PKG inhibits TCF signaling in colon cancer cells by blocking β-catenin expression and activating FOXO4
Introduction
Guanylin/uroguanylin are endogenous peptides that regulate electrolyte homeostasis and control differentiation along the crypt-villus axis in the intestine (Forte, 2004; Pitari et al., 2001). These ligands bind to receptor guanylyl-cyclase (GC-C) on the intestinal epithelium causing increased cGMP levels and can suppress tumor burden in the Apc mouse cancer model (Shailubhai et al., 2000; Uzzau & Fasano, 2000). The nature of the tumor suppressive properties of uroguanylin is presently not known, but as central mediators of cGMP signaling in cells, the cyclic-GMP-dependent protein kinases (PKG) are likely mediators (reviewed in Francis & Corbin, 1999; Lincoln et al., 2001; Ruth, 1999). Type 2 PKG phosphorylates ion exchangers and channels in the intestinal epithelium to carryout the natriuretic functions of uroguanylin (Forte et al., 2000). The function of type 1 PKG in epithelial cells is not known, but its expression is reduced in colon tumors and cell lines relative to matched normal tissue (Browning, 2008). Evidence for tumor suppressive functions of PKG in colon cancer cells is derived from ectopically expressed PKG, which can increase apoptosis, and reduce tumor growth and angiogenesis in xenografts (Deguchi et al., 2004; Hou et al., 2006a; Hou et al., 2006b; Kwon et al., 2008). Further evidence for growth-inhibitory effects of cGMP comes from studies with exisulind, which increases cGMP levels by inhibiting phosphodiesterases (Deguchi et al., 2004; Liu et al., 2001; Zhu et al., 2005). This drug can induce apoptosis in colon cancer cells in a type 1 PKG-dependent manner (Goluboff, 2001; Haanen, 2001; Liu et al., 2001). The signal transduction pathways involved in the antitumor effects of PKG are not fully understood, but activation of cJun-N-terminal Kinase (JNK) (Soh et al., 2000), downregulation of β-catenin (Liu et al., 2001; Thompson et al., 2000), and more recently activation of SP1 (Cen et al., 2008) have all been implicated as important.
The regulation of β-catenin levels by PKG is particularly significant because aberrant overexpression of this protein is a hallmark of intestinal tumorigenesis (Giles et al., 2003) and also some breast and bone marrow derived cancers (Clevers, 2006; Lin et al., 2000; Schlange et al., 2007). Elevations in β-catenin expression promote interaction with TCF/LEF transcription factors to activate growth-related target genes such as c-Myc, cyclin D1, and c-Jun (Behrens, 2000; Lustig & Behrens, 2003). In most non-cancer cells the β-catenin levels are minimal because it associates with the adenomatous polyposis coli (APC) complex leading to phosphorylation by glycogen synthase kinase 3β (GSK-3β) and subsequent ubiquitination and degradation in proteasomes (Behrens, 1999; Behrens, 2000). Many colorectal tumors have truncating mutations in APC that render it unable to bind β-catenin, or in the phosphorylation sites of β-catenin itself, which leads to excessive levels (Sparks et al., 1998; Strate & Syngal, 2005). It has been suggested that PKG has a similar role to GSK3β and directly phosphorylates β-catenin/TCF leading to proteasomal degradation (Liu et al., 2001; Thompson et al., 2000). However, in addition to the canonical pathway involving degradation of β-catenin, TCF signaling is subject to strict regulation by diverse factors that interfere with Wnt signaling or with the interaction of β-catenin with TCF in the nucleus (reviewed recently (Jin et al., 2008). Evidence is accumulating that oxidative stress can inhibit of TCF signaling by activating forkhead box O (FOXO) transcription factors, which compete with TCF for β-catenin (Hoogeboom & Burgering, 2009). In cancer cells, Akt phosphorylates FOXO, which leads to interactions with 14-3-3 that sequesters them in the cytosol (Burgering, 2008; Huang & Tindall, 2007). However, in response to oxidative stress, phosphorylation by JNK causes dissociation from 14-3-3 and FOXO activation (Essers et al., 2004). The present study has examined the mechanism of β-catenin/TCF downregulation by PKG in colon cancer cells and has identified two pathways. PKG was found to reduce the steady-state levels of β-catenin protein by inhibiting transcription of the CTNNB1 gene and not by stimulating protein degradation. In addition, we show that PKG can activate FOXO4 in colon cancer cells, and this pathway inhibits TCF-dependent transcription by recruiting β-catenin to FOXO4. These results support previous work highlighting the anti-tumor properties of cGMP signaling in colon cancer cells and for the first time show activation FOXO by PKG.
Results
Downregulation of β-catenin by PKG does not involve increased protein turnover
The tumor suppressive properties of PKG have been attributed in part to its ability to block signaling through β-catenin/TCF and the present study addresses the mechanism. Activation of PKG was first reported to reduce β-catenin levels in SW480 and HT29 cells grown in vitro (Liu et al., 2001; Thompson et al., 2000). More recently, reduced β-catenin levels were observed in SW620 xenografts with inducible PKG expression (Kwon et al., 2008). In support of previous studies with SW480 cells, activating PKG in the inducible SW620 cells grown in vitro resulted in a dose-dependent inhibition of β-catenin protein on Western blots (Fig. 1a). It has been suggested that PKG can phosphorylate β-catenin directly, leading to increased proteasomal degradation in a manner mimicking the effect of GSK3β (Li et al., 2002; Thompson et al., 2000). To test this idea we used phospho-specific antibodies specific to the amino-terminal Ser33/37/Thr41 residues that are involved in targeting β-catenin for ubiquitinylation (Fig. 1b). The constitutive phosphorylation of β-catenin was readily observed in HEK293 cells when treated with the phosphatase inhibitor calyculin, and it was blocked by pretreatment with lithium, indicating the involvement of GSK3β. However, we were unable to detect any Ser33/37/Thr41 phosphorylation in SW620 cells overexpressing PKG. In support of this result, analysis of the β-catenin sequence using a phosphorylation target database (Blom et al., 2004) revealed that the amino terminal GSK3β target sites are likely to be poor substrates for PKG, but Ser552 and Ser675 were identified as potential PKG targets. Indeed, we found that 8Br-cGMP could increase phosphorylation of these sites in PKG-expressing SW620 colon cancer cells (data not shown). To examine the possibility that an unconventional route to degradation was involved, the effect of PKG on β-catenin protein levels was measured in the presence of cycloheximide (Fig. 1c). These experiments revealed a nearly linear decrease in β-catenin protein levels over time, with approximately 30% remaining after 20 h in both parental cells and those expressing PKG. Treatment with 8-Br-cGMP did not alter the rate of protein degradation in the parental SW620 cells (3.1 +/−0.7 % per hour), but slightly increased stability in the PKG-expressing cells (3.2 +/−0.8% to 2.9 +/−0.1% per hour). Although this stabilization effect of cGMP did not attain statistical significance in our studies, it demonstrates that PKG signaling does not increase β-catenin degradation in SW620 colon cancer cells.
PKG inhibits transcription from the CTNNB1 gene in colon cancer cells
The inability of PKG to induce β-catenin degradation indicated a mechanism different from GSK3β and underscored the possibility that PKG could affect expression at the transcriptional or translational levels. As a first step, the effect of PKG activation on the steady-state levels of β-catenin mRNA was measured using semi-quantitative RT-PCR (Fig. 2a). Activation of PKG in several inducible clones of SW620 resulted in a dramatic reduction of β-catenin message, whereas treatment of parental SW620 cells with inducer and 8Br-cGMP had no effect. In order to determine whether the reduced β-catenin mRNA levels were due to PKG-dependent regulation of transcription, a reporter construct previously shown to encode the essential promoter regions of the β-catenin gene (CTNNB1) was used (Li et al., 2004). Transient transfection studies with the CTNNB1-luciferase reporter demonstrated that PKG activation in SW620 cells inhibited basal transcription by approximately 60%, whereas the PKG activator 8Br-cGMP or PKG induction alone has little effect (Fig. 2b). Suppression of CTNNB1 activation by PKG was not unique to the SW620 cells, since transient transfection of PKG could also inhibit CTNNB1 transcription in several other colon cancer cell lines (Fig. 2c). Although the degree of inhibition varied between the different cells, the most dramatic reduction in CTNNB1 transcription was observed in SW480 cells (80%).
Inhibition of TCF transcription by PKG requires activation of JNK
The inhibition of CTNNB1 activation by PKG is the most likely cause of the reduced β-catenin expression in colon cancer cells. Downregulation of β-catenin expression by PKG has been shown to block transcription of TCF-dependent target genes such as cyclin D1 in SW480 cells (Deguchi et al., 2004; Thompson et al., 2000). In our inducible SW620 colon cancer cells, activation of PKG inhibited TCF-activity by 65% as measured using the TOP-flash TCF-luciferase reporter system (Fig. 3a). In these experiments PKG overexpression by itself, or treatment with 8Br-cGMP alone did not affect the basal TCF activity or the β-catenin protein levels, which underscores the requirement for active PKG. We also examined the inhibition kinetics of both β-catenin protein expression and TCF activity by PKG in SW620 cells. In these experiments approximately 60% of the total TCF inhibition (measured at 24 h) was observed after 6 h stimulation with 8Br-cGMP, at a time when there was negligible effect on β-catenin protein (Fig. 3b). This apparent disconnect between the downregulation of β-catenin protein and TCF activity was also observed in other colon cancer cells, including SW480, HCT116 and HT29. Cell lines were created with doxycycline-inducible PKG expression, and similar to SW620 cells, activation of PKG resulted in reduced β-catenin protein and TCF activity (Fig. 3c). The dramatic inhibition of TCF activity in SW480 and HCT116 cells (more than 80%) was not paralleled by the comparatively small effect of PKG on the β-catenin protein levels. Taken together, these results indicate that mechanisms in addition to the regulation of β-catenin expression, must contribute to the inhibition of TCF signaling by PKG.
The regulation of β-catenin/TCF signaling can be complex as there are families of diffusible Wnt antagonists as well as diverse endogenous inhibitors of the β-catenin/TCF interaction (reviewed recently in Jin et al., 2008). To identify additional pathways that contribute to TCF-inhibition by PKG, we first ruled out paracrine mechanisms by growing cells on trans-well filters (data not shown). It was then determined that PKG did not significantly block the nuclear localization of β-catenin, or increase the expression of endogenous TCF inhibitors (data not shown). These observations led us to consider the involvement of FOXO transcription factors, which have been reported to inhibit TCF activity in response to oxidative stress (Almeida et al., 2007; Essers et al., 2005). FOXO regulation was particularly attractive because activation of FOXO4 by oxidative stress requires JNK (Essers et al., 2004), which has previously been shown to be activated by PKG in colon cancer cells (Soh et al., 2000). Using phospho-JNK antibodies, we first confirmed that PKG can activate JNK (Fig. 4a), and then sought to verify that oxidative stress could inhibit TCF activity in our cells. In transiently transfected SW480 cells, stimulation with either 100 µM 8Br-cGMP or 200 µM H2O2 for 6 h resulted in 50% inhibition of the basal TCF activity (Fig. 4b). Cotransfection of the cells with a dominant negative JNK (JNK-DN) construct blocked the TCF-inhibition by PKG (23%) and H2O2 (10%). These data indicate that PKG and H2O2 use a common mechanism to inhibit TCF activity in colon cancer cells, and this pathway requires activation of JNK. The ability of the JNK-DN to only partially block TCF-inhibition by PKG (56% rescue) compared to H2O2 (90% rescue), supports the idea that downregulation of β-catenin expression also plays a role downstream of PKG. The inhibition of TCF activity by H2O2 was blocked by co-incubation of the cells with the antioxidant N-acetylcysteine (NAC) (Fig. 4c). In contrast, NAC did not affect the ability of PKG to inhibit TCF activity, which indicates that JNK activation by PKG is not due to a redox effect. This is consistent with the idea that PKG can activate JNK by another pathway such as MEKK phosphorylation as has been reported previously (Soh et al., 2001).
PKG activates FOXO4 in colon cancer cells
The FOXO transcription factors are a recently recognized family of tumor suppressors that are gaining significant interest as targets for cancer therapy (Arden, 2006; Dansen & Burgering, 2008; Maiese et al., 2008; Weidinger et al., 2008). These proteins are generally inactive in cancer cells owing to constitutive Akt, which promotes interaction with 14-3-3 proteins and cytosolic localization (Dansen & Burgering, 2008; Fu & Tindall, 2008; Maiese et al., 2008). Our studies focused on FOXO4 because this isoform has unique JNK phosphorylation sites that are involved in H2O2 induced FOXO activation in colon cancer cells (Essers et al., 2004). Transient transfection of a flag-FOXO4 construct into colon cancer cells showed both cytosolic and nuclear localization of the protein by immunofluorescence (Fig. 5a). However, in cells cotransfected to express FOXO4 with PKG, treatment with 8Br-cGMP caused a striking mobilization of the cytosolic fraction to the nucleus. FOXO can activate several classes of genes, including those involved in cell cycle arrest (p27), DNA repair (GADD45), and antioxidant pathways (catalase, MnSOD) (Burgering, 2008; Calnan & Brunet, 2008; Ho et al., 2008; Huang & Tindall, 2007). Activation of PKG in SW620 colon cancer cells caused increased expression of the classical FOXO target genes encoding catalase and MnSOD (Fig. 5b, c), but in these studies PKG did not activate other target genes such as p27 or GADD45 (not shown). These data are evidence that PKG can activate FOXO4 in colon cancer cells.
Activation of FOXO4 is required for the inhibition of TCF by PKG
FOXO proteins regulate gene expression not only by activating FOXO target genes, but also by interacting with other transcription factors (reviewed in (van der Vos & Coffer, 2008). Several groups have shown that FOXO binds strongly to β-catenin, which increases the transcription of FOXO target genes but inhibits TCF-transcription by diverting β-catenin (Almeida et al., 2007; Essers et al., 2005; Hoogeboom et al., 2008). Results shown here demonstrate activation of FOXO4 in colon cancer cells but the importance of this pathway to the inhibition TCF activity by PKG required further study. Immunoprecipitation experiments demonstrated that transient expression of flag-FOXO4 was able to bind to endogenous β-catenin under control conditions, but activation of cotransfected PKG dramatically increased the level of β-catenin bound (Fig. 6a). In keeping with the importance of JNK downstream of PKG, cotransfection of the JNK-DN completely blocked the PKG-dependent increase in β-catenin bound to FOXO4. These results strengthen the notion that PKG can activate FOXO in colon cancer cells and supports previous work by other groups that have reported that activated FOXO binds to β-catenin. We found that overexpression of FOXO4 could inhibit basal TCF activity with similar effectiveness as PKG (data not shown). In order to determine whether FOXO4 activation is involved in TCF-inhibition by PKG, we generated siRNA to knockdown FOXO4 expression. Compared to a non-targeting siRNA, the FOXO4 siRNA-1 generated from position 829 of the coding region had better than 90% knockdown of transiently expressed FOXO4, whereas siRNA-2 (position 1267) was less efficient with close to 70% knockdown (Figure 6b). The effect of silencing FOXO4 in the SW480 colon cancer cells on the ability of PKG to inhibit TCF transcription was examined using TCF-luciferase assays subsequent to siRNA transfection (Fig. 6c). Results showed that the more efficient siRNA-1 completely blocked the effect of PKG and actually raised the basal TCF-activity slightly higher than control. The siRNA-2 was also able to block the inhibitory effect of PKG on TCF activity from 60% inhibition to only 20% inhibition. These data demonstrate that FOXO4 is necessary for PKG to inhibit TCF activity in colon cancer cells. Taken together our results outline a model in which PKG activity can inhibit TCF activity by two mechanisms (Fig. 6d). One effect is to block transcription from the CTNNB1 gene, resulting in reduced total β-catenin levels. Another mechanism is to activate FOXO4, which by binding to β-catenin, limits the pool available to interact with TCF.
Downregulation of β-catenin by PKG does not involve increased protein turnover
The tumor suppressive properties of PKG have been attributed in part to its ability to block signaling through β-catenin/TCF and the present study addresses the mechanism. Activation of PKG was first reported to reduce β-catenin levels in SW480 and HT29 cells grown in vitro (Liu et al., 2001; Thompson et al., 2000). More recently, reduced β-catenin levels were observed in SW620 xenografts with inducible PKG expression (Kwon et al., 2008). In support of previous studies with SW480 cells, activating PKG in the inducible SW620 cells grown in vitro resulted in a dose-dependent inhibition of β-catenin protein on Western blots (Fig. 1a). It has been suggested that PKG can phosphorylate β-catenin directly, leading to increased proteasomal degradation in a manner mimicking the effect of GSK3β (Li et al., 2002; Thompson et al., 2000). To test this idea we used phospho-specific antibodies specific to the amino-terminal Ser33/37/Thr41 residues that are involved in targeting β-catenin for ubiquitinylation (Fig. 1b). The constitutive phosphorylation of β-catenin was readily observed in HEK293 cells when treated with the phosphatase inhibitor calyculin, and it was blocked by pretreatment with lithium, indicating the involvement of GSK3β. However, we were unable to detect any Ser33/37/Thr41 phosphorylation in SW620 cells overexpressing PKG. In support of this result, analysis of the β-catenin sequence using a phosphorylation target database (Blom et al., 2004) revealed that the amino terminal GSK3β target sites are likely to be poor substrates for PKG, but Ser552 and Ser675 were identified as potential PKG targets. Indeed, we found that 8Br-cGMP could increase phosphorylation of these sites in PKG-expressing SW620 colon cancer cells (data not shown). To examine the possibility that an unconventional route to degradation was involved, the effect of PKG on β-catenin protein levels was measured in the presence of cycloheximide (Fig. 1c). These experiments revealed a nearly linear decrease in β-catenin protein levels over time, with approximately 30% remaining after 20 h in both parental cells and those expressing PKG. Treatment with 8-Br-cGMP did not alter the rate of protein degradation in the parental SW620 cells (3.1 +/−0.7 % per hour), but slightly increased stability in the PKG-expressing cells (3.2 +/−0.8% to 2.9 +/−0.1% per hour). Although this stabilization effect of cGMP did not attain statistical significance in our studies, it demonstrates that PKG signaling does not increase β-catenin degradation in SW620 colon cancer cells.
PKG inhibits transcription from the CTNNB1 gene in colon cancer cells
The inability of PKG to induce β-catenin degradation indicated a mechanism different from GSK3β and underscored the possibility that PKG could affect expression at the transcriptional or translational levels. As a first step, the effect of PKG activation on the steady-state levels of β-catenin mRNA was measured using semi-quantitative RT-PCR (Fig. 2a). Activation of PKG in several inducible clones of SW620 resulted in a dramatic reduction of β-catenin message, whereas treatment of parental SW620 cells with inducer and 8Br-cGMP had no effect. In order to determine whether the reduced β-catenin mRNA levels were due to PKG-dependent regulation of transcription, a reporter construct previously shown to encode the essential promoter regions of the β-catenin gene (CTNNB1) was used (Li et al., 2004). Transient transfection studies with the CTNNB1-luciferase reporter demonstrated that PKG activation in SW620 cells inhibited basal transcription by approximately 60%, whereas the PKG activator 8Br-cGMP or PKG induction alone has little effect (Fig. 2b). Suppression of CTNNB1 activation by PKG was not unique to the SW620 cells, since transient transfection of PKG could also inhibit CTNNB1 transcription in several other colon cancer cell lines (Fig. 2c). Although the degree of inhibition varied between the different cells, the most dramatic reduction in CTNNB1 transcription was observed in SW480 cells (80%).
Inhibition of TCF transcription by PKG requires activation of JNK
The inhibition of CTNNB1 activation by PKG is the most likely cause of the reduced β-catenin expression in colon cancer cells. Downregulation of β-catenin expression by PKG has been shown to block transcription of TCF-dependent target genes such as cyclin D1 in SW480 cells (Deguchi et al., 2004; Thompson et al., 2000). In our inducible SW620 colon cancer cells, activation of PKG inhibited TCF-activity by 65% as measured using the TOP-flash TCF-luciferase reporter system (Fig. 3a). In these experiments PKG overexpression by itself, or treatment with 8Br-cGMP alone did not affect the basal TCF activity or the β-catenin protein levels, which underscores the requirement for active PKG. We also examined the inhibition kinetics of both β-catenin protein expression and TCF activity by PKG in SW620 cells. In these experiments approximately 60% of the total TCF inhibition (measured at 24 h) was observed after 6 h stimulation with 8Br-cGMP, at a time when there was negligible effect on β-catenin protein (Fig. 3b). This apparent disconnect between the downregulation of β-catenin protein and TCF activity was also observed in other colon cancer cells, including SW480, HCT116 and HT29. Cell lines were created with doxycycline-inducible PKG expression, and similar to SW620 cells, activation of PKG resulted in reduced β-catenin protein and TCF activity (Fig. 3c). The dramatic inhibition of TCF activity in SW480 and HCT116 cells (more than 80%) was not paralleled by the comparatively small effect of PKG on the β-catenin protein levels. Taken together, these results indicate that mechanisms in addition to the regulation of β-catenin expression, must contribute to the inhibition of TCF signaling by PKG.
The regulation of β-catenin/TCF signaling can be complex as there are families of diffusible Wnt antagonists as well as diverse endogenous inhibitors of the β-catenin/TCF interaction (reviewed recently in Jin et al., 2008). To identify additional pathways that contribute to TCF-inhibition by PKG, we first ruled out paracrine mechanisms by growing cells on trans-well filters (data not shown). It was then determined that PKG did not significantly block the nuclear localization of β-catenin, or increase the expression of endogenous TCF inhibitors (data not shown). These observations led us to consider the involvement of FOXO transcription factors, which have been reported to inhibit TCF activity in response to oxidative stress (Almeida et al., 2007; Essers et al., 2005). FOXO regulation was particularly attractive because activation of FOXO4 by oxidative stress requires JNK (Essers et al., 2004), which has previously been shown to be activated by PKG in colon cancer cells (Soh et al., 2000). Using phospho-JNK antibodies, we first confirmed that PKG can activate JNK (Fig. 4a), and then sought to verify that oxidative stress could inhibit TCF activity in our cells. In transiently transfected SW480 cells, stimulation with either 100 µM 8Br-cGMP or 200 µM H2O2 for 6 h resulted in 50% inhibition of the basal TCF activity (Fig. 4b). Cotransfection of the cells with a dominant negative JNK (JNK-DN) construct blocked the TCF-inhibition by PKG (23%) and H2O2 (10%). These data indicate that PKG and H2O2 use a common mechanism to inhibit TCF activity in colon cancer cells, and this pathway requires activation of JNK. The ability of the JNK-DN to only partially block TCF-inhibition by PKG (56% rescue) compared to H2O2 (90% rescue), supports the idea that downregulation of β-catenin expression also plays a role downstream of PKG. The inhibition of TCF activity by H2O2 was blocked by co-incubation of the cells with the antioxidant N-acetylcysteine (NAC) (Fig. 4c). In contrast, NAC did not affect the ability of PKG to inhibit TCF activity, which indicates that JNK activation by PKG is not due to a redox effect. This is consistent with the idea that PKG can activate JNK by another pathway such as MEKK phosphorylation as has been reported previously (Soh et al., 2001).
PKG activates FOXO4 in colon cancer cells
The FOXO transcription factors are a recently recognized family of tumor suppressors that are gaining significant interest as targets for cancer therapy (Arden, 2006; Dansen & Burgering, 2008; Maiese et al., 2008; Weidinger et al., 2008). These proteins are generally inactive in cancer cells owing to constitutive Akt, which promotes interaction with 14-3-3 proteins and cytosolic localization (Dansen & Burgering, 2008; Fu & Tindall, 2008; Maiese et al., 2008). Our studies focused on FOXO4 because this isoform has unique JNK phosphorylation sites that are involved in H2O2 induced FOXO activation in colon cancer cells (Essers et al., 2004). Transient transfection of a flag-FOXO4 construct into colon cancer cells showed both cytosolic and nuclear localization of the protein by immunofluorescence (Fig. 5a). However, in cells cotransfected to express FOXO4 with PKG, treatment with 8Br-cGMP caused a striking mobilization of the cytosolic fraction to the nucleus. FOXO can activate several classes of genes, including those involved in cell cycle arrest (p27), DNA repair (GADD45), and antioxidant pathways (catalase, MnSOD) (Burgering, 2008; Calnan & Brunet, 2008; Ho et al., 2008; Huang & Tindall, 2007). Activation of PKG in SW620 colon cancer cells caused increased expression of the classical FOXO target genes encoding catalase and MnSOD (Fig. 5b, c), but in these studies PKG did not activate other target genes such as p27 or GADD45 (not shown). These data are evidence that PKG can activate FOXO4 in colon cancer cells.
Activation of FOXO4 is required for the inhibition of TCF by PKG
FOXO proteins regulate gene expression not only by activating FOXO target genes, but also by interacting with other transcription factors (reviewed in (van der Vos & Coffer, 2008). Several groups have shown that FOXO binds strongly to β-catenin, which increases the transcription of FOXO target genes but inhibits TCF-transcription by diverting β-catenin (Almeida et al., 2007; Essers et al., 2005; Hoogeboom et al., 2008). Results shown here demonstrate activation of FOXO4 in colon cancer cells but the importance of this pathway to the inhibition TCF activity by PKG required further study. Immunoprecipitation experiments demonstrated that transient expression of flag-FOXO4 was able to bind to endogenous β-catenin under control conditions, but activation of cotransfected PKG dramatically increased the level of β-catenin bound (Fig. 6a). In keeping with the importance of JNK downstream of PKG, cotransfection of the JNK-DN completely blocked the PKG-dependent increase in β-catenin bound to FOXO4. These results strengthen the notion that PKG can activate FOXO in colon cancer cells and supports previous work by other groups that have reported that activated FOXO binds to β-catenin. We found that overexpression of FOXO4 could inhibit basal TCF activity with similar effectiveness as PKG (data not shown). In order to determine whether FOXO4 activation is involved in TCF-inhibition by PKG, we generated siRNA to knockdown FOXO4 expression. Compared to a non-targeting siRNA, the FOXO4 siRNA-1 generated from position 829 of the coding region had better than 90% knockdown of transiently expressed FOXO4, whereas siRNA-2 (position 1267) was less efficient with close to 70% knockdown (Figure 6b). The effect of silencing FOXO4 in the SW480 colon cancer cells on the ability of PKG to inhibit TCF transcription was examined using TCF-luciferase assays subsequent to siRNA transfection (Fig. 6c). Results showed that the more efficient siRNA-1 completely blocked the effect of PKG and actually raised the basal TCF-activity slightly higher than control. The siRNA-2 was also able to block the inhibitory effect of PKG on TCF activity from 60% inhibition to only 20% inhibition. These data demonstrate that FOXO4 is necessary for PKG to inhibit TCF activity in colon cancer cells. Taken together our results outline a model in which PKG activity can inhibit TCF activity by two mechanisms (Fig. 6d). One effect is to block transcription from the CTNNB1 gene, resulting in reduced total β-catenin levels. Another mechanism is to activate FOXO4, which by binding to β-catenin, limits the pool available to interact with TCF.
Discussion
Elevated β-catenin expression is a common feature of colon cancer cells (Sparks et al., 1998), where it promotes proliferation and angiogenesis and blocks differentiation (Fodde & Brabletz, 2007; Katoh & Katoh, 2007). The importance of β-catenin to colon tumorigenesis and progression has made this protein an important therapeutic target (Luu et al., 2004; Takahashi-Yanaga & Sasaguri, 2007). There is growing evidence that type 1 PKG has anti-tumor effects in colon cancer cells, including promoting apoptosis and inhibiting growth and angiogenesis (Deguchi et al., 2004; Hou et al., 2006a; Hou et al., 2006b; Kwon et al., 2008; Soh et al., 2008). Some of these effects may to be due inhibition of β-catenin/TCF signaling by PKG, which has been reported by independent laboratories (Kwon et al., 2008; Thompson et al., 2000). The present study has focused on the mechanism of β-catenin/TCF inhibition by PKG in order to better understand the therapeutic potential of this enzyme.
Previous studies have suggested that PKG overrides APC deficiency by mimicking GSK3β and directly phosphorylating β-catenin leading to its degradation (Liu et al., 2001; Thompson et al., 2000). The present study makes this model unlikely because overexpression of PKG in SW620 colon cancer cells did not increase protein turnover or cause phosphorylation of the sites known to target β-catenin for degradation. This point is strengthened by the observation that PKG is also an effective inhibitor of TCF-activity in HCT116 cells, which possess mutant β-catenin that is a poor substrate for GSK3b (Sparks et al., 1998). The phosphorylation of β-catenin by PKG as reported previously is most likely due to Ser552 and Ser675, which are also targeted by PKA (Taurin et al., 2006). Phosphorylation of these residues does not promote degradation, but instead have been found to increase β-catenin stability and enhance transcriptional activity (Fang et al., 2007; Hino et al., 2005; Xu & Kimelman, 2007). Results shown here indicate that downregulation of β-catenin protein by PKG is due to inhibition of CTNNB1 gene transcription. The mechanism underlying regulation of CTNNB1 transcription by PKG is not presently known, but as an alternative approach to normalizing β-catenin levels in tumors, further attention to this pathway is warranted.
Downregulation of β-catenin could not account for the dramatic inhibition of TCF-activity resulting from short-term activation of PKG, and led to the novel finding that PKG activates FOXO4 in colon cancer cells. The important role of JNK in FOXO4 activation by PKG is consistent with previous studies showing that PKG activates JNK in colon cancer cells (Soh et al., 2000; Soh et al., 2001), and that JNK activates FOXO4 (Essers et al., 2004). Activation of FOXO4 can increase resistance to oxidative stress but may slow growth by competing with TCF for β-catenin binding (Almeida et al., 2007; Essers et al., 2005; Hoogeboom et al., 2008). Type 1 PKG activity is stimulated by redox stress, suggesting that activation of FOXO4 by PKG is a key component of the larger cellular antioxidant response (Burgoyne et al., 2007). Oxidative stress is carcinogenic in the colon (Federico et al., 2007), and we have found that 8Br-cGMP can activate catalase expression in the mucosa of colon explants (Browning unpublished observations). The ability of FOXO4 activation in the intestinal epithelium to protect against oxidative stress and also suppress aberrant β-catenin/TCF signaling, suggests a possible tumor prevention role for PKG. This is consistent with the ability of cGMP-elevating uroguanylin to suppress tumor burden in the Apc mouse (Shailubhai et al., 2000), but the role for PKG/FOXO in this phenomenon will require further study.
In summary, we have examined the regulation of β-catenin/TCF signaling by PKG in colon cancer cells, and report dual inhibitory activities. PKG can inhibit transcriptional activation of the CTNNB1 gene, which can lead to reduced β-catenin protein levels. A potentially more important finding is that PKG can activate FOXO4, which leads to a more rapid inhibition of TCF-activity by sequestering β-catenin. These findings are consistent with reports of antitumor properties of PKG in colon cancer cells, and further suggest the possibility of a tumor preventative role.
Materials and methods
Tissue culture and Reagents
All cell lines were obtained from the American Type Culture Collection and maintained in 5% CO2 in RPMI-1640 medium containing 10% FBS, and supplemented with 200 µM L-glutamine, 10 IU/mL penicillin, 10 µg/mL streptomycin. We have previously described the creation and characterization of SW620 colon carcinoma cell lines made inducible for type1 PKG expression in response to nanomolar levels of mifepristone (Hou et al., 2006a; Hou et al., 2006b; Kwon et al., 2008). The medium used to maintain stocks of the H3Z6 clone of inducible SW620 was supplemented with 300 µg/ml each of hygromycin and zeosin. Prior to experimentation the cells were grown up for at least one passage in the absence of antibiotics. Additional clones of SW480, HCT116 and HT29 cells made inducible for PKG1β expression by doxacycline were created using the Lenti-X™ Tet-On Advanced Inducible Expression System (Clontech, Mountain View, CA). Recombinant virus were produced in HEK-293T cells according to the manufacturer instructions. Cells were first infected with a virus encoding the dox-inducible PKG1β, and then following several days selection with puromycin, the surviving cells were re-infected with virus encoding tet-activator and colonies growing in G418 were isolated for analysis. The mifepristone, calyculin, DAPI and 8-Br-cGMP were from Calbiochem (San Diego, CA). NP-40, Tween-20, Lithium, doxacycline, and cycloheximide were from Sigma (St. Louis, MO), and unless specified, all other chemicals were from Fisher Scientific (Pittsburgh, PA).
Antibodies and constructs
The antibody against Catalase was from R&D systems (Minneapolis, MN). The antibodies specific to phospho-JNK and β-catenin (N-terminus, and phosphorylated) were all from Cell Signaling (Beverly, MA). The β-actin and anti-flag epitope antibodies were from Sigma (St. Lois, MO). The polyclonal anti-PKG1 antibodies raised against the common C-terminus, and the PKG expression vectors have been described previously (Browning et al., 2001). The TopFlash reporter system was from UpState, (Billerica, MA) and the CTNNB1 reporter has been described previously (Li et al., 2004). The flag-JNK-DN (dominant negative) and the flag-FOXO4 constructs were purchased from Addgene (Cambridge, MA).
Immunofluorescence
Cells were grown on coverslips in 6-well dishes until 50% confluent and then transiently transfected to express flag-FOXO4 and PKG1β. After treatment, the cells were washed with PBS, fixed in 4% paraformaldehyde and permeablized with 0.1% Triton-X-100. After blocking 1 h at 37°C in PBS containing 5% goat serum, the coverslips were incubated overnight at 4°C in blocking buffer containing 1/500 anti-Flag-M2 mAb (Sigma, St. Louis MO). Coverslips were then incubated for 2 h at room temperature with Alexa Fluor 568 goat anti-mouse secondary antibodies (Invitrogen) and counterstained with DAPI for 5 min, before affixing to slides using Mowiol (Calbiochem, La Jolla CA). Images were captured using a Nikon TE-300 inverted epifluorescence microscope equipped with a SPOT RT3camera and SPOT Software version 4.7 (Diagnostic Instruments Inc).
Western blotting and Immunoprecipitation
After various treatments the tissue culture dishes were placed on ice and the medium (per well of a 6 well dish) was replaced with 250 µl ice-cold lysis buffer (50 mM HEPES pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.25% deoxycholate) supplemented with phosphatase and protease inhibitor cocktails (Calbiochem, La Jolla CA). Cell suspensions were rocked for 20 min at 4°C followed by clarification of the extracts by centrifugation. For electrophoretic analysis, 30 µl of the homogenates were mixed with 10 µl 5X PAGE sample buffer, boiled for 10 min, then 20 µl was loaded per lane. Electrophoresis of proteins was performed on 10% mini-gels followed by electrophoretic transfer to nitrocellulose. The blots were blocked with 5% BSA in PBS containing Tween 20 (PTS) for 20 min at room temperature and then antibodies were added overnight at 4°C. Following addition of 1/3000 peroxidase conjugated secondary antibody (Bio-Rad Laboratories, Hercules CA) for 1 h, the bands on the blots were visualized using chemiluminescence according to manufacturer’s instructions (Pierce, Rockford IL). At least 3 washes (5 min each) using excess PTS buffer were performed between each incubation step.
The turnover of β-catenin protein in SW620 cells was determined by incubating the cells in 10 µg/ml of cycloheximide to block protein synthesis. At different times over a 24 h period, the cells were harvested and lysates were analyzed for β-catenin content by Western blotting. The intensity of the β-catenin band was quantitated using Image-Quant (Molecular Dynamics Inc.) and the levels were normalized to similarly quantitated β-actin, and expressed as a percentage of the starting amount. Results were reproduced in at least 3 independent experiments, the slopes from best-fit lines were determined for each replicate.
Interactions between FOXO4 and β-catenin were determined by immunoprecipitation using anti-flag-M2-sepharose beads (Sigma, St. Louis MO). Cells were transiently transfected with expression vectors for flag-FOXO4 and either empty pCDNA3 or PKG1β. In these experiments 20 µl of beads were added to 1 ml clarified lysate (10 cm dish of cells), which was rocked for 2 h at 4°C and then washed 3X in RIPA buffer before analysis of β-catenin content by Western blot.
Measurement of steady-state message levels
Semi-quantitative reverse transcriptase polymerase chain reaction (RT-PCR) was used to determine the relative steady-state levels of β-catenin mRNA (foreword ATCCCACTGGCCTCTGATAAA, reverse CAATAGCTTCTGCAGCTTCCT), PKG mRNA (foreword GAGCGAACTGGAGGAAGAC, reverse GGTACACAACTTCACACCTTCT), catalase mRNA (forward CGTGCTGAATGAGGAACAGA and reverse TCTTCATCCAGTGATGAGCG), mnSOD mRNA (forward AATCAGGATCCACTGCAAGG and reverse AAGGCATCCCTACAAGTCCC). Total RNA was isolated using TRIzol reagent according to manufacturer’s instructions (Invitrogen), and cDNA generated using the GeneAmp reverse transcriptase system (Roche). PCR was performed using 1 µl RT-product in reactions with 0.2 U Taq (TaKara Bio Inc), 30 cycles at 60°C anneal temperature. Parallel control reactions were performed using primers specific for the hypoxanthineguanine-phosphoribosyltransferase (HPRT1; forward GATGAAGAGCAAGGTTATGAC, reverse ACACAGAGCAACGATATGG).
Luciferase reporter assays
For measurements of transcription, the cells were cultured in 12-well plates, and at 80% confluence the assays were performed by transfecting triplicate wells with luciferase reporter plasmids using Lipofectamine 2000™ reagent (Invitrogen). Measurement of β-catenin transcriptional activity was carried out with TCF-reporter plasmids using the TOP flash system according to the manufacturer’s protocols (UpState, Billerica, MA). In these experiments cells were cotransfected with 0.2 µg TOP flash luciferase reporter/well (or mutant control FOP flash vector) and 0.2 µg CMV-β-galactosidase to control for cell number and transfection efficiency. After 16 h the medium was changed and cells were stimulated with 100 µM 8-Br-cGMP for an additional 6-8 h before enzyme assay. Cell extracts were prepared and analyzed for luciferase and β-galactosidase activities as described previously (Taurin et al., 2006). The luciferase activity was standardized with regard to respective β-galactosidase activity, and the TCF-specific luciferase was expressed as the net TOP flash luciferase after subtraction of the associated FOP flash activity. Measuring the transcription of the β-catenin gene followed an almost identical procedure except that 0.2 µg CTNNB1-luciferase was used, and basal luciferase was controlled using the empty pGL3 vector. This construct encodes the promoter regions of human β-catenin and has been described previously (Li et al., 2004). The effect of FOXO4 knockdown on TCF activity was done by creating specific short-interfering RNA (siRNA) using BLOCK-iT™ RNAi Designer (Invitrogen, Carlsbad, CA). The siRNAs used correspond to position 829 (CCCUGCACAGCAAGUUCAU; designated FOXO4-1) and 1268 (GCUGUUAGAUGGGCUCAAU; designated FOXO4-2) and a non-targeting control (GCUGAUGUACGGCUGUAAU). The cells were transfected with the siRNAs using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) and after 48 h, they were transfected again with TCF reporter as described above. After an additional 24 h the cells were harvested for luciferase determinations.
Statistical Analysis
Results are expressed as the mean +/− SEM of at least three independent experiments and statistical comparisons employed Student’s t test. A probability value of less than 0.05 was considered to be significant.
Tissue culture and Reagents
All cell lines were obtained from the American Type Culture Collection and maintained in 5% CO2 in RPMI-1640 medium containing 10% FBS, and supplemented with 200 µM L-glutamine, 10 IU/mL penicillin, 10 µg/mL streptomycin. We have previously described the creation and characterization of SW620 colon carcinoma cell lines made inducible for type1 PKG expression in response to nanomolar levels of mifepristone (Hou et al., 2006a; Hou et al., 2006b; Kwon et al., 2008). The medium used to maintain stocks of the H3Z6 clone of inducible SW620 was supplemented with 300 µg/ml each of hygromycin and zeosin. Prior to experimentation the cells were grown up for at least one passage in the absence of antibiotics. Additional clones of SW480, HCT116 and HT29 cells made inducible for PKG1β expression by doxacycline were created using the Lenti-X™ Tet-On Advanced Inducible Expression System (Clontech, Mountain View, CA). Recombinant virus were produced in HEK-293T cells according to the manufacturer instructions. Cells were first infected with a virus encoding the dox-inducible PKG1β, and then following several days selection with puromycin, the surviving cells were re-infected with virus encoding tet-activator and colonies growing in G418 were isolated for analysis. The mifepristone, calyculin, DAPI and 8-Br-cGMP were from Calbiochem (San Diego, CA). NP-40, Tween-20, Lithium, doxacycline, and cycloheximide were from Sigma (St. Louis, MO), and unless specified, all other chemicals were from Fisher Scientific (Pittsburgh, PA).
Antibodies and constructs
The antibody against Catalase was from R&D systems (Minneapolis, MN). The antibodies specific to phospho-JNK and β-catenin (N-terminus, and phosphorylated) were all from Cell Signaling (Beverly, MA). The β-actin and anti-flag epitope antibodies were from Sigma (St. Lois, MO). The polyclonal anti-PKG1 antibodies raised against the common C-terminus, and the PKG expression vectors have been described previously (Browning et al., 2001). The TopFlash reporter system was from UpState, (Billerica, MA) and the CTNNB1 reporter has been described previously (Li et al., 2004). The flag-JNK-DN (dominant negative) and the flag-FOXO4 constructs were purchased from Addgene (Cambridge, MA).
Immunofluorescence
Cells were grown on coverslips in 6-well dishes until 50% confluent and then transiently transfected to express flag-FOXO4 and PKG1β. After treatment, the cells were washed with PBS, fixed in 4% paraformaldehyde and permeablized with 0.1% Triton-X-100. After blocking 1 h at 37°C in PBS containing 5% goat serum, the coverslips were incubated overnight at 4°C in blocking buffer containing 1/500 anti-Flag-M2 mAb (Sigma, St. Louis MO). Coverslips were then incubated for 2 h at room temperature with Alexa Fluor 568 goat anti-mouse secondary antibodies (Invitrogen) and counterstained with DAPI for 5 min, before affixing to slides using Mowiol (Calbiochem, La Jolla CA). Images were captured using a Nikon TE-300 inverted epifluorescence microscope equipped with a SPOT RT3camera and SPOT Software version 4.7 (Diagnostic Instruments Inc).
Western blotting and Immunoprecipitation
After various treatments the tissue culture dishes were placed on ice and the medium (per well of a 6 well dish) was replaced with 250 µl ice-cold lysis buffer (50 mM HEPES pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.25% deoxycholate) supplemented with phosphatase and protease inhibitor cocktails (Calbiochem, La Jolla CA). Cell suspensions were rocked for 20 min at 4°C followed by clarification of the extracts by centrifugation. For electrophoretic analysis, 30 µl of the homogenates were mixed with 10 µl 5X PAGE sample buffer, boiled for 10 min, then 20 µl was loaded per lane. Electrophoresis of proteins was performed on 10% mini-gels followed by electrophoretic transfer to nitrocellulose. The blots were blocked with 5% BSA in PBS containing Tween 20 (PTS) for 20 min at room temperature and then antibodies were added overnight at 4°C. Following addition of 1/3000 peroxidase conjugated secondary antibody (Bio-Rad Laboratories, Hercules CA) for 1 h, the bands on the blots were visualized using chemiluminescence according to manufacturer’s instructions (Pierce, Rockford IL). At least 3 washes (5 min each) using excess PTS buffer were performed between each incubation step.
The turnover of β-catenin protein in SW620 cells was determined by incubating the cells in 10 µg/ml of cycloheximide to block protein synthesis. At different times over a 24 h period, the cells were harvested and lysates were analyzed for β-catenin content by Western blotting. The intensity of the β-catenin band was quantitated using Image-Quant (Molecular Dynamics Inc.) and the levels were normalized to similarly quantitated β-actin, and expressed as a percentage of the starting amount. Results were reproduced in at least 3 independent experiments, the slopes from best-fit lines were determined for each replicate.
Interactions between FOXO4 and β-catenin were determined by immunoprecipitation using anti-flag-M2-sepharose beads (Sigma, St. Louis MO). Cells were transiently transfected with expression vectors for flag-FOXO4 and either empty pCDNA3 or PKG1β. In these experiments 20 µl of beads were added to 1 ml clarified lysate (10 cm dish of cells), which was rocked for 2 h at 4°C and then washed 3X in RIPA buffer before analysis of β-catenin content by Western blot.
Measurement of steady-state message levels
Semi-quantitative reverse transcriptase polymerase chain reaction (RT-PCR) was used to determine the relative steady-state levels of β-catenin mRNA (foreword ATCCCACTGGCCTCTGATAAA, reverse CAATAGCTTCTGCAGCTTCCT), PKG mRNA (foreword GAGCGAACTGGAGGAAGAC, reverse GGTACACAACTTCACACCTTCT), catalase mRNA (forward CGTGCTGAATGAGGAACAGA and reverse TCTTCATCCAGTGATGAGCG), mnSOD mRNA (forward AATCAGGATCCACTGCAAGG and reverse AAGGCATCCCTACAAGTCCC). Total RNA was isolated using TRIzol reagent according to manufacturer’s instructions (Invitrogen), and cDNA generated using the GeneAmp reverse transcriptase system (Roche). PCR was performed using 1 µl RT-product in reactions with 0.2 U Taq (TaKara Bio Inc), 30 cycles at 60°C anneal temperature. Parallel control reactions were performed using primers specific for the hypoxanthineguanine-phosphoribosyltransferase (HPRT1; forward GATGAAGAGCAAGGTTATGAC, reverse ACACAGAGCAACGATATGG).
Luciferase reporter assays
For measurements of transcription, the cells were cultured in 12-well plates, and at 80% confluence the assays were performed by transfecting triplicate wells with luciferase reporter plasmids using Lipofectamine 2000™ reagent (Invitrogen). Measurement of β-catenin transcriptional activity was carried out with TCF-reporter plasmids using the TOP flash system according to the manufacturer’s protocols (UpState, Billerica, MA). In these experiments cells were cotransfected with 0.2 µg TOP flash luciferase reporter/well (or mutant control FOP flash vector) and 0.2 µg CMV-β-galactosidase to control for cell number and transfection efficiency. After 16 h the medium was changed and cells were stimulated with 100 µM 8-Br-cGMP for an additional 6-8 h before enzyme assay. Cell extracts were prepared and analyzed for luciferase and β-galactosidase activities as described previously (Taurin et al., 2006). The luciferase activity was standardized with regard to respective β-galactosidase activity, and the TCF-specific luciferase was expressed as the net TOP flash luciferase after subtraction of the associated FOP flash activity. Measuring the transcription of the β-catenin gene followed an almost identical procedure except that 0.2 µg CTNNB1-luciferase was used, and basal luciferase was controlled using the empty pGL3 vector. This construct encodes the promoter regions of human β-catenin and has been described previously (Li et al., 2004). The effect of FOXO4 knockdown on TCF activity was done by creating specific short-interfering RNA (siRNA) using BLOCK-iT™ RNAi Designer (Invitrogen, Carlsbad, CA). The siRNAs used correspond to position 829 (CCCUGCACAGCAAGUUCAU; designated FOXO4-1) and 1268 (GCUGUUAGAUGGGCUCAAU; designated FOXO4-2) and a non-targeting control (GCUGAUGUACGGCUGUAAU). The cells were transfected with the siRNAs using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) and after 48 h, they were transfected again with TCF reporter as described above. After an additional 24 h the cells were harvested for luciferase determinations.
Statistical Analysis
Results are expressed as the mean +/− SEM of at least three independent experiments and statistical comparisons employed Student’s t test. A probability value of less than 0.05 was considered to be significant.
Supplementary Material
S-Fig1
S-Fig2
S-Fig3
S-Materials
S-Fig1
S-Fig2
S-Fig3
S-Materials
Abstract
Activation of cGMP-dependent protein kinase (PKG) has antitumor effects in colon cancer cells but the mechanisms are not fully understood. The present study has examined the regulation of β-catenin/TCF signaling, since this pathway has been highlighted as central to the antitumor effects of PKG. We show that PKG activation in SW620 cells results in reduced β-catenin expression and a dramatic inhibition of TCF-dependent transcription. PKG did not affect protein stability, nor did it increase phosphorylation of the amino-terminal Ser33/37/Thr41 residues that are known to target β-catenin for degradation. However, we found that PKG potently inhibited transcription from a luciferase reporter driven by the human CTTNNB1 promoter, and this corresponded to reduced β-catenin mRNA levels. While PKG was able to inhibit transcription from both the CTNNB1 and TCF reporters, the effect on protein levels was less consistent. Ectopic PKG had a marginal effect on β-catenin protein levels in SW480 and HCT116 but was able to inhibit TCF-reporter activity by over 80%. Investigation of alternative mechanisms revealed that cJun N-terminal kinase (JNK) activation was required for the PKG-dependent regulation of TCF activity. PKG activation caused β-catenin to bind to FOXO4 in colon cancer cells, and this required JNK. Activation of PKG was also found to increase the nuclear content of FOXO4 and increase the expression of the FOXO target genes MnSOD and catalase. FOXO4 activation was required for the inhibition of TCF activity since FOXO4-specific siRNA completely blocked the inhibitory effect of PKG. These data illustrate a dual inhibitory effect of PKG on TCF activity in colon cancer cells that involves reduced expression of β-catenin at the transcriptional level, and also β-catenin sequestration by FOXO4 activation.