The Retinoblastoma Family Proteins Bind to and Activate Diacylglycerol Kinaseζ*

The retinoblastoma protein (pRB) is a tumor suppressor and key regulator of the cell cycle. We have previously shown that pRB interacts with phosphatidylinositol-4-phosphate 5-kinases, lipid kinases that can regulate phosphatidylinositol 4,5-bisphosphate levels in the nucleus. Here, we investigated pRB binding to another lipid kinase in the phosphoinositide cycle, diacylglycerol kinase (DGK) that phosphorylates the second messenger diacylglycerol to yield phosphatidic acid. We found that DGKζ, but not DGKα or DGKθ, interacts with pRB in vitro and in vivo. Binding of DGKζ to pRB is dependent on the phosphorylation status of pRB, since only hypophosphorylated pRB interacts with DGKζ. DGKζ also binds to the pRB-related pocket proteins p107 and p130 in vitro and in cells. Although DGKζ did not affect the ability of pRB to regulate E2F-mediated transcription, we found that pRB, p107, and p130 potently stimulate DGKζ activity in vitro. Finally, overexpression of DGKζ in pRB-null fibroblasts reconstitutes a cell cycle arrest induced by γ-irradiation. These results suggest that DGKζ may act in vivo as a downstream effector of pRB to regulate nuclear levels of diacylglycerol and phosphatidic acid.

Diacylglycerol (DAG) 3 regulates many cellular processes, including proliferation, differentiation, and cell migration, by modulating the activity of several proteins, such as protein kinase C (PKC), Ras guanyl nucleotidereleasing proteins, chimaerins, and Munc 13 (1). DAG can be produced by the action of several different signal transduction pathways, including phospholipase C-mediated hydrolysis of phosphoinositides or phosphatidylcholine and phospholipase D-mediated hydrolysis of phosphatidylcholine followed by dephosphorylation of phosphatidic acid (PA), and during de novo synthesis of phospholipids (2).
DAG is not only produced at the plasma membrane but at other intracellular sites as well, including the nucleus. Nuclear DAG levels are increased in liver as a consequence of two-thirds partial hepatectomy (3) and in cell cultures treated with insulin-like growth factor 1, which stimulates proliferation (4,5). This suggests that nuclear DAG levels are intimately linked with cell cycle progression, but a causal relationship has not been firmly established. An attractive hypothesis is that nuclear DAG stim-ulates cell cycle progression via a DAG-binding protein such as PKC (1,6). Indeed, DAG in the nucleus recruits and activates PKC in response to insulin-like growth factor 1 stimulation of Swiss 3T3 cells, which is required for G 1 to S phase transition (4,7). However, the role of PKC in regulating the cell cycle is complex, with different PKC isoforms inducing a cell cycle arrest or stimulation of cell cycle progression. Furthermore, the same PKC isoform is able to induce both an arrest and progression through the cell cycle when expressed in different cell types (8).
In the nucleus, DAG kinase (DGK) controls the levels of DAG generated from PI-phospholipase C-mediated hydrolysis of PI(4,5)P 2 (9), and nuclear DGK activity can be stimulated in response to both growth factor (10) and peptide-hormone treatment (11). DGK (Fig. 1A) is one of 10 different DGK isoforms identified to date (12,13). DGK contains a nuclear localization signal (14,15), and DGK has indeed been shown to be nuclear in some cell types (16,17). The nuclear localization signal sequence in DGK overlaps with a motif similar to the PKC phosphorylation site domain (PSD) within the myristoylated alanine-rich protein kinase C substrate (MARCKS) protein (DGK-MARCKS-PSD). The DGK-MARCKS-PSD can be phosphorylated by PKC, which prevents nuclear accumulation of DGK (15). Furthermore, PKC-mediated phosphorylation of the DGK-MARCKS-PSD also inhibits DGK activity (18). Importantly, overexpression of DGK within the nucleus inhibits cell cycle progression (15). Thus, the levels and activity of DGK in the nucleus are subject to regulation by PKC, whereas, conversely, DGK may regulate nuclear DAG levels and consequently PKC activity.
We previously demonstrated that, in vivo, the level of nuclear PI(4,5)P 2 can be modulated by the interaction of Type I PIP-kinases (enzymes that synthesize PI(4,5)P 2 ) with pRB (19). Together with its family members p107 and p130, pRB regulates cell cycle progression by interacting with and attenuating the activity of the E2F transcription factor family (20,21). Since PI(4,5)P 2 is hydrolyzed by phospholipase C, which generates DAG, and since this nuclear DAG was shown to be subsequently phosphorylated by a DGK (9, 10), we questioned whether pRB may act as a nuclear scaffold to regulate PI signaling and DAG phosphorylation.
In this study, we show that GST-pRB fusion proteins can bind and extract DGK, PIP-kinase, and PI-kinase activities from cell lysates. We identify DGK as the DGK isoform that specifically interacts both in vitro and in vivo with pRB and its family members p107 and p130 and show that this interaction potently enhances DGK activity. Finally, we demonstrate that DGK probably lies downstream of pRB signaling in a DNA damage signaling pathway. Our data would imply that disruption of pRB function, which frequently occurs in human cancers, may lead to enhanced nuclear DAG levels and, in turn, to uncontrolled nuclear PKC activity. tagged DGK were published previously (15,22). NH 2 -terminal HAtagged DGK and GFP-DGK were cloned via three-point ligations into pMT2SM-HA and pEGFP-C2, respectively, using internal NdeI and XmaI sites, respectively. The 5Ј fragment was generated by PCR, and the 3Ј fragment was digested from wild-type DGK. GST-DGK and VSV-DGK (both COOH terminus) were cloned by inserting DGK in pMT2SM-GST and pMT2SM-tag, respectively, via a three-point ligation using the internal SphI site. The 3Ј fragment was generated by PCR, and the 5Ј fragment was derived from wild-type DGK. The GST-pRB C terminus (amino acids 767-928) was generated by PCR and inserted into pGEX-4T-2. Wild-type and kinase-inactive DGK were cloned into pBabe by PCR.
Affinity Purifications-GST-DGK (expressed in COS-7 cells) and GST fusion proteins of pRB, p107, and p130 (expressed in bacteria and induced with 200 M isopropyl 1-thio-␤-D-galactopyranoside) were purified using glutathione-Sepharose 4B beads (Amersham Biosciences) according to the manufacturer's instructions. Approximately 200 g of cell lysate was incubated with 4 g of immobilized GST fusion proteins for 2 h at 4°C, and beads were then washed with 1% Nonidet P-40 lysis buffer. For DGK activity assays, equal amounts of GST-protein complexes were washed once in PIPkinase buffer, resuspended in 20 l of 10 mM Tris (pH 7.4), and assayed for DGK activity. For Western blotting, affinity-purified proteins were separated by SDS-PAGE, transferred to nitrocellulose, and probed with an anti-DGK polyclonal antibody or anti-VSV monoclonal antibody P5D4 (Roche Applied Science).
DGK Activity Assay-Immunoprecipitates, GST-protein complexes, or purified proteins were assayed for DGK activity as described by Divecha et al. (4). Lipid vesicles were prepared by sonicating 1 nmol of dioleoylglycerol (from Sigma), 1 nmol of PIP (Biomol), 1 nmol of PI (Sigma), and 3 nmol of PA (Sigma) in 10 mM Tris (pH 7.4). Reactions were performed at 30°C for 10 min (GST pull-downs and immunoprecipitates) or 5 min (purified proteins) in PIPkinase buffer containing 10 M cold ATP and 5 Ci of [␥-32 P]ATP in a final volume of 100 l. Lipids were extracted with 0.5 ml of chloroform/methanol (1:1, v/v), followed by the addition of 125 l of 2.4 M HCl and phase separation. Lipid extracts were dried and separated by thin layer chromatography (silica gel 60 TLC plates (Sigma) soaked in 1 mM EDTA and 1 mM potassium oxaloacetate and heat-activated) using chloroform/methanol/water/ ammonia (45:35:7.5:2.5, v/v/v/v). Lipids were visualized by autoradiography and quantified using phosphoimaging.
E2F Luciferase Activity Assay-C33A cells were seeded in 6-well plates and transiently transfected with the indicated plasmids using the calcium phosphate precipitation method. 48 h after transfection, cells were lysed in 250 l of 1ϫ passive lysis buffer (Promega), and 50 l was used to measure first firefly luciferase and then Renilla luciferase activity using the dual luciferase reporter assay system (Promega). Luciferase activity was detected using a Wallac 1420 multilabel counter (PerkinElmer Life Sciences) according to the manufacturer's instructions. To control for transfection efficiency, firefly luciferase activity was corrected for Renilla luciferase activity.
Preparation of Purified Proteins for DGK Activity Assay-In order to determine DGK activation as a result of direct protein-protein interaction (DGK and GST fusion proteins of pocket proteins), the following purification steps were performed. HA-DGK was immunoprecipitated from 600 g of COS-7 lysate and eluted from the beads using elution buffer (50 mM Tris, pH 8.0, 300 mM NaCl) containing 1 mg/ml HA peptide (YPYDVPDYA). Purified GST fusion proteins were eluted using GST elution buffer (50 mM Tris, pH 8.0, 300 mM NaCl, 20 mM glutathione) and quantified using a bovine serum albumin standard on a Coomassie-stained gel. Purified HA-DGK was incubated with GST fusion proteins at 4°C for 1 h. All samples contained equal amounts of GST elution buffer and 5 g of bovine serum albumin.
Generation of pRBϪ/Ϫ MEFs Stably Expressing DGK and Cell Cycle Analysis after ␥-Irradiation-To immortalize mouse embryonic fibroblasts (MEFs) and to generate cells stably expressing DGK, LZRS-TBX2-iresGFP (kindly provided by M. van Lohuizen (The Netherlands Cancer Institute)) or DGK-pBabe constructs were transfected in Phoenix packaging cells and used to transduce MEFs. Ecotropic retroviral supernatants were collected 48 h after transfection, filtered through a 0.45-m filter, and incubated with 4 g/ml Polybrene (Sigma) before adding to the cells. Viral supernatants were diluted 6 h after transduction. After immortalizing cells with TBX2, cells were transduced with DGK and, after 48 h, were selected with 200 g/ml puromycin (Sigma). For irradiation experiments, 50,000 cells were plated per well (6-well plates). Two days after seeding, cells were ␥-irradiated using a 137 Cs radiation source. 30 min after irradiation, cells were treated with nocodazole (1 g/ml; Sigma) for 30 h. For cell cycle analysis, cells were trypsinized, fixed in ice-cold 70% ethanol, and resuspended in 200 l of phosphate-buffered saline containing 50 g/ml propidium iodide and 50 g/ml RNase. Cell cycle distribution was determined by FACScan analysis and quantified using FCS Express 2.

RESULTS
DGK Interacts with pRB, p107, and p130 in Vitro-To determine which lipid kinases interact with pRB, we used three different GST-pRB fusion proteins (Fig. 1B): the large pocket region that includes the small pocket domain and the C terminus (GST-pRB), the small pocket domain (GST-pRB(AϩB)), and the pRB C terminus (GST-pRB(C)). These GST fusion proteins were used to affinity-purify lipid kinases from lysates of rat brain, murine erythroleukemia (MEL) cells, or human MCF7 breast cancer cells. GST-Cdc42 served as a negative control for nonspecific binding. Fig. 2A shows that, in each cell lysate, each of the three pRB constructs bound three different lipid kinase activities (i.e. PI-kinase (yielding PIP), PIP-kinase (yielding PIP 2 ), and DGK (yielding PA)), as assessed by 32 P incorporation into the respective lipid products. The ability of pRB to interact with PIP-kinases is in agreement with our previous data (19). Here, we focus on the novel finding that pRB binds to DGK. Fig. 2A shows that rat brains contain much more pRB-binding lipid kinase activity than MEL and MCF7 cells. In each of the cell lysates, GST-pRB(AϩB) (the small pocket of pRB; lanes 2, 6, and 10) shows slightly less binding of each of lipid kinase activities than the other two pRB constructs (large pocket and C terminus) (for rat brain, this is better seen at the lower exposure; lanes 1*-4*). Thus, in three different cell lysates, three different lipid kinases of the canonical PI cycle (i.e. a DGK, PI-kinase, and PIP-kinase) all associate with GST-pRB fusion proteins. These data suggest that pRB may act as a scaffold protein in order to regulate nuclear PI signaling.
To define which DGK isoform interacts with pRB, we used GST-pRB to extract DGKs from lysates of COS-7 cells overexpressing VSV-tagged DGK␣, DGK, or DGK. Interaction between DGKs and GST-pRB was assessed using in vitro DGK activity assays and Western blotting. GST-pRB extracted 30-fold more DGK activity from lysates expressing DGK compared with those expressing DGK␣ or DGK (Fig. 2B, top, compare lane 3 with lanes 1 and 5), although less DGK activity was present in the lysates compared with DGK␣ or DGK activity (compare lane 10 with lanes 9 and 11). Similarly, more DGK bound to GST-pRB than DGK␣ or DGK, as revealed by Western blotting, despite the lower expression of DGK in cell lysates (Fig. 2B, bottom). These data indicate that DGK is the predominant isotype that binds to pRB.
To assess if DGK activity was required for interaction with pRB, we mutated a conserved glycine residue within the ATP binding site. This mutant showed less than 1% of the activity in vitro but was still able to interact with GST-pRB as well as the wild-type enzyme (data not shown).
To more carefully examine the region of pRB where DGK binds, we performed a titration experiment in which increasing amounts of GST  . DGK associates with the retinoblastoma gene product (pRB) in vitro. A, endogenous DGK, PIP-kinase, and PI-kinase activities associate with GST-pRB fusion proteins. GST fusion proteins of the pRB large pocket region (GST-pRB), pRB small pocket domain (GST-pRB(AϩB)), pRB C terminus (GST-pRB(C)), and Cdc42 (GST-Cdc42; used as a control) were isolated from bacteria, and equal quantities were incubated for 2 h with cell lysates of rat brain, MEL cells, or MCF7 cells (500 g each) and assayed for lipid kinase activity by adding lipid vesicles composed of DAG, PI, PIP, and PA together with [␥-32 P]ATP and cold ATP, as detailed under "Experimental Procedures." The radiolabeled lipid products, PIP 2 , PIP, and PA, were separated by TLC and visualized by autoradiography (positions indicated). Lanes marked 1*-4* represent a short exposure of lanes 1-4 of the same TLC plate. B, GST-pRB fusion proteins specifically bind to DGK. VSV-tagged DGK␣, -, and -(or empty vector) were overexpressed in COS-7 cells, and lysates were incubated for 2 h with the denoted GST fusion proteins. 15% of affinity-purified proteins was assayed for DGK activity as described in A. Autoradiographs of the PA spots on the TLC plates are shown (top). DGK activity associated with the GST proteins is shown in lanes 1-8. DGK activity in the total lysates (2.5% of input lysate) is shown in lanes 9 -12. 85% of the affinity-purified proteins was separated by SDS-PAGE, transferred to nitrocellulose, and visualized using the VSV tag-specific antibody P5D4 (bottom). Data are representative of three experiments. C, DGK has highest affinity for the large pocket domain and the C terminus of pRB. Increasing amounts of GST-pRB, GST-pRB(AϩB), GST-pRB(C), or GST-Cdc42 (negative control) were incubated with lysates of DGK-or vector-overexpressing COS-7 cells. Associated DGK was detected by Western blotting with a DGK-specific antibody (top). Blots were reprobed with a GST-specific antibody (bottom). Total lysates represent one-tenth of input lysate.

pRB-type Proteins Bind to and Activate DGK
fusion proteins were used to bind DGK from COS-7 lysates. DGK has similar affinities for GST-pRB and GST-pRB(C) (Fig. 2C, compare lanes  3 and 8), whereas GST-pRB(AϩB) only weakly bound to DGK. The substantial DGK activity bound to GST-pRB(AϩB) from endogenous cell systems ( Fig. 2A) compared with the small amounts of DGK bound to pRB(AϩB) found on Western blot may be due to differences in the assays utilized. Based on Western blot analysis, we conclude that the C terminus of pRB is the major binding site for DGK.
Since pRB is a member of the family of pocket proteins that also include p107 and p130, we assessed if p107 and p130 also interact with DGK. Since GST-p130 and DGK have the same molecular weight, we used lysates expressing GFP-DGK to discriminate between them on SDS-PAGE. Like GST-pRB, GST-p107 and GST-p130 also bound to GFP-DGK (Fig. 3A). However, GST-p130 extracted less DGK compared with GST-pRB and GST-p107, which might reflect the lower amount of GST protein used or that p130 has a lower affinity for DGK. These data show that DGK can interact with all members of the pRB pocket protein family.
DGK Binds pRB, p107, and p130 in Cells-To demonstrate that DGK binds to pRB also in cells, we expressed pRB and FLAG-tagged DGK (FLAG-DGK), alone or in combination, in HEK293T cells and immunoprecipitated the lipid kinase from cell lysates using a specific anti-FLAG antibody. pRB and DGK were expressed in total lysates of transfected HEK293T cells (Fig. 3B, bottom), and FLAG-DGK could be immunoprecipitated. However, pRB was only detected in the immunoprecipitates from lysates expressing both pRB and DGK (Fig. 3B, top). We also tested the pRB family members p107 and p130 for DGK binding in cells. Total lysates revealed that DGK, p107, and p130 were expressed in transfected HEK293T cells (bottom of Fig. 3, C and D). Similar to pRB, p107 and p130 were only detected in the anti-FLAG immunoprecipitates when they were co-expressed with FLAG-DGK (top of Fig. 3, C and D). The interaction of p130 with DGK was lower in both the in vitro GST pull-downs and in the co-immunoprecipitations, suggesting that p130 binds DGK with a lower affinity than pRB and p107.
To demonstrate that endogenous DGK and pRB interact with each other, we immunoprecipitated DGK from MEL lysates and subjected the precipitate to immunoblotting with a pRB-specific antibody. pRB was present in the DGK immunoprecipitate but not in a control precipitate (Fig. 4A, top). Fig. 4A (bottom) confirms the presence of DGK activity and the presence of DGK protein in the DGK immunopre- . pRB co-immunoprecipitates with endogenous DGK, and vice versa. A, DGK was immunoprecipitated (IP) from 2 mg of MEL lysate using a DGK-specific antibody. Preimmune serum (control) was used as a control. Immunoprecipitates were split; 85% was used for Western blotting with a pRB-specific antibody (top), and 15% of the DGK precipitate was assayed for DGK activity, and [ 32 P]PA was separated by TLC and visualized by autoradiography (bottom). The lanes marked total lysate represent one-two thousandth of input lysate and were present on the same blot as the immunoprecipitates. The Western blot in the middle shows that DGK is indeed immunoprecipitated. This blot is from a different immunoprecipitate, because pRB and DGK have almost the same molecular weight. Total lysate represents one-twenty-fifth of input lysate. B, pRB was immunoprecipitated from 2.5 mg of lysate of differentiated MEL cells. 20% of the immunoprecipitate was assayed for DGK activity, and 80% was analyzed on Western blot. First, blots were stained for DGK using a DGK-specific antibody. Second, blots were stripped and reprobed for pRB. The lanes marked total lysates contain one-one thousandth of the input lysate for pRB and one-five hundredth for DGK and were on the same blot.

pRB-type Proteins Bind to and Activate DGK
cipitate but not in the control precipitate. To successfully immunoprecipitate pRB and determine DGK co-immunoprecipitation, we used MEL lysates of differentiated cells that contain pRB predominantly in the hypophosphorylated status, the status of pRB that binds DGK (see Fig. 6). A small fraction of pRB was immunoprecipitated (Fig. 4B, top), and DGK protein and DGK activity were specifically co-immunoprecipitated (Fig. 4B, bottom). These results indicate that endogenous pRB and DGK interact with each other.
The MARCKS-PSD of DGK Is a pRB-binding Site-The MARCKS-PSD of DGK has previously been shown to be a major determinant for the localization of DGK in the nucleus (15). We therefore postulated that the MARCKS-PSD may be important in the interaction of DGK with pRB. To test this hypothesis, we tested the interaction between pRB and a DGK mutant in which the MARCKS domain was deleted (DGK-⌬MARCKS). Whereas both wild type (wtDGK) and DGK-⌬MARCKS were equally expressed (Fig. 5A, compare lane 17 with lane 18), DGK-⌬MARCKS was hardly detectable compared with wtDGK in the GST-pRB fusion protein precipitates (lanes 5-8). Since the MARCKS-PSD contains a large number of basic amino acids that might be important for electrostatic interaction between pRB and DGK, we tested a DGK-MARCKS-PSD mutant in which all basic amino acids of the nuclear localization signal were substituted for alanines (DGK-K/ R3 A) for pRB binding. Similar to the MARCKS-PSD deletion mutant, the DGK-K/R3 A mutant failed to interact with GST-pRB (lanes 9 -12), although the protein was expressed at a higher level compared with wtDGK (compare lane 19 with lane 17).
To test whether the DGK-MARCKS-PSD was sufficient to mediate interaction with pRB, we used a biotinylated DGK-MARCKS-PSD peptide to affinity-purify pRB from MCF7 cell lysates. The DGK-MARCKS-PSD peptide bound to pRB, whereas a biotinylated control peptide was unable to bind pRB (Fig. 5B). Furthermore, the interaction between full-length DGK and GST-pRB was inhibited by the DGK-MARCKS-PSD peptide (Fig. 5C, lane 3) but not by the control peptide (Fig. 5C, lane 4). Together, these results indicate that the DGK-MARCKS-PSD is important in mediating the interaction between pRB and DGK.
The MARCKS-PSD of DGK Binds to pRB Directly-To determine whether pRB directly binds to DGK, we tested whether purified and eluted GST-pRB could be extracted by the biotinylated DGK-MARCKS-PSD peptide. Purified GST-pRB specifically bound to the DGK-MARCKS-PSD peptide coupled to streptavidin-agarose (Fig.  5D), whereas GST-Cdc42 did not bind. GST-pRB was not extracted by a control peptide. These results suggest that pRB binds to the MARCKS-PSD of DGK directly.
DGK Binds Hypophosphorylated pRB-pRB regulates cell cycle progression through its interaction with the transcription factor E2F. During G 1 , pRB exists in a hypophosphorylated state and can bind to and inactivate E2F. When cells progress to S-phase, pRB becomes highly phosphorylated by cyclin-cyclin-dependent kinase complexes, which leads to the release of E2F, enabling E2F-mediated transcription of genes required for S-phase progression (24). In total cell lysates, pRB is Lysates were incubated with the indicated GST fusion proteins, and associated proteins were analyzed by immunoblotting using a DGK-specific antibody. GST fusion protein precipitates are shown in lanes 1-16, and total lysates are shown in lanes 17-20. B, DGK-MARCKS-PSD peptide specifically binds to pRB. Biotinylated DGK-MARCKS-PSD peptide or biotinylated control peptide were incubated overnight with MCF7 cell lysates and immobilized on streptavidin-agarose beads. Associated pRB was visualized by immunoblotting using an anti-pRB antibody. Total lysates contain one-twenty-fifth of input lysate. C, full-length DGK binding to pRB is blocked by a DGK-MARCKS-PSD peptide. Lysates of COS-7 cells overexpressing DGK were incubated for 2 h with 1 or 10 g of indicated peptides together with GST-pRB immobilized on glutathione-Sepharose 4B beads. Affinity-purified DGK was visualized by immunoblotting using an anti-DGK antibody (lanes 1-6). 10% of input lysates are shown in lanes 7 and 8. D, purified eluted GST-pRB directly binds to DGK-MARCKS-PSD peptide. 250 ng of purified eluted GST-pRB or GST-Cdc42 was incubated overnight with 100 g of biotinylated DGK-MARCKS-PSD or control peptide. Peptides were immobilized on streptavidin-agarose beads and associated GST proteins were analyzed by Western blotting using a GST-specific antibody. The lanes marked input contain one-tenth of the GST fusion proteins used in the assay. present in both the low and highly phosphorylated state (see the doublet in Fig. 3B, bottom), whereas only one band is detectable in the DGK immunoprecipitate (Fig. 3B, top), suggesting that the interaction between DGK and pRB may be dependent on the phosphorylation status of pRB. To further test this, we used the osteosarcoma cell line SAOS-2 that lacks functional pRB. When pRB is overexpressed in SAOS-2 cells, it is not phosphorylated (Fig. 6, lane 7) and causes a cell cycle arrest. Co-expression of pRB with cyclin E, however, leads to hyperphosphorylation of pRB (lane 8), which attenuates the pRB-mediated cell cycle arrest (25). We purified GST-DGK from COS-7 cells and used it to affinity-purify pRB from lysates of SAOS-2 cells expressing pRB alone, or co-expressing pRB and cyclin E. Hypophosphorylated pRB (lane 1) specifically bound to GST-DGK, but in the presence of cyclin E when pRB is hyperphosphorylated (lane 2), binding was almost undetectable (the minor amount of pRB in lane 2 is hypophosphorylated). These results indicate that, similar to E2F, DGK preferentially binds to the hypophosphorylated form of pRB.
DGK Does Not Affect Sequestering and Inactivation of E2F by pRB-Since DGK interacts specifically with hypophosphorylated pRB, we questioned if DGK might influence pRB-mediated regulation of E2F transcriptional activity. Therefore, we used a reporter construct with a promotor containing six E2F binding sites upstream from the firefly luciferase reporter coding region. Binding of E2F to the promotor drives transcription of the luciferase reporter. C33A cells were cotransfected with the E2F luciferase reporter construct, a control Renilla luciferase receptor construct, E2F1, and DP1, in the absence or presence of pRB and/or DGK. E2F and DP1 caused a 4.5-fold stimulation of E2F promoter activity compared with background E2F activity. This stimulation was 50 and 75% reduced by co-transfection of 100 and 250 ng, respectively, of pRB expression plasmid (Fig. 7). The addition of DGK plasmid did not affect basal E2F activity or pRB-mediated inhibition of E2F activity, indicating that DGK does not affect the regulation of E2F activity by pRB.
pRB, p107, and p130 Stimulate DGK Activity-To explore the function of the interaction between DGK and pRB, we tested whether pRB could regulate DGK activity. Therefore, we compared the activity of HA-DGK immunoprecipitated with an anti-HA antibody with HA-DGK affinity-purified by GST-pRB. Immunoprecipitation with the anti-HA antibody yielded DGK that was not bound to pRB, whereas affinity purification with GST-pRB ensured that all of the DGK present on the beads interacted with pRB. The amount of DGK protein on the beads was determined by immunoblotting, whereas DGK activity was assessed by an in vitro assay. As shown in Fig. 8A, the amount of immunoprecipitated HA-DGK in lane 2 was comparable with the amount of HA-DGK affinity-purified by GST-pRB in lane 6. However, 7.5-fold more DGK activity was associated with GST-pRB-bound HA-DGK compared with HA-immunoprecipitated DGK. Since the HA tag antibody did not interfere with HA-DGK activity (data not shown), these results suggest that DGK is more active when in a complex with pRB.
To further verify that all of the pocket protein family members could stimulate DGK activity, we purified GST-pRB, -p107, and -p130 and assessed their effects on DGK activity in vitro. HA-DGK was immunoprecipitated from COS-7 cell lysates and eluted from the beads using an HA-peptide. Purified GST-pRB, GST-p107, and GST-p130 were eluted from the beads using glutathione. Purified HA-DGK and GST fusion proteins were combined on ice, and complexes were allowed to form prior to the DGK activity assay. GST-pRB, GST-p107, and GST-p130 enhanced DGK activity 5-, 3.5-, and 4.5-fold, respectively, in a concentration-dependent manner, whereas a GST-Cdc42 control did not affect DGK activity (Fig. 8B). Together, these results indicate that pRB family members activate DGK in vitro.
Overexpression of DGK Can Partially Rescue the Loss of a G 1 Arrest after ␥-Irradiation in pRB-null MEFs-In order to establish a physiological role for DGK in pRB-dependent signaling pathways, we studied the G 1 arrest induced by ␥-irradiation. Cell cycle arrest in response to ionizing radiation is a well established tumor-suppressive pathway that is dependent on the growth-suppressive activity of pRB. This pathway is completely blocked in pRB-null mouse embryonic fibroblasts (MEFs) (26). We postulated that, if DGK kinase activity is enhanced by pRB, then overexpression of DGK may partially substitute for the loss of pRB in ␥-irradiation-induced cell cycle arrest. MEFs isolated from pRB-null mice were transduced with viral constructs encoding vector (pRBϪ/Ϫ vector), kinase-inactive DGK (pRBϪ/Ϫ kdDGK), or wild-type DGK (pRBϪ/Ϫ wtDGK). As a control, MEFs were isolated from wild-type mice. In all cases, MEFs were immortalized by prior transduction with TBX2, which blocks passage-induced senescence. MEFs were ␥-irradiated and, after 30 min, treated with nocodazole to arrest them in G 2 /M. Cells arrested in G 1 were assessed by fluorescence-activated cell sorting analysis. As shown in Fig. 9, ␥-irradiation of wild-type MEFs (pRBϩ/ϩ) led to a dose-dependent increase in the number of cells arrested in G 1 . As expected, irradiation of pRB-null MEFs (pRBϪ/Ϫ vector) did not lead to any increase in the number of cells in G 1 . Consistent with a role for DGK in pRB signaling, overexpression of DGK in pRB-null MEFs led to a partial rescue of the arrest defect at all doses tested. The rescue was dependent on the activity of DGK, since it was not observed in cells transduced with the kinase-inactive DGK. . DGK binds to hypophosphorylated pRB. GST-DGK was isolated from COS-7 cells and incubated with lysates of SAOS-2 cells that were transfected with pRB and/or cyclin E as the indicated constructs. Affinity-purified proteins (lanes 1-6) and total lysates representing one-twentieth of input lysates (lanes 7-9) were subjected to Western blotting using a pRB-specific (top), GST-specific (bottom left), or cyclin E-specific antibody (cycE, top right). FIGURE 7. DGK does not affect regulation of E2F by pRB. C33A cells were co-transfected with indicated constructs. Cells were lysed 48 h after transfection and assayed for firefly luciferase and Renilla luciferase activity using a luminometer. The firefly luciferase data were corrected for Renilla luciferase activity and plotted in the histogram shown as means Ϯ the range of the duplicates (n ϭ 2).

DISCUSSION
In this study, we show that pRB specifically interacts with DGK in vitro and in vivo. The C terminus of pRB is required for the interaction with DGK, whereas the MARCKS-PSD of DGK is sufficient to mediate the interaction. Furthermore, the interaction between pRB and DGK is dependent on the phosphorylation status of pRB, since DGK only binds active hypophosphorylated pRB. The interaction between DGK and pRB does not appear to modulate the repression of E2F transcriptional activity by pRB. However, we show that pRB and other pocket protein family members are potent activators of DGK activity in vitro.
We previously demonstrated that Type I phosphatidylinositol 4-phosphate 5-kinase, an enzyme that converts phosphatidylinositol 4-phosphate into PI(4,5)P 2 , interacts with and is activated by pRB (19). In this paper, we now demonstrate that a PI 4-kinase that will generate phosphatidylinositol 4-phosphate and a DGK that phosphorylates DAG to PA can also interact with pRB. Furthermore, the p55 subunit of PI 3-kinase, which can generate PI(3,4,5)P 3 via the 3Ј phosphorylation of PI(4,5)P 2 , was also shown to interact with pRB (27). We suggest that pRB may act as a scaffold protein to integrate nuclear PI signaling and may provide a link between cell cycle regulation and changes in nuclear signaling lipids. For example, as cells progress from G 1 to S phase, nuclear levels of DAG, PA, PIP, and PIP 2 have been shown to change (9,28).
In addition to pRB binding, DGK also interacts with the pRB-related pocket proteins p107 and p130. pRB, p107, and p130 are highly similar within the pocket region, but also regions beyond the pocket domain are conserved. To date, almost all p107-and p130-binding proteins also bind to pRB, whereas most pRB-binding proteins have not been tested for binding to p107 and p130 (21). All pocket proteins show substantial functional overlaps as well as some unique functions (21,29). When overexpressed, they all can arrest the appropriate cells in G 1 -phase of the cell cycle and inhibit E2F-mediated gene transcription and are all phosphorylated by cyclin-cyclin-dependent kinase complexes. However, p107 and p130 bind different E2F family members than pRB and therefore regulate transcription of different sets of genes. Also, their expression patterns differ during the cell cycle; the levels of pRB are  3, 7, and 8. Immunoprecipitates and affinity-purified proteins were split, and 20% was assayed for DGK activity, using substrate and [␥-32 P]ATP. [ 32 P]PA was separated by TLC and quantified using phosphoimaging. The data are plotted in the histogram and are shown as means Ϯ S.D. (n ϭ 3). The other 80% was used to visualize the amount of DGK affinity-purified by either the HA-specific antibody or GST-pRB by immunoblotting using a DGK-specific polyclonal antibody (bottom). Note that lane 2 and lane 6 contain the same amount of DGK protein, whereas the activity of HA-DGK bound to GST-pRB was 7.5-fold higher than the immunoprecipitated HA-DGK. B, HA-DGK overexpressed in COS-7 cells was immunoprecipitated using an anti-HA antibody and eluted from the beads using HA peptide. Eluted HA-DGK was incubated with the indicated amounts of GST fusion proteins that were isolated from bacteria, eluted from the beads using glutathione, and quantified relative to bovine serum albumin standards on a Coomassiestained gel. In vector controls, 100 ng of the indicated GST fusion proteins was added. All samples contained 5 g of carrier bovine serum albumin. Mixtures were assayed for DGK activity as in A.
Results are the means Ϯ the range of the duplicates (n ϭ 2) and representative of three experiments.

pRB-type Proteins Bind to and Activate DGK
constant during the cell cycle, whereas p107 expression peaks during S-phase of the cell cycle and p130 is highly expressed in quiescent cells. Furthermore, they appear to have specific functions in differentiation. For example, deletion of pRB attenuates adipocyte differentiation, whereas deletion of p107 acts to sensitize MEFs to adipocyte differentiation (30).
In this study, we show that overexpression of DGK is able to partially rescue a cell cycle arrest defect in response to ␥-irradiation in pRB-null MEFs. Irradiation of MEFs is known to induce pRB activity, which leads to a cell cycle arrest required to prevent cells from entering S-phase with damaged DNA (26,31). The arrest is thought to allow DNA repair and thus ensure the survival of the cell. How irradiation induces a cell cycle arrest is not clear. In response to irradiation, p53 is induced and up-regulates the levels of the cyclin kinase inhibitor p21 WAF1/CIP1 . This inhibits phosphorylation of pRB, leading to its activation. The cell cycle arrest is thought to be induced by pRB-mediated attenuation of E2F transcriptional activity. However, in pRB-negative C33A cells, expression of a mutant of pRB in the LXCXE binding site, while maintaining the ability of pRB to interact with and repress E2F activity, is unable to reconstitute a DNA damage arrest, whereas wild-type pRB expression can (32). This suggests that yet another factor besides E2F repression determines induction of the cell cycle arrest in response to DNA damage. This factor could be DGK, since overexpression of DGK can partially rescue the loss of a cell cycle arrest in pRB-null cells. This would suggest that DGK acts either on a parallel pathway or lies directly downstream of pRB. Since our studies also demonstrate that the active (hypophosphorylated) form of pRB interacts with and stimulates DGK activity, we favor the latter suggestion. Furthermore, since a kinase-inactive DGK is unable to reinitiate a cell cycle arrest, it appears that either the removal of DAG or the generation of PA is important for the cell cycle arrest.
An interesting possibility by which overexpression of DGK allows regulation of a cell cycle arrest in response to irradiation may be through the other pocket protein family members, p107 and p130. The use of triple-knock-out MEFs may be useful for testing this possibility.
In accord with a role for DGK to attenuate DAG as possible (co-)regulator of the cell cycle, previous studies have demonstrated that overexpression of a phospholipase C, an enzyme that generates DAG by hydrolysis of PI(4,5)P 2 in the nucleus, increases the number of cells entering S-phase in response to insulin-like growth factor 1 (33,34). In contrast, suppression of nuclear phospholipase C activity inhibits entry of Swiss 3T3 cells into S-phase (7,35). Furthermore, overexpression of phospholipase C in the nucleus inhibits terminal differentiation of MEL cells in response to Me 2 SO treatment, whereas phospholipase C suppression augments differentiation (36). These data are consistent with a role for nuclear DAG in modulating cell cycle progression and differentiation. A more direct study has demonstrated that overexpression of DGK within the nucleus slows down cell cycle progression with an accumulation of cells in G 1 phase (15).
How nuclear DAG/PA levels regulate cell cycle progression is not clear. DAG is a potent activator of PKC, an enzyme capable of regulating cell cycle progression indirectly by modulating cyclin-dependent kinase inhibitors, including p21 WAF1/CIP1 (37) and/or directly via its ability to phosphorylate pRB (38). In contrast to DAG, there are no clear roles for PA in cell cycle progression. However, it is noteworthy that protein phosphatase PP1␥ is potently inhibited by its interaction with PA (39). Interestingly, PP1␥ is a nuclear protein, whose nuclear location is regulated during cell cycle progression (40). Furthermore, studies in yeast have shown that PA may act as a regulator of transcription factors (41).
Interestingly, PKC itself is also able to phosphorylate DGK within the MARCKS-PSD, which leads to an inhibition of DGK nuclear accumulation and an inhibition of its activity (15,18). This suggests that pRB and PKC have opposing effects on DGK activity and, by consequence, on nuclear DAG/PA levels. Such a mechanism would allow tight regulation of nuclear DAG/PA levels. It will be interesting to determine what role PKC-mediated phosphorylation of the DGK-MARCKS-PSD plays in modulating the interaction between DGK and pRB. Altogether, our data show that DGK binds to and is activated by pRB, p107, and p130. We suggest that pocket proteins act as scaffolds to regulate nuclear inositide metabolism and to regulate the levels of the second messengers DAG and PA. Our data are consistent with a physiological role for either DAG or PA in modulating a pRB-mediated cell cycle arrest in response to DNA damage.