Opposing Effects of KLF5 on the Transcription of MYC in Epithelial Proliferation in the Context of Transforming Growth Factor β

The proto-oncogene MYC plays a critical role in cell proliferation and tumorigenesis, and its down-regulation by transforming growth factor β (TGFβ) signaling is necessary for TGFβ to inhibit cell proliferation. KLF5, on the other hand, is a pro-proliferative basic transcription factor that reverses function to become an anti-proliferative TGFβ cofactor upon TGFβ stimulation in epithelial homeostasis. In this study we investigated whether KLF5 directly regulates MYC transcription in epithelial cells in the context of TGFβ. Knockdown of KLF5 significantly reduced MYC expression in the HaCaT epidermal epithelial cells. When TGFβ was applied, however, whereas MYC expression was significantly inhibited, knockdown of KLF5 increased MYC expression. Furthermore, re-expression of KLF5 restored the inhibitory effect of TGFβ on MYC expression in two cancer cell lines. Chromatin immunoprecipitation and oligo pulldown experiments demonstrated that whereas binding of KLF5 to both KLF5 binding element (KBE) and TGFβ inhibitory element (TIE) DNA elements was necessary for MYC transcription, binding to KBE was decreased by TGFβ, and binding to TIE was increased by TGFβ. These results suggest that KLF5 is not only essential for MYC transcription in proliferating epithelial cells but also mediates the inhibitory effect of TGFβ on MYC transcription. Furthermore, different binding sites mediate different effects of KLF5 in the context of TGFβ.

The proto-oncogene MYC plays a critical role in cell proliferation and tumorigenesis, and its down-regulation by transforming growth factor ␤ (TGF␤) signaling is necessary for TGF␤ to inhibit cell proliferation. KLF5, on the other hand, is a pro-proliferative basic transcription factor that reverses function to become an anti-proliferative TGF␤ cofactor upon TGF␤ stimulation in epithelial homeostasis. In this study we investigated whether KLF5 directly regulates MYC transcription in epithelial cells in the context of TGF␤. Knockdown of KLF5 significantly reduced MYC expression in the HaCaT epidermal epithelial cells. When TGF␤ was applied, however, whereas MYC expression was significantly inhibited, knockdown of KLF5 increased MYC expression. Furthermore, re-expression of KLF5 restored the inhibitory effect of TGF␤ on MYC expression in two cancer cell lines. Chromatin immunoprecipitation and oligo pulldown experiments demonstrated that whereas binding of KLF5 to both KLF5 binding element (KBE) and TGF␤ inhibitory element (TIE) DNA elements was necessary for MYC transcription, binding to KBE was decreased by TGF␤, and binding to TIE was increased by TGF␤. These results suggest that KLF5 is not only essential for MYC transcription in proliferating epithelial cells but also mediates the inhibitory effect of TGF␤ on MYC transcription. Furthermore, different binding sites mediate different effects of KLF5 in the context of TGF␤.
The c-myc (MYC) gene encodes a short-lived transcription factor (MYC) which heterodimerizes with Max. MYC/Max heterodimers activate or repress two distinct pools of target genes that elicit a variety of biological responses, including cell cycle progression, cellular growth, differentiation, and apoptosis/survival (1)(2)(3). Physiologically, MYC is broadly expressed during embryogenesis and in the compartments of adult tissues that possess high proliferative capacity, including skin epidermis and gut, and its role in the regulation of cell proliferation, differentiation, and apoptosis has been demonstrated (3). In epithelial homeostasis, MYC has a positive function in cell proliferation which involves its interaction with the zinc finger protein Miz-1 and Smads to repress the cyclin-dependent kinase inhibitor p15 Ink4b (4,5). More specifically, MYC appears to promote the switch from stem cell to transit amplifying cell (6 -9), and ectopic expression of MYC disrupts terminal differentiation of epithelial cells (7). During TGF␤ 2 -induced epithelial differentiation, MYC is rapidly down-regulated, and its down-regulation is necessary for TGF␤ to induce cyclin-dependent kinase inhibitors p15 Ink4b and p21 Cip1 and to inhibit cell cycle progression from G 1 to S phase (10,11). Although the down-regulation of MYC involves a number of transcription factors including Smads and E2F4/5 (12)(13)(14)(15), regulation of MYC in proliferating cells is not well understood.
Human Krüppel-like factor 5 (KLF5, also named IKLF or BTEB2) belongs to the Sp/KLF zinc finger transcription factor family, which is composed of about 20 mammalian members that share three C2H2-type zinc fingers at the carboxyl terminus (16,17). KLF5 is a basic transcription factor necessary for cell proliferation in epithelial homeostasis. For example, it is highly expressed in proliferating epithelial cells including immortal but untransformed epithelial cell lines as well as proliferating primary cultures of epithelial cells, which mostly represent progenitor cells (18 -20). Consistently, KLF5 is expressed at a higher level in basal rapidly proliferating cells of normal intestine, but its expression is reduced in mature and differentiated cells (21). In addition, knock-out of one KLF5 allele significantly reduced the size of villi in mouse intestine, whereas knock-out of both KLF5 alleles was embryonically lethal (21). Furthermore, overexpression of KLF5 in the epidermis of transgenic mice caused hyperplasia of basal cells (22), further indicating a pro-proliferative role of KLF5 in epithelial homeostasis.
Our recent study demonstrates that, although essential for cell proliferation, KLF5 reverses function and becomes a cofactor for TGF␤ to inhibit cell proliferation upon TGF␤-mediated acetylation in epithelial cells (23). Based on the correlation between KLF5-and MYC-proliferating epithelial cells and that between the functional reversal of KLF5 and down-regulation of MYC in TGF␤-induced cell cycle arrest in epithelial cells, we hypothesize that, as a basic transcription factor, KLF5 could function in the transcription of MYC in epithelial homeostasis. In this study we tested this hypothesis using an in vitro model of epithelial homeostasis. We found that KLF5 binds to MYC pro-moter to induce MYC expression, and the induction is reversed when the TGF␤ signaling is activated.

EXPERIMENTAL PROCEDURES
Cell Lines and Other Materials-The HaCaT epidermal epithelial cell line was established by Dr. Norbert E. Fusenig of the German Cancer Research Center (24) and was kindly provided to us by Dr. Robert A. Swerlick of Emory University. It was maintained following established procedures (24). The MDA-MB-231 breast cancer cell line, PC-3 prostate cancer cell line, and HepG2 hepatoma cell line were purchased from the American Type Culture Collection (ATCC) (Manassas, VA) and propagated following the ATCC instructions. The TGF␤ used in this study was TGF␤1 from R&D Systems (Minneapolis, MN). The siRNA for KLF5 had a sequence of 5Ј-AAGCTCAC-CTGAGGACTCA-3Ј) and was described previously (25). The antibody against KLF5 was prepared as previously described (26).
Promoter-Luciferase Reporter Assay-Both wild type and mutant plasmids were transfected into HaCaT cells using Lipofectamine reagent (Invitrogen) following the manufacturer's instructions. Twenty-four hours later different concentrations of siRNA for KLF5 and a negative control, 5Ј-AAGUCCAAG-UUAAGCAUGGCCTT-3Ј (Dharmacon, Lafayette, CO), were transfected using siPORT Amine (Ambion, Austin, TX). For HepG2 cells and MDA-MB-231 cells, plasmids of MYC promoter reporter and pcDNA3-FLAG-KLF5 (26), pcDNA3-FLAG-KLF5-K369R (23), or pcDNA3-FLAG (Invitrogen) were transfected into cells with the Lipofectamine 2000 reagent (Invitrogen). On the following day, 100 pM (2 ng/ml) TGF␤ was added and incubated with cells for 20 h. Luciferase assay was carried out using the Promega luciferase assay kit as previously described (27). Three wells of cells were used for each data point, and each experiment was repeated at least once.
Oligonucleotide Pulldown Assay with in Vitro Translated KLF5-Full-length and truncated KLF5 (amino acids 1-367) proteins were produced by in vitro translation in the presence of [ 35 S]methionine (GE Healthcare) using the TNT Quickcoupled Transcription/Translation Systems (Promega) as described in our previous study (26). Two l of in vitro translated product were used in each oligonucleotide pulldown assay as described above.

KLF5 Plays a Necessary Role in the Transcription of MYC in
Proliferating Epithelial Cells-The HaCaT epidermal epithelial cell line was used in this study because it is highly proliferative without TGF␤ and its proliferation is inhibited upon TGF␤ activation, it has been used in the study of multiple factors including TGF␤ signaling and MYC in epithelial homeostasis (10,28,29), and both KLF5 and MYC are expressed at higher levels in these cells (13,30). To examine whether KLF5 is required for basal transcription of MYC, siRNA targeting KLF5 (siKLF5) was used to knock down endogenous KLF5 in HaCaT cells, and expression analysis was conducted. As shown in Fig.  1A, expression of MYC mRNA was significantly reduced upon the knockdown of KLF5, as detected by real-time PCR. Further-more, knockdown of KLF5 reduced mRNA levels of two MYC target genes, human telomerase reverse transcriptase (hTERT) and cyclin-dependent kinase 4 (20), as determined by real-time PCR assay (Fig. 1A). Consistent with RNA results, protein levels of MYC were also remarkably down-regulated when KLF5 was knocked down, and different concentrations of KLF5 siRNA had different effects on MYC expression in HaCaT cells (Fig.  1B), as detected by Western blotting analysis. Transfection of a mixture of two siRNAs against sequences in the 5Ј-and 3Ј-un-translated regions of KLF5 (23) also decreased MYC and hTERT levels (Fig. 1C). These results indicate that KLF5 is required for basic MYC transcription in proliferating epithelial cells.

Inductive Function of KLF5 in MYC Transcription Is Reversed to Suppressive in TGF␤-induced Cell Cycle Arrest-TGF␤
induces cell cycle arrest in HaCaT cells through gene regulation including suppression of MYC and induction of p15. To determine whether KLF5 is also involved in MYC regulation when TGF␤ is present, we analyzed MYC expression in HaCaT cells treated with TGF␤ and transfected with siRNA against KLF5. Although knockdown of KLF5 reduced MYC mRNA levels in the absence of TGF␤ ( Fig. 2A, white bars), as detected by realtime PCR, it showed an opposite effect by increasing MYC transcription in TGF␤-treated HaCaT cells ( Fig. 2A). Consistent with previous findings, TGF␤ still inhibited MYC transcription when HaCaT cells were transfected with negative control siRNA, but the inhibitory effect of TGF␤ was diminished by 50 nM siRNA for KLF5 and even reversed by 100 nM siRNA for KLF5 ( Fig. 2A). Consistent with RNA expression, Western blot analysis showed that the protein level of MYC was down-regulated upon the knockdown of KLF5 in HaCaT cells without TGF␤ (Fig. 2B). Considering that MYC is rapidly down-regulated by TGF␤ (10, 11), we selected two time points of TGF␤ treatment, 1 and 4 h, for detecting MYC protein expression, so that the effect of TGF␤ on MYC protein expression could be more accurately revealed. MYC protein expression was obviously reduced by TGF␤ without the knockdown of KLF5, but TGF␤ increased MYC protein levels in cells where KLF5 was knocked down (Fig. 2B), further suggesting that knockdown of KLF5 reverses TGF␤-caused inhibition of MYC expression in HaCaT cells.
To further clarify whether KLF5 plays a necessary role in the inhibition of MYC induced by TGF␤, we modulated the expression of KLF5 in the MDA-MB-231 breast cancer cell line and the PC-3 prostate cancer cell line and treated them with TGF␤. Both MDA-MB-231 and PC-3 cell lines express KLF5 at a lower level (18,26) and have an intact TGF␤ pathway (13,31). As shown in Fig. 2C, expression of MYC mRNA was slightly induced by 0.2 ng/ml but not by 2 ng/ml TGF␤ in MDA-MB-231 cells transfected with empty pcDNA3 vector, where the KLF5 protein level was hardly detectable (Fig. 2D). In MDA-MB-231 cells where KLF5 expression was restored by plasmid transfection, however, MYC mRNA expression was remarkably down-regulated by TGF␤ treatment at both 0.2 and 2 ng/ml for 20 h (Fig. 2C). At the protein level, TGF␤ had no effect on MYC expression in KLF5-negative control cells, but MYC expression was obviously reduced by TGF␤ treatment at 2 ng/ml for 4 h in MDA-MB-231 cells expressing KLF5 (Fig. 2D). Similar results were obtained in PC-3 cells in which KLF5 expression was restored by the infection of adenoviruses expressing either GFP alone (AdGFP, negative control) or both GFP and KLF5 (AdKLF5). Without KLF5 expression, the mRNA but not the protein level of MYC was slightly increased by TGF␤ compared with the control (Fig. 2, E and F). When KLF5 expression was restored, MYC expression was significantly increased in PC-3 cells without TGF␤ treatment, but MYC expression was inhibited by TGF␤ at both the mRNA and protein levels (Fig. 2, E and F). These results further indicate that KLF5 induces MYC  (23). The concentration for SiLuc is 100 nM, whereas those for SiKLF5 mixtures are indicated. The same method of real-time PCR as in panel A was used. Asterisks indicate statistically significant differences (p Ͻ 0.05) in gene expression between knockdown (SiKLF5) and control (SiLuc). OCTOBER 9, 2009 • VOLUME 284 • NUMBER 41 expression without TGF␤ but rescues the inhibitory effect of TGF␤ on MYC expression in tumor cells.

Opposite Effects of KLF5 on MYC Transcription
KLF5 and TGF␤ Co-regulate MYC Promoter in the Same Pattern as Their Effect on MYC Expression-To further understand the role of KLF5 in MYC regulation in the context of TGF␤, we examined the effect of KLF5 and TGF␤ on the activities of MYC promoters. As a basic transcription factor, KLF5 usually binds to GC-rich and CACCC element (CAbox) of promoter DNA (17). Recently, the CA-box was extended to CCCCACCC in the Nanog enhancer region that was shown to bind to KLF5 (32). We analyzed MYC promoter and found that the sequence 5Ј-CCCACCTTCCCCACCCTC-CCCACCCTCCCCAT-3Ј (nucleotides Ϫ319 to Ϫ288) in the MYC promoter contains two tandem CCCCACCC motifs or tandem CA boxes. We defined these two motifs as KLF5 binding elements. In addition, the TGF␤ inhibitory element, which is at nucleotides Ϫ92 to Ϫ63 relative to the P2 transcription initiation site of MYC promoter, has been established as an essential region for TGF␤ to suppress MYC expression (13)(14)(15). The core sequence of the TIE (5Ј-GGCTTGGCGG-3Ј, nucleotides Ϫ84 to Ϫ75) contains GC-rich elements that could also be bound by KLF5. Fig. 3A illustrates the promoter of MYC with relative locations of potential KLF5 bindings KBE (CA boxes) and TIE, the P1 and P2 initiation sites, and PCR primers that were used in both the ChIP assay and the construction of different sizes of MYC promoter. We cloned four MYC promoter fragments into the pGL3-basic vector to construct four promoter-reporter plasmids, including pMyc9 (containing KBE and TIE), pMyc10 (containing TIE), pMyc11 (containing TIE but is 188-bp shorter than pMyc10), and pMyc3 (containing only KBE) (Fig.  3B). We first investigated the response of these promoters to TGF␤ signals by transfecting pMyc3, pMyc9, pMyc10, and pMyc11 into HaCaT cells individually and performing luciferase reporter assays. Regardless of TGF␤, pMyc9 showed the highest promoter activity, whereas pMyc3 showed the weakest (Fig. 3C). TGF␤ treatment at 2 ng/ml for 20 h suppressed promoter activity in pMyc9, pMyc10, and pMyc11 but not in pMyc3 in HaCaT cells (Fig. 3C).
We then determined whether KLF5 regulates MYC promoter activity in the context of TGF␤ using pMyc3, pMyc9, pMyc10, and pMyc11. Reporter plasmids were co-transfected with siRNA for KLF5 (siKLF5) or siRNA for scramble sequence (SiS, negative control) into HaCaT cells (Fig. 3D), and TGF␤ treatment was applied at 2 ng/ml for 20 h. In control cells without TGF␤ treatment, knockdown of KLF5 significantly reduced the promoter activity of pMyc9 and pMyc11, and the reduction in pMyc3 and pMyc10 was not statistically significant. TGF␤ had no effect on the activity of pMyc3 vector. Treatment with TGF␤, on the other hand, significantly inhibited the promoter activity for pMyc9, pMyc10, and pMyc11 in cells with normal levels of KLF5 but significantly induced the promoter activity for each of the of promoter constructs in cells where KLF5 was knocked down. Consistent with endogenous MYC expression results in HaCaT cells ( Fig. 2A), these results further indicate that, whereas KLF5 alone transactivates MYC transcription, KLF5 reverses function when TGF␤ is present and is essential for TGF␤ to inhibit MYC transcription.
KLF5 Directly Binds to MYC Promoter, and Bindings to Different Sites Are Affected Differently by TGF␤-As a basic transcription factor, KLF5 most likely binds to MYC promoter directly to regulate its transcription. To test this hypothesis, we conducted a ChIP assay to scan the proximal region of MYC promoter (Ϫ414 to ϩ54 nucleotides) for KLF5 binding. HepG2 cells were transfected with pcDNA3-FLAG-KLF5 or pcDNA3-FLAG vector and treated with TGF␤ at 5 ng/ml for 1 h before collecting cells. The FLAG antibodies were used to immunoprecipitate the KLF5⅐DNA complex, and PCR was performed with different primer pairs from MYC promoter (Fig. 3A). As shown in Fig. 4A, ChIP assay detected the binding of KLF5 to the regions between F6 and R6 (Ϫ318 to Ϫ208), F6 and R3 (Ϫ318 to Ϫ88), and F5 and R4 (Ϫ108 to ϩ54) but not those between F3 and R5 (Ϫ414 to Ϫ318), F7 and R3 (Ϫ208 to Ϫ88), and F8 and R4 (Ϫ71 to ϩ54). In addition, binding of KLF5 to MYC promoter containing the KBE site, as shown in fragments between F6 and R6 F6 (Ϫ318 to Ϫ208) and between F6 and R3 (Ϫ318 to Ϫ88), was decreased by TGF␤ treatment, whereas the binding to MYC promoter containing the TIE site, as shown in fragment between F5 and R4 (Ϫ108 to ϩ54), was increased by TGF␤ treatment. Therefore, KLF5 binds to different regions of MYC promoter, and bindings to different regions respond differently to TGF␤.
As illustrated in Fig. 3A, the MYC promoter has two potential KLF5 binding sites, KBE and TIE. In addition, DNA fragments containing KBE and TIE had different patterns of binding to KLF5 in the context of TGF␤, as shown in the ChIP assay (Fig.  4A). To further investigate whether KLF5 directly binds to KBE and TIE on MYC promoter, we performed oligo pulldown assays for both of them. We incubated the lysates from HaCaT cells treated with or without TGF␤ with biotin-labeled oligonucleotides for wild type and mutant KBE or TIE and performed immunoprecipitation and Western blotting to detect KLF5 in the DNA-protein complexes pulled down by the oligos. As shown in Fig. 4B, whereas binding of Smad4 to KBE was undetectable, binding of KLF5 to KBE was detected, and the binding was reduced by TGF␤, which is consistent with the ChIP data. The mutation in KBE decreased its binding to KLF5 and made the binding unresponsive to TGF␤. For the TIE site, although its binding to KLF5 was also detected, the binding was increased by TGF␤, which is consistent with the ChIP data but different from KBE. Mutation of CGG to TAT in the TIE site decreased its binding to KLF5, whereas mutation of GGCT to TTAA in TIE abolished the binding (Fig. 4C). Consistent with a previous report (13), we also detected the binding of Smad4 to wild type TIE, and the binding was increased by TGF␤ (Fig. 4C). Unlike KLF5, Smad4 showed no binding to mutant 1 of TIE (CGG Ͼ TAT) but a weak binding to mutant 2 (GGCT Ͼ TTAA) (Fig. 4C).
To examine whether the binding of KLF5 to MYC promoter is direct or mediated by other proteins, we performed oligo pulldown assays using in vitro translated 35 S-labeled full-length KLF5 protein, with a truncated form of KLF5 lacking the car- suppresses MYC promoter activity in pMyc9, pMyc10, and pMyc11 but not in pMyc3 in HaCaT cells transfected with plasmids and treated with TGF␤ at 2 ng/ml for 20 h, as detected by luciferase activity assay. D, knockdown of KLF5 inhibits MYC promoter activity without TGF␤ but increases the activity of MYC promoter containing TIE in TGF␤-treated cells. HaCaT cells were transfected with 150 nM scramble siRNA (SiS, negative control) or siRNA to KLF5 (SiKLF5) and treated with TGF␤ at 2 ng/ml for 20 h. Promoter reporter activity was measured by luciferase activity assay. Asterisks indicate statistically significant differences (p Ͻ 0.05) in reporter activities between TGF␤-treated and untreated cells, whereas pound signs (#) indicate statistically significant differences between cells with KLF5 knockdown and control cells. OCTOBER 9, 2009 • VOLUME 284 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 28247
boxyl-terminal DNA binding zinc-finger domains as the negative control. Consistent with the results from HaCaT cells, in vitro translated full-length KLF5 bound to wild type KBE and TIE, but the truncated form did not (Fig. 4, D and E). Mutants of KBE and TIE also showed the same pattern of binding as in cell lysate assays. These results from both cell lysates and in vitro translated KLF5 indicate that KLF5 directly binds to both KBE and TIE, and TGF␤ reduces the binding of KLF5 to KBE but increases the binding to TIE on MYC promoter.
Our previous study showed that TGF␤ causes the acetylation of KLF5, and the acetylation alters the bindings of KLF5 to p15 promoter and other transcription factors to reverse the function of KLF5 in p15 regulation and cell proliferation (23,33). The same mechanism could be also responsible for TGF␤-determined opposing effects of KLF5 on MYC transcription. To investigate whether KLF5 acetylation affects the binding of KLF5 to KBE and TIE of MYC promoter, we conducted a ChIP assay in HepG2 cells transfected with pcDNA3-FLAG-KLF5, pcDNA3-FLAG-KLF5K369R (acetylation deficient KLF5), or pcDNA3-FLAG vector. The cells were also treated with TGF␤ at 5 ng/ml for 1 h. Antibodies against FLAG or Smad4 were used to immunoprecipitate the protein-DNA complexes. Real-time PCR was performed with different primer pairs from the MYC promoter, including F5 and R4 flanking TIE, F6 and R6 flanking KBE, and F7 and R3 not flanking either. As a positive control, Smad4 bound to the TIE-containing region (Ϫ108 to ϩ54) at a higher level upon TGF␤ treatment (Fig. 4G). Consistent with the results from ChIP assay combined with regular PCR, whereas KLF5 bound to both the TIE-containing region (Ϫ108 to ϩ54) and the KBE-containing region (Ϫ318 to Ϫ208) (Fig. 4, F and H), TGF␤ dramatically increased the binding of KLF5 to the TIE-containing region (Fig. 4F) but decreased the binding of KLF5 to the KBE-containing region (Fig. 4H). KLF5 did not bind to the region between TIE and KBE (Ϫ208 to Ϫ88) (data not shown).
Interestingly, although the K369R mutation of KLF5 increased the binding of KLF5 to the TIEcontaining fragment without TGF␤, the same mutation dramatically decreased the binding of KLF5 to the same element when TGF␤ was present (Fig. 4F). The K369R mutation also decreased the binding of Smad4 to the TIE-containing fragment when TGF␤ was present (Fig. 4G). For the KBE-containing promoter fragment, although wild type KLF5 bound to FIGURE 4. KLF5 binds to MYC promoter at different sites, and the bindings to different sites are affected differently by TGF␤. A, binding of KLF5 to MYC promoter containing the KBE site is decreased by TGF␤, whereas the binding to that containing the TIE site is increased by TGF␤, as detected by ChIP assay in HepG2 cells transfected with pcDNA3-FLAG-KLF5 or pcDNA3-FLAG plasmid. TGF␤ treatment at 5 ng/ml for 1 h was applied 40 h after transfection. M, molecular weight marker. Water served as a negative control in PCR. B, confirmation of TGF␤-reduced KLF5-KBE binding by oligo pulldown assay combined with Western blot analysis using lysates from HaCaT cells treated with or without TGF␤ (5 ng/ml for 1 h). Binding of mutant KBE to KLF5 is weaker and TGF␤-independent. Input indicates cell lysates not subjected to oligo pulldown assay. C, confirmation of TGF␤-increased KLF5-TIE binding in the same cell lysates from panel B and by the same methods. TIE mutant 1 (CGG Ͼ TAT) showed weaker response, but mutant 2 (GGCT Ͼ TTAA) had no binding. D and E, binding of in vitro translated full-length KLF5 (K5) but not carboxyl terminus-truncated KLF5 (⌬C) to both KBE and TIE of MYC promoter, as detected by the oligo pulldown assay. F-H, binding of wild type KLF5, Smad4, and acetylation-deficient mutant KLF5 (K369R) to TIE-or KBE-containing MYC promoter DNA, as detected by CHIP assay combined with real-time PCR analysis in HepG2 cells transfected with different plasmids. the promoter at a higher level and TGF␤ significantly reduced the binding, mutation of KLF5 interrupted the inhibitory effect of TGF␤ on KLF5 binding (Fig. 4H). These results suggest that acetylation of KLF5 is essential not only for TGF␤-induced binding of KLF5 to the TIE element but also for TGF␤-mediated reduction in the binding of KLF5 to the KBE element in the MYC promoter.
KLF5 Regulates MYC Promoter Activity in the Context of TGF␤ through Different Binding Sites-To further evaluate the role of KBE and TIE in KLF5-mediated MYC transcription in the context of TGF␤, we mutated KBE and/or TIE in pMyc9 MYC promoter reporter plasmid, transfected resultant reporter plasmids and siRNA of KLF5 into HaCaT cells, and performed luciferase reporter assays. In the control HaCaT cells where control siRNA (SiS) was transfected and no TGF␤ treatment was applied, whereas the activity of MYC promoter was at a higher level and mutation 1 of TIE (CGGϾTAT) did not change the activity (Fig. 5A), the activity was significantly reduced by mutation 2 of TIE (GGCT Ͼ TTAA), the mutation in KBE (ACCCTCCCCA Ͼ TTCCCTCCCTT), and the combination of KBE mutation with either TIE mutation (Fig. 5A). In the same cells treated with TGF␤, however, mutation 1 of TIE made the response of promoter to TGF␤ insignificant, whereas reporter activity for MYC promoter with mutation 2 of TIE or KBE mutation was still inhibited by TGF␤ (Fig. 5A). When both TIE and KBE were mutated, the inhibitory effect of TGF␤ on promoter activity was abolished (Fig. 5A). Knockdown of KLF5 reduced the activities of wild type pMyc9 and its mutants except the reporter vector containing both the KBE mutation and mutation 2 of TIE (Fig. 5A). We also mutated TIE in pMyc10 MYC promoter reporter plasmid and performed luciferase reporter assay (Fig. 5B). Consistent with the results from pMyc9, mutation 1 of TIE made the response of promoter to TGF␤ insignificant, whereas reporter activity for mutation 2 of TIE was still inhibited by TGF␤ (Fig.  5B). Moreover, KLF5 knockdown decreased the activity of wild type and mutant pMyc10 (Fig. 5B). These results indicate that KLF5 is necessary for MYC transcription and that both KBE and  vector control (pcDNA3). D, effect of KLF5 acetylation on MYC promoter activity in the context of TGF␤. HepG2 cells were transfected with different MYC promoter-reporter plasmids and expression plasmids for wild type KLF5 (KLF5), the acetylation-deficient mutant of KLF5 (K369R), and the vector control (pcDNA3). TGF␤ treatment was at 2 ng/ml for 20 h. Promoter reporter activity was measured by the luciferase reporter assay. Asterisks indicate statistically significant differences (p Ͻ 0.05) in reporter activities between TGF␤-treated cells and control. The pound signs (#) indicate statistically significant differences in reporter activities between a wild type promoter and a mutant promoter. TIE are important for MYC transcription regulated by KLF5 and TGF␤.
The MDA-MB-231 breast cancer cell line was used to investigate the effect of KLF5 overexpression on MYC promoter activity in the context of TGF␤ (Fig. 5C). Wild type pMyc9 MYC promoter reporter plasmid and its mutants were transfected with either pcDNA3-KLF5 or pcDNA3 vector into MDA-MB-231 cells, and luciferase reporter assays were performed. When TGF␤ was absent, overexpression of KLF5 significantly increased reporter activities in wild type pMyc9 and the mutant with mutation 1 of TIE but did not cause significant changes in pMyc9 mutants with TIE mutation 2, KBE mutation, or combined KBE and TIE mutations (Fig. 5C). When TGF␤ was present, only the wild type pMyc9 but not any of the pMyc9 mutants showed a TGF␤-mediated inhibition of promoter activity (Fig. 5C). These results are consistent with those from HaCaT cells (Fig. 5, A and B), further indicating that both KBE and TIE are important for MYC transcription regulated by KLF5 and TGF␤.
To determine whether and how acetylation of KLF5 affects KLF5-mediated MYC transcription in the context of TGF␤, we transfected different MYC promoter reporter plasmids, including pMyc9 that contained both TIE and KBE, pMyc11 containing only the TIE, and pMyc3 containing only the KBE into HepG2 cells along with plasmids for wild type KLF5 (pcDNA3-FLAG-KLF5), acetylation-deficient mutant of KLF5 (pcDNA3-FLAG-KLF5K369R), or vector control (pcDNA3-FLAG). Luciferase reporter assays were then performed (Fig. 5D). Without TGF␤, expression of wild type KLF5 but not its mutant significantly increased promoter reporter activity in each of the three MYC reporter plasmids (Fig. 5D). Interestingly, when TGF␤ treatment was applied, inhibition of reporter activity was achieved only in cells with wild type KLF5 for promoters with TIE (i.e. pMyc9 and pMyc11) but not in cells with mutant KLF5 or in cells with only the KBE (i.e. pMyc11) (Fig. 5D). These results not only indicate that KLF5 is necessary for transcriptional activation of MYC; they further indicate that acetylation of KLF5 is indispensable for KLF5 to modulate MYC transcription in the context of TGF␤.

DISCUSSION
KLF5 Is Required for the Maintenance of MYC Transcription during KLF5-mediated Epithelial Proliferation-Recently, different groups including our own demonstrated that KLF5 is required for the proliferation of non-tumorigenic epithelial cells from different tissues (20,23,33,34). MYC is also a proproliferative factor in different types of cells including epithelial cells, as demonstrated in previous studies (3,4). In this study we demonstrate that knockdown or overexpression of KLF5 inhibits or induces, respectively, the transcription of MYC in epithelial cells and that KLF5 directly binds to the promoter of MYC to regulate MYC transcription (Figs. 1-5). These results indicate a necessary role of KLF5 in the maintenance of MYC transcription in proliferating epithelial cells. We noticed that in MDA-MB-231 cells KLF5 did not induce an obvious increase in MYC transcription (Fig. 2C), which could be caused by the higher level of endogenous MYC expression and/or rapid deg-radation of MYC mRNA in this cell line (35,36). Several studies have shown that KLF5-mediated cell proliferation involves the regulation of different genes including the induction of cyclin D1 (20,34) and growth factor platelet-derived growth factor ␣ (25) as well as the inhibition of cyclin-dependent kinase inhibitors p15 and p27 (23,33,37). Our results in this report suggest that MYC, which itself plays a role in the regulation cell cyclerelated genes (3,4), is another functional mediator of KLF5 in epithelial proliferation.
Recently, expression of MYC and three other transcription factors (KLF4, Oct4, and Sox2) was shown to reverse differentiated cells to a pluripotent state (38 -40). On the other hand, KLF5 is involved in self-renewal and pluripotency maintenance in stem cells, which may be through regulating the transcription of Nanog and Oct3/4 (32,41,42). Regulation of MYC by KLF5 could also play a role in the maintenance of self-renewal and pluripotency of embryonic stem cells.
Down-regulation of MYC in TGF␤-induced Cell Cycle Arrest also Requires KLF5-As a potent inducer of cell cycle arrest, TGF␤ down-regulates the expression of MYC in different types of human cells including epithelial cells (14), and down-regulation of MYC is necessary for TGF␤ to inhibit cell cycle progression by inducing cyclin-dependent kinase inhibitors p15 and p21 (10,11,43,44). In addition, overexpression of MYC abrogates the growth inhibition induced by TGF␤ in epidermal epithelial cells (13). Our findings in this study, including those showing the necessity of KLF5 for TGF␤ to suppress MYC transcription (Fig. 2), different effects of KLF5 on the transcription and promoter activity of MYC with and without TGF␤ (Figs. [2][3][4][5], and different bindings of KLF5 to MYC promoter in the context of TGF␤ signaling (Fig. 4) indicate that KLF5 is also necessary for the down-regulation of MYC mediated by TGF␤ in epithelial cells. When the knockdown of KLF5 was more extensive, the inhibitory effect of TGF␤ on MYC transcription could be even reversed to a stimulatory effect ( Fig. 2A). These results are consistent with our previous study establishing KLF5 as an essential cofactor for TGF␤ signaling in the regulation of p15 (23). Furthermore, these results indicate that the function of KLF5 in MYC regulation reverses upon the activation of TGF␤ signaling, which is consistent with our previous study demonstrating the reversal of KLF5 function in cell proliferation and p15 regulation in TGF␤-treated epithelial cells (23).
Two KLF5 Binding Elements of MYC Promoter Respond Differently to TGF␤ Signaling and Are Necessary for MYC Transcription Regulated by KLF5 and TGF␤-Based on previously characterized MYC promoter (12)(13)(14)(15) and consensus sequences for KLF5 binding, we constructed reporter plasmids covering different sizes of MYC promoter, analyzed their promoter activities under different treatments of TGF␤ and KLF5 (Fig. 3), performed ChIP analysis to locate KLF5 binding elements on MYC promoter (Fig. 4), and carried out oligo pulldown assays to characterize the binding sites of KLF5 (Fig. 4). These experiments led to the identification of two KLF5 binding sites, KBE and TIE, on the MYC promoter that function in transcriptional regulation of MYC mediated by TGF␤ and KLF5 (Fig. 5). These findings indicate that KLF5 binds to specific elements of MYC promoter to regulate MYC transcription.
Binding of KLF5 to the KBE element, which also occurred to in vitro translated KLF5 (Fig. 4D), was decreased by TGF␤ (Fig. 4,  A, B, and H). In addition, Smad4, a typical component of the TGF␤ signaling, did not bind to KBE (Fig. 4B), and binding of acetylation-deficient mutant KLF5 to KBE was not reduced by TGF␤ (Fig. 4H). These results suggest that the KBE element is more important for KLF5 to maintain MYC expression without TGF␤. For the TIE, the binding of both KLF5 and Smad4 was increased by TGF␤, as demonstrated in the ChIP and oligo pulldown experiments (Fig. 4, A, C, and F), and the TIE-containing promoter showed a TGF␤-dependent decrease in activity (Fig. 5D). Therefore, the TIE element is more critical for KLF5 to mediate TGF␤-suppressed MYC transcription. However, both KBE and TIE are clearly necessary for either MYC maintenance or MYC repression by TGF␤, as demonstrated in the luciferase reporter assays where mutation of either KBE or TIE could attenuate the promoter activities of MYC without TGF␤, and it was necessary to mutate both KBE and TIE to abolish the inhibitory effect of TGF␤ on MYC promoter (Fig. 5, A-C).
TGF␤ induces a transcriptional complex on the TIE element to repress MYC transcription (12)(13)(14)(15), which involves Smads including Smad4. In our oligo pulldown experiment for TIE (Fig. 4C), although both KLF5 and Smad4 were detected in the protein complex precipitated with wild type TIE oligo, mutant 1 of TIE only pulled down KLF5, and mutant 2 of TIE only pulled down Smad4, suggesting that KLF5 and Smad4 bind to TIE at different nucleotides. In addition, in vitro translated KLF5, which does not contain other DNA or transcription factors, still bound to TIE and mutant 1 of TIE in the oligo pulldown assay (Fig. 4E), and KLF5 enhanced the binding of Smad4 to TIE in the presence of TGF␤, as demonstrated by ChIP assay (Fig. 4G). Taken together with our previous finding that KLF5 and Smad4 interact with each other in epithelial cells (33), our results suggest that KLF5 and Smad4 co-exist in the same transcriptional complex on MYC promoter.
Acetylation of KLF5 at Lysine 369 Is Necessary for TGF␤ and KLF5 to Regulate MYC Transcription-Previously we found that TGF␤ induces the acetylation of KLF5 and that acetylation of KLF5 is essential for TGF␤ and KLF5 to inhibit the transcription of p15 and cell proliferation, primarily by modulating the assembly of transcriptional complex (23,33). Our results in this paper further demonstrate that acetylation of KLF5 is also involved in TGF␤-KLF5-regulated MYC transcription. For example, the K369R mutation of KLF5 significantly decreased TGF␤-induced binding of KLF5 to TIE (Fig. 4F) but interrupted TGF␤-mediated inhibition of KLF5 binding to KBE (Fig. 4H). Consistently, the mutation also abolished the inhibitory effect of TGF␤ on MYC promoter activities (Fig. 5D).
We noticed that without TGF␤, mutant KLF5 still bound to MYC promoter (Fig. 4, F and H), and the binding to the TIEcontaining region was even at a higher level for mutant KLF5 than for the wild type (Fig. 4F), suggesting that the K369R mutation does not reduce the binding of KLF5 to MYC promoter DNA. Increased binding of mutant KLF5 to the TIE-containing fragment (Fig. 4F) could be because of structural changes in KLF5 caused by the mutation. We also noticed that although the K369R mutation did not reduce the binding of KLF5 to MYC promoter, it reduced MYC promoter activities in the luciferase assay (Fig. 5D). One explanation is that lysine 369 might undergo other TGF␤-independent modifications that modulate the transactivation activity of KLF5, as suggested by a previous study (45).

Loss of KLF5 in Tumor Cells May Attenuate the Function of TGF␤ in Proliferation Inhibition and MYC Down-regulation-
MYC is a proto-oncogene activated in various animal and human tumors through different mechanisms (14,46). Upon TGF␤ treatment, MYC expression decreases rapidly, relieving MYC-mediated repression of p15 to inhibit cell proliferation (10). Artificial avert of MYC down-regulation blocks the ability of TGF␤ to induce p15 and inhibit cell proliferation (13). Thus, down-regulation of MYC is a key event in the TGF␤ program of growth inhibition.
Selective loss of MYC down-regulation in response to TGF␤ was identified as a mechanism for the loss of TGF␤ growth inhibitory effect in cancer cells, including MDA-MB-231 breast cancer cells in which TGF␤ still induces the binding of Smad4 to the Smad binding element but has no effect on MYC expression (13). Mechanistically, TGF␤ induces the formation of a Smad complex on the TIE element of MYC promoter, and the formation of this complex is selectively interrupted in breast cancer cells, leading to the loss of TGF␤-mediated MYC downregulation (13). In the present study we found that re-expression of KLF5 rescued the inhibitory effect of TGF␤ on MYC expression in both MDA-MB-231 and PC-3 cancer cells (Figs. 2, C and D, and 5C). While further confirming that KLF5 is essential for TGF␤ to function in inhibiting the proliferation of epithelial cells, these results also suggest that loss of KLF5 in cancer cells, which could be mediated by frequent genomic deletion and excessive protein degradation in prostate and breast cancers (18,19,26,47), is a mechanism that attenuates the inhibitory effect of TGF␤ on proliferation inhibition and MYC down-regulation.
In summary, we found that KLF5 is a necessary factor for the transcription of MYC in proliferating epithelial cells and that KLF5 is also necessary for TGF␤-mediated MYC downregulation. The opposing functions of KLF5 in MYC regulation are mediated by its bindings to different elements on MYC promoter, which are modulated differently by TGF␤ signaling. These findings will help us understand the role of KLF5, MYC, and TGF␤ in the development and progression of human cancer.