Diacylglycerol Kinase δ Modulates Akt Phosphorylation through Pleckstrin Homology Domain Leucine-rich Repeat Protein Phosphatase 2 (PHLPP2)*

Background: Diacylglycerol kinases phosphorylate diacylglycerol to terminate its signaling. Results: Diacylglycerol kinase δ regulates a diacylglycerol feedback loop that promotes dephosphorylation of Akt. Conclusion: Diacylglycerol signaling limits Akt activation through PHLPP2. Significance: Receptor tyrosine kinases, which initiate Akt signaling, also activate a diacylglycerol signaling pathway that serves to modulate the levels of Akt activity. Discovering proteins that modulate Akt signaling has become a critical task, given the oncogenic role of Akt in a wide variety of cancers. We have discovered a novel diacylglycerol signaling pathway that promotes dephosphorylation of Akt. This pathway is regulated by diacylglycerol kinase δ (DGKδ). In DGKδ-deficient cells, we found reduced Akt phosphorylation downstream of three receptor tyrosine kinases. Phosphorylation upstream of Akt was not affected. Our data indicate that PKCα, which is excessively active in DGKδ-deficient cells, promotes dephosphorylation of Akt through pleckstrin homology domain leucine-rich repeats protein phosphatase (PHLPP) 2. Depletion of either PKCα or PHLPP2 rescued Akt phosphorylation in DGKδ-deficient cells. In contrast, depletion of PHLPP1, another Akt phosphatase, failed to rescue Akt phosphorylation. Other PHLPP substrates were not affected by DGKδ deficiency, suggesting mechanisms allowing specific modulation of Akt dephosphorylation. We found that β-arrestin 1 acted as a scaffold for PHLPP2 and Akt1, providing a mechanism for specificity. Because of its ability to reduce Akt phosphorylation, we tested whether depletion of DGKδ could attenuate tumorigenic properties of cultured cells and found that DGKδ deficiency reduced cell proliferation and migration and enhanced apoptosis. We have, thus, discovered a novel pathway in which diacylglycerol signaling negatively regulates Akt activity. Our collective data indicate that DGKδ is a pertinent cancer target, and our studies could lay the groundwork for development of novel cancer therapeutics.

modulating PHLPP expression through mechanisms that regulate their translation (7) and stability (8). In addition, there are also mechanisms to regulate the ability of PHLPP enzymes to dephosphorylate substrates. For example, their localization is modulated in part by their PH domains and through interactions with scaffolding proteins that bring PHLPP enzymes in close physical proximity to their substrates (9 -11). There are likely additional means of regulation, such as modifications that modulate their catalytic activity, but these mechanisms have not yet been identified.
Often simultaneous with formation of PIP 3 is the generation of DAG by phospholipase C enzymes. DAG is a lipid second messenger that, like PIP 3 , recruits numerous signaling proteins to membranes and can activate them as well. DAG targets include members of the PKC family, chimaerins, protein kinase D, and RasGRP enzymes (12). Because of its numerous targets, DAG levels must be tightly regulated. It is metabolized by a family of proteins called the diacylglycerol kinases (DGKs), which are lipid kinases that phosphorylate DAG to produce phosphatidic acid. Ten mammalian DGK isoforms have been identified, and gene deletion studies suggest that they have unique functions (13). We discovered that DGK␦ deficiency led to reduced phosphorylation of Akt following activation of the epidermal growth factor receptor (EGFR) (14). In additional collaborative work, we found that heterozygous deletion of the gene encoding DGK␦ in mice reduced Akt phosphorylation downstream of another RTK, the insulin receptor (15). Collectively, these observations suggested that DGK␦ might broadly regulate Akt signaling downstream of RTKs. Akt is an important cancer target, and our observations suggested that we had found a novel way to regulate its signaling. Thus, we set out to discover the mechanism by which DGK␦ modulates Akt activation.
To immunoprecipitate ␤-arrestins, cells transfected with PHLPP2 and Akt along with empty vector, FLAG-␤-arrestin 1, or FLAG-␤-arrestin 2 were treated as indicated, collected in lysis buffer (Cellular Signaling Technology, Inc., catalog no. 9803), incubated on ice for 10 min, and then centrifuged to remove debris. The lysates (300 g of protein) were incubated with anti-FLAG overnight (4°C), and then protein A/G-agarose (25 l, Santa Cruz Biotechnology, Inc.) was added for 2 h. After three washes with lysis buffer, the pellets and lysates (10 g) were separated by SDS-PAGE and then immunoblotted to detect PHLPP2, Akt, or FLAG. HA-PHLPP2 was immunoprecipitated similarly using anti-HA antibodies.
Generation of Stable DGK␦ Knockdown Cell Lines and Transformation Assays-To generate stable DGK␦ knockdown H1650 cell lines, Dharmacon SMARTvector 2.0 lentiviral particles (catalog no. SK-006713-00) were used according to the instructions. Non-targeting particles (catalog no. S-005000-01) were used as a control cell line. Polyclonal stable cell lines were isolated under puromycin (2 g/ml) selection. Cell proliferation was assessed by plating 10,000 cells in a 10 cm-diameter plate, growing them in 1% FBS with antibiotics for 5-7days, and then counting the number of cells. Cell migration was measured by growing HeLa cells to 80% confluency, performing RNAi, starving the cells overnight, scratching the monolayer with a pipette tip, washing, adding growth factor (25 ng/ml) for 24 h, fixing with 10% formalin for 1 h at 4C, and then staining the cells with 0.1% crystal violet. Growth in soft agar was performed as described (18).

RESULTS
DGK␦ Deficiency Reduces Phosphorylation of Akt-We first asked whether DGK␦ modulated Akt phosphorylation downstream of other RTKs in addition to EGFR and the insulin receptor. We found reduced phosphorylated Akt (pAkt) downstream of c-Met in DGK␦-deficient HeLa cells (Fig. 1A). A second siRNA duplex targeting another region of DGK␦ similarly reduced pAkt downstream of both EGFR and c-Met (Fig. 1B). We also tested PDGF receptor and again found reduced pAkt in DGK␦-deficient cells (Fig. 1C). Next, we examined two additional cell lines and found similar reductions in pAkt in DGK␦deficient cells (Fig. 1D). Finally, we assayed primary keratinocytes and tissue harvested from newborn wild-type or DGK␦ knockout mice. In the keratinocytes, DGK␦ deficiency led to reduced Akt phosphorylation following treatment with either TGF␣ or lysophosphatidic acid (Fig. 1E). In tissue harvested from DGK␦ knockout mice, we found reduced pAkt in several organs (Fig. 1F). Together, these data indicate that DGK␦ deficiency broadly attenuates Akt phosphorylation in vitro and in vivo.
There are two alternatively spliced products of the DGK␦ gene, DGK␦1 and DGK␦2, which differ at their amino termini.
The alternative splicing does not change their DGK activity but appears to alter their subcellular localization (19). HeLa cells express both DGK␦ splice variants (data not shown). To test which DGK␦ splice variant could modulate Akt phosphorylation, we used RNAi to reduce the levels of endogenous DGK␦ in HeLa cells and then tested whether we could rescue Akt phosphorylation by expressing RNAi-resistant DGK␦1 or DGK␦2 cDNA constructs. We consistently observed that DGK␦2, but not DGK␦1, rescued Akt phosphorylation (Fig. 2, A and B), indicating that DGK␦2 was the relevant splice variant. Affirming this conclusion, we found that transient overexpression of kinase-dead DGK␦2 reduced phosphorylation of Akt (Fig. 2C). Collectively, these data indicate that DGK␦2 modulates Akt phosphorylation.
The Defect in DGK␦-deficient Cells Occurs at the Level of Akt-Although most RTKs activate PI3K/Akt signaling, they do so using different mechanisms. c-Met uses the adaptor protein Gab1 as a scaffold to recruit PI3K (20), whereas EGFR does not necessarily require Gab1 (21). The diverse mechanisms by which these receptors activate PI3K/Akt signaling contrasted with the similarity of responses to DGK␦ depletion suggested that the defect caused by DGK␦ deficiency occurred at, or near, the level of Akt. To determine where in the RTK signaling pathway the phosphorylation defect occurred, we focused on signaling from c-Met because, in contrast to its effects on EGFR, DGK␦ deficiency does not affect c-Met levels. We first examined the amount of tyrosine-phosphorylated c-Met and found no differences between control and DGK␦ siRNA-treated cells (Fig. 1A). Next, we assayed the levels of phosphorylated Gab1 and again found no differences in its phosphorylation (Fig. 1A), indicating that the defect likely occurred after PI3K was recruited to Gab1. The PIP 3 generated by PI3K serves to recruit numerous signaling proteins to the plasma membrane, including Akt. To determine whether the reduced Akt phosphorylation in DGK␦-deficient cells was due to reduced membrane association of Akt, we examined the abundance of membrane-FIGURE 1. Reduced Akt phosphorylation in DGK␦-deficient cells. A, control or DGK␦ knockdown HeLa cells were treated with EGF (10 ng/ml) or HGF (25 ng/ml) for 10 min. The levels of indicated proteins were detected by Western blotting. p473 and p308 refer to Akt. B, two unique siRNA duplexes were used to knock down DGK␦, and then pAkt (Ser-473) was measured by Western blotting. C, PDGF (25 ng/ml) was used to activate HeLa cells, and then pSer-473 Akt was detected by Western blotting. The vertical lines indicate that intervening lanes were omitted. D, pSer-473 in Akt was measured in H1650 or HepG2 cells. The vertical lines indicate that intervening lanes were omitted. E, primary keratinocytes isolated from either Dgkd (Ϫ/Ϫ) or Dgkd (ϩ/ϩ) newborn mice were treated with either TGF␣ (5 ng/ml) or lysophosphatidic acid (50 M) for the indicated times, and then total and pAkt (Ser473) were detected by Western blotting. F, tissues harvested from Dgkd (Ϫ/Ϫ) or Dgkd (ϩ/ϩ) newborn mice were used for Western blotting to detect total and pAkt (Ser-473). The numbers below the blots throughout this figure indicate relative band intensity normalized to total Akt.

FIGURE 2. Effects of manipulating DGK␦ on Akt phosphorylation. A and B,
RNAi-resistant constructs of either DGK␦1 or DGK␦2 were used to test for rescue of pSer-473 in Akt. C, HeLa cells transfected with control vector or kinase-dead (⌬ATP) constructs of DGK␦ were treated with 25 ng/ml HGF, and then the indicated proteins were detected by Western blotting. D, membrane fractions from control or DGK␦ RNAi HeLa cells treated with 25 ng/ml for the indicated times were isolated by 100,000 ϫ g centrifugation, and the indicated proteins were detected by Western blotting. Numbers below the blots throughout this figure indicate relative band intensity normalized to total Akt.
bound Akt before and after activation of c-Met and found no differences in membrane-associated Akt between control and DGK␦-deficient cells (Fig. 2D). These results, combined with the observation that Akt phosphorylation was significantly reduced in DGK␦-deficient cells, indicated that the defect in Akt phosphorylation occurred after its recruitment to the membrane.
Once Akt associates with the membrane, it is phosphorylated first on Thr-308 by 3-phosphoinositide-dependent protein kinase-1 (PDK1). A second phosphorylation, regulated by mTORC2 at Ser-473 in Akt, yields fully activated Akt (2). We examined phosphorylation of both Thr-308 and Ser-473 in Akt and found that phosphorylation at both of sites was reduced to similar extents in DGK␦-deficient cells (Fig. 1A). Next, we examined the levels of phosphorylated Ser-241 in PDK1, which is required for PDK1 activity, and found no differences in phosphorylated Ser-241 between control and DGK␦-deficient cells (Fig. 1A), indicating that the defect in Akt phosphorylation occurred at a later point in the signaling cascade. To confirm this possibility, we examined another PDK1 substrate, p70 S6kinase (p70S6K), and found that its phosphorylation on Thr-229 was not reduced in DGK␦-deficient cells (Fig. 1A). Additionally, phosphorylation at Thr-389 in p70S6K, a prerequisite for phosphorylation of Thr-229 (22), was also not affected by DGK␦ deficiency (not shown). The normal phosphorylation patterns of PDK1 and its downstream target p70S6K were in contrast to the reduced phosphorylation of Akt. This led us to hypothesize that DGK␦ might modulate Akt dephosphorylation.
DAG Signaling Promotes Dephosphorylation of Akt-To test whether DGK␦ modulated Akt dephosphorylation, we reasoned that DGK␦ deficiency would reduce phosphorylation of myristoylated Akt (Myr-Akt), which is associated with the plasma membrane and is constitutively phosphorylated. In DGK␦-deficient cells, we found reduced basal phosphorylation of Myr-Akt and observed that upon activation of c-Met, phosphorylation of Myr-Akt was reduced further (Fig. 3A). In contrast, the levels of phosphorylated Myr-Akt in control cells increased upon activation. The reduction in Akt phosphorylation following activation of c-Met in DGK␦-deficient cells suggested that DGK␦ deficiency promoted dephosphorylation of Akt.
The rapid dephosphorylation of Myr-Akt in DGK␦-deficient cells following activation of c-Met indicated that receptor activation initiated a phosphatase feedback mechanism to limit Akt phosphorylation. To better understand the dynamics of Akt phosphorylation, we performed a time course experiment and observed, after c-Met activation, that phosphorylation of Akt peaked at 5 min and then returned to near base line by 20 min, whereas phosphorylation of c-Met peaked at 10 min and remained quite high, even after 20 min (Fig. 3B and data not  shown). The more rapid reduction in phosphorylated Akt compared with phosphorylated c-Met suggested that activation of c-Met initiated a feedback mechanism that led to dephosphorylation of Akt. Although the phosphatases involved, including protein phosphatase 2A (PP2A) and PHLPP enzymes, have been studied thoroughly, the mechanisms that initiate this phosphatase feedback loop are not well understood. Our data indicate that DGK␦ might be involved.
We have shown previously that DGK␦ deficiency increases cellular DAG levels (23), which led us to hypothesize that DAG signaling might have a prominent role in modulating Akt dephosphorylation. To test the importance of DAG signaling, we determined the effects of increasing DAG levels by depleting PLC␥1, which is activated downstream of c-Met, or PLC␤1, which is not. We found that PLC␥1 depletion, but not PLC␤1 depletion, prolonged Akt phosphorylation (Fig. 3, C and D), implicating DAG signaling as a negative regulator of Akt phosphorylation. As a second approach to test the effects of DAG signaling, we treated cells with PMA, a long-lived DAG analog, and found that it reduced the phosphorylation of Akt following activation of c-Met (Fig. 3E). Collectively, our data indicate that phosphorylation of Akt is modulated by a DAG negative feedback loop initiated by PLC␥1 and negatively regulated by DGK␦.
DGK␦ Modulates Akt Dephosphorylation through PHLPP2-Two different types of phosphatases regulate Akt. Those in the PP2A family predominantly dephosphorylate Thr-308 (3), whereas those in the PHLPP family dephosphorylate Ser-473 under both basal and activated conditions and can also mediate dephosphorylation of Thr-308 under activated conditions (4). We first tested whether we could rescue Akt phosphorylation in DGK␦-deficient cells using calyculin A to inhibit PP2A phosphatases and found that this inhibitor failed to rescue HGFinduced phosphorylation of either Thr-308 or Ser-473 (Fig.  4A). Okadaic acid, another PP2A inhibitor, also did not rescue FIGURE 3. DAG signaling modulates Akt phosphorylation. A, HeLa cells were transfected with wild-type Akt or Myr-Akt followed by control or DGK␦ RNAi. The cells then were treated with 25 ng/ml HGF for 10 min, and levels of the indicated proteins were detected by Western blotting. Short and long refer to the times that the blot was exposed to film. The vertical lines indicate that intervening lanes were omitted. B, HeLa cells were treated with 25 ng/ml HGF for the indicated times, and then the proteins were measured by Western blotting. C and D, same as B, except that PLC␥1 or PLC␤1 were knocked down using RNAi. The vertical lines indicate that intervening lanes were omitted. E, HeLa cells were pretreated with 10 nM PMA for 5 min, exposed to 25 ng/ml HGF for 10 min, and then the proteins were detected by Western blotting. The numbers below the blots throughout this figure indicate relative band intensity normalized to total Akt. Akt phosphorylation (not shown). PHLPP enzymes are insensitive to PP2A inhibitors, so we next tested the effects of RNAi knockdown of PHLPP enzymes. We found that RNAi knockdown of PHLPP2, but not that of PHLPP1, rescued Akt phosphorylation in DGK␦-deficient cells (Fig. 4, B and C). Collectively, these data indicate that the reduced Akt phosphorylation in DGK␦-deficient cells is mediated by PHLPP2.
Specificity in signal transduction is achieved by gathering together signaling proteins in common pathways along with their regulators (24). PHLPP2 has been shown to dephosphorylate Akt 1 but not Akt 2 (4). We hypothesized that a scaffolding protein regulates this specificity by juxtaposing these proteins. ␤-arrestins act to scaffold Akt to the insulin receptor (25), and disrupting this interaction using molecular approaches or by deleting the gene encoding ␤-arrestin 2 in mice reduced Akt activation and caused insulin resistance (25). Both reduced Akt activation and insulin resistance were also evident in DGK␦deficient mice (15), and DGK␦ is known to bind ␤-arrestins 1 and 2 (16). Collectively, these observations indicated that ␤-arrestins and DGK␦ might impact the same Akt signaling path-way. This led us to test the possibility that ␤-arrestins scaffold PHLPP2 and Akt1. We found that ␤-arrestin 1, but not ␤-arrestin 2, coimmunoprecipitated with both PHLPP2 and Akt1 (Fig. 4D). This scaffolding interaction was not affected by treatment with the DAG analog PMA, suggesting that DAG signaling likely affects the specific activity of PHLPP2 rather than its ability to access Akt1. Further supporting its role as a scaffold, we found that RNAi mediated depletion of ␤-arrestin 1-rescued Akt phosphorylation in DGK␦-deficient cells (Fig. 4E).
PKC␣ Is a Target of Excess DAG That Modulates Akt Phosphorylation-Protein kinase C isoforms are major targets of DAG (12). In DGK␦ knockout mice we found evidence of increased PKC activation detected by increased phosphorylation of PKC substrates and enhanced phosphorylation of PKC isoforms (14,23). Additionally, we observed that DGK␦ coimmunoprecipitated with several PKC isoforms (23). Together, these data led us to examine whether PKCs were the DAG targets that modulated phosphorylation of Akt downstream of c-Met. Prior data have implicated PKC␣ as a regulator of Akt signaling (26,27), and we found that an inhibitor of conven- tional PKCs ␣ and ␤ rescued Akt phosphorylation in DGK␦deficient cells (14). Thus, we focused our efforts on PKC␣. To test whether this PKC isoform modulates Akt phosphorylation, we knocked down PKC␣ in HeLa cells and evaluated the extent of stimulated Akt phosphorylation over 30 min. We found that PKC␣ knockdown augmented and prolonged Akt phosphorylation in response to HGF stimulation (Fig. 5A). This effect on Akt was similar to the effects of knocking down PLC␥1 (see Fig.  3C), suggesting that PKC␣ is an important component of the mechanism by which DAG and DGK␦ modulate Akt dephosphorylation downstream of c-Met.
To test this possibility, we reasoned that knocking down PKC␣ could rescue phosphorylation of Akt in DGK␦-deficient cells. After simultaneously reducing expression of both DGK␦ and PKC␣, we found that this manipulation normalized Akt phosphorylation (Fig. 5B). Collectively, these data suggest that the DAG generated after activation of c-Met modulates Akt phosphorylation in part through PKC␣ and that DGK␦ helps modulates the levels of this DAG. In the context of DGK␦ defi-ciency, DAG levels rise, which activates PKC␣ and promotes dephosphorylation of Akt. Finally, to test for a link between PKC activity and PHLPP2, we treated cells with HGF and IPed PHLPP2 and then detected phosphorylated PHLPP2 by immunoblotting with PKC substrate antibodies. We found that HGF increased phosphorylation of PHLPP2 (Fig. 5C). This observation, together with our coimmunoprecipitation data in Fig. 4D, suggests that phosphorylation of PHLPP2 might affect its activity. We are in the process of testing this possibility more rigorously.
DGK␦ Depletion Limits Tumorigenic Properties-Because of the reduced Akt phosphorylation in DGK␦-deficient cells, we hypothesized that depleting the levels of DGK␦ would attenuate tumorigenic properties of cultured cells. Akt1 modulates cell survival (28), so we first determined whether DGK␦ deficiency promoted apoptosis by starving control or DGK␦ RNAi knockdown HeLa cells and then assessing the level of apoptosis by measuring cleaved poly ADP ribose polymerase (PARP). We found that knockdown of DGK␦ significantly enhanced apoptosis under these conditions (Fig. 5D). Cell migration is another important tumorigenic property, so we also tested the ability of control or DGK␦-deficient HeLa cells to migrate into a wounded area and found that the DGK␦-deficient cells migrated more slowly compared with control cells in serumfree conditions (not shown) and during exposure to either EGF or HGF (Fig. 5E). Together, these data indicated that DGK␦ modulates tumorigenic properties and that inhibiting its function could be used to limit tumor growth or metastasis.
Targeted strategies to limit tumor growth are more likely to succeed if the target is a primary driver of tumor growth. Recently, activating mutations in EGFR have been discovered in lung adenocarcinomas (29). These mutations in EGFR increase its kinase activity, which promotes proliferation and survival, leading to a state of oncogene addiction. Because DGK␦ modulates both EGFR expression levels and signaling downstream of EGFR (14,23), we reasoned that inhibiting the function of DGK␦ in cells harboring an activating EGFR mutation would significantly reduce cell proliferation. To test this possibility, we used H1650 lung cancer cells, which harbor an activating mutation in EGFR, to generate cell lines with stable knockdown of DGK␦ (Fig. 5F). We found that depletion of DGK␦ reduced growth of the cells on plastic and in soft agar (Fig. 5G). Unfortunately, we could not maintain knockdown of DGK␦ during growth of the cell lines as xenografts in immunodeficient mice, so we were unable to examine the effects of DGK␦ deficiency on tumorigenesis in vivo. However, our collective in vitro data strongly suggest that DGK␦ deficiency attenuates oncogenic properties, justifying further preclinical studies to assess the therapeutic potential of DGK␦ inhibitors.

DISCUSSION
We have discovered a novel mechanism where DAG signaling modulates PHLPP2 to limit the activation of Akt. RTKs initiate DAG signaling by activating PLC␥1, but the effects of this DAG have not been extensively evaluated. Early studies using phorbol esters, which are long-lived DAG analogues, suggested that excessive DAG signaling might universally promote tumorigenesis (30,31). More recent evidence has indicated that the effects of excessive DAG signaling depend on the specific DAG targets that are abnormally activated (12). PKCs are major DAG targets, and they offer an enlightening example of the diverse effects that excessive DAG signaling can have. PKC⑀, for example, generally promotes cell transformation (12,32,33). We determined whether DGK␦ affects the activity of PKC⑀ by measuring its kinase activity in vitro with or without overexpression of DGK␦ and found no changes in its activity (not shown). Nor was phosphorylation of the PKC⑀ hydrophobic motif affected by DGK␦ deficiency (not shown). Thus, DGK␦ does not appear to modulate PKC⑀. In contrast to PKC⑀, PKC␣ generally attenuates transforming properties (12). The inhibitory effects of PKC␣ on transformation were evident in a recent study where deletion of the gene encoding PKC␣ doubled the number of polyps in the Apc Min/ϩ mouse model of familial adenomatous polyposis (34). The mechanisms by which PKC␣ deficiency promoted tumorigenesis in Apc Min/ϩ mice were not clear, but excessive EGFR signaling was implicated (34).
On the basis of the excessive tumorigenesis in PKC␣ null mice, one would predict that increasing the activity of PKC␣ could be a valid therapeutic strategy to limit EGFR and Akt signaling. Specifically increasing PKC␣ activity is difficult to achieve, given the broad and often tumor-promoting effects of phorbol esters and related compounds that activate PKCs (12). Our data indicate that an alternative approach to capitalize on the antitumorigenic effects of PKC␣ would be to disrupt DGK␦ activity. Indeed, we have found that DGK␦ deficiency led to activation of PKC␣, which then reduced EGFR expression, limited the specific activity of EGFR, and reduced Akt phosphorylation (Refs. 14, 23 and Figs. [1][2][3][4][5]. Collectively, these changes caused Dgkd knockout mice to have a phenotype very similar to that of Egfr knockout mice (23). Moreover, the effects of disrupting DGK␦ were not limited to EGFR signaling. Akt activation was also diminished downstream of the insulin receptor (15), c-Met, and PDGFR (Fig. 1). Given the importance of these RTKs and Akt in cell proliferation, migration, apoptosis, and transformation, our collective data suggest that disrupting the activity of DGK␦ could be a viable therapeutic strategy to treat cancer. Consistent with this possibility, we found that DGK␦ deficiency reduced proliferation and migration and enhanced apoptosis of cancer cells.
Central to the ability DGK␦ to modulate Akt is PHLPP2. We found that depleting PHLPP2, but not PHLPP1, rescued Akt phosphorylation in DGK␦-deficient cells. This specificity of PHLPP function suggested that scaffolding interactions, which are known to regulate PHLPPs (9 -11), might be important. Indeed, we discovered that ␤-arrestin 1 could scaffold PHLPP2 and Akt1. Again, there was specificity in this interaction. ␤-arrestin 2 did not bind PHLPP1 or Akt1. We also found evidence that PHLPP2 was phosphorylated by PKCs but that PMA did not affect the PHLPP2/␤-arrestin/Akt1 interaction. Together, these data indicate that DAG signaling likely modulates the specific activity of PHLPP2 rather than affecting the ability of PHLPP2 to access Akt. We are currently developing an assay that will allow us to measure the specific activity of PHLPP2 in cultured cells.
PHLPP enzymes can also be regulated by means that modulate their abundance (7,8). We found no changes in the abundance of PHLPP2 in DGK␦ or PKC␣ depletion experiments, suggesting that changes in the abundance of PHLPP2 are not the cause of reduced Akt activation in DGK␦-deficient cells. The rapid dephosphorylation of Myr-Akt that we observed in DGK␦-deficient cells also argues against this possibility and suggests that direct modifications, binding partners, or localization of PHLPP2 are more likely causes. All of these changes can be achieved through appropriate scaffolding interactions, such as the ␤-arrestin 1 scaffold that we show here.
PHLPP enzymes have been shown to dephosphorylate numerous proteins, including PKCs, Akt, and p70S6K (3,35,36). Given the diverse targets of PHLPP enzymes, it is not surprising that their specificity might be modulated by scaffolding interactions. Such interactions could provide distinct targets sets for PHLPP1 and PHLPP2 as well as unique targets for a single PHLPP enzyme that depend on the cell context. Although we found reduced Akt phosphorylation in DGK␦deficient cells, we found no differences in p70S6K phosphorylation and excessive phosphorylation of several PKC isoforms (23), even though PHLPP enzymes have been shown to dephosphorylate all of these kinases (35,36). Collectively, these data demonstrating that DGK␦ specifically affects Akt lend credence to the possibility that ␤-arrestin 1 and other scaffolding proteins provide specificity for PHLPP enzymes.
Like PHLPP and PKC enzymes, DGK enzymes appear to have distinct functions. In published work, we demonstrated that similar to DGK␦, DGK also inhibited the function of PKC␣ (37). However, the phenotype of DGK-deficient mice was not similar to that of DGK␦ null mice, and DGK deficiency did not affect EGFR or Akt signaling (Ref. 38 and data not shown). DGK, a close relative of DGK␦, was recently found to affect ERK phosphorylation downstream of EGFR, but it did not alter Akt signaling (39). DGK␣ has also been shown to affect RTK signaling (40 -42), but like DGK knock-out mice, DGK␣ null mice displayed no evidence of altered EGFR or Akt signaling (43). We and others have additionally generated mouse knockouts of DGKs ␤, ⑀, , and , all of which show no evidence of EGFR or Akt signaling defects [(44 -46 and data not shown).
Thus, it appears that DGKs have unique functions. We have now discovered that DGK␦ specifically regulates Akt signaling downstream of potentially oncogenic RTKs. Its novel mechanism of action offers unique opportunities for therapeutic applications and we believe that further studies are warranted to determine the suitability of DGK␦ as a cancer target.