Molecular Cloning of a Novel Diacylglycerol Kinase Isozyme with a Pleckstrin Homology Domain and a C-terminal Tail Similar to Those of the EPH Family of Protein-tyrosine Kinases*

A fourth member of the diacylglycerol kinase (DGK) gene family termed DGK (cid:100) was cloned from the human testis cDNA library. The cDNA sequence contains an open reading frame of 3,507 nucleotides encoding a putative DGK protein of 130,006 Da. Interestingly, the new DGK isozyme contains a pleckstrin homology domain found in a number of proteins involved in signal trans- duction. Furthermore, the C-terminal tail of this isozyme is very similar to those of the EPH family of receptor tyrosine kinases. The primary structure of the (cid:100) -isozyme also has two cysteine-rich zinc finger-like structures (C3 region) and the C-terminal C4 region, both of which have been commonly found in the three isozymes previously cloned (DGKs (cid:97) , (cid:98) and (cid:103) ). However, DGK (cid:100) lacks the EF-hand motifs (C2) and contains a long Glu- and Ser-rich insertion (317 residues), which divides the C4 region into two portions. Taken together, these structural features of DGK (cid:100) indicate that this isozyme belongs to a DGK subfamily distinct from that consisting of DGKs (cid:97) , (cid:98) , and (cid:103) . Increased DGK activity without marked preference to arachidonoyl type of diacylglyc- erol was detected in the particulate fraction of COS-7 cells expressing the transfected DGK (cid:100) cDNA. The enzyme activity was independent of phosphatidylserine, which is a common activator for the previously sequenced DGKs. Northern blot analysis showed that the DGK (cid:100) mRNA ( (cid:59) 6.3 kilobases) is most abundant in human skeletal muscle but undetectable in the brain, thy- mus, and retina. This expression pattern is different from those of the previously cloned DGKs. Our results show that the DGK gene family consists of at least two subfamilies consisting of enzymes with distinct struc- tural characteristics and that each cell type probably expresses its own characteristic repertoire of DGKs whose functions may be regulated through different signal transduction pathways.

A fourth member of the diacylglycerol kinase (DGK) gene family termed DGK␦ was cloned from the human testis cDNA library. The cDNA sequence contains an open reading frame of 3,507 nucleotides encoding a putative DGK protein of 130,006 Da. Interestingly, the new DGK isozyme contains a pleckstrin homology domain found in a number of proteins involved in signal transduction. Furthermore, the C-terminal tail of this isozyme is very similar to those of the EPH family of receptor tyrosine kinases. The primary structure of the ␦-isozyme also has two cysteine-rich zinc finger-like structures (C3 region) and the C-terminal C4 region, both of which have been commonly found in the three isozymes previously cloned (DGKs ␣, ␤ and ␥). However, DGK␦ lacks the EF-hand motifs (C2) and contains a long Glu-and Ser-rich insertion (317 residues), which divides the C4 region into two portions. Taken together, these structural features of DGK␦ indicate that this isozyme belongs to a DGK subfamily distinct from that consisting of DGKs ␣, ␤, and ␥. Increased DGK activity without marked preference to arachidonoyl type of diacylglycerol was detected in the particulate fraction of COS-7 cells expressing the transfected DGK␦ cDNA. The enzyme activity was independent of phosphatidylserine, which is a common activator for the previously sequenced DGKs. Northern blot analysis showed that the DGK␦ mRNA (ϳ6.3 kilobases) is most abundant in human skeletal muscle but undetectable in the brain, thymus, and retina. This expression pattern is different from those of the previously cloned DGKs. Our results show that the DGK gene family consists of at least two subfamilies consisting of enzymes with distinct structural characteristics and that each cell type probably expresses its own characteristic repertoire of DGKs whose functions may be regulated through different signal transduction pathways.
Diacylglycerol kinase (DGK, 1 EC 2.7.1.107) phosphorylates diacylglycerol to produce phosphatidic acid (1). DGK is thought to play an important role in the signal transduction linked to phospholipid turnover. The roles of diacylglycerol and phosphatidic acid as lipid second messengers have been attracting much attention. Diacylglycerol is known to be an activator of protein kinase C (2,3), and phosphatidic acid has been reported in a number of studies to modulate a ras GTPase-activating protein (4), phosphatidylinositol (PI)-4phosphate kinase (5) and many other important enzymes. Phosphatidic acid is also known to have mitogenic effects in a variety of cells (6). Thus, DGK is thought to be one of the key enzymes involved in the cellular signal transduction (1,7). For instance, DGK was found to be involved in interleukin-2 production in T-lymphocytes (8) and retinal degeneration of Drosophila melanogaster (9).
Many DGK isozymes have been purified from various animal sources, such as two isozymes from porcine brain and thymus (10,11), three from human platelets (12), two from rat brain (13), and one from bovine testis (14). These isozymes have been described to differ from each other with respect to molecular masses, enzymological properties, activators, and substrate specificity. It is thus likely that these isozymes are operated under distinct regulatory mechanisms. To date, three mammalian DGK genes, DGKs ␣ (15-17), ␤ (18), and ␥ (19,20) have been isolated. The protein products of these genes contain four conserved regions, i.e. the N-terminal region (C1), two sets of Ca 2ϩ -binding EF-hand motifs (C2), two cysteine-rich zinc finger-like structures (C3), and the C-terminal C4 regions. These isozymes have been shown to exhibit different tissue-and cellspecific modes of expression. Namely, DGK␣ is most abundant in thymus (15) and oligodendrocytes of brain (17), and DGK␤ is particularly enriched in rat neuron (18). DGK␥, on the other hand, is highly expressed in the human retina (19) and rat cerebellar Purkinje cells (20). Despite different expression patterns, the structural and enzymatic properties of the three cloned DGKs are very similar to each other, and all of them are characteristically expressed in the central nervous system. Thus, it seems quite likely that DGK gene family may possibly include additional hitherto unknown members with distinct structural features. Identification of novel DGK isozymes and determination of the complete DGK repertoire represent an important and necessary step for elucidating the exact physiological role of DGK.
In the present investigation, we cloned a novel DGK isozyme, DGK␦, that apparently belongs to a subfamily distinct from that consisting of DGKs previously cloned. Interestingly, the novel isozyme contains a pleckstrin homology (PH) domain at the N terminus and a C-terminal tail similar to those of EPH family of protein-tyrosine kinases. Moreover, its expression pattern was markedly different from those of other DGKs.
General manipulation of DNA and RNA was carried out according to the standard procedures (21).
Poly(A) ϩ RNA Preparation-Total RNAs were extracted by the guanidium/cesium chloride method (22) from Jurkat and HepG2 cells. RNAs obtained were enriched in poly(A) ϩ RNA content by oligo(dT)cellulose column chromatography (Pharmacia Biotech Inc. mRNA purification kit). Poly(A) ϩ RNAs from human retina, testis, and HL-60 cells were purchased from Clontech.
Reverse  (17), rat DGK␤ (18), Drosophila DGK1 (23), and Drosophila DGK2 (9). At this stage of investigation, the primary structure of human DGK␥ (19) had not been revealed. These sequences correspond to positions 523-529 and 693-699, respectively, of the amino acid sequence of porcine DGK␣ (15). Sufficient degeneracy was incorporated into the primer sequences to encompass divergence among the various isozymes at the nucleotide level. To facilitate subcloning, XhoI endonuclease sites (underlined) were added to the 5Ј-ends of both primers.
To obtain DGK-related core sequences, reverse transcriptase PCR amplification was carried out using GeneAmp RNA PCR kit (Perkin-Elmer Corp.). In brief, first strand cDNA template was synthesized from 0.1 g of poly(A) ϩ RNA prepared from human testis in 20 l of reverse transcriptase buffer containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 5 mM MgCl 2 , 1 mM each dNTP, 1 unit/l RNase inhibitor, 2.5 units/l Moloney murine leukemia virus reverse transcriptase, and 2.5 M random hexamers. The reaction mixture was incubated for 10 min at 25°C and then for 30 min at 42°C. After heat treatment (97°C for 5 min), the mixture was added with 80 l of Taq polymerase buffer containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.25 mM MgCl 2 , 2.5 units of Taq polymerase, and 1 nmol each of primers P1 and P2. Amplification reactions were carried out for 40 cycles on a GeneAmp PCR System 2400 (Perkin-Elmer Corp.) under the following conditions: 94°C for 1 min, 52°C for 2 min, and 72°C for 2 min, followed by a final elongation step at 72°C for 7 min. Amplified PCR products were subsequently separated by preparative agarose gel electrophoresis. Bands in the position (ϳ500 bp) predicted by reference to the cDNA sequences of DGKs so far cloned were recovered, digested with XhoI, and subcloned into the XhoI site of pBluescript II SKϩ (Stratagene).
Screening of cDNA Libraries-Randomly primed and oligo(dT)primed cDNA libraries were constructed in phage gt10 (Pharmacia TimeSaver cDNA synthesis kit and Stratagene Gigapack® II Gold packaging extract) using poly(A) ϩ RNA prepared from human testis and HepG2 cells. Approximately 8 ϫ 10 5 (testis) and 1 ϫ 10 6 (HepG2 cells) bacteriophage plaques were screened under the stringent condition of hybridization (21). The PCR amplification product (DGKD core) was labeled with [␣-32 P]dCTP by random oligonucleotide primer extension (24) and used as a probe (Fig. 1). Two representative overlapping clones were designated DGKD3 and DGKD10 ( Fig. 1) Rapid Amplification of cDNA Ends (RACE)-RACE, an anchored PCR procedure, was performed to extend the 5Ј-end of the novel DGKrelated cDNA (DGKD10) using 5Ј-AmpliFINDER RACE kit (Clontech). Briefly, cDNA was first synthesized with antisense gene-specific primer, R1 (5Ј-ATTTCGGAGAGGACCCAGCCAACAC-3Ј), from HepG2 poly(A) ϩ RNA (see Fig. 1). A single-stranded anchor oligonucleotide included in the kit was ligated to the 3Ј-end of the cDNA by T4 RNA ligase. Following anchor ligation, the cDNA was used as a template for PCR amplification using a primer (containing an EcoRI site) comple-mentary to the anchor and a second nested gene-specific primer, R2 (5Ј-GCGAATTCATCCCCGCCACAAACCAGAATCCG-3Ј, EcoRI site underlined). The PCR was conducted for 45 cycles (94°C for 1 min, 60°C for 2 min, and 72°C for 2 min) as above. The PCR products were digested with EcoRI and separated by preparative agarose gel electrophoresis. The largest fragment (ϳ1100 bp) was recovered and subcloned into the EcoRI site of pBluescript II. The three longest cDNA clones (DKD11-13) were selected for sequence analysis.
DNA Sequencing-Purified plasmid DNAs from positive clones were sequenced by the dideoxy chain termination method (25) using a Sequenase 2.0 kit (U. S. Biochemical Corp.). Sequences were analyzed using the GeneWorks sequence analysis software (IntelliGenetics Inc.). Deduced amino acid sequences were analyzed and aligned with related sequences included in the DNA Data Bank of Japan using the program FASTA (26).
RNA Blot Hybridization Analysis-Poly(A) ϩ RNAs (2 g each) of HepG2 cells, Jurkat cells, HL-60 cells, and retina were separated on 1% agarose gel and transferred to a nylon membrane (Nytran, Schleicher and Schuell) as described previously (15). The other poly(A) ϩ RNAs (2 g each) were analyzed using multiple tissue Northern (MTN) blot I and II (Clontech). The hybridization probe used was the 402-bp XhoI-NcoI fragment (nucleotide 2,606 -3,007) of DGKD3 (see Fig. 1). The nylon membranes were hybridized and washed as described previously (15). Autoradiography was performed at Ϫ80°C for 60 h with intensifying screens.
COS-7 Cell Transfection and Determination of DGK Activity-To construct the cDNA containing the full length of an open reading frame, DGKD3 and DGKD10 were joined at the KpnI site, and the resultant construct and DGKD11 were combined at the BspEI site. (see Fig. 1). In order to maximally fulfill Kozak consensus translation initiation sequence (27) and also to create EcoRI sites in both 5Ј-and 3Ј-flanking regions, several mutations were introduced into the composite cDNA (DGKD103) by in vitro mutagenesis system (Amersham Corp.) employing complementary oligonucleotides with nucleotide substitutions as indicated in Fig. 1. The nucleotide substitutions were confirmed by sequencing. The full coding region (DGKD104) was subcloned into the EcoRI site of the simian virus 40-based expression vector pSRE (28), which was derived from the original pcDL-SR␣-296 vector (29). Approximately 10 g each of the resultant construct, pSRE-DGK␦, and pSRE vector alone were transfected into COS-7 cells (100-mm dish) using DEAE-dextran (30) as described previously (28). After 3 days, the cells were harvested and lysed by sonication in the lysis buffer (0.5 ml/100 mm dish) containing 20 mM Tris-HCl (pH 7.4), 0.25 M sucrose, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 20 g/ml each of leupeptin, pepstatin, aprotinin, and soybean trypsin inhibitor, and 1 mM phenylmethylsulfonyl fluoride. After a low speed centrifugation (550 ϫ g for 10 min), the soluble and particulate fractions were separated by a centrifugation at 100,000 ϫ g for 30 min. The particulate fraction was suspended in the same volume of the lysis buffer. The aliquots of the two fractions were subjected to DGK activity assay and immunoblot analysis.
The octyl glucoside mixed micellar assay of DGK activity was done as described previously (31). 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 , 1 mM EGTA, 1.5 mM diacylglycerol, and 1 mM [␥-32 P]ATP (100,000 cpm/nmol). The reaction was initiated by adding the particulate or soluble fraction (10 g of protein) from COS-7 cells and continued for 5 min at 30°C. Lipids were extracted from the mixture, and phosphatidic acid separated by thin layer chromatography (32) was scraped and counted by a liquid scintillation spectrophotometer.
Preparation of Anti-DGK␦ Antibodies and Immunoblot Analysis-The cDNA fragment (644 bp) encoding the C-terminal peptide (amino acids 1079 -1169) of DGK␦ was obtained by digestion of DGKD3 with DraI and SmaI (Fig. 1). The fragment was subcloned into the T7 RNA polymerase-dependent pET3c vector (33). This construct was introduced into Escherichia coli strain BL21(DE3)/pLysS and induced in the presence of 1 mM isopropyl-D-thiogalactopyranoside as described previously (33). The expressed fusion protein recovered in insoluble fraction was solubilized with 8 M urea and dialyzed against 30 mM Tris-HCl (pH 7.4) containing 30 mM NaCl and 1 mM dithiothreitol. The insoluble fraction was removed by centrifugation, and the resultant supernatant was used as an antigen. Rabbits were immunized by subcutaneous multiple injections of 150 g of the fusion protein emulsified with an equal volume of Freund's complete adjuvant (Wako Pure Chemicals). Injections were repeated at 2-week intervals, and the serum was obtained after the fifth injection. The IgG was purified by a column chromatography on Protein A-Sepharose (Pharmacia Biotech Inc.).
The soluble and particulate fractions of COS-7 cells (10 g of protein each) were separated on an SDS, 7.5% polyacrylamide gel (34). The proteins were transferred to a nitrocellulose membrane (Schleicher and Schuell) and blocked for 1 h in phosphate-buffered saline (pH 7.4) containing 0.1% (v/v) Tween 20 and 20% (w/v) skim milk. The membrane was incubated with the anti-DGK␦ antibody (0.7 g/ml) in the same buffer for 1 h. The immunoreactive bands were visualized using peroxidase-conjugated, anti-rabbit IgG antibody (Jackson ImmunoResearch Laboratories) and enhanced chemiluminescence (ECL kit, Amersham Corp.).

RESULTS
Isolation and Identification of DGK␦ cDNA-In order to identify novel genes potentially related to the DGK family, a PCR-based amplification strategy was performed. We designed two degenerate oligonucleotide primers, P1 and P2 (Figs. 1 and 2), that corresponded to the amino acid sequences, Tyr-Phe-Ser-(Phe/Ile/Val)-Gly-Val-Asp and Pro-Met-Gln-(Val/Ile)-Asp-(Gly/Val)-Glu (residues 523-529 and 693-699 in porcine DGK␣, respectively; Ref. 15). The amino acid sequences are highly conserved among all mammalian DGK isozymes and Drosophila homologs cloned so far (9,(15)(16)(17)(18)23). The reverse transcriptase PCR amplification of poly(A) ϩ RNA derived from human testis resulted in several products having the expected size of ϳ500 bp (not shown). The amplification products were shortened by digestion with XhoI (see Fig. 1), which was used to produce cohesive-ends for subcloning, and the resultant ϳ350-bp fragments were subcloned into pBluescript. Sequence analysis identified one 363-bp clone encoding a peptide of 121 amino acids. The sequence of this peptide was 43 and 45% identical to the corresponding sequences in porcine DGK␣ and rat DGK␤, respectively. Therefore, the PCR product was judged to correspond to a novel DGK gene. Since the new DGK clone is the fourth to be reported to our knowledge, it is designated DGK␦.
This cDNA clone (the DGKD core sequence, Fig. 1) was used to screen ϳ800,000 clones of the testis library and ϳ1,000,000 clones of the HepG2 cell library. Seven positive clones (DGKD1-7) and three positive clones (DGKD8 -10) were isolated from the testis and the HepG2 libraries, respectively (Fig.  1). Only the most 5Ј-and 3Ј-stretched clones, DGKD10 and -3, respectively, were indicated in Fig. 1. However, sequence analysis revealed that these clones lacked the N-terminal part of the open reading frame including the initiation methionine codon. Although we further screened the same libraries with DGKD10 cDNA as a probe, we could not obtain clones covering the 5Ј-region. We, therefore, carried out the RACE-anchored PCR procedure using poly(A) ϩ RNA isolated from HepG2 cells as a template. The 5Ј-anchored PCR employing primers R1 and R2 (Figs. 1 and 2) gave rise to three clones (DGKD11-13) with the same length, all of which were found to cover the complete 5Ј-region.
The composite DGK␦ cDNA sequence obtained from clones DGKD3, DGKD10, and DGKD11 (DGKD103, 6214 bp, Fig. 1) has an open reading frame starting from nucleotide 81 to nucleotide 3590 (Fig. 2). The initiation ATG of the open reading frame was identified as the first ATG sequence following an in-frame termination codon at nucleotide 66 to 68. The open reading frame is flanked by 80-(5Ј-) and 2624-bp (3Ј-) noncoding nucleotide sequences. A polyadenylation signal (AATAAA) is found 22 nucleotides upstream of the poly(A) sequence at the 3Ј-end of the cDNA.
Analysis of the Deduced Amino Acid Sequence for DGK␦-The DGK␦ cDNA encodes a putative protein of 1169 amino acid residues having a calculated M r of 130,006 (Fig. 2). The primary structure displays several characteristic features shared by other DGK isozymes cloned so far (Fig. 3A). DGK␦ contains a tandem repeat of cysteine-rich sequence (residues His 120 -Cys 169 and His 192 -Cys 242 ), 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 is conserved in all of the different DGK isozymes (Fig. 3, A and B). This motif was also found in protein kinase C (35), n-chimaerin (36), and a vav oncogene product (37). The sequence identities of zinc fingers of DGK␦ were 33, 33, and 32%, respectively, when compared with those of porcine DGK␣ (15), rat DGK␤ (18), and human DGK␥ (19). Although the relative locations of critical residues like His and Cys are well conserved, the overall identity of amino acid sequences of the two zinc fingers is not particularly high among different DGKs (Fig. 3B).
One of the striking structural features of DGK␦ is that there is a long insertion (317 residues) that divides the C4 region into two portions, designated C4-a and C4-b subregions (Fig. 3A). A similar division of the C4 region was also noted previously for Drosophila DGK1. However, both sequences of the C4 subregions are very similar to the corresponding parts of the C4 region of other DGKs, exhibiting 51-55% (C4-a) and 38 -44% (C4-b) identities, respectively (Fig. 3, C and D). Since the C4 region is well conserved in all mammalian DGK isozymes and Drosophila homologs, it seems to be the catalytic domain. Indeed, we recently found that a DGK␣ mutant containing only the C4 region showed phospholipid-dependent DGK activity. 2 The insert separating the two C4 subregions displayed no significant sequence similarity to that of Drosophila DGK1, and the effect of this insert on the catalytic activity remains unknown. The insertion contains a Glu-and Ser-rich sequence (residues 416 -538). In this case, 15 (12.2%) and 22 (17.9%) out of 123 residues are Glu and Ser residues, respectively. The additional characteristic of the sequence of DGK␦ in comparison with other DGKs is that DGK␦ has neither the N-terminal C1 region nor the EF-hand motifs (C2) (Figs. 2 and 3A).
Interestingly, this novel DGK isozyme contains the PH domain (38 -40) (residues 10 -100) at the N terminus (Figs. 2 and  3E). The PH domain, comprising about 100 amino acids, was originally detected as an internal repeat in pleckstrin, a 47-kDa protein that is the major substrate of protein kinase C in platelets (41). At present, the PH domains have been found in a number of proteins involved in signal transduction (38 -40) such as a ras-GTPase activating protein, AKT/RAC protein kinase, ␥ and ␦ types of PI-specific phospholipase C, a vav oncogene product and protein kinase C. All of the most highly conserved residues (Gly 14 , Leu 16 , Leu 33 , Leu 54 , Phe 71 , and Trp 93 in DGK␦) of the PH domains (38,40) are completely conserved also in DGK␦ (Fig. 3E). The extents of identity and similarity between the PH domain of DGK␦ and the corresponding regions of several proteins such as pleckstrin, AKT1, protein kinase C and phospholipase Cs, are 17-31% and 35-52%, respectively (Fig. 3E). The C-terminal tail (residues 1105-1169) of DGK␦ is unexpectedly similar to those of the EPH family of receptor tyrosine kinases (42)(43)(44)(45)(46)(47)(48) (Figs. 2 and 3F). The C terminus of the EPH family, which follows the catalytic domain, is thought to be a regulatory domain. The EPH family (the fourth family of receptor protein-tyrosine kinases) is distinct from the epidermal growth factor receptor, insulin receptor, and platelet-derived growth factor receptor families (42)(43)(44)(45)(46)(47)(48). Most of the invariant residues (Trp 1108 , Leu 1109 , Phe 1121 , Asp 1138 , Gly 1143 , and Ile 1153 , except for Met 1150 , in DGK␦) found in all EPH members cloned so far (48) are conserved also in DGK␦. The extents of identity and similarity between DGK␦ and the receptor family in the C-terminal tail are 26 -32% and 48 -57%, respectively, as shown in Fig. 3F.
Transient Expression of DGK␦ in COS-7 Cells-In order to confirm that the translational product of the newly isolated cDNA indeed possesses DGK activity, the cDNA (DGKD104) was subcloned into the pSRE expression vector and transfected into COS-7 cells. In order to obtain a maximum expression level, most of the 5Ј-and 3Ј-untranslation sequences were removed (Fig. 1). Moreover, several nucleotide substitutions were introduced into the DGK␦ cDNA to maximally fulfill Kozak consensus translation initiation sequence (27). The pSRE vector alone was also transfected into COS-7 cells as a control. As shown in Fig. 4A, the expression of the DGK␦ enzyme protein with the expected molecular size (130 kDa) could be confirmed by immunoblot analysis with antibodies raised against the C-terminal portion of the enzyme, whereas no band was detected in the control cells. Approximately 80% of the enzyme protein was found to be recovered in the particulate fraction. Two bands of lesser M r may be proteolytic products of the parent band. As shown in Fig. 4B, DGK activity measured with the particulate fraction of COS-7 cells expressing the DGK␦ cDNA was approximately 2.5-fold greater than that measured in cells transfected with the vector alone, whereas there was no corresponding difference in DGK activity in the soluble fraction. Since most of the membrane-associated DGK activity was released when treated with 1 M NaCl, DGK␦ appeared to be loosely bound to the membranes as has been noted for rat DGK␤ (18) and human DGK␥ (19).
Northern Blot Analysis-The expression pattern of the corresponding transcripts was analyzed in a variety of human tissues and cell lines (Fig. 6). The size of the DGK␦ mRNA is ϳ6.3 kb, which coincides with the composite length of the cDNAs (6214 bp). A relatively high expression level was found in skeletal muscle. The mRNA could be detected to a lesser extent in testis, colon, peripheral blood leukocytes, and HepG2 cells. On the other hand, the brain, retina, thymus, and Jurkat cells displayed extremely low or undetectable levels of the DGK␦ mRNA. This expression profile is quite different from those described for DGKs ␣, ␤ and ␥, which are most abundant in thymus, brain, and retina, respectively (15,18,19). DISCUSSION Our work adds a new member, DGK␦, to the growing list of mammalian DGK genes. Although rat DGK-III cDNA has recently been cloned (20), this isozyme shared 88% identical amino acid sequence with human DGK␥ (19). We thus believe that the DGK-III should be a rat counterpart of human DGK␥. It is of particular interest to note that DGK␦ contains the PH domain at the N terminus and the C-terminal tail very similar to those of the protein-tyrosine kinases of EPH receptor family (Figs. 2 and 3A). Such structural features enabled us to classify FIG. 3. Sequence comparison of DGK␦ with other DGKs, the members of the EPH receptor family and several proteins containing the PH domain. A, the schematic structure of DGK␦ is shown with reference to previously sequenced porcine DGK␣ (15), rat DGK␤ (18), human DGK␥ (19), Drosophila DGK1 (23), and Drosophila DGK2 (9). The mammalian DGK gene family is proposed to consist of at least two enzyme subfamilies tentatively designated Type I and Type II. B, amino acid alignment of the cysteine-rich sequences of DGK␦ and other DGK isozymes. Asterisks indicate the conserved cysteine and histidine residues. Since the motifs of the cysteine-rich sequences in Drosophila DGK2 (9) are considerably different, their alignments are not shown. C, alignment of the C4-a of DGK␦ and C4 regions of other DGK isozymes and Drosophila homologs. D, alignment of the C4-b of DGK␦ and C4 regions of other DGK isozymes and Drosophila homologs. E, alignment of the PH domains of DGK␦ and several proteins containing the PH domain(s) (38 -40). The consensus sequence (38 -40), including alternatives, is shown below; the most highly conserved residues are highlighted. F, alignment of the C-terminal tails of DGK␦ and the members of the tyrosine kinase EPH receptor family (42)(43)(44)(45)(46)(47)(48). Highlighted letters in the consensus sequence indicate invariant residues found in all members of the EPH family. B-F, the residue number of the first amino acid is given in parentheses for each sequence. Identical amino acids are indicated in reverse type. Similar amino acids are hatched. The groups of similar amino acids were defined as follows: small R groups with near neutral polarity (G, A, P, S, and T); acidic and uncharged polar R groups (D, E, N, and Q); basic polar R groups (H, R, and K); nonpolar chain R groups (M, I, L, and V); and aromatic nonpolar or uncharged R groups (F, W, and Y). The extent of identity (I) and similarity (S) are shown at the end of each column. Dashes indicate gaps inserted to maximize alignment. DGK␦ into the new DGK subfamily (type II), which is distinct with respect to domain structures from that (type I) consisting of DGKs ␣, ␤, and ␥ previously cloned. It should be noted here that DGKs belonging to the type I subfamily possess basically the same domain structures despite their markedly different expression pattern. Moreover, DGK␦ is highly expressed in the skeletal muscle with a very low expression in the brain (Fig. 6). This expression pattern of DGK␦ is quite different from those of the type I DGK isozymes, which are typically expressed in the central nervous system. It is interesting to see if the type II DGK subfamily also consists of multiple isozymes with related structural characteristics. We may hypothesize that the DGK gene family probably consists of several different subfamilies, as has been known for protein kinase C (3,35) and PI-specific phospholipase C (50). The divergence of DGK isozymes may reflect their physiological importance and the needs to respond to a variety of signaling pathways operating under distinct regulatory mechanisms.
The distribution of the PH domain has so far been limited to cytoskeletal proteins and other proteins involved in cellular signal transductions, many of which are associated with the cell surface and intracellular membranes (40). This domain is thought to serve as the site of protein-protein interaction, which is crucial to the function of these proteins (40). The presence of the PH domain in DGK␦ suggests that this enzyme would share similar regulatory mechanisms with the signaling and cytoskeletal proteins. Lefkowitz and co-workers (51) have recently found that the PH domains in a number of proteins such as ␤-adrenergic receptor kinase and PI-specific phospholipase C␥ can bind the ␤␥ complexes of heterotrimeric G proteins. On the other hand, the PH domains of pleckstrin (49) and the AKT protein kinase family (52) have been found to bind PI-4,5-bisphosphate and PI-3-phosphate, respectively. Although it is possible that the activity of DGK␦ may be regulated by binding to regulatory proteins or phospholipids, we could not detect at present the effects of at least PI-4,5-bisphosphate on the DGK␦ activity.
The EPH family with a dozen members is currently known to be the largest subfamily of receptor protein-tyrosine kinases (53). Oncogenic activities of this receptor family have also been New Member of DGK Family described previously (54). Recently, one of their ligands has been found to be B61, the protein product of an early response gene induced by tumor necrosis factor-␣ (55). It is also found that PI-3-kinase serves as a downstream target of this receptor family (56). Our work is the first to report that the homology region to the C-terminal tail of the EPH protein-tyrosine kinases exists in the C terminus of an apparently unrelated enzyme protein, DGK. This finding led us to speculate that this domain may be distributed in a wider range of proteins than has been previously anticipated. We thus tentatively propose that the conserved C-tail region is designated an EPH C-terminal tail homology domain. Although the function of the EPH tyrosine kinases has been suggested to be regulated by phosphorylation of the conserved tyrosine residue of the C-terminal tail (tyrosine 1117 in the case of DGK␦) (44), the exact function(s) of this domain remains to be explored. We are currently investigating the possibility that the action of DGK␦ is regulated by tyrosine phosphorylation.
DGK␦ has a long insertion composed of 317 residues between the C4-a and C4-b subregions. This insertion occurs only in DGK␦ among the sequenced mammalian DGKs, but a similar insertion was described previously for Drosophila DGK1. However, there was no significant sequence homology between the insertions of DGK␦ and the Drosophila homolog. Moreover, the insertion did not have significant similarities to other sequences deposited in protein data base. We noted that the insertion of DGK␦ contains Glu-and Ser-rich sequences. It is interesting to note that casein kinases I, II and mammary gland casein kinase phosphorylate Ser residues in the sequences, Glu-X-X-Ser, Ser-X-X-Glu, and Ser-X-Glu, respectively (57). DGK␦ contains several of these sequences (Fig. 2). It is therefore possible that the insertion of DGK␦ serves as a phosphorylation site(s) of such protein kinases. The sequences of C4 region were most highly conserved in all DGK isozymes and Drosophila homologs. This region is therefore quite likely to serve as the catalytic domain common to all DGKs. There are several sequences perfectly conserved in all DGKs (Figs. 3, C and D), and these are probably important for the catalytic action including DG and ATP bindings. Although our earlier study noted in the C4 region of DGK␣ a putative ATP-binding site with a motif of Gly-X-Gly-X 2 -Gly-. . . -Lys (15), this motif is not conserved in other DGKs including DGK␦. A comparison of the amino acid sequence of the C4 region with other nucleotidebinding proteins (58) did not reveal candidate ATP-binding site(s) of DGK. The present work showed that the zinc finger structures are conserved also in DGK␦, the domain structure of which is markedly deviated from those of the type I isozymes.
The zinc fingers together with the highly conserved C4 region thus appear to be essential for the action of different DGKs. It was previously shown that the zinc fingers of human DGK␣ is incapable of phorbol ester binding (59). Furthermore, we found that a DGK␣ mutant lacking the zinc fingers exhibited a phospholipid-dependent DGK activity. 2 These results indicate that the DGK zinc fingers are not the sites of phorbol ester or DG binding. The function of this motif is thus entirely unknown for DGKs. Recently the zinc fingers of protein kinase C⑀ and Raf-1 kinase have been shown to be a subcellular localization domain Poly(A) ϩ RNA (2 g each) was analyzed as described under "Experimental Procedures." Except for poly(A) ϩ RNAs from retina, HL-60, Jurkat, and HepG2 cells, commercially available multiple tissue Northern Blot I and II (Clontech) were employed. The filters were hybridized with the XhoI-NcoI fragment of DGKD3 labeled with 32 P and exposed for 60 h with intensifying screens. Almost equal amount of RNA had been loaded, as judged from hybridization of the same filter with a human ␤-actin cDNA probe (data not shown). (60) and an interaction site with p21 ras (61,62), respectively. It is therefore likely that some novel and unexpected functions would be ascribed to the DGK zinc fingers in future studies.
The DGK␦ cDNA gave only 2.5-fold increase of DGK activity in the transfected COS-7 cells (Fig. 4B). This contrasts with the high enzyme activities of other DGKs transfected under the same experimental conditions (19). This low level of enzyme activity hindered us from defining detailed enzymological properties of DGK␦. This may be partly due to the low expression of the enzyme protein or to proteolytic degradation of the expressed enzyme as noted in immunoblotting (Fig. 4A). In view of the apparent lack of phosphatidylserine dependence of the DGK␦ activity (Fig. 5), the enzyme assay conditions developed for the type I enzymes may not be suitable for this isozyme. It is also possible that DGK␦ requires unknown activator(s) or its phosphorylation on serine or tyrosine residues to exert full catalytic activity. In this respect, we could not deny the possibility that DGK␦ might prefer substrates other than DG. However, we could not detect the enzyme activity when the COS cell extracts were tested with PI, 1-or 2-monoacylglycerol, ceramide, and sphingosine as substrates. The enzymological studies using purified DGK␦ are needed to define the regulatory mechanisms of this isozyme.
Despite interesting domain structures, the function of DGKs in cellular phospholipid metabolism is largely unknown. The present work further points to the potential importance of DGKs as signaling molecules by disclosing the existence of a novel isozyme possessing the PH and EPH C-terminal tail homology domains in addition to the previously identified functional regions. The expression pattern of DGK␦, when taken together with those of other DGKs, suggests that each DGK species may play a specialized role in differentiated cells and that each cell type probably expresses its own characteristic repertoire of DGK isozymes. Elucidation at the molecular level of the regulatory mechanisms of DGKs would reveal the physiological significance of these proteins with unique structures.