|Home | About | Journals | Submit | Contact Us | Français|
Receptor tyrosine kinases of the Eph family play multiple roles in the physiological regulation of tissue homeostasis and in the pathogenesis of various diseases, including cancer. The EphA2 receptor is highly expressed in most cancer cell types, where it has disparate activities that are not well understood. It has been reported that interplay of EphA2 with oncogenic signaling pathways promotes cancer cell malignancy independently of ephrin ligand binding and receptor kinase activity. In contrast, stimulation of EphA2 signaling with ephrin-A ligands can suppress malignancy by inhibiting the Ras-MAP kinase pathway, integrin-mediated adhesion, and epithelial to mesenchymal transition. Here we show that ephrin-A1 ligand-dependent activation of EphA2 decreases the growth of PC3 prostate cancer cells and profoundly inhibits the Akt-mTORC1 pathway, which is hyperactivated due to loss of the PTEN tumor suppressor. Our results do not implicate changes in the activity of Akt upstream regulators (such as Ras family GTPases, PI3 kinase, integrins, or the Ship2 lipid phosphatase) in the observed loss of Akt T308 and S473 phosphorylation downstream of EphA2. Indeed, EphA2 can inhibit Akt phosphorylation induced by oncogenic mutations of not only PTEN but also PI3 kinase. Furthermore, it can decrease the hyperphosphorylation induced by constitutive membrane-targeting of Akt. Our data suggest a novel signaling mechanism whereby EphA2 inactivates the Akt-mTORC1 oncogenic pathway through Akt dephosphorylation mediated by a serine/threonine phosphatase. Ephrin-A1-induced Akt dephosphorylation was observed not only in PC3 prostate cancer cells but also in other cancer cell types. Thus, activation of EphA2 signaling represents a possible new avenue for anti-cancer therapies that exploit the remarkable ability of this receptor to counteract multiple oncogenic signaling pathways.
The serine/threonine kinase mTOR (mammalian Target of Rapamycin), which is of major importance for cell growth, has recently received much attention as a possible novel target for anti-cancer drugs [1–4]. mTOR functions downstream of the serine/threonine kinase Akt as part of the mTOR complex 1 (mTORC1) protein complex and upstream of Akt as part of the mTORC2 complex [5–8]. Typically, growth factor receptors activate Akt through PI3 kinase, which phosphorylates the phospholipid PI(4,5)P2 to produce PI(3,4,5)P3. Binding to PI(3,4,5)P3 causes relocalization of Akt to the plasma membrane. Here, Akt is activated through phosphorylation at T308 by the PDK1 kinase, which is also anchored to the plasma membrane by PI(3,4,5)P3, and through phosphorylation at S473 by mTORC2. Activated Akt in turn phosphorylates and inactivates Tuberous sclerosis complex 2 (TSC2), which is a GTPase-activating protein for the Ras family protein Rheb. This leads to activation of Rheb and its downstream target mTORC1. Two major downstream targets of mTORC1 that regulate mRNA translation are the 4E-BP translational repressor and S6 kinase, which phosphorylates the S6 ribosomal protein to promote protein synthesis. The Akt-mTORC1 pathway is often activated in cancer cells due to loss of the tumor suppressor PTEN, a lipid phosphatase that dephosphorylates PI(3,4,5)P3 to PI(4,5)P2 [9, 10]. PTEN loss is prevalent in prostate cancer, and reducing PTEN levels in mouse prostate epithelial cells is sufficient to induce cancer development through hyperactivation of Akt and mTORC1 [11–13]. Activating mutations in PI3 kinase or Akt, and deregulation of growth factor receptors, can also result in activation of the Akt-mTORC1 pathway in cancer cells [10, 14]. This pathway can promote cancer cell growth as well as migration and invasiveness, and often cooperates with the Ras-MAP kinase pathway to induce malignant transformation [6, 15, 16].
Receptor tyrosine kinases of the Eph family can suppress cancer cell growth, migration and invasiveness through multiple signaling pathways activated by ephrin ligands and whose underlying mechanisms are not completely understood . Eph receptors can, for example, inhibit the Ras-MAP kinase pathway [18, 19], the Crk proto-oncogene [20–22], integrin-mediated adhesion [21, 23–25], and epithelial-mesenchymal transition [26, 27]. Furthermore, a recent report has shown that the EphA2 receptor can also inhibit Akt phosphorylation in cancer cells, but the molecular mechanisms involved and the downstream pathways affected were not elucidated . Here we show that ligand-dependent activation of EphA2 decreases the growth of PC3 prostate cancer cells, which lack PTEN . In these and other cancer cell types, ephrin-A1 stimulation inactivates the Akt-mTORC1 pathway. Our results suggest that Akt dephosphorylation downstream of ligand-activated EphA2 depends on a novel mechanism involving crosstalk with a serine/threonine phosphatase.
PC3 and WM793 cells were cultured in RPMI-1640 medium with 10% fetal bovine serum (FBS); Lu1205, UACC903, and HT-29 cells in DMEM medium with 10% FBS; SKOV-3 cells in McCoy’s 5a modified medium with 10% FBS; MDA-MB-231 cells in DMEM/F12 medium with 10% FBS; MCF-10A cells in DMEM/F12 medium with 10 ng/ml VEGF, 5 μg/ml insulin and 5% FBS.
siGENOME SMARTpool siRNAs (Dharmacon) were used for knockdown of EphA2, Ship2, PHLPP1 and PHLPP2. Dharmacon siCONTROL non-targeting siRNA, which engages the RISC complex but does target any mouse or human genes, was used as a control. The siRNA transfection protocol was optimized for PC3 cells. The cells were transfected with 40 nM EphA2 siRNA, 80 nM Ship2 siRNA, or 62 nM PHLPP1 and 62 nM PHLPP2 siRNAs using Lipofectamine 2000 or Lipofectamine RNAiMax (Invitrogen Life Technologies) according to manufacturer’s instructions. The cells were then stimulated with ephrins or antibodies 2 days after transfection.
For plasmid transfections in PC3 cells, cells in 60 mm plates were transfected with 2 μg total plasmid DNA and 8 μl Lipofectamine 2000 (Invitrogen Life Technologies) according to the manufacturer’s instructions. The plasmids used include: pcDNA3 vector, pCMV vector, HA-tagged wild-type and myristoylated Akt1 in pcDNA3, wild-type and constitutively active CAAX-PI3 kinase in pcDNA3, wild-type and constitutively active H-Ras G12V in pcDNA3, wild-type and constitutively active R-Ras G38VY66F in pcDNA3. Cells were used 2 days after transfection.
PC3 cells grown in medium containing 10% FBS were stimulated with unclustered ephrin-A1 Fc or Fc as a control, or left unstimulated. In some cases, the cells were also treated with various inhibitors, including LY294002 (Promega, 3 mM stock dissolved in DMSO); PD98059 (LC Laboratories, 20 mM stock dissolved in DMSO); rapamycin (LC Laboratories, 50 mM stock dissolved in ethanol). For 2D growth on tissue culture plates, cells were counted in a hemocytometer or viable cells were quantified using the MTT assay (Calbiochem). Briefly, 5,000 cells/well were seeded in 96-well plates, allowed to attach, and then treated with the various inhibitors. For focus formation assays, cells plated at low density were grown for 11 days and stained with crystal violet. The plates were scanned to visualize the foci and then the cells were solubilized and the absorbance at 570 nm was measured. To measure 3D growth in Matrigel (BD Bioscience), cells were plated at low density and spheroids were photographed at different time points. Spheroid volume was estimated as (d max × d min2 × π)/6 (d = diameter). To measure 3D growth in soft agar, cells were plated at low density in 6-well plates on 1.5 ml 0.5% low melting agarose (Gibco) and covered with 1 ml 0.3% low melting agarose. After 3 weeks, the wells were photographed under a 10× objective and colonies were counted from 10 photographs per condition (5 photographs/well, 2 wells/experiment).
For ephrin stimulation experiments, cells grown in 10% FBS were stimulated with ephrin Fc fusion proteins (R&D Systems), which in some cases were pre-clustered with 1/10 concentration of goat anti-human IgG antibody (Jackson ImmunoResearch). For stimulation with immobilized ephrin-A1 Fc, PC3 cells were allowed to attach for 15 min on Petri dishes coated with 3 μg/ml ephrin-A1 Fc or Fc as a control. In the experiment shown in Fig. 7B, some of the cells were grown overnight in the absence of FBS before stimulation. In some experiments, cells were pretreated with calyculin (Calbiochem/EMB Bioscience; 20 μM stock in DMSO), tautomycin (Calbiochem/EMB Bioscience, 1mM stock in ethanol), okadaic acid (MP Biomedicals, 150 μM stock in DMSO), LY294002, PD98059, or rapamycin (see previous section), dasatinib (LC laboratories; 50 μM stock in DMSO), or Gleevec (LC laboratories, 10 mM stock in DMSO).
For immunoblotting, ephrin-stimulated cells and cells transfected with siRNAs or plasmids were lysed in modified RIPA buffer (50 mM TrisHCl, pH 7.6, 150 mM NaCl, 1% TritonX-100, 0.5% sodium deoxycholate, 0.1% SDS, 2 mM EDTA with protease and phosphatase inhibitors) and analyzed by SDS-PAGE followed by immunoblotting with various primary antibodies. Antibodies to phospho-TSC2 T1462, phospho-Akt T308, phospho-Akt S473, Akt, phospho-threonine, phospho-Erk T202/Y204, Erk1/2, phospho-S6 kinase T389, S6 kinase and Ship2 were from Cell Signaling Technology; antibodies to TSC2 and R-Ras were from Santa Cruz Biotechnology; the 9EG7 anti-β1 integrin-activating antibody, the anti-phosphotyrosine antibody conjugated to horseradish peroxidase (HRP) and the anti-Ras antibody were from BD Biosciences; antibodies to PHLPP1 and PHLPP2 were from Bethyl Laboratories; antibodies to EphA2 were from Zymed/Invitrogen (polyclonal, used for immunoblotting) and Upstate Biotechnology/Millipore (monoclonal D7, used for immunoprecipitation); the EphA4 monoclonal antibody was from Zymed/Invitrogen. Goat anti-rabbit and sheep anti-mouse secondary antibodies conjugated to HRP were from Millipore.
For immunoprecipitations, PC3 cells were lysed in modified RIPA buffer and EphA2 was immunoprecipitated with 2.5 μg anti-EphA2 monoclonal antibody (Upstate Biotechnology/Millipore) bound to GammaBind Plus sepharose beads (GE Healthcare). The immunoprecipitates were separated by SDS-PAGE and probed by immunoblotting with an anti-phosphotyrosine or anti-phosphoAkt substrate antibody and reprobed with an anti-EphA2 polyclonal antibody. Other cell types used for immunoprecipitations were lysed in in HEPES buffer (25 mM HEPES pH 7.5, 150 mM NaCl, 10 mM MgCl2, 1% NP-40, 0.25% Na deoxycholate with phosphatase and protease inhibitors (Sigma)) and Eph receptors were immunoprecipitated using 2 μg anti-EphA2 or anti-EphA4 antibodies bound to 15 μl anti-mouse IgG beads (Sigma).
This was carried out with a pAktS473 assay kit (Meso Scale Discovery) according to manufacturer’s instructions.
To examine the effects of integrin-mediated adhesion on Akt phosphorylation, PC3 and WM793 cells were trypsinized, washed in DMEM containing 1% BSA and kept in suspension at 37°C in the same medium for 30 min. Cells (1.5 × 106 cells per 60 mm plate) were then allowed to attach for 15 min at 37°C to plates coated with poly-L-lysine (10 μg/ml for PC3 cells and 5 μg/ml WM793 cells), 10 μg/ml fibronectin (for PC3 cells), or 2.5 μg/ml vitronectin (for WM793 cells) in the presence of 7.5 μg/ml preclustered ephrin-A1 Fc or control Fc. To determine Akt phosphorylation, both adherent and non-adherent cells were pooled, lysed in HEPES buffer and probed by immunoblotting.
To measure the effects of maintaing integrin activity with Mn2+ on ephrin-A1-dependent Akt inactivation, PC3 and WM793 cells were seeded on non-coated tissue culture plates the day before the experiment. Sub-confluent cultures were pre-treated with 1 mM Mn2+ for 30 min at 37°C before they were stimulated for 30 min at 37°C with 4 μg/ml preclustered ephrin-A1 Fc or control Fc. To measure the effects of maintaing integrin activity with an integrin-activating antibody, PC3 and WM793 cells were plated on 10 μg/ml fibronectin overnight at 37°C. Cells were washed in pre-warmed PBS before incubation with 20 μg/ml 9EG7 anti-β1 integrin-activating antibody in RPMI for 30 min at 37°C. Cells were then stimulated for 30 min at 37°C with 4 μg/ml preclustered ephrin-A1 Fc or control Fc followed by a washing step with pre-warmed PBS and cell lysis in HEPES buffer.
To confirm inhibition of ephrin-A1-induced integrin inactivation, PC3 and WM793 cells (3 × 104 cells per well of a 24-well plate) were plated on coverslips coated with 5 μg/ml poly-L-lysine or 10 μg/ml fibronectin and pre-treated either with Mn2+ or anti-β1 integrin antibodies before ephrin-A1 Fc or control Fc stimulation as described above. Cells were then fixed in 4% paraformaldehyde and stained with DAPI and FITC-conjugated phalloidin. The percentage of retracted cells was quantified from 30 20× microscope fields in 3 independent experiments.
It has been recently reported that in neurons ephrin-A-induced activation of EphA receptors inhibits mTORC1 activity . Since mTORC1 is a critical regulator of cancer cell proliferation, particularly in cells in which it is hyperactivated by oncogenic mutations [4, 7], we examined whether ephrin-A ligand stimulation inhibits the growth of PC3 prostate cancer cells. The Akt-mTORC1 pathway is hyperactivated in these cells due to a frameshift mutation that abrogates PTEN lipid phosphatase expression . PC3 cells have been widely used to investigate EphA2 signaling pathways because they express high levels of this receptor [18, 21, 24, 31, 32].
To activate EphA2, we used a soluble activating form of ephrin-A1 (ephrin-A1 Fc). Ephrin-A1, which is a major ligand for EphA2, is anchored to the plasma membrane by a GPI linkage but soluble forms of this ligand released from the cell surface can also activate EphA2 [33, 34]. We found that treatment with ephrin-A1 Fc inhibits PC3 cells 2-dimensional growth on tissue culture plates and 3-dimensional growth in focus formation assays, spheroid formation assays in Matrigel, and colony formation in soft agar (Fig. 1A-D). Ephrin-A1 stimulation also inhibited phosphorylation of S6 kinase at T389 (Fig. 1E), which is a sensitive readout for mTORC1 activity. Thus, ephrin-A1 stimulation can overcome the constitutive activation of mTORC1 caused by loss of PTEN. Consistent with previous findings , ephrin-A1 also decreased phosphorylation of Erk1/2 at T202 and Y204, indicating inhibition of the Ras-Erk MAP kinase pathway.
Interestingly, ephrin-A1 Fc concentrations that caused high acute EphA2 tyrosine phosphorylation (indicative of activation) also induced marked receptor degradation and, therefore, short-lived EphA2 signaling (Suppl. Fig. S1). In contrast, lower ephrin concentrations caused more prolonged persistence of phosphorylated EphA2 due to lower receptor degradation. This may explain why higher ephrin concentrations resulted in less pronounced growth inhibition (Fig. 1A–D).
To examine the involvement of the Akt-mTORC1 and Ras-MAP kinase pathways in PC3 cell growth, we used chemical inhibitors of these pathways. The PI3 kinase inhibitor LY294002 and the mTORC1 inhibitor rapamycin dramatically reduced PC3 cell growth (Fig. 2A, B, D). In contrast, the PD98059 Mek inhibitor only slightly decreased growth (Fig. 2C, D), consistent with the fact that the Ras-Erk pathway is not highly activated by oncogenic mutations in PC3 cells. Thus, these cells appear to be critically dependent on the Akt-mTORC1 pathway for their growth, which is consistent with previous findings and the notion that cancer cells become dependent on hyperactivated oncogenic pathways [16, 35, 36].
In neurons, ephrin-A stimulation inhibits mTORC1 by decreasing Erk1/2-dependent phosphorylation of TSC2 at S664, which results in increased TSC2 activity . Interestingly, this occurs without inhibition of Akt, a kinase that can also inactivate TSC2 by phosphorylating different sites . In contrast, ephrin-A1 treatment has recently been shown to inhibit Akt phosphorylation in several cancer cell types , although increased Akt phosphorylation downstream of EphA2 has also been reported . In PC3 cells treated with ephrin-A1 Fc, we detected a dramatic loss of Akt phosphorylation at both T308 and S473, suggesting decreased Akt activity (Fig. 3A). Indeed, phosphorylation of TSC2 at T1467 and GSK3β at S9 – both of which are Akt target sites – was also reduced (Fig. 3A and data not shown). Furthermore, we observed decreased Akt phosphorylation when PC3 cells were stimulated by contact with immobilized ephrin-A1 Fc to mimic the characteristic mode of Eph receptor activation through contact with ephrins immobilized on adjacent cell surfaces (Fig. 3B). Thus, EphA receptor activation by either soluble or immobilized ephrin-A1 inhibits the Akt-mTORC1 signaling pathway in PC3 cells.
Ephrin-A4 had similar effects as ephrin-A1, consistent with the ability of these ligands to promiscuously activate EphA receptors (Fig. 3C). Activation of EphA2 appears to be sufficient to inhibit Akt because an EphA2-specific activating antibody  also caused loss of Akt phosphorylation (Fig. 3C; Suppl. Fig. 2). Furthermore, the YSA peptide – which is also a selective agonist for EphA2  – reduced Akt phosphorylation not only in PC3 cells but also in WM793 melanoma cells (Fig. 3D) (Mitra et al., submitted). Thus, EphA2 signaling inhibits Akt phosphorylation in different cancer cell types. In addition, siRNA-mediated downregulation of EphA2 abolished the effect of ephrin-A1 on Akt phosphorylation (Fig. 3D), indicating that EphA2 signaling is required for Akt inhibition in PC3 cells. The critical involvement of EphA2 in ephrin-A1-induced Akt inactivation is consistent with a previous report suggesting that EphA2 is the most abundant EphA receptor expressed in PC3 cells . Interestingly, Erk MAP kinases were still inhibited by ephrin-A1 and ephrin-A4 in siRNA-transfected PC3 cells, suggesting the presence of other EphA receptors that can inhibit Erk1/2 but not Akt.
We also found that dasatinib, a potent EphA2 receptor kinase inhibitor originally identified as a Src and Abl kinase inhibitor [38, 42], blocked loss of Akt phosphorylation in PC3 cells stimulated with ephrin-A1 Fc, whereas the Abl inhibitor Gleevec and the Src inhibitor PP2 were ineffective (Fig. 3E). This suggests that EphA2 kinase activity is required for Akt inhibition.
EphA2 signaling inhibits H-Ras, a GTPase that can bind to the p110 catalytic subunit of PI3 kinase and enhance its activity [18, 43]. EphA2 may therefore cause Akt inactivation through inhibition of H-Ras and PI3 kinase. However, we found that ephrin-A1 stimulation of the MDA-MB-231 breast cancer cell line, which expresses the constitutively active K-Ras G13D mutant , still inhibits Akt phosphorylation (Fig. 4A). In contrast, Erk1/2 phosphorylation was not affected, as expected because K-Ras G13D activates Erk1/2 and cannot be inhibited. In comparison, ephrin-A1 inibited both Akt and Erk in the non-transformed MCF-10A mammary epithelial cells, which do not harbor mutated Ras GTPases. Ephrin-A1 stimulation also caused Akt but not Erk1/2 inactivation in several melanoma cell lines expressing the B-Raf V600E mutant, which constitutively activates Erk1/2 (Fig. 4B). Whether Akt inactivation in the melanoma cells treated with ephrin-A1 depends only on EphA2 or also other EphA receptors remains to be determined.
Transfection of wild-type H-Ras or the constitutively active H-Ras G12V mutant increased basal Akt phosphorylation, indicating that activated H-Ras can indeed promote Akt activation in PC3 cells (Fig. 4C). However, neither wild-type nor constitutively active H-Ras blocked Akt inactivation by ephrin-A1. In comparison, the constitutively active H-Ras G12V enhanced basal Erk1/2 phosphorylation in PC3 cells much more than wild-type H-Ras and abolished ephrin-A1-dependent Erk1/2 inactivation. This indicates that inactivation of Ras GTPases by EphA2 can explain Erk1/2 but not Akt inactivation.
EphA2 signaling also inhibits R-Ras, a more distant Ras family member known to activate PI3 kinase but not the Erk MAP kinase pathway . We therefore also expressed constitutively active R-Ras G38VY66F together with low levels of HA-tagged Akt in PC3 cells to preferentially monitor Akt phosphorylation in the transfected cells (representing ~40% of the cells). Expression of constitutively active R-Ras somewhat increased basal Akt phosphorylation but only slightly reduced the ephrin-dependent decrease in Akt phosphorylation detected in cell lysates (Fig. 4D) and in anti-HA antibody immunoprecipitates from cells transfected with HA-tagged Akt (data not shown). These results suggest that inactivation of Ras family GTPases does not play a major role in the loss of Akt phosphorylation downstream of EphA2. Therefore, other pathways must be involved.
Integrin-mediated adhesion can increase Akt phosphorylation through PI3 kinase activation , and ephrin-A1 Fc stimulation of PC3 cells has been shown to inhibit β1 integrins [21, 24]. Thus, EphA2 signaling might decrease Akt phosphorylation indirectly, through inhibition of integrin activity. Consistent with this, we found that Akt phosphorylation dramatically increases in PC3 cells upon attachment to the β1 integrin ligand fibronectin and in WM793 melanoma cells upon attachment to the β3 integrin ligand vitronectin (Fig. 5A). Furthermore, Akt phosphorylation was reduced in these cells by ephrin-A1 stimulation. However, manganese treatment to prevent integrin inactivation only slightly reduced the ephrin-A1-dependent loss of Akt phosphorylation (Fig. 5B). The efficacy of the manganese treatment was confirmed by the observed inhibition of retraction of the cell periphery (Fig. 5B) [21, 45]. Treatment with the 9EG7 β1 integrin-activating antibody to maintain β1 integrin activity in PC3 and WM793 cells plated on fibronectin also partially inhibited cell retraction but not ephrin-A1-dependent loss of Akt phosphorylation (Fig. 5C). Hence, loss of integrin-mediated cell substrate adhesion does not play a critical role in Akt inactivation downstream of EphA2.
To determine whether EphA2 may regulate Akt by inhibiting PI3 kinase through other pathways, we examined the HT-29 colorectal cancer and SKOV-3 ovarian cancer cell lines, which respectively express the constitutively active P449T and H1047R PI3 kinase mutants . Ephrin-A1 stimulation decreased Akt and Erk1/2 kinase phosphorylation in these cells (Fig. 6A), suggesting that inhibition of PI3 kinase activity is not essential for loss of Akt phosphorylation downstream of EphA2. We also expressed in PC3 cells a prenylated form of the p110α catalytic subunit of PI3 kinase, which is constitutively active because its farnesylation mediates permanent membrane association . To preferentially monitor Akt phosphorylation in the transfected cells, we also co-expressed low levels of wild-type Akt (Fig. 6B). As expected, we observed enhanced Akt phosphorylation in cells transfected with both Akt and constitutively active PI3 kinase, compared to cells transfected only with Akt. Akt phosphorylation in cells co-expressing constitutively active PI3 kinase was only slightly decreased by treatment with 0.1 μg/ml ephrin-A1 Fc (not shown). However, treatment with 1 μg/ml ephrin-A1 Fc substantially reduced Akt phosphorylation, albeit less than in cells transfected only with Akt (Fig. 6B). This suggests that even the high levels of Akt phosphorylation resulting from concomitant transfection of constitutively active PI3 kinase and Akt in cells lacking PTEN can be overcome by high levels of stimulation of EphA2-dependent pathways. Hence, even if inactivation of PI3 kinase contributed to Akt inactivation downstream of EphA2, other pathways must also be involved. It should also be noted that EphA2 has been reported to activate – rather than inhibit – PI3 kinase [48–50].
Since ephrin-A1 decreases Akt phosphorylation in PC3 prostate cancer cells as well as WM793, LU1205 and UACC903 melanoma cells, all of which lack PTEN [29, 51], EphA2 does not function by activating PTEN. On the contrary, a recent study in C. elegans suggests that Eph receptors may negatively regulate PTEN . However, EphA2 has been shown to associate with Ship2, another lipid phosphatase that can functionally compensate for the loss of PTEN by dephosphorylating PI(3,4,5)P3 [29, 50]. We therefore investigated whether enhanced Ship2 activity downstream of EphA2 may be responsible for Akt inhibition in cells treated with ephrin-A1 through a reduction of PI(3,4,5)P3 levels. We found that Ship2 downregulation by siRNA interference increases basal Akt phosphorylation, indicating that Ship2 can indeed regulate Akt activity in PC3 cells (Fig. 6C). However, Ship2 knock down did not prevent ephrin-A1-dependent Akt inactivation (Fig. 6C), suggesting that regulation of Ship2 activity by EphA2 is not critical for Akt inhibition.
Ephrin-A1 treatment also decreased phosphorylation of myristoylated Akt, which is constitutively active due to its permanent membrane localization [53, 54] (Fig. 6D). Thus, signaling events occurring downstream of PI3 kinase and independent of PI(3,4,5)P3 levels can lead to Akt dephosphorylation downstream of EphA2.
Treatment of PC3 cells with ephrin-A1 can cause an almost complete loss of Akt phosphorylation, similar to that induced by the potent PI3 kinase inhibitor Wortmannin (Fig. 7A). Furthermore, the loss occurs rapidly because Akt phosphorylation is already drastically reduced within 5 min of stimulation with 1 μg/ml ephrin-A1 Fc (Fig. 6B and Suppl. Fig. S3). Interestingly, EphA2 also inhibits Akt phosphorylation when the cells are cultured in medium without serum (Fig. 7B). The high Akt phosphorylation that is still observed even under serum-free conditions is likely explained by the high PI(3,4,5)P3 levels due to lack of PTEN expression. In contrast, activation of various growth factor receptors is presumably very low in the absence of serum, resulting in very low activity of PI3 kinase upstream regulatory pathways. Thus, it seems unlikely that EphA2 might decrease Akt phosphorylation by inhibiting a pathway upstream of Akt. Rather, a plausible explanation of our findings is that EphA2 regulates a serine/threonine phosphatase that can dephosphorylate Akt.
Several serine/threonine phosphatases could function with EphA2 to inactivate Akt. For example, PHLPP1 and PHLPP2 are two widely expressed phosphatases known to dephosphorylate S473 of Akt [55, 56]. However, siRNA-mediated knockdown of these phosphatases did not prevent EphA2-dependent Akt dephosphorylation in PC3 cells (Fig. 7C). Thus, PHLPP phosphatases do not play a critical role in Akt inactivation by EphA2.
To examine the involvement of PP1 and PP2A, two very abundant phosphatases responsible for the dephosphorylation of many cellular proteins [57–60], we examined the effects of calyculin. This inhibitor, which targets both PP1 and PP2A [61–63], completely blocked Akt, TSC2, and S6 kinase dephosphorylation in PC3 cells treated with ephrin-A1 (Fig. 7D). Calyculin also inhibited ephrin-A1-induced Akt dephosphorylation in WM793 and Lu1205 melanoma cells (Suppl. Fig. 4), indicating that this effect is not limited to PC3 cells. Inhibition of Akt dephosphorylation was observed even at the low calyculin concentration of 10 nM, which only slightly affected overall protein threonine phosphorylation. Interestingly, calyculin did not detectably increase the basal level of Akt phosphorylation in control cells not stimulated with ephrin-A1. This suggests that Akt is not constitutively dephosphorylated by a calyculin-sensitive phosphatase in PC3 cells, but becomes a target of the phosphatase when EphA2 is activated by ephrin-A1. Therefore, EphA2 activation by ephrin-A1 triggers Akt dephosphorylation by a calyculin-sensitive serine/threonine phosphatase.
To discriminate between PP1 and PP2A, we used the more selective inhibitors tautomycin, and okadaic acid [62, 64–66]. Tautomycin blocked Akt dephosphorylation induced by ephrin-A1 at the concentration of 5 μM (Fig. 7E), which preferentially inhibits PP1 over PP2A [62, 64]. Okadaic acid also blocked Akt dephosphorylation induced by ephrin-A1, but only at the high concentrations (300 nM) that have been reported to inhibit not only PP2A but also PP1 [64, 65, 67] (Fig. 7F). In contrast, lower concentrations that should preferentially inhibit PP2A (100 nM) did not block Akt dephosphorylation. These results implicate a PP1-like phosphatase, or another phosphatase that is sensitive to calyculin and tautomycin but not to low concentrations of okadaic acid, in Akt dephosphorylation downstream of EphA2.
It has been recently reported that Akt phosphorylates EphA2 at S879 and that this phosphorylation can be detected with an antibody recognizing Akt substrate motifs . We found that treatment with calyculin, tautomycin or high concentrations of okadaic acid substantially increases EphA2 recognition by the anti-Akt substrate antibody in ephrin-A1-stimulated as well as control cells (Fig. 8). The increased EphA2 phosphorylation in cells treated with the phosphatase inhibitors suggests that EphA2 continuously undergoes dephosphorylation by a phosphatase with the same inhibitor sensitivity profile as the phosphatase that dephosphorylates Akt. This further supports a functional interplay between EphA2 and one or more serine/threonine phosphatases.
Both tumor promoting and tumor suppressing activities have been reported for Eph receptors, and the molecular mechanisms responsible for one versus the other outcome are under intense investigation. EphA2 is the Eph receptor that has attracted most attention in the cancer field. Its expression is upregulated downstream of the Ras-MAP kinase pathway, which is often hyperactivated by oncogenic mutations [17, 68]. This may explain the high EphA2 levels found in many cancer types. Ephrin-dependent activation of EphA2 can in turn suppress the Ras-MAP kinase pathway by activating the Ras GTPase-activating protein, p120RasGAP [17, 45, 69]. We report here that, remarkably, the EphA2 receptor also suppresses another major oncogenic pathway, the Akt-mTORC1 pathway through a novel signaling mechanism.
While our work was in progress, ephrin-A1-dependent activation of EphA2 has been reported to decrease Akt phosphorylation in PTEN-deficient cancer cells . However, the mechanism involved was not elucidated. It has also been recently shown that ephrin-A-mediated activation of neuronal EphA receptors inhibits mTORC1 by decreasing Erk1/2-dependent phosphorylation of TSC2 without affecting Akt phosphorylation . Taken together, these findings suggest that different EphA receptors can inactivate mTORC1 by using different mechanisms. According to a recent report, EphB receptors can also inhibit Akt .
Our evidence suggests that the major mechanism by which EphA2 reduces Akt phosphorylation in PC3 cells does not involve inhibition of upstream regulatory pathways. Neither expression of constitutively active Ras proteins nor blocking integrin inactivation prevented ephrin-A1-induced loss of Akt phosphorylation, suggesting that the main underlying mechanism does not rely on decreased PI3 kinase function due to inhibition of Ras or integrin activity. Furthermore, neither the lack of PTEN in PC3 and melanoma cells nor the PI3 kinase activating mutations in SKOV-3 and HT-29 cells prevented ephrin-A1-induced inactivation of Akt. This is in contrast to the strong effect of K-Ras and B-Raf oncogenic mutations or transfection of constitutively active H-Ras, all of which abolished ephrin-A1-dependent inactivation of Erk1/2. Hence, constitutively activating mutations of upstream regulators can prevent Erk1/2 but not Akt inactivation downstream of EphA2. We also found that although Ship2 knock down in PC3 cells increases basal Akt phosphorylation, it did not prevent Akt inactivation by EphA2. Our experiments also show that EphA2 can decrease phosphorylation of myristoylated Akt, suggesting that EphA2-dependent inhibition of Akt does not occur through a reduction in its membrane association. In addition, we did not detect EphA2-induced downregulation of PDK1 levels or phosphorylation at S241 (data not shown), suggesting that EphA2 also does not inhibit PDK1 .
Interestingly, a recent analysis of the signaling networks activated downstream of another Eph receptor, EphB2, has suggested that protein phosphatases are important effectors of Eph receptors . In agreement with this, the low-molecular-weight protein tyrosine phosphatase functions downstream of EphA2 in cancer cells . We therefore examined whether EphA2 may promote Akt dephosphorylation by a serine/threonine phosphatase. The recently discovered PHLPP1 and PHLPP2 phosphatases are known to dephosphorylate S473 of Akt and their regulation is poorly understood [55, 56]. Knock down of both PHLPP1 and PHLPP2, however, did not prevent Akt dephosphorylation by EphA2.
PP1 and PP2A are the major serine/threonine phosphatases found in eukaryotic cells [57–60]. They dephosphorylate a multitude of cellular proteins, and both are capable of dephosphorylating Akt. The substrate selectivity of PP1 and PP2A is mainly controlled through protein-protein interactions. PP1-interacting proteins, some of which are also substrates, contain consensus binding motifs such as RV×F and (S/G)ILK [57, 74, 75]. Interestingly, the EphA2 kinase domain contains a similar GMLK sequence in a loop within the N-terminal lobe of the kinase domain, although it is not known whether this motif is functional and whether its PP1 binding ability may be affected by receptor activation. On the other hand, a RVDF sequence also found in the N-terminal kinase lobe is unlikely to bind PP1 because it is not in a loop and the aspartic acid at the variable position likely does not support PP1 binding. While both PP1 and PP2A are known to associate with Akt and have been implicated in its dephosphorylation [59, 60, 62, 65, 76, 77], our inhibitor selectivity profile suggests the involvement of a PP1-like phosphatase in Akt inhibition by EphA2 in PC3 and several other cancer cell lines, consistent with the idea that PP1 may play a preferential role in Akt dephosphorylation in cells of epithelial origin .
The phosphatase inhibitors did not affect Akt phosphorylation under basal conditions, suggesting that Akt is dephosphorylated by a PP1-like phosphatase only when the cells are stimulated with ephrin-A1. Alternatively, feedback loops may keep Akt phosphorylation low when the inhibitors are present. In contrast, phosphatase inhibition greatly enhanced EphA2 phosphorylation at the Akt substrate motif, suggesting that a continuous functional interplay between EphA2 and a phosphatase keeps phosphorylation of this site low. Interestingly, a similar scenario has been reported for the Ron receptor tyrosine kinase, which is phosphorylated by Akt on S1394 near the carboxy terminus . This phosphorylation is also increased by phosphatase inhibitors and promotes the binding of PP1, which in turn dephosphorylates the Ron receptor. Therefore, a plausible model is that EphA2 facilitates the functional interaction of a PP1-like phosphatase with activated Akt. Consistent with this model, overexpressed EphA2 has been shown to colocalize with activated Akt at the leading edge of polarized cells . Furthermore, EphA2 activation does not appear to increase overall cellular levels of threonine-phosphorylated proteins detected by immunoblotting and measurements in extracts of PC3 cells stimulated with ephrin-A1 did not reveal overall decreases in phosphatase activity (data not shown). These findings suggest a localized rather than global effect of EphA2. However, we could not conclusively identify the specific phosphatase(s) involved because siRNA-mediated downregulation of PP1 or PP2A catalytic subunits caused extensive PC3 cell death (data not shown). Since both phosphatases can associate with many regulatory subunits that direct them to different substrates [57–60], and since EphA2 itself may fullfill the function of a PP1 regulatory protein, additional work will be required to identify the mechanism of Akt dephosphorylation downstream of EphA2 and the specific phosphatase involved. Other serine/threonine phosphatases, such as PP4 or PP6, cannot be completely discounted because they also have some sensitivity to the inhibitors used .
The ability of EphA2 to cause Akt dephosphorylation may be also affected by the cellular context, because we did not detect substantial Akt dephosphorylation in a few of the EphA2-expressing cell lines examined, including U251 glioma cells, MCF7 breast cancer cells, and ES2 and HEYA8 ovarian cancer cells (data not shown). Ephrin-A1-dependent stimulation of EphA2-transfected B16 melanoma, LNCaP prostate cancer, COS and 293 HEK cells, which do not endogenously express the receptor, also did not cause Akt dephosphorylation. Moreover, EphA2 was shown to activate Akt in pancreatic cancer cells . It will be interesting to elucidate the mechanisms underlying this differential responsiveness.
Loss of PTEN is particularly critical for prostate cancer development and malignancy due to hyperactivation of the Akt-mTORC1 pathway [12, 13, 80, 81]. Akt can also promote cell survival, proliferation and invasiveness in many other types of cancer and its oncogenic effects can be enhanced by concomitant hyperactivation of the Ras-MAP kinase pathway [16, 82]. Our data suggest that EphA2 activation can overcome the effects of oncogenic mutations in the PI3 kinase-PTEN-Akt pathway. EphA2 acutely inhibits Akt-mTORC1 as effectively as LY294002, Wortmannin or rapamycin, and also inhibits the Ras-MAP kinase pathway. However, EphA2-dependent inhibition of cell growth is lower than that of the chemical inhibitors. This is likely due to the transient nature of the EphA2 signals, because ephrin-A stimulation triggers EphA2 internalization and degradation, which ultimately results in at least partial restoration of Akt and Erk1/2 phosphorylation. Recent data suggest that lack of PTEN in PC3 cells may even accelerate EphA2 degradation . Ephrin-induced EphA2 degradation likely explains our observation that ephrin-A1 has a more pronounced effect on PC3 cell growth at concentrations that do not maximally activate EphA2 because, over prolonged periods, lower ephrin concentrations result in higher steady-state leveld of activated EphA2 due to decreased receptor degradation. Thus, EphA2-based anti-cancer treatments could be more effective if activation of EphA2 forward signaling can be achieved without inducing drastic receptor downregulation.
This work shows that activation of the EphA2 receptor tyrosine kinase by ephrin-A ligands in cancer cells can inhibit a major oncogenic signaling pathway, the Akt-mTORC1 pathway. This tumor suppressor activity of EphA2 was observed even when Akt and mTOR are hyperactivated due to mutations in the PTEN lipid phosphatase and in cells harboring constitutive activating mutations of PI3 kinase, Ras or B-Raf. Our data suggest that crosstalk of EphA2 with a serine/threonine phosphatase plays a critical role in ephrin-A-dependent Akt inactivation.
High ephrin-A1 concentrations cause high but transient EphA2 activation. PC3 cells were stimulated for 20 or 120 min with the indicated concentrations of ephrin-A1 Fc, 0.1 μg/ml control Fc or 100 nM rapamycin, and EphA2 immunoprecipitates were probed with anti-phosphotyrosine antibodies (PTyr) and reprobed for EphA2.
The 1C1 EphA2 agonistic antibody inactivates Akt and Erk1/2 in a concentration-dependent manner. Cells were treated with the indicated concentrations of the 1C1 antibody or an irrelevant antibody (IgG), and lysates were probed as indicated. EphA2 immunoprecipitates were probed with anti-phosphotyrosine antibodies (PTyr) and reprobed for EphA2.
Ephrin-A1 stimulation rapidly inactivates Akt. PC3 cells were transfected with vector control or wild-type Akt. Cells were then stimulated with 1 μg/ml ephrin-A1 Fc for various time periods and lysates were probed as indicated. Shown are longer exposures of the phosphoAkt blots from Fig. 6B.
Calyculin prevents ephrin-A1-dependent Akt dephosphorylation in melanoma cells. Cells were incubated for 30 min with the indicated concentrations of the phosphatase inhibitor calyculin and stimulated for 15 min with 1 μg/ml ephrin-A1 Fc or control Fc. Lysates were probed as indicated.
The authors thank Z. Ronai for the melanoma cell lines; A. Newton for PHLPP1 and PHLPP2 siRNAs and antibodies; T. Franke for the WT and Myr-Akt constructs; S. Field for the WT and CAAX-PI3 kinase construct; J. Stebbins for help with MesoScale experiments and E.M. Lisabeth for helpful comments on the manuscript. This work was supported by DOD grant W81XWH-07-1-0462 (to EBP), grants from MedImmune and the NIH (to EBP), DOD postdoctoral fellowship W81XWH-09-1-0665 (to NYY), and a postdoctoral fellowship from the Spanish Ministry of Education (to CF).
Authors’ contributionsAll authors designed and interpreted some of the experiments. NYY performed most of the experiments involving PC3 cell transfections and phosphatase inhibitor treatments and helped write the manuscript. CF performed the PC3 cell growth experiments. MR performed all the experiments with cells other than PC3 cells and the experiments to investigate the involvement of integrins. ZX performed initial experiments identifying inhibition of Akt phosphorylation by EphA2, experiments with the 1C1 antibody, and phosphatase inhibitor experiments. FV performed initial experiments identifying inhibition of Akt phosphorylation by EphA2, including the ephrin concentration, time and serum dependence of the effects. DAT and EPB conceived the project and designed and intepreted experiments. EBP wrote the manuscript with help from NYY and DAT. All authors read and approved the final manuscript.
Conflict of interest
ZX and DAT are employees of MedImmune/AstraZeneca. EBP received a grant from MedImmune/AstraZeneca that partially supported this research. The other authors declare no potential conflict of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.