ABSTRACT
Epithelial cells of the intestinal mucosa undergo a continual process of proliferation, differentiation, and apoptosis which is regulated by multiple signaling pathways. The Wnt/β-catenin pathway plays a critical role in this process. Mutations in the Wnt pathway, however, are associated with colorectal cancers. Krüppel-like factor 4 (KLF4) is an epithelial transcriptional factor that is down-regulated in many colorectal cancers. Here, we show that KLF4 interacts with β-catenin and represses β-catenin-mediated gene expression. Moreover, KLF4 inhibits the axis formation of Xenopus embryos and inhibits xenograft tumor growth in athymic nude mice. Our findings suggest that the cross talk of KLF4 and β-catenin plays a critical role in homeostasis of the normal intestine as well as in tumorigenesis of colorectal cancers.
Wnt/β-catenin signaling plays essential roles in both development and tumorigenesis (12, 26, 46). Wnt signaling is mediated by β-catenin. In the absence of Wnt stimulation, a protein complex consisting of glycogen synthase kinase 3 (GSK-3), casein kinase I alpha, and the tumor suppressor proteins axin and adenomatous polyposis coli (APC) phosphorylates β-catenin (40). The phosphorylated β-catenin is recognized by β-TrCP, a ubiquitin ligase, and degraded by the ubiquitin/proteasome pathway (40). Wnt stimulation results in an inhibition of GSK-3 activity and stabilization of β-catenin. Accumulated β-catenin enters the nucleus and forms a complex with T-cell factor (TCF) and activates TCF target genes, which regulate cell growth and cell fate (5).
Wnt/β-catenin signaling plays important roles in normal intestinal homeostasis and colorectal cancers. The intestine can be divided into crypts, which contain proliferating cells, and villi, which contain differentiated cells. The differentiated cells can be further divided into absorptive cells (enterocytes) and secretory cells (Paneth, goblet, and enteroendocrine cells). Wnt signaling maintains the proliferation of crypt progenitor cells and controls the differentiation of enterocytes, Paneth cells, goblet cells and enteroendocrine cells (1, 15, 27, 28, 30, 41). Disturbance of intestinal homeostasis can lead to a number of human diseases, including colorectal cancers. Mutations in the Wnt/β-catenin signaling pathway have been found in most colorectal cancers (20, 29). These mutations activate Wnt/β-catenin signaling, promote cell proliferation, and inhibit differentiation, ultimately leading to tumor formation (27).
Normal intestinal homeostasis is regulated by multiple signaling pathways that have been intensively studied; however, the cross talk between these pathways is less understood. KLF4, previously known as gut-enriched Krüppel-like factor or epithelial zinc-finger protein, is a member of the Krüppel-like transcription factor family (4). Like all Krüppel-like factors, KLF4 has three zinc-finger domains in its C terminus. Depending on the cellular context, KLF4 expression can lead to either transcriptional activation or repression (11). KLF4 is expressed in the epithelial cells of the gastrointestinal tract, skin, and several other organs (35), and its expression is associated with terminal differentiation (6, 33, 34, 49, 50). KLF4-null mice die within 15 h of birth, presumably due to severe defects in the late-stage differentiation of skin (31). Moreover, KLF4-null mice demonstrate a markedly reduced number of goblet cells in the colon (18).
KLF4 expression has been studied in adenomatous polyps and colon cancers; RNA and protein levels of KLF4 were significantly decreased in the dysplastic epithelium of the colon (33). In the multiple intestinal neoplasm mouse, KLF4 levels inversely correlate with the size of the intestinal adenoma (7). In adenomas removed from patients with familial adenomatous polyposis, KLF4 mRNA levels are significantly lower than in the adjacent normal mucosa (33). Loss of heterozygosity of the KLF4 locus, mutations in the KLF4 open reading frames, and hypermethylation of the KLF4 promoter have been identified in a number of human colorectal cancers and cell lines (51). Overexpression of KLF4 in the colon cancer cell line RKO reduces colony formation, cell migration, and invasion, suggesting that KLF4 is a tumor suppressor protein (8). Recently, similar findings have been noted in gastric cancer; KLF4 is down-regulated in gastric cancer, and KLF4 knockout mice demonstrate defects in gastric differentiation and have precancerous changes in the stomach (17, 45). Here, we report that KLF4 interacts with β-catenin and inhibits β-catenin signaling. We found that KLF4 binds the transcriptional activation domain of β-catenin and inhibits β-catenin-mediated transcription. Our results suggest that the cross talk between KLF4 and β-catenin plays important roles in intestinal homeostasis and colorectal carcinogenesis.
MATERIALS AND METHODS
Cells and plasmids.HEK293T, HeLa, HCT116T, DLD-1, and HT29 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum at 37°C in 5% CO2. LS174T was cultured in RPMI 1640 medium. Cells were transfected by the CaCl2 method or with Lipofectamine (Invitrogen) as previously described (25). The colon cancer cell lines DLD-1-TR7 and LS174T-TR4, which express the Tet repressor, were provided by M. van de Wetering (Hubrecht Laboratory, The Netherlands). Flag-tagged KLF4 and myc-tagged dominant negative TCF (DnTCF) were cloned into the pcDNA4/TO (Tet-on) vector, transfected into DLD-1-TR7 or LS174T-TR4 cells, and selected with zeocin. After several weeks of selection, individual colonies were analyzed by Western blotting. Flag-tagged KLF6 was from Scott L. Friedman at Mount Sinai School of Medicine.
TAP assay.β-Catenin with an N-terminal Flag tag followed by a hemagglutinin (HA) tag was transfected into HeLa cells. Stable cell lines were selected using the method described by Shi et al. (32). Nuclear proteins were isolated from 20 liters of HeLa cells. The tagged β-catenin and its binding proteins were purified by sequential Flag and HA affinity columns and eluted by Flag and HA peptides, respectively. Protein samples were separated by 4 to 20% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and stained by Coomassie blue. Proteins bands specifically present in β-catenin-expressing samples were analyzed by a mass spectrometer in the UTMB Proteomics Core.
Xenopus embryo injection and cap assay.KLF4 cDNA was cloned into pCS2+ vector, and KLF4 mRNA was synthesized using an SP6 in vitro transcription kit (Ambion). Embryo injection, phenotype analysis, and the animal cap assay were performed as previously described (16, 24, 37). The expression of Wnt/β-catenin target genes, Siamois and Xnr3, was analyzed by real-time PCR as described by Xanthos et al. (47). The PCR primers for Xnr3 were 5′-TGAGGCACCATGAAGAGATG-3′ (forward) and 5′-ATGGATCGGCACAACAGATT-3′ (reverse); the PCR primers for Siamois were 5′-CAGACATCTGCCAAGAGCAA-3′ (forward) and 5′-CAGTTTGGGTAGGGCTGTGT-3′ (reverse).
Coimmunoprecipitation and Western blotting.Cells were lysed in 1% Triton lysis buffer (37). Immunoprecipitations were performed as previously described (25). KLF2 and KLF4 antibodies were purchased from Cemines (Golden, CO) and were used for immunoprecipitation. A novel KLF4 antibody generated in this study was used for Western blotting (see Fig. S1 in the supplemental material). Western blotting was performed as previously described (24).
Immunohistochemistry, immunofluorescence, and enzymatic assays.Tissues fixed in 10% formalin or 4% paraformaldehyde were paraffin embedded and sectioned (19). The sections (5 μm) were stained with anti-β-catenin antibody (dilution of 1:100; Upstate), anti-CD44 antibody (dilution of 1:100; Rockland, Gilbertsville, PA), and the KLF4 antibody specially generated in this study. Cells grown on coverslips were fixed for 10 min with 4% paraformaldehyde. The cells were permeabilized with phosphate-buffered saline containing 0.2% (wt/vol) Triton X-100 for 20 min and then blocked by serum-free protein blocking buffer (DakoCytomation, Denmark) for 1 h. Anti-β-catenin antibody (1:600; Sigma, St. Louis, MO) and anti-Flag antibody (1:600; Sigma) were diluted in antibody diluent (DakoCytomation) and incubated with cells overnight. The cells were washed three times with Tris-buffered saline-Tween and then incubated with Alexa 594-labeled anti-mouse immunoglobulin G (1:800) and Alexa 488-labeled anti-rabbit immunoglobulin G (1:800) diluted in Tris-buffered saline-Tween for 1 h. The coverslips were washed, mounted on glass slides, viewed, and photographed with a Zeiss LSM510 confocal microscope (Germany). Intestinal alkaline phosphatase (IAP) activity was detected by an IAP kit from Sigma. Luciferase activity was analyzed by a dual-luciferase reporter assay system (Promega, Madison, WI). Cell proliferation was analyzed by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide] cell proliferation reagent (Sigma).
RNA interference (RNAi).The KLF4 small interfering oligonucleotides were purchased from Dharmacon (Chicago, IL). The oligonucleotides were designed according to a previous study (48) and transfected into cells using Lipofectamine as described (25).
Reverse-transcription PCR (RT-PCR).Total RNA was isolated from cultured cells using an RNeasy minikit (QIAGEN, Valencia, CA). The RNA samples were converted into cDNA using oligo(dT)18 primer (NEB, Beverly, MA) and reverse transcriptase (Roche). The PCR consisted of 28 cycles with a fast-start high-fidelity PCR system (Roche). Amplified products were analyzed on a 1.5% agarose gel. Primers were ordered from Sigma-Genosys (The Woodlands, TX). Primer sequences were as follows: for Axin2, 5′-CACCACCACCACCACCATTC-3′ (forward) and 5′-GCATCCACTGCCAGACATCC-3′ (reverse); for c-myc, 5′-TATGTGGAGCGGCTTCTCG-3′ (forward) and 5′-TCCGCTGTGAGGAGGTTTG-3′ (reverse); for IAP, 5′-CCATTGCCGTACAGGATGGAC-3′ (forward) and 5′-CGCGGCTTCTACCTCTTTGTG-3′ (reverse); and for β-actin, 5′-CAACCGCGAGAAGATGAC-3′ (forward) and 5′-AGGAAGGCTGGAAGAGTG-3′ (reverse).
Tumor xenografts.LS174T-KLF4 or DLD-1-KLF4 cells (2 × 106) were injected subcutaneously into both flanks of athymic nude mice as previously described (10, 43). Drinking water, with or without 200 μg/ml doxycycline, was given to the mice. The tumor growth was analyzed twice weekly. After 3 weeks, the tumor xenografts were harvested and embedded in paraffin. Cross-sections (5 μm) of each tumor were analyzed by histochemical and immunohistochemical methods. Hematoxylin and eosin (HE) staining and periodic acid-Schiff (PAS) staining were performed in the UTMB Histopathology Core. KLF4 and Ki67 immunostaining was performed using the polyclonal KLF4 antibody generated in this study and an MIB-1 monoclonal antibody against Ki67 (Beckman Coulter, Fullerton, CA).
RESULTS
KLF4 interacts with β-catenin in the nucleus.To identify novel β-catenin binding proteins in the nucleus, we generated a stably transfected HeLa cell line expressing β-catenin with N-terminal Flag and HA tags. β-Catenin and its binding proteins were purified from nuclear extracts by HA and Flag affinity chromatography. Proteins were then separated by 4 to 20% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and analyzed by mass spectrometry. An approximately 55-kDa protein was identified to be KLF4 (Fig. 1A). Since KLF4 has important functions in epithelial cell differentiation and gastrointestinal cancer (17, 45, 51), we further characterized the interaction of KLF4 and β-catenin. Flag-tagged KLF4 and myc-tagged β-catenin were transfected into HEK 293T cells. When KLF4 protein was immunoprecipitated with a Flag antibody, β-catenin was also detected in the precipitate. As a control, Flag antibody was not able to precipitate β-catenin in the cells without Flag-tagged KLF4 (Fig. 1B). We analyzed another Krüppel-like factor, KLF6, and did not detect any binding between β-catenin and KLF6, suggesting that β-catenin specifically binds KLF4 (Fig. 1B). To confirm that endogenous KLF4 binds β-catenin, KLF4 was immunoprecipitated from the lysates of mouse intestine, and the presence of β-catenin in the immunoprecipitate was detected by Western blotting. As a control, KLF2 antibody was used, and no precipitation of β-catenin was detected (Fig. 1C).
KLF4 interacts with β-catenin in the nucleus. (A) KLF4 was identified as a component of β-catenin complex. The β-catenin complex was purified by TAP assay from the nuclear extracts of HeLa cells that express Flag-tagged and HA-tagged β-catenin. A 55-kDa protein was identified to be KLF4. (B) KLF4 precipitated β-catenin. β-Catenin (myc tagged) was cotransfected with empty vector or KLF4 (Flag tagged) into HEK 293T cells. Immunoprecipitation (IP) was performed with a Flag antibody. β-Catenin was analyzed with β-catenin antibody, and KLF4 was analyzed with a Flag antibody. As a control, Flag-tagged KLF6 was transfected into HEK 293T cells; KLF6 does not precipitate β-catenin. (C) Endogenous KLF4 precipitated endogenous β-catenin. Immunoprecipitation (IP) was performed with KLF4 antibody and KLF2 antibody (control) from mouse intestinal tissue lysates. β-Catenin was analyzed with β-catenin antibody. KLF4 was analyzed with KLF4 antibody. (D) KLF4 and β-catenin were colocalized in the cell nucleus. KLF4 (Flag tagged) was transfected into HeLa cells. The cells were immunostained with a Flag antibody and a β-catenin antibody. Nucleus was stained with DAPI (4′,6′-diamidino-2-phenylindole). Colocalization (yellow fluorescence) of KLF4 (red fluorescence) and β-catenin (green fluorescence) was detected in the cell nucleus (blue fluorescence).
To further confirm that KLF4 binds β-catenin in the nucleus, Flag-tagged KLF4 was transfected into HeLa cells. The cells were fixed and immunostained with an anti-β-catenin antibody (green fluorescence) and an anti-Flag antibody (red fluorescence for KLF4) (Fig. 1D). β-Catenin was detected on the cell membrane as well as in the nucleus, while KLF4 was only detected in the nucleus. Consistent with our binding results, β-catenin and KLF4 proteins colocalized in the nucleus (yellow fluorescence). Taken together, these data further corroborate our finding that KLF4 binds β-catenin in the nucleus.
KLF4 inhibits Wnt/β-catenin signaling.To study the functional effects of the KLF4/β-catenin interaction, we performed several TCF/β-catenin reporter assays. The TopFlash plasmid contains a luciferase reporter under the control of three copies of the TCF binding element upstream of the thymidine kinase minimal promoter, and it is specifically regulated by Wnt/β-catenin signaling (22). Without Wnt signaling, the TopFlash reporter only had minimal activity in HEK 293T cells; Wnt3A transfection stabilized β-catenin and increased the TopFlash reporter activity (Fig. 2A, top panel). We found that KLF4 inhibited Wnt3A-induced TopFlash reporter activity, suggesting that KLF4 inhibits Wnt signaling. It has been previously reported that KLF4 can decrease β-catenin mRNA levels in the human colon cancer cell line HT29 (36). To examine whether KLF4 inhibits Wnt signaling by regulating β-catenin protein levels, we analyzed cytoplasmic β-catenin levels by Western blotting (Fig. 2A, bottom panel). Whereas Wnt3A stabilized β-catenin, KLF4 had no effect on Wnt3A-induced β-catenin stabilization.
KLF4 inhibits Wnt/β-catenin signaling. (A) KLF4 inhibited Wnt3A-induced TCF/β-catenin reporter activity. Normalized luciferase activities are reported as arbitrary units, with the control (transfected with empty vector) defined as 1 (top). Wnt3A-induced β-catenin stabilization was not inhibited by KLF4 (bottom). (B) KLF4 inhibited β-catenin induced TCF/β-catenin reporter activity. β-Catenin levels were not affected by KLF4 (bottom). (C) KLF4 inhibited β-catenin(S33A)-induced TCF/β-catenin reporter activity. β-Catenin(S33A) protein levels were not affected by KLF4 (bottom) (D) Depletion of KLF4 increased TCF/β-catenin reporter activity in HCT116T cells. KLF4 double-stranded RNAi duplexes decreased KLF4 protein levels (bottom) (E) KLF4 inhibited axis formation in Xenopus embryos. KLF4 mRNA (2 ng) was injected into the dorsal side of 47 embryos. The table shows the dorsal-anterior index numbers of these embryos. (F) Examples of dorsal-injected embryos: wild-type embryo (top) and KLF4-injected embryos (bottom). (G) KLF4 inhibited Wnt/β-catenin-induced axis duplication. Ventral injection of Wnt8 or β-catenin mRNA induced dorsal axis duplication. Coinjection of KLF4 inhibited Wnt8A- or β-catenin-induced a double-axis phenotype. The table shows the numbers of embryos with each phenotype. (H) KLF4 inhibited β-catenin target gene expression in Xenopus embryos. β-catenin target genes, Xnr3 and Siamois, were analyzed by cap assay and real-time PCR. KLF4 inhibited Xnr3 and Siamois expression induced by Wnt8A or β-catenin. Cat, β-catenin; Cat33A, β-catenin(S33A); CS2, pCS2+ vector; C-Ri, control RNAi; KLF4-Ri, KLF4 RNAi.
Next, we tested the effect of KLF4 on wild-type β-catenin and a β-catenin(S33A) mutant (serine 33 changed to alanine), both cloned downstream of a cytomegalovirus promoter. The serine 33 mutation prevents GSK-3 phosphorylation and proteasome degradation of β-catenin (25). We found that KLF4 significantly inhibited TopFlash reporter activity induced by both wild-type β-catenin and β-catenin(S33A) (Fig. 2B and C, top panel) but had no effect on the protein levels of either wild-type or mutant β-catenin (Fig. 2B and C, bottom panel), suggesting that KLF4 inhibits β-catenin function without affecting β-catenin levels. Other Krüppel-like factors, such as KLF5 and KLF6, did not inhibit Wnt/β-catenin signaling in HEK 293T cells (data not shown). To test whether KLF4 inhibits β-catenin in vivo, we performed RNAi experiments in the colon cancer cell line HCT116T. Knockdown of KLF4 by RNAi enhanced Wnt/β-catenin signaling (Fig. 2D), suggesting that KLF4 inhibits β-catenin function in vivo.
Wnt/β-catenin signaling controls Xenopus embryogenesis (46). Dorsal accumulation of β-catenin is required for axis formation (46). The ventral injection of either Wnt or β-catenin mRNA induces dorsal axis duplication (double-axis phenotype) (46). The Xenopus axis formation is a standard in vivo assay for Wnt/β-catenin function (46). To further confirm that KLF4 inhibits β-catenin signaling, KLF4 mRNA was injected into the dorsal side of Xenopus embryos at the 4-cell stage. We found that primary axis formation was inhibited by KLF4 (Fig. 2E and F). The dorsal-anterior index number for embryos injected with KLF4 mRNA was 0 to 1, representing a high degree of ventralization. Ventral injection of Wnt or β-catenin mRNA induced axis duplication in the Xenopus embryos. Coinjection of KLF4 mRNA blocked the Wnt/β-catenin-induced double-axis phenotype (Fig. 2G). To confirm that inhibition of axis formation is specific to Wnt/β-catenin signaling, the expression of two Wnt/β-catenin target genes, Xnr3 and Siamois, was analyzed by real-time PCR (Fig. 2H). Injection of Wnt or β-catenin mRNA induced Xnr3 and Siamois expression in the animal caps of Xenopus embryos. Coinjection of KLF4 mRNA inhibited the expression of Xnr3 and Siamois, suggesting that KLF4 inhibits axis formation by inhibiting Wnt/β-catenin signaling. KLF4 inhibited formation of both the primary axis and the ectopic axis, thus further confirming that KLF4 inhibits Wnt/β-catenin signaling and indicating that KLF4 or its homologue plays an important role in Xenopus embryogenesis. These results demonstrate that the interaction of KLF4 and β-catenin leads to repression of β-catenin-mediated transcription.
KLF4 binds and inhibits the β-catenin transactivating domain.Wnt/β-catenin signaling can be regulated at multiple steps (e.g., regulation of β-catenin protein levels, β-catenin nuclear localization, β-catenin/TCF interaction, or interactions between β-catenin and transcriptional mediator proteins) (12). The above experiments suggested that KLF4 has no effect on β-catenin protein levels or its localization. To determine whether KLF4 prevents binding between β-catenin and TCF, we tested the effects of KLF4 on a lymphoid enhancer factor (LEF)-β-catenin fusion protein (Fig. 3A). LEF is a member of the TCF family, and a LEF-β-catenin fusion protein can activate the TopFlash reporter (42). If KLF4 inhibits β-catenin signaling by altering β-catenin and TCF/LEF binding, KLF4 should not inhibit the LEF-β-catenin fusion protein. However, we found that KLF4 inhibits the TopFlash activity induced by the LEF-β-catenin fusion protein, indicating that KLF4 inhibits the function, but not the formation, of the β-catenin/LEFcomplex (Fig. 3B). An LEF-VP16 fusion protein, which strongly activates the TopFlash reporter (42), was used as a control. KLF4 has no significant effect on the LEF-VP16 fusion protein, suggesting that KLF4 inhibits TopFlash reporter through β-catenin. β-Catenin has several functional domains; the N-terminal S/T domain controls β-catenin stability, the central armadillo domains interact with TCF/LEF and E-cadherin, and the C-terminal domain interacts with the transcriptional mediator complex and activates gene transcription (13). A fusion protein of LEF and the C-terminal domain of β-catenin can efficiently activate the TopFlash reporter. We found that this fusion protein was also inhibited by KLF4, suggesting that KLF4 can suppress the function of the β-catenin C-terminal transactivating domain (Fig. 3B). We tested the binding between KLF4 and the β-catenin C terminus by immunoprecipitation and found that KLF4 indeed binds the β-catenin C terminus (Fig. 3C).
KLF4 binds and inhibits the β-catenin transactivating domain. (A) Schematic representation of KLF4, β-catenin, their derivatives, and LEF fusion proteins. LEF-CAT is a fusion protein of LEFΔN (lacking the N-terminal β-catenin binding domain) and full-length β-catenin. LEF-CAT-C is a fusion protein of LEFΔN and the β-catenin transactivating domain. LEF-VP16 is a fusion protein of LEFΔN and the VP16 transactivating domain. (B) KLF4 inhibited LEF-CAT- and LEF-CAT-C-induced TCF/β-catenin reporter activity but not LEF-VP16-induced TCF/β-catenin reporter activity. (C) KLF4 interacted with the β-catenin C terminus. β-catenin C terminus (myc tagged) was cotransfected with empty vector or KLF4 (Flag-tagged) into HEK 293 cells. Immunoprecipitation (IP) was performed with a Flag antibody. The β-catenin C terminus was analyzed with a myc antibody (MT-Cat-C). KLF4 was analyzed with a Flag antibody. (D) KLF4ΔC interacted with the β-catenin C terminus. KLF4ΔC (Flag tagged) was cotransfected with Cat-N (myc tagged) or Cat-C (myc tagged). Immunoprecipitation (IP) was performed with a Flag antibody. Cat-N and Cat-C were analyzed with a myc antibody. KLF4ΔC was analyzed with a Flag antibody. (E) Wild-type KLF4 but not KLF4ΔC inhibited β-catenin(S33A)-induced TCF/β-catenin reporter activity. (F) KLF4 inhibits and KLF4ΔC activates TCF/β-catenin reporter activity. Cat, β-catenin; Cat33A, β-catenin(S33A); IB, immunoblot; CS2, pCS2+ vector.
To determine whether the zinc finger domains of KLF4 are required for β-catenin binding, we deleted the C-terminal zinc finger domains from KLF4. We found that a KLF4 construct lacking any zinc finger domains (KLF4ΔC) was sufficient for β-catenin binding and that KLF4ΔC specifically binds the β-catenin C terminus, which contains the transactivation domain (Fig. 3D). Although KLF4ΔC contains the β-catenin binding domain, KLF4ΔC did not inhibit β-catenin-induced TopFlash reporter activity (Fig. 3E). Interestingly, KLF4ΔC further increased TopFlash reporter activity (Fig. 3E and F), suggesting that KLF4ΔC is a dominant negative mutant of KLF4 and that the KLF4 C terminus is required for β-catenin inhibition.
KLF4 function in colon cancer cells.Since β-catenin is an oncogene and KLF4 is a potential tumor suppressor gene for colorectal cancer (51), we analyzed the functional correlation of KLF4 and β-catenin in colon cancer cells. Wnt/β-catenin signaling can be blocked by DnTCF. The effects of DnTCF have been well studied in colon cancer cells (39). Both KLF4 and DnTCF inhibit Wnt/β-catenin signaling. Therefore, to determine the functional correlation of KLF4 and DnTCF in the intestinal cells, we generated DLD-1 stable cell lines that express doxycycline-induced KLF4 or DnTCF using a Tet-on system (39) (Fig. 4A). KLF4 and DnTCF have similar effects on cell growth. Induced expression of KLF4 or DnTCF inhibits cell proliferation (Fig. 4B and C). The expression of Wnt/β-catenin target genes, c-myc and Axin2, was analyzed by semiquantitative RT-PCR. As expected, DnTCF decreased the mRNA levels of both target genes. Importantly, we also found that KLF4 inhibits both Axin2 and c-myc expression (Fig. 4D). As a control, actin mRNA levels were not affected by KLF4 or DnTCF expression. IAP is a target gene of KLF4 (14). We found that IAP expression was up-regulated by either KLF4 or DnTCF (Fig. 4D). KLF4 and DnTCF also increased the enzymatic activity of IAP (Fig. 4E). Since IAP is a differentiation marker for intestinal cells (44), these results suggest that both KLF4 and DnTCF can induce intestinal cell differentiation. Collectively, the above results suggest that KLF4 and β-catenin play antagonistic roles in regulating intestinal cell proliferation and differentiation, further confirming that KLF4 inhibits Wnt/β-catenin signaling in intestinal cells.
KLF4 functions in colon cancer cells. (A) Stable cell lines that express inducible KLF4 and DnTCF. KLF4 (Flag tagged) and DnTCF (myc tagged) were transfected into DLD-1-TR7 cells. Stable cell lines were analyzed by Western blotting. The expression of KLF4 and DnTCF was induced by doxycycline. (B) Induced expression of KLF4 or DnTCF4 in DLD-1 cells inhibited cell proliferation. The cells were stained by crystal violet after 5 days of induction by doxycycline. (C) MTT cell proliferation assay. Induced KLF4 expression inhibits DLD-1 cell proliferation. (D) KLF4 inhibited β-catenin target genes. Axin2 and c-myc are β-catenin target genes. IAP is a differentiation marker and is a target gene of KLF4. The mRNA levels of these genes were analyzed by semiquantitative RT-PCR; β-actin mRNA levels were analyzed as a control. (E) IAP enzymatic activities in the DLD-1 cells with or without doxycycline were analyzed. IAP activity was induced by KLF4 expression or DnTCF expression. Dox, doxycycline; IB, immunoblot.
KLF4 inhibits the tumorigenic activity of colon cancer cells that contain APC or β-catenin mutations.Colorectal cancers are initiated by APC or β-catenin mutations which cause abnormal activation of Wnt/β-catenin signaling. Since KLF4 inhibits Wnt/β-catenin signaling and colon cancer cell proliferation, KLF4 may inhibit β-catenin-mediated tumorigenesis. It has been reported that KLF4 inhibits tumorigenesis of RKO colon cancer cells, which have both wild-type APC and β-catenin (8). Since most colon cancer cells have either an APC mutation or a β-catenin mutation, it is necessary to determine whether KLF4 inhibits tumorigenesis of APC- or β-catenin-mutated colon cancer cells. Whereas APC is mutated in DLD-1 cells and β-catenin is mutated in LS174T cells, we generated two stable cell lines (LS174T-KLF4 and DLD-1-KLF4), using the Tet-on system (39), that express doxycycline-inducible Flag-tagged KLF4 (Fig. 5A). LS174T-KLF4 and DLD-1-KLF4 cells were injected subcutaneously into the flanks of athymic nude mice given drinking water with or without doxycycline. Tumor growth was measured twice weekly. Both LS174T-KLF4 and DLD-1-KLF4 xenograft tumors were significantly smaller in mice treated with doxycycline, suggesting that KLF4 inhibits tumor growth (Fig. 5B).
KLF4 inhibits xenograft tumor growth. LS174T-KLF4 or DLD-1-KLF4 cells (2 × 106) were injected subcutaneously into the flanks of athymic nude mice. Drinking water with or without 200 μg/ml doxycycline was given to the mice. The tumor volume was measured two times per week. A portion of each tumor was removed and stored in liquid nitrogen. Protein expression was analyzed by Western blotting. Each tumor was fixed and embedded in paraffin. Tumor sections were analyzed by histochemical and immunohistochemical methods. (A) LS174T-KLF4 and DLD-1-KLF4 cell lines. The KLF4 expression was detected by anti-Flag antibody. Flag-tagged KLF4 was only expressed in the doxycycline-treated cells. (B) Representative nude mice photographed 3 weeks after injection showing that the xenograft tumors treated with doxycycline were significantly smaller than the xenograft tumors without doxycycline treatment. (C) KLF4 expression in xenograft tumors. A portion of each tumor was homogenized and analyzed by Western blotting with anti-KLF4 Ab. (D) Quantification of tumor volume. The average volume for four xenograft tumors from each cell line with or without doxycycline was determined. (E) Immunohistochemical studies of xenograft tumors. The tumors were fixed and paraffin embedded and sectioned at 5 μm. KLF4 expression was analyzed by immunostaining with the anti-KLF4 antibody generated in this study. Cell proliferation marker Ki67 was analyzed with an anti-Ki67 antibody, MIB-1. (F). Histochemical studies of xenograft tumors. Tumor sections were analyzed by HE staining and PAS staining. Dox, doxycycline.
Since the C-terminal zinc finger domains are conserved among the Krüppel-like factor family (4), we generated a novel polyclonal antibody against the N terminus of KLF4 (see Fig. S1 in the supplemental material). Using this antibody, KLF4 expression in the tumors was analyzed by Western blotting (Fig. 5C). The average tumor volume, with or without doxycycline, was measured. Tumors that expressed KLF4 grew significantly more slowly than tumors that did not express KLF4 (Fig. 5D). Two of the doxycycline-treated mice demonstrated no tumor growth (these two mice were excluded from further analysis). Doxycycline did not inhibit the growth of control xenograft tumors (see Fig. S2 in the supplemental material).
Cross-sections of these tumors were analyzed by immunohistochemistry (Fig. 5E). KLF4 expression was detected in the nucleus of doxycycline-treated tumors, thus confirming the Western blotting results that doxycycline induced KLF4 expression in these tumors. Expression of Ki67, a cell proliferation marker, was also analyzed in these cross-sections. Tumors expressing KLF4 showed less Ki67 staining compared with tumors without KLF4 expression, suggesting that KLF4 expression inhibits tumor proliferation. The tumor sections were also analyzed with HE and PAS staining (Fig. 5F). HE staining was not significantly different among these tumors. However, in LS174T xenograft tumors, the PAS staining in KLF4-expressing tumors was much stronger than in tumors without KLF4 expression. Since PAS stains mucins, these findings indicate that KLF4 increased mucin production in these tumor cells.
KLF4 expression in the normal human intestine and colon cancers.To further study the correlation between KLF4 and Wnt/β-catenin signaling in the intestine, we examined the expression of KLF4 in normal human intestine and colon cancers. In the normal human colon mucosa, KLF4 expression was noted in the most differentiated portion of the crypts (i.e., apex) and was decreased in the proliferative portion (i.e., bottom of the crypts) (Fig. 6A and B). In normal human small bowel tissue, KLF4 expression gradually increased from the crypt base to the villus and was strongly expressed in the differentiated villus epithelial cells (Fig. 6C). In contrast, the β-catenin target gene, CD44, was strongly expressed in the crypts, and this expression was decreased as cells extended the crypt-villus axis (Fig. 6D), indicating that the interaction of KLF4 and β-catenin contributes to the expression pattern of β-catenin target genes. As the cells migrate up the crypt-villus axis, Wnt/β-catenin signaling gradually decreases (39), and KLF4 expression increases; our data suggest that KLF4 contributes to the inhibition of Wnt/β-catenin signaling and regulates the smooth transition between proliferation and differentiation zones along the crypt-villus axis. In colon cancers, β-catenin levels were increased compared with the adjacent normal mucosa (Fig. 6E). In contrast, KLF4 levels were decreased in a majority cancers compared with normal intestinal epithelial cells (Fig. 6F). These results support the role of KLF4 as a potential tumor suppressor gene in colorectal cancer (51) and suggest that the inhibition of β-catenin contributes to the tumor suppressor function of KLF4.
KLF4 expression in normal human intestine and colon cancers. (A) KLF4 expression was detected predominantly in the apex of the crypts in normal human colon tissue and was noted to be decreased toward the bottom of the crypts. (B) Magnified (×100) picture of nuclear KLF4 in the apex of colon (the magnification of the other panels is ×40). (C) KLF4 was strongly expressed in the villi of the small bowel with decreased expression noted in the crypts. (D) CD44 is a target gene of β-catenin. CD44 protein was detected in the crypts of the small bowel. (E) β-Catenin expression in human colon cancers. β-catenin levels were increased in tumor tissues. (F) KLF4 levels were decreased in human colon cancer tissues.
DISCUSSION
β-Catenin is encoded by an oncogene and KLF4 is encoded by a potential tumor suppressor gene (8, 17, 29, 51). In our current study, we provide compelling evidence that KLF4 binds β-catenin and inhibits β-catenin function. Furthermore, KLF4 inhibited the tumorigenic activity of APC or β-catenin mutant cells. Our findings identify KLF4 as a novel antagonist of β-catenin in the nucleus and demonstrate a cross talk mechanism between KLF4 and β-catenin in the normal intestine and in colorectal cancer.
Depending on the cellular context, KLF4 can act as either a transcriptional activator or repressor (11). KLF4 represses β-catenin transcription in colon cancer HT29 cells (36). However, in our experiments KLF4 inhibited β-catenin signaling but did not affect β-catenin protein levels. Furthermore, KLF4 inhibited a fusion protein of β-catenin and LEF, suggesting that KLF4 inhibits the function of the β-catenin/LEF complex but not the formation of the complex. We found that KLF4 specifically interacted with and inhibited the β-catenin C terminus, which contains the transcriptional activation domain. Since the β-catenin C terminus interacts with transcriptional mediators such as p300/CBP (13) or Brg-1 (2), KLF4 may inhibit β-catenin function by preventing β-catenin binding to transcription mediators. For example, another β-catenin binding protein, ICAT, binds the β-catenin C terminus and inhibits β-catenin/p300 binding (9). It is also possible that KLF4 recruits other transcriptional repressors to the β-catenin/TCF complex and inhibits transcription. We are currently investigating these possible mechanisms.
The function of KLF4 in intestinal development is not entirely understood, due to the early lethality of mice with homozygous KLF4 mutation. The KLF−/− mice have normal colonocytes and enteroendocrine cells but demonstrate a significant decrease in the number of goblet cells in the colon, suggesting that KLF4 regulates goblet cell differentiation (18). We found that KLF4 induced mucin production in LS174T xenograft tumors. LS174T cells are goblet-like cells (23). This result supports the idea that KLF4 regulates goblet cell differentiation. The role of the Wnt/β-catenin signaling pathway in goblet cell differentiation is not entirely known. Inhibition of Wnt/β-catenin signaling by β-catenin RNAi increased mucin production in LS174T cells (38). Activation of Wnt signaling by selective APC knockout leads to goblet cell depletion (30). These data suggest that inhibition of Wnt/β-catenin signaling contributes to goblet cell differentiation. However, overexpression of the Wnt antagonist, DKK1, also reduces goblet cell numbers (28). Since depletion of intestinal stem cells results in depletion of the differentiated cells of intestinal mucosa including goblet cells (15, 21) and inhibition of goblet cell differentiation also leads to decreased goblet cell number, it will be necessary to generate intestinal-specific knockout mice to fully understand the in vivo function of KLF4.
Nuclear β-catenin is gradually decreased (39), and KLF4 expression is increased along the crypt-villus axis. Similarly, ephrin-B1 is expressed in the villi and its receptor EphB2/B3 is expressed in the crypts. The interaction of ephrin-B1 and EphB2/B3 regulates the formation of the crypt-villus boundary and restricts cell intermingling (3). The interaction of KLF4 and β-catenin may have a similar physiological role but acts within the nucleus. This expression pattern suggests that KLF4 contributes to the maintenance of intestinal homeostasis by preventing abnormal activation of β-catenin signaling in the differentiated cells and, thus, inhibiting tumor formation.
The function of KLF4 is cell type specific. KLF4 expression is down-regulated in colon and gastric cancers and is considered a tumor suppressor for these cancers. In contrast, KLF4 is up-regulated in breast and pancreatic cancers. In these contexts, it is considered as an oncogene. The mechanisms regulating KLF4 function in different cell types are not known. A further assessment of the cross talk between KLF4 and β-catenin will improve our understanding of tumor formation in different tissues.
In summary, our results suggest that the cross talk of KLF4 and β-catenin plays a critical role in normal intestinal development and colorectal cancers (Fig. 7). In the normal intestine, β-catenin regulates cell proliferation in the crypts and inhibits enterocyte differentiation. KLF4 is expressed predominantly in differentiated epithelial cells, with decreased expression noted in the crypts. Our findings suggest that KLF4 inhibits Wnt/β-catenin signaling and induces cell cycle arrest and differentiation. Mutations of APC and β-catenin in the intestine constitutively activate β-catenin target genes which promote cell proliferation and inhibit cell differentiation, ultimately leading to cancer. KLF4 inhibits β-catenin-mediated gene expression and thus inhibits the increased tumorigenesis noted by APC or β-catenin mutation. Down-regulation of KLF4 in the intestine enhances β-catenin-mediated gene expression. Our findings suggest that genetic events that both activate β-catenin and inactivate KLF4 are required for colorectal carcinogenesis. Since KLF4 inhibits both mutant β-catenin and β-catenin in APC mutant cells, small molecules which mimic the function of KLF4 could be useful as therapeutic agents for colorectal cancers.
Model for β-catenin/KLF4 cross talk in the intestine.
ACKNOWLEDGMENTS
We thank Hans Clevers, Mark van de Wetering, Scott L. Friedman, and Bert Vogelstein for stable cell lines and plasmids; Mark Hellmich and Qingding Wang for helpful discussions; and Xiaofu Wang, Zhaoyong (Daniel) Hu, and Hui-Qun Wang for technical assistance.
The tandem affinity purification of β-catenin complex was initiated by C.L. in the laboratory of X.H., who is supported by R01 GM057603 and R01 GM074241 from the National Institutes of Health (NIH); X.H. is a W. M. Keck Foundation Distinguished Young Scholar. V.W.Y. is supported by R01 DK52230 and R01 CA84197 from NIH. B.M.E. is supported by R01 DK48498, R01 CA104748, and P01 DK35608 from the NIH. C.L. is supported by a John Sealy Memorial Fund Recruitment Award and by R21 CA112007 from the NIH.
FOOTNOTES
- Received 29 September 2005.
- Returned for modification 28 November 2005.
- Accepted 22 December 2005.
- Copyright © 2006 American Society for Microbiology