Identification and Characterization of a Novel Human Type II Diacylglycerol Kinase, DGKκ*

Diacylglycerol kinase (DGK) plays an important role in signal transduction through modulating the balance between two signaling lipids, diacylglycerol and phosphatidic acid. Here we identified a tenth member of the DGK family designated DGKκ. The κ-isozyme (1271 amino acids, calculated molecular mass, 142 kDa) contains a pleckstrin homology domain, two cysteine-rich zinc finger-like structures, and a separated catalytic region as have been found commonly for the type II isozymes previously cloned (DGKδ and DGKη). The new DGK isozyme has additionally 33 tandem repeats of Glu-Pro-Ala-Pro at the N terminus. Reverse transcriptase-PCR showed that the DGKκ mRNA is most abundant in the testis, and to a lesser extent in the placenta. DGKκ, when expressed in HEK293 cells, was persistently localized at the plasma membrane even in the absence of cell stimuli. Deletion analysis revealed that the short C-terminal sequence (amino acid residues 1199–1268) is necessary and sufficient for the plasma membrane localization. Interestingly, DGKκ, but not other type II DGKs, was specifically tyrosine-phosphorylated at Tyr78 through the Src family kinase pathway in H2O2-treated cells. Moreover, H2O2 selectively inhibited DGKκ activity in a Src family kinase-independent manner, suggesting that the isozyme changes the balance of signaling lipids in the plasma membrane in response to oxidative stress. The expression patterns, subcellular distribution, and regulatory mechanisms of DGKκ are distinct from those of DGKδ and DGKη despite high structural similarity, suggesting unique functions of the individual type II isozymes.

Diacylglycerol kinase (DGK) plays an important role in signal transduction through modulating the balance between two signaling lipids, diacylglycerol and phosphatidic acid. Here we identified a tenth member of the DGK family designated DGK. The -isozyme (1271 amino acids, calculated molecular mass, 142 kDa) contains a pleckstrin homology domain, two cysteine-rich zinc finger-like structures, and a separated catalytic region as have been found commonly for the type II isozymes previously cloned (DGK␦ and DGK). The new DGK isozyme has additionally 33 tandem repeats of Glu-Pro-Ala-Pro at the N terminus. Reverse transcriptase-PCR showed that the DGK mRNA is most abundant in the testis, and to a lesser extent in the placenta. DGK, when expressed in HEK293 cells, was persistently localized at the plasma membrane even in the absence of cell stimuli. Deletion analysis revealed that the short C-terminal sequence (amino acid residues 1199 -1268) is necessary and sufficient for the plasma membrane localization. Interestingly, DGK, but not other type II DGKs, was specifically tyrosine-phosphorylated at Tyr 78 through the Src family kinase pathway in H 2 O 2treated cells. Moreover, H 2 O 2 selectively inhibited DGK activity in a Src family kinase-independent manner, suggesting that the isozyme changes the balance of signaling lipids in the plasma membrane in response to oxidative stress. The expression patterns, subcellular distribution, and regulatory mechanisms of DGK are distinct from those of DGK␦ and DGK despite high structural similarity, suggesting unique functions of the individual type II isozymes.
Diacylglycerol kinase (DGK) 2 phosphorylates diacylglycerol (DAG) to generate phosphatidic acid (PA) (1). DAG, which is liberated from phosphatidylinositols and other phospholipids upon cell stimulation by growth factors and other agonists, regulates a wide range of cellular functions (2,3). For instance, DAG is an allosteric activator of conventional and novel protein kinase Cs (PKCs) (3). Moreover, protein kinase D (PKD), Unc-13, chimaerins, and Ras guanyl nucleotide-releasing pro-tein have been recently found to be regulated by DAG (4 -7). Thus, DGK consumes DAG and, as a result, is responsible for attenuating DAG-mediated signals (8 -11). PA, the reaction product of DGK, has also been reported to regulate a number of signaling proteins such as phosphatidylinositol-4-phosphate 5-kinase (12,13), Ras GTPase-activating protein (14), Raf-1 kinase (15), mTOR (mammalian target of rapamycin) (16), and atypical PKC (17). Therefore, DGK is thought to play roles not only in the down-regulation of DAG signaling, but also in the production of another lipid mediator, PA. Because the cellular concentration of these signaling lipids must be strictly regulated, their interconversion by DGK is likely to be one of key processes in cellular signal transduction.
It is now recognized that DGK represents a large enzyme family. The DGK isoforms differ from each other remarkably with respect to their structures, the modes of tissue expression and enzyme properties. To date, nine mammalian DGK isozymes ␣ (18,19), ␤ (20), ␥ (21,22), ␦ (23), ⑀ (24), (25,26), (27), (28), and (29) containing in common two or three characteristic cysteine-rich, zinc finger-like structures and the catalytic region are subdivided into five groups according to their structural features (8 -11). Each subgroup is characterized by the subtypespecific functional domains such as calcium-binding EF-hand motifs and a recoverin homology domain (type I, DGKs ␣, ␤, and ␥), pleckstrin homology (PH) and sterile ␣ motif domains, and the separated catalytic region (type II, DGKs ␦ and ). There are no recognizable regulatory domains except for the zinc finger structures and the catalytic region (type III, DGK⑀), a MARCKS (myristoylated alanine-rich C kinase substrate) phosphorylation site domain and four ankyrin repeats (type IV, DGKs and ), and three (instead of two) zinc finger structures, a PH domain-like region, and a Ras-associating domain (type V, DGK). Moreover, the occurrence of alternative splicing was recently detected for six mammalian DGK genes (␥ (21), (30), ␤ (31), ␦ (32), (33), and (34) isoforms). Thus, the list of the mammalian DGK gene family is still growing. In contrast to mammals, only a few DGK isozymes have been identified in organisms such as Caenorhabditis elegans, Drosophilamelanogaster, and Arabidopsis thaliana (9 -11). Moreover, it is noteworthy that the DGK gene has not yet been detected in a unicellular organism, yeast. It is therefore likely that many mammalian DGK isozymes have essential roles in biological processes specific to higher vertebrates, such as development/differentiation, neural networking, the immune system, and tumorigenesis through modulating the balance between two bioactive lipids, DAG and PA, in microenvironments within the cells undergoing signal transduction.
Here, we identified a tenth member of the DGK family in mammals, DGK, which is classified into the type II DGK subfamily together with DGK␦ and DGK already cloned. The novel DGK isozyme contains 33 tandem repeats of Glu-Pro-Ala-Pro at the N terminus, a PH domain, two cysteine-rich structures, and the highly conserved DGK catalytic domain separated into two portions. Despite high structural similarity, the subcellular distribution of DGK was clearly different from those of DGK␦ and DGK. Moreover, DGK, but not other type II isozymes, was specifically tyrosine-phosphorylated and down-regulated in H 2 O 2treated cells.
Expression Profiling-cDNAs from human normal tissues and tumor-derived cells (Human MTC Panels I and II and Human Tumor MTC Panel) were purchased from BD Biosciences. PCR amplification was performed with Takara Ex Taq (Takara Bio Inc., Tokyo, Japan) using DGK gene-specific forward (3-RACE-2) and reverse (nucleotide positions 3040 -3066, 5Ј-ACCATCAATGGTTATCATCACCT-CATG-3Ј) primers. PCR conditions were as follows: 94°C for 4 min; 37 cycles of 94°C for 30 s, 68°C for 2 min; and 68°C for 5 min. For normalization, human glyceraldehyde phosphate dehydrogenase mRNA was simultaneously amplified (25 cycles). PCR products amplified were separated by agarose gel electrophoresis and stained with ethidium bromide.
The octyl glucoside mixed micellar assay of DGK activity was done as described previously (23). In brief, the assay mixture (50 l) contained 50 mM MOPS, pH 7.4, 50 mM octyl glucoside, 1 mM dithiothreitol, 100 mM NaCl, 20 mM NaF, 10 mM MgCl 2 , 3 mM EGTA, 10 mM phosphatidylserine, 2 mM dioleoylglycerol, and 1 mM [␥-32 P]ATP (100,000 cpm/ nmol). The reaction was initiated by adding the suspension of the immunoprecipitates (agarose beads), and continued for 5 min at 30°C. Lipids were extracted from the mixture, and PA separated by thin layer chromatography was visualized and quantified using a BAS1800 Bioimaging Analyzer (Fuji Film, Tokyo, Japan).
Protein Phosphorylation Analysis in Intact Cells-COS-7 cells (35-mm dish) expressing 3xFLAG-tagged DGK were used at 40-h post-transfection. Protein phosphorylation analysis in intact cells labeled with [ 32 P]orthophosphate was carried out as described previously (35).
In the case of tyrosine phosphorylation analysis, COS-7 cells (60-mm dish) expressing Myc-, 3xFLAG-or GFP-tagged DGK constructs were used at 24 h post-transfection. The cells were washed four times with phosphate-buffered saline and then cultured in Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin. After 12 h, the medium was exchanged with the fresh medium containing 2 mM H 2 O 2 , or 1 mM pervanadate. In this experiment, pervanadate was freshly prepared by treating 100 mM Na 3 VO 4 in phosphate-buffered saline with 200 mM H 2 O 2 for 5 min at room temperature and excess H 2 O 2 was removed by catalase (100 g/ml). After 30 min, cells were harvested in the lysis buffer (0.5 ml/60-mm dish). Myc-, 3xFLAG-, or GFP-tagged DGK was immunoprecipitated using anti-Myc 9E10 monoclonal antibody (10 g, Roche Applied Sciences), anti-FLAG (M2, 2 g), or anti-GFP antibody (2 l, Living Colors full-length A.v. polyclonal antibody, BD Biosciences) as described above and processed for subsequent Western blotting using anti-phosphotyrosine antibody (4G10, Upstate, Lake Placid, NY).
Preparation of Pig Testis Homogenates-Pig testis was obtained at a local slaughterhouse. Pig testis was homogenized in buffer containing 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.25 M sucrose, 1 mM phenylmethylsulfonyl fluoride, complete protease inhibitor mixture, and phosphatase inhibitor mixture II. After removal of debris by a low speed centrifugation (600 ϫ g for 10 min), the homogenate was used for Western blotting analysis.
Generation of Anti-DGK Antibody-The plasmid encoding a DGK antigen (amino acid residues 1050 -1271) fused with glutathione S-transferase (GST) was obtained by inserting the antigen fragment into the pGEX-6P-1 vector (Amersham Biosciences). XL1-Blue cells (Stratagene) were transformed by the pGEX-6P-1 construct, and the GST fusion protein was expressed and purified by affinity chromatography using glutathione-Sepharose according to the procedure recommended by the manufacturer (Amersham Biosciences). Rabbits were immunized by intramuscular multiple injections of 500 g of the purified fusion protein. The serum was obtained 2 weeks after the fifth injection (day 70). Anti-GST antibody was depleted from the serum by passing it on an immobilized GST column (Pierce). The anti-DGK antibody was then affinity-purified by column chromatography on Sepharose 4B (Amersham Biosciences) coupled to the GST fusion protein originally used as an antigen.
Western Blot Analysis-Cell lysates, immunoprecipitates, and pig testis homogenates were separated on SDS-PAGE. The separated proteins were transferred to a polyvinylidene difluoride membrane (Bio-Rad) and blocked with Block Ace (Dainippon Pharmaceutical, Tokyo, Japan). The membrane was incubated with anti-FLAG M2 monoclonal antibody, anti-c-Myc monoclonal antibody (9E10), anti-phosphotyrosine monoclonal antibody (4G10), anti-GFP monoclonal antibody (B-2, Santa Cruz Biotechnology), or anti-DGK polyclonal antibody in Block Ace for 1 h. The immunoreactive bands were visualized using Nucleotides and amino acids are numbered at the right, respectively. Dotted underline, in-frame stop codon in the 5Ј-untranslated sequence; wavy underline, Pro-rich sequence; dashed underline EPAP-repeats; double wavy underline, Ser-Pro repeats; thick underline, PH domain; gray underline, cysteine-rich, zinc finger-like sequences; closed circles, conserved cysteine and histidine residues in the cysteine-rich, zinc finger-like sequences; double underlines, C4-a and C4-b (catalytic subregions); box, putative PDZ-binding sequence; asterisk, the translation termination codon; arrowheads, positions of exon/intron boundaries (the exons are numbered on both sides of the arrowheads). DECEMBER 2, 2005 • VOLUME 280 • NUMBER 48 peroxidase-conjugated anti-mouse or anti-rabbit IgG antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) and ECL (Amersham Biosciences). Densitometric analysis to quantify the total and tyrosinephosphorylated DGK was performed using NIH Image 1.62 software.

Human Diacylglycerol Kinase
Fluorescence Microscopy-HEK293 cells were grown on poly-L-lysine-coated glass coverslips and transiently transfected with cDNAs encoding various DGK constructs fused N-terminally with YFP or GFP. After 24 h, HEK293 cells were washed four times with phosphate-buffered saline and then cultured in Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin. After 3 h, the medium was exchanged with the fresh medium containing 100 nM 12-O-tetradecanoylphorbol 13-acetate (TPA) for 30 min or 0.5 M sorbitol for 5 min. Stimulated cells were fixed with 3.7% formaldehyde in phosphate-buffered saline for 10 min and then washed twice with phosphate-buffered saline at room temperature. The cover slips were mounted using Vectashield (Vector Laboratories, Burlingame, CA). Cells were examined using an inverted confocal laser scanning microscope (Zeiss LSM 510). Images were processed using Adobe Photoshop CS (Adobe Systems, San Jose, CA).
Co-immunoprecipitation Analysis-COS-7 cells (60-mm dish, ϳ1 ϫ 10 7 cells) co-expressing 3xFLAG-DGK and GFP-DGK were harvested in the lysis buffer. The mixture was centrifuged at 100,000 ϫ g for 30 min at 4°C to give cell lysates. FLAG-tagged DGK was immunoprecipitated as described above and processed for Western blotting to detect co-precipitated proteins.

RESULTS
Identification and Isolation of Human DGK cDNA-In the course of performing a genome data base search, we found a predicted protein product (core region in Fig. 1A, XP 066534, locus: chromosome X p11. 22), in which sequences of cysteine-rich zinc finger-like structures and an N-terminal half of catalytic domain (C4-a) are highly similar to those of DGKs ␦ and (Fig. 1, A and B). This high sequence similarity strongly suggests that the gene product is a novel member of the DGK family. However, in the predicted gene product, the PH domain and the C-terminal half of catalytic region (C4-b) are only partly similar to those of DGK␦ or DGK (Fig. 1A), implying that the exon/intron boundaries predicted to compose XP 066534 are not correct. Thus, to reveal the complete sequence of this gene, we performed 5Ј-and 3Ј-RACE using human testis cDNA as a template (Fig. 1A). In the 5Ј-RACE products, we found three independent cDNA clones that have an in-frame stop codon located upstream of the first ATG codon (Fig. 1, A and C). Moreover, three independent cDNA clones containing a stop codon and a poly(A) tract were identified in the 3Ј-RACE products.
The composite cDNA contains a 3816-nucleotide open reading frame (nucleotide positions 61-3876) (Fig. 1). Analysis of two genomic clones (GenBank TM accession no. AL357894 and AL591367) showed that the open reading frame of the gene utilizes 28 exons (Fig. 1A), and that all of the exon/intron boundaries fulfill the GT/AG rule (data not shown). The flanking sequence of the initiation ATG of the open reading frame conforms well to the Kozak consensus sequence (CC(A/G)C-CATG(A/G)) (Fig. 1C). The deduced amino acid sequence encodes a  protein of 1271 amino acid residues with calculated molecular mass of 142 kDa.
Analysis of the Deduced Amino Acid Sequence for DGK-The primary structure of the gene product contains the complete DGK catalytic region and two cysteine-rich zinc finger-like structures (His 328 -Cys 377 and His 399 -Cys 449 ), namely, His-X 10 -12 -Cys-X 2 -Cys-X 12-14 -Cys-X 2 -Cys-X 4 -His-X 2 -Cys-X 5-7 -Cys, which are conserved in all of the DGK isozymes cloned so far (Fig. 1, B and C). Therefore, we designated this protein DGK. The catalytic domain (C4) of DGK is separated into two subdomains, C4-a and C4-b. Such a separated C4 domain has already been found in the type II DGK isozymes, DGK␦ and DGK, previously cloned (23,27,32,33). Moreover, DGK has a PH domain highly similar to those of the type II DGKs. Thus, DGK is classified into the type II DGK subfamily (Fig. 1B). The COILS software (www.ch.embnet.org/software/COILS_form.html) revealed that, as observed in DGKs ␦ and , DGK contains two putative coiled-coil structures (also called leucine zipper, amino acid residues 1065-1103 and 1163-1193) in the C-terminal region (Fig. 1B).
One of the striking structural features of DGK is that there are 33 tandem repeats of Glu-Pro-Ala-Pro (EPAP repeats, Glu 48 -Pro 179 ) at the N terminus (Fig. 1). In addition to the EPAP repeats, the N-terminal region of this isozyme contains a Pro-rich region (Pro 24 -Pro 44 ) and five Ser-Pro repeats (Ser 196 -Pro 205 ). DGKs ␦1, ␦2, and 2 have a sterile ␣ motif domain at their C termini (Fig. 1B) (23,32,33). However, DGK does not contain this domain. We noted that the last three residues (Ser-Gln-Leu) at the C terminus represent a potential target sequence, (Ser/Thr)-X-⌽ for class I PDZ domains (⌽, a hydrophobic amino acid, usually Val, Ile, or Leu; X, unspecified amino acid) (37).
Tissue Distribution of DGK-To investigate the expression pattern of DGK, panels of cDNAs synthesized from human normal and tumor tissues were analyzed by PCR. As shown in Fig. 2, the DGK mRNA was expressed in the testis, and to a lesser extent in the placenta. Moreover, the DGK transcript was not detected in any of the tumor-derived cells (Human Tumor MTC Panel) examined (data not shown). With regard to the expression patterns of the type II DGKs, we previously reported that the DGK␦2 mRNA, an alternative splicing product, was detectable in all normal and tumor tissues and cell lines examined, whereas the DGK␦1 mRNA was detected only in the ovary (32). The DGK1 transcript was detectable in most of the normal tissues and all of the tumorderived cells examined, and that of DGK2, an alternative splicing product, was detected in testis, kidney, and colon (33). Therefore, the expression pattern of DGK is quite different from those of other type II DGK isoforms, DGKs ␦1, ␦2, 1, and 2.   To confirm the protein expression of endogenous DGK, we performed immunoblot analysis of pig testis using anti-DGK antibody raised against a C-terminal portion of the isozyme. First, we characterized the antibody using COS-7 cells expressing 3xFLAG-tagged DGK or other type II DGK isoforms. The anti-DGK antibody recognized only 3xFLAG-DGK with an apparent molecular mass of 180 kDa, whereas no bands were detected in the control cells expressing 3xFLAG-tagged DGK␦1 (DGK␦2 shares the same C-terminal sequence, see Fig. 1B), DGK1 or DGK2 (Fig. 3A). Moreover, the recognition of DGK was blocked by preincubation of the antibody with the peptide antigen, confirming the specificity of the anti-DGK antibody. An apparent molecular mass of intact DGK without tag was 170 kDa (Fig.  3B). As shown in Fig. 3C, we detected the expression of endogenous DGK with an apparent molecular mass of 170 kDa in pig testis. We further confirmed that the recognition of the protein was blocked by the peptide antigen. A lower band (155 kDa) may be a proteolytic product of the parent band or an alternative splicing product.
Catalytic Activity of DGK-To address whether the isolated cDNA encodes a catalytically active enzyme, we measured DGK activity in immunoprecipitates obtained by using anti-FLAG antibody from COS-7 cells expressing 3xFLAG-DGK (Fig. 4). The precipitates from vector-transfected cells showed no detectable activity. However, DGK activity toward 1,2-dioleoyl-sn-glycerol was clearly detected in the immunoprecipitates containing DGK, confirming that the DGK cDNA encodes a functional enzyme. We did not observe a preference of DGK for 1-stearoyl-2-arachidonoyl-sn-glycerol (data not shown).
No Detectable Homo-oligomer Formation of DGK-We recently found that DGK␦1, DGK␦2, and DGK2 formed oligomeric structures in vitro and in vivo and that the SAM domain played a critical role in the oligomer formation (32,33,35). To examine whether DGK forms homo-oligomer structures in vivo, we performed co-immunoprecipitation analysis using COS-7 cells co-expressing 3xFLAG-tagged and GFPtagged DGK proteins. As reported previously (35), when 3xFLAG-DGK␦1 was immunoprecipitated with anti-FLAG antibody, GFP- DGK␦1 was co-immunoprecipitated (Fig. 5). However, GFP-DGK was not co-immunoprecipitated with 3xFLAG-DGK (Fig. 5). This result indicates that DGK is unable to form homo-oligomer structures.
Subcellular Localization of DGK-We next investigated the subcellular localization of DGK. Interestingly, DGK expressed in HEK293 cells was localized at the cell periphery, probably at the plasma membrane, even in the absence of cell stimuli (Fig. 6A). Because we have recently found that, when expressed in HEK293 or COS-7 cells, DGKs, ␦1, ␦2, 1, and 2, were all distributed in the cytoplasm under unstimulated conditions (32,33,35) (also see Fig. 6A for DGK␦1 and Fig. 6B for DGK2, respectively), the subcellular localization of DGK is clearly distinct from those of other type II DGKs.
We previously reported that TPA stimulation induced translocation of DGK␦1 from the cytoplasm to the plasma membrane (35,36). Thus, we examined whether TPA stimulation affected the localization pattern of DGK. We confirmed that DGK␦1 was translocated from the cytoplasm to the plasma membrane in TPA-stimulated cells (Fig. 6A). However, TPA stimulation did not affect the plasma membrane localization of DGK.
Because osmotic shock (500 mM sorbitol) induced translocation of DGKs 1 and 2 from the cytoplasm to endosomes in COS-7 cells (33), we intended to see the subcellular distribution of DGK under osmotic shock. We confirmed that DGK2 was translocated from the cytoplasm to punctate vesicles in osmotically shocked HEK293 cells (Fig. 6B) as previously reported (33). On the contrary, the plasma membrane localization of DGK was not markedly affected in sorbitol-treated cells (Fig.  6B). Taken together, it is revealed that the subcellular localization of DGK is clearly distinct from those of other type II DGKs in both quiescent and/or stimulated cells.
To map the domain of DGK responsible for its persistent cell peripheral localization, we examined in HEK293 cells the subcellular localization of a series of DGK deletion mutants. We found that a mutant lacking the C-terminal region (DGK-(1-1033)) failed to be localized at the plasma membrane (Fig. 7). We next attempted to narrow down the C-terminal region responsible for its plasma membrane localization. The last three residues (Ser-Gln-Leu) of DGK are a potential target sequence of class I PDZ-domains (Figs. 1 and 7A). However, a mutant lacking the PDZ binding motif (DGK-(1-1268)) exhibited the same localization as that given by wild-type DGK. The coiled-coil structure is known as a protein-protein interaction motif (38). However, we found that a mutant still containing the coiled-coil structures (DGK-(1-1198)) failed to be localized at the plasma membrane (Fig.  7). Moreover, although the short C-terminal portion (DGK-(1199 -1268)) lacks the coiled-coil structures, this mutant was clearly located at the plasma membrane. These results indicate that the 70 amino acids (amino acids 1199 -1268), but not the coiled-coil structures or the PDZ binding motif, in the C-terminal region of DGK is necessary and sufficient for the plasma membrane localization. In this case, a mutant lacking the N-terminal EPAP repeats gave essentially the same localization pattern as that of wild-type enzyme (data not shown). We previously reported that the PH domain of DGK␦1 was necessary and sufficient for the plasma membrane localization in TPA-stimulated cells (32,36). However, in contrast to DGK␦1, the PH domain alone of DGK failed to be localized at the plasma membrane (data not shown).
Phosphorylation of DGK in Unstimulated Cells-We have demonstrated that DGK␦1 and DGK1 were phosphorylated upon TPA cell stimulation (33,35,36). Thus, we examined protein phosphorylation of DGK expressed in COS-7 cells. In contrast to DGK␦1 used as a positive control, DGK was phosphorylated even without cell stimuli, and the phosphorylation was not significantly affected by TPA stimulation (Fig.  8). Because tyrosine phosphorylation of DGK was not detected in resting cells (Fig. 9, A-C), it is likely that the basal phosphorylation occurred at serine and/or threonine residues.
Tyrosine Phosphorylation of DGK in Response to H 2 O 2 -We next examined tyrosine phosphorylation of DGK induced by growth factor stimulation or oxidative stress. The addition of epidermal growth factor (its receptor was expressed in COS-7 cells) did not induce detectable tyrosine phosphorylation of DGK (data not shown). On the other hand, we found that, when cells were treated with H 2 O 2 , the isozyme was tyrosine-phosphorylated in both dose-and time-dependent manners (Fig. 9, A and B). Maximum tyrosine phosphorylation of DGK was reached within 30 min of incubation in the presence of 2 mM H 2 O 2 . Moreover, the tyrosine phosphorylation was significantly enhanced by treatment with pervanadate, a protein-tyrosine phosphatase inhibitor (Fig. 9C), indicating that tyrosine phosphorylation of DGK was negatively regulated by protein-tyrosine phosphatase.
It is known that H 2 O 2 induces activation of Src family and c-Abl tyrosine kinases in some cells (39). Thus, we next examined the effect of PP2, a specific Src family kinase inhibitor, on tyrosine phosphorylation of DGK (Fig. 9D). In the presence of PP2, the tyrosine phosphorylation level of DGK was significantly reduced in H 2 O 2 -treated cells. Essentially the same result was obtained using pervanadate-treated cells. However, PP3, an inactive PP2 analog, did not prevent the DGK tyrosine phosphorylation. AG957, a selective c-Abl inhibitor, failed to block the phosphorylation (data not shown). These data indicate that DGK is a new downstream target of the Src family tyrosine kinase pathway.
We next asked whether the other type II DGK isoforms were also tyrosine-phosphorylated in response to H 2 O 2 . As shown in Fig. 9E, all of the other type II DGKs, ␦1, ␦2, 1, and 2 failed to exhibit detectable tyrosine phosphorylation in H 2 O 2 -treated cells. Moreover, DGK␣ (type I isozyme) also did not show detectable tyrosine phosphorylation in the same experiment (data not shown). Essentially the same results were obtained using pervanadate-treated cells (data not shown). These results strongly suggest that the tyrosine phosphorylation in response to oxidative stress occurs in a isozyme-specific manner. . FLAG-tagged proteins were immunoprecipitated with anti-FLAG antibody, and the aliquots were analyzed by SDS-PAGE. The radioactive signal was detected by a BAS1800 Bio-imaging analyzer (Phosphorylation), and precipitated 3xFLAG-tagged proteins were visualized by Western blotting with anti-FLAG antibody (WB). The bottom panel shows the relative phosphorylation levels. The phosphorylation levels were normalized for signal intensities of DGK protein bands visualized by Western blotting. In this case, the value obtained for DGK␦1ϩTPA was set at 100.
Motif Scan (scansite.mit.edu) showed that Tyr 78 (Ala-Thr-Glu-Leu-Tyr*-Thr-Glu-Pto-Thr, where the asterisk indicates the putative phosphorylation site) and Tyr 1075 (Ser-Asp-Glu-Glu-Tyr*-Ala-Gln-Met-Gln) are putative phosphorylation sites by Src family tyrosine kinases. To determine whether these tyrosine residues are phosphorylated in response to oxidative stress, we generated two point mutants, DGK-Y78F and DGK-Y1075F, by replacing Tyr 78 and Tyr 1075 with Phe. As shown in Fig. 9F, the phosphorylation level of DGK-Y78F was markedly reduced in both H 2 O 2 -and pervanadate-treated cells. However, the phosphorylation level of DGK-Y1075F was not attenuated (data not shown). We confirmed that DGK-Y78F was localized at the plasma membrane as observed for the wild-type enzyme (data not shown). These results strongly suggest that Tyr 78 is the major tyrosine phosphorylation site of DGK in response to oxidative stress.
Inhibition of DGK Activity in H 2 O 2 -treated Cells-We next attempted to reveal further effects of the oxidative stress on DGK. Although we first examined subcellular localization of the isozyme in H 2 O 2 -treated cells, no significant changes were observed (data not shown). Interestingly, we found that the catalytic activity of DGK was suppressed in response to H 2 O 2 in both dose-and time-dependent manners (Fig. 10, A and B). We next asked whether the catalytic activities of the other type II DGK isoforms were also affected by H 2 O 2 treatment. As shown in Fig. 10C, although the enzyme activity of DGK was markedly reduced (ϳ50% inhibition), those of DGK␦2 and DGK1 remained unchanged. In these experiments, the amount of enzyme protein expressed was not affected by the H 2 O 2 -treatment. These results suggest that the inhibition induced by H 2 O 2 is a unique event of the -isozyme and that the isozyme changes the balance of DAG and PA in the plasma membrane in response to oxidative stress.
We next tested the effect of PP2 on the enzyme activity in H 2 O 2treated cells. As shown in Fig. 10D, PP2 failed to block the H 2 O 2 -induced suppression of the enzyme activity. This result strongly suggests that the H 2 O 2 -induced suppression of DGK activity is regulated via the Src family kinase-independent pathway. To confirm this conclusion, we determined the enzyme activity of DGK-Y78F, which exhibited significantly reduced tyrosine phosphorylation (Fig. 9F), in H 2 O 2 -treated cells. H 2 O 2 treatment inhibited the enzyme activities of wild-type DGK and DGK-Y78F to a similar extent (ϳ60% inhibition) (Fig. 10E). Taken together, it was revealed that the tyrosine phosphorylation and the inhibition of DGK are regulated through different signaling pathways.

DISCUSSION
Our work adds a tenth member, DGK, to the list of mammalian DGK family. The cysteine-rich, zinc finger-like structures and the catalytic region of DGK have highest sequence similarities with those of the type II DGKs (DGK␦ and DGK) (Fig. 1). Moreover, DGK shares the domain motifs that typify the type II DGKs, i.e. a PH domain and a separated catalytic domain. Thus, DGK is classified into the type II subfamily. Worthy of note is that, in addition to the common domains, DGK has a Pro-rich region, 33 tandem EPAP repeats, and Ser-Pro repeats at the N terminus. When expressed in COS-7 cells, a mutant lacking these repeats gave a DGK activity similar to that of wild-type enzyme (data not shown). Furthermore, these repeats are not involved in the regulation of the intracellular enzyme distribution (data not shown). Thus, functions of these sequences remain unclear at present. As shown in Fig. 3B, the apparent molecular mass of DGK (170 kDa) was markedly larger than the expected size (142 kDa). However, the apparent increase in size of a mutant lacking the EPAP repeats was not detected (both the apparent and expected molecular masses were 120 kDa) (data not shown). The result indicates that the slow migration rate on SDS-PAGE is because of the EPAP repeats, a highly acidic region; acidic proteins usually migrate with apparent molecular weights higher than expected on electrophoresis (40).
Interior to the catalytic domain (C4-a), all of the DGK isozymes identified so far have the putative ATP binding consensus motif, Gly-X-Gly-X 2 -Gly, in which a mutation at the second glycine in Drosophila DGK2 causes its enzymatic inactivation and, as a result, the rdgA phenotype (41). In DGK, the third glycine in the consensus motif is replaced with serine (Ser 554 ) (Fig. 1C). However, DGK clearly showed catalytic activ-ity (Fig. 4). Moreover, we found that, compared with the wild-type enzyme, a mutant (DGK-S554G) that had Ser 554 in the consensus replaced with glycine exhibited an indistinguishable level of activity (data not shown). Thus, the putative ATP binding site conserved for DGK family members is now represented by Gly-X-Gly-X 2 -Gly/Ser.
We previously reported that DGK␦2 and DGK1 were ubiquitously expressed in all tissues examined (32,33). On the other hand, the mRNAs of DGK␦1 (ovary and spleen (32)), DGK2 (testis, kidney, and colon, Ref. 33) and DGK (testis and placenta, see Fig. 2) showed a limited tissue distribution. The results suggest that DGKs ␦1, 2, and play more specialized roles in regulating cellular DAG and PA levels when compared with DGKs ␦2 and 1. Moreover, although the mRNA levels of DGKs ␦1, ␦2, 1, and 2 were significantly or moderately augmented in many tumor-derived cells (32,33), the DGK transcript was not detected in any of the tumor-derived cells examined (data not shown). This suggests that although specialized functions of DGK remain unknown, it plays important roles in quiescent cells, but not in rapidly growing cells.
Many DGK isozymes such as DGK␣ (42) 36), DGK (44), and DGK (45) were reported to be translocated from the cytoplasm to the plasma membrane, in which they are supposed to accomplish their roles in response to various stimuli such as TPA and hormones. However, DGK was found to be persistently located at the plasma membrane even in the absence of cell stimuli (Fig. 6), suggesting again that the isozyme has important functions in resting cells. The PH domain of DGK␦1 and the cysteine-rich, zinc finger-like structures of DGKs ␥ (43), (44), and (45) are critical for their translocation to the plasma membrane. On the other hand, DGK was localized at the plasma membrane through its C-terminal portion (amino acid residues 1199 -1268), implying that, despite the same distribution, its target(s) on the plasma membrane is different from those of DGKs ␦, ␥, , and .
Because the C-terminal sequence of DGK responsible for the plasma membrane localization contains neither prominent hydrophobic regions nor lipid binding motifs identified thus far (Fig. 1C), it is not likely that the protein directly interacts with membrane lipids. Thus, DGK is thought to bind to certain target proteins located at the plasma membrane. However, its target protein remains unknown at present. Although a type I DGK (DGK␣) (46,47) has been reported to be phosphorylated by tyrosine kinases in response to growth factors and hormones, there is no report describing tyrosine phosphorylation of type II DGK so far. Therefore, our work is the first report describing tyrosine phosphorylation of type II DGK. Because it was revealed that the tyrosine phosphorylation and the inhibition of DGK are regulated through independent pathways (Figs. 9 and 10), the effects of tyrosine phosphorylation on physiological functions of DGK remain unclear at present. The Motif Scan software predicted that the phosphorylated Tyr 78 interacts with several SH2 domains contained in c-Abl, Nck, Fyn, and Lck. Thus, DGK may interact with these SH2 domain-containing proteins in response to oxidative stress.
DGK␣ (42, 46 -48) and DGK (49) have been reported to be activated in response to growth factors and hormones. Interestingly, in contrast to DGKs ␣ and , DGK activity was inhibited in response to H 2 O 2 (Fig.  10). This result allows us to speculate that DGK keeps DAG concentration low in the plasma membrane in resting cells, whereas the suppression of the DAG level is released by inactivating this isozyme in cells exposed to oxidative stress. In contrast, PA concentration is thought to be reduced in H 2 O 2 -treated cells. In contrast to DGK, conventional and novel PKCs (50,51) and PKD (39,52), which are activated by DAG (3)(4)(5)(6)(7), were reported to be tyrosine-phosphorylated and, consequently, activated in response to H 2 O 2 . Because the inhibition of DGK, which consumes DAG, should result in DAG accumulation in the plasma membrane, the oxidative stress may synergistically activate PKC/PKD through the tyrosine phosphorylation and the accumulation of the activator (DAG) of the protein kinases. Although tyrosine phosphorylation of PKC and PKD is critical for activation of these protein kinases, the phosphorylation of DGK does not seem to be directly involved in its inhibition. Further studies are needed to reveal the inhibition mechanisms of DGK and the functional relationship between DGK and PKC/PKD. Characterization of the oxidative stress-dependent inhibition mechanisms of DGK and identification of downstream effectors of DAG and PA, quantitatively regulated by DGK, are interesting targets of future investigation.
In conclusion, we have identified a novel type II DGK that is markedly different from previously cloned DGKs based on its domain structures and tissue distribution. Moreover, its subcellular localization was clearly distinct from those of other DGKs. Furthermore, DGK was uniquely tyrosine-phosphorylated and inhibited in cells challenged with oxidative stress, and these modifications are dually regulated through Src family kinase-dependent and -independent pathways. These results indicate that, even among the same subfamily members, DGK isoforms are quite heterogeneous, further suggesting the importance of spatiotemporally segregated DAG pools regulated by individual isozymes in DAG-and PA-dependent cellular processes.