Research ArticleCancer

β-Catenin Promotes Colitis and Colon Cancer Through Imprinting of Proinflammatory Properties in T Cells

See allHide authors and affiliations

Science Translational Medicine  26 Feb 2014:
Vol. 6, Issue 225, pp. 225ra28
DOI: 10.1126/scitranslmed.3007607

Abstract

The density and type of lymphocytes that infiltrate colon tumors are predictive of the clinical outcome of colon cancer. High densities of T helper 17 (TH17) cells and inflammation predict poor outcome, whereas infiltration by T regulatory cells (Tregs) that naturally suppress inflammation is associated with longer patient survival. However, the role of Tregs in cancer remains controversial. We recently reported that Tregs in colon cancer patients can become proinflammatory and tumor-promoting. These properties were directly linked with their expression of RORγt (retinoic acid–related orphan receptor-γt), the signature transcription factor of TH17 cells. We report that Wnt/β-catenin signaling in T cells promotes expression of RORγt. Expression of β-catenin was elevated in T cells, including Tregs, of patients with colon cancer. Genetically engineered activation of β-catenin in mouse T cells resulted in enhanced chromatin accessibility in the proximity of T cell factor-1 (Tcf-1) binding sites genome-wide, induced expression of TH17 signature genes including RORγt, and promoted TH17-mediated inflammation. Strikingly, the mice had inflammation of small intestine and colon and developed lesions indistinguishable from colitis-induced cancer. Activation of β-catenin only in Tregs was sufficient to produce inflammation and initiate cancer. On the basis of these findings, we conclude that activation of Wnt/β-catenin signaling in effector T cells and/or Tregs is causatively linked with the imprinting of proinflammatory properties and the promotion of colon cancer.

INTRODUCTION

The gastrointestinal tract is poised in a state of equilibrium that permits rapid protective responses against pathogens but curtails damage by hindering long-lasting vigorous inflammatory processes. This balance is achieved through interactions between proinflammatory T helper 17 (TH17) cells and anti-inflammatory regulatory T cells (Tregs) (1) that suppress TH17 inflammation in an interleukin-10 (IL-10)–dependent manner (25). Autoimmune disorders, in particular human inflammatory bowel disease (IBD), are etiologically associated with chronically deregulated inflammation (6, 7). Both the progression of IBD to cancer (8) and the initiation and progression of sporadic colon cancer are driven by inflammation (912). Accordingly, infiltration of colon cancer tumors with TH17 cells negatively correlates with patient survival (13), whereas high densities of Tregs predict better clinical outcomes (1315). The protective role of Tregs in colon cancer is, however, controversial, and other reports suggest a negative correlation of high Treg densities with disease outcome (16). We reported earlier that in human colon cancer, there is preferential expansion of a Treg subset that is potently T cell–suppressive but has TH17 characteristics (11, 12, 1719). These Tregs express the signature TH17 transcription factor retinoic acid–related orphan receptor γt (RORγt) and promote inflammation and tumor growth (11, 12, 18). Expression of RORγt by T cells, including Tregs, is pivotal for sustaining pathologic inflammation in mouse polyposis, and genetic ablation of RORγt in these cells protects against polyposis (12, 17). It is unclear what triggers up-regulation of RORγt in T cells in the course of polyposis and colon cancer. Elucidating the molecular mechanisms that shift the lymphocyte balance from anti-inflammatory to proinflammatory is expected to improve diagnosis and treatment of IBD and colon cancer.

Inactivation of the adenomatous polyposis coli (APC) gene is the initiating event in about 80% of human colon cancer cases (20), inducing the development of aberrant crypt foci and polyps (21, 22). Polyp growth is directly linked with stabilization of β-catenin (21, 22), the central effector of the Wnt signaling pathway. Focal inflammatory reactions in response to the oncogenic event (22) and to the gut microbiota (23) also contribute to disease progression. In thymocytes, β-catenin is activated by T cell receptor (TCR) signaling, and together with its T cell–specific DNA binding partner T cell factor-1 (Tcf-1), β-catenin promotes thymic development and selection (2429). The transgenic overexpression of β-catenin in thymocytes promotes expression of RORγt, which, in turn, controls the expression of prosurvival genes (30). Accordingly, enhanced β-catenin activity is suggested to promote survival of ex vivo–generated mouse Tregs (31). By contrast, more recent findings suggest that pharmacologic activation of Wnt signaling suppresses Foxp3 and compromises the function of ex vivo–differentiated human Tregs (32). Furthermore, the ex vivo differentiation of TH17 cells coincides with up-regulation of β-catenin and Wnt signaling genes (33), and ablation of Tcf-1 promotes expression of IL-17 by T cells (34, 35). These findings are consistent with the notion that Wnt/β-catenin signaling promotes TH17 differentiation.

Here, we evaluated the role of Wnt/β-catenin in dictating T cell functions in colitis and colon cancer and the pathogenic consequences thereof. We found that the expression of RORγt and gain of proinflammatory functions by T cells, including Tregs, in colitis and colon cancer are regulated through β-catenin–mediated epigenetic reprogramming. Through the combined use of mouse models and patient specimens, we demonstrate the relevance of these findings to IBD and to colon cancer in humans. These findings provide a mechanism for the chronic shift in lymphocyte properties from anti-inflammatory to proinflammatory and highlight the critical role of Wnt signaling within T cells in the epigenetic imprinting of inflammation in autoimmunity and cancer.

RESULTS

In human colitis and colon cancer, T cells express elevated levels of β-catenin

Earlier, we provided evidence that β-catenin is activated downstream of the TCR (25). Both IBD and colon cancer involve activation of T cells; therefore, we investigated whether the Wnt/β-catenin pathway was up-regulated in T cells in these diseases. We examined tissue and tumor sections from patients with long-standing ulcerative colitis (UC), UC-associated colon cancer, and sporadic colon cancer. Colitis and tumor samples were compared to control colon tissue from patients with arteriovenous malformation (AVM) or diverticular disease. The control specimens were selected on the basis of their noninflamed and noncancerous characteristics. Colonic tissues were sectioned and stained with specific antibodies to CD3 (T cells) and β-catenin, and expression of CD3 and β-catenin in the same cell was confirmed by immunofluorescence confocal microscopy.

We found that tumor tissues were enriched for T cells exhibiting strong membrane as well as cytoplasmic coexpression of β-catenin and CD3 (Fig. 1A and fig. S1). We quantified the frequencies of β-catenin–expressing T cells in UC-associated colon cancer specimens (Fig. 1A, a and b), UC (Fig. 1A, c and d), control uninflamed noncancerous colons (Fig. 1A, e and f), sporadic colon cancer tumors (Fig. 1A, g and h), and healthy margin of sporadic tumor cancers (Fig. 1A, i and j). These analyses established that significantly more T cells infiltrated tumors (P < 0.0001) as compared to control and colitis tissue (P = 0.0062) (Fig. 1B, left panel, and table S1). A significantly increased fraction of infiltrating T cells in both colitis (P < 0.0001) and tumor tissue (P < 0.0001) expressed β-catenin compared to T cells in control colon or in the margins of tumors obtained from sporadic colon cancer patients (P = 0.0007) (Fig. 1B, right panel, and table S1). To further relate these changes within the tumor to systemic immunity, we determined the levels of β-catenin in lysates of purified CD4+ T cells and CD4+CD25+ Tregs from peripheral blood of colon cancer patients. Western blot analysis showed that T cells from cancer patients, including both Tregs (P = 0.0023) and non-Treg CD4+ T cells (P = 0.0004), had significantly higher levels of β-catenin compared to healthy donors (Fig. 1C and table S1). These findings suggest that T cells up-regulate expression of β-catenin in UC and colon cancer.

Fig. 1. β-Catenin is up-regulated in T cells and Tregs of colon cancer patients.

(A) (Upper) Representative confocal images of CD3 (green), β-catenin (red), and DAPI (blue), or isotype controls rabbit immunoglobulin G (IgG) (green), mouse IgG1 (red), and DAPI (blue) immunofluorescence. Scale bars, 50 μm. CD3, β-catenin, and DAPI immunostaining is shown for UC-associated colon cancer (a), UC (c), control uninflamed and nonmalignant lesions from patients with AVM, or diverticular disease (“healthy”) (e), sporadic colon cancer (g), and healthy margin of sporadic colon cancer (i). Immunostaining of the respective sections with isotype controls is shown in (b), (d), (f), (h), and (j). Arrowheads show CD3+ β-catenin+ T cells (also shown in inset images, which are three times enlarged), and arrows depict CD3+ β-catenin T cells. Independent fluorescence channels are shown in fig. S1. (B) Quantification of colon infiltration by CD3+ and β-catenin+CD3+ T cells in stained sections as in (A). Quantified were control (healthy) AVM and diverticular disease patients (n = 7), UC patients (n = 5), UC colon cancer tumors (n = 7), healthy margin of sporadic colon cancer (n = 3), and sporadic colon cancer tumors (n = 4). Four to six independent fields per patient were quantified. Two-tailed t test statistics: CD3+ healthy/UC colon cancer (***P < 0.0001), UC/UC colon cancer (**P = 0.0062). %β-Catenin+ among CD3+ healthy/UC (***P < 0.0001), healthy/UC colon cancer (***P < 0.0001), healthy margin/colon cancer (***P = 0.0007). (C) Western blot analysis of purified CD4+Foxp3 and CD4+Foxp3+ T cells from colon cancer patients (n = 10 and 7, respectively) and healthy donors (n = 7 and 6, respectively) normalized for β-tubulin. Western blots for representative patients are shown on the left, and graphs of all patients’ cells are shown on the right. Each dot represents one patient. Two-tailed Mann-Whitney test: ***P = 0.0004 (CD4+Foxp3), **P = 0.0023 (CD4+Foxp3+). Mean ± SEM is shown.

TH17 and Wnt/β-catenin signature genes are up-regulated in intestinal T cells during polyposis

We investigated whether β-catenin activity in T cells contributes to inflammation and cancer, using APC+/Δ468 mice, which have a heterozygous deletion of the APC gene and develop hereditary polyposis (22, 36, 37). Polyp-ridden mice at 3 months of age had increased frequencies of activated T cells in both spleen (50% activated versus 10% in healthy animals) and intestine (80% activated versus 50% in healthy animals) (Fig. 2, A and B, and table S1). To measure expression of β-catenin, we sorted CD4+ T cells and CD4+Foxp3+ Tregs (>97% purity) from Foxp3-GFP (green fluorescent protein) reporter mice on the wild-type or APC+/Δ468 background (fig. S2A). Lysates of the purified cells were analyzed by Western blot. Gut-infiltrating CD4+CD25 T cells and CD4+CD25+ Tregs had elevated levels of β-catenin as compared to spleen and mesenteric lymph nodes, and these levels increased during polyposis (Fig. 2, C and D, table S1, and fig. S2B). We established that β-catenin was stabilized in T cells during polyposis in response to cell-extrinsic stimuli and not because of the mutated APC allele, by targeting the APC ablation to epithelial cells. To this end, we used conditional APClox468 mice (37, 38) crossed to Ts4Cre transgenic mice (39) that express Cre specifically in gut epithelial cells. Western blot analyses showed that CD4+ T cells isolated from aged polyp-ridden Ts4CreAPC+/lox468 mice also expressed elevated levels of β-catenin, although these cells have wild-type APC alleles (fig. S3).

Fig. 2. CD4 T cells up-regulate activation markers and β-catenin during polyposis.

(A) Histogram overlays show surface expression of CD69 by wild-type (WT) (gray) and APC+/Δ468 (clear) CD4+ T cells from mouse spleen and small intestine. (B) Frequency of CD69+CD4+ T cells in the indicated tissues and mice (n = 4). *P < 0.05, two-tailed Mann-Whitney test. Data are means ± SEM. (C) Representative β-catenin Western blot of lysates of sorted CD4+ T cells from WT and APC+/Δ468 mice (M. lymph node, mesenteric lymph nodes). β-Tubulin was probed as loading control. (D) Quantification of β-catenin protein revealed by Western blot of intestinal CD4+ cells as in (C). *P = 0.0401, two-tailed Student’s t test. Data are means ± SEM. (E) GSEA of Wnt pathway genes (Biocarta) in APC+/Δ468 compared to WT CD4+ T cells from small intestine [normalized enrichment score (NES) = 2.2692885, nominal P = 0.0, false discovery rate (FDR) q = 0.0]. (F) Heatmap shows expression of Wnt pathway genes that are up-regulated in APC+/Δ468 compared to WT small intestine. CD4+ T cells. (G) Heatmap of Affymetrix array data depicts the expression of TH17 signature genes in WT and APC+/Δ468 CD4+ T cells. Each rectangle represents the average of three to five values. Gene IDs and fold change in expression in APC+/Δ468 versus WT CD4+ T cells are shown. More than 97% sample purity for Western blots and microarrays (see fig. S2A).

To determine how the tumor microenvironment affected gene expression, mRNA was prepared from CD4+Foxp3GFP T cells and CD4+Foxp3GFP+ Tregs sorted to ~97% purity and interrogated by ImmGen using Affymetrix arrays (12). Expression of Wnt pathway genes was compared between polyp-ridden APC+/Δ468 and wild-type mice by Gene Set Enrichment Analysis (GSEA; Massachusetts Institute of Technology) using a Wnt pathway gene set (KEGG_WNT_SIGNALING_PATHWAY). This analysis revealed significant (P < 0.001) enrichment in the expression of Wnt pathway genes in CD4+ T cells infiltrating the intestine of APC+/Δ468 mice (Fig. 2, E and F). A weaker but significant (P = 0.02) enrichment of Wnt pathway genes was detected in Tregs infiltrating the intestinal tumors (fig. S2C). Multiple genes associated with the TH17 lineage, including IL-17 and RORγt, were also up-regulated in T cells infiltrating the wild-type intestine. Expression of most of these genes increased further during polyposis (Fig. 2G). These findings are in line with an earlier report that Wnt pathway genes are up-regulated during ex vivo TH17 commitment (33) and with our earlier findings that RORγt+ T cells are more frequent in the intestine of polyp-ridden APC+/Δ468 (12) and Ts4CreAPC+/lox468 mice (37). Together, these observations connect Wnt/β-catenin signaling with the gain of TH17 characteristics by T cells and Tregs during polyposis.

Activation of β-catenin in T cells predisposes mice to intestinal inflammation, colitis, and cancer

To understand the biological outcome of β-catenin expression in T cells, we activated β-catenin specifically in T cells using compound mutant CD4Cre (40) Ctnnb1ex3 (21) mice or in Tregs using Foxp3Cre (5) Ctnnb1ex3 (21) mice. The Cre-mediated excision of β-catenin exon 3 removes phosphorylation sites that target the protein for degradation, thereby producing stable, dominant, constitutively active β-catenin. Both mouse models developed intestinal and colonic inflammation and polyps but with different kinetics and severity. The CD4CreCtnnb1ex3 compound mutant progeny developed cachexia and rectal prolapse as early as 8 to 10 weeks of age. The Foxp3CreCtnnb1ex3 progeny were healthy until 4 months but then also developed inflammation and polyps. In both instances, the penetrance was 100% with all mice developing disease.

Histologic analysis revealed crypt elongation in the colon and crypt and villus elongation in the small intestine of CD4CreCtnnb1ex3 mice compared to littermate CD4Cre controls (Fig. 3A). Accordingly, starting at 6 weeks of age, epithelial cells in both the small intestine and colon of CD4CreCtnnb1ex3 mice showed increased mitotic activity (Fig. 3, B and C, and table S1). By 8 to 10 weeks of age, CD4CreCtnnb1ex3 mice began to develop active small intestine inflammation and chronic colitis. We observed progressive leukocyte infiltration that promoted ulcers and crypt distortion (Fig. 3D, a and b), granulomas (Fig. 3D, c), and crypt abscesses (Fig. 3D, d). Ulcers together with dysplasia and focal invasion were also observed (fig. S4, A and B). In many of these lesions, the crypt epithelial cells had peripheral localization of β-catenin (fig. S4, C and D), suggesting that the invasion of crypts into the submucosa was a reaction to the formation of ulcers rather than cancer. By 4 to 8 months of age, more than half of the mice (7 of 13 mice) had developed from one to three adenomatous polyps, which were identified morphologically and by nuclear β-catenin staining (fig. S4, E and F). Polyps were concentrated in the distal ileum and proximal colon close to the cecum (fig. S5). Small intestine polyps were histologically indistinguishable from those detected in mouse models of hereditary polyposis, which we have characterized in detail in our earlier studies (22, 36, 37) (Fig. 3, E and F), whereas colonic polyps had a serrated architecture that is typical of more aggressive adenomatous polyps (40, 41) (Fig. 3G).

Fig. 3. CD4CreCtnnbex3 mice have intestinal pathologies.

(A) Stabilization of β-catenin in T cells causes inflammation, as well as crypt and villus elongation in CD4CreCtnnbex3 compared to WT. (B) Increased mitosis of intestinal epithelial cells revealed by bromodeoxyuridine (BrdU) staining with hematoxylin counterstain. (C) Quantification of epithelial mitotic activity (BrdU staining) in small intestine and colon of WT and CD4CreCtnnbex3 mice. Values represent the average count of dividing cells/crypt in 10 to 30 independent fields. Mean ± SEM is shown. Two-way analysis of variance (ANOVA): intestine, P < 0.0001; colon, P = 0.0003. (D) Representative images of ulcers (a and b), granulomas (c), and crypt abscesses (d) in CD4CreCtnnbex3 intestine. (E) Small intestinal polyp (arrow). (F) High-power view of small intestinal polyp of CD4CreCtnnbex3 mice. Scale bars, 25 μm. (G) Adenoma with serrated features and invasion in CD4CreCtnnbex3 colon with hyperplastic surface epithelium (black arrow) overriding a large lymphoid aggregate (white arrowhead). (H) Small intestinal polyp (arrow). (I) High-power view of small intestinal polyp in Foxp3CreCtnnbex3 mice. Scale bars, 25 μm. Black arrows in (F) and (I) demonstrate nuclear atypia and nuclear stratification compared to overriding normal epithelium (white arrow) or a mildly reactive adjacent crypt, respectively (J). Additional polyp with hyperplastic surface epithelium (black arrow) overriding a large lymphoid aggregate (white arrowhead) in Foxp3CreCtnnbex3 mouse. Scale bars, 200 μm, unless otherwise indicated.

Analysis of Foxp3CreCtnnb1ex3 compound mutant mice revealed that Wnt/β-catenin signaling in Tregs critically contributed to colitis and cancer. Foxp3-CreCtnnb1ex3 mice showed progressive inflammation leading to polyp formation. At 4 months, the mice had growing lymphoid structures in the small intestine (fig. S6, A and B). By 6 months of age, the follicles had become abnormally enlarged, covered by crypt and villus structures (fig. S6, C and D). Abnormally enlarged lymphoid follicles were also apparent in the colon (fig. S6, E and F). Nine-month-old mice had typical adenomatous polyps (Fig. 3H) consisting of aberrant epithelial cells with abnormal hyperchromatic nuclei (Fig. 3I). In addition, hyperproliferative crypts were also seen, at times growing over enlarged lymphoid follicles (Fig. 3J). Thus, sustained β-catenin activity in the T cell compartment causes inflammation, ulcers, and ultimately epithelial transformation and polyposis in the small intestine and colon.

Polyps in the small intestine and colon were densely infiltrated with mast cells, which localized to the parenchyma, stroma, and submucosa of the lesions (fig. S7, A and B). Mastocytosis was focal, and mast cell numbers declined outside the polyps (fig. S7C). CD11b+ myeloid cells, B220+ B cells, and CD11c+ antigen-presenting cells were increased in numbers in the spleen and lymph nodes of the mice, indicating both local and systemic inflammation (fig. S7D).

Constitutive activation of β-catenin in T cells promotes TH17 commitment and sustained inflammation

To confirm that gut pathologies were induced by constitutive activation of β-catenin in T cells rather than by leaky expression of Cre in gut epithelial cells, we depleted T cells and B cells by introducing a homozygous Rag2 mutation in CD4CreCtnnb1ex3 mice. CD4CreCtnnb1ex3 Rag2−/− mice (n = 10) observed until 10 months of age did not exhibit inflammation or polyps (Fig. 4A), demonstrating that progressive colitis and growth of polyps in CD4CreCtnnb1ex3 mice were strictly lymphocyte-dependent. Additionally, intracellular β-catenin staining and fluorescence-activated cell sorting analysis of leukocytes derived from CD4CreCtnnb1ex3 mice revealed that β-catenin protein levels were elevated in circulating CD4 and CD8 T cells but not in macrophages, dendritic cells, or B cells (Fig. 4B). Stabilization of β-catenin was confirmed by Western blot analysis of sorted CD4+ T cells, which showed elevated expression of β-catenin in CD4+ T cells from the spleen and small intestine (Fig. 4C).

Fig. 4. Stabilization of β-catenin in T cells induces inflammation and polyposis.

(A) Gut dysplasia in CD4CreCtnnbex3 mice is T cell–dependent; hematoxylin and eosin–stained sections of gut rolls from the indicated age-matched mice (n = 10). Note that no polyps developed in CD4CreCtnnbex3 Rag2−/− mice. Images are to scale. Scale bars, 0.5 mm. (B) Histogram overlays show intracellular (IC) β-catenin staining in mesenteric lymph node cells from WT (gray line) and CD4CreCtnnb1ex3 (black line) mice. Cells were gated for CD4+ T cells, CD8+ T cells, macrophages (MΦ), dendritic cells (DC), and B cells. Filled gray histogram shows isotype control staining. Results are representative of n ≥ 3 experiments. (C) Western blot of CD4+ T cell lysates sorted from the indicated tissues and mice. β-Tubulin was probed as loading control. Results are representative of two experiments. (D) Histogram overlays comparing surface expression of activation markers (CD69, NKG2D, and CD122) and naïve cell marker CD62L in CD4 and CD8 mononuclear cells from WT and CD4CreCtnnbex3 mice. Results are representative of four experiments. (E) Upper histograms show frequencies and cell numbers of CD4+ and CD8+ cells from mesenteric lymph nodes as indicated at 4 weeks of age (n = 3 to 8). Lower histograms show frequencies and cell numbers of CD4+ and CD8+ cells from WT and CD4CreCtnnbex3 mesenteric lymph nodes as indicated at 8 to 10 weeks of age (n = 3 to 8). Data are means ± SEM.

To investigate the mechanism of action of β-catenin in T cells, we determined their activation status and numbers. T cells with constitutively active β-catenin expressed several activation markers including CD69, CD122, and NKG2D, and they down-regulated CD62L (Fig. 4D). We reported earlier that stabilization of β-catenin during thymic development stalls differentiation of T cells at the double-positive stage (25, 26). Accordingly, at 4 weeks of age, CD4CreCtnnb1ex3 mice had fewer thymic and peripheral CD4+ and CD8+ T cells; however, by 8 weeks of age, the absolute numbers of peripheral T cells had increased to near-normal levels (Fig. 4E and table S1).

An abnormally large fraction of CD4+ T cells from CD4CreCtnnb1ex3 mice expressed IL-17 in the thymus and in peripheral lymphoid organs, including mesenteric lymph nodes and intestine (Fig. 5, A and B, and table S1). Furthermore, the proinflammatory cytokines IL-17, tumor necrosis factor–α (TNF-α), and IL-6 were elevated in the small intestine and cecum of CD4CreCtnnb1ex3 mice (Fig. 5C and table S1). To demonstrate their proinflammatory properties, we transferred total peripheral T cells (106 cells per mouse) from CD4CreCtnnb1ex3 or control CD4Cre mice to Rag2−/− recipients. Three weeks after transfer, TH17 and TH1 cytokines, as well as IL-10, but not IL-2, were significantly higher in the serum of mice that received T cells with constitutively active β-catenin as compared to mice that received control T cells (Fig. 5D and table S1). Mice receiving CD4CreCtnnb1ex3 T cells did not survive beyond 4 weeks after transfer, whereas those receiving CD4Cre T cells remained healthy. These observations demonstrate that T cells with elevated levels of β-catenin are TH17-biased and proinflammatory. This is in line with an earlier report that β-catenin is up-regulated in ex vivo–differentiated TH17 cells (33). On the basis of these observations, we conclude that high levels of β-catenin in T cells cause chronic T cell activation, TH17 commitment, and pathogenic inflammation that predispose the small intestine and colon to cancer.

Fig. 5. CD4CreCtnnbex3 mice have increased amounts of proinflammatory cytokines.

(A) Dot plots show IL-17–expressing CD4+ T cells in the indicated organs and mice. Note increased frequency of CD4+IL-17+ T cells in CD4CreCtnnbex3 compared to WT. (B) Histograms show frequencies of IL-17–expressing CD4+ T cells in the indicated organs from WT and CD4CreCtnnbex3 mice (n = 3 to 5). Two-tailed Student’s t test: thymus, P = 0.0100; lymph nodes, P = 0.0021; gut, P = 0.0029. (C) Histograms show the concentration of the indicated cytokines in lysates from the indicated parts of the intestine of WT versus CD4CreCtnnbex3 mice (n = 4 to 6). Representative of two independent experiments. Two-tailed t test: TNF-α (intestine, P = 0.0036; cecum, P = 0.00036); IL-17 (intestine, P = 0.0042; cecum, P = 0.00026); IL-6 (intestine, P = 0.0008; cecum, P = 0.0032). (D) Histograms show concentrations of the indicated cytokines in the serum of recipient Rag2−/− mice injected with 106 T cells isolated from WT or CD4CreCtnnbex3 mice, quantified 3 weeks after transfer (n = 3). Results are representative of two independent experiments. Two-tailed t test: IL-17, P = 0.0155; TNF-α, P = 0.0088; interferon-γ (IFN-γ), P = 0.0101; IL-10, P = 0.0094; IL-2, P = 0.3757. Representative of two independent experiments. For (B) to (D), data are means ± SEM.

Activation of β-catenin impairs normal Treg development and function

Expression of β-catenin enhances survival of ex vivo–differentiated Tregs (31), raising the possibility that constitutive activation of β-catenin may lead to expansion of Tregs. We observed that CD4CreCtnnb1ex3 mice had reduced numbers of thymic Foxp3+CD4+ Tregs compared to controls (Fig. 6A and table S1). However, a Treg reduction was not apparent in spleen and lymph nodes (Fig. 6A and table S1), suggesting a developmental defect that is compensated in the periphery, perhaps through generation and/or survival of extrathymic Tregs. To confirm that the defect was cell-intrinsic, we generated competitive bone marrow (BM) chimeras in which BM progenitors from CD4CreCtnnb1ex3 (Thy1.2) and wild-type (Thy1.1) mice mixed at 1:1 ratio were injected into lethally irradiated syngeneic wild-type hosts. At 6 weeks after transfer, CD4CreCtnnb1ex3 BM progenitors gave rise to significantly (P < 0.001) fewer thymic Tregs than did wild-type BM progenitors (fig. S8). Therefore, we conclude that a cell-intrinsic defect accounts for the reduced generation of thymic Tregs in CD4CreCtnnb1ex3 mice.

Fig. 6. Tregs with constitutively active β-catenin show impaired development and function.

(A) Frequency and number of CD4+Foxp3+ among CD4+ T cells in the indicated organs and mice (n = 3 to 11). CD4CreCtnnb1ex3 and littermate control mice were analyzed at 4 weeks of age before the onset of inflammation. Two-tailed t test: Foxp3 frequency (thymus, P = 0.0001; spleen, P = 0.0003; lymph nodes, P = 0.7758); Foxp3 numbers (thymus, P = 0.0030; spleen, P = 0.0004; lymph nodes, P < 0.0001). Data are means ± SEM. (B) Scheme of in vivo colitis induction in Rag2−/− mice to assess the anti-inflammatory properties of Tregs. Arrows indicate adoptive transfer of the indicated sorted cell populations. (C) (Left) Weight loss (as a percentage of starting weight) over time after injection of naïve T cells followed by the specified Tregs 4 weeks later. Two-way ANOVA test: naïve T cells + CD4Cre-Tregs/naïve T cells + CD4CreCtnnb1ex3-Tregs; week 5 (P < 0.05), week 6 (P < 0.0001), week 7 (P < 0.0001). (Right) Colitis score was calculated on the basis of percent colon involvement, severity of inflammation, level of involvement, and extent of damage. Data are means ± SEM. (D) Contour plots show expression of the indicated proinflammatory cytokines in CD45.2+CD4+Foxp3+ donor Tregs retrieved from the gut of Rag2−/− recipients at 7 weeks. (E) Histograms show average frequencies of cytokine expression in transferred Tregs (n = 3). Data are means ± SEM. (F) In vitro CD4 T cell proliferation inhibition assay. Carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled naïve WT T cells were stimulated with IL-2 and anti-CD3 and anti-CD28, and then plated at a 1:2 ratio with CD4Cre or CD4CreCtnnbex3 Tregs. Proliferation was assessed by CFSE dilution. Filled histogram shows proliferation of activated T cells in the absence of Tregs. Note that CD4CreCtnnb1ex3 Tregs only partially inhibit proliferation of activated T cells compared to WT Tregs. The result shown is representative of two independent experiments done in triplicates.

Normally, Tregs have potent anti-inflammatory properties. Adoptive transfer of Tregs from healthy mice to lymphopenic mice suppresses the cytokine storm caused by previous or simultaneous transfer of naïve CD4 T cells (42). The ability of Tregs to suppress inflammation is protective, and adoptive Treg transfer into polyp-ridden mice suppresses cancer-associated inflammation and causes regression of the polyps (18). However, we previously showed that Tregs from mice with polyposis (18) or from colon cancer patients are functionally altered (11, 12), exhibit TH17 characteristics, and promote inflammation and tumor growth. Therefore, we considered that Tregs with stabilized β-catenin may have lost their anti-inflammatory functions, and that this impairment is an underlying mechanism in colitis and colon cancer.

To evaluate this possibility, we first induced colitis by transferring naïve CD4+ T cells (CD4+CD45RBhiCD25 Thy1.1) into Rag2−/− mice. Four weeks later, when the mice started losing weight, we transferred CD4CreCtnnb1ex3 or CD4Cre Tregs (CD4+CD25hi or CD4+Foxp3-GFP+ Thy1.2+) to suppress the colitis. Tregs with constitutively active β-catenin were significantly (P < 0.0001) less effective than control (CD4Cre) Tregs in protecting the recipient mice from colitis (Fig. 6, B and C, table S1, and fig. S9). Analysis of Thy1.2+CD4+Foxp3-GFP+ Tregs retrieved from the colons of Rag2−/− recipients at the end point showed that a much larger fraction of the transferred Tregs with constitutively active β-catenin expressed proinflammatory cytokines including IL-17, IFN-γ, and TNF-α as compared to transferred wild-type Tregs (Fig. 6, D and E). Tregs are also potent suppressors of T cell functions, which normally are assayed by their ability to suppress proliferation of stimulated CD4+ T cells in vitro. Foxp3+ Tregs with constitutively active β-catenin inhibited proliferation of CD4+ T cells, although with less potency than control CD4Cre Tregs (Fig. 6F). These observations demonstrate that activation of β-catenin in Tregs impairs their anti-inflammatory functions and contributes to systemic inflammation, colitis, and polyposis in CD4CreCtnnb1ex3 mice.

β-Catenin increases chromatin accessibility and expression of target genes

We gained insight into the molecular mechanisms by which β-catenin alters the properties of T cells, by comparing gene expression of CD4CreCtnnb1ex3 and wild-type thymocytes using Affymetrix arrays (26). T cell–specific activation of β-catenin resulted in robust expression of TH17 family genes including RORγt, the signature transcription factor of the TH17 lineage (Fig. 7A). This expression pattern showed a clear overlap with that of T cells from polyp-ridden APC+/Δ468 mice (Fig. 2G). To investigate the underlying reason for the change in gene expression, we performed ChIP-seq analyses of Tcf-1 and the distribution of histone marks. In wild-type thymocytes, Tcf-1 preferentially bound to consensus TCF motifs (P < 10−940) within enhancers and promoters, and associated genome-wide with marks of open chromatin such as lysine acetylation of histone 3 (H3KAc) (Fig. 7B, upper panels, black lines). In contrast, there was little association of Tcf-1 with marks of closed chromatin such as lysine 27 trimethylation of histone 3 (H3K27me3) (Fig. 7B, lower panels, black lines). Stabilization of β-catenin in CD4CreCtnnb1ex3 thymocytes enhanced H3KAc marks in both promoters and enhancers spanning more than 3 kb from the Tcf-1 binding sites (Fig. 7B, upper panels, red lines). Conversely, H3K27me3 marks (Fig. 7B, lower panels, red lines) were reduced. These observations are consistent with the notion that the observed activation of gene expression involves interaction of β-catenin with its T cell–specific DNA binding partner Tcf-1. Presumably, Tcf-1 binds accessible loci, and β-catenin enhances their accessibility. Therefore, we tested the possibility that enhanced accessibility could translate into elevated gene expression.

Fig. 7. β-Catenin stabilization induces sustained RORγt expression in T cells and Tregs.

(A) Heatmap depicts expression changes of TH17 lineage genes induced by β-catenin stabilization in CD4CreCtnnb1ex3 thymocytes. Columns represent independent arrays for the indicated mouse strain. Gene IDs and average fold expression changes are shown. Red letters depict genes similarly changed in gut CD4 T cells during polyposis as shown in Fig. 2G. (B) Histograms illustrate normalized average tag density distribution of the indicated histone marks in CD4Cre (black) and CD4CreCtnnb1ex3 (red) thymocytes centered on shared Tcf-1 peaks. Tcf-1 binding at sites <3 kb or >3 kb from the nearest transcription start site was considered promoter or enhancer binding, respectively. (C) Peaks of Tcf-1 binding and H3KAc marks in the Rorc locus of CD4Cre and CD4CreCtnnb1ex3 thymocytes. Diagrams depict the Rorc locus. RORγt is in cyan. Gray bars indicate more than threefold increase in H3KAc marks. (D) Histogram overlays show expression of RORγt-GFP reporter in gated CD4+ SP thymocytes, and histogram bars show average mean fluorescence intensity (MFI) of RORγt-GFP expression (n = 3 to 6) in the indicated mouse strain. ***P < 0.0001, two-tailed t test. Data are means ± SEM. (E) (Left) Contour plots of intracellular Foxp3/RORγt staining in gated CD4+ splenic T cells from the indicated mouse strains. (Right) Frequency of RORγt+ cells among CD4+ or CD4+Foxp3+ splenocytes as indicated (n = 3). One-way ANOVA, Bonferroni’s multiple comparison test: %RORγt+CD4+ (P = 0.0030), %RORγt+Foxp3+ (P = 0.0018). Representative of two independent experiments, done in triplicate. Data are means ± SEM.

Given that RORγt is the signature transcription factor for TH17 differentiation and was found by our expression arrays to be up-regulated in CD4CreCtnnb1ex3 thymocytes, we examined the epigenetic state of the Rorc locus, which encodes RORγt. Our ChIP-seq analysis revealed comparable Tcf-1 binding to bona fide promoter and intragenic regions of the RORγt gene in wild-type and CD4CreCtnnb1ex3 thymocytes (Fig. 7C). Marks of open chromatin such as H3KAc were significantly elevated in Tcf-1 bound regions (Fig. 7C), covering a large part of the Rorc locus. This finding is in line with earlier findings that β-catenin induces histone acetylation by recruiting the histone acetyltransferases (HATs), CBP (cyclic adenosine monophosphate response element–binding protein), and p300 to Tcf sites (43).

We confirmed the up-regulation of RORγt expression in T cells that had constitutively activated β-catenin using a transgenic RORγt reporter, by intracellular staining of RORγt and by reverse transcription polymerase chain reaction (PCR). We crossed reporter mice that contained an insertion of the GFP (G) gene in the RORγt locus (44) with CD4CreCtnnb1ex3 and CD4Cre control mice. The resulting heterozygous RORγtG/+ mice had one wild-type (+) and one truncated (G) RORγt allele, but were healthy with no detectable hematopoietic abnormalities or other pathologies. RORγt is required during thymic development, but its expression is shut down in mature CD4+ single-positive (SP) thymocytes. Accordingly, the RORγt-GFP reporter was no longer expressed in control CD4Cre CD4+ SP. However, expression of RORγt-GFP was detectable in CD4CreCtnnb1ex3 CD4+ SP that had stabilized β-catenin (Fig. 7D and table S1). A significantly larger fraction of peripheral CD4+Foxp3 T cells (P = 0.0148) and CD4+Foxp3+ Tregs (P = 0.0076) in CD4CreCtnnb1ex3 mice exhibited RORγt expression compared to the same cells in control CD4Cre mice (Fig. 7E and table S1). CD4CreRORγtG/G and CD4CreCtnnb1ex3RORγtG/G RORγt-deficient mice were used as intracellular staining controls. Consistently, quantitative PCR showed that RORγt expression was about fourfold higher in thymic and threefold higher in peripheral Tregs with stabilized β-catenin (Fig. 7E, table S1, and fig. S10). Thus, three independent assays showed increased expression of RORγt in thymocytes and T cells that had constitutively active β-catenin.

These findings demonstrate that activation of β-catenin in T cells introduces global changes in chromatin landscape, increases accessibility, and enhances expression of target genes including RORγt.

RORγt functions downstream of β-catenin compromise Treg properties

RORγt is the signature transcription factor of proinflammatory TH17 cells, raising the possibility that T cell abnormalities upon constitutive activation of β-catenin are, in part, caused by deregulated expression of RORγt. Therefore, we analyzed CD4CreCtnnb1ex3RORγtG/G mice. In the absence of RORγt, normal development and function of T cells and Tregs were partially restored despite the presence of constitutively active β-catenin. In particular, ablation of RORγt reduced the frequencies of IL-17–expressing peripheral CD4+ T cells and Tregs to levels comparable to wild type (Fig. 8A and table S1). This confirmed that RORγt was indispensable for the deregulation of TH17 inflammation by β-catenin. Ablation of RORγt also increased the frequency of thymic Tregs expressing the Treg markers Foxp3 and GITR (Fig. 8B and table S1). It further restored the expression of Foxp3 protein, which was reduced in Tregs with activated β-catenin, to normal levels (Fig. 8, C and D, and table S1). Ablation of RORγt in CD4CreCtnnb1ex3 Tregs with constitutively active β-catenin significantly improved their ability to suppress proliferation of in vitro–activated T cells (P < 0.0001) (Fig. 8E and table S1). These findings are consistent with the interpretation that Treg dysfunction in CD4CreCtnnb1ex3 mice is the result of persistent β-catenin activity and activation of RORγt.

Fig. 8. β-Catenin–mediated up-regulation of RORγt affects the development and function of Tregs.

(A) Histograms show the frequency (%) of IL-17+CD4+ cells (left) and IL-17+Foxp3+ Tregs (right) in the spleen of the indicated mice (n = 3 to 4). Representative of two independent experiments. Two-tailed t test: %CD4+IL-17+ (CD4CreCtnnb1ex3RORγG/+/CD4CreCtnnb1ex3RORγG/G, P = 0.0002; CD4CreRORγG/+/CD4CreCtnnb1ex3RORγG/+, P = 0.0045; CD4CreCtnnb1ex3RORγG/+/CD4CreRORγG/G, P < 0.0001); %IL-17+Foxp3+ (CD4CreCtnnb1ex3RORγG/+/CD4CreCtnnb1ex3RORγG/G, P = 0.0449). (B) Contour plots of CD4/Foxp3 (upper) and CD25/GITR (glucocorticoid-induced TNF receptor) (lower) show gated CD4+ SP thymocytes of the indicated mice. Histogram below shows the frequency of Foxp3+ cells among gated CD4+ SP thymocytes of the indicated mice (n = 3 to 4). Two-tailed t test: CD4CreCtnnb1ex3RORγtG/+/CD4CreCtnnb1ex3RORγtG/G, P = 0.0011; CD4CreRORγtG/G/CD4CreCtnnb1ex3RORγtG/+, P < 0.0001; CD4CreRORγtG/+/CD4CreCtnnb1ex3RORγt +/G, P < 0.0001. Representative of two independent experiments. (C) Histogram overlay shows intracellular Foxp3 staining in CD4+Foxp3+ splenic T cells. (D) Histogram bars show average Foxp3 MFI in the indicated mouse strains (n = 3 to 4). Note the comparable increase in Foxp3 upon RORγt ablation both in Tregs with physiological β-catenin levels and in Tregs with stabilized β-catenin. Two-tailed t test: CD4CreRORγtG/+/CD4CreCtnnb1ex3RORγtG/+, P = 0.0473; CD4CreCtnnb1ex3RORγtG/+/CD4CreCtnnb1ex3RORγtG/G, P = 0.0001; CD4CreRORγtG/G/CD4CreCtnnb1ex3RORγtG/G, P = 0.0301; CD4CreRORγtG/G/CD4CreCtnnb1ex3RORγtG/+, P = 0.0004. (E) Histogram bars show the fraction of CFSE-labeled effector T cells (CD4+CD25) that have divided when cultured at a 1:2 or 1:1 ratio of Treg/effector T cells as indicated. Tregs were isolated from the indicated mice (n = 3). Note that ablation of RORγt in Tregs with stabilized β-catenin significantly enhances their ability to inhibit proliferation of activated CD4 T cells. Two-tailed t test: Treg:CD4-(1:2) CD4Cre/CD4CreCtnnb1ex3, P = 0.0002; CD4CreCtnnb1ex3/CD4CreRORγtG/G, P < 0.0001; CD4CreCtnnb1ex3/CD4CreCtnnb1ex3RORγtG/G, P = 0.0022; Treg:CD4-(1:1) no-Tregs/CD4CreCtnnb1ex3, P = 0.0006. Representative of two independent experiments. For graphs in (A), (B), (D), and (E), data are means ± SEM.

DISCUSSION

We recently reported that in human colon cancer, a subset of Tregs that coexpress Foxp3 and RORγt expand preferentially in a tumor-dependent manner (12). RORγt is the signature transcription factor of TH17 cells and promotes expression of the proinflammatory cytokine IL-17 (45). Here, we observed that expression of β-catenin was increasingly elevated in colon-infiltrating CD4+ T cells, including Tregs of patients with colon cancer and patients with long-lasting UC, as the chronically inflamed tissues progressed to cancer. These findings led us to question a mechanistic link between the expression of β-catenin and the expression of RORγt in T cells during colitis and colon cancer (12).

The biological role of β-catenin activation in T cells was revealed by analysis of mouse models. Previously, we had shown that APC+/Δ468 mice with hereditary polyposis depend on TH17 cytokines to develop polyps and, in addition, have a subset of RORγt+ Tregs with proinflammatory properties (12). The APC+/Δ468 model allowed us to examine how expression of RORγt was related to expression of β-catenin. We demonstrated by independent assays that in polyp-ridden APC+/Δ468 mice, effector T cells and Tregs express high levels of β-catenin and Wnt pathway genes. Using the CD4CreCtnnb1ex3 and Foxp3CreCtnnb1ex3 models of sustained β-catenin activity, we showed that persistent Wnt/β-catenin signaling in all T cells or only Tregs culminated in long-lasting colitis, eventually progressing to intestinal and colon adenomatous polyps. We ruled out the possibility that this pathology resulted from leaky expression of stable β-catenin in gut epithelial cells, by showing absence of disease in lymphocyte-deficient Rag2−/− CD4CreCtnnb1ex3 mice. It is well known that chronic inflammation in the colon predisposes to cancer, and our mouse models provided insight into the mechanistic relevance of β-catenin signaling in T cells to this process. In particular, the mouse model demonstrated that sustained activation of β-catenin in human T cells is biologically relevant. This was further substantiated with the APC+/Δ468 mouse model of hereditary polyposis. T cells and, in particular, Tregs in these mice had similar characteristics to their counterparts in CD4CreCtnnb1ex3 mice and in colon cancer patients, including overexpression of β-catenin.

Our earlier studies highlighted the strategically important role of Tregs in the control of TH17 inflammation during mouse polyposis (18) and also pointed to a similar function in human colon cancer (12). We previously showed that Treg-targeted ablation of RORγt protects APC+/Δ468 mice against polyposis (12). Here, we demonstrated that Treg-targeted activation of β-catenin increased gut infiltration by lymphocytes and culminated in reactive tissue as well as bona fide adenomatous polyps. These observations agree with the notion that Wnt/β-catenin signaling in Tregs is critical for gain of Treg proinflammatory properties and contributes to the carcinogenic processes in the intestine and colon.

Stabilized β-catenin mediates its effects by translocating to the nucleus, where it interacts with DNA binding Tcf/Lef (lymphoid enhancer factor) factors and regulates transcription of target genes. We showed that expression of genes associated with T cell activation and TH17 lineage commitment is up-regulated after stabilization of β-catenin in CD4CreCtnnb1ex3 thymocytes. Many of these genes were naturally elevated in gut-infiltrating T cells during polyposis. These global changes in gene expression involved β-catenin–induced changes in chromatin accessibility. ChIP-seq of Tcf-1 and chromatin marks revealed that in thymocytes, Tcf-1 binds primarily to its conserved binding motifs in bona fide promoter and enhancer regions of target genes (46). Activation of β-catenin increased the accessibility of these sites through the deposition of H3KAc marks over long distances from the Tcf-1 binding. Our results are consistent with earlier findings in Drosophila, showing recruitment of β-catenin to TCF DNA binding sites in complex with the histone H3 acetyltransferase (HAT) CBP, and rapid, extensive histone acetylation (47).

We showed that Tcf-1 binds to consensus motifs in the Rorc locus. Activation of β-catenin enhanced H3KAc in this locus and promoted the expression of RORγt in T cells including Tregs. Our findings confirm and extend recent observations that in thymocytes, RORγt gene expression is regulated by Wnt/β-catenin signaling (30). However, we did not find increased Tcf-1 binding to DNA in response to stabilization of β-catenin; rather, we observed that β-catenin enhances accessibility of the locus. Our findings are consistent with reports that Wnt/β-catenin genes are up-regulated during in vitro TH17 differentiation of T cells (33). Our results are also consistent with reports that Tcf-1 suppresses IL-17 gene expression and with the notion that elevated levels of β-catenin convert this suppression to activation (34, 35). Finally, our results agree with a recent report indicating that Treg differentiation involves the programmed epigenetic closing of sequences with Tcf/Lef binding motifs (48), and further suggest that elevated levels of β-catenin prevent this process, resulting in TH17 commitment of thymocytes and defective Treg development.

In summary, we demonstrated that T cells and Tregs have tumor-promoting roles in colon cancer that are epigenetically imprinted by Wnt/β-catenin signaling, the same pathway that initiates colon cancer in intestinal epithelial cells. Because β-catenin is also up-regulated in proinflammatory monocytes and antigen-presenting cells (49), it is an attractive candidate molecule for shaping a tumor-promoting cancer field (50) in inflammation-driven malignancies. One potential caveat is that up-regulation of β-catenin in T cells and Tregs may be a common feature of all chronic inflammatory diseases.

MATERIALS AND METHODS

Detailed Materials and Methods are provided in the Supplementary Materials.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/6/225/225ra28/DC1

Materials and Methods

Fig. S1. β-Catenin is up-regulated in T cells and Tregs of colon cancer patients (independent fluorescence channels shown).

Fig. S2. β-Catenin and Wnt pathway genes are up-regulated in CD4CreCtnnb1ex3 Tregs.

Fig. S3. T cells from Ts4CreAPC+/lox468 mice express elevated levels of β-catenin.

Fig. S4. The colon of CD4CreCtnnb1ex3 mice develops ulcers and reactive tissue.

Fig. S5. CD4CreCtnnb1ex3 polyps are localized in distal ileum, cecum, and proximal colon.

Fig. S6. Stabilization of β-catenin in Tregs induces lymphadenopathy and crypt hyperplasia.

Fig. S7. Inflammatory cells accumulate in mice with T cell–specific stabilization of β-catenin.

Fig. S8. Stabilization of β-catenin in T cells induces a cell-intrinsic defect in Treg development.

Fig. S9. CD4CreCtnnb1ex3 Tregs fail to suppress inflammation.

Fig. S10. CD4CreCtnnb1ex3 Tregs express RORγt and IL-17.

Table S1. Original data values (provided as a separate Excel file).

Table S2. Antibodies used in flow cytometry.

REFERENCES AND NOTES

  1. Acknowledgments: Tcf-1 antibody was a gift of H. Kawamoto (Riken Research Center for Allergy and Immunology, Yokohama, Japan). We thank S. Rosen and C. Benoist for guidance and encouragement and M.-L. Alegre for suggestions and comments. We are grateful to A. Saadalla and Z. Mojtahedi for help with processing of patient specimens. Funding: This work was supported by NIH grant R21AI076720, P30-DK42086-pilot, and American Cancer Society grant ACS/RSG-LIB-113428 (to F.G.) and NIH grant 1R01CA160436 and Circle of Service award (Robert H. Lurie Comprehensive Cancer Center, to K.K.). S.K. was supported by T32HL007381 Cardiovascular Sciences Training Grant. Author contributions: S.K. designed and performed the experiments, analyzed the data, and helped in manuscript preparation. K.A. designed and performed the experiments, analyzed the data, and helped in manuscript preparation. M.D. performed epigenetic studies. L.M. provided expertise and analyzed the data. C.W. analyzed histological samples. T.S. conducted the experiments. A.O.E. performed epigenetic analyses. N.B. gave conceptual and technical advice and performed the experiments. M.W.K. and V.V. performed immunostainings and analyzed the data. E.M.R. consented patients, acquired specimens, and helped with preparation and analysis of blood. D.J.B., M.M., and A.K. provided patient specimens and advice for analysis. K.K. provided conceptual and technical advice and helped with the design of the experiments and writing of the paper. F.G. designed the experiments, analyzed the data, and wrote the paper. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The data generated by this study have been deposited in the Gene Expression Omnibus database (GSE41229 and GSE7050). We have reported Tcf-1 ChIP-seq analyses (46), and the data are deposited in GSE32311.
View Abstract

Navigate This Article