HIC1 Tumor Suppressor Loss Potentiates TLR2/NF-κB Signaling and Promotes Tissue Damage–Associated Tumorigenesis

This article is featured in Highlights of This Issue, p. 1057

The HIC1 gene was isolated as a candidate tumor suppressor gene during tumor DNA hypermethylation screen of the chromosome 17 short arm, a chromosomal region which is frequently reduced to homozygosity in human cancers (1). Hic1^−/− mice die prenatally due to severe developmental defects of craniofacial structures and limbs (2). Hic1^+/− mice are viable; however, they develop spontaneous malignant tumors that are Hic1 deficient due to the intact Hic1 allele methylation (3). Hic1 protein functions as an evolutionarily conserved transcription repressor, which cooperates with several partners to regulate expression of multiple target genes (4). The protein is composed of three structural domains. The Broad complex, Tramtrack, Bric à brac/POx viruses, and Zinc finger (BTB/POZ) domain responsible for Hic1 multimerization is situated N-terminally, followed by the central region binding co-repressors such as C-terminal binding protein (CtBP). The C-terminal domain consists of five zinc fingers providing affinity to the specific Hic1-responsive (HiRE) sequence motif in DNA (5). The known Hic1 target genes participate in diverse cellular processes, including cell-cycle regulation, cell differentiation, DNA damage response, and metastatic invasion (reviewed in ref. 4). Hic1 transcription is positively regulated by p53 protein, a key molecule inducing either cell-cycle arrest or apoptosis upon various cellular stress-inducing insults (6). Conversely, the p53 activity is restrained by a protein deacetylase encoded by the sirtuin 1 (Sirt1) gene, whose expression is blocked by Hic1. Hic1 inactivation thus leads to functional suppression of p53, allowing damaged cells to escape the p53-mediated response (7). Besides the direct regulation of gene expression, Hic1 attenuates transcription via interaction with other transcription factors. For example, association with Wnt pathway effector T-cell factor 4 (TCF4) sequesters TCF4 (and its transcriptional activator β-catenin) to nuclear speckles called Hic1 bodies. Subsequently, expression of the TCF4/β-catenin-responsive genes is inhibited (8).

The tissue maintenance of the single-layer intestinal epithelium is sustained by intestinal stem cells (ISC) that reside at the bottom of invaginations called intestinal crypts, where ISCs divide regularly and give rise to transit amplifying cells (TA). Rapidly dividing TA cells migrate upward and while exiting the crypt, they differentiate to absorptive enterocytes, mucus-producing goblet cells, and hormone-secreting enteroendocrine cells. In the small intestine, differentiated cells cover finger-like protrusions called villi; the surface of the colon is flat. The intestinal epithelium self-renewing in 3 to 5 days represents one of the most rapidly self-renewing tissues in the mammalian body. One exception from the outlined scheme are Paneth cells that secrete bactericidal cryptidins, defensins, or lysozyme. These relatively long-lived cells are present in the small intestine only. Moreover, during maturation, the Paneth cell does not migrate from the crypt but stays at the crypt bottom, where it persists for 6 to 8 weeks (reviewed in ref. 9). Owing to the dynamic turnover, the intestinal epithelium is at high risk of carcinogenesis. Although aberrant activation of the Wnt pathway initiates the majority of colorectal carcinomas (reviewed in ref. 10), CRC development is a multistep process that entails accumulation not only of genetic, but also of epigenetic changes in epithelial cells (11). The physiologic role of HIC1 in the intestine has not yet been elucidated in detail. However, in the mouse, Hic1 represses atonal homolog 1 (Atoh1) and SRY-box containing gene 9 (Sox9) genes, which are involved in the cell fate determination of secretory cell lineages in the small intestine (12–14).

In the present study, we used a conditional knock-out of the Hic1 gene (15) to identify genes repressed by Hic1. Expression profiling of mouse embryonic fibroblasts (MEF) revealed six novel Hic1 target genes, including Tlr2. Tlr2 functions as a microbial sensor to initiate inflammation and immune responses. In addition, the receptor recognizes endogenous inflammatory mediators released from dead cells (review in ref. 16). Upon ligand binding, Tlr2 triggers several signal transduction pathways, including NF-κB signaling that activates expression of proinflammatory cytokines and enzymes, such as tumor necrosis factor-alpha (TNFα), interleukin 6 (IL6), and cyclooxygenase-2 (Cox2) (17). Solid tumors contain inflammatory infiltrates, and many recent studies have shown association between inflammation and increased risk of cancer development and progression. Furthermore, there is growing evidence that Tlr2 activators released from cancer cells might initiate persistent inflammation found in many tumors (reviewed in ref. 18). However, two recent studies documented an inflammation-independent Tlr2 role in promoting gastric and intestinal cancer (17, 19). Here, we show that Hic1 depletion in the intestinal epithelium resulted in increased Tlr2 expression that promoted proliferation of colonic tumors induced by chemical carcinogenesis.

Housing of mice and in vivo experiments were performed in compliance with the European Communities Council Directive of 24 November 1986 (86/609/EEC) and national and institutional guidelines. Animal care and experimental procedures were approved by the Animal Care Committee of the Institute of Molecular Genetics (Ref. 82/2011). Generation and genotyping of Hic1^flox/flox and Hic1 citrine reporter (Hic1^cit/+) mice was described previously (15). The Rosa26-CreERT2 [B6.129-Gt(ROSA)26Sor^{tm1(cre/ERT2)Tyj}/J] mouse strain was purchased from The Jackson Laboratory and was genotyped as recommended by the provider. Villin-CreERT2 and Villin-Cre transgenic mice (20) were kindly provided by Sylvie Robine (Institut Curie, Centre de Recherche, Paris, France). Animals were housed in specific pathogen-free conditions. Tumors of the colon and rectum were collected from adult Hic1^flox/flox Villin-Cre⁺ mice 5 weeks after a single subcutaneous injection of azoxymethane [(AOM); 10 mg/kg; purchased from Sigma] that was followed by a 5-day dextran sodium sulfate (DSS) treatment in drinking water [2% (w/v) DSS; MW 36–50 kDa; MP Biomedicals]. The mice were euthanized and the intestines were dissected, washed in PBS, and fixed in 4% formaldehyde (v/v) in PBS for 3 days. Fixed intestines were embedded in paraffin, sectioned and stained. The number and size of the neoplastic lesions were quantified using Ellipse software (ViDiTo). Colitis was induced by DSS (2% in the drinking water for 5 days) without AOM treatment. Colons were collected 2, 6, and 9 days upon DSS withdrawal.

MEFs were isolated from embryos obtained at embryonic day (E) 11 to E14, details of the procedure are given in Supplement. For Cre-mediated recombination, cells were cultured in the presence of 4-OHT at a final concentration of 2 μmol/L (prepared from 1 mmol/L solution in ethanol; Sigma). Control cells were treated with the corresponding amount of ethanol. Small intestinal crypts were isolated and cultured according to the previously published protocol (21). Colonic crypts were isolated using the same procedure, but the culture medium was additionally supplemented with conditioned medium obtained from mouse Wnt3a-producing L cells (L-Wnt3a; ref. 22). L-Wnt3a, HEK293, and BJ-Tert cells were purchased from the ATCC (Cat. No.: CRL-2647, CRL-1573, and CRL-4001, respectively). All cell lines were obtained in 2006 and maintained in DMEM (Sigma) supplemented with 10% FBS (Gibco), penicillin, streptomycin, and gentamicin (Invitrogen). Upon receipt, cells were expanded and aliquots of cells at passage number <10 were stored frozen in liquid nitrogen. Cells from one aliquot were kept in culture for less than 2 months after resuscitation. The cell identity was not authenticated by the authors.

Total RNA was isolated from MEFs harvested 24, 48, 72, and 120 hours upon 4-OHT addition using RNeasy Plus Mini Kit (Qiagen). Control cells were grown with the same volume of vehicle (ethanol). The quality of isolated mRNA was checked using Agilent Bioanalyzer 2100; RNAs with RNA integrity number (RIN) above 8 were further processed. Two biologic replicates were used for each time point and treatment. The RNA samples were analyzed using MouseRef-8 v2.0 Expression BeadChip (Illumina). Raw data were processed using the beadarray package of Bioconductor and analyzed as described previously (23). Gene set enrichment analysis (GSEA) was performed using the Enricher gene analysis tool (http://amp.pharm.mssm.edu/Enrichr; ref. 24). Microarray data were deposited in ArrayExpress (http://www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-3486.

MEFs were electroporated (details are given in Supplement); transfection of HEK293 cells was performed using Lipofectamine 2000 reagent (Invitrogen). Luciferase reporter constructs NF-κB-Luc and pRL-TK were purchased from Promega. To generate Tlr2-Luc reporter plasmid, genomic DNA containing the mouse Tlr2 promoter region encompassing nucleotides −796 to +52 (the transcription start site corresponds to position +1) was amplified by PCR and cloned into the pGL4.26 luciferase reporter vector (Promega). The HIC1 construct was described previously (8). Details of the luciferase assay and NF-κB pathway stimulation are given in Supplement. For RNAi, BJ-Tert fibroblasts were reverse-transfected with Lipofectamine RNAiMax (Invitrogen) according to manufacturer's instructions using 10 nmol/L small interfering RNA (siRNA) targeting HIC1 (HIC1 siGENOME SMART Pool M-006532-01, Dharmacon) or a scrambled control siRNA (siCtrl; siGENOME RISC free control siRNA, Dharmacon) and harvested 2 days upon transfection.

Chromatin immunoprecipitation (ChIP) using chromatin obtained from immortalized BJ-Tert fibroblasts was performed as described previously (25). Occupancy of gene regulatory regions by HIC1 was assayed by ddPCR (QX200, Bio-Rad) using EvaGreen master mix (Bio-Rad). PCR primers are listed in Supplement.

Paneth cell sorting was performed according to the previously published protocol (26). Antibodies used for flow cytometry: phycoerythrin (PE)-conjugated anti-CD24 (12-0242-81, eBioscience), allophycocyanin (APC)-conjugated anti-EpCam (17-5791-80, eBioscience), FITC-conjugated anti-CD45 (ED7018, ExBio).

The technique was performed as described previously (27, 28). Hematoxylin and eosin (Sigma) were used for counterstaining. Antibodies are given in Supplement. For visualization of citrine fluorescence, intestines dissected from Hic1^cit/+ and wild-type (wt) mice were snap frozen in liquid nitrogen and immediately sectioned. Specimens were stained with 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI; Life Technologies) and the citrine fluorescence signal was evaluated using a laser scanning confocal microscope (Leica TCS SP5).

Hic1-specific polyclonal antisera were generated in rabbit or chicken immunized with the recombinant His-tagged fragment of human HIC1 protein (amino acids 154-396). Commercially available antibodies are given in Supplement.

Total RNAs were isolated from cells and tissues using RNeasy Mini Kit (Qiagen) and reversely transcribed and analyzed by qRT-PCR as described previously (29). The primers are listed in Supplementary Table S1.

Results of the gene reporter assay, Ellipse, and qRT-PCR analysis were evaluated by the Student t test. Datasets obtained using DNA microarrays were analyzed in the R environment using the package LIMMA (Linear Models for Microarray Data LIMMA; ref. 30).

To investigate the biologic function of transcription repressor Hic1, we have recently developed the Hic1^flox/flox mouse strain that enables conditional inactivation of the Hic1 gene (15). These mice were crossed to Rosa26-CreERT2 animals expressing Cre recombinase estrogen receptor T2 fusion protein (CreERT2) from the Rosa26 locus (31). The fusion protein resides in the cytoplasm until an antagonist of the estrogen receptor, tamoxifen (or its active metabolite 4-OHT), is administered. Subsequently, Cre enzyme is translocated into the nucleus, where it allows excision of DNA sequences flanked, that is, “floxed”, by loxP sites (32). MEFs were isolated from embryos, cultured with 4-OHT or vehicle (ethanol) and genotyped by PCR to confirm locus recombination and generation of the Hic1^delEx2 allele. Simultaneously, the presence of Hic1 protein was detected by immunoblotting (Fig. 1A). Subsequently, total RNA was isolated at four time points after the addition of 4-OHT (or vehicle) into the culture media, and microarray analysis of the expression profile was performed. The analysis revealed genes whose expression was changed significantly (q < 0.05) after the Hic1 locus recombination. Furthermore, six genes were upregulated in at least two time points and more than twice [the binary logarithm of fold change (log FC) > 1; Fig. 1B and Supplementary Fig. S1 and Supplementary Table S2]. These genes were angiopoietin-like 7 (Angptl7), carbonyl reductase 2 (Cbr2), death-associated protein kinase 2 (Dapk2), HSP, α-crystallin-related, B6 (Hspb6), Tlr2, and WAP four-disulfide core domain 2 (Wfdc2). In addition, increased expression of previously identified HIC1 target genes cyclin-dependent kinase inhibitor 1a (CDKN1A; ref. 33), Sox9 (13), and Sirt1 (7) was observed, but these genes did not satisfy the significance criteria (not shown). Rather unexpectedly, the GSEA analysis using the Enricher gene library online tool (24) revealed that the inactivation of the Hic1 gene in MEFs mainly altered expression of genes involved in lipid metabolism (Supplementary Table S3).

As recent evidence supports the role of TLR signaling in inflammation-associated tumorigenesis in the gastrointestinal tract (reviewed in ref. 34), we further investigated the relationship between Hic1 and Tlr2 expression. First, we generated a Tlr2-Luc reporter by subcloning genomic DNA harboring the Tlr2 promoter region before the luciferase sequence. Then, we cotransfected the Tlr2-Luc reporter together with the expression construct encoding HIC1 into HEK293 cells and performed the luciferase reporter assay. As expected, increasing amounts of HIC1 resulted in reduced activity of the reporter, whereas cotransfection with a control “empty” plasmid had no significant effect (Fig. 1C). Moreover, we used a HIC1-specific antibody to perform ChIP using chromatin isolated from BJ fibroblasts that express detectable amounts of endogenous HIC1 (25). Droplet digital PCR confirmed the enrichment of genomic DNA encompassing the promoter regions of TLR2 and SIRT1 (used as positive control) genes in the precipitate. Importantly, none of these regions was precipitated using a control “irrelevant” antibody and, furthermore, the anti-HIC1 antibody did not pull down the β-ACTIN promoter used as negative control. In addition, upregulation of TLR2 and SIRT1 was detected upon HIC1 mRNA knockdown (Fig. 1D and data not shown). The Tlr2 gene is activated by NF-κB signaling (35); therefore, the luciferase assay was performed in MEFs stimulated with NF-κB pathway activators TNFα, LTα, and with synthetic bacterial lipopeptide Pam3csk4, which functions as the Tlr2 ligand (36). Interestingly, Pam3csk4 appeared to be the most potent enhancer of the Tlr2-Luc activity, especially upon Hic1 deletion; however, the Tlr2-Luc activity was also significantly increased upon stimulation with unrelated LTα. In addition, the luciferase reporter assay clearly showed increased Tlr2-Luc activity in Hic1-deficient MEFs when compared with cells with the intact Hic1 locus (Fig. 1E). Several studies indicate the NF-κB pathway activation via Tlr2 signaling (37, 38). Thus, we next examined the consequence of Hic1 deficiency for the NF-κB pathway. Luciferase reporter assay using the NF-κB-Luc reporter showed increased NF-κB signaling even in unstimulated Hic1^{delEx2/delEx2} cells (compared with MEFs before Cre-mediated Hic1 inactivation). Treatment of cells with Pam3csk4 further potentiated the reporter activity. However, increased NF-κB signaling was also observed upon stimulation with LTα and TNFα, indicating that the NF-κB pathway output was not completely dependent on Tlr2 (over)expression (Fig. 2A). The augmentation of the NF-κB pathway activity in Hic1-deficient MEFs was confirmed by qRT-PCR analysis that showed increased expression of the putative NF-κB target genes Cox2 and TNFα (Fig. 2B). In agreement with these results, immunoblotting revealed increased levels of the phosphorylated transcriptionally active form of the nuclear mediator of NF-κB signaling p65 (Fig. 2C). In summary, all the data supported direct transcriptional repression of Tlr2 by Hic1 and, moreover, potentiation of the NF-κB pathway activity upon loss of the Hic1 gene.

In the intestinal epithelium, the Hic1 expression was examined using previously generated Hic1-citrine “reporter” mice. In this mouse strain, the sequence encoding the central and C-terminal portions of Hic1 protein was replaced by cDNA encoding citrine (yellow) fluorescent protein (15). In both the small and large intestine, native citrine fluorescence was observed in epithelial cells (Fig. 3A). Subsequently, production of Hic1 protein in the gut epithelia was confirmed by immunoblotting using cell lysates prepared from crypts and differentiated cells lining the villi in the small intestine or intercrypt regions (ICR) in the colon. The analysis revealed higher amounts of Hic1 in the villus and ICR fractions when compared with the crypts. This result was confirmed by qRT-PCR using total RNA isolated from the same epithelial fractions (Fig. 3B). In order to evaluate the expressional changes of the putative Hic1 target genes in epithelial cells, we established three-dimensional “organoid” cultures (21) from the small intestinal crypts of Hic1^flox/flox Villin-CreERT2⁺ mice producing tamofixen-regulated Cre enzyme in all intestinal cell lineages (20). Interestingly, Cre-mediated inactivation of the Hic1 gene increased mRNA levels of all genes identified by the microarray analysis. Furthermore, like in MEFs, Hic1 deficiency resulted in upregulation of the NF-κB-responsive genes Cox2 and TNFα (Fig. 3C). In addition, Tlr2 was analyzed in Hic1^flox/flox Villin-Cre⁺ mice expressing the constitutively active form of Cre enzyme in the embryonic and adult gut epithelia (20). Although efficient recombination and generation of the Hic1^delEx2 allele along the rostro-caudal axis of the gastrointestinal tract was confirmed by PCR genotyping (data not shown), these mice—lacking Hic1 in the intestinal epithelium—were viable, showing no signs of any health problems. However, IHC staining showed increased Tlr2 positivity in the small intestine and colon (Fig. 3D).

As morphology and proliferation of the Hic1^loxP/loxPVillin-Cre⁺ intestinal epithelium appeared to be normal, we performed detailed analysis of all major cellular lineages present in the small intestine. Maturation of Paneth cells is driven by transcription factors Atoh1 and Sox9 (14). Interestingly, Hic1 is a transcriptional repressor of Atoh1 and SOX9 in the mouse developing cerebellum and in U2OS osteosarcoma cells, respectively (12, 13). Using expression profiling of sorted Paneth cells isolated from the Hic1^flox/flox Villin-Cre⁺ and control (Hic1^flox/flox) small intestinal crypts, we observed a significant increase in Atoh1, Sox9, and Tlr2 mRNA levels upon loss of Hic1. Moreover, FACS analysis showed a slight increase in Paneth cell counts in Hic1^flox/flox Villin-Cre⁺ animals (Fig. 4A). Increased Paneth cell numbers along the rostro-caudal axis of the small intestine were also recorded using paraffin sections stained with an antibody recognizing Paneth cell-specific marker lysozyme (Fig. 4B). In addition, the numbers of mucin-producing goblet cells and chromogranin A-positive enteroendocrine cells were also elevated. In contrast, the differentiation and number of absorptive enterocytes seemed unchanged in the Hic1-deficient intestine (Supplementary Fig. S2).

Recently, Mohammad and colleagues described accelerated formation of tumors upon loss of Hic1 in the Apc^Min/+ mouse model of intestinal cancer (39). Because inflammation is an important tumor promoter in colorectal neoplasia, the role of Hic1 was examined utilizing colitis-associated tumorigenesis. In the acute colitis phase – induced by DSS treatment – no differences in the extent of tissue damage were observed in histologic specimens obtained from Hic1^flox/flox Villin-Cre⁺ and Hic1^flox/flox mice. However, Hic1^flox/flox Villin-Cre⁺ individuals displayed more robust DSS treatment-associated transcriptional response of the Cox2, Tlr2, and TNFα genes, that is, genes upregulated upon Hic1 loss. Analysis of the colon during the regenerative phase, that is, 6 and 9 days upon DSS withdrawal, showed a more robust hyperproliferative response of the colonic epithelium of Hic1^flox/flox Villin-Cre⁺ individuals when compared with their Hic1^flox/flox littermates (Fig. 5A). Moreover, a continuous increase of Cox2, Tlr2, and TNFα expression was observed at these time points (Supplementary Fig. S3). Strikingly, upon combined AOM/DSS treatment, Hic1 deficiency significantly increased the size of colonic and rectal tumors (Fig. 5B; average tumor size in Hic1^flox/flox mouse = 3.1 mm² vs. 8.7 mm² in Hic1^flox/flox Villin-Cre⁺; P = 0.00862). IHC examination revealed decreased numbers of p53-positive cells within the tumor mass of Hic1-negative mice. Nevertheless, even in Hic1^flox/flox mice, cells displaying nuclear p53 staining were rare and scattered throughout the neoplastic tissue. However, expression of the p21 cell-cycle inhibitor, the most prominent in wt cells at the surface area of the tumors, was reduced in Hic1 knock-outs. Strikingly, proliferating cell nuclear antigen (PCNA) expressing cells were localized in wt mice in areas clearly distinguished from the regions containing p21-positive cells. In contrast, in Hic1-deficient mice, proliferating cells were more abundant and dispersed throughout the neoplastic tissue. Finally, no differences were noted in the numbers of apoptotic cells (Fig. 5C and not shown).

In the present study, we used a gene inactivation-based screen to identify genes regulated by the Hic1 tumor suppressor. Interestingly, none of the identified presumptive Hic1 target genes was described previously. As the expression of all tested genes reacted to the Hic1 presence not only in MEFs, but also in intestinal organoids, we can exclude the possibility that the genes represent a small (sub)set of tissue-specific Hic1 targets. More likely, because the majority of previous studies utilized cells (over)producing ectopic HIC1 or cells treated with HIC1-specific siRNA (33), we suggest that (our) genes represent Hic1 target genes whose expression is relieved when the Hic1 steady-state levels are diminishing. On the other hand, the previously identified targets might represent genes efficiently repressed (or activated) immediately upon perturbation of the Hic1 protein levels. We used Hic1-reporter mice, Western blotting, and FACS analysis to demonstrate that in the adult intestine, Hic1 is mainly expressed in the differentiated epithelial cells. The expression of its target gene, Tlr2, was documented in ISCs (19). However, in the Hic1-deficient gut, Tlr2 was upregulated in differentiated cells, confirming the inverse Hic1-Tlr2 relationship. Moreover, intestinal ablation of Hic1 resulted in a moderate increase in Paneth, goblet, and enteroendocrine cell numbers. Such phenotype might be attributed to upregulation of Atoh1, the Hic1 target gene functioning as the master regulator of secretory cell lineages in the small intestine (14). Upon ligand binding, Tlr2 triggers several signal transduction pathways, including NF-κB signaling (17). Interestingly, qRT-PCR analysis and luciferase reporter assays revealed that not only ligand-induced, but also basal levels of the NF-κB signaling are elevated in Hic1-deficient cells (Fig. 2A and B). As the NF-κB pathway activity was more pronounced even in cells stimulated with other NF-κB inducers, the observed NF-κB (hyper)activity cannot be solely attributed to the increase in Tlr2 expression or function. Indeed, Western blotting confirmed that Hic1^{delEx2/delEx2} cells displayed higher amounts of phosphorylated nuclear NF-κB mediator p65 (Fig. 2C). The exact reason(s) why Hic1 deficiency potentiates NF-κB signaling remains to be determined. Nevertheless, since several studies reported interaction between the STAT3 and NF-κB pathways (reviewed in ref. 40), we suggest that the autocrine IL6-STAT3 signaling might also be linked to the observed “boosts” in cellular reactivity to the stimuli activating the NF-κB pathway and to the increase in p65 stability and phosphorylation.

Several studies have demonstrated that Tlr2 expression is increased during intestinal inflammation (41) or in human patients with ulcerative colitis (42). In addition, Maeda and colleagues showed that conditioned medium obtained from cultured colon cancer cells might activate NF-κB signaling via Tlr2 (43). These data are in accordance with our observation that in damaged tissue, transcription of Tlr2 and genes linked to active NF-κB signaling is elevated. Importantly, in the Hic1-deficient colon, the damage-induced transcriptional response was more robust than in wt mice, supporting the Hic1 repressive role in Tlr2 expression and NF-κB activation (Fig. 5A and Supplementary Fig. S3). Interestingly, several recent studies reported decreased tumor burden in Apc^+/Min mice deficient in Tlr2 or its intracellular adaptor myeloid differentiation primary response protein-88 (Myd88; refs. 19, 44). Furthermore, deletion of Tlr2 (or Myd88) reduced development of colonic tumors in DSS-treated animals. Moreover, functional blocking of TLR2 inhibited in vitro growth of cancer cells (19). In another study, epithelial expression of Tlr2 and Tlr2/NF-κB signaling promoted growth and survival of gastric cancer cells (17). In summary, all these results support the cell-autonomous and inflammation-independent role of Tlr2 in cancer.

HIC1 participates in the cellular network regulating the DNA damage response (7). HIC1-mediated repression of deacetylase SIRT1 gene promotes p53 stability and potentiates transcription of p53-dependent genes such as p21 (45) and, in a positive regulatory feedback loop, also HIC1 itself (1, 46). Unexpectedly, only a small fraction of cells in tumors generated by AOM/DSS treatment displayed stabilized p53 (Fig. 5C). Although the portion of p53-positive cells was decreased in Hic1^−/− neoplastic tissues, it is unlikely that the increased tumor size observed upon Hic1 deletion is linked to the attenuated p53-mediated response. In human cells, HIC1 directly represses transcription of the CDKN1A gene encoding p21 (33). Strikingly, in Hic1-deficient tumors, we observed the opposite effect of Hic1 absence, that is, reduced p21 staining (Fig. 5C). Nevertheless, the observed discrepancy of the results might be explained by the fact that next to p53 and HIC1, p21 levels might be controlled by many other stimuli and also by post-transcriptional mechanisms (reviewed in ref. 47). As we noted increased cell proliferation upon DSS-induced epithelial damage and in Hic1-negative AOM/DSS tumors, we suggest that besides compromising the p53-mediated tumor-suppressive mechanisms the loss of Hic1 might potentiates tumor-promoting proproliferative Tlr2/NF-κB signaling.

No potential conflicts of interest were disclosed.

Conception and design: L. Janeckova, V. Pospichalova, V. Korinek

Development of methodology: V. Pospichalova, M. Dubuissez, D. Leprince, A. Hlavata, V. Korinek

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Janeckova, V. Pospichalova, M. Vojtechova, J. Tureckova, J. Dobes, M. Dubuissez, D. Leprince, N. Baloghova, M. Horazna, A. Hlavata, J. Stancikova, E. Sloncova

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Janeckova, V. Pospichalova, A. Hlavata, H. Strnad, V. Korinek

Writing, review, and/or revision of the manuscript: L. Janeckova, V. Pospichalova, D. Leprince, V. Korinek

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Galuskova

Study supervision: V. Korinek

The authors thank S. Robine for Villin-Cre and Villin-CreERT2 mice, and S. Takacova for critically reading the manuscript.

This work was supported by Grant Agency of the Czech Republic Grant No. P305/12/2347, institutional Grant No. RVO 68378050, and project BIOCEV – Biotechnology and Biomedicine Centre of the Academy of Sciences and Charles University (CZ.1.05/1.1.00/02.0109) from the European Regional Development Fund.

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.