Transcriptional Regulation of Oncogenic Protein Kinase Cϵ (PKCϵ) by STAT1 and Sp1 Proteins*

Background: PKCϵ, a kinase widely implicated in tumorigenesis and metastasis, is overexpressed in many cancers. Results: Transcription factors Sp1 and STAT1 control the expression of PKCϵ in cancer cells. Conclusion: Up-regulation of PKCϵ is mediated by dysregulated transcriptional mechanisms. Significance: Our results may have significant implications for the development of approaches to target PKCϵ and its effectors in cancer therapeutics. Overexpression of PKCϵ, a kinase associated with tumor aggressiveness and widely implicated in malignant transformation and metastasis, is a hallmark of multiple cancers, including mammary, prostate, and lung cancer. To characterize the mechanisms that control PKCϵ expression and its up-regulation in cancer, we cloned an ∼1.6-kb promoter segment of the human PKCϵ gene (PRKCE) that displays elevated transcriptional activity in cancer cells. A comprehensive deletional analysis established two regions rich in Sp1 and STAT1 sites located between −777 and −105 bp (region A) and −921 and −796 bp (region B), respectively, as responsible for the high transcriptional activity observed in cancer cells. A more detailed mutagenesis analysis followed by EMSA and ChIP identified Sp1 sites in positions −668/−659 and −269/−247 as well as STAT1 sites in positions −880/−869 and −793/−782 as the elements responsible for elevated promoter activity in breast cancer cells relative to normal mammary epithelial cells. RNAi silencing of Sp1 and STAT1 in breast cancer cells reduced PKCϵ mRNA and protein expression, as well as PRKCE promoter activity. Moreover, a strong correlation was found between PKCϵ and phospho-Ser-727 (active) STAT1 levels in breast cancer cells. Our results may have significant implications for the development of approaches to target PKCϵ and its effectors in cancer therapeutics.

Overexpression of PKC⑀, a kinase associated with tumor aggressiveness and widely implicated in malignant transformation and metastasis, is a hallmark of multiple cancers, including mammary, prostate, and lung cancer. To characterize the mechanisms that control PKC⑀ expression and its up-regulation in cancer, we cloned an ϳ1.6-kb promoter segment of the human PKC⑀ gene (PRKCE) that displays elevated transcriptional activity in cancer cells. A comprehensive deletional analysis established two regions rich in Sp1 and STAT1 sites located between ؊777 and ؊105 bp (region A) and ؊921 and ؊796 bp (region B), respectively, as responsible for the high transcriptional activity observed in cancer cells. A more detailed mutagenesis analysis followed by EMSA and ChIP identified Sp1 sites in positions ؊668/؊659 and ؊269/؊247 as well as STAT1 sites in positions ؊880/؊869 and ؊793/؊782 as the elements responsible for elevated promoter activity in breast cancer cells relative to normal mammary epithelial cells. RNAi silencing of Sp1 and STAT1 in breast cancer cells reduced PKC⑀ mRNA and protein expression, as well as PRKCE promoter activity. Moreover, a strong correlation was found between PKC⑀ and phospho-Ser-727 (active) STAT1 levels in breast cancer cells. Our results may have significant implications for the development of approaches to target PKC⑀ and its effectors in cancer therapeutics.
The serine-threonine kinase protein kinase C⑀ (PKC⑀), a phorbol ester receptor, has been widely implicated in numerous cellular functions, including cell cycle progression, cytokinesis, cytoskeletal reorganization, ion channel control, and transcription factor activity regulation (1)(2)(3)(4)(5)(6). This ubiquitously expressed kinase has been associated with multiple disease conditions, including obesity, diabetes, heart failure, neu-rological diseases, and cancer (7)(8)(9)(10). PKC⑀ is primarily activated by the lipid second messenger diacylglycerol (11), a product of phosphatidylinositol 4,5-bisphosphate hydrolysis by phospholipase C, which, like phorbol esters, binds to the C1 domains located in the N-terminal regulatory region. Receptors coupled to diacylglycerol generation, including tyrosine kinase and G-protein-coupled receptors, cause the intracellular mobilization of PKC⑀ to the plasma membrane and other intracellular compartments, where it associates with interacting partners and phosphorylates specific substrates (12).
It is widely recognized that distinct members of the diacylglycerol/phorbol ester-regulated PKCs act either as promoters or suppressors of growth and tumorigenesis (13,14). In that regard, work from several laboratories identified PKC⑀ as an oncogenic kinase and established important roles for this kinase in the development and progression of cancer. Early studies revealed that ectopic overexpression of PKC⑀ leads to malignant transformation in some cell types (11,15,16). PKC⑀ confers growth advantage and survival through the activation of Ras/Raf/ERK, PI3K/Akt, STAT3, and NF-B pathways (17,18). PKC⑀ also mediates resistance to chemotherapeutic agents and ionizing radiation, and inhibition of its activity or expression sensitizes cancer cells to cell death-inducing agents (19 -21). Most remarkably, PKC⑀ emerged as a cancer biomarker, as it is markedly up-regulated in most epithelial cancers (22,23). For example, the vast majority of prostate tumors, in particular those from advanced and recurrent patients, display elevated PKC⑀ levels (24). Prostate-specific PKC⑀ transgenic mice develop prostatic neoplastic lesions with elevated Akt, STAT3, and NF-B activity (17). Another remarkable example of PKC⑀ up-regulation is in lung cancer; the vast majority (Ͼ90%) of primary human non-small cell lung cancers show significant PKC⑀ overexpression compared with normal lung epithelium, and knockdown of PKC⑀ from non-small cell lung cancer cells impairs their ability to form tumors and metastasize in nude mice (25). Likewise, depletion of PKC⑀ from breast cancer cells impairs growth, tumorigenicity, and invasiveness. Accordingly, PKC⑀ up-regulation has been associated with poor disease-free and overall survival of breast cancer patients (22). More recently, a PKC⑀ ATP mimetic inhibitor was found to impair the growth of breast cancer cells in vitro and in vivo, highlighting the potential of PKC⑀ as a breast cancer therapeutic target (26). Regardless of the well accepted fact that disregulation in PKC⑀ expression plays a causative role in cancer progression, little is known regarding the mechanisms that control the expression of this pro-oncogenic and metastatic kinase. To our knowledge, the transcriptional mechanisms controlling the expression of the PRKCE promoter in humans or other species have not yet been studied. To characterize the regulation of PKC⑀ expression, we cloned a fragment of the promoter region of the human PRKCE gene and investigated the critical determinants controlling transcriptional activation of this gene. Our analysis revealed key cis-acting elements in the PRKCE promoter and candidate transcription factors, particularly Sp1 and STAT1, that contribute to PKC⑀ overexpression in breast cancer. Furthermore, we identified a self-controlled mechanism that significantly contributes to the up-regulation of PKC⑀ in breast cancer cells. Cancer cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) or RPMI 1640 medium supplemented with 10% FBS, L-glutamine (500 M), and penicillin/ streptomycin (100 units/100 g/ml). Normal immortalized MCF-10A, HBEC, and RWPE-1 cells were cultured as described previously (18,27). All cells were grown at 37°C in a humidified 5% CO 2 incubator.
Transient Transfection and Luciferase Assays-Cells in 12well plates (ϳ2 ϫ 10 5 cells/well) were co-transfected with 450 ng of a PRKCE promoter Firefly luciferase reporter vector and 50 ng of the Renilla luciferase expression vector (pRL-TK) using Lipofectamine 2000 (Invitrogen) or X-tremeGENEHP DNA transfection reagent (Roche Applied Science). After 48 h, cells were lysed with passive lysis buffer (Promega, Madison, WI). Luciferase activity was determined in cell extracts using the Dual-Luciferase TM reporter assay kit (Promega). Data were expressed as the ratio between Firefly and Renilla luciferase activities. In each experiment, the pGL3-positive control vector (Promega) was used as a control. Promoter activity of each PRKCE promoter luciferase reporter construct was expressed as follows: (Firefly (sample)/Renilla (sample))/(Firefly (positive)/Renilla (positive)) ϫ 100%.
RNA Interference-RNAi duplexes were transiently transfected using Lipofectamine RNAiMax. For transient depletion of PKC⑀, STAT1, and Sp1, we used ON-TARGET Plus RNAi duplexes purchased from Dharmacon (Waltham, MA). Silencer control RNAi from Ambion was used as a nontarget control. Twenty four h after RNAi delivery, cells were transfected with different luciferase reporters, and luciferase activity was determined 48 h later.
Real Time Quantitative PCR (qPCR) 2 -Total RNA was extracted from subconfluent cell cultures using the RNeasy kit (Qiagen, Valencia, CA). One g of RNA/sample was reversetranscribed using the TaqMan reverse transcription reagent kit (Applied Biosystems, Branchburg, NJ) with random hexamers used as primers. PCR primers and a 5Ј end 6-carboxyfluorescein-labeled probe for PKC⑀ were purchased from Applied Biosystems. PCR was performed using an ABI PRISM 7700 detection system in a total volume of 25 l containing TaqMan universal PCR MasterMix (Applied Biosystems), commercial target primers (300 nM), the fluorescent probe (200 nM), and 1 l of cDNA. PCR product formation was continuously monitored using the sequence detection system software version 1.7 (Applied Biosystems). The 6-carboxyfluorescein signal was normalized to endogenous tRNA 18 S or ubiquitin C. ⌬Ct was obtained by subtracting the circle threshold (CT) of tRNA 18 S or ubiquitin C from that of PKC⑀. ⌬(⌬Ct) was determined by subtracting the control ⌬Ct from the sample ⌬Ct. Fold-changes were expressed as (2) Ϫ⌬(⌬Ct) .
RNA Stability Assay-5 ϫ 10 5 cells seeded into 35-mm plates were treated with actinomycin D (2.5 g/ml) for 16 h. Total RNA from different cell lines was extracted at different times using TRIzol (Invitrogen). cDNA was synthesized using the TaqMan reverse transcription reagent kit (Applied Biosystems). PKC⑀ mRNA levels were determined by qPCR as described above. For each cell line, mRNA levels at time ϭ 0 h was set as 100%.
In Silico PKC⑀ mRNA Profiling in Breast Cancer Cells-Analysis of PRKCE gene expression in breast cancer was done from three independent studies (GSE10843, GSE12777, and GSE41445) using inSilicoDb and inSilicoMerging R/Bioconductor packages (29). These gene expression profiles were developed using the Affymetrix HG -U133 Plus2 platform (GPL570). Briefly, the frozen RMA preprocessed expression profiles of these studies were downloaded from the InSilico database and merged using the COMBAT algorithm as the batch removal method. Visualization and statistical analysis of PKC⑀ expression profile were done with R.
Analysis of Methylation of the PRKCE Promoter-The presence of CpG islands in the human PRKCE promoter (NC_000002.11) was determined using the Methyl Primer Express software (Applied BioSystems). For the analysis of PKC⑀ mRNA expression after demethylation, MCF-10A cells were treated with different concentrations (1-100 M) of 5-aza-2Ј-deoxycytidine (96 h or 7 days) and/or trichostatin A (100 ng/ml, 24 h). Total mRNA was extracted, and PKC⑀ mRNA levels were determined by qPCR as described above.
Cell Migration Assay-Cell migration was determined with a Boyden chamber, as described previously (31). Briefly, MCF-7 cells (3 ϫ 10 4 cells/well) were seeded in the upper compartment of a Boyden chamber (NeuroProbe). A 12-m pore polycarbonate filter (NeuroProbe) coated overnight with type IV collagen in cold PBS was used to separate the upper and lower compartments. In the lower chamber, 0.1% BSA/DMEM with or without FBS (5%) was used. After 24 h of incubation at 37°C, nonmigrating cells on the upper side of the membrane were wiped off the surface, and migrating cells on the lower side of the membrane were fixed, stained with DIFF Quik Stain Set (Dade Behring), and counted by contrast microscopy in five independent fields.
Statistical Analysis-Results are the means Ϯ S.E. of at least three individual experiments. Student's t test was used for statistical comparison. A p value Ͻ 0.05 was considered statistically significant.

Overexpression of PKC⑀ in Breast Cancer Cells and Initial
Characterization of the PRKCE Promoter-PKC⑀, a kinase broadly implicated in tumorigenesis and metastasis, is overexpressed in multiple cancers. Elevated PKC⑀ levels have been associated with poor outcome in prostate, breast, lung, and head and neck cancer (22,24,32,33); however, the mechanisms behind the control of PKC⑀ expression remain to be established. A comparative analysis of PKC⑀ protein levels by Western blot shows that this kinase is overexpressed in multiple breast cancer cell lines (MCF-7, T-47D, BT-474, HCC-1419, MDA-MB-231, MDA-MB-453, and MDA-MB-468 cells) relative to MCF-10A cells, an immortalized nontumorigenic mammary cell line (Fig. 1A). qPCR assays also revealed significantly higher PKC⑀ mRNA levels in breast cancer cells compared with MCF-10A cells (Fig. 1B). To determine whether overexpression of PKC⑀ is associated with altered mRNA stability, we assessed mRNA levels at different times after treatment with the transcriptional inhibitor actinomycin D. As shown in Fig. 1C, the decay in mRNA levels is essentially the same in breast cancer cell lines (MCF-7, T-47D, and MDA-MB-453) and MCF-10A cells. Thus, the differential expression of PKC⑀ may involve a dysregulation of transcriptional mechanisms. Likewise, and in agreement with previous studies (18,27), PKC⑀ is overexpressed in lung and prostate cancer cell lines relative to corresponding normal "nontransformed" cell lines (Fig. 1A).
To investigate the transcriptional mechanisms involved in PKC⑀ expression, we cloned a 2.1-kb fragment of the human PRKCE gene from genomic DNA using PCR. This fragment includes 1933 bp of the putative PRKCE promoter as well as 219 bp after the putative transcription start site. We also cloned four fragments encompassing shorter regions of the putative PRKCE promoter (1416/ϩ219 bp, 808/ϩ219 bp, 320/ϩ219 bp, and 105/ϩ219 bp, respectively). The different DNA fragments were subcloned into the pGL3-enhancer luciferase reporter vector to generate the plasmids pGL3Ϫ1933/ϩ219, pGL3Ϫ1416/ϩ219, pGL3Ϫ808/ϩ219, pGL3Ϫ320/ϩ219, and pGL3Ϫ105/ϩ219. Plasmids were transiently transfected into MCF-7 breast cancer cells along with pRL-TK (Renilla luciferase vector) for normalization of transfection efficiency. The pGL3Ϫ1416/ϩ219 reporter construct exhibited the highest luciferase activity, which was ϳ40 times higher than pGL3enhancer empty vector, therefore confirming that it possesses functional PRKCE promoter activity. A progressive loss in luciferase activity was observed upon deletions of fragments Ϫ1416/ Ϫ809, Ϫ1416/Ϫ321, and Ϫ1416/Ϫ106. A significant loss of promoter activity was also observed with pGL3Ϫ1933/ϩ219, suggesting repressive transcriptional elements within the Ϫ1933/Ϫ1417 bp region (Fig. 1D). A comparison of PRKCE promoter activity in different cell lines using pGL3Ϫ1416/ ϩ219 revealed a manifest elevation in luciferase activity in breast cancer cells relative to normal immortalized MCF-10A cells. Similarly, lung and prostate cancer cell lines exhibited higher promoter activity than the corresponding nontumorigenic counterparts (Fig. 1E).
A comparative analysis of PRKCE gene expression in 48 breast cancer cell lines (24 luminal-like and 24 basal-like) obtained from three independent studies (GSE10843, GSE12777, and GSE41445) was performed using inSilicoDb and inSilicoMerging R/Bioconductor packages (29). This analysis showed no statistically significant differences between luminal and basal-like breast cancer cell lines (p ϭ 0.673) (Fig. 1F).
Differential Expression of PKC⑀ Is Not Related to Promoter Methylation-It is well established that epigenetic mechanisms control the expression of key oncogenic and tumor-suppressing proteins. To determine whether methylation of the PRKCE promoter could be implicated in the differential expression between normal mammary and breast cancer cells, we first examined if the promoter was rich in CpG islands using the Methyl Primer Express software (Applied Biosystems). This analysis revealed two regions in the PRKCE promoter that were very rich in CpG islands, a proximal region between Ϫ2.6 and ϩ0.9 kb and a distal region between Ϫ8.9 and Ϫ7.7 kb ( Fig. 2A).
To determine whether the reduced PKC⑀ expression in MCF-10A cells could be due to promoter methylation, we used the demethylating agent 5-aza-2Ј-deoxycytidine (AZA). qPCR analysis revealed that PKC⑀ mRNA levels remain essentially unchanged in MCF-10A cells treated with different concentrations of AZA, either in the presence or absence of the HDAC inhibitor trichostatin A (Fig. 2B). A similar treatment in MCF-10A cells caused a significant rescue in the expression of the oncogenic protein P-Rex1, a gene that is regulated by methyla-tion. 3 Therefore, overexpression of PKC⑀ in breast cancer cells does not seem to be related to demethylation of the PRKCE gene promoter.
Identification of Key Transcriptional Regions in the Human PKC⑀ Promoter-To characterize the human PRKCE promoter in more detail and to identify positive regulatory elements responsible for transcriptional activation, a series of 5Ј-unidirectional deletions was generated from the pGL3Ϫ1416/ ϩ219 luciferase reporter vector using the Erase-a-Base system. The resulting constructs were transfected into MCF-7 cells, and luciferase activity was determined. Fig. 3 shows that promoter activities of pGL3Ϫ1319/ϩ219, pGL3Ϫ1224/ ϩ219, pGL3Ϫ1121/ϩ219, pGL3Ϫ1032/ϩ219, pGL3Ϫ1028/ ϩ219, and pGL3Ϫ921/ϩ219 constructs were essentially similar to that of pGL3Ϫ1416/ϩ219. However, a significant reduction in transcriptional activity was observed upon serial deletions starting from bp Ϫ887. Indeed, pGL3Ϫ887/ϩ219, pGL3Ϫ873/ϩ219, and pGL3Ϫ819/ϩ219 display 77, 58, and 37% activity, respectively, compared with that of pGL3Ϫ1416/ ϩ219. No additional changes in reporter activity were observed with pGL3Ϫ808/ϩ219. Constructs pGL3Ϫ796/ϩ219 and pGL3Ϫ777/ϩ219 display slightly lower luciferase activity than pGL3Ϫ808/ϩ219. Luciferase activity drops significantly with constructs pGL3Ϫ320/ϩ219 (91% reduction) and pGL3Ϫ105/ϩ219 (98% reduction). To summarize these initial observations, the deletional analysis delineated two prominent regions in the PRKCE promoter containing positive regulatory elements that we defined as region A (Ϫ777 to Ϫ105 bp) and region B (Ϫ921 to Ϫ796 bp). In subsequent sections, a more detailed characterization of the cis-acting elements in these two regions will be shown.
Analysis of Region A Revealed a Crucial Role for Sp1 in PKC⑀ Transcription-To identify putative transcriptional elements in region A of the PRKCE promoter, we initially used the PROMO software. This analysis revealed the presence of seven putative Sp1-responsive elements that we named Sp1-1 (the most distal site, bp Ϫ716 to Ϫ707) to Sp1-7 (the most proximal site, bp Ϫ256 to Ϫ247) (Fig. 4A, left panel). The putative Sp1-binding sequences are shown in Fig. 4A, right panel. To define the relevance of the different Sp1-binding sites, additional truncated mutants for region A were generated using pGL3Ϫ777/ϩ219 as a template (pGL3Ϫ644/ϩ219, pGL3Ϫ531/ϩ219, and pGL3Ϫ401/ϩ219), and we examined for their luciferase activity upon transfection into MCF-7 cells. Fig. 4B shows that deletion of region comprising bp Ϫ777 to Ϫ664 (which includes Sp1-1 and Sp1-2 sites) caused a 65% drop in luciferase activity. No additional changes in reporter activity were observed upon deletions of regions comprising bp Ϫ644/ Ϫ532, Ϫ644/Ϫ402, and Ϫ644/Ϫ321, which include sites Sp1-3, Sp1-4, and Sp1-5. However, when fragment Ϫ320/Ϫ105 (which includes Sp1-6 and Sp1-7) was deleted, an additional reduction in luciferase activity was observed. These results suggest that multiple Sp1 sites in region A contribute to the transcriptional activity of the PRKCE promoter.  To further determine the contribution of the different Sp1 sites in the transcriptional activation of the PRKCE promoter, we performed site-directed mutagenesis of these sites in the context of the pGL3Ϫ777/ϩ219 construct. Essential residues GGCG in Sp1 sites were mutated to TTAT, and luciferase activities of the corresponding constructs were determined after transfection into MCF-7 cells. As shown in Fig. 4C, mutation of Sp1-1 in pGLϪ777/ϩ219 had no effect; however, mutation of Sp1-2 caused a 62% reduction in reporter activity. Sp1-6 and Sp1-7 were only 4 bp apart, and therefore we decided to mutate them together. When we mutated Sp1-6/7 in pGL3Ϫ777/ ϩ219, a significant reduction (50%) in luciferase activity was observed. We further mutated Sp1-6/7 sites in pGL3Ϫ320/ ϩ219, and observed a significant reduction in reporter activity compared with the wild-type pGL3Ϫ320/ϩ219 construct. However, it did not reach complete inhibition, thus arguing for the presence of other relevant transcriptional element(s) within the Ϫ320/Ϫ105 region that remain to be identified. The deletional and mutational analyses of region A indicate that multiple Sp1 sites control the transcriptional activation of the PRKCE promoter.
To confirm the relevance of the Sp1-binding sites in transcriptional activation of the PRKCE gene, we used a number of additional approaches. First, we examined the effect of mithramycin A (MTM), an agent that prevents binding of Sp1 to its transcription binding site (34,35). As shown in Fig. 4D, MTM markedly reduced luciferase activity of reporters pGL3Ϫ777/ ϩ219 and pGL3Ϫ320/ϩ219. As a second approach, and to address whether Sp1 proteins associate with the PRKCE promoter in vivo, we performed a chromatin immunoprecipitation (ChIP) assay using an anti-Sp1 antibody. As a negative control, we used IgG. Three sets of primers were utilized in these experiments as follows: one encompassing bp Ϫ772 to Ϫ615 (for site Sp1-2); a second encompassing bp Ϫ320 to Ϫ186 (for Sp1-6 and Sp1-7), and a third for bp Ϫ443 to Ϫ286 (for site Sp1-5). Sp1 immunoprecipitation revealed the expected bands for regions Ϫ772/Ϫ615 and Ϫ320/Ϫ186, and no band was observed for region Ϫ443/Ϫ286 (Fig. 4E). Thus, the Sp1 transcription factor binds in vivo to the sites identified in our deletional/mutational analysis. Finally, to confirm the involvement of Sp1, we knocked down this transcription factor using RNAi. Sp1 RNAi depletion from MCF-7, T-47D, MDA-MB-231, and BT-474 breast cancer cell lines significantly reduced the expression of PKC⑀ protein (Fig. 4F) and PKC⑀ mRNA, as determined by qPCR (Fig. 4G). Altogether, these results demonstrate the relevance of Sp1 in transcriptional activation of the PRKCE promoter. Fig. 3, region B located between bp Ϫ921 and Ϫ796 plays a positive role in transcriptional activation of the PRKCE promoter. Analysis using the PROMO program revealed two putative STAT1 sites in this region, which we named STAT1-1 (Ϫ916 to Ϫ905 bp) and STAT1-2 (Ϫ880 to Ϫ869 bp). There is also a third STAT1 site (STAT1-3) at the edge of region B (Ϫ793 to Ϫ782 bp) (Fig. 5A). To determine the potential relevance of these sites, essential residues TTTCC in STAT1 sites were mutated to T¡C in pGL3Ϫ921/ϩ219. The resulting mutant constructs were transfected into MCF-7 cells and assessed for their luciferase reporter activity. As shown in Fig.  5B, mutation of the most distal STAT1 site (STAT1-1) had no significant effect on luciferase activity. Conversely, mutation of STAT1-2 site caused a 44% reduction in reporter activity. A slight, yet statistically significant reduction in luciferase activity was observed upon mutation of the STAT1-3 site. A double mutant for STAT1-2 and STAT1-3 sites was generated, and its activity was examined in MCF-7 cells, which revealed a 61% reduction in luciferase activity compared with the pGL3Ϫ921/ ϩ219 construct. Therefore, the STAT1-2 and STAT1-3 sites are involved in the regulation of PKC⑀ promoter activity.

STAT1-binding Sites in Region B Control PKC⑀ Transcriptional Activation-As established in the deletional analysis shown in
The program PROMO also identified two additional STAT1 sites outside region B, which were named STAT1-4 (Ϫ401 to Ϫ390 bp) and STAT-5 (Ϫ227 to Ϫ216 bp). These two sites were actually located within the region A and in close proximity to Sp1 sites (Fig. 5A). We mutated STAT1-4 and STAT1-5 sites and found these mutations do not alter reporter activity (Fig.  5B), suggesting that only STAT1-2 and STAT1-3 sites are involved in transcriptional control of the PRKCE promoter in breast cancer cells.
Next, to confirm the relevance of STAT1 in the control of PKC⑀ transcriptional activity, we used RNAi (Fig. 5C). MCF-7 cells were transfected with a STAT1 SMARTpoolRNAi, which caused Ͼ90% depletion in STAT1 levels (Fig. 5C, inset), or a SMARTpool control RNAi and then transfected with the pGL3Ϫ921/ϩ219 luciferase reporter vector. As expected from the deletional and mutational analyses, silencing STAT1 inhibited transcriptional activity of the PKC⑀ reporter (54% reduction, which is in the same range as the reduction in activity observed upon mutation of STAT1-2 and STAT1-3 sites combined, see Fig. 5B). Moreover, when we assessed the activity of the STAT1-2/3-mutated pGL3Ϫ921/ϩ219 construct, STAT1 RNAi depletion failed to cause an additional reduction in luciferase activity (Fig. 5C), thus confirming the importance of STAT1-2 and STAT1-3 sites in the control of PRKCE promoter activity. To further confirm the relevance of the STAT1 sites, we used ChIP. For this analysis, we used a set of primers encompassing Ϫ949 to Ϫ751 bp in the PRKCE promoter, a region that includes both STAT1-2-and STAT1-3-binding sites. Results shown in Fig. 5D revealed a band of the expected size (199 bp) when an anti-STAT1 antibody was used in the immunoprecipitation, whereas no band was observed using control IgG, thus suggesting direct binding of STAT1 to the Ϫ949 to Ϫ751-bp promoter region. Furthermore, STAT1 RNAi depletion from MCF-7 cells caused a significant reduction in PKC⑀ mRNA (Fig. 5E) and protein levels (Fig. 5F). Altogether, these results indicate that STAT1-2-and STAT1-3-binding sites are involved in the transcriptional control of the PRKCE promoter. An additive effect between STAT1 RNAi depletion and MTM treatment was observed (Fig. 5F).
STAT1 and Sp1 Contribute to the Elevated PKC⑀ Transcriptional Activity in Breast Cancer Cells-Once we identified relevant Sp1 and STAT1 sites in the PRKCE promoter, we asked if these sites mediate PKC⑀ up-regulation in breast cancer cells relative to nontumorigenic mammary cells. To address this issue, we compared the activities of the different deleted reporters between MCF-7 versus MCF-10A cells. As shown previously in Fig. 1E with reporter pGL3Ϫ1416/ϩ219, activity of pGL3Ϫ921/ϩ219 reporter was also higher in MCF-7 cells relative to MCF-10A cells (Fig. 6A). Deletion of fragment Ϫ921 to Ϫ777 bp, which includes STAT1-2/3 sites in region B, diminished luciferase activity in MCF-7 cells by 61%, an effect that was not seen in MCF-10A cells (Fig. 6, A and B). To verify the relevance of the STAT1 sites in PKC⑀ up-regulation in breast cancer cells, we compared the activity of pGL3Ϫ921/ϩ219 (wild type) versus pGL3Ϫ921/ϩ219 (STAT-2/3-mutated) in MCF-7 and MCF-10A cells. Whereas mutation of STAT1-2 and STAT1-3 sites failed to reduce reporter activity in MCF-10A cells, a marked reduction in activity (ϳ70% reduction) was observed in MCF-7 cells (Fig. 6C) as well as in T-47D cells (data not shown). To validate the relevance of the STAT1-2/3 sites in PKC⑀ up-regulation, we used an EMSA approach. Nuclear extracts from MCF-10A, MCF-7, or T-47D cells were incubated with 25-bp double-stranded radiolabeled probes for either the STAT1-2 site or a standard STAT1 binding consensus. As shown in Fig. 6D, a shift protein-DNA complex band was detected after incubation of nuclear extracts from either probe both in MCF-7 (lanes 3 and 6) and T-47D cells (lanes 4  and 7). However, this effect was not seen in nontumorigenic MCF-10A cells (Fig. 6D, lanes 2 and 5). The shift band was competed by co-incubation with an excess (50-fold molar) of   unlabeled probes for either STAT1-2 (Fig. 6D, lane 8) or a standard STAT1-binding consensus sequence (lane 9) but not with an excess unlabeled probe for AP-1 (lane 10), thereby confirming the specificity of the interaction. A similar result was observed using a probe for site STAT1-3 (data not shown). Thus, STAT1-2/3 sites contribute to the up-regulation of PKC⑀ transcriptional activity in breast cancer cells. Next, we carried out similar experiments to determine whether the Sp1-2 site was implicated in PKC⑀ up-regulation in breast cancer cells relative to nontumorigenic MCF-10A cells. As shown in Fig. 6A, deletion of fragment Ϫ777 to Ϫ531 bp, which includes relevant Sp1-2 site in region A (position Ϫ668 to Ϫ659), reduced luciferase reporter activity in MCF-7 cells but not in MCF-10A cells. No additional changes were found upon deletion of region Ϫ531 to Ϫ320 bp in either cell line. To verify the relevance of the Sp1-2 site in PKC⑀ up-regulation in breast cancer cells, we compared the activity of pGL3Ϫ777/ ϩ219 (wild type) versus pGL3Ϫ777/ϩ219 (Sp1-2-mutated) in MCF-7 and MCF-10A cells. Fig. 7A shows that mutation of Sp1-2 significantly reduced luciferase activity in MCF-7 cells, whereas this mutation had no effect in MCF-10A cells. As expected, mutation of the Sp1-1 site, which was dispensable for transcriptional activity (see Fig. 4C), did not alter reporter activity in MCF-7 or MCF-10A cells. To further verify the relevance of the Sp1-2 site in PKC⑀ up-regulation in breast cancer, we used an EMSA approach. Nuclear extracts from MCF-10A, MCF-7, or T-47D cells were incubated with radiolabeled probes for either the Sp1-2 site or a standard Sp1 binding consensus. As shown in Fig. 7B, a shift protein-DNA complex band was detected after incubation of nuclear extracts from either probe both in MCF-7 (lanes 3 and 6) and T-47D cells (lanes 4 and 7) but not in nontumorigenic MCF-10A cells (lanes 2 and  5). The specificity of the interaction was confirmed by competition of the shift band with an excess (50-fold molar) of unlabeled probes for either Sp1-2 (Fig. 7B, lane 8) or a standard Sp1 binding consensus (lane 9) but not with an unlabeled probe for AP-1 (lane 10).

S T A T 1 -1 S T A T 1 -2 S T A T 1 -4 S T A T 1 -5 S T
We also found that deletion of fragment Ϫ320 to Ϫ105 bp, which comprises proximal Sp1-binding sites (Sp1-6/7), essentially abolished luciferase activity both in MCF-7 and MCF-10A cells (see Fig. 6A). Mutation of Sp1-6/7 sites significantly reduced the activity of the pGL3Ϫ320/ϩ219 reporter in MCF-7 and MCF-10A cells (Fig. 7C), suggesting that Sp1-6/7 may control constitutive expression both in normal and cancer cells. The large drop in activity by deletion of fragment Ϫ320 to Ϫ105 bp compared with the mutation of Sp1-6/7 sites (Fig. 6A see also Fig. 3) argues for additional elements in this region controlling basal promoter activity.
PKC⑀ Controls Its Own Expression in Breast Cancer Cells-There is evidence that PKC⑀ controls the phosphorylation status and activity of STAT1 in several cellular models (36 -38). Ser-727 phosphorylation in STAT1 is required for its maximal transcriptional activity (39). Likewise, we found that PKC⑀ controls the activation status of STAT1 in breast cancer cells, as judged by the reduction in phospho-Ser-727-STAT1 levels upon PKC⑀ depletion in MCF-7, T-47D, MDA-MD-231, MDA-MB-453, and MDA-MB-468 breast cancer cell lines (Fig.  8A). Similar results were observed in prostate and lung cancer models (data not shown). Treatment of MCF-7 cells with the pan-PKC inhibitor GF 109203X or the specific PKC⑀ inhibitor ⑀V1-2 also reduced phospho-Ser727-STAT1 levels (Fig. 8B). Given our finding that STAT1 transcriptionally regulates PKC⑀ expression, we speculated that PKC⑀ controls its own expression via STAT1. Treatment of MCF-7 cells with ⑀V1-2 ( Fig. 8C) or GF 109203X (data not shown) significantly reduced pGL3Ϫ1416/ϩ219 luciferase reporter activity. To examine the potential involvement of PKC⑀ in controlling its own promoter activity, we used PKC⑀ RNAi. PKC⑀ expression was silenced from MCF-7 cells by Ͼ90% upon delivery of two different PKC⑀ RNAi duplexes (⑀1 and ⑀2), as we did previously in other models (18,25). Notably, luciferase activity of the pGL3Ϫ1416/ϩ219 reporter was significantly decreased in PKC⑀-depleted MCF-7 cells (Fig. 8D), indicating that the elevated levels of PKC⑀ in breast cancer cells positively control its own expression at a transcriptional level. The results described above argue for a mutual dependence between PKC⑀ expres- sion and STAT1 activation. We decided to formally test this hypothesis in mammary cellular models (Fig. 8E). We observed that normal immortalized MCF-10A cells, which express low PKC⑀ levels, display low levels of phospho-Ser-727-STAT1. Western blots, we found a strong correlation between PKC⑀ and phospho-Ser-727-STAT1 levels (R 2 ϭ 0.90) (Fig. 8F). Altogether, these results argue for a positive feedback between PKC⑀ expression and STAT1 activation in breast cancer cells. PKC⑀ Mediates Migration of Breast Cancer Cells-PKC⑀ has been implicated in tumor initiation, progression, and metastasis (22,25,27). Fig. 9A shows that PKC⑀ RNAi depletion significantly reduced the motility of cells in response to 5% FBS, as determined with a Boyden chamber. The Sp1 inhibitor MTM, which significantly reduces PKC⑀ expression (Fig. 9B, see also    Figs. 4F and 5F) also significantly impaired MCF-7 cell migration (Fig. 9A). Adenoviral overexpression of PKC⑀ overcame the effect of PKC⑀ RNAi on cell migration. The impaired cell migration caused by MTM could be partially restored by adenoviral overexpression of PKC⑀, thus arguing that the expression levels of PKC⑀ are crucial for the ability of breast cancer cells to migrate.

DISCUSSION
PKC⑀, a member of the novel PKCs, has been extensively characterized as a mitogenic/survival kinase that activates pathways linked to malignant transformation and metastasis, including Ras/Raf/Erk, PI3K/Akt, and NF-B (17,18). Pharmacological inhibition or RNAi silencing of PKC⑀ expression impairs the ability of cancer cells to form tumors in nude mice and metastasize to distant sites (22). Overexpression of PKC⑀ in nontransformed cells confers growth/survival advantage or leads to malignant transformation (16). In an in vivo scenario, transgenic overexpression of PKC⑀ in the mouse prostate leads to a preneoplastic phenotype, and skin transgenic overexpression of this kinase leads to the development of metastatic squamous carcinoma (40). Therefore, there is significant evidence that overexpression of PKC⑀ is causally associated with the development of a malignant and metastatic phenotype. This is highly relevant in the context of human cancer, as a vast majority of cancers displays PKC⑀ up-regulation, including breast, prostate, and lung cancer (18,22,25). Increased PKC⑀ expression in breast cancer correlates with high histological grade, positive ErbB2/Her2 status, and hormone-independent status (22). Despite the wealth of functional information regarding PKC and cancer, both in vitro and in vivo, as well as the established mechanistic links with proliferative pathways, the causes behind the up-regulation of PKC⑀ in human cancer remained elusive.
In this study we report that PKC⑀ up-regulation in breast cancer cells occurs through dysregulation of transcriptional mechanisms. An ϳ1.6-kb fragment of human genomic DNA encompassing the 5Ј-flanking region and part of the first exon (Ϫ1.4 to ϩ0.2 kb) of the PRKCE gene was isolated and cloned into a luciferase reporter vector. This fragment displayed significantly higher transcriptional activity when expressed in breast cancer cells relative to normal immortalized MCF-10A cells. However, the elevated PKC⑀ mRNA levels in breast cancer cells do not seem to be related to changes in mRNA stability. Our deletional and mutagenesis studies combined with in silico analysis identified key positive regulatory cis-acting Sp1 and STAT1 elements in two regions (regions A and B) that we defined as responsible for the up-regulation of PKC⑀ transcriptional activation in breast cancer cells, and their functional relevance was confirmed by EMSA and ChIP. A region that negatively regulates transcription located upstream from the 1.6-kb fragment, specifically between Ϫ1.4 and Ϫ1.9 kb, was also identified. Studies to dissect and characterize these negative elements are underway.
From the seven putative Sp1-responsive elements located in region A of the PRKCE gene, only one located between bp Ϫ668 and Ϫ659 contributes to the differential overexpression of PKC⑀ in MCF-7 cells. The two most proximal Sp1 sites located in positions Ϫ269/Ϫ260 and Ϫ256/Ϫ247 contribute to transcriptional activation of the PRKCE gene both in MCF-7 and MCF-10A cells, suggesting that these sites control basal expression both in normal and cancer cells. The Sp1 transcription factor has been widely implicated in cancer and is up-regulated in human tumors. For example, it has been reported that Sp1 protein and binding activity are elevated in human breast carcinoma (41,42). Sp1 is highly expressed both in estrogen receptor-positive and -negative cell lines (43), and its depletion using RNAi leads to reduced G 1 /S progression of breast cancer cells (44). Sp1 controls the expression of genes implicated in breast tumorigenesis and metastatic dissemination, including ErbB2 (45), EGF receptor (46), IGF-IR (47,48), VEGF (49,50), cyclin D1 (51), and urokinase-type plasminogen activator receptor (42). The transcription factor Sp1 binds to GC-rich motifs in DNA, and DNA methylation of CpG islands can inhibit Sp1 binding to DNA (52)(53)(54). Nevertheless, our studies show that the demethylating agent AZA could not up-regulate PKC⑀ mRNA levels in MCF-10A cells. Thus, despite the presence of CpG-rich regions in the PRKCE promoter, repression by methylation does not seem to take place in normal mammary cells. It is interesting that a recent study in ventricular myocytes showed PRKCE gene repression through methylation of Sp1 sites via reactive oxygen species in response to norepinephrine or hypoxia (55,56), suggesting that epigenetic regulation of the PRKCE gene can take place in some cell types under specific conditions. Notably, functional Sp1-binding sites have been identified in the promoters of PKC␤ and PKC␦ isozymes, and Sp1 binding to the PKC␤ gene is repressed by hypermethylation and re-expressed by AZA treatment (57,58). The most notable characteristic of region B in the PRKCE promoter is the presence of three STAT1-binding sites. Two of those sites located in position Ϫ880/Ϫ869 and Ϫ793/Ϫ782 are functionally relevant in breast cancer cells. Indeed, a marked reduction (Ͼ50%) of promoter activity was observed upon mutation of these sites. Moreover, STAT1 RNAi caused a significant reduction in PKC⑀ mRNA and protein levels. The elevated PKC⑀ levels in breast cancer cell lines strongly correlate with the activation status of STAT1. Activation of STAT transcription factors involves the phosphorylation of tyrosine residues either by JAK or independently of JAK by tyrosine kinase receptors such as EGF receptor (59). To date, the role of STAT1 in cancer progression remains controversial. Based on its canonical role in IFN-␥ signaling and loss of function studies using STAT1 knock-out mice, it has been postulated that STAT1 acts as a tumor suppressor (60). However, a large number of studies link STAT1 with tumor promotion as well as with resistance to chemotherapy and radiotherapy. Moreover, STAT1 is up-regulated and/or hyperactive in many cancers, including breast cancer (61,62). STAT1 up-regulation in human breast cancer is associated with metastatic dissemination and poor outcome in patients (62)(63)(64). In addition, STAT1 overexpression has been linked to aggressive tumor growth and the induction of proinflammatory factors, whereas STAT1 knockdown delays tumor progression (61). Inhibition of STAT1 in breast cancer prevents the homing of suppressive immune cells to the tumor microenvironment and enables immune-mediated tumor rejection (61). ErbB receptor activation, a common event in human breast cancer, significantly enhances STAT1 expression (65). In other models, such as melanoma, suppression of STAT1 expression reduces cell motility, invasion, and metastatic dissemination (66). STAT1 expression correlates with resistance to chemotherapeutic agents such as doxorubicin, docetaxel, and platinum compounds and is elevated in resistant tumors (67)(68)(69)(70)(71)(72). STAT1 also promotes radioresistance of breast cancer stem cells (73). Notably, PKC⑀ has been linked to chemo-and radio-resistance (19,20); thus, it is conceivable that PKC⑀ up-regulation mediated by STAT1 may play a role in this context. The fact that PKC⑀ controls its own expression in breast cancer cells suggests the possibility of a vicious cycle that contributes to the overexpression of this kinase. It is unclear at this stage what pathways are controlled by PKC⑀ that lead to its own transcriptional activation. One possibility is that PKC⑀ controls the expression of factors that influence STAT1 activation status, such as growth factors or cytokines that signal via this transcription factor.
In summary, this study identified relevant mechanisms that control PKC⑀ expression in breast cancer cells. As PKC⑀ overexpression has been linked to an aggressive phenotype and metastatic dissemination, our study may have significant therapeutic implications. In this regard, several studies suggested that targeting PKC⑀ could be an effective anticancer strategy. Indeed, the PKC⑀ translocation inhibitor ⑀V1-2 has anti-tumorigenic activity in non-small cell lung cancer and head and neck squamous cell carcinoma models (25,27). More recently, an ATP mimetic inhibitor with selectivity for PKC⑀ was shown to impair the growth of MDA-MB-231 breast cancer xenografts in mice as well as to reverse Ras-driven and epithelial-mesenchymal transition-dependent phenotypes in breast cancer cells (26). Thus, targeting PKC⑀ or the mechanisms responsible for its up-regulation in tumors may provide novel means for the treatment of cancer types driven by PKC⑀ overexpression.