Opposite Regulation of Myc and p21 waf1 Transcription by STAT3 Proteins*

Activated forms of STAT3 transcription factors are often found in various cancers and tumor cell lines, indicating that this signaling pathway is involved in tumorogenesis. At the molecular level, STAT3 proteins function as transcriptional activators and up-regulate several growth-promoting genes such as myc,pim-1, or cyclin D1. However, these transcription factors have also proapoptotic functions and can activate the expression of the cell-cycle inhibitor p21 waf1 , suggesting that STAT3 can also block cell-cycle progression and prevent abnormal cell proliferation. To reconcile these observations, one would predict that the STAT3-mediated activation of p21 waf1 is lost during cell transformation. In this study, we show that upon IL-6 stimulation of glioblastoma cells, STAT3 does not activate the expression of the p21 waf1 gene, whereas the expression of the myc gene remains unaltered. Chromatin immunoprecipitation experiments show that STAT3 and its cofactor NcoA/SRC1a are effectively recruited to the p21 waf1 promoter but that this is not followed by the association of the CREB-binding protein (CBP) histone acetylase and the type II RNA polymerase as normally seen on the myc promoter. Whereas the PI-3K/Akt pathway is constitutively activated in these cells, inactivation of this pathway restores the loading of CBP and the RNA polymerase and the expression of the p21 waf1 gene without having any effect onmyc regulation. Moreover, this effect was recapitulated in HepG2 cells expressing an activated form of the Akt kinase. In these cells, the kinase blocked the STAT3-mediated expression of the p21 waf1 gene by inhibiting the recruitment of CREB-binding protein and the type II RNA polymerase, without having any effects on the loading of STAT3 and its cofactor NcoA/SRC1a. Together, these findings suggest that the phosphatidylinositol 3-kinase/Akt pathway inhibits the transcriptional activation of the p21 waf1 gene by STAT3 proteins without altering the regulation of themyc promoter.

STAT3 proteins are cytoplasmic transcription factors that are phosphorylated upon activation, translocate into the nucleus, and activate target genes. A large number of growth factors activates these transcription factors since they play an important role in a wide variety of physiological pathways such as cell proliferation, differentiation, and apoptosis. Among the STAT 1 family, STAT3 is the only gene for which ablation leads to embryonic lethality (1), and tissue-specific ablation of the transcription factor yields important defects in hepatocytes, macrophages, keratinocytes, and thymic or mammary epithelial cells (2,3). Besides these normal functions, STAT3 participates in cellular transformation and tumorigenesis. Constitutive activation of STAT3 has been reported in several primary cancers, in tumor cell lines, and in many oncogene-transformed cells, and inactivation of STAT3 in these cell lines leads to an inhibition of cell proliferation (4 -6). Moreover, a constitutively activated form of STAT3 induces cell transformation, growth in soft agar, and tumors in nude mice, further confirming the importance of the activated form detected in tumors (7). Together, these observations raise important questions concerning the gene program activated by STAT3 proteins, because the target genes activated by its oncogenic form probably favor cell transformation. Effectively, STAT3 activates several genes involved in cell cycle progression such as fos, cyclin D, myc or, pim-1, and up-regulates antiapoptotic genes such as Bcl-2 and Bcl-XL (7-11) Therefore, through a combined inhibition of apoptosis and activation of cell-cycle progression, constitutively activated forms of STAT3 are believed to participate in cell transformation (4,12). However, these observations are complicated by the fact that STAT3 is involved in contradictory cell responses. Conditional inactivation of STAT3 shows that it has proapoptotic functions during mammary-gland involution (2) and that it inhibits neutrophil proliferation (13). STAT3 activation leads to down-regulation of myc genes and induction of junB and IRF1, thereby inducing cell-cycle arrest of monocytic cells (3). Inhibition of melanoma cells proliferation also relies on STAT3 activity, and, importantly, STAT3 activates the expression of the cell-cycle inhibitor p21 waf1 (14 -16). How a single transcription factor can mediate such contradictory responses is an important issue to be resolved because STAT3mediated cell transformation is probably related to a specific set of activated genes.
In general, high expression of p21 waf1 inhibits the cdk2 kinase to block cell-cycle progression (17). This predicts that this gene should be a frequent target of mutations or inactivations in the neoplastic process. Although p21 waf1 mutations are extremely rare, its promoter is effectively silenced during carcinogenesis through CpG hypermethylation (18,19). In light of these observations, we hypothesized that STAT3 should be prevented from activating p21 waf1 in tumor cell lines. In this study, we report that STAT3 does not activate the expression of the p21 waf1 gene in T98G glioblastoma cells because of the constitutive expression of the PI3K/Akt signaling pathway. Using chromatin immunoprecipitation experiments, we found that STAT3 and its transcriptional cofactor NcoA/SRC1a are normally recruited to the promoter of p21 waf1 but that the histone acetylase CBP and the type II RNA polymerase do not associate with the promoter. Inactivation of PI3K/Akt reinduces the binding of CBP, the binding of the polymerase, and the expression of the p21 waf1 mRNA. Importantly, the transcriptional events leading to the activation of the myc gene remains normal in glioblastoma cells, suggesting that the inhibition of the p21 waf1 gene is specific.
Transient Transfections and Preparation of Nuclear Extracts-Transfection experiments and nuclear extracts were done as described before (16) using the calcium phosphate precipitation method and were repeated at least five times. The amount of transfected DNA was kept constant by addition of appropriate amounts of the parental empty, expression vector.
Electrophoretic Mobility-shift Assay, Western Blot, and Northern Blot Analysis-Nuclear extracts were preincubated for 5 min at room temperature in 25 mM NaCl, 10 mM Tris pH 7.5, 1 mM MgCl 2 , 5 mM EDTA pH 8, 5% glycerol, and 1 mM dithiothreitol with 1 g of poly(dI-dC) as a nonspecific competitor. Where indicated, 5-g extracts were preincubated for 1.5 h at 4°C with 1 g of polyclonal antibodies (C20) directed against STAT3. 10 pg of probe (5Ј-CATTTCCCGTAAATCTTGTCG-3Ј, 20,000 cpm) was then added to the protein mixture for 15 min followed by loading on a 5% polyacrylamide gel (30:1) and separated by electrophoresis in 50 mM Tris, 0.38 M glycine, 1 mM EDTA (pH 8.5). Western blot and luciferase assays were performed as described (16). Northern blots were performed with 20 g of RNA using either a human p21 or myc cDNA probes labeled with [ 32 P]dCTP using the random-priming labeling kit from Amersham Biosciences (20).

RESULTS
The IL-6 Signaling Pathway Does Not Activate the Expression of the p21 waf1 Gene in Glioblastoma Cells-To characterize STAT3 activity in glioblastoma cells, we first analyzed the expression of the p21 waf1 mRNA after IL-6 stimulation of glioblastoma T98G or control cell lines. As previously reported (14 -16), Northern blot analysis showed that IL-6 induced a significant induction of the p21 waf1 mRNA in HepG2 and NGP cells (Fig. 1A, lanes 5 and 8). Surprisingly, IL-6 stimulation of T98G cells did not lead to a detectable p21 waf1 mRNA expression (Fig. 1A, compares lane 1 and 2). Importantly, the same effect was also observed in different glioblastoma cell lines (data not shown). As a control and as previously reported (20), serum stimulation induced p21 waf1 mRNA expression to the same extent in the HepG2 and T98G cell lines (Fig. 1A, lanes 3 and 6). Equal loading was verified by visualization of 18S ribosomal RNA expression (Fig. 1A, bottom). To investigate whether the regulation of p21 waf1 mRNA expression occurred at the transcriptional level, we monitored the expression of a p21 waf1 promoter luciferase construct (20) after IL-6 or 10% serum stimulation. As shown in Fig. 1B, we found that the Cells were serum-starved for 2 days and stimulated for 6 h with IL-6 (20 ng/ml) or 10% serum as indicated. Total RNA was prepared, and 20 g of RNA was subjected to Northern blot analysis using a human cDNA probe. The membrane was stripped and reprobed with a 18S oligonucleotide (bottom panel). B, cells were transfected with 500 ng of a vector encoding the p21 waf1 promoter linked to a luciferase reporter gene. After transfection, cells were serum-starved for 24 h and stimulated with IL-6 (20 ng/ml, lanes 2, 4) or 10% serum (lane 6) for 6 h. Cytoplasmic extracts were then prepared and processed to measure luciferase activity. The mean of 5 transfections is shown. C, T98G or HepG2 cells were serum-starved for 48 h and stimulated with IL-6 (20 ng/ml, lanes 2, 4) for 15 h. After stimulation, nuclear extracts were prepared, and p21 waf1 expression was analyzed by Western blot with polyclonal antibodies directed against the protein.
p21 waf1 promoter-driven luciferase activity was up-regulated 4-fold in HepG2 cells (Fig. 1B, lanes 3-4), however, this activation was not observed in T98G cells after IL-6 stimulation ( Fig. 1B, lanes 1-2) but remained functional upon serum stimulation (Fig. 1B, lanes [5][6]. To extend these results, we evaluated the expression of the p21 waf1 protein in these cell lines. Using nuclear cell extracts, we found that the p21 waf1 protein was induced in HepG2 cells after IL-6 stimulation (Fig. 1C, lanes [3][4]. However, under the same conditions, this expression was not detectable in T98G cells (Fig. 1C, lanes 1-2). We conclude from these results that IL-6 stimulation did not activate to a detectable level the expression of the p21 waf1 promoter, mRNA, and protein in T98G cells.
IL-6 Stimulation Induces the Expression of c-myc mRNA and Promoter in T98G Cells-To determine whether STAT3-mediated gene activation was inhibited as a general phenomenon or whether this observation was only related to the p21 waf1 gene, we evaluated the activation of myc, a second target gene of the transcription factor (8). As previously reported, IL-6 induced a significant induction of the c-myc mRNA in HepG2 cells, and the same effect was observed in T98G cells ( Fig. 2A, lanes 2-5  and 7). Maximal c-myc mRNA expression was observed after 1 h of IL-6 stimulation, a time during which the expression of the p21 waf1 mRNA was not detected in both cell lines (data not shown). To confirm these results at the transcriptional level, we monitored the expression of the c-myc HBM promoter luciferase construct (21) after IL-6 stimulation. As shown in Fig.  2B, the c-myc promoter activity was up-regulated to the same extent, 6-to 8-fold, in T98G or HepG2 cells.
Altogether, these results suggest that STAT3 activates the c-myc gene to the same extent in HepG2 and T98G cells, suggesting that the inhibition of STAT3 signaling is specific to the p21 waf1 promoter.
IL-6 Activates STAT3 Phosphorylation and DNA-binding Ac-tivity in T98G Cells-To explain the absence of p21 waf1 expression in T98G cells, we first hypothesized that the signaling pathway leading to STAT3 activation was modified. To this end, STAT3 nuclear phosphorylation was analyzed in nuclear extracts of T98G cells as compared with NGP and HepG2 control cell lines. After starvation and IL-6 stimulation, STAT3 translocated to the nucleus and was phosphorylated on Tyr 705 and Ser 727 in each cell lines (Fig. 3, A and B, lanes 3-4). Hence, neither the nuclear localization nor the phosphorylation status of STAT3 was inhibited in T98G cells. In addition, electrophoretic mobility-shift assay experiments indicated that endogeneous STAT3 associated with DNA upon IL-6 stimulation in T98G cells (Fig. 3C, lanes 1-2). The presence of STAT3 in the complex was verified by supershift with a polyclonal STAT3 antibody (Fig. 3C, lanes 5-6). Thus, we conclude from these results that STAT3 activation is functional in T98G cells.
Occupancy of the p21 waf1 and myc Promoters by STAT3, CBP, NcoA/SRC1a, and Type II RNA Polymerase-Using chromatin immunoprecipitation (ChIP), we have recently shown that the transcriptional activity of STAT3 on the p21 waf1 promoter depends on the recruitment of the NcoA/SRC1a and CBP coactivators (16). After STAT3 binding to DNA, the recruitment of a SRC1a-CBP complex is believed to lead to the activation of STAT3-responsive genes. The recruitment of these complexes was analyzed on the STAT3-responsive regions of the p21 waf1 (Fig. 4A) and myc (Fig. 4B) promoters in T98G cells. ChIPs experiments show that STAT3 and NcoA/SRC1a were recruited to the p21 waf1 and c-myc promoters after IL-6 stimulation of glioblastoma cells (Fig. 4, A and B, lanes 4 and 6). As expected, the c-Myc occupancy by STAT3 and NcoA/SRC1a was associated with the recruitment of CBP and the RNA polymerase II (Fig. 4B, lanes 8 and 10). However, under the same conditions (i.e. the same extraction), STAT3 activation failed to and stimulated for 20 mn with IL-6 (20 ng/ml). A and B, after stimulation, nuclear cell extracts were prepared and activation of STAT3 was analyzed using antibodies directed against STAT3 (Fig. 3A, bottom  panel), its Tyr 705 (Fig. 3A, top panel), or Ser 727 (Fig. 3B), phosphorylated forms. C, STAT3 DNA-binding activity present in T98G cells was determined by electrophoretic mobility-shift assay after IL-6 stimulation as described above (lanes 2, 3, 5, and 6). Supershift of STAT3 was performed (lane 5) with the addition of 2 g of polyclonal STAT3 antibodies. Note that the gel was run for a longer time on lanes 3-6 to detect the supershift so that the free probe is no longer apparent.
induce any CBP or RNA polymerase II association with the p21 waf1 promoter (Fig. 4A, lanes 7-8 and 11-12). By contrast, serum stimulation induced the recruitment of CBP and RNA polymerase II to the p21 waf1 and myc promoters (Fig. 4C, lanes  2 and 4 for p21 waf1 , lane 8 for myc). Finally, as a control for DNA sonication efficiency, PCR analysis did not detect any increase in the STAT3 occupancy of a region 2 kb upstream of the STAT3-responsive region of the p21 waf1 promoter (Fig. 4A,  lanes 13-22; Fig. 4C, lanes 5-6). Taken together, these results suggest that STAT3 and NcoA/SRC1a are recruited to the promoter of the p21 waf1 gene in T98G cells, however, the CBP histone acetylase and the RNA polymerase II are not recruited. By contrast, the recruitment of these complexes remains normal on the myc promoter.
The Akt Pathway Inhibits STAT3 Transcriptional Activity-PTEN is a tumor suppressor that inhibits tumor cell proliferation through inactivation of the PI3K/Akt signaling pathway. Loss of heterozygosity at the 10q23 PTEN locus occurs frequently in glioblastoma (22) and has been reported in T98G cells to induce the constitutive activation of Akt (23). In our hands, Western blot experiments effectively confirmed that Akt is constitutively phosphorylated on Ser 473 in T98G cells (Fig. 5A, lanes 1-2) and that IL-6 did not further activate the kinase. By contrast, IL-6 stimulation was necessary to activate Akt in control HepG2 cells (Fig. 5A, lanes 3-4). Because Akt down-regulates STAT3 transcriptional activity (10, 24), we explored the possible role of this kinase in the modification of STAT3 transcriptional activity in T98G cells. To this end, we first confirmed in HepG2 cells the effect of a constitutively active form of Akt (Akt-CA; Ref. 25) on STAT3 transcriptional activity. As previously reported (10, 24), we found that Akt-CA inhibits the activity of a Gal4-STAT3 fusion protein (Fig. 5B,  compare lanes 2 and 4). These results suggested that Akt might regulate the STAT3 functions in T98G cells. To test this hypothesis, we then monitored the effect of the LY294002 pharmacological inhibitor of the PI3K/Akt pathway on the STAT3mediated regulation of the p21 waf1 gene in T98G cells. ChIP analysis indicated that treatment of T98G cells with LY294002 before IL-6 stimulation restored the association of CBP and the RNA polymerase II with the p21 waf1 promoter (Fig. 6A, compare lanes 1-2 and 9 -10 with lanes 3-4 and 11-12). PCR analysis did not detect any increase in the STAT3 occupancy of a control p21 waf1 DNA region (Fig. 6A, lanes 5-8 and 13-16). We then evaluated the activity of the p21 waf1 promoter after IL-6 stimulation and LY294002 pretreatment. Results showed that PI3K/Akt inhibition allowed a 3-fold activation of the p21 waf1 promoter in response to STAT3 activation, whereas it remained otherwise inactive in the absence of drug pretreatment (Fig. 6B, compare lanes 3 and 6). Further confirming FIG. 4. ChIPs analysis of the recruitment of STAT3, NcoA/ SRC1a, CBP, and RNA polymerase II on the p21 waf1 and myc promoters. Soluble chromatin was prepared from cells treated with IL-6 or untreated for 10 -20 min and immunoprecipitated with the indicated antibodies. Final DNA extractions were amplified using two pairs of primers that cover the STAT3 (A, lanes 1-8) and RNA polymerase II binding sites of p21 waf1 (A, lanes 9 -12), or one pair of primers that covers both sites on the myc promoter (B, lanes 1-10). As a control, a distal region of the p21 waf1 promoter was analyzed (A, lanes 13-22 and C, lanes [5][6]. Amplifications shown in C were done with the same pair of primers, respectively, except that the cells were stimulated with 10% serum. Note that RNA polymerase II association was detected after 20 min of IL-6 or serum stimulation, whereas the other complexes are present as early as 10 min.

FIG. 5. Akt inhibits the transcriptional activity of a Gal4-STAT3 fusion protein.
A, T98G or HepG2 cells were serum-starved for 2 days and stimulated or untreated for 30 min with IL-6 (20 ng/ml). Nuclear extracts were prepared and activation of Akt was analyzed using antibodies directed against Akt (A, lanes 1-2 top), or its Ser 473phosphorylated form (A, lanes 1-2 bottom and lanes 3-4). B, HepG2 cells were co-transfected with vectors expressing a Gal4 luciferase reporter gene (100 ng) together with vectors expressing a Gal4 fusion protein linked to the STAT3 activation domain (300 ng, lanes 2 and 4, with a plasmid encoding an activated form of the Akt kinase (Akt-CA, lanes [3][4]. After transfection, cells were serum-starved for 24 h and stimulated for 6 h with IL-6 (20 ng/ml). Cytoplasmic extracts were then prepared and processed to measure luciferase activity. The mean of 5 transfections is shown. these observations, Northern blot experiments showed that inactivation of the PI3K/Akt signaling pathway restored the expression of the p21 waf1 mRNA upon STAT3 activation, whereas this was not observed under control conditions (Fig.  6C, compare lanes 5 and 6). Finally, expression of the p21 waf1 protein was also recovered when STAT3 was activated after LY294002 pretreatment (Fig. 6D, compare lanes 5 and 6). Importantly, inactivation of the PI3K/Akt signaling pathway by LY294002 had no significant effect on the expression of myc mRNA in T98G cells (Fig. 6C, lanes 7-11, compare the top and  middle panels). Moreover, the activity of the p21 waf1 promoter was not affected by the inhibitor in HepG2 cells (Fig. 6B,  compare lanes 10 and 12).
Therefore, we concluded from these results that the PI3K/ Akt signaling pathway inhibits the activation of the p21 waf1 gene but not of myc in T98G cells. To further confirm these results, we made the hypothesis that we should be able to recapitulate the inactivation of the p21 waf1 gene in HepG2 cells by using the constitutively active form of Akt, Akt-CA. To this end, HepG2 cells were stably transfected with an expression vector encoding the activated form of Akt, Akt-CA, or the corresponding parental plasmid. Following G418 selection, Akt-CA was clearly overexpressed and phosphorylated in Akt-CA expressing cells as compared with control clones (Fig.  7A, lanes 1-2). Given that the PI3K/Akt signaling pathway inhibited the transcriptional activity of STAT3 in T98G cells, we hypothesized that the expression of p21 waf1 should be downregulated in Akt-CA-expressing clones. Effectively, although IL-6 and 10% serum stimulation significantly induced p21 waf1 mRNA in control HepG2 cells (Fig. 7B, lanes 1-3), the STAT3mediated induction was not detected in CA-Akt-expressing cells (Fig. 7B, lanes 4 -5). Importantly, the kinase did not affect  1-8) on the p21 waf1 promoter was then analyzed by ChIPs analysis as described above. B, T98G (lanes 1-6) or HepG2 cells (lanes 7-12) were transfected with 500 ng of a vector encoding the p21 waf1 promoter linked to a luciferase reporter gene. Cells were then serum-starved for 24 h, pretreated for 1 h with LY294002 (lanes 3, 6, 9, and 12; 50 M) or with Me 2 SO (vehicle, lanes 2, 5, 8, and 11) and were then stimulated with IL-6 for 6 h. Cytoplasmic extracts were then prepared and processed to measure luciferase activity. The mean of 5 transfections is shown. C, Northern blot analysis of the expression of the p21 waf1 (lanes 1-6) or myc mRNAs (lanes 7-11) after IL-6 stimulation of T98 G cells, in the absence ( lanes  1-2, 4 -5, lanes 7-11 middle panel) or presence (lanes 3 and 6, lanes  7-11 top panel) of LY294002 pretreatment. D, Western blot analysis of the expression of the p21 waf1 protein. T98G cells were serum-starved for 48 h and stimulated with IL-6 (20 ng/ml, lanes 4 -6) for 15 h in the absence (lanes 1-2, 4 -5) or presence (lanes 3, 6) of LY294002. After stimulation, nuclear extracts were prepared and p21 waf1 expression was analyzed by Western blot with polyclonal antibodies directed against the protein. Equal loading was verified using antibodies directed against tubulin.  1). After G418 selection, overexpression was verified using nuclear extracts analyzed by Western blot and polyclonal antibodies against the Ser 473 -phosphorylated form of the kinase. B, HepG2 cells described in A were serumstarved for 48 h and then stimulated with IL-6 (10 ng/ml, lanes 2 and 5) or 10% serum for 4 h. Northern blot analysis of p21 waf1 expression was then performed in control clones (lanes 1-3) or in Akt-CA-expressing cells (lanes 4 -6). C, soluble chromatin was prepared from stable cell lines treated with IL-6 or untreated and immunoprecipitated with the indicated antibodies. Final DNA extractions were amplified using pairs of primers that cover the STAT3 (lanes 1-4) and RNA polymerase II-binding sites (lanes 5-8) of the p21 waf1 promoter as described above. the serum-mediated activation of the p21 waf1 gene, further indicating the specificity of the kinase effects toward the STAT3 transcriptional activity. Equal loading was verified with control 18S ribosomal RNA staining. ChIPs experiments were then performed to determine whether the constitutive expression of the kinase could mimic the observations made on the transcriptional cofactor recruitment in T98G cells. As described for the glioblastoma cell line, STAT3 and NcoA/SRC1a were recruited to the p21 waf1 promoter after IL-6 stimulation to the same extent in control or in Akt-CA-expressing cell lines (Fig. 7C, compare lanes 1-2 and 3-4, middle panels). Importantly, although STAT3 activation was associated with the recruitment of CBP and RNA polymerase II in control cell lines, these proteins were not recruited by STAT3 on the p21 waf1 promoter in cells expressing the constitutive form of the Akt kinase (Fig. 7C, compare lanes 1-2 and 3-4, bottom  panel, and compare lanes 5-6 and 7-8). Therefore, we conclude from these results that Akt blocks the STAT3-mediated activation of the p21 waf1 gene by inhibiting the recruitment of the histone acetylase CBP and the RNA polymerase II to the promoter.

DISCUSSION
STAT3 proteins are mediators of cell survival and proliferation that play important roles in cellular transformation and tumorigenesis (4 -6). Therefore, it is important to determine the subset of target genes that are differentially activated by the oncogenic form of the transcription factor as opposed to the genes activated by STAT3 during normal growth conditions. One could predict that oncogenic forms of STAT3 activate growth-promoting genes either directly involved in cell-cycle progression or in prosurvival functions. However, STAT3 has also been reported to activate the expression of p21 waf1 , which inhibits cell-cycle progression when highly expressed. Therefore, this inhibitor theoretically should block the effect of the oncogenic form of STAT3 on cell proliferation. In this study, we have shown that the activation of p21 waf1 by STAT3 is inhibited by the constitutively activated form of Akt present in T98G cells. Moreover, these results were extended to a different model, because HepG2 cells that overexpress a constitutively activated form of the kinase do not activate the expression of p21 waf1 upon STAT3 activation. Importantly, activation of Akt is a common event in glioblastomas as a consequence of LOH at 10q23, the locus of the PTEN phosphatase. Confirming this study, STAT3 activation of p21 waf1 is also lost during progression of human malignant melanomas (26), whereas Akt confers resistance to the antiproliferative effect of IL-6 on these cells (27). Finally, the regulation of the myc gene remained unaltered in T98G cells, suggesting that Akt specifically targets the inhibitory functions of STAT3 on cell-cycle progression.
A few possibilities can be raised concerning the molecular mechanisms whereby activation of the p21 waf1 gene is inhibited. First, methylation of the p21 waf1 proximal promoter could lead to silencing of the gene. We consider this possibility unlikely because mRNA expression is induced after serum stimulation, and preliminary experiments in our laboratory indicate that the promoter is not methylated in T98G cells. Alternatively, we have recently shown that cyclin D1 inhibits the transcriptional activity of STAT3, and it is known that Akt activates the expression of the cyclin through protein stabilization (28). Therefore, the effect of Akt could be indirect and mediated through an enhanced interaction between cyclin D1 and STAT3. Interestingly, both cyclin D1 and Akt affect the activation domain of STAT3, and both proteins have the same effect on the recruitment of STAT3 transcriptional cofactors (29). 2 ChIP experiments are actually conducted to determine whether cyclin D1 is present on the promoter of the p21 waf1 gene but not on myc in glioblastoma cells.
It remains also to be determined why Akt inhibits the activity of STAT3 on the p21 waf1 without having any effect on the myc gene. Most enhancers contains DNA-binding sites for multiple transcription factors so that differential gene expression is activated by the combination of the corresponding proteins. The presence or absence of these factors and their regulated association in a three-dimensional complex maintained by architectural proteins (enhanceosome) achieve a specific pattern of gene expression. For instance, viral induction of the interferon-␤ gene requires the coordinate activation of NFB, interferon regulatory factors, and ATF-2/c-Jun dimers with the architectural protein HMGI(Y) to form an enhanceosome that recruits the transcriptional machinery (30 -33). Such an enhanceosome remains to be characterized in the case of the STAT3-regulation of the p21 waf1 and myc genes. However, STAT3 cooperates with c-jun to achieve a maximal expression of the ␣2-macroglobulin gene (34,35), and the same complex regulates transcription from the Fas promoter. Interestingly, in this different experimental model, the STAT3-jun complex is also inhibited by the PI3K/Akt-signaling pathway (10,24). In light of these observations, it is tempting to speculate that the regulation of the p21 waf1 promoter relies on the formation of a c-Jun-STAT3 heterodimer that is negatively regulated by Akt. By contrast, the activation of myc would not rely on such a complex. Alternatively, we also found that Akt blocks the STAT3-mediated activation of the p21 waf1 gene by inhibiting the recruitment of CBP and the RNA polymerase II to the 2 F. Bienvenu and O. Coqueret, unpublished observations. FIG. 8. Proposed model for STAT3-mediated activation of the p21 waf1 gene during carcinogenesis. Under normal conditions, the STAT3-mediated activation of the p21 waf1 and myc genes relies on the formation of a STAT3/SRC1a/CBP complex that participates in the promoter loading of the type II RNA polymerase. After constitutive activation of the PI3K/Akt signaling pathway, the association of CBP with STAT3/SRC1a is inhibited and the RNA polymerase II does not bind to the p21 waf1 promoter anymore. By contrast, the regulation of the myc gene remains normal. The expression of p21 waf1 is prevented, which might lead to an enhanced proliferation.
promoter, whereas the loading of STAT3 and its cofactor NcoA/ SRC1a is not modified. The NcoA/SRC1a coactivator is thought to contribute to transcriptional activation by recruiting the CBP histone acetylase to transcription factors. Activation of the p21 waf1 gene probably relies on the recruitment of a SRC1a-CBP complex that initiates the loading of the initiation complex, either through histone acetylation and chromatin remodeling or through a direct contact with the RNA helicase A/RNA polymerase II complex (36). We can speculate that the PI3K/ Akt signaling pathway would block the association between NcoA/SRC1a and CBP on the p21 waf1 promoter, thereby releasing the histone acetylase and inhibiting the association with the polymerase complex on the p21 waf1 promoter. Finally, it is also possible that the PI3K/Akt signaling pathway activates a transcriptional repressor that would inhibit only the p21 waf1 promoter without having any effect on the myc gene. Alternatively, these kinases could also target an unknown transcription factor activated by IL-6 that would be different from STAT3 and that would specifically regulate the p21 waf1 gene but not the myc promoter.
Altogether, these results elucidate a new pathway induced by oncogene activation that would displace the transcriptional activity of STAT3 toward the activation of growth-promoting genes at the expense of cell-cycle inhibitory proteins. We propose that this could be a necessary step in the activation of the STAT3 oncogenic potential. Inactivation of p21 waf1 would remove a major inhibitory signaling pathway responsible for cell-cycle arrest, so that oncogenic STAT3, in cooperation with other oncogenes such as Akt, could activate only growth-promoting genes (Fig. 8). Because this hypothesis is for the moment based on the regulation of only two genes, further experiments are needed to determine whether these observations are related to a general phenomena or are only specific to the myc and p21 waf1 promoters.