Phosphatase and Tensin Homolog Deleted on Chromosome 10 (PTEN) and PH Domain and Leucine-rich Repeat Phosphatase Cross-talk (PHLPP) in Cancer Cells and in Transforming Growth Factor β-Activated Stem Cells*

Background: The oncogene Akt is regulated by phosphatases. Results: Akt phosphatases cross-talk in prostate cancer cells and in transforming growth factor β1-activated stem cells but not in non-transformed cells. Conclusion: This back-up mechanism facilitates invasive migration of prostate stem and cancer cells. Significance: Characterization of Akt regulation may lead to a better understanding of tumor development and to novel strategies for treatment. Akt kinase controls cell survival, proliferation, and invasive growth and is a critical factor for cancer development. Here we describe a cross-talk between phosphatases that may preserve levels of activated/phosphorylated Akt and confer aggressive growth of cancer cells. In prostatic cancer cells, but not in non-transformed cells or in prostate stem cells, we found that the phosphatase and tensin homolog deleted on chromosome 10 (PTEN) overexpression down-regulated PH domain and leucine-rich repeat phosphatase (PHLPP) and that PHLPP overexpression down-regulated PTEN. We also show that silencing PTEN by siRNA increased the levels of PHLPPs. This cross-talk facilitated invasive migration and was mediated by epigenetic alterations, including activation of miR-190, miR-214, polycomb group of proteins, as well as DNA methylation. A role for the purinergic receptor P2X4, previously associated with wound healing, was indicated. We also show that TGF-β1 induced cross-talk concomitant with epithelial-mesenchymal transition in stem cells. The cross-talk emerged as an integrated part of epithelial-mesenchymal transition. We conclude that cross-talk between PTEN and PHLPPs is silenced in normal prostate cells but activated in TGF-β1 transformed prostate stem and cancer cells and facilitates invasive growth.

The serine/threonine kinase Akt is a critical factor for cancer development and an attractive target for cancer therapy. Deregulated and increased levels of activated/phosphorylated Akt are common in a plethora of tumors and it is generally associated with poor prognosis (1). Akt is involved in cell functions critical for cancer development such as survival, proliferation, and migration (2). It has also been shown to be involved in self-renewal of stem cells (3) and epithelial-mesenchymal transition (EMT) 2 (4).
EMT plays an important role in cancer progression as changes in cellular phenotypes, from differentiated epithelial to dedifferentiated and migrating cells, are induced. Many studies show that activation of the PI3 kinase/Akt pathway is involved in EMT (4). Thus TGF␤-1 is a potent inducer of EMT and activates the PI3K/Akt/mammalian target of rapamycin axis in many cell types. In prostate cancer stem cells activation of the Akt signaling pathway by PTEN loss promotes EMT and metastasis (5).
Several phosphatases negatively regulate the PI3K/Akt pathway. Two isoforms of PHLPP, PHLPP1 and PHLPP2, have been shown to directly dephosphorylate Akt (6), whereas PTEN dephosphorylates the second messenger inositol 1,4,5-trisphosphate and hinders phosphorylation of Akt. PTEN is well studied and its activity is regulated by phosphorylations, ubiquitinations, and protein complex formation (7), whereas the regulation of PHLPP expression is largely unknown (6). Furthermore, possible interplays between PTEN and PHLPP have been studied only to a limited extent (8). Recently, we found that PTEN, PHLPP1, and PHLPP2 acted in a coordinated manner to deplete nuclear pAkt (9). Extracellular ATP, a natural ligand for purinergic receptors and a "danger signal" in innate immunity (10), rapidly triggered these events via P2X7 (9). Another study documented the cross-talk between PI3K and AR pathways leading to down-regulation of PHLPP and Akt-stimulated proliferation of PTEN null cells in mouse prostate (11). There is also indirect evidence for reversed relationships between PTEN and PHLPP expression in some cell models, and for synergistic effects between co-deletions of PTEN and PHLPP1 on metastasis in human prostate cancer (12).
A role of Akt in prostatic cancer (PC) development is well documented (6), and here we show evidence for an epigeneti-cally regulated cross-talk between PTEN and PHLPP in prostatic cancer cells that may preserve levels of Akt. This cross-talk is not detected in non-transformed cells but becomes active in stem cells undergoing EMT. This process depends on the purinergic receptor P2X4. We also show evidence that the cross-talk facilitates invasive migration. Our data might be important for understanding how prostatic cancers acquire aggressive characteristics and for the development of therapeutic strategies.

EXPERIMENTAL PROCEDURES
Cell Culture-The human prostate carcinoma cell lines DU145, 22RV1, LNCaP, PC3, prostate stem cell line WPE, immortalized prostate luminal epithelial (nontumorigenic) RWPE-1, and human breast adenocancinoma MCF7 cells were purchased from American Type Culture Collection (Manassas, VA). Mouse embryonic fibroblasts (MEFs), were kindly provided by Dr. Jing Zhang (Harvard Medical School, Boston, MA). TRL 1215 cells were provided as a generous gift from Dr. Michael P. Waalkes from the National Cancer Institute. This non-tumorigenic cell line was originally derived from the livers of 10-day-old Fischer F344 rats (13). Human embryonic kidney (HEK) 293 cells stably expressing human P2X4 and P2X7 were kindly provided by A. Surprenant, Sheffield University, UK. HEK293 cells stably expressing human empty vector were kindly provided by Prof. A. North, University of Manchester, United Kingdom. DU145, MCF7, and MEF cells were grown in Dulbecco's modified Eagle's medium with 10% inactivated fetal bovine serum (FBS), penicillin/streptomycin, and 1 mM sodium pyruvate. 22RV1 cells were grown in RPMI 1640 supplemented with 10% inactivated FBS and penicillin/streptomycin. PC3 cells were grown in RPMI 1640 supplemented with 10% inactivated FBS, 1 mM sodium pyruvate, 2 mM L-glutamine and penicillin/ streptomycin. LNCaP cells were additionally supplemented with 1 mM HEPES. The RWPE-1 and WPE cells were grown in kerantinocyte SFM (Invitrogen 17005), with bovine pituitary extract, EGF human recombinant and antibiotic-antimycotic (Invitrogen 15240). The culturing of WPE cells was according the procedure for ATCC CRL-2887. TRL 1215 cells were grown in William's EϩGlutaMax TM -I with penicillin/streptomycin and 10% inactivated FBS. HEK293 P2X4 and P2X7 cells were grown in DMEM/F-12 with 1 mM L-glutamine, 10% inactivated FBS, and 300 g/ml of G418. HEK293 control cells were grown in DMEM/F-12 (Invitrogen 21331), 2 mM L-glutamine, 10% inactivated FBS.
Chromatin Isolation-Chromatin was isolated essentially as described in Ref. 14. Cells were lysed in IPB-7. Two fractions were isolated by centrifugation (14,100 ϫ g). Supernatant containing the cytoplasm and the soluble nuclear fraction and nonsoluble pellet contained the chromatin (14).
Immunoprecipitation-Immunoprecipitation was performed by using Evi1, Bmi1, chromatin, or Akt and protein A/G PLUSagarose (Santa Cruz, CA). Cells were washed with PBS and lysed in IPB-7. The cell lysates were incubated for 1 h with antibodies and thereafter with protein A/G PLUS-agarose for 24 h at 4°C.
Small Interference RNA Transfection-Cells were transfected with P2X4, P2X7, PTEN, or control small interference RNA (siRNA) (Santa Cruz Biotechnology, Santa Cruz, CA) for 40 h or for times indicated in the figures according to the TranIT-TKO protocol (Lipofectamine TM 2000, Invitrogen).
Inhibition of MicroRNA-Cells were transfected with anti-microRNA 190 (anti-miR190), anti-microRNA 214 (anti-miR214), and microRNA negative control (non-targeting, NT) inhibitor (mirVana TM miRNA inhibitors, Ambion Life Technologies, Bleiswijk, Netherlands) for 40 h according to the Ambion protocol for mirVana TM miRNA inhibitors. Cells were transfected using Lipofectamine TM 2000 (Invitrogen) as the transfection reagent. RNA Purification and Real-time RT-PCR-Total RNA was prepared using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and cDNA was generated with the High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA) according to protocol. Subsequently, quantification of gene expression was performed in duplicates using Maxima TM SYBR Green qPCR Master Mix (Fermentas, St. Leon-Rot, Germany) with detection on an Applied Biosystems 7500 Realtime PCR System (Applied Biosystems). The reaction cycles used were 95°C for 2 min, and then 40 cycles at 95°C for 15 s and 60°C for 1 min followed by melt curve analysis. Primer sequences are given in Table 1. Relative gene expression quan-tification was based on the comparative threshold cycle method (2 Ϫ⌬⌬Ct ) with normalization of the raw data to the included housekeeping gene (GAPDH). Quantification of miR was performed using a miRCRY LNA Universal RT miR cDNA synthesis kit, SYBR Green master mix, Universal RT, and LNA PCR primer set for miR16, miR21, miR26a, miR107, miR190 and miR214, normalized to miR103 (Exiqon, Vedbaek, Denmark). Relative gene expression quantification was based on the comparative threshold cycle method (2 Ϫ⌬⌬Ct ).
Cell Invasion Assay-The cell invasion assay was performed using 8-m pore size Transwell Biocoat Control inserts (BD Biosciences) according to the manufacturer's instructions. The cells were fixed with methanol and thereafter stained with toluidine blue from Merck (Darmstadt, Germany). The number of transmembrane cells was counted.
MTT Assay-Cell proliferation was determined by a 3-(4,5dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) assay detecting the cellular mitochondrial capacity to convert MTT tetrazolium salt to formazan. Cells were incubated with the medium containing MTT (Sigma) for 4 h. The cells were then lysed in dimethyl sulfoxide. Absorbance was measured at 570 -620 nm.
Animal Experiments-Female Sprague-Dawley rats were injected intraperitoneally with diethyl nitrosamine (300 mol/kg body weight) (Sigma), dissolved in 0.15 M NaCl within 24 h after birth. At 3 weeks of age, these rats were weaned and injected thereafter with the same dose of diethyl nitrosamine once every other week. After 11 additional doses visible hepatic lesions (preneoplastic tissue) and hepatic tissue without visible lesions (control tissue) were dissected out and homogenized in 0.25 M sucrose. Samples from control and preneoplastic tissue were analyzed by Western blotting. All experiments involving animals were approved by the local ethical committee according to the guidelines of the Swedish National Board of Laboratory Animals. Institutional guidelines for the proper humane use of animals in research were followed.
Statistical Analysis-Statistical analysis was conducted using Student's t test. The data were presented as mean Ϯ S.D. Experiments were performed at least three times with different batches of cells. Results were considered to be statistically significant at p Յ 0.05.

Cross-talk between PHLPP and PTEN in Cancer Cells but Not in Non-transformed Cells-
We have previously shown that statins and ATP inhibited nuclear Akt in several cancer cell lines and that this effect was dependent on coordinated activation of phosphatases (9). PTEN was one of the phosphatases required for depletion of nuclear pAkt, and in a PTEN negative PC cell line, LNCaP, transfection of PTEN restored the statininduced pAkt depletion (9). When restoring PTEN in PC3 cells we observed that PTEN transfection decreased or depleted basic protein levels of PHLPP2 (Fig. 1A).
To further investigate this observation we overexpressed both PHLPP2 and PTEN in PC3 cells. As shown in Fig. 1B, both PTEN and PHLPP2 overexpression by single transfections or by double transfection decreased the levels of PHLPP1 (Fig. 1B). The effect of transfections was also investigated in PTEN-ex-pressing 22RV1 PC cells. In this cell line overexpression of PHLPP1 or PHLPP2 alone (Fig. 1C) or in combination (Fig. 1D) decreased PTEN levels. We also found that transfection of PHLPP1 decreased levels of both PTEN and PHLPP2 (Fig. 1C). Similarly PHLPP2 overexpression decreased basal levels of PTEN and PHLPP1 (Fig. 1C). These experiments suggested a phosphatase cross-talk, i.e. PTEN down-regulated PHLPPs and vice versa, and that this cross-talk also balanced the expression of the two isoforms of PHLPP, PHLPP1 and PHLPP2.
Next we quantified the level of PTEN and PHLPPs in different PC cell lines. As shown in Fig. 1E the basic levels of PHLPP1 are higher in PTEN-deficient cell lines, PC3 and LNCaP, whereas the basic levels of PHLPP1 are lower in PTEN-expressing DU145 and 22RV1 cells. This is in line with the results above and consistent with a cross-talk between PTEN and PHLPPs in PC cells. As expected, all transfections led to decreased levels of pAkt and its downstream target pGSK3␤ Ser-9 (data not shown). We also tested the effect of transfections in MCF-7 breast cancer cells and found that also in these cells overexpression of PTEN decreased the levels of PHLPPs and vice versa (data not shown). This indicates that cross-talk between PTEN and PHLPP might not be specific for prostate cells.
To explore whether this phosphatase cross-talk is associated with a malignant phenotype we studied non-transformed RWPE-1 prostate cells. In this cell line no negative regulation between PTEN and PHLPP1 was detected. In contrast the level of PHLPP2 was elevated by PTEN and PHLPP transfection (Fig.  1F). The study was further extended to other non-malignant cell lines, including mouse fibroblasts (MEF) and TRL-1215 rat liver cells. Similar results were obtained in those cells as in RWPE-1 cells, i.e. overexpression of PTEN or PHLPP1 or 2 did not repress the level of the other phosphatases ( Fig. 1, G and H). In TRL-1215 cells PHLPP2 transfection increased the level of PTEN (Fig. 1H). siRNA PTEN was employed to investigate whether the cross-talk is activated by PTEN suppression in 22RV1 cells. We found that depletion of PTEN lead to increased levels of PHLPPs ( Fig. 1I).
In further studies we investigated if lower levels of PTEN expression induced cross-talk. We used different concentrations of plasmids in PC3 cells and compared the protein levels with that in PTEN expressing 22RV1 cells (Fig. 1J). We found that increasing PTEN levels from zero to moderate or seemingly physiological levels of PTEN in PC3 (Fig. 1J) induced cross-talk between PTEN and PHLPPs (Fig. 1K).
In summary, we found a cross-talk between PTEN and PHLPPs in cancer cell lines, but not in non-malignant cells. Rather, a reversed response to transfection was seen in nonmalignant cells.
Cross-talk Is Mediated by Transcriptional and Epigenetic Changes and Is Dependent on PTEN C Terminus-The above data prompted us to examine whether the effect was regulated on a transcriptional level. Gene-specific primers for PHLPP1, PHLPP2, and PTEN were used (Table 1). RT-PCR results show that PTEN transfection led to down-regulation of mRNA levels of PHLPP1 and PHLPP2 in PC3 cells ( Fig. 2A). In 22RV1 cells, similar results were found and mRNA levels of PHLPP1 and PTEN were suppressed by PHLPP2 transfection (Fig. 2B). miR-21, -26a, and -214 have been shown to regulate PTEN (15)(16)(17) and miR-190 has been shown to regulate PHLPP (18). As shown in Fig. 2C the level of miR-190 was significantly increased by PTEN transfection in PC3 cells, whereas the other tested miRs were not significantly changed (Fig. 2C). We also found that in 22RV1 cells miR-214 levels were increased when transfected with PHLPP2 (Fig. 2D). This is in line with the finding that miR-214 has been shown to negatively regulate PTEN in several cancer cells (17,19,20). Together these data suggest that protein levels in our models were repressed by miRs that previously have been shown to regulate PTEN or PHLPP, respectively.
It has been shown that the C terminus of PTEN plays important roles for its phosphatase activity, membrane recruitment (21), and binding properties to several proteins (22). We investigated whether the PTEN binding capacity is important for cross-talk by transfecting PC3 cells with CD-PTEN, which is binding deficient and lacks phosphatase activity. The use of an antibody against the HA-probe shows that CD-PTEN is nearly half the size of wild-type PTEN (Fig. 2E). Furthermore, overexpression of CD-PTEN failed to repress PHLPPs. As expected, overexpression of CD-PTEN increased the level of pAkt and its downstream targets (Fig. 2E), but did not decrease PHLPP1 or PHLPP2 mRNA in PC3 cells (Fig. 2F). Neither was any effect on miR190 expression detected (Fig. 2G). Also, no changes in miRs were detected when RWPE-1 cells were transfected with CD-PTEN (Fig. 2H). These data indicate that the C terminus of PTEN is essential for the epigenetic suppression of PHLPPs.
The effect of transfections on mRNA and miR expression was also studied in non-transformed cells. RT-PCR results show PTEN transfection rather increased the mRNA levels of PHLPP1 and PHLPP2 in RWPE-1 cells (Fig. 2I). PHLPP2 transfection did not have any effect on PHLPP1 and PTEN mRNAs levels, whereas PHLPP1 overexpression increased PHLPP2 mRNA (Fig. 2I). In RWPE-1 cells the effect of transfection on tested miRs also deviated from results obtained in cancer cell lines. Thus in RWPE-1 cells PHLPP2 increased miR-190 expression, whereas PTEN transfection decreased miR-190 (Fig. 2J), which correlates with results shown in Fig. 1F. To further confirm the involvement of miR-190 and miR-214 in the cross-talk, we transfected the PC3 cells with PTEN and 22RV1 cells with PHLPP2 in combination with miR inhibitors. As shown in Fig. 2, K and L, inhibition of miR-190 or miR-214 prevented the cross-talk between PTEN and PHLPP2.
PHLPP and PTEN Overexpression Leads to Activation of Polycomb Group Proteins and Epigenetic Silencing-miR-214 is involved in the regulation of the polycomb group (PcG) of proteins in both cancer and embryonic stem cells (23). Evi1 has been shown to bind to several PcG proteins, e.g. Bmi1, and to repress PTEN expression (24). PcG proteins are up-regulated in PTEN-null aggressive prostate cancer and inhibition of Bmi1 inhibits growth of aggressive PTEN deletion-induced prostate cancer (25). We tested whether PcG proteins were activated by PHLPP2 overexpression in 22RV1 cells and found that the binding of Evi1 to chromatin was increased (Fig. 3A). Evi1 has been shown to bind to PcG proteins and recruit polycomb repressive complexes leading to PTEN down-regulation (26). Immunoprecipitation experiments show that the binding between Evi1 and Bmi1 was increased in PHLPP2-transfected cells (Fig. 3A). These results indicate that PcG was formed and activated. Similar complexes were detected when PC3 cells were transfected for PTEN (Fig. 3B). Thus, PTEN transfection leads to increased binding between Evi1 and Bmi1 (Fig. 3B). Densitometric analysis from three separate experiments show increased binding between complexes in both 22RV1 and PC3 cells (Fig. 3, A and B).
Activation of PcG proteins has been shown to silence genes by DNA methylation. To investigate whether DNA methylation was involved we used a specific DNA methylation inhibitor, 5-aza-2Ј-deoxycytidine (5-Aza) (27). Cells were treated with 5-Aza during the transfection, and PHLPP2 transfectioninduced down-regulation of PTEN was blocked by 5-Aza (Fig.  3C). A statistically significant difference was seen in three different experiments. This indicates that the silencing of genes involved DNA methylation. Taken together these data indicate an involvement of epigenetic mechanisms in cross-talk leading to down-regulation of PHLPP2 and PTEN after transfection with PTEN and PHLPP2, respectively.
Epithelial-Mesenchymal Transition of Prostate Stem Cells Activates Cross-talk between PHLPP and PTEN-We detected cross-talk in cancer cell lines, but not in non-malignant cells. PTEN depletion may induce EMT (28), and next we studied whether stem cells or stem cells undergoing EMT exhibit crosstalk between PHLPP and PTEN. Bmi1 and miR-214 are activated in several cancer stem cells, including prostate cells (29). Furthermore, Bmi1 plays a regulatory role in self-renewal of stem cells and cancer stem cell (25,30,31). Because our data indicated an involvement of Bmi1 and miR-214 and that immortalized prostate cells (RWPE-1) lack phosphatase cross-  talk, we studied whether EMT affected the cross-talk in prostate stem cells. TGF␤-1 induces EMT (4, 32) as indicated by cancer stem cell markers, morphological changes, and increased invasiveness (33,34). We used WPE prostate stem cells and similar to RWPE-1 cells, WPE cells did not respond to PHLPP1 or -2 transfections by changing PTEN levels (Fig. 4E). TGF␤-1 induced morphological changes already after 3 days and cells acquired an elongated appearance (Fig. 4A). After 8 days of TGF␤-1 treatment, the morphology of WPE cells was changed into more aberrant phenotypes (Fig. 4A) and the proliferation of WPE cells was significantly increased already after 24 h (Fig. 4B). MMP9, MMP2, Snail, Slug, and ␤-catenin are induced during TGF␤-1-induced EMT (35,36), and we found that TGF␤-1 increased levels of these and other EMT markers (Fig. 4C). Furthermore, invasiveness was dramatically increased (Fig. 4D). These data indicated EMT and we now tested  the phosphatase cross-talk. TGF␤-1-treated cells were transfected with PHLPP1 or PHLPP2 and PHLPP2 transfection reduced PTEN levels (Fig. 4F). These results suggest that activation of phosphatase cross-talk is part of EMT in prostatic stem cells.
Cross-talk-induced Increased Invasiveness of PC3 Cells-Overexpression of PTEN has been shown to decrease invasiveness of prostate cells (37) and we examined whether cross-talk affected the invasiveness of PC3 cells. Furthermore, TGF␤-1induced EMT was associated with increased invasiveness (Fig.  4D) and factors such as Bmi1, Evi1, and miR-214 regulating PTEN and PHLPPs also regulate invasive properties (38). miR-214 is up-regulated in many cancers, including PC (39) and targets PTEN leading to increased invasiveness. We first studied the time-response of PTEN transfection on markers of invasiveness (Fig. 5, A and B). MMP9, MMP2, and pAkt levels were reduced as detected 6, 24, and 40 h after transfection in PC3 cells (Fig. 5, A and B). However, after an additional 1-3 days the levels of MMP9, MMP2, and pAkt were elevated (Fig. 5A). The effect of transfections was also analyzed by invasion assay and the results were in accordance to MMP levels. Thus, PTEN transfection reduced invasiveness after 6 h transfection (Fig.  5C), which is in line with previous publications (37). However, invasiveness increased 2 days post-transfection (Fig. 5D). The PC3 cells expressed PTEN after both 6 h of transfection and 2 days post-transfection (Fig. 5, C and D). PTEN transfection was transient and the highest levels of PTEN were detected 24 to 40 h after transfection (Fig. 5A). We then double-transfected PTEN and PHLPP1 or PHLPP2 and also the double transfections decreased MMPs initially (Fig. 5B) but increased MMPs 3 days after transfection (Fig. 5E). These data corresponded to changes in invasiveness (data not shown). Transfection with all three phosphatases, PHLPP1, PHLPP2, and PTEN did not increase MMPs levels 3 days post-transfection (Fig. 5E) and importantly, in triple transfected cells the invasiveness was reduced also after 3 days post-transfection (Fig. 5F). The MTT assay was performed to exclude a role of cell proliferation in reduced invasiveness. After 6, 24, and 40 h of transfection, no significant changes in cell proliferation were observed between empty vector and PTEN vector overexpression (data not shown). Together these data suggest that cross-talk increase invasiveness of PC3 cells.
Phosphatase Cross-talk Is Dependent on P2X4 Receptor-We have shown an involvement of the purinergic receptor P2X7 in nuclear pAkt depletion induced by PTEN and PHLPPs (9). To study the role of purinergic receptors in phosphatase cross-talk we investigated their involvement in cross-talk in PC3 cells, activated stem cells, and in virally transformed HEK293 cells. HEK293 cells lack high expression of purinergic receptors and in control HEK293 cells overexpression of PTEN did not reduce the level of PHLPP2 (Fig. 6A). We then tested HEK293 cells heterologously expressing P2X7 (40). Neither in these cells did overexpression of PTEN reduce the level of PHLPP2 (Fig. 6B). Similarly, PHLPP1 or -2 transfections did not decrease PTEN levels. P2X7 may form aggregates with P2X4 (41) and we also tested HEK293 cells expressing P2X4, which expressed PTEN, PHLPP1, and PHLPP2 to a similar extent as P2X7 expressing HEK293 cells (Fig. 6C). Interestingly, in P2X4 expressing cells overexpression of PTEN decreased PHLPP2 levels suggesting a role for P2X4 in cross-talk between PHLPP and PTEN (Fig. 6B). To further elucidate the role of P2X4 and P2X7 in the PTEN overexpression-induced decrease of PHLPP2 we transfected PC3 cells with siRNA for P2X4 or P2X7. As shown in Fig. 6D, in P2X4-silenced cells PTEN transfection did not down-regulate PHLPP2, but rather increased the level of PHLPP2. In contrast, P2X7 siRNA did not block cross-talk between PTEN and PHLPP2, but rather augmented it. Both siRNAs decreased the levels of the respective target protein (Fig. 6D). As shown in Fig.  4, TGF␤-1 activated EMT in WPE cells to EMT and these activated cells had the capacity to cross-talk. Next, we studied the role of P2X4 for cross-talk in these activated cells. Also in these cells, expressing high levels of P2X4 (Fig. 6E), P2X4 seemed necessary for the phosphatase cross-talk (Fig. 6F). TGF␤-1 increased the level of PHLPPs (Fig. 6E), which is partially in line with findings by others (42). Thus as shown in Fig. 6F, silencing P2X4 with siRNA abrogated the effect of PHLPP2 overexpression on PTEN (this difference was statistically different). These results indicate that the cross-talk between PTEN and PHLPPs remains silent in the absence of P2X4 in cancer cells and in activated stem cells.
P2X4 Is Essential for TGF␤-1-induced Invasiveness and Epithelial-Mesenchymal Transition-Considering that P2X4 was needed for the cross-talk, that TGF␤-1-induced EMT, and that invasiveness activated the cross-talk, the role of P2X4 in invasiveness was studied. TGF␤ increases the invasiveness of PC3 cells (43) and we first studied the role of the P2X4 receptor in TGF␤-1-induced invasiveness. As expected, TGF␤-1 increased the invasiveness and MMP9 expression of PC3 cells (Fig. 7, A  and B). When cells were treated with an inhibitor of P2X4, TNT-ATP, the effect of TGF␤-1 on invasiveness was abrogated (Fig. 7A), and this was associated with inhibition of MMP9 expression (Fig. 7B). The role of P2X4 in TGF␤-1-induced invasiveness was further investigated by using siRNA for P2X4. As shown in Fig. 7C siRNA lowered the levels of P2X4 and inhibited TGF␤-1-induced MMP9 expression in PC3 cells.
P2X4 signaling is Ca 2ϩ -dependent (44) and we tested the effect of an inhibitor of calcium release, BAPTA-AM. As shown in Fig. 7D, BAPTA-AM inhibited the cross-talk. This indicates a role for Ca 2ϩ signaling.
EMT-related invasiveness in mesenchymal stem cells involves purinergic receptors (45), and next we studied the role of P2X4 and TGF␤-1-mediated EMT in WPE cells. When cells were preincubated with an inhibitor of P2X4, TNP-ATP, the TGF␤-1-induced morphological changes were not detected (Fig. 7E). Also the TGF␤1-induced induction of MMP9 and MMP2 was inhibited (Fig. 7F) and the effect on invasiveness was abrogated (Fig. 7G). Together these results suggest that the P2X4 receptor is needed for TGF␤ driven invasiveness in PC3 cells and TGF␤-activated stem cells.
Cross-talk-like Expression of PTEN and PHLPP in Regenerative Preneoplastic Rat Liver Lesions-An involvement of P2X4 in phosphatase cross-talk suggest a coupling to cell death and repair processes (46), and we analyzed regenerative preneoplastic liver lesions in rats, induced after repeated toxic doses of the potent genotoxic carcinogen diethylnitrosamine (47). The mechanistic scenario for these chemically induced lesions includes several rounds of cellular damage-repair cycles and a dependence on TGF␤ signaling (48). Stem cell origin has also been indicated (49). The levels of PTEN, PHLPP1, and PHLPP2 revealed differences between preneoplastic and normal tissue and between those seen in TRL1215 cells (Fig. 1H). Strikingly, preneoplastic tissue exhibited higher levels of PTEN compared with normal tissue (Fig. 7H) and PHLPP1 and PHLPP2 were expressed at lower levels than in control tissue. The high PTEN expression is consistent with a previous report (48) and the expression pattern suggests an activated phosphatase cross-talk.

DISCUSSION
In this work we present evidence for an epigenetic mechanism governing a negative cross-talk between pAkt phosphatases that may serve to prevent pAkt depletion. In prostatic cancer cell lines we found that PTEN overexpression down-regulated PHLPPs and vice versa. Furthermore, we documented that silencing PTEN also enhanced PHLPPs. The crosstalk-induced down-regulation was mediated by genetic and epigenetic alterations involving miR-190, miR-214, the PcG complex activation, and DNA methylation. This phosphatase cross-talk was not seen in non-transformed cells but in cancer cell lines and in prostate stem cells activated by TGF␤-1. This cross-talk is dependent on PTENs C-terminal containing phos-phatase and binding domains. Our results also suggest that this cross-talk is dependent on the purinergic receptor P2X4 and develops as a component in EMT, facilitating invasive growth.  Ͻ 0.05). G, WPE cells were treated with TGF␤-1 (4 ng/ml) and TNP-ATP (30 M) for 8 days. Invasiveness was analyzed and representative images (original magnification ϫ10) are shown as well as data from three different experiments. Results are presented as mean Ϯ S.D. *, significantly different from controls (p Ͻ 0.05); #, significantly different from TGF␤-1-treated cells (p Ͻ 0.05). H, levels of PTEN, PHLPP1, and PHLPP2 in liver preneoplastic tissue and control tissues from rats treated with diethyl nitrosamine (300 mol/kg body weight) weekly for 11 weeks. Samples were analyzed for PTEN, PHLPP1, and PHLPP2. Cdk2 were used as a loading control. Densitometric analysis of data from three different rats. Results are presented as mean Ϯ S.D. I, negative cross-talk between PTEN and PHLPP in cancer cells or activated stem cells. A scheme depicting factors implicated in the down-regulation of PHLPP in response to PTEN transfection and a similar scenario involving miR214 is envisioned for PTEN down-regulation in response to PHLPP transfection.
Although our intention was not to explain all mechanistic aspects of the complex cell signaling we describe here, we show that the cross-talk between PHLPP and PTEN was mediated by epigenetic silencing involving miRs, PcG proteins, and DNA methylation. miR-190 was activated by PTEN transfection and has been shown to down-regulate PHLPP (6), whereas PHLPP transfection activated miR-214 shown to repress PTEN expression (24). Moreover blocking miRs with anti-miRs abrogated the cross-talk further illustrating the central role of miRs in the process. miR-214 has also been shown to be involved in the regulation of PcG proteins in both cancer and embryonic stem cells (23). The role we ascribe to PcG proteins is in line with previous data and we conclude that the phosphatase cross-talk was regulated by complex epigenetic events, previously implicated in PTEN and PHLPP regulation. The detailed mechanism(s) triggering the cross-talk remains to be characterized in future studies.
We found that overexpression of PTEN initially decreased MMPs and invasiveness and later increased pAkt levels, MMP levels, and invasiveness. This paradoxical response was not seen when all three phosphatases were overexpressed. Instead invasiveness and MMPs were reduced. These data indicate that the cross-talk might increase invasive growth capacities. An earlier  APRIL 25, 2014 • VOLUME 289 • NUMBER 17 study shows that permanently transfected PTEN prevented metastasis in mice of orthotopically implanted PC3 cells. However, PTEN did not prevent tumorigenicity and invasive growth was not studied (50) so its relationship to our work is not clear. Our data on prostatic stem cells and EMT further support a role for cross-talk in invasiveness. The data suggest that activation of the phosphatase cross-talk is related to EMT induced by TGF-␤1. The finding that EMT, cross-talk, and invasiveness were prevented by targeting a single receptor, P2X4, suggests that cross-talk is a key event in EMT-related invasiveness. The results are also in line with previous work indicating a critical role of Akt in cell invasion. Thus TGF␤-1 is a potent inducer of EMT and also induces cancer progression and invasiveness (34,36), and Akt is important for EMT in several types of cells (51) and for several downstream targets involved in invasion and metastases (52). Akt has also been shown to play a role in TGF␤-1-induced EMT, probably by supporting cell invasion (4). It is thus plausible that EMT and invasive growth demands an altered regulation of Akt. It is, e.g. possible that the cross-talk we describe here ensures sufficient pAkt levels in restricted compartments important for invasive growth.

Cross-talk between PTEN and PHLPP
The role for P2X4 remains to be characterized. P2X4 is implicated in wound healing (46) and is activated by extracellular ATP. ATP may serve as an alarm signal when damaged cells leak ATP. ATP may induce remodeling of the tissue and TGF␤-1 has been shown to enhance extracellular ATP levels (53). The involvement of P2X4 thus suggests that the epigenetic control of this cross-talk might have evolved to counteract apoptotic signals or to facilitate invasive growth in, e.g. wound healing or during embryonic development. This reasoning is supported by the altered expression of PTEN and PHLPP in preneoplastic liver lesions, which suggest that repeated rounds of carcinogen-induced cell death-repair cycles might lead to an early development of phosphatase cross-talk and perhaps partially explain the tumor promoting effect of such treatments (54). There are also data showing that nerve injury converts certain cells, microglia, to a "P2X4 state" in which they overexpress and redistribute P2X4 to the plasma membrane (55). These data might provide clues to an understanding of the role of P2X4 in phosphatase cross-talk.
In conclusion, we have shown that there is a cross-talk between phosphatases regulating Akt. It is interesting to note that the cross-talk we describe here is not the only cross-talk anticipated to ensure a stable Akt signaling in aggressive cancer. For example, it has been discovered that inhibition of PI3K may activate Wnt cross-talk. This cross-talk may promote invasiveness of cancer cells and explain why therapeutic efforts to target Akt signaling have not been entirely successful (56). It is thus possible that Akt signaling is safeguarded in cancer cells by back-up loops that need to be controlled in successful therapy of invasive tumors.