p21-activated Kinase-1 Signaling Mediates Cyclin D1 Expression in Mammary Epithelial and Cancer Cells*

p21-activated kinase 1 (Pak1) has been shown recently to induce hyperplasia in the mammary epithelium, a phenotype also manifested by overexpression of cyclin D1, a known indicator of the proliferative stage. Here we investigated the role of the Pak1 pathway in the expression of cyclin D1 using tissue culture models and transgenic mice expressing activated Pak1 in mammary glands. We found that hyperplastic mammary glands from catalytically active Pak1 transgenic mice exhibit a 5- to 7-fold increased expression of cyclin D1 as compared with stage-matched wild-type mice. In addition, Pak1 levels were elevated in human breast tumors and also correlated well with increased cyclin D1 expression. Increased expression of Pak1 in breast cancer cells stimulated cyclin D1 promoter activity, elevated levels of cyclin D1 mRNA, protein, and nuclear accumulation of cyclin D1. Conversely, Pak1 inhibition by an auto-inhibitory peptide (amino acids 83–149) or Pak1 knock-down by short interference RNA markedly reduced the expression of cyclin D1, suggesting a requirement of a functional Pak1 pathway for optimal expression of cyclin D1. Results from deletion and mutant analysis indicate that Pak1 regulates cyclin

The small GTPases, including Cdc42 and Rac1, have been implicated in the regulation of mammalian cell morphology and motility (1). More specifically, Rac1 induces cortical actin polymerization, which is seen as membrane ruffling and lamellipodia, and Cdc42 induces the formation of peripheral actin microspikes and filopodia (2)(3)(4). The small GTPases regulate the formation of cytoskeletal structures by means of a family of serine/threonine kinases known as p21-activated kinases (Paks). 1 Activation of Pak1 is accompanied by the disassembly of stress fibers and focal adhesion complexes, as well as by maintenance of the integrity of the motile leading edge (5,6). Pak1 is activated by a number of extracellular signals, including heregulin-␤1 and epidermal growth factor, which are potent inducers of Pak1 activity and cell motility of breast cancer cells (7,8). Activation of Pak1 involves autophosphorylation of several sites, including Thr-423 within the auto-inhibitory loop of the kinase (9). Accordingly, Pak1 phosphorylation at Thr-423 has been linked with its activation, as substitution of the acidic residue glutamic acid at this site yields a constitutively active T423E Pak1 enzyme (9,10). In addition to its cytoskeleton effects, Pak1 also activates c-Jun NH 2 -terminal kinase and extracellular signal-regulated kinase kinases, and thus influences nuclear signaling (11,12).
Several recent studies have suggested that, in addition to cell motility, Pak1 is also involved in breast cancer progression. Adam et al. (13) have shown a mechanistic role for Pak1 activation in the increased cell invasion of breast cancer cells by heregulin. Furthermore, expression of a kinase-dead Pak1 mutant in the highly invasive breast cancer cell lines MDA-MB-435 and MDA-MB-231 led to stabilization of stress fibers, enhanced cell spreading, and reduction in invasiveness (14). Conversely, hyperactivation of the Pak1 pathway by conditional expression of catalytically active T423E Pak1 in the non-invasive breast cancer cell line MCF-7 promotes cell migration and anchorage-independent growth (15). Furthermore, increased Pak1 activity correlates well with the invasiveness of human breast cancer cells and tumors (15). Emerging data suggest that Pak1 may be overexpressed in human cancers. For example, Pak1 gene amplification has also been reported in ovarian (16) and breast (17) cancers. Furthermore, Pak1 protein has been shown to be up-regulated in ovarian tumors (16) and breast cancer (15,18, and this study).
More recently, Pak1 has been shown to directly phosphorylate estrogen receptor-␣ (ER) at Ser-305 and to promote its transactivation functions (19). Additionally, expression of kinase-active T423E Pak1 transgene in mammary glands induces hyperplasia in the mammary epithelium (19), a phenotype manifested by several other oncogenes including cyclin D1 (20).
Overexpression of cyclin D1 has been noted in over 50% of human breast tumors of all histological types (20 -22). Cyclin D1 overexpression is found at the earliest stages of breast cancer progression such as ductal carcinoma in situ and maintained in all stages of the metastasis (23). Accordingly, overexpression of murine mammary tumor virus-cyclin D1 in mammary glands leads to breast cancer (20). The expression of cyclin D1 is regulated by diverse signaling cascades. For example, growth factor-dependent growth stimulation of hematopoietic cells has been shown to be dependent on STAT5 regulation of the cyclin D1 promoter (21). Also, NF-B interaction with the NF-B binding sites in the cyclin D1 promoter is required for cyclin D1 expression, leading to cell cycle progression. The small GTPase Rac1 signaling has been found to activate cyclin D1 transcription by means of an NF-B-dependent pathway in murine NIH3T3 cells (22,23). In addition, up-regulation of cyclin D1 by ER␣ signaling is accompanied by an increased proliferative response in breast cancer cells (24,25), as ER␣stimulated proliferation could be effectively blocked by antisense cyclin D1 or by microinjection of anti-cyclin D1 antibodies (25,26) and reversed by cyclin D1 overexpression (27). Upregulation of cyclin D1 expression has also been found in hyperplastic mammary glands and proliferative human breast disease (28,29). Together, these observations suggest that cyclin D1 may constitute an important downstream target of diverse upstream signals, with a role in mammary gland development and tumorigenesis. Despite the widespread role of cyclin D1 in the biology of breast cancer, its involvement in Pak1 signaling, a common point of convergence of growth factor signaling in breast cancer cells, remains unknown. Here we investigated the role of Pak1 signaling pathways in the expression of cyclin D1 using breast cancer cell culture models and transgenic mice expressing activated Pak1 in mammary glands.

MATERIALS AND METHODS
Cell Cultures and Reagents-MCF-7 cells, MDA-MB-231 human breast cancer cells, HeLa cells, and endometrial Ishikawa cells were maintained in Dulbecco's modified Eagle's medium F12 (1:1) supplemented with 10% fetal calf serum. HC11 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and epidermal growth factor (10 ng/ml). Antibodies were purchased from the following companies: anti-Pak1 from Cell Signaling, anti-cyclin D1 from Santa Cruz Biotechnology, anti-vinculin from Sigma, and anti-HA from Roche Applied Science.
Metabolic Labeling-MCF-7 cells constitutively expressing active Pak1 were grown to 50% confluency and were metabolically labeled with 20 Ci/ml of [ 35 S]methionine for 24 h in a methionine-free medium containing 2% dialyzed fetal bovine serum in the absence or presence of doxycycline (Dox) (30). Conditioned media with equal trichloroacetic acid precipitable counts were immunoprecipitated with the desired or control antibody, resolved on PAGE gels, and analyzed using autoradiography.
Cell Extracts, Immunoprecipitation, Immunoblotting Assays-For preparation of cell extracts, cells were washed 3ϫ with phosphatebuffered saline and lysed in buffer (50 mM Tris-HCl, pH 7.5, 120 mM NaCl, 1% Triton X-100, 100 mM NaF, 200 mM NaVO 5 , 1 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture (GIBCO) for 15 min on ice. The lysates were centrifuged in an Eppendorf centrifuge at 4°C for 15 min. Cell lysates containing equal amounts of protein were immunoprecipitated, resolved on a 10% SDS-PAGE, transferred to nitrocellulose, and probed with the appropriate antibodies using an ECL method.
Northern Blotting-Total cytoplasmic RNA (20 g) was analyzed by Northern blot analysis. Northern blots were probed with 32 P-labeled cyclin D1 and Pak1 cDNA probe. Actin mRNA was used to assess the integrity of the RNA and to control for the RNA loading.
Cyclin D1 Promoter-Reporter Assays-Subconfluent cells cultured in 6-well plates were transiently cotransfected with 1 g of full-length cyclin D1 promoter-luciferase reporter construct (23), active pak1 (15), and 20 ng of ␤-galactosidase using the Fugene-6 reagent according to manufacturers protocol (Roche Applied Science). Twenty-four hours after transfection, cells were lysed with passive lysis buffer, and luciferase assay was performed using luciferase reporter assay kit (Promega, Madison, WI). ␤-galactosidase activity was used to normalize the transfection. Each transfection was performed in triplicate wells (31).
Transgenic Studies-Generation of transgenic mice expressing constitutively kinase active T423E Pak1 has been described previously (19).
Short Interference RNA (siRNA) Design and Treatment-For Pak1 siRNA, the siRNA was synthesized by using the silencer siRNA construction kit (Ambion, Texas) according to the manufacturer's instructions. The Pak1 targeted sequences were described earlier (19).
Immunofluorescence and Confocal Imaging-Cells were plated on glass coverslips in 6-well culture plates and treated with or without Dox. After 24 h, the cells were rinsed in PBS, fixed in cold 100% methanol for 10 min, processed for immunofluorescent localization of cyclin D1 and HA-tagged Pak1 or cyclin D1 and F-actin (using Alexa-568-labeled phalloidin, Molecular Probes). The DNA was visualized by counter-staining with ToPro3. Fluorescent labeling was visualized by using a Zeiss LSM 510 microscope and a 63ϫ objective. Immunohistochemistry-For immunohistochemical detection of cyclin D1 and Pak1 sections were deparaffinized with xylene and rehydrated using graded ethanol. Sections were incubated in 0.3% H 2 O 2 and methanol for 30 min to inactivate endogenous peroxidase. The sections were then boiled for 10 min in 0.01 M citrate buffer and cooled for 30 min at room temperature to expose antigenic epitopes. The sections were incubated with 2% normal goat serum in 1% bovine serum albumin, PBS for 30 min. The sections were incubated with a mouse-cyclin D1 (1:25 diluted in 2% normal goat serum, 1% bovine serum albumin, and PBS) at a dilution of 1:50 in breast tumor arrays, and with Pak1 at 1:25 dilution and incubated overnight and at room temperature. The sections were washed three times with 0.05% Tween in PBS for 10 min, incubated with secondary antibody, developed with DAB-H 2 O 2 , and then counter-stained with Mayer's hematoxylin. Negative controls were performed by replacing the primary antibody with corresponding IgG.

Pak1 Regulates Cyclin D1 Expression in a Transgenic
Model-In an attempt to define the role of Pak1 in the mammary gland, we recently demonstrated that overexpression of a kinase-active T423E Pak1 transgene in the murine mammary gland leads to a widespread hyperplasia (19). Because cyclin D1 overexpression in the mammary tissues has been reported to cause hyperplasia (28,29) and the finding that Rac, an upstream effector of Pak1, regulates cyclin D1 transcription activity (23), we hypothesized that Pak1 might also regulate cyclin D1 expression. Therefore, we explored the possibility of involvement of cyclin D1 in the action of Pak1 by examining the status of cyclin D1 expression in mammary glands from T423E transgenic and stage-matched wild-type mice. Immunohistochemical analysis revealed a background level of cyclin D1 staining in wild-type mammary glands. In contrast, mammary epithelium from T423E Pak1 transgenic mice exhibited an intense cyclin D1 nuclear staining (Fig. 1A). Quantitation of signals in three independent transgenic lines indicated an ϳ5to 7-fold-increased expression of cyclin D1 in T423E-Pak1 transgenic mammary glands (Fig. 1B). These results suggest that hyperplasia in mammary glands of Pak1-TG mice may be associated, at least in part, with the potential up-regulation of cyclin D1.
Overexpression of Pak1 in Breast Tumors-To assess the significance of Pak1 in breast cancer, we next determined the status of Pak1 expression in breast tumor specimens. We performed immunohistochemical staining of Pak1 and cyclin D1 in tumor arrays containing 60 breast cancer sections. Overexpression of Pak1 was observed in 34 tumors (55%) and cyclin D1 in 42 tumors (70%). Subcellular localization of Pak1 was variable. Of the 34 Pak1-positive tumors, an intense Pak1 cytoplasmic staining was observed in 19 tumors (56%) ( Fig. 2A), nuclear staining in 5 tumors (15%) (Fig. 2B), and both nuclear and cytoplasmic Pak1 immunoreactivity in 10 breast tumor specimens (29%) (Fig. 2C). Pak1 expression also correlated well with the overall cyclin D1 expression in about 40% of tumors. Representative examples of co-expression of Pak1 and cyclin D1 in the same tumor specimen are shown in Fig. 2 (I-L).
Pak1 Stimulates Cyclin D1 Promoter Activity-To investigate the potential Pak1 regulation of cyclin D1 expression, we next examined the effects of wild-type (WT) Pak1 or catalytically active T423E Pak1 on the transcriptional regulation of cyclin D1 using the cyclin D1-promoter luciferase reporter system (Ϫ1745 cyclin D1 luciferase) (23). Cotransfection of WT-Pak1 or T423E Pak1 in murine normal mammary epithelial HC11 cells stimulated transcription from the cyclin D1 promoter as compared with the transcription driven by control vector (Fig. 3). The stimulatory effect of Pak1 signaling on the cyclin D1 promoter was a wide-spread phenomenon, as WT-Pak1 or T423E Pak1 efficiently stimulated cyclin D1 promoter activity in invasive breast cancer MDA-MB-231, cervical cancer HeLa, and in well differentiated endometrial Ishikawa cells (Fig. 3). Together, these results implied a role of the Pak1 pathway in the expression of cyclin D1.
Effect of Kinase-Active T423E Pak1 on Cyclin D1 Expression-To further implicate a regulatory role for Pak1 signaling in the regulation of cyclin D1 expression, we next used a previously characterized MCF-7 breast cancer cell clone expressing HA-tagged-T423E Pak1 under the control of an inducible tetracycline promoter (15). As expected, induced expression of HA-tagged T423E-Pak1 in MCF-7 cells stimulated transcription from the cyclin D1 promoter-luciferase reporter (Fig. 4A). The increased expression of kinase-active Pak1 in MCF-7 cells was also accompanied by elevated steady-state levels of cyclin D1 mRNA (Fig. 4B) and cyclin D1 protein (Fig. 4C), as well as in the levels of newly synthesized metabolic-labeled 35 S-cyclin D1 level (Fig. 4D). To gain clues about the functional conse- quence of Pak1-mediated increased expression of cyclin D1, we determined the subcellular distribution of cyclin D1 in MCF-7 cells expressing kinase-active Pak1 by quantitative confocal scanning microscopy. After 24 h of Dox treatment, both Pak1 and cyclin D1 protein expression were greatly induced, with the majority of cyclin D1 accumulating in the nucleus. (Fig.  5A). Quantitation of cellular localization of cyclin D1 upon Pak1 signaling is presented in Fig. 5B. The observed nuclear accumulation of cyclin D1 seems to be due to increased cyclin D1 expression as well as to increased levels of phosphorylated cyclin D1 in MCF-7 cells with a hyperactivated Pak1 pathway (Fig. 5C). Taken together, these results suggest that Pak1 signaling regulates the expression of cyclin D1 and that Pak1 may be an important mediator of upstream signals leading to cyclin D1 expression.
Pak1 Status Modulates Cyclin D1 Expression-To validate the effect of Pak1 on cyclin D1 expression, we next used previously characterized MCF-7 clones 10 and 17, which ectopically express an auto-inhibitory peptide fragment of Pak1 (amino acids 83-149) that does not interact with GTPases and thus acts as a dominant-negative Pak1 (7,32). We observed that inhibition of the Pak1 pathway in MCF-7 cells substantially suppressed cyclin D1 transcription but not transcription from a control luciferase reporter (Fig. 6A). The expression of Pak1 inhibitory domain 83-149 in stable clones is shown in Fig. 6A, inset. Consistent with this finding, we also observed a marked reduction in the baseline level of cyclin D1 protein expression in clones 10 and 17 (Fig. 6B). These findings suggested that the Pak1 pathway might have an important role in the biology of cyclin D1 in breast cancer cells. To further confirm the significance of Pak1 signaling in the optimal expression of cyclin D1, we next selectively knocked down the endogenous Pak1 expression in MCF-7 breast cancer cells by using the siRNA methodology (33). We demonstrated previously the efficacy of Pak1specific siRNA in down-regulating endogenous Pak1 using Western blotting and confocal microscopy and the ability of Pak1-specific siRNA to suppress Pak1-driven pathways (19). Interestingly, reducing Pak1 expression by Pak1-siRNA but not by control siRNA was accompanied by a significant reduction of cyclin D1 expression (Fig. 6C), suggesting a potential role of functional Pak1 pathway for optimal expression of cyclin D1. To study further whether the inhibitory effect of Pak1 (83-149) on cyclin D1 expression can be relieved by an activating point mutation (Pak1 L107F), transfection studies were performed using plasmids containing Pak1 (83-149) and Pak1 (83-149) L107F on cyclin D1 promoter. Interestingly, activating point mutant Pak1 (83-149) L107F partially but significantly relieved the repression induced by Pak1 (83-149) (Fig.  6D), validating the requirement of a functional Pak1 pathway for optimal cyclin D1 expression.
Pak1 Regulates Cyclin D1 Transcription via an NF-B-Dependent Pathway-Data from the literature suggest that cyclin D1 transcription could be up-regulated through multiple signaling pathways, including Stat5a and NF-B (21,23). Because Pak1 signaling has been shown to modulate NF-B activation (34,35), and because Pak1 also stimulated cyclin D1 transcription (this study), we wished to investigate the potential significance of these mediatory signaling molecules in Pak1 regulation of cyclin D1 transcription. As illustrated in Fig. 7A, there was no significant stimulatory or inhibitory effects of co-expression of WT or dominant-negative Stat5a on cyclin D1 transcription in HC11 cells. Interestingly, co-expression of dominantnegative IB (36) substantially inhibited both basal as well as kinase-active Pak1-induced stimulation of cyclin D1 promoter activity, suggesting a mechanistic involvement of NF-B pathway in the Pak1 regulation of cyclin D1 transcription. Cotrans-fection of WT Pak1 also yielded similar results (Fig. 7B). To investigate the mechanism of this regulation, cotransfection studies were performed using reporter plasmids containing the human cyclin D1 promoter-deletion constructs and T423E-Pak1. The results showed that Pak1 strongly activated cyclin D1 gene expression and that this regulation mapped to regions between Ϫ1745 to Ϫ630 and Ϫ261 to Ϫ66 within the promoter (Fig. 7C). Sequence analysis of the human cyclin D1 promoter identified three potential NF-B binding sites at positions Ϫ858, Ϫ749, and Ϫ39 (37), so we extended our study to examine which promoter regions were critical for Pak1 regulation of cyclin D1. Accordingly, deletion of Ϫ858 and Ϫ749 sites in Ϫ630 cyclin D1-luciferase resulted in a significant reduction in promoter activity. Also, site-directed mutagenesis of one NF-B binding site at Ϫ39 (23) in the Ϫ66 cyclin D1 IBMT luciferase abolished cyclin D1 promoter activity (Fig. 7D), strongly sug-FIG. 6. Pak1 status modulates cyclin D1 expression. A, stable clones expressing Pak1 inhibitor (83-149) and pcDNA control cells were transfected with 1.0 g of cyclin D1-luciferase reporter (left panel) and with a control cytomegalovirus-luciferase reporter (right panel). After 24 h, the cells were lysed, and luciferase activity was measured (n ϭ 3). The activity was normalized with ␤-galactosidase activity. B, Western blot analysis of cyclin D1 in stable clones expressing Pak1 inhibitor (83-149). C, reduction in cyclin D1 protein expression by Pak1-siRNA. MCF-7 cells were cotransfected with 10 nM Pak1, control, or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) siRNA for 48 h and analyzed for the status of cyclin D1 and Pak1 protein expression (n ϭ 2). D, HC11 cells were transfected with 0.5 g of cyclin D1-luciferase reporter and with either 1.5 g of glutathione S-transferase (GST)vector, 1.5 g of GST-Pak1 (83-149), or with 1.5 g of GST-Pak1 (83-149) L107F. After 24 h, the cells were lysed, and luciferase activity was measured (n ϭ 3). The activity was normalized with ␤-galactosidase activity.
gesting that Pak1 regulates the transcription of cyclin D1 through NF-B.
In summary, the results presented here demonstrate that (i) Pak1 regulates the expression of cyclin D1 in diverse cell-types; (ii) Pak1 signaling is an important regulator of cyclin D1 expression and functions; (iii) Pak1 regulates cyclin D1 transcription by means of an NF-B-dependent pathway; and (iv) Pak1 is highly expressed in breast tumors. These data provide evidence that up-regulation of the Pak1 pathway in breast cancer epithelial cells may have functional implications in the enhancement of the growth-rate of breast cancer cells. Any potential up-regulation of cyclin D1 by Pak1 and its upstream activators in breast cancer epithelial cells is likely to sustain and perhaps further promote the ability of tumor cells to grow by supporting the putative functions of cyclin D1 and might contribute toward the noticed hyperplasia in mammary epithelium from kinase-active Pak1 transgenic mice (19). FIG. 7. Pak1 regulates cyclin D1 transcription via an NF-B-dependent pathway. A, HC11 cells were cotransfected with 0.5 g of cyclin D1-luciferase reporter, 0.5 g of T423E Pak1, 0.5 g of Stat5a, 0.5 g of dominant-negative Stat5a, and 0.5 g of dominant-negative NF-B. After 24 h, the cells were lysed, and luciferase activity was measured (n ϭ 3). The activity was normalized with ␤-galactosidase activity. B, HC11 cells were cotransfected with 0.5 g of cyclin D1luciferase reporter, 0.5 g of WT Pak1, 0.5 g of Stat5a, 0.5 g of dominant-negative Stat5a, and 0.5 g of dominant-negative IBMT. After 24 h, the cells were lysed, and luciferase activity was measured (n ϭ 3). The activity was normalized with ␤-galactosidase activity. C, MCF-7 cells were cotransfected with 1.0 g of reporter plasmids containing various deletions of the human cyclin D1 promoter and 0.5 g of T423E Pak1. After 24 h, the cells were lysed, and luciferase activity was measured (n ϭ 3). The activity was normalized with ␤-galactosidase activity. D, MCF-7 cells were cotransfected with 1.0 g of reporter plasmids containing Ϫ66 wild type and Ϫ66 B mutant of the human cyclin D1 promoter and 0.5 g of T423E Pak1. After 24 h, the cells were lysed, and luciferase activity was measured (n ϭ 3). The activity was normalized with ␤-galactosidase activity.