Toll-Interacting Protein Regulates Immune Cell Infiltration and Promotes Colitis-Associated Cancer.

Expression of Toll-interacting protein (Tollip), a potent TLR modulator, decreases in patients with inflammatory bowel diseases (IBD), whereas Tollip-/- mice are susceptible to colitis. Tollip expression was shown to be reduced in sporadic adenoma. In contrast, we found variable Tollip expression in patients with colitis-associated adenomas. In Tollip-/- mice challenged to develop colitis-associated cancer (CAC), tumor formation was significantly reduced owing to decreased mucosal proliferative and apoptotic indexes. This protection was associated with blunt inflammatory responses without significant changes in microbial composition. mRNA expression of Cd62l and Ccr5 homing receptors was reduced in colons of untreated Tollip-/- mice, whereas CD62L+ CD8+ T cells accumulated in the periphery. In Tollip-deficient adenomas Ctla-4 mRNA expression and tumor-infiltrating CD4+ Foxp3+ regulatory T cell (Treg) were decreased. Our data show that protection from CAC in Tollip-deficient mice is associated with defects in lymphocyte accumulation and composition in colitis-associated adenomas.

Introduction

Patients with inflammatory bowel diseases (IBD) with long-standing and extensive colitis are at higher risk of developing a distinct type of colorectal cancer, termed colitis-associated cancer (CAC) (Canavan et al., 2006, Rutter et al., 2004). Although CAC etiology is complex and still incompletely understood, compromised epithelial integrity and deregulated Toll-like receptors (TLRs)-mediated microbiota sensing are important risk factors (Danese and Mantovani, 2010). Disruption of this microbiota sensing mechanism either at the receptor level or via downstream adaptor molecules, such as Myd88 ablation, results in increased production of pro-inflammatory cytokines and colitis development (Rakoff-Nahoum et al., 2004), favoring CAC onset (Lowe et al., 2010, Salcedo et al., 2010).

Dysbiosis and abundance of specific microbiota members can also influence carcinogenesis. Genomic analysis of human CRC biopsies detected a high prevalence of bacterial species such as Fusobacterium nucleatum and Bacteroides fragilis and have the ability to aggravate tumor propensity in mice (Kostic et al., 2012, Toprak et al., 2006). Consistently, blocking of TLR/MyD88-mediated sensing abrogated colorectal carcinogenesis (Rakoff-Nahoum and Medzhitov, 2007).

Regulation of NF-κB activity is essential to avoid spontaneous and prolonged inflammatory responses upon receptor binding. In mice lacking IκB kinase β (IKKβ), CAC susceptibility (Greten et al., 2004) was attributed to enhanced NF-κB-mediated IL-6 production and activation of STAT3-dependent epithelial neoplastic changes (Bollrath et al., 2009). Similarly, ablation of TLR4 or chemokine receptor signals results in impaired immune cell infiltration and protection of mice against CAC (Popivanova et al., 2009, Katoh et al., 2013, Fukata et al., 2007). Consistently, lack of negative regulators of innate immune signals, such as A20, single immunoglobulin receptor related molecule (SIGIRR), and interleukin receptor associated kinase-M (IRAK-M), was shown to enhance susceptibility to CAC in both humans and mice (Shao et al., 2013, Xiao et al., 2007, Kesselring et al., 2016, Begka et al., 2016). In CAC, although epithelial changes are essential to promote colon cancer, immune cell recruitment into the inflamed mucosa fuels tumor growth through perpetuation of inflammation (Greten et al., 2004). Regulatory T cells (Tregs) control intestinal inflammation, but their contribution in colorectal carcinogenesis remains elusive and controversial (Whiteside, 2012, Ladoire et al., 2011).

Toll-interacting protein (Tollip) is an intracellular, ubiquitously expressed adaptor protein, initially described as a negative regulator of NF-κB signaling (Burns et al., 2000, Bulut et al., 2001). Upon activation of IL-1R, TLR2, and TLR4, Tollip recruits interleukin receptor associated kinase 1 (IRAK1) but inhibits its disassociation, depending on the stimulus strength, thus resulting in discontinuation of NF-κB pathway (Bulut et al., 2001, Burns et al., 2000, Zhang and Ghosh, 2002, Begka et al., 2016). Tollip also regulates the turnover of ubiquitinated receptors, such as IL-1R and TGFβ-RI, through clathrin-mediated endosomal trafficking and sorting to late endosomes (Brissoni et al., 2006, Zhu et al., 2012, Begka et al., 2016).

We have previously shown that Tollip deficiency increases susceptibility of mice to a dextran sodium sulfate (DSS) colitis model (Maillard et al., 2014). Tollip ablation resulted in disruption of tight junctions and increased intestinal permeability upon acute DSS chemical injury, whereas non-hematopoietic expression of Tollip partially restored disease susceptibility (Maillard et al., 2014). Tollip has also been linked to human IBD as its gene lies within one of the IBD susceptibility loci (van Heel et al., 2004, Ishihara et al., 2009). In addition, Tollip expression was downregulated in biopsies from both active and inactive colonic segments from patients with ulcerative colitis (UC) and Crohn's disease (CD) compared with healthy subjects (Fernandes et al., 2016).

Despite the demonstrated role of Tollip in colitis, its influence on inflammation-associated cancer remains unknown. In this study, we report that Tollip promotes colitis-driven carcinogenesis. We found that Tollip-deficient mice fail to mount efficient chronic inflammatory responses resulting in reduced tumor risk. Those global defects were also accompanied by qualitative changes in cellular composition with a reduction in tumor-infiltrating regulatory T cells.

Results

Variable Tollip mRNA Expression in Patients with CAC

Although Tollip expression is decreased in sporadic colorectal cancer (CRC) adenomas (Pimentel-Nunes et al., 2012), there are no data available for human colitis-associated cancer (CAC) lesions. Given the link between inflammation and inflammation-driven cancer, we hypothesized that Tollip should be downregulated in CAC lesions. Therefore, we investigated Tollip mRNA expression in patients with IBD with active or quiescent UC or CD, as well as in CAC or sporadic CRC adenomas, compared with adjacent non-tumorous (NT) mucosal biopsies from sporadic CRC. Consistent with previous reports, we observed reduced Tollip mRNA expression in mucosa of patients with quiescent and active UC and CD compared with the control NT mucosal biopsies (Figure 1). Moreover, Tollip expression was downregulated in paired mucosal samples of sporadic CRC adenomas compared with NT normal adjacent mucosa (Figure 1). Surprisingly, Tollip expression in patients with CAC was variable and not significantly decreased, as we initially hypothesized. Tollip expression was higher in 2/6 CAC samples, histologically characterized as adenocarcinoma, compared with 4/6 CAC samples of low-grade dysplastic lesions, as well as with patients with quiescent and active UC and CD, whereas it was comparable with NT normal adjacent mucosa biopsies (Figure 1). In summary, Tollip expression is differentially regulated in CAC lesions compared with sporadic adenomas. We decided to further investigate the role of Tollip in an in vivo mouse model of chronic colitis-associated cancer onset, in which adenomas do not develop a differential dysplastic grade.

Figure 1

Tollip Expression in Biopsies from Patients with IBD and Colon Cancer

qRT-PCR analysis of Tollip mRNA expression relative to Gapdh in active and inactive gut segments from patients with UC and CD. n = 14 and n = 8 for UC and n = 15 and n = 10 for CD, respectively. Tollip expression was also assessed in biopsies from adenomas developed in patients with IBD (CAC, n = 6) and sporadic colorectal cancer (CRC, n = 10). As control, RNA extracts from the non-tumoral colon adjacent to the CRC (n = 10) were taken. Data are mean ± SEM. Differences were analyzed by one-way ANOVA test for CAC versus non-active UC and CAC versus non-active CD; p = 0.0497 and p = 0.0458, were calculated, respectively. Differences between CRC and NT were analyzed by Wilcoxon matched-pairs test, p = 0.0020. *p < 0.05, **p < 0.01.

Tollip Ablation Is Protective against Colitis-Associated Cancer

To decipher the role of Tollip in CAC and avoid the variability introduced by human biopsies, we employed a CAC model in mice lacking Tollip expression. Tollip and wild-type (WT) littermates were challenged with the azoxymethanone (AOM)/dextran sodium sulfate (DSS) model to recapitulate CAC development (Figure 2A). Adenoma growth and development were monitored by mouse colonoscopy at day 63 (Figures 2A and 2B). We observed significantly fewer adenomas and reduced endoscopic tumor scores (Becker et al., 2005) in the AOM/DSS-exposed Tollip compared with WT mice, indicating that Tollip ablation resulted not only in reduced tumor incidence but also in smaller lesions (Figures 2B and 2C). Consistently, macroscopic examination of the excised colons demonstrated reduced tumor number upon Tollip ablation (Figures 2D and 2E). We also observed a significant increase in small (<2 mm) lesions and decrease in larger (2–4 mm) lesions in Tollip compared with WT mice (Figure 2F), indicating that Tollip ablation results in the formation of smaller adenomas, in accordance with the observed endoscopic score (Figure 2C). Histological analyses confirmed the development of adenomatous AOM/DSS-induced polyps (Figure 2G), whereas frequency of dysplastic low-high grade or adenocarcinoma development was similar between Tollip and WT mice (Figure 2H). These data, together with our previous findings, suggest that Tollip has a dual role in favoring CAC development despite being protective against acute colitis.

Figure 2

Tollip Deficiency Attenuates Colitis-Associated Carcinogenesis

(A) Tollip and WT littermates C57BL/6 mice received a single intraperitoneal AOM injection (10 mg/kg) followed by three cycles of 2.5% DSS in drinking water. Unchallenged Tollip and WT littermates were used as controls. On day 63, mice were subjected to colonoscopy to assess adenoma development and size.

(B) An endoscopic score was determined according to a standardized grading system (Becker et al., 2005). Representative endoscopic images from AOM/DSS exposed WT (upper panel) and Tollip littermates at day 63. Tumors are depicted with blue arrows and ulcer formation with green arrowheads.

(C) Endoscopic tumor number and score in AOM/DSS-exposed WT and Tollip littermates and unchallenged animals. Each dot represents data from an individual mouse. Data are mean ± SEM; n = 3–5 for unchallenged and n = 8–14 for treated groups. Differences were analyzed by Mann-Whitney test. p = 0.0318, p = 0.0106, for tumor number and score, respectively.

(D) Macroscopic evaluation of adenomatous lesions in excised colons from treated Tollip and WT littermates.

(E and F) Quantification of overall macroscopic tumor number and size in AOM/DSS exposed Tollip and WT littermates. Data are mean ± SEM; n = 8–14. Differences were analyzed by Mann-Whitney test, p = 0.0299 for macroscopic tumor number. Differences were analyzed with multiple t tests for macroscopic tumor size; p = 0.02 and p = 0.0001 for <2 and 2–4 mm, respectively.

(G) Histological evaluation of tubular adenomas formation by hematoxylin and eosin staining (H&E) in paraffin-embedded colons of Tollip WT (i–iii) and Tollip treated mice (iv–vi) in increasing magnifications of 10×, 20×, and 40×. Scale bars: 100 μm.

(H) Histopathological score of adenomas in H&E staining of paraffin-embedded colons, LG, low-grade dysplasia; HG, high-grade dysplasia; ADK, adenocarcinoma. Differences were analyzed by two-tailed unpaired t test. Data are mean ± SEM. n = 8–19. ns p > 0.5, *p < 0.05, ***p < 0.001.

Tollip Deficiency Is Associated with Reduced Cell Turnover in Colonic Adenomas

Given the protective role of Tollip in colonic epithelium during acute inflammation (Maillard et al., 2014), we hypothesized that reduced tumor incidence in Tollip mice could be related to defects in cell turnover rates. To address this hypothesis, we assessed apoptosis during tumor initiation and late tumor stage in the CAC model. At the beginning of AOM/DSS treatment (day 8) we observed a significant increase in epithelial apoptotic Tunel+ cells in Tollip-deficient colons (Figures 3A and 3B). However, at late tumor stage (day 63) the apoptotic index was significantly decreased in both Tollip-deficient adenomas (Figures 3C and 3E) and crypts of adjacent non-adenomatous mucosa of Tollip-treated animals compared with WT-treated mice (Figures 3D and 3F). Notably, the density of apoptotic cells was higher at the tip of colonic crypts in acute inflammation/early stage carcinogenesis on day 8 (Figure 3A), whereas analysis of late-stage carcinogenesis on day 63 showed a shift toward the lamina propria (Figure 3D). The apoptotic index was similar in unchallenged mice (Figure 3F) and our multi-screen gene analysis showed a downregulation of additional apoptosis-related genes in Tollip-deficient adenomas compared with WT (Figure S1A). We also observed a significantly reduced proliferative index in Tollip-deficient compared with WT adenomas (Figure 3G). We next examined STAT-3 activation (Figures S1B–S1D), along with important upstream and downstream targets involved with neoplastic properties of the STAT3 pathway (Figure S1E). Immunohistochemical analyses of the active form of STAT3 (P-STAT3) (Figure S1B) and relative P-STAT3 expression levels, although variable, were comparable between Tollip and WT adenomas (Figures S1C and S1D) and in colons of untreated animals (Figure S1D). Similarly, Bcl-xL, c-myc, Bax, IL-6, and IL-11 mRNA expression was comparable between Tollip and WT adenomas (Figure S1E). In unchallenged Tollip mice, Bcl-xL and c-myc mRNA expression was lower in whole colon homogenates compared with WT mice, indicating that the mucosal environment upon homeostatic conditions tends to be skewed toward a pro-apoptotic response (Figure S1E). However, this was not correlated with an impaired STAT3 signaling during tumorigenesis as both total and P-STAT3 protein abundance were comparable in Tollip and WT adenomas, as well as in unchallenged mucosa (Figures S1B–S1D). Taken together, these results indicate that Tollip deficiency induces pronounced epithelial apoptosis upon early tumorigenesis but, in the late tumor stage, Tollip deficiency results in retarded development and progress of adenomas with low turnover rates.

Figure 3

Tollip-Deficient Adenomas Present Reduced Proliferative and Apoptotic Indexes

(A) Tollip and WT C57BL/6 littermates received a single i.p. AOM injection (10 mg/kg). After 5 days mice received oral treatment of 2.5% DSS in the drinking water and were euthanized on day 8; n = 5–6 mice. Cells apoptosis was quantified via immunofluorescent analysis of TUNEL+ cells on day 8. DAPI was used for nuclear staining. Scale bars: 100 μm. Insets show a magnified view of the TUNEL+ apoptotic cells.

(B) Quantification of TUNEL+ apoptotic cells in 20× field for at least three representative pictures/mouse. Data are mean ± SEM. Differences were evaluated with Mann-Whitney test. p = 0.0001.

(C and D) (C) Immunofluorescent evaluation of apoptosis by TUNEL staining in adenomas and (D) normal adjacent mucosa of AOM/DSS-exposed mice. DAPI was used for nuclear staining. Scale bars: 100 μm. Insets show a magnified view of the TUNEL+ apoptotic cells.

(E and F) (E) Quantification of TUNEL+ apoptotic cells in adenomas and (F) normal adjacent mucosa of treated as well as in colons of untreated mice in 10× field view for at least three representative pictures per mouse. Data are mean ± SEM. n = 5–14. Differences were analyzed by Mann-Whitney test. p = 0.0059 and p < 0.0001 for adenomas and normal mucosa crypts, respectively.

(G) Representative immunohistochemical Ki67 and nuclear staining with hematoxylin in adenomas of paraffin-embedded colons from AOM/DSS-treated mice. Scale bars: 100 μm. Analysis of proliferative index by measuring the percentage of surface of Ki67/nuclear staining ratio specifically for the adenomas by ImageJ algorithm Immunoratio. Counting was performed in 10× field view for at least three representative pictures/adenoma. Data are mean ± SEM. n = 5–7 mice and n = 21–34 adenomas. Differences were analyzed by Mann-Whitney test, p = 0.0357. **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Reduced Cancer Risk in Tollip-Deficient Animals Is Not Associated with Changes in Gut Microbial Composition

Given the role of Tollip as regulator of the Toll/IL-1R signaling and the implication of several of those pathways in microbiota shaping, we hypothesized that changes in gut microbial composition might account for reduced cancer occurrence in Tollip-deficient mice. We performed 16S rRNA amplicon sequencing on feces of co-housed Tollip and WT littermates prior and after AOM/DSS treatment. There was no difference in microbiota diversity between the two groups both at baseline (Figure S2A) and after treatment (Figure S2B). As inflammation-induced colon tumorigenesis has been linked with shifts in microbiota composition, we next addressed differences in microbiota composition between Tollip and WT littermates. There were no differences in overall microbiota composition both before (ANOSIM p = 0.474) (Figure S2C) and following AOM/DSS treatment (ANOSIM, p = 0.5083, Figure S2D). The relative abundance of bacterial taxa at the phylum level was comparable between Tollip and WT mice both before and after treatment (Figure S2E). Finally, we evaluated changes in specific bacterial taxonomic groups using differential abundance testing. We detected a minor enrichment of unclassified genus of Peptococcaceae (p = 0.019, LDA score = 1.5326) in WT mice and Rikenellaceae family of also unclassified genera and of unclassified families within the Bacteroidales order (p = 0.028) in stools of Tollip mice (p = 0.034, LDA score = 2.1055 and 2.0850) (Figure S2F) following AOM/DSS treatment. Finally, AOM/DSS treatment itself led to a decrease in families within the Proteobacteria phylum, independently of genotype (Figure S2G). Altogether, these data demonstrate that reduced tumor incidence upon Tollip deficiency does not correlate with a tumor-protective shift in microbiota diversity or global composition.

Tollip Modulates Migration Patterns and Cell Infiltration in Colonic Mucosa

We next aimed to investigate whether Tollip ablation could influence mucosal immune inflammation-related cell composition. Multi-gene RNA screening in healthy colons of unchallenged animals demonstrated a significant reduction in L-selectin (Cd62L) mRNA expression (Figure 4A), an adhesion molecule known to regulate trafficking of naive or central memory T (TN or TCM) cells from blood to secondary lymphoid organs (Masopust and Schenkel, 2013). We also observed a non-significant reduction in several T cell homing markers such as Cd44, Il7r, and Ccr7, as well as the integrins Itgβ2 and Itgα4 that form the heterodimers LFA-1 and the gut-homing α4β7 (Masopust and Schenkel, 2013), respectively (Table S1). Next, we assessed whether T cell recruitment to the healthy gut mucosa was influenced by Tollip ablation under basal conditions. We found no differences in the CD3+/CD4+/CD8+ T cell abundance in the colons of unchallenged mice (Figure 4B). However, investigation of the activation status of Tollip-deficient T cells in the periphery demonstrated a significant increase in the percentage of CD8+ CD44- CD62L+ TN (naive) cells in spleen, blood, and peripheral lymph nodes (pLNs) (Figures 4C and 4D). The percentages of CD8+ CD44hi CD62L− T effector (Teff) (Figures 4C and 4D) and CD4+ CD44- CD62L+ TN and CD4+ CD44hi CD62L− Teff cell were comparable between Tollip and WT mice (Figures S3A and S3B). Interestingly, we also observed a significant reduction in Ccr5 mRNA expression in Tollip mice (Figure 4A, Table S1), a chemokine receptor known to mediate the recruitment of immune cells upon inflammatory conditions (Griffith et al., 2014). Collectively, these results suggest that Tollip deficiency favors naive cell accumulation in peripheral lymphoid organs upon homeostatic conditions.

Figure 4

Tollip Deficiency Alters Migration Patterns upon Homeostatic Conditions

(A) Volcano plot depicts differential gene expression in colonic mucosa of untreated Tollip and WT littermates (data obtained from three biological replicates per group).

(B) Flow cytometry analysis of colonic lamina propria T cells frequency and absolute cell numbers in untreated Tollip and WT littermates. Data are mean ± SEM. n = 5–7 mice.

(C) Representative FACS plots of CD44+ and CD62L+ T cells frequencies in spleen and blood, pre-gated on CD8+ T lymphocytes.

(D) Flow cytometry analysis of naive and effector CD8+ T cells frequencies isolated from peripheral (pLN) and mesenteric (mLN) lymph nodes, spleen, and blood. Data are mean ± SEM; n = 4–5 mice. Differences were analyzed by one-way ANOVA test. p = 0.045, p = 0.0031, and p = 0.0015 values were calculated for pLN, spleen, and blood, respectively. *p < 0.05, **p < 0.01.

Tollip Deficiency Impedes Infiltration of CD4+ and Regulatory T Cells in Adenomas

We next questioned whether the altered cell trafficking noted upon Tollip ablation during homeostatic conditions would impede immune cell accumulation and attenuate inflammatory responses upon AOM/DSS challenge. Despite Tollip commonly being described as a negative regulator (Zhang and Ghosh, 2002), several pro-inflammatory genes were downregulated in Tollip-deficient compared with WT adenomas (Figure S4A). Investigation of other inflammatory indexes, such as colonic shortening, was moderate in treated Tollip mice compared with the significantly reduced colon length observed in WT mice (Figure S4B). We also noticed a significant reduction in CD45+ cells frequency and absolute number infiltrates in Tollip-deficient adenomas compared with WT (Figure S4C). In addition, we also observed a prominent reduction of IFNγ- and TNFα-producing inflammatory cells infiltration into colons of challenged Tollip mice compared with WT mice (Figure S4D). Collectively, these data suggest that Tollip aggravates inflammatory responses at late tumor stage development (day 63), whereas no differences in CD45+ frequency and infiltration were observed in pLNs, mLNs, ileum, and spleen (Figure S5A).

Given our observation of reduced immune cell infiltration in Tollip adenomas (Figure S4C), we further questioned whether reduced CD45+ cell accumulation was associated with defects in CD3+ T cell infiltration. As expected, we observed a marked colon-specific decrease in the frequency and number of CD3+ T cell infiltrates in Tollip compared with WT adenomas (Figure 5A). This decrease was mainly due to reductions in CD4+ T cells frequency and infiltrates (Figure 5A). Importantly, this defect was only seen in the colonic adenomas but not in peripheral lymphoid organs (Figure S5B), suggesting that the observed changes are restricted to the mucosal environment during carcinogenesis.

Figure 5

Tollip Deficiency Impairs T Regulatory Cells Accumulation into Colitis-Associated Adenomas

(A) Representative FACS plots of tumor-infiltrating T cells and analysis of CD3+, CD4+, and CD8+ T cells frequencies and absolute cell numbers isolated from adenomas of Tollip and WT mice. Data are mean ± SEM; n = 6 mice. Differences were analyzed by multiple t test. p = 0.03 and p = 0.003 values were calculated for CD3+ and CD4+ T cell frequencies, and p = 0.004 and p = 0.004 values were calculated for CD3+ and CD4+ T absolute cells numbers, respectively.

(B) Cell type mRNA profiling scores for CD45+, T cells, and T regulatory (Treg) cells are calculated for adenomas and colon mucosa of treated and untreated Tollip and WT mice.

(C) Volcano plot depicts differential gene expression in adenomas of AOM/DSS-exposed Tollip versus WT littermates (data obtained from three biological replicates per group).

(D) Relative mRNA expression of Foxp3, TGF-β, and IL-10 analyzed by real-time PCR. Gadph was used as a house keeping gene. Data are mean ± SEM; n = 5–19 samples. Differences were analyzed by two-tailed unpaired t test. p = 0.0228 for colons of untreated mice and p = 0.0027 Foxp3 values for adenomas were calculated and p = 0.0317 for TGF-β for adenomas.

(E) Immunoblot analysis for phosphorylated Smad-2 (P-Smad-2, band at 60kD) and total Smad-2 (band at 60kD) protein extracted from whole colon homogenates from tumors (T) of both treated mice. β-Actin (at 42 kD) was used as a loading control. The pre-stained protein ladder appears before the protein bands of interest. The molecular bands of 100, 55, and 35 kD are noted in the pre-stained protein ladder that appears before the protein bands of interest. The same nitrocellulose membrane was used for the immunoblotting of all three markers.

(F) Representative FACS plots of tumor-infiltrating CD4+ CD25+ Foxp3+ T regulatory cells and analysis of CD4+ Foxp3+ and CD4+ CD25+ Foxp3+ Treg cells frequencies and absolute numbers isolated from adenomas of Tollip and WT mice. Data are mean ± SEM; n = 6 mice. Differences were analyzed by t test. p = 0.023 and p = 0.0041 values were calculated for Treg frequencies and absolute cell numbers, respectively. *p < 0.05, **p < 0.01.

To further investigate this reduced colon-specific infiltration of inflammatory cells, we performed a multi-screen RNA analysis. In line with the flow cytometry data (Figure S4C), a lower CD45 mRNA cell score was predicted for the challenged Tollip compared with WT mice (Figure 5B), indicative of attenuated inflammatory responses. Tollip expression was used as internal control and was, as expected, not expressed in Tollip-deficient samples (Figure 5C). We observed a non-significant trend of reduced expression of genes relevant to the immune cells homing to inflammatory sites in colons of treated Tollip versus WT mice (Table S2). Additionally, we found a significant reduction in expression of three key T cell receptor signaling-associated genes in Tollip-deficient adenomas (Tanoue et al., 2016, Hill et al., 2002, van de Ven and Borst, 2015), namely, Ctla4, Cd27, and Ptpn22 (Figure 5C, Table S2).

Ctla4 is a costimulatory molecule that prevents excessive T cell activation upon TCR ligation. It has also been shown to be constitutively expressed on regulatory T cells (Sansom, 2000). Given the observed reduction in Ctla4 expression in Tollip-deficient adenoma, we hypothesized that defects in immunosuppressive pathways could be observed upon Tollip ablation. Indeed, cell type mRNA profiling suggested reduced T lymphocytes and regulatory T cell abundance in both whole colon homogenates of unchallenged and adenomas of AOM/DSS-challenged Tollip mice (Figure 5B). Accordingly, we noted a significant reduction in Foxp3 mRNA expression in both healthy mucosa and adenomas of Tollip compared with WT mice (Figure 5D). Furthermore, TGF-β but not Il10 mRNA expression was downregulated in Tollip-deficient adenomas (Figure 5D). Consistently, we found a significant decrease in phospho-Smad2 and total Smad2 protein abundance that are known downstream effectors of TGF-β (Figure 5E). Altogether, those data demonstrate that Tollip ablation leads to reduced immunoregulatory T cell accumulation and blunted TGF-β/TGFβR-mediated responses.

Finally, we sought to determine the Treg frequencies and cell numbers systemically (Figure S5C) and in adenomas (Figure 5F). Indeed, we wanted to examine whether reduced CD3+ T cell accumulation in adenomas was also associated with an imbalance between effector and regulatory T cells. In association with our prior results, we detected reduced CD4+ Foxp3+ and CD4+ Foxp3+ CD25+ frequencies and cell numbers infiltrating Tollip-deficient adenomas compared with WT (Figure 5F). Conversely, no differences in Treg frequency or cell number were noted in other lymphoid organs (Figure S5C). Altogether, our data show that Tollip deficiency leads to reduced CD3+ T cell accumulation in colonic adenoma together with a local imbalance between effector and regulatory T cells.

Collectively, these data suggest that decreased adenoma risk in Tollip-deficient mice is associated with impeded immune cell infiltration and attenuated inflammatory responses together with abrogated Treg infiltration.

Discussion

Aberrant activation of TLR and NF-κB signaling modifies risk of colitis and cancer development in mice (Shao et al., 2013, Xiao et al., 2007, Kesselring et al., 2016, Begka et al., 2016). Tollip is an intracellular adaptor able to negatively regulate IL-1R and TLR2/TLR4-mediated NF-κB activation and to promote tolerance in intestinal epithelial cells (IECs) (Bulut et al., 2001, Burns et al., 2000, Otte et al., 2004). We hypothesized that Tollip deficiency would promote CAC risk, based on our previous findings (Maillard et al., 2014) and what had been previously reported on other innate negative regulators (Salcedo et al., 2010, Xiao et al., 2007, Begka et al., 2016). However, we observed that Tollip mice were partially protected against CAC and that Tollip expression tended to be downregulated in dysplastic lesions from patients with UC.

Although prominent epithelial damage was observed early on the AOM/DSS regime, this only evolved to blunted tumor development characterized by reduced cell turnover. This could be in part due to a reduced number of surviving epithelial cells subjected to AOM-induced mutagenesis. Although increased epithelial apoptosis activates STAT3-mediated pro-repair and neoplastic responses (Bollrath et al., 2009), histologically we did not find any difference in STAT3 protein expression. Similarly, analysis of STAT3 activation pathway showed that it remained unchanged upon Tollip deficiency.

Following epithelial destruction and microbial incursion, infiltrating leukocytes not only support anti-microbial responses but also fuel tumorigenesis via excessive cytokine and growth factor production (Danese and Mantovani, 2010, West et al., 2015). Interestingly, Tollip deficiency led to poor leukocyte recruitment to tumoral tissues despite prominent epithelial disruption at early stages of acute DSS-induced injury. Our findings correlate with prior reports on TLR4 signaling, in which Tollip actively participates. Indeed, TLR4 deficiency promoted colitis induction but attenuated tumor development and growth due to reduced leukocyte infiltration into inflamed mucosa (Fukata et al., 2005, Fukata et al., 2007).

Consistently, Diao and colleagues reported decreased numbers of Ly-6G+ cells infiltrating the colonic mucosa of Tollip mice upon acute colitis. In addition, Tollip-deficient CCR5+ neutrophils accumulated in the blood due to decreased chemotactic receptor FPR2 levels, indicating defective migration of these cells into inflamed colonic mucosal sites (Diao et al., 2016). CCR5 is an inflammatory chemokine receptor enabling recruitment of diverse immune cell types (Griffith et al., 2014). In a phase I clinical trial of metastatic colorectal cancer, drug-induced blockade of CCR5+CD4+ and CD8+ lymphocyte infiltration with maraviroc reduced interaction of lymphocytes with MΦs, leading to reduced inflammation and better clinical responses (Halama et al., 2016). Ablation of chemokine receptors, such as CCR2 and CXCR2, was shown to severely abrogate leukocyte infiltration, resulting in reduced inflammation-driven colon cancer (Popivanova et al., 2008, Popivanova et al., 2009, Katoh et al., 2013). Notably, we demonstrated a pronounced decrease in Ccr5 expression in the colonic mucosa of unchallenged Tollip mice, indicating a potential correlation between Tollip-Ccr5 expression and modulation of leukocytes trafficking, which deserves further study.

Tollip ablation influenced not only chemokine receptors expression but also lymphocyte homing molecules causing altered T cells migratory patterns. As such, a pronounced Cd62l mRNA decrease in colonic mucosa correlated with increased accumulation of naive CD8+ CD62L+ T cells in the blood of Tollip unchallenged mice. This effect was even more pronounced in Tollip-deficient adenomas, where we observed a prominent reduction in CD3+ T lymphocyte infiltrates. This colon-specific reduction was restricted to colorectal carcinogenesis, as it could not be detected in peripheral lymphoid organs or upon homeostatic conditions. Remarkably, decreased CD3+ T lymphocyte infiltration was mainly due to reduced CD4+ Foxp3+ Treg cells recruitment in the tumor microenvironment. In line with this observation, we observed a significant decrease in Ctla4 mRNA expression in Tollip-deficient adenomas, a receptor expressed on Treg cells. Finally, defects in TGFβR signaling activation further argued for reduced immune-suppressive responses in tumor microenvironment upon Tollip ablation. We suggest that, already during homeostasis, Tollip ablation predisposes leukocytes accumulation in the periphery. In addition, Tollip deficiency also affects leukocyte recruitment, predominantly Treg cells, to adenomatous lesions.

Although Treg cells are key factors of intestinal homeostasis and protect against colitis (Mottet et al., 2003, Tanoue et al., 2016), their role in colitis-induced cancer was recently associated with poor prognosis (Pastille et al., 2014). As such, depletion of Tregs during acute CAC resulted in increased intestinal inflammation, whereas ablation of Tregs at late CAC state promoted CD8+-induced IFNγ/granzyme B antitumoral responses, resulting in hindered tumor progress (Pastille et al., 2014). Hence, our observations support the concept that reduced Treg infiltration upon Tollip ablation into the inflamed colonic mucosa might underlie early CAC events by facilitating epithelial destruction, whereas attenuated cell infiltration, due to Tollip ablation, on late CAC phase seems to play a protective role, resulting in reduced immune suppressive responses and partial protection from CAC onset and development.

The exact molecular mechanisms through which Tollip may alter leukocytes migratory capacity remains unknown. Several studies have shown that Tollip promotes endosomal trafficking of receptors for either lysosomal degradation or nuclear translocation (Brissoni et al., 2006, Zhu et al., 2012, Ciarrocchi et al., 2009). As such, Tollip deficiency resulted in impaired trafficking and accumulation of IL-1R in late endosomes (Brissoni et al., 2006). Therefore, we hypothesize that Tollip deficiency may impair endosomal trafficking of several chemokine receptors and T cell homing factors, such as CCR5 and CD62L.

Altogether, our human and murine data indicate that Tollip plays a dual role in being protective against colitis while exacerbating colitis-induced cancer in mice. Our findings implicate Tollip in the expression of several homing/trafficking molecules of relevance to anti-tumoral immunity, including T lymphocytes and Treg subset infiltration. Thus, shifts in anti-tumoral immunity profile together with defects in TGFβR signaling underlie partial protection from colitis-associated cancer. Future studies should assess the role of Tollip in the turnover of inflammatory and suppressive receptors, as well as in leukocyte migration to inflammatory and cancerous sites.

Limitations of the Study

Our initial findings using human samples were only exploratory due to limited sample size. This was in part due to the rare incidence of colitis-associated cancer in human subjects and also to a limited number of patients screened in the GETAID cohort. In addition, samples obtained from the SIBDCS were only taken in active versus inactively inflamed colon segments prompting us to complete our dataset with samples coming from different sources.

To explore the mechanisms underlying colitis-associated cancer development, we employed the widely accepted and established AOM/DSS mouse model. Although this model is thought to most closely reflect human CAC development, kinetics of adenoma development is more rapid in mice than in humans, putting some limitations in the analyses of cell-recruitment dynamics in tumors. Moreover, AOM/DSS-treated mice mostly develop non-invasive adenomas, whereas in humans the carcinogenic process may evolve to a fully blown invasive adenocarcinoma. As Tollip expression may be different between low-grade and advanced adenomas, more detailed analyses could not be done in the mouse setting.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Consortia

Members of the SIBDCS Study Group

Claudia Anderegg, Peter Bauerfeind, Christoph Beglinger, Stefan Begré, Dominique Belli, José M. Bengoa, Luc Biedermann, Beat Bigler, Janek Binek, Mirjam Blattmann, Stephan Boehm, Jan Borovicka, Christian P. Braegger, Nora Brunner, Patrick Bühr, Bernard Burnand, Emanuel Burri, Sophie Buyse, Matthias Cremer, Dominique H. Criblez, Philippe de Saussure, Lukas Degen, Joakim Delarive, Christopher Doerig, Barbara Dora, Gian Dorta, Mara Egger, Tobias Ehmann, Ali El-Wafa, Matthias Engelmann, Jessica Ezri, Christian Felley, Markus Fliegner, Nicolas Fournier, Montserrat Fraga, Pascal Frei, Remus Frei, Michael Fried, Florian Froehlich, Christian Funk, Raoul Ivano Furlano, Suzanne Gallot-Lavallée, Martin Geyer, Marc Girardin, Delphine Golay, Tanja Grandinetti, Beat Gysi, Horst Haack, Johannes Haarer, Beat Helbling, Peter Hengstler, Denise Herzog, Cyrill Hess, Klaas Heyland, Thomas Hinterleitner, Philippe Hiroz, Claudia Hirschi, Petr Hruz, Rika Iwata, Res Jost, Pascal Juillerat, Céline Keller, Christina Knellwolf, Christoph Knoblauch, Henrik Köhler, Rebekka Koller, Claudia Krieger-Grübel, Gerd Kullak-Ublick, Patrizia Künzler, Markus Landolt, Rupprecht Lange, Frank Serge Lehmann, Andrew Macpherson, Philippe Maerten, Michel H. Maillard, Christine Manser, Michael Manz, Urs Marbet, George Marx, Christoph Matter, Rémy Meier, Martina Mendanova, Pierre Michetti, Benjamin Misselwitz, Bernhard Morell, Patrick Mosler, Christian Mottet, Christoph Müller, Pascal Müller, Beat Müllhaupt, Claudia Münger-Beyeler, Leilla Musso, Andreas Nagy, Michaela Neagu, Cristina Nichita, Jan Niess, Andreas Nydegger, Nicole Obialo, Carl Oneta, Cassandra Oropesa, Ueli Peter, Daniel Peternac, Laetitia Marie Petit, Franziska Piccoli-Gfeller, Julia Beatrice Pilz, Valérie Pittet, Nadia Raschle, Ronald Rentsch, Sophie Restellini, Jean-Pierre Richterich, Sylvia Rihs, Marc Alain Ritz, Jocelyn Roduit, Daniela Rogler, Gerhard Rogler, Jean-Benoît Rossel, Vanessa Rueger, Gaby Saner, Bernhard Sauter, Mikael Sawatzki, Michela Schäppi, Michael Scharl, Sylvie Scharl, Martin Schelling, Susanne Schibli, Hugo Schlauri, Sybille Schmid Uebelhart, Jean-François Schnegg, Alain Schoepfer, Frank Seibold, Mariam Seirafi, Gian-Marco Semadeni, David Semela, Arne Senning, Marc Sidler, Christiane Sokollik, Johannes Spalinger, Holger Spangenberger, Philippe Stadler, Michael Steuerwald, Alex Straumann, Bigna Straumann-Funk, Michael Sulz, Alexandra Suter, Joël Thorens, Sarah Tiedemann, Radu Tutuian, Stephan Vavricka, Francesco Viani, Jürg Vögtlin, Roland Von Känel, Alain Vonlaufen, Dominique Vouillamoz, Rachel Vulliamy, Jürg Wermuth, Helene Werner, Paul Wiesel, Reiner Wiest, Tina Wylie, Jonas Zeitz, Dorothee Zimmermann.