Transforming Growth Factor-β2 Is a Transcriptional Target for Akt/Protein Kinase B via Forkhead Transcription Factor*

Tumors evade cell death by constitutively activating cell survival pathways and suppressing intrinsic death machinery. Activation of cell survival pathways leads to transcriptional repression of genes associated with cell death and activation of ones promoting anti-apoptosis. Akt/protein kinase B phosphorylates forkhead transcription factors and prevents their nuclear localization, leading to repression of genes involved in apoptosis, such as Fas ligand (FasL). Using bioinformatic approaches, we have identified three consensus sequences for forkhead transcription factor binding in transforming growth factor β2 (TGF-β2) promoter. TGF-β inhibits cell proliferation and induces apoptosis in many cell types, and acquisition of TGF-β resistance is linked to tumorigenesis. In this study, we show that activated Akt down-regulates TGF-β2 promoter, and sequences within the promoter that are related to consensus forkhead binding sites are necessary for repression. Forkhead factor FKHRL1 binds in vitro to the three consensus sequences and can activate TGF-β2 promoter in normal and Akt-transformed cell lines. In human breast and pancreatic tumors, activated Akt expression correlated with down-regulation of TGF-β 2 mRNA levels. A number of tumor cells expressing activated Akt were responsive to TGF-β addition, indicating the presence of an intact TGF-β-signaling pathway. These results suggest that repression of TGF-β 2 promoter activity in cells expressing activated Akt may play a role in promoting tumorigenesis and escape from the growth-inhibitory and/or apoptotic effects of TGF-β.

Akt/protein kinase B is a cellular homologue of the viral oncogene v-Akt of the transforming retrovirus Akt-8 (1). Akt is a serine/threonine kinase that plays a critical role in cell survival signaling, and its activation has been linked to tumorigenesis (for reviews, see Refs. 2 and 3). Up-regulation of Akt as well as its upstream regulator phosphatidylinositol 3-kinase has been found in many tumors, and the negative regulator of this pathway, PTEN/MMAC, is a tumor suppressor (4). The mechanism by which Akt activity contributes to cellular transformation is generally thought to be by inhibiting apoptosis. Several substrates of Akt have been recently identified, and these include two components of the intrinsic cell death ma-chinery, BAD and caspase-9, transcription factors of the forkhead family, and a kinase IKK-␣, which regulates the NFB transcription factor (for reviews, see Refs. 2 and 3). In the case of BAD and caspase-9, phosphorylation by Akt suppresses their proapoptotic functions, accounting at least in part for the potent survival functions of Akt. In addition, Akt phosphorylates proteins that control the cell cycle progression, such as glycogen synthase kinase 3 and p21 Cip/WAF1 (5). It is clear that Akt is a focal point for deregulation in cancer and can impact on cell proliferation and survival of cancer cells through a growing list of key targets.
The best-characterized nuclear substrates of Akt are transcription factors of the Forkhead family, FKHR, FKHRL1, and AFX (6). According to the new nomenclature these three proteins have been assigned to the FOXO subfamily of Forkhead transcription factors (7). Phosphorylation of forkhead transcription factor by Akt results in its retention in the cytoplasm and repression of its target genes (6,8). Under conditions of growth factor deprivation, the phosphatidylinositol 3-kinase/ Akt pathway is inactivated, and forkhead factor is unphosphorylated, leading to activation of genes associated with programmed cell death. Genes that might be targets for transcriptional regulation by the forkhead transcription factors have been identified in several systems. These include the insulin-like growth factor-binding protein 1 (IGFBP-1) 1 (9), Fas ligand (FasL) (10), Bim (11), and p27 (12). Evidence that forkhead family members can regulate apoptosis stems from the observation that the non-phosphorylatable mutants of FKHR or FKHRL1, which are potent transcriptional activators in the nucleus, trigger apoptosis in multiple cell types (8). In the case of FKHRL1-induced apoptosis, a target gene suggested to be able to mediate this effect is the Fas ligand gene (FASL), but other genes involved in forkhead-induced apoptosis have yet to be identified. Hence, we set out to identify other candidate target genes that may be under the control of the Akt/forkhead pathway.
In this study, using bioinformatic approaches we have identified candidate target genes that contain forkhead binding sites in their promoter region. One candidate gene identified in this search is transforming growth factor ␤2 (TGF-␤2). TGF-␤ is of particular interest because it is known to inhibit cell proliferation and induce apoptosis in many cell types (13). In mammals there are three isoforms of TGF-␤, -␤1, -␤2, and -␤3, encoded by three different genes, all of which function through the same signaling systems (14). TGF-␤ signaling is initiated by binding to two receptors, termed type I and type II, and mediated by a family of effector proteins known as Smads (15,16). The TGF-␤-signaling pathway has been implicated in tumor suppression, and acquisition of TGF-␤ resistance has been linked to tumorigenesis (15,16). We show that TGF-␤2 promoter contains three consensus sequences for forkhead transcription factor binding, and TGF-␤2 is a transcriptional target for Akt/phosphatidylinositol 3-kinase signaling pathway. TGF-␤2 mRNA levels are down-regulated in several human breast and pancreatic cell lines expressing activated Akt. Growth of a number of tumor cells expressing activated Akt was inhibited by exogenous TGF-␤ addition, indicating the presence of an intact TGF-␤-signaling pathway. These studies suggest that repression of TGF-␤ 2 promoter through forkhead factor may be a novel mechanism for tumors to escape from the growth inhibition and/or apoptotic effects of TGF-␤.

MATERIALS AND METHODS
Transient Reporter Assays-Chloramphenicol acetyltransferase (CAT) reporter gene plasmids pTGF-␤2-CAT (Ϫ1729 to ϩ63) and pTGF-␤2-CAT (Ϫ28 to ϩ63) were gifts from Dr. Seong-jin Kim at NCI, National Institutes of Health. Akt expression plasmid pEE-Myr-Akt (EE epitope-tagged) was a gift from Dr. Phil Hawkins (Babraham Institute, Cambridge, UK), and kinase dead Akt pHA-Akt-AA (hemagglutininepitope tagged) was a gift from Dr. Richard Roth at Stanford University. The primer pair used for amplification of the regions containing the three forkhead binding sites (Ϫ579 to Ϫ212) in TGF-␤2 was 5Ј-A-TATAAGCTTGCAGCAAATTATAAAGGTGACCA-3Ј (forward primer) and 5Ј-CGACGGATCCCGCCCTGACAACAGTGATT-3Ј (reverse primer). The PCR product was digested with HindIII and BamHI restriction enzymes and ligated into the HindIII-BamHI sites of the plasmid pBLCAT2 containing minimal thymidine kinase (TK) promoter linked to CAT reporter gene. Transient transfections were performed using the LipofectAMINE method (Invitrogen) as described by the manufacturer. CAT assays were performed as described previously (17).
Isolation of Total RNA and Real-time Quantitative Reverse Transcription-PCR-Total RNA was isolated from cells using RNAeasy reagent (Qiagen) as per the manufacturer's protocol. The mRNA for TGF-␤2 and TGF-␤1 were measured by real-time quantitative reverse transcription-PCR (Taqman) using PE Applied Biosystems prism model 7700 sequence detection instrument. The sequences of the primers (using primer express-PE ABI) were TGF-␤1 forward primer (5Ј-CTC-TCCGACCTGCCACAGA-3Ј) and reverse primer (5Ј-TAACCTAGATG-GGCGCGATCT-3Ј). The primers for TGF-␤2 were 5Ј-GTCGCGCTCA-GCCTGTCT-3Ј (forward primer) and 5Ј-CCTCGATCCTCTTGCGCA-T-3Ј (reverse primer). The Taqman Tm fluorogenic probes used for TGF-␤1 and TGF-␤2 are 5Ј-6FAM-CCCTATTCAAGACCACCCACCTT-CTGGTA-TAMRA-3Ј and 5Ј-6FAM-CCTGCAGCACACTCGATATGGA-CCAGT-TAMRA-3Ј respectively. During PCR amplification, 5Ј nucleolytic activity of Taq polymerase cleaves the probe separating the 5Ј reporter fluorescent dye from the 3Ј quencher dye. Threshold cycle C T , which correlates inversely with the target mRNA levels, was measured as the cycle number at which the reporter fluorescence emission increases above a threshold level. The comparative C T method was used to derive relative quantitation of the mRNA levels. This takes into account differences in concentration between RNA samples by normalizing to endogenous ribosomal S9 mRNA that is not transcriptionally regulated. The TGF-␤1 and TGF-␤2 expression was relative to TGF-␤2 mRNA level in Miapaca cells.

Identification of Genes Containing Forkhead Factor Binding
Sites in Their Promoters-We undertook a search for genes containing forkhead factor binding sites in their promoter regions using computational methods (18). The forkhead factor binding sequences from IGFBP1 and FasL promoters have been characterized (8,9). The DNA binding specificities of four of the FREAC proteins belonging to the forkhead family were determined by selection of binding sites from random sequence oligonucleotides (19). Recently, the binding sites for the forkhead transcription factors have also been experimentally defined (20,21). We aligned the six known forkhead transcription factor binding sequences and derived a consensus sequence for forkhead transcription factor binding ( Fig. 1). The consensus sequence, WAARYAAAYW (W ϭ A or T, R ϭ A or G, Y ϭ C or T) was used to search the GenBank TM data base. The informatics approach focused on genes with annotated gene structure, particularly in the 5Ј-flanking region. The GCG FIND-PATTERNS program was used to match the consensus sequence of the forkhead factor binding site as described above. We analyzed the human subset (93,412 sequences) of the primate division of GenBank TM and identified 16,511 forkhead binding sites (or 6,406 sequences) that match the consensus sequence. The GenBank TM feature fields that define the transcript start site of a gene were used as coordinates to locate the matched sequences in the 5Ј-flanking region of a gene. The parsing of the GenBank TM report was carried out using script language PERL (22). This resulted in a forkhead target reference data base that is composed of 437 genes with at least one forkhead binding site in the promoter region. 2 A partial list of genes that contain the forkhead factor binding sites is shown in Table I. We selected TGF-␤2 from this list to determine whether its transcription is under the control of Akt/protein kinase B-signaling pathway. TGF-␤ is of particular interest since it is the prototype of a family of structurally related cytokines capable of inducing diverse cellular responses, including apoptosis, differentiation, and cell cycle arrest.
Activated Akt Expression Leads to Transcriptional Repression of TGF-␤2-Our analysis showed that TGF-␤2 promoter contains three consensus forkhead binding sites at Ϫ212, Ϫ480, and Ϫ579 positions from the transcription start site (Table I) (23). The core forkhead consensus binding site is not found in other isoforms of TGF-␤ such as TGF-␤1 and -␤3. The three putative forkhead binding sites in TGF-␤2 promoter matched perfectly with the consensus sequence (Fig. 2a). The sequences also matched the experimentally derived consensus site for FOXO sub-family (21). To determine whether TGF-␤2 promoter activity is regulated by activated Akt in mammalian cells, a plasmid containing TGF-␤2 promoter linked to CAT reporter gene was transiently transfected into NIH 3T3 cells, and its response to activated Akt expression was measured. Two forms of Akt expression vectors were tested. One encodes myristoylated Akt (pEE-Myr-Akt), in which a membrane-targeting myristoylation sequence and an EE epitope tag were fused to Akt-coding sequences at the N terminus. The addition of myristoylation signal to Akt-coding sequence has been previously shown to lead to constitutive activation of Akt due to membrane localization (24). The other encodes a mutant form (pHA-Akt-AA, with a hemagglutinin epitope tag at the N terminus) in which the two phosphorylation sites (Thr-308 to and Ser-473) have been mutated to generate an inactive version of the protein (25). Coexpression of constitutively active Myr-Akt in NIH 3T3 cells leads to strong repression of TGF-␤2 promoter, whereas the kinase-dead mutant of Akt had very little effect (Fig. 2b). Western blot analysis of total cell lysates using epitope-specific antibodies indicates (Fig. 2d) that the two Akt proteins are expressed at equal levels in this transient transfection experiment. Next, we transfected TGF-␤2-CAT plasmid into parental Rat-1 and myristoylated Akt-transformed Rat-1 (Myr-Akt-Rat-1) cells. As a control, plasmid pLW2 containing herpes immediate early promoter linked to the CAT reporter gene was transfected into these cells. Fig. 2c shows that TGF-␤2-CAT reporter gene expression was selectively repressed in Myr-Akt-Rat-1 cells compared with parental Rat-1 cells. In contrast, transfection of a deletion mutant of TGF-␤2 (Ϫ28, TGF-␤2-CAT) lacking all three forkhead binding sites into Rat-1 and Myr-Akt-transformed Rat-1 cells showed basal CAT activity in both cell lines. Roughly a 3.5-fold higher amount of deletion mutant was transfected in this experiment. This result shows that deletion of region containing forkhead binding sites in TGF-␤2 promoter leads to loss of Akt responsiveness. Together these results show that TGF-␤2 promoter is repressed in cells expressing activated AKT, and the region containing the three forkhead binding sites is required for its repression.
Consensus Forkhead Factor Binding Sites Mediate Repression of TGF-␤2 Promoter by Activated Akt-To determine whether the three forkhead binding sites in TGF-␤2 promoter can confer Akt responsiveness to heterologous promoters, we isolated the region containing the three forkhead binding sites (Ϫ579 to Ϫ212) in TGF-␤2 promoter by PCR methods. This region, which does not contain the binding sites for other known transcription factors, was subcloned into plasmid pBLCAT2 containing minimal TK promoter linked to CAT reporter gene (26). This plasmid, designated 3XFH-TK-CAT, was transiently transfected into NIH 3T3 cells along with myr-Akt and Akt kinase dead mutant expression vectors. Coexpression of myr-Akt, but not the kinase dead mutant of Akt, resulted in strong repression of the 3XFH-TK-CAT expression (Fig. 3a). Similarly, CAT gene expression by this plasmid was also strongly repressed in myr-Akt-Rat-1 cells compared with parental Rat-1 cells (Fig. 3b).
Forkhead Factor Binds to the Three Consensus Sequences in TGF-␤2 Promoter in Vitro and Activates in Vivo-To analyze the ability of forkhead transcription factor to bind in vitro to the putative forkhead binding sites in TGF-␤2 promoter, we synthesized oligonucleotides corresponding to the three forkhead factor binding sites in TGF-␤2 promoter. Electrophoretic mobility shift assays were performed using GST-tagged FKHRL1protein purified from an E. coli expression system. The results indicate specific binding of forkhead protein to the three forkhead consensus binding sites which can be competed out by unlabeled oligonucleotides and also by oligonucleotides corresponding to forkhead binding site in IGFBP-1 promoter ( Fig. 4). Nonspecific oligonucleotides did not compete for the mobility shift in these experiments, suggesting that forkhead factor binds specifically to these sites. Similar results were seen with all three forkhead binding sites, although the forkhead binding site 3 bound less protein compared with sequences corresponding to binding sites 1 and 2 (Fig. 4).
Next we wanted to see if forkhead transcription factor could activate TGF-␤2 promoter in vivo. Previous studies show that forkhead, when not phosphorylated by Akt, is an activator of transcription, and phosphorylation by Akt leads to its retention in cytoplasm, resulting in repression of forkhead target genes (8). A mutant of FKHRL1 in which all three putative Akt phosphorylation sites have been converted to alanines localizes to the nucleus and induces transcription of genes even in the presence of activated Akt. To determine whether forkhead transcription factor expression can regulate TGF-␤2 promoter activity in vivo, we cotransfected TGF-␤2-CAT plasmid along with expression plasmids encoding wild type or mutant FKHRL1 into Myr-Akt-transformed Rat-1 cells, ZR-75, and LNCaP tumor cells. ZR-75 is a human breast carcinoma cell line, whereas LNCaP cell line is derived from prostate carcinoma. Both these cells are null for PTEN/MMAC gene, indicating that Akt activation in these cells is primarily due to inactivation of PTEN/MMAC tumor suppressor gene product (27). TGF-␤2-CAT activity is repressed in Myr-Akt-Rat-1, ZR-75, and LNCaP cells (Fig. 5a). Cotransfection of wild type or mu-FIG. 2. TGF-␤2 promoter contains three forkhead consensus sites and is repressed by activated Akt expression. a, alignment of putative forkhead binding sites from TGF-␤2 promoter with the consensus sequence. b, NIH 3T3 cells were transfected in duplicate with pTGF-␤2-CAT (6 g) plasmid encoding TGF-␤2 promoter (Ϫ1729 to ϩ63) linked to CAT reporter gene along with Myr-Akt (activated Akt) or kinase-dead mutant of Akt (6 g each) expression plasmids. pCDNA3 (Invitrogen) DNA was added to cells designated as vector control. Two days after transfection, cells were lysed, and CAT activity was determined. c, pTGF-␤2 CAT (Ϫ1729 to ϩ63) (6 g) and p-28 TGF-␤2-CAT (Ϫ28 to ϩ63) were transfected into Rat-1 and Myr-Akt-Rat-1 cells, and CAT activities were analyzed. pLW2 plasmid was used as a control. d, cell lysates containing equal amounts of protein were electrophoresed and immunoblotted for Myr-Akt and Akt kinase dead protein levels using antibodies raised against tagged epitopes. DN, dominant negative; HA, hemagglutinin. tant FKHRL1 expression plasmids stimulated TGF-␤2-CAT activity in all three cell types (Fig. 5a). The mutant FKHRL1 protein was slightly more active than the wild type protein in stimulating TGF-␤2-CAT activity in these cell lines (Fig. 5a). Next, we cotransfected plasmid p3XFHTK-CAT into ZR-75 and LNCaP cells along with FKHRL1 expression plasmids. Wild type and mutant FKHRL1 protein stimulated CAT activity in both cell types (Fig. 5b). These results show that the TGF-␤2 promoter is repressed in tumor cells in which Akt is activated primarily due to inactivation of PTEN/MMAC tumor suppressor gene product, and forkhead factor can activate the promoter activity in vivo.
To further confirm that activated Akt expression mediates the repression of TGF-␤2-CAT activity in tumor cells, we treated LNCaP cells with phosphatidylinositol 3-kinase inhibitor LY2940002 to inhibit Akt activation and determine the effect on TGF-␤2 promoter activity. TGF-␤2-CAT and 3XFHT-KCAT plasmids were transfected into LNCaP cells and then treated with LY2940002 for 24 h. As shown in Fig. 5c, TGF-␤2-CAT and 3XFHTKCAT activities were significantly activated after treatment with LY2940002, indicating that inhibition of Akt activation by treatment with a phosphatidylinositol 3-kinase inhibitor leads to activation of TGF-␤2 promoter activity. Taken together, these results demonstrate that TGF-␤2 is a transcriptional target for Akt signaling via forkhead transcription factor.
TGF-␤2 mRNA Is Down-regulated in Tumor Cells Expressing Activated Akt-Next, we examined the relationship between Akt activation and TGF-␤2 mRNA levels in several human tumor cell lines. Total RNA from pancreatic and breast tumor cell lines was used for quantitation of both TGF-␤2 and TGF-␤1 mRNA levels by real-time quantitative reverse transcription-PCR method using TaqMan fluorogenic probes. The results showed that TGF-␤2 mRNA levels are selectively downregulated in Panc-1, Miapaca-2, Aspc-1, LNCAP, PC3, and MCF-7 cells and remained unchanged in Capan-1, Cfpac-II, FIG. 3. Forkhead binding sites in TGF-␤2 promoter can confer Akt responsiveness to heterologous promoters. The region containing the three forkhead binding sites (Ϫ579 to Ϫ212) in TGF-␤2 promoter was PCR-amplified and subcloned into the HindIII-BamHI sites of the plasmid pBLCAT2 containing minimal TK promoter linked to CAT reporter gene. a, the resulting plasmid p3XFHTK-CAT was transfected into NIH 3T3 cells along with Myr-Akt and Akt K D expression plasmids, and CAT assays were performed on cell lysates as described in the legend to Fig. 1. b, plasmid p3XFHTK-CAT was transfected into Rat-1 and Myr-Akt Rat-1 cells, and CAT activities were determined 48 h later. Plasmid pLW2 was transfected as a positive control in these cells to show equal expression of this promoter in both these cells.

FIG. 4. Forkhead transcription factor (FKHRL1) binds to the three consensus sequences from TGF-␤2 promoter in vitro.
Oligonucleotides corresponding to the three consensus Forkhead binding sites (FBS) sites from the TGF-␤2 promoter were end-labeled with [␥-32 P]ATP and subjected to electrophoretic mobility assay using 100 ng of GST-FKHRL1 fusion protein purified from an E. coli expression system. For competition experiments, a 100-fold molar excess of unlabeled FBS oligonucleotides or the consensus site from IGFBP1 promoter or nonspecific oligonucleotides were added to the reaction mixes 10 min before adding the labeled oligos. Purified GST protein was also included as a control. The DNA-protein complexes were resolved on a 4% non-denaturing polyacrylamide gel in 0.5ϫ Tris-buffered EDTA running buffer at 4°C. The gels were dried and exposed to x-ray film overnight. and MDAMB-231 cells (Fig. 6a). The fold repression observed in these cells ranged from 100-to 150-fold. There was no change in TGF-␤1 mRNA levels in these cell lines (Fig. 6a). We also determined the relative levels of TGF-␤1 and -␤2 mRNAs in these experiments and observed that in Cfpac-II and MDAMB-231 cells TGF-␤2 mRNA is expressed ϳ5-fold higher than TGF-␤1 (Fig. 6a). Recently we carried out a genome-wide gene expression analysis of breast tumor cell lines using microarrays that contain ϳ60,000 cDNAs including TGF-␤1 and TGF-␤2 (18). Fig. 5b shows the results obtained using microarray experiments, indicating that TGF-␤2 mRNA levels are down-regulated in T47D, MCF-7, ZR-75-1, BT-474, and BT-20 cells but not in MDAMB-231 and HMEC21 (normal mammary epithelial) cells. TGF-␤1 mRNA levels did not change appre-ciably in different tumor cell lines (results not shown). These results were also confirmed using Taqman analysis (Fig. 6b). Analysis of total cell extracts by Western blot using phospho-(Ser-473)Akt-specific antibodies showed that Miapaca-2, Aspc-1, Panc-1, T47D, BT-20, ZR-75-1, BT474, and MCF-7 cell extracts showed significant levels of phosphorylated Akt protein but not Capan-1, Cfpac-II, and MDAMB-231 cells (Fig. 6c). These results demonstrate that expression of activated Akt in tumor cells is accompanied by selective down-regulation of TGF-␤2 mRNA level.
TGF-␤ Sensitivity in Tumor Cells Expressing Active Akt-What is the biological significance of repression of TGF-␤2 promoter in tumor cells expressing Akt? Loss of TGF-␤ sensitivity is frequently observed in human tumors derived from FIG. 5. Forkhead transcription factor (FKHRL1) regulates TGF-␤2 promoter activity in vivo. a, Myr-Akt-Rat-1, ZR-75, and LNCAP cells were transiently transfected with pTGF-␤2-CAT plasmid DNA along with expression plasmids encoding wild type and triple mutant FKHRL1 cDNA expression vectors. Plasmid pCDNA3 was used in vector control transfections, and cells were processed for CAT activity as described in the legend to Fig. 3. b, ZR-75 and LNCAP cells were transiently transfected with p3XFHTKCAT plasmid along with FKHRL1 wild type and FKHRL1 triple mutant expression plasmids. Plasmid pCDNA3 was used in vector control transfections, and cells were processed for CAT activity. c, LNCAP cells were transfected with pTGF␤2-CAT and p3XFHTK-CAT plasmids and treated with either Me 2 SO (control) or LY2940002 (25 M) for 24 h. Cells were processed for CAT activity. cells that are normally sensitive, and the extent of TGF-␤ resistance often correlates with malignancy (14,28). Our results suggest that repression of TGF-␤ at the transcriptional level by forkhead transcription factor may be an additional mechanism to escape from the antiproliferative and/or apoptotic effects of TGF-␤. This would predict that tumor cells expressing activated Akt may have a normal TGF-␤-signaling pathway and be responsive to exogenous addition of TGF-␤. The primary mechanism of escape from the effects of TGF-␤ may be by direct repression of its promoter activity. To investigate this we tested several pancreatic and breast tumor cells for their sensitivity to TGF-␤. For this purpose, a cell viability assay was used to see if the growth of the tumor cells was inhibited by exogenous addition of TGF-␤. The results shown in Fig. 7, a and b, indicate that Panc-1, MCF-7, BT-474, and MDA MB-435 S cells, which express activated Akt, retain a normal TGF-␤-signaling pathway and are sensitive to TGF-␤. In contrast, Miapaca-2, AspC-1, Capan-1, and MDA MB-231 were resistant to TGF-␤, since these cells contain defective TGF-␤-signaling components (29 -32). Miapaca-2 cells are characterized by a lack of expression of TGF-␤-type RII receptor and are reported to be resistant to the growth inhibitory effects of TGF-␤. Similarly, Capan-1 cells have been shown to express a truncated form of Smad4 mRNA (33). Panc-1 cells express normal levels of Smad4 mRNA and are responsive to TGF-␤ (29). These results indicate that some of the tumor cell lines expressing activated Akt retain an intact TGF-␤ signaling, and the escape from the effects of TGF-␤ may be primarily due to repression at the transcriptional level.
FIG. 6. TGF-␤2 mRNA levels are selectively down-regulated in tumor cells expressing activated Akt. a, pancreatic, prostate, and breast tumor cells were grown to confluence in 100-mm dishes, and total RNA was isolated using RNAeasy reagent (Qiagen). The mRNA levels for TGF-␤2 and TGF-␤1 were measured by real-time quantitative reverse transcription-PCR (Taqman) using a PE Applied Biosystems prism model 7700 sequence detection instrument. b, total RNA samples isolated from tumor cells were subjected to microarray analysis, and the results for TGF-␤2 are indicated. The RNA samples were also analyzed by real-time quantitative PCR to quantitate TGF-␤2 mRNA levels as described above. c, tumor cells were grown to confluence in 60-mm dishes, and cell lysates containing equal amounts of protein were electrophoresed and immunoblotted for phospho-Akt and Akt protein levels using appropriate antibodies. Blots were developed using ECL reagent from Amersham Biosciences.

DISCUSSION
In this study we have shown that TGF-␤2 promoter contains three consensus binding sites for forkhead transcription factor and is down-regulated by activated Akt expression. We have shown that repression is mediated by the forkhead binding sites and that forkhead transcription factor FKHRL1 can bind to these sites in vitro and regulate TGF-␤2 promoter in vivo. TGF-␤2 promoter is repressed in tumors expressing activated Akt, and inhibition of Akt activation leads to its derepression. TGF-␤2 mRNA levels were found to be down-regulated in several human breast and pancreatic tumor cell lines expressing activated Akt. In addition, we have shown that human tumor lines expressing activated Akt retain a normal TGF-␤-signal-ing pathway and remain sensitive to exogenous addition of TGF-␤.
TGF-␤ is a potent inducer of growth inhibition in several cell types, and TGF-␤-signaling pathway has been implicated in tumor suppression (16,34). TGF-␤ has also been recently reported to play a role in apoptosis in a number of cell types including hepatocytes and several hepatoma cells (35)(36)(37). Increased levels of Smad3 or Smad4 can induce apoptosis, and dominant negative interference with Smad3 function protects against apoptosis (38). The loss of responsiveness to the antiproliferative effects of TGF-␤ results in unrestrained cell growth and correlates with the malignant progression of several tumors including breast carcinomas, hepatomas, gastric colon carcinomas, and skin tumors as well as B and T lymphomas (16,34). The role of TGF-␤ signaling as a tumor suppressor pathway is best illustrated by the presence of inactivating mutations in genes encoding components of the TGF-␤-signaling pathway (16,34). A substantial portion of colorectal and pancreatic cancers harbor inactivating mutations in the genes encoding the TGF-␤ type II receptor or its mediators, Smad2 and Smad4 (39). However, such alterations cannot account for the majority of cases in which TGF-␤ resistance is lost. Therefore, TGF-␤ resistance must also be achieved by other mechanisms. In particular, direct regulation of TGF-␤ at the transcriptional level has not been documented in human tumors. Our studies show that TGF-␤2 is a transcriptional target for activated Akt via forkhead transcription factor.
Previous studies show that TGF-␤ is a tumor suppressor that shows true haploid insufficiency in its ability to protect against tumorigenesis, and mice with one inactivated allele of the gene encoding TGF-␤ show increased propensity for developing carcinoma (40) Thus, repression of TGF-␤ synthesis at the transcriptional level may indeed favor malignant transformation. Three major isoforms of TGF-␤ (TGF-␤1, -␤2, and -␤3) have been identified in many tissues (41). TGF-␤2 is 71% homologous with human TGF-␤1. It appears that both TGF-␤1 and TGF-␤2 are equally potent in inhibiting cell growth, although striking differences have also been observed (42). A major question that arises from this study is what is the physiological significance of the selective regulation of TGF-␤2, and in particular, is TGF-␤2 more critical than ␤1? Our studies show that TGF-␤2 but not TGF-␤1 is transcriptionally repressed by Akt signaling via forkhead. Previous studies show that primary cultures of breast tumor fibroblasts from benign tumors express significantly higher amounts of TGF-␤2 protein compared with TGF-␤1 (43). Our results confirm and extend this observation and show that TGF-␤2 mRNA levels are significantly higher in breast and pancreatic tumor lines in which Akt is not activated. Thus, transcriptional regulation of TGF-␤2 in tumors expressing activated Akt is likely to be physiologically significant. Akt isoforms have been found to be amplified or overexpressed to different degrees in various human tumors (1,2). In addition to gene amplification of Akts in many tumors, it has been shown that Akt is constitutively activated in human tumors that express inactivating mutations in the tumor suppressor gene PTEN/MMAC1 (44). The gene encoding the catalytic subunit of phosphatidylinositol 3-kinase is amplified in a number of ovarian cancers leading to activation of Akt (45). Thus, tumors have developed mechanisms to evade cell death by constitutively activating cell survival pathways mediated by phosphatidylinositol 3-kinase/Akt. Activation of cell survival pathways leads to transcriptional repression of genes associated with cell death and activation of ones promoting antiapoptosis. A large body of evidence has accumulated to suggest that TGF-␤ plays two distinct roles in promoting tumorigenesis (16). First, it is necessary for cells to lose responsiveness to TGF-␤ and thereby circumvent the antiproliferative effects of TGF-␤. These resistant lines often then activate TGF-␤ expression, which is thought to act on neighboring cells to stimulate angiogenesis, inhibit immune responses, and thereby promote tumorigenic processes. Thus, transcriptional regulation of different TGF-␤ isoforms at different stages of tumor development may provide a mechanism for differential regulation of TGF-␤ expression. In this study we have shown that TGF-␤2 is transcriptionally repressed in tumor cells expressing activated Akt. Human tumor cells expressing activated Akt were found to be sensitive to TGF-␤ addition, indicating the presence of normal TGF-␤-signaling pathway. Repression of TGF-␤2 promoter in tumors expressing activated Akt may contribute toward tumor-igenesis at earlier stages of tumor development and escape from growth inhibitory and/or apoptotic effects of TGF-␤.