Cross-talk between p21-activated kinase 4 and ERα signaling triggers endometrial cancer cell proliferation

Cross-talk between estrogen receptor alpha (ERα) and signal transduction pathways plays an important role in the progression of endometrial cancer (EC). Here, we show that 17β-estradiol (E2) stimulation increases p21-activated kinase 4 (Pak4) expression and activation in ER-positive EC cells. The estrogen-induced Pak4 activation is mediated via the PI3K/AKT pathway. Estrogen increases Pak4 and phosphorylated-Pak4 (p-Pak4) nuclear accumulation, and Pak4 in turn enhances ERα trans-activation. Depletion or functional inhibition of Pak4 abrogates EC cell proliferation induced by E2, whereas overexpression of Pak4 rescues cell proliferation decreased by inhibiting the estrogen pathway. Pak4 knockdown decreases cyclin D1 expression and induces G1-S arrest. Importantly, Pak4 suppression inhibits E2 induced EC tumor growth in vivo, in a mouse xenograft model. These data demonstrate that estrogen stimulation increases Pak4 expression and activation, which in turn enhances ERα transcriptional activity and ERα-dependent gene expression, resulting in increased proliferation of EC cells. Thus inhibition of Pak4-ERα signaling may represent a novel therapeutic strategy against endometrial carcinoma.


INTRODUCTION
Endometrial carcinoma (EC) is the most common malignancy of the female genital tract [1][2][3]. Estrogen stimulation is an important pathogenic factor contributing to endometrial carcinoma [4,5]. Estrogen receptors (ERα and ERβ) are ligand-dependent transcription factors that mediate the effects of estrogen via gene regulation [6]. In addition, ERα can initiate several non-genomic signaling events in an estrogen-independent manner via posttranslational modifications, such as phosphorylation. The cross-talk between ERα and signal transduction pathways leads to a series of cellular effects, including adhesion, migration, survival, and proliferation, which play a key role in tumorigenesis [7,8].
The p21-activated kinases (Paks) are a family of serine/threonine kinases comprising six isoforms (group I includingPak1-3, and group II including Pak4-6) in humans, based on their structure and function [9,10]. Interestingly, Paks have been found to mediate tamoxifen resistance in breast cancer [11,12]. Pak1 signaling promotes trans-activation of ERα in breast cancer cells by phosphorylating ERα at serine 305 in the absence of its ligand estrogen, resulting in tamoxifen resistance [11]. Pak4 binds to ERα and promotes its transcriptional activity by phosphorylating ERα-Ser 305, and ERα in turn binds to the Pak4 promoter and induces Pak4 transcription. The Pak4-ERα interaction decreases sensitivity to tamoxifen in MCF-7 human breast cancer cells [12]. However, tamoxifen use increases the risk of endometrial cancer due to its estrogenic effects on the endometrium [13,14], suggesting different regulatory mechanisms of estrogen signaling in breast and endometrial cancer.
Pak4 is one of the major downstream kinases in oncogenic signaling [15,16]. Pak4 is upregulated and activated by various stimuli [17,18]. For example, it promotes prostate cancer cell migration in response to hepatocyte growth factor (HGF) [19]. In gestational trophoblastic disease, Pak4 is activated by human chorionic gonadotropin (hCG) via PI3K/PKB signaling [20]. We have previously demonstrated that the Pak4 expression increases with the progression of EC [21]. Furthermore, we have observed a nuclear localization of Pak4, especially the activated, phosphorylated Pak4 form (p-Pak4ser 474 ) in endometrial cancer tissues [21], suggesting that Pak4 might activate ERα and contribute to estrogen-induced EC pathogenesis.
To explore this possibility, we have investigated the relationship between Pak4 and estrogen signaling in endometrial cancer. We tested the hypothesis that a positive feedback loop exists in which estrogen stimulates Pak4 expression and activation, which in turn promotes ERα trans-activation, and endometrial cancer cell proliferation. This feedback loop also involves PI3K/AKT signaling, cyclin D1, and cell cycle progression. These studies define a novel mechanism underlying estrogen signaling regulation, and suggest that Pak4 might be an important therapeutic target in endometrial cancer.

Estrogen up-regulates Pak4 expression and activation
ER-positive human Ishikawa and RL95-2 endometrial cancer cells, as well as estrogen-responsive breast cancer MCF-7 cells were treated with a low-dose E 2 (10 nM). We observed that E 2 treatment led to a timedependent increase in both Pak4 mRNA and protein levels ( Figure 1A and 1B). In Ishikawa cells, the Pak4 protein levels started to rise after 2 days, and gradually peaked in 6 days. Similar trends were also found in RL95-2 cells. As for MCF-7 breast cancer cells, the levels of Pak4 mRNA and protein increased after 3 days of E 2 stimulation.
Western blotting revealed a time-dependent increase in the levels of p-Pak4 Ser 474 (the activated form) in Ishikawa and RL95-2 cells in the presence of E 2 . The level of p-Pak4 was increased after 5 min of E 2 stimulation, and lasted for at least 90 min ( Figure 1C), indicating that estrogen activates Pak4.

Estrogen activates Pak4 via PI3K/AKT signaling
We next investigated the estrogen downstream signaling involved in the Pak4 activation. We found that estrogen increased AKT phosphorylation within 15min in Ishikawa cells, and 5 min in RL95-2 cells, and lasted for at least 90 min (Figure 2A). In order to elucidate the role of PI3K/AKT in estrogen-induced Pak4 activation, we treated RL95-2 cells with LY 294002, a specific PI3K inhibitor, in the presence of estrogen. LY 294002 significantly blocked the E 2 mediated AKT stimulation, and partially blocked Pak4 phosphorylation ( Figure 2B and 2C), suggesting that PI3K/AKT signaling mediates the estrogen-induced Pak4 activation.

Estrogen induces Pak4 and p-Pak4 nuclear levels in EC cells
We have previously observed mild nuclear and strong cytoplasmic Pak4 levels, and strong nuclear and moderate cytoplasmic p-Pak4 levels in EC tissues [21]. In this study, we further investigated the subcellular localization of Pak4 and p-Pak4 in EC cells by immunofluorescence staining. As shown in Figure 3A and 3B, Pak4 was found in the cytoplasm and in the nucleus, whereas mild cytoplasmic and strong nuclear p-Pak4 immunostaining was observed in human RL95-2 endometrial cancer cells. Moreover, we observed that estrogen treatment stimulated Pak4 and p-Pak4 nuclear accumulation ( Figure 3A and 3B). Increased expression of Pak4 and p-Pak4 in cytoplasmic and nuclear fractions of RL95-2 in the presence of E 2 was also confirmed by western analysis ( Figure 3C).

Pak4 promotes ERα trans-activation
Following the nuclear accumulation of Pak4 and p-Pak4 after E 2 stimulation, we then investigated the role of Pak4 in ERα trans-activation. Ishikawa cells that have relatively low Pak4 levels, were stably transfected with wild-type (wt) Pak4, constitutively active (ca) Pak4, or kinase-dead Pak4, whereas RL95-2 cells were stably transfected with two different shRNA constructs against human Pak4. The Pak4 mRNA and protein levels were substantially enhanced by wt Pak4 overexpression ( Figure  4A) and reduced by Pak4 depletion ( Figure 4B). Neither overexpression nor depletion of Pak4 affected the ERα mRNA levels ( Figure 4A and 4B).
In order to elucidate the effect of abnormal Pak4 expression on ERα transcription, we performed an estrogen response element (ERE) luciferase assay. We found that wt Pak4 and ca Pak4 significantly increased ERE-dependent trans-activation of ERα in estrogen- treated Ishikawa cells ( Figure 4C). Similarly, ca Pak4 also increased ERα target gene expression, including cyclin D1, PR, and pS2, both with and without E 2 treatment. Wild-type Pak4 also induced mRNA expression of cyclin D1, PR, and pS2 in the presence of E 2 , but had no effect on PR expression in the absence of E 2 ( Figure 4C). Conversely, a stable knockdown of Pak4 decreased ERE luciferase activity in RL95-2 cells, both in the presence and absence of E 2 ( Figure 4D). Significantly reduced cyclin D1, PR, and pS2 mRNA levels were also observed in RL95-2 cells after shPak4 knockdown in the presence of E 2 . Interestingly, without E 2 treatment, Pak4 inhibition only decreased cyclin D1 mRNA expression, but had no effect on PR and pS2 ( Figure 4D).
To investigate whether Pak4 regulates the recruitment of ERα to ERE of target genes, we performed ChIP-qPCR using primers for ERE of CCND1, PGR, and PSEN2 genes ( Figure 4E). Consistent with the results above, knockdown of Pak4 with shRNA ( Figure 4F) or inhibition of Pak4 with Pak4-inhibitor PF 3758309 ( Figure 4G) impaired the recruitment of ERα to ERE of the target genes.

Pak4 inhibition decreases E 2 -induced cell proliferation via alteration of G0/G1-phase cell cycle progression
We next investigated whether estrogen increases EC cell proliferation by activating Pak4. As shown in Figure  5A and 5C, E 2 treatment promoted RL95-2 cell colony formation in soft agar, whereas depletion or functional inhibition of Pak4 (with Pak4 inhibitor PF 3758309) almost completely abrogated this effect. In addition, using MTT proliferation assay, we found that E 2 induced RL95-2 cell proliferation, and that the E 2 -induced proliferation was reversed by shPak4 or PF 3758309 ( Figure 5D). We next inhibited the estrogen pathway with a selective ERα inhibitor ICI 182,780, and then rescued the phenotype with overexpression of wt Pak4, ca Pak4, or kinase-dead Pak4. We observed that ICI 182,780 treatment inhibited the E 2 -induced colony formation, and overexpression of wt Pak4 or ca Pak4 partly rescued the number of colonies ( Figure 5A and 5B). MTT proliferation assay provided similar results ( Figure 5E).
To assess the mechanism of Pak4-induced cellular proliferation, the cell-cycle profiles of RL95-2 cells were determined using propidium iodide staining and fluorescence-activated cell sorter analysis. Knockdown of Pak4 in RL95-2 cells resulted in accumulation of cells in G0/G1 phase and a decrease in S phase compared with controls, both in the presence and absence of E 2 ( Figure 5F).

Pak4 inhibition suppresses estrogen-induced tumor growth in nude mice
To investigate the role of Pak4 in vivo, RL95-2 cells with stably knocked-down Pak4 (shPak4) or control cells were injected subcutaneously into the flanks of nude mice. Both groups were administered subcutaneously 17β-estradiol 90-day-release pellets. The growth rate of tumors in mice inoculated with shPak4 RL95-2 cells was significantly lower than in tumors formed by control cells (P=0.0198; Figure 6A). The Pak4 suppression in shPak4 tumors was confirmed by immunohistochemistry staining ( Figure 6B).

DISCUSSION
Cross-talk between ERα and signaling pathways contributes to the pathogenesis of endometrial carcinoma [22], but the underlying mechanisms are not clear. In the present study, we identified a positive feedback loop between Pak4 kinase and ERα signaling that promotes EC cell proliferation, and determined the role of PI3K/AKT signaling pathway and cyclin D1 in the process (Figure 7).
Pak4 is overexpressed and/or activated in various human cancer cells [16,23,24]. It is upregulated in cancers of breast, stomach, ovary, pancreas, and prostate [12,[25][26][27][28][29][30]. Although in our previous study, we observed overexpression and activation of Pak4 in estrogen-induced EC cells, it was not clear whether it was related to estrogen signaling [21]. In this study, we found that the expression and activation of Pak4 was increased in response to estrogen stimulation in EC cells, and that the activation occurred via PI3K/AKT pathway. In HGF signaling, Pak4 is activated through PI3K in MDCK cells [17]. However, LY294002, the inhibitor of PI3K, had no effect on HGFinduced Pak4 activation in prostate cancer cell lines [29]. In this study, we observed that LY294002 treatment partially blocked E 2 -induced Pak4 phosphorylation. These findings suggest that the relationship between Pak4 and PI3K/AKT pathway may be cell specific.
In the present study, E 2 treatment also induced Pak4 phosphorylation, suggesting that in addition to the increased expression, the activation of Pak4 may play an important role in EC progression. In our previous study, we found a positive correlation between nuclear p-Pak4 expression and EC progression [21]. Here, we observed Pak4 and p-Pak4 nuclear accumulation after E 2 stimulation. We also demonstrated that wild-type Pak4 increased ERα trans-activation in the presence or absence of estrogen. A more pronounced effect was observed when cells were transfected with constitutively active Pak4, suggesting that Pak4 may activate ERα in an estrogen-independent manner, and may contribute to the development of hormone independence. Therefore, the protein level and activation of Pak4 may potentially affect the effectiveness of anti-estrogen therapies; this is consistent with previous reports of Pak1 signalingdependent activation of ER-S305 leading to enhanced S118 phosphorylation, and development of tamoxifen resistance in breast cancer [11]. Another study reported that GNE-2861, a small molecule selectively inhibiting group II PAKs, overcomes tamoxifen resistance in breast cancer [12]. Therefore, the emergence of Pak4 inhibitors [31,32] suggests that targeting Pak4 expression or activity may represent a novel strategy to increase the response to hormonal treatment in endometrial cancer.
Furthermore, we demonstrated that depletion or functional inhibition of Pak4 abrogated cell proliferation induced by E 2 , whereas increased Pak4 expression and activation rescued the impaired cell proliferation caused by inhibiting the estrogen pathway. We also demonstrated that Pak4 knockdown decreased cyclin D1 expression with and without E 2 treatment. Cyclin D1, a D-type cyclin regulating G1-phase cell-cycle progression, has been identified as a critical downstream effector of estrogen signaling and is frequently overexpressed in endometrial cancer [33][34][35][36]. Pak4 has been shown to control cell cycle progression by regulating the cyclin-dependent kinase inhibitor p21 [37]. Downregulation of Pak4 decreases proliferation and increases apoptosis and S phase arrest in Hep-2 cells via activation of the ATM/Chk1/2/p53   pathway [38]. Our results show that knockdown of Pak4 induces G1-S arrest, which is consistent with the effect of cyclin D1 downregulation [39].
In summary, our results demonstrate the presence of a positive feedback loop between Pak4 and ERα signaling in endometrial cancer. Estrogen stimulation leads to increased Pak4 expression, as well as hyper-activation via PI3K/ AKT pathway, and the increased and activated Pak4 in turn enhances ERα trans-activation. This positive feedback loop promotes endometrial cancer cell proliferation by increasing cyclin D1 expression and altering cell cycle progression. The correlation between Pak4 and ERα signaling not only reveals an underlying mechanism of estrogen-related tumor progression, but also provides a rationale for multi-targeted therapies against endometrial cancer.

Plasmids and transfection
The Pak4 mutants including ca Pak4 (E474) and kinase-dead Pak4 (M350) were generated by site-specific mutagenesis from the Pak4 wild-type using the Quick Change Site Directed Mutagenesis Kit (Stratagene, LaJolla, CA, USA). The ca Pak4 mutant was generated by mutating Ser474 to glutamic acid. The resulting Pak4 (E474) mutant showed enhanced auto-phosphorylation activity [40]. The kinase-dead Pak4 (M350) mutant carries a mutation in which the conserved lysine in subdomain II is converted to a non-phosphorylated methionine, resulting in a completely inactive kinase [40].

Estrogen treatment, PI3K, Pak4 and ERα inhibitor treatment
For western blot analysis, cells were seeded in 6-well plates at 70% confluency and cultured for 24 h in serum and phenolred-free medium, then treated with 10 nM 17β-estradiol (SIGMA Chemical, St Louis, MO, USA) or vehicle (DMSO, 0.1%) for the indicated times. For long-term treatment, cell were seeded in 6-well plates, at 70% confluency at different time points, cultured in a medium containing 10% charcoal-stripped FBS (HyClone, Logan, UT

Biochemical fractionation
Cytoplasmic and nuclear extracts from RL95-2 cells were isolated using the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce; Thermo Scientific, Rockford, IL, USA). Primary antibodies were the following:

Luciferase reporter assays
The reporter gene ERE-Luc was constructed using the enhanced luciferase reporter gene pGL3-promotor. Three tandem repeats of the consensus ERE oligo (GGTCACTGTGACC) were inserted into the Mlu I-Bgl II site of the multiple cloning site of pGL3-promotor, upstream of SV40 promoter. The ERE-Luc reporter and phRL/CMV (Renilla luciferase) plasmids were cotransfected into Pak4 overexpressing Ishikawa cells and Pak4 knockdown RL95-2 cells. After 24 h, cells were treated with or without E 2 . Luciferase assay was performed using Dual-Luciferase Reporter Assay System (E1531, Promega, Mannheim, Germany) after 48 h treatment with E 2 .

Soft agar colony assay
Cells were seeded on 0.3% top agar in a growth medium over a layer of 0.6% agar in a 6-well plate, at a density of 1×10 4 cells/mL (100 μL/well). Growth medium with or without E 2 or E 2 +ICI 182,780 or PF 3758309 was added to the wells every 3 to 4 days. After 2 weeks of incubation, colonies of more than 50 cells were produced. Colonies were photographed and counted with an inverted microscope.

RNA extraction and qRT-PCR
Total RNA was extracted from cell lines using TRIzol reagent (Invitrogen, Life Technologies; Shanghai, PR China). cDNA was reverse-transcribed from total RNA using Prime Script RT reagent Kit (Takara, Dalian, PR China). Real-time PCR was performed using SYBR Premix Ex Taq (Takara, Dalian, PR China) and analyzed with an ABI Prism 7000 Sequence Detection System. The oligonucleotide primers used were: Pak4: 5'-ATGT GGTGGAGATGTACAACAGCTA-3', 5'-GTTCATCC TGGTGTGGGTGAC-3'. The primers for ERα, cyclin D1, PR, and pS2 were described previously [41]. Gel electrophoresis was used to confirm PCR purity. All data were obtained in triplicates in three independent experiments.

Western blot
Cells were lysed using ProteoJET Mammalian Cell Lysis Reagent (MBI Fermentas, ON, Canada) with a protease inhibitor cocktail (Roche Diagnostics, Basel, Switzerland). Total protein concentration was estimated using the BCA method (Pierce, Rockford, IL, USA). A total of 60 μg of protein was separated on an 8% sodium dodecyl sulfate-polyacrylamide gel and transferred to a polyvinylidene fluoride (PVDF) membrane. Membranes were incubated with primary antibodies. Signal was detected using the BeyoECL Plus (Beyotime, Shanghai, PR China).

Cell cycle analysis
Cells were fixed in 70% ice-cold ethanol. The cells were stained with 25μg/mL propidium iodide (KeyGen Biotech, Shanghai, PR China) in fluorescenceactivated cell sorting buffer (PBS containing 0.1% bovine serum albumin, 0.05% of Triton X-100, and 50 μg/mL of RNaseA). After incubation for 30 min at room temperature, the cells were analyzed by flow cytometry (Becton Dickinson FACScan). Tests were performed in triplicates.

Proliferation assay
Cells (2 ×10 3 cells/well) were seeded in 96-well culture plates in a growth medium. Cell viability was measured every 24 h by MTT assay following the manufacturer's instructions (Beyotime, Shanghai, PR China). Medium was changed every other day. Each experiment was repeated in triplicates.

Studies in vivo
The 5 × 10 6 shPak4 or vector-transfected RL95-2 cells were inoculated s.c. (five mice per group) into four-week-old BALB/c female nude mice. Estrogen was administered to the animals subcutaneously as 17-beta-estradiol 90-day-release pellets (0.72 mg/pellet; IRA, Toledo, OH) as described previously [42]. The diameters of s.c. tumors were measured perpendicularly weekly, and volumes were calculated using the following standard formula: tumor volumes (cm 3 ) = (the longest diameter) × (the shortest diameter) 2 × 0.5. Mice were sacrificed at 28 days post-injection. Tumors were excised and measured. All experimental protocols were approved by the Ethics Committee for Animal Experimentation at Tongji University.

Statistical analysis
All statistical analyses were performed using SPSS 16.0 (Microsoft, Redmond, WA, USA) or Prism (GraphPad, San Diego, CA, USA). Each experiment was performed as least three times, and data were expressed as the means ± SD where applicable. A P value < 0.05 was considered to be significant.

Author contributions
WL and XPW conceived and designed the experiments. WL, TS, JJQ, KW and DZ performed the experiments. WL, YPZ and LY analyzed the data. WL wrote the manuscript. All authors read and approved thefinal manuscript.