L. Huso, D.M. Pardoll, C.L. Sears

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Housseau, H. Fan, T. Iyadorai, A.C. Roslani, J.S. Vadivelu

Study supervision: F. Housseau, A.C. Roslani, J.S. Vadivelu, C.L. Sears

The authors thank the SKCCC Flow Core for support.

This work was supported by grants from the NIH (RO1CA151393 to C.L. Sears and D.M. Pardoll; R01DK080817and R01CA151325 to C.L. Sears; P30DK089502 GI Core; P30CA006973 SKCCC core; K08 DK087856 to E.C. Wick). Funding for this work was also provided by University of Malaya Research Grant (UMRG RP016A-13HTM, JSV).

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.