Signal Transducer and Activator of Transcription 3 (STAT3) Protein Suppresses Adenoma-to-carcinoma Transition in Apcmin/+ Mice via Regulation of Snail-1 (SNAI) Protein Stability*

Background: STAT3 suppresses carcinogenesis of intestinal tumors in Apc min mice. Results: STAT3 suppresses expression of SNAI in intestinal epithelium by regulating GSK3β activity. Conclusion: STAT3 induces degradation of SNAI by promoting GSK3β activity and thereby suppresses adenoma-to-adenocarcinoma transition in Apc min mice. Significance: Our data provide a new insight into the role of STAT3 in colorectal cancer biology. STAT3 was recently reported to suppress tumor invasion in Apcmin/+ mice. We investigated the mechanisms by which STAT3 inhibits intestinal epithelial tumors using Apcmin/+/Stat3IEC-KO mice (intestinal epithelial cell (IEC)-specific deletion of STAT3 in the Apcmin/+ background) to determine the role of STAT3 in carcinogenesis in vivo as well as colorectal cancer cell lines in vitro. To inhibit invasion of IEC tumors, STAT3 functions as a molecular adaptor rather than a transcription factor. Accordingly, the tumors in Apcmin/+/Stat3IEC-KO mice undergo adenoma-to-carcinoma transition and acquire an invasive phenotype. Similarly, STAT3 knockdown in a colorectal cell line enhances IEC invasion. We demonstrate that STAT3 down-regulates SNAI (Snail-1) expression levels and hence suppresses epithelial-mesenchymal transition of colorectal cancer cells. Mechanistically, STAT3 facilitates glycogen synthase kinase (GSK) 3β-mediated degradation of SNAI by regulating phosphorylation of GSK3β. Our data identified a new role for STAT3 in the adenoma-to-carcinoma sequence of intestinal tumors.


STAT3 was recently reported to suppress tumor invasion in Apc min/؉ mice. We investigated the mechanisms by which STAT3 inhibits intestinal epithelial tumors using Apc min/؉ / Stat3 IEC-KO mice (intestinal epithelial cell (IEC)-specific deletion of STAT3 in the Apc min/؉ background) to determine the role of STAT3 in carcinogenesis in vivo as well as colorectal cancer cell lines in vitro.
To inhibit invasion of IEC tumors, STAT3 functions as a molecular adaptor rather than a transcription factor. Accordingly, the tumors in Apc min/؉ /Stat3 IEC-KO mice undergo adenoma-to-carcinoma transition and acquire an invasive phenotype. Similarly, STAT3 knockdown in a colorectal cell line enhances IEC invasion. We demonstrate that STAT3 down-regulates SNAI (Snail-1) expression levels and hence suppresses epithelial-mesenchymal transition of colorectal cancer cells. Mechanistically, STAT3 facilitates glycogen synthase kinase (GSK) 3␤-mediated degradation of SNAI by regulating phosphorylation of GSK3␤. Our data identified a new role for STAT3 in the adenoma-to-carcinoma sequence of intestinal tumors.
STAT3 is a multifunctional transcription factor that is activated by various growth factors (e.g. epidermal growth factor and hepatocyte growth factor) and cytokines (e.g. IL-6 and IL-10). Consequently, STAT3 plays a key role in many biological processes such as cell growth, apoptosis, and inflammation (1). STAT3 is phosphorylated on a tyrosine residue (Tyr-705) by an upstream kinase JAK2 (Janus kinase 2). The phosphorylated protein forms either a homo-or heterodimer with other STAT proteins, then translocates to the nucleus and transcribes target genes. Rare nontranscriptional activities of STAT3 have also been reported. For example, STAT3 functions as an adaptor protein connecting IFNAR1 (interferon ␣ receptor 1) and the p85 regulatory subunit of PI3K (phosphoinositide 3-kinase) (2), and it inhibits stathmin that depolymerizes microtubules (3). Recently, STAT3 was shown to inhibit intestinal epithelial cell (IEC) 4 tumor invasion in Apc min/ϩ mice (4), but the underlying mechanisms remain unclear.
The development of human colon adenomas is initially induced by mutations in APC or ␤-catenin, and its transition to carcinoma is followed by sequential genetic mutations in K-Ras, SMAD2 or SMAD4, and p53. Whereas deletion of p53 does not provoke malignant transformation in Apc min/ϩ mice (5), at least two genes were found critical for adenoma-to-adenocarcinoma transition. Tumors in Smad4 ϩ/Ϫ /Apc min/ϩ mice become malignant, showing an extensive stromal cell proliferation and submucosal invasion (6), and Ephb3 Ϫ/Ϫ / Apc min/ϩ mice develop carcinomas that invade the muscle layer (7). However, the mechanisms by which these genes suppress IEC carcinogenesis are not fully understood.
The invasion of epithelial tumors to the surrounding tissues is associated with epithelial-mesenchymal transition (EMT), a process in which the epithelial tumor cells lose their epithelial phenotype and acquire a mesenchymal phenotype. Epithelial cells that have undergone EMT display reduced intercellular interactions but increased motility and invasiveness. Several transcriptional repressors were identified as inducers of EMT, including SNAI, Slug (Snail-2), E47, Twist, and the Zeb factors (8). Main targets of these proteins are cell adhesion molecules such as E-cadherin and claudins (CLDNs) that link epithelial cells together (8 -11).
Glycogen synthase kinase 3␤ (GSK3␤) performs multiple cellular functions (12). As an important regulator of metastasis, it suppresses the level of SNAI by being phosphorylated on two different sites: phosphorylation on the first site leads to ubiquitination and proteasomal degradation of SNAI whereas that on the second site controls the subcellular localization of SNAI (13). GSK3␤ is inactivated by phosphorylation by several different kinases at the Ser-9 residue (12,14) but also can be regulated by phosphorylation at other residues: it is inactivated by phosphorylation at Thr-390 (15) and is activated by phosphorylation at Tyr-216 (16,17). In addition, GSK3␤ activity is modulated by protein complex formation mediated by GSK3␤-binding proteins and its subcellular localization (18). GSK3␤ is also a key regulator of ␤-catenin (19,20), an effector molecule for the Wnt signaling pathway. Constitutive activation of ␤-catenin due to a mutation in the APC gene is the major cause of colorectal cancers (CRCs) (21) and induces the multiple intestinal neoplasia (min) phenotype in Apc min/ϩ mice (21)(22)(23) Here, we identified a novel adaptor function for STAT3 in IECs, which has a broad implication to CRC. Our data show that in a nontranscriptional fashion, STAT3 suppresses expression of SNAI, a major driving force in EMT and in the invasion of CRC cells. STAT3 interacts with GSK3␤ and negatively regulates phosphorylation of GSK3␤, thereby inducing degradation of SNAI.
Isolation of IECs from Mouse Intestines-Isolation was performed as described previously (25,26).
Cell Culture-Culture conditions of IEC cells have been described previously (25). HCA-7 cells were cultured in low glucose DMEM, and HCT116 and RKO cells in high glucose DMEM (Mediatech, Manassas, VA).
Cell Invasion Assay-The cell invasion assay was performed with a basement membrane-coated CytoSelect TM 24-well cell invasion assay kit according to the manufacturer's instructions (Cell Biolabs, San Diego, CA). IECs (1 ϫ 10 5 cells/well) were plated into each well 3 days after siRNA transfection and cultured for 2 days. The number of invaded cells per 4 high powered fields/well was visually counted using a microscope. The results are the average of two independent experiments, and the statistical analysis was performed by one-way ANOVA.
Cell Proliferation Assay in Mice-Mice were administered BrdU intraperitoneally (2 mg/mouse) and killed 2 h after the injection. BrdU incorporation (proliferation) in paraffinized colonic tissues and the TUNEL assay in paraffinized colonic tissues were performed and analyzed according to the manufacturer's instructions (BD Biosciences).
Generation of STAT3 IEC-KO Mice-To generate STAT3 deficiency specifically in IEC in mice, Stat3 flox/flox mice (a gift from Shizou Akira, Osaka University, Osaka, Japan) were crossed to villin-Cre mice (on the C57BL/6J background; The Jackson Laboratory, Bar Harbor, ME), to create Stat3 IEC-KO mice. Genotyping was performed as described (27). Apc min/ϩ mice were obtained from The Jackson Laboratory. Apc min/ϩ / Stat3 IEC-KO mice were generated by crossing Apc min/ϩ mice with Stat3 IEC-KO mice. All animal protocols received prior approval by the Institutional Animal Care and Use Committee.

STAT3 Suppresses Invasion of Intestinal Tumors in Apc min/ϩ
Mice-To determine the role of STAT3 in intestinal tumorigenesis, we conditionally deleted STAT3 in IEC (STAT3 IEC-KO ) in Apc min/ϩ mice (supplemental Fig. 1, A and B). At 20 weeks of age, the tumor count in Apc min/ϩ /Stat3 IEC-KO mice was comparable with that in Apc min/ϩ mice (supplemental Fig. 1C). In addition, proliferation of IECs in normal or tumor IEC was not significantly affected by STAT3 deletion (supplemental Fig.  1D).
Whereas the IEC tumors in Apc min/ϩ mice were strictly confined to the mucosal crypts, those in the Apc min/ϩ /Stat3 IEC-KO mice invaded the mucosal stroma, submucosa, and muscle (Fig.  1A). Such invasive tumors were found in all of the Apc min/ϩ / Stat3 IEC-KO mice examined at 20 weeks (n ϭ 20) of age and were confirmed by ␤-catenin immunostaining (Fig. 1A). At 25 weeks of age, the IEC tumors in Apc min/ϩ /Stat3 ⌬1EC mice penetrated through the intestinal serosa (Fig. 1B). Although some of the  -CAT GTT CGT CAT GGG TGT GAA CCA-3Ј  5Ј-AGT GAT GGC ATG GAC TGT GGT CAT-3Ј  h-SNAI  5Ј-TAC AGC GAG CTG CAG GAC TCT AAT-3Ј  5Ј-AGG ACA GAG TCC CAG ATG AGC ATT-3Ј tumors penetrating serosa were necrotic, live tumor cells (␤-catenin-positive) were also found surrounded by stromal tissues (Fig. 1B). However, STAT3 deletion did not significantly affect the levels of ␤-catenin in tumor cells (Fig. 1C). In a human colorectal cell line, STAT3 knockdown slightly enhanced the levels of ␤-catenin, phospho-␤-catenin, and axin (a target of ␤-catenin), indicating that it can suppress the ␤-catenin pathway (Fig. 1D). Taken together, these findings validate that STAT3 in this animal model inhibits invasion of IEC tumors without affecting tumor induction (4). STAT3 Suppresses EMT-Because EMT often precedes tumor invasion, we next investigated whether STAT3 deletion induces EMT in IEC tumors. Expression of E-cadherin or occludin was largely unaffected by STAT3 deletion (Fig. 2A), but expression of CLDN-3 and CLDN-5 was markedly decreased in tumor cells of Apc min/ϩ /Stat3 IEC-KO mice compared with those in Apc min/ϩ mice (Fig. 2B). Vimentin was expressed mainly in mesenchymal cells and rarely in tumor cells in Apc min/ϩ mice; however, all of the tumor cells in the Apc min/ϩ /Stat3 IEC-KO mice expressed vimentin at a high level (Fig. 2C). Fibronectin was also up-regulated in the tumor cells of Apc min/ϩ /Stat3 IEC-KO mice whereas it was not detectable in tumor from Apc min/ϩ mice (Fig. 2D). Interestingly fibronectin in Apc min/ϩ /Stat3 IEC-KO was observed only in IEC tumor cells around the necrotic regions (Fig. 2D). Collectively, these data highly suggest that STAT3 suppresses EMT in IEC tumor cells in Apc min/ϩ mice.
STAT3 Suppresses Expression of MMP14-We next checked the expression of MMPs, which are important for tumor inva-sion (28,29). Whereas MMP7 expression in tumor cells was significantly lower in Apc min/ϩ /Stat3 IEC-KO mice than that in Apc min/ϩ mice, MMP9 or MMP15 (MT2-MMP) expression was minimal and observed mostly in nonepithelial cells in both strains of mice (supplemental Fig. 2). In contrast, MMP14 (MT1-MMP) expression in tumor cells was highly induced in Apc min/ϩ /Stat3 IEC-KO mice but undetectable in Apc min/ϩ mice (Fig. 2E). Interestingly, MMP14, a membrane type metalloproteinase, is particularly efficient in hydrolyzing basement membranes (30,31), and its expression can be induced by SNAI in carcinomas (30).
STAT3 Suppresses Tumor Invasion via Down-regulation of SNAI-To examine further the suppressive effect of STAT3 on EMT, known regulators of EMT were examined. Whereas expression of Slug and Zeb-1 were barely detectable, and their levels were not significantly different in IEC tumor cells from both strains of mice (supplemental Fig. 3), SNAI expression was significantly elevated in Apc min/ϩ /Stat3 IEC-KO tumor cells (Fig.  3A). To investigate further the role of STAT3 in SNAI expression and the role of SNAI in invasion, we used CRC cell lines in the subsequent studies. In three different CRC cell lines, HCT116, HCA-7, and RKO, STAT3 knockdown increased the expression of SNAI (Fig. 3B). Conversely, ectopic expression of STAT3-WT suppressed expression of SNAI (Fig. 3C).
We next tested whether the elevated SNAI was responsible for the invasive phenotype of STAT3-depleted CRC cells. STAT3, SNAI, or STAT3 with SNAI was silenced in HCT116 cells, and the transfected cells were subjected to an in vitro cell invasion assay. STAT3 knockdown significantly enhanced invasiveness compared with control. In addition, SNAI knockdown alone or a combination of STAT3 and SNAI knockdown significantly curtailed the invasion (Fig. 3D). Thus, these data indicate that SNAI promotes invasion and that STAT3 downregulates SNAI and thereby suppresses invasion of CRC cells.
STAT3 Promotes GSK3␤ Activity to Induce Degradation of SNAI-We next studied how STAT3 regulates SNAI expression. STAT3 deletion or knockdown in IEC did not significantly affect the transcription of SNAI (Fig. 4, A and B). We therefore examined whether STAT3 regulated SNAI via post-translational modification. SNAI in HCT116 cells was constitutively ubiquitinated and degraded by proteasomes (Fig. 4C). STAT3 knockdown significantly diminished the levels of SNAI ubiquitination, even though the levels of SNAI were higher than control (Fig. 4C). Because GSK3␤ phosphorylates SNAI, which leads to ubiquitination and proteasomal degradation (13), we investigated whether STAT3 is involved in the ubiquitin-mediated proteasomal degradation of SNAI via GSK3␤. First, we found that GSK3␤ also down-regulates SNAI in CRC cell lines (Fig. 5A) and that inhibition of GSK3␤ activity with pharmacological inhibitors significantly increases SNAI (Fig. 5B). In addition, STAT3 forms a complex with GSK3␤ (Fig. 5C). Because STAT3 deletion or knockdown did not significantly affect the expression of GSK3␤ (Fig. 5C), we examined whether STAT3 regulates activation (or phosphorylation) of GSK3␤ by Phostag SDS-PAGE (13,32). Whereas there was no noticeable difference in GSK3␤ expression level between control and STAT3-deleted cells in a traditional SDS-PAGE analysis, the putative phosphorylated (retarded in mobility) GSK3␤ level was much higher in STAT3-deleted cells compared with that in control cells (Fig. 5D). STAT3 knockdown significantly increased the phosphorylation level on Ser9 residue of GSK3␤ (Fig. 5E). This result indicated that STAT3 negatively affects phosphorylation of GSK3␤ and suggested that STAT3 regu-  lates SNAI expression and tumor invasion by promoting GSK3␤ activity. Therefore, we tested whether a constitutively active GSK3␤ (GSK3␤-CA) mutant can reverse the effects of STAT3 depletion in HCT116 cells. Indeed, expression of GSK3␤-CA reversed the increased SNAI expression and the increased invasiveness induced by STAT3 depletion (Fig. 5F). Taken together, these data provide evidence for a nontranscriptional and inhibitory role of STAT3 in IEC tumor invasion.

DISCUSSION
Although no naturally occurring mutations in STAT3 have been identified as a cause of any human cancer, it is considered as a strong promoter of carcinogenesis. STAT3 is activated in multiple solid and hematologic human malignancies, and its role in carcinogenesis has been validated in animal models (1,33). For example, STAT3 was found to be crucial in initiation and progression of skin carcinoma (34), and the growth and survival of lymphomas are dependent on STAT3 (35). Recently, constitutively activating STAT3 mutations were identified in human hepatocellular adenomas, where the Tyr-640 mutant homodimerizes independently of upstream signals and is hypersensitive to IL-6 stimulation (36). In CRC, STAT3 activation (phosphorylation) was considerably up-regulated during adenoma-to-carcinoma progression (37,38), but the precise role of STAT3 in CRC has not been yet defined.
Our data provide the mechanisms by which STAT3 suppresses tumor invasion both in an animal model and in CRC cell lines. STAT3 in IEC tumors of Apc min/ϩ mice inhibits expression of an EMT inducer SNAI and mesenchymal markers, such as vimentin and fibronectin. Similarly, in CRC cell line FIGURE 5. STAT3 promotes GSK3␤ activity to induce degradation of SNAI. A, GSK3␤ suppresses SNAI expression in CRC lines. The indicated protein levels were measured by immunoblotting 3 days after siRNA transfection. B, GSK3␤ suppresses SNAI via its kinase activity. HCT116 cells were treated with either LiCl (4 mM) or SB216763, GSK3␤ inhibitors as indicated for 4 h, and the indicated proteins were measured by immunoblotting. C, STAT3 interacts with GSK3␤. GSK3␤ in isolated mouse tumor cells or in HCT116 cells was immunoprecipitated and immunoblotted for STAT3 (S3, STAT3; G3, GSK3␤). D, STAT3 regulates phosphorylation levels of GSK3␤. Lysates from HCT116 cells transfected with the indicated siRNA were divided and separated by either traditional SDS-PAGE or Phos-tag SDS-PAGE, and the indicated proteins were measured by immunoblotting. E, STAT3 inhibits phosphorylation at Ser-9 of GSK3␤. HCT116 cells were transfected with the indicated siRNA (C, control; S3, STAT3), total cell lysates were collected 1 or 3 days after transfection, and the levels of indicated proteins were measured by immunoblotting. F, GSK3␤-CA reverses the effects of STAT3 depletion. HCT116 cells were transfected with control siRNA (C), STAT3 siRNA (S3), GSK3␤-CA plasmid, or STAT3 siRNA (S3) plus GSK3␤-CA plasmid. Cell invasion assay (n ϭ 3) and immunoblotting were performed 3 days after transfection.
HCT116, STAT3 suppresses invasiveness by limiting expression of SNAI. Overexpression of a dominant negative STAT3 mutant in a CRC cell line was reported to cause down-regulation of E-cadherin (39). Although SNAI is widely known as a suppressor of E-cadherin, our data and another recent report (4) show that its expression is not significantly affected by STAT3 deletion in IEC. However, overexpression of SNAI did not suppress E-cadherin in a breast cancer cell line, whereas a mutant form (6SA) was able to do so (13). Therefore, it is possible that Ser-6 of SNAI may be phosphorylated in STAT3deficient tumor cells.
Like STAT3, GSK3␤ is also a multifunctional protein involved in a wide range of physiological activities such as metabolism, cell development, and body pattern formation (12). It also plays a key role in Wnt signaling whose dysregulation is the most common cause of human CRC (2,20). Our data show that STAT3 regulates GSK3␤ activity to control SNAI expression. We also found that STAT3 inhibits activation of ␤-catenin in a CRC cell line, but the mechanisms behind this phenomenon are yet to be investigated.
Interestingly, STAT3 in IECs is required for tumor induction in models of colitis-associated cancer. In colitis-associated cancer, STAT3 activation in IEC regulates cell survival and cell cycle progression through its transcriptional activation of downstream targets such as Bcl-X L , c-Myc, and cyclin D1 (40 -42). This colitis-associated cancer model employs a chemical mutagen, azoxymethane, and three cycles of the chemical irritant, dextran sodium sulfate, to induce colorectal tumors. Despite the observation that Stat3 IEC-KO mice are much more susceptible to dextran sodium sulfate-induced colitis (41), it appears that STAT3 in IEC in Apc min model is necessary for their malignant transformation.
Overall, our data document the restraining role of STAT3 on the acquisition of an invasive phenotype of IEC tumors and explain the molecular basis that controls this process. We therefore propose that the inhibition of STAT3 in patients with CRC especially in advanced stages should be practiced with caution.