Cloning and characterization of a glucocorticoid-induced diacylglycerol kinase.

Diacylglycerol kinase (DGK) plays a key role in cellular processes by regulating the intracellular concentration of the second messenger diacylglycerol. We screened a hamster DDT1 smooth muscle cell library and isolated a unique, glucocorticoid-inducible cDNA with substantial homology to known DGKs. DGK activity was increased in lysates of insect cells infected with recombinant baculovirus containing this cDNA. Antibodies raised against expressed sequences recognized a glucocorticoid-inducible 130-140-kDa protein on immunoblots of DDT1 cell lysates. Thus, this sequence appears to be a new member of the DGK family that we refer to as DGKη. Homology to other DGKs was apparent in domains that are thought to be important for DGK function including the cysteine-rich motifs and potential catalytic domains. DGKη shares substantial homology with DGKδ including the N-terminal pleckstrin homology domain. The tissue distribution of DGKη message (determined by ribonuclease protection assays) and protein (determined by immunoblots) was broader than reported for other DGKs, indicating that DGKη may play a more general role in regulating cellular DG levels than other DGKs. Heterogeneity among DGK family members indicates that individual DGKs may have unique functions.

Diacylglycerol kinase (DGK) plays a key role in cellular processes by regulating the intracellular concentration of the second messenger diacylglycerol. We screened a hamster DDT1 smooth muscle cell library and isolated a unique, glucocorticoid-inducible cDNA with substantial homology to known DGKs. DGK activity was increased in lysates of insect cells infected with recombinant baculovirus containing this cDNA. Antibodies raised against expressed sequences recognized a glucocorticoid-inducible 130 -140-kDa protein on immunoblots of DDT1 cell lysates. Thus, this sequence appears to be a new member of the DGK family that we refer to as DGK. Homology to other DGKs was apparent in domains that are thought to be important for DGK function including the cysteine-rich motifs and potential catalytic domains. DGK shares substantial homology with DGK␦ including the N-terminal pleckstrin homology domain. The tissue distribution of DGK message (determined by ribonuclease protection assays) and protein (determined by immunoblots) was broader than reported for other DGKs, indicating that DGK may play a more general role in regulating cellular DG levels than other DGKs. Heterogeneity among DGK family members indicates that individual DGKs may have unique functions.
Diacylglycerol (DG) 1 is a neutral lipid that regulates a variety of intracellular processes. One important function is as a second messenger that regulates activity of protein kinases C (PKCs), a multienzyme family of serine/threonine kinases important for cell growth and differentiation (1,2). DG has also been shown to regulate cytoskeletal structures, e.g. DG directly stimulates actin nucleation and, hence, polymerization (3). Furthermore, DG is a key intermediate in lipid metabolic pathways for phospholipid, prostaglandin, and leukotriene synthesis. Thus, regulation of intracellular DG levels is critical for normal cellular function. One of the major routes of DG metabolism is via DG kinase (DGK) that phosphorylates DG to generate phosphatidic acid (4). Recent evidence indicates that phosphatidic acid and its metabolite lysophosphatidic acid may be second messengers as well (5)(6)(7)(8). Thus, DGK has two important functions, first, to limit cellular levels of DG and second, to generate additional second messengers.
Several DGK activities have been purified or partially purified (4,9), and a number of DGK and DGK-related clones have been isolated (10 -18). DGKs appear to differ with respect to their molecular weight, cofactor regulation, tissue distribution, and substrate specificity; however, direct comparisons of DGK activities have been hampered by the lack of a standard assay technique and availability of specific antibodies. Alignment of DGK sequences has been used to identify motifs that may be important for DGK function and regulation. These include two calcium-binding EF hands that confer calcium-dependent activity (19), two cysteine-rich motifs (CRMs) that are similar to those found in PKC and Raf kinase (2), and a putative catalytic domain in the C-terminal half of the protein. The CRMs and the putative catalytic domain sequences are conserved in all DGKs including those from Drosophila (14) and Caenorhabditis elegans (13). Genetic analysis of Drosophila rdgA mutants has linked mutations in the DGK2 gene to retinal degeneration (20) indicating that this retinal specific form of DGK is essential for normal retinal function. Specific functions of other DGKs have not yet been identified.
In the course of studying PKC␣ regulation in hamster DDT1 smooth muscle cells, we identified a high molecular weight, glucocorticoid-inducible protein (referred to as DGK) that cross-reacted with a PKC␣-specific monoclonal antibody. Because of the possibility that this could have been a unique form of PKC, we immunoscreened a glucocorticoid-induced DDT1 cell cDNA library with the PKC␣ antibody. Several clones were isolated, none of which had appreciable homology to PKCs or other protein kinases. However, they were homologous to DGK sequences within CRMs and the putative catalytic domains. Expression of a partial cDNA confirmed that the sequence encodes a DGK activity. Our results demonstrate that DGK is a unique form of DGK with broad tissue distribution. Differences from previously reported DGKs suggest that DGK may have a specific role in cellular DG metabolism.

EXPERIMENTAL PROCEDURES
Materials-Male Syrian hamsters weighing 100 -150 g were purchased from Charles River Breeding Lab Inc., Wilmington, MA. Fetal bovine serum, Dulbecco's modified Eagle's medium, and Ham's F12 were from Life Technologies, Inc. Insulin and transferrin were from Sigma. Vitrogen 100 was from Collagen Corp., Palo Alto, CA. Escherichia coli diacylglycerol kinase was from Lipidex, Inc., Westfield, NJ. 1,2,-Dioctanoyl-sn-glycerol (diolein) was from Avanti Polar Lipids, Inc., Alabaster, AL. [␥-32 P]ATP (3000 Ci/mmol) and [␣-32 P]UTP (3000 Ci/ mmol) were from DuPont NEN. All restriction enzymes, Taq polymerase, Prime-a-Gene Kit, and alkaline phosphatase-conjugated goat antirabbit or anti-mouse immunoglobulins were from Promega, Madison, WI. The MAXIscript T7/T3 in vitro transcription and RPA II Kits were from Ambion, Austin, TX. The TA Cloning Kit was from Invitrogen Corp., San Diego, CA. ECL Western blot reagents were from Amersham Corp. Affi-Gel 10 and protein assay reagents were from Bio-Rad. Nitro-PLus was from Micron Separations Inc., Westborough, MA. Recombi-* This work was supported by National Institutes of Health Grants CA53841 and CA65874 (to S. J.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U59429.
cDNA Cloning and Sequencing-A Lambda Zap II expression library (kindly provided by Dr. Steve Harris, University of Texas, San Antonio) prepared from glucocorticoid-induced DDT1 cells was immunoscreened with a PKC␣-specific monoclonal antibody (M6) (21) according to standard protocols. Three positive clones were isolated out of two million plaques screened. Bluescript ® SK phagemid (pSK) containing the positive inserts were excised from lambda phage and used for sequencing. Additional clones were isolated by rescreening the library with 32 Prandom-labeled cDNA probes prepared from pSK4 (see Fig. 2) according to standard protocols. To obtain additional 5Ј sequences, a second DDT1 library (kindly provided by Dr. Jim Norris, Medical University of South Carolina, Charleston, SC) was screened with a cDNA probe prepared from a 5Ј fragment of pSKA21, (pSKA21a) and two identical overlapping clones were isolated. mRNA from glucocorticoid-treated DDT1 cells was used as a template in 5Ј RACE protocols to obtain additional 5Ј sequence according to the method of Frohman et al. (22). The sequences of the three nested gene-specific antisense primers designed to the 5Ј end of the target cDNA (pSKA21) were as follows: GP16 (5Ј-AGGTCGCTCTACAGAAAC-3Ј), GP17 (5Ј-CAACAGAGGGCTGACAA-3Ј), and GP18 (5Ј-TACACACCGCCTGCAAAGAT-3Ј). The final product was gel-purified, blunt-ended, and subcloned into pSK for further sequencing.
Both 5Ј RACE and inverse PCR strategies were used to obtain the remaining 5Ј end. The 5Ј Amplifinder RACE kit (Clontech Laboratories, Inc., Palo Alto, CA) was used with GP45 (5Ј-TCATAGGGTTCTCT-GCTCTGTACTGAC-3Ј) for reverse transcription and the nested primer GP23 (5Ј-GCTGATCCAGTCTTCCATCT-3Ј) and an anchor primer for amplification. For inverse PCR, cDNA was reverse-transcribed from GP43 (5Ј-CTCTCTGCTGCACACATTACAGAATGTGGG-3Ј) and ligated intramolecularly to form a closed circular double-stranded DNA molecule. This was amplified by PCR using a sense primer GP26 (5Ј-TGTCGCTGAAGCAAGCAC-3Ј) and an antisense primer GP62 (5Ј-GGGAAACCGCGCTGGTCCCAACGG-3Ј) to obtain unknown 5Ј sequence (23). The PCR products from 5Ј Amplifinder and inverse PCR were cloned directly into pCRII with the TA Cloning kit.
Inserts of positive clones were sequenced on both strands using a combination of manual and automated dideoxy-chain termination reactions. A USB Sequenase Version 2.0 kit was used for the manual sequencing. Automated sequencing was done using an Applied Biosystems 370A Automated DNA Sequencer (Applied Biosystems, San Francisco, CA) using either dye-primer or dye-terminator protocols. Doublestranded DNA for sequencing was prepared by Promega Magic Miniprep columns. Sequencing data were analyzed using GeneWorks (IntelliGenetics, Inc., Mountain View, CA) and DNASIS (Hitachi Software Engineering America, Ltd., Brisbane, CA).
Source of Primary Antibodies-The M6 monoclonal antibody used in this study was prepared to purified rabbit PKC␣ and recognizes the catalytic domain of PKC␣ (21). The antipeptide antibody AJ10 was prepared to a synthetic peptide consisting of the 15 C-terminal amino acids of the DGK⑀ predicted amino acid sequence. AJ10 antisera was purified against the peptide coupled to Affi-Gel 10 (Bio-Rad). The polyclonal antibodies AJ12 and AJ21 were raised in rabbits against purified fusion proteins produced in bacteria expressing the pSK4 or the pSKA21b insert, respectively. The pSK4 insert was expressed as a glutathione S-transferase fusion protein from a pGEX3X vector (Pharmacia Biotech Inc.) and purified by electroelution. Subclone A21b was expressed as a histidine-tagged fusion protein from pQE31 (Qiagen, Inc., Chatsworth, CA) and purified under nondenaturing conditions with a nickel affinity column according to the Qiagen protocol. Both antisera were affinity purified on Affi-Gel-10 columns containing the cognate fusion protein.
Western Blots-Proteins were separated by SDS-polyacrylamide gel electrophoresis (7.5% unless indicated otherwise) (24). The proteins were transferred to NitroPlus, and blots were immunostained as described previously (25) and developed with either alkaline phosphatase substrates or enhanced chemiluminescence reagents.
Baculovirus Expression-A partial DGK construct (pSKDGK-p) was prepared in pSK by ligating together pSK5ЈR1 and pSKB24 using convenient restriction sites (see Fig. 2 for a schematic map of DGK clones). The total size of the insert was 4.112 kb, and the predicted protein size was 116 kDa corresponding to an open reading frame of 3.1 kb. This insert was cloned into the pBlueBacHisC transfer vector (Invitrogen Corp., San Diego, CA) for preparation of recombinant baculovirus (BvDGK-p). The full-length DGK baculovirus (BvDGK) was constructed by addition of a pCRII5ЈR4 fragment to the 5Ј end of pSKK⑀-p. Subsequently, the BamHI/HindIII fragment containing the full-length DGK coding region was cloned into pBlueBacHisA for preparation of recombinant baculovirus.  (26).
DGK Assay-The DGK assay was carried out as described previously with some minor changes (27). Samples were collected in sucrose/ATP buffer described above. The assay was performed in a 50-l reaction volume containing 100 mM Tris-Cl, pH 7.4, 1 mM sodium deoxycholate, 0.5 mM dithiothreitol, 1 mM diolein, 1.6 mM [␥- 32  Tissue Distribution Analysis-Male Syrian hamsters (100 -150 g) were anesthetized with 90 mg/kg pentobarbital by injection into the lower abdominal cavity. For immunoblotting, the appropriate tissues were removed, rinsed in ice-cold homogenization buffer (20 mM Tris-Cl, pH 7.4, 5 mM EDTA, 2 mM dithiothreitol, 0.25 M sucrose), weighed, macerated, and placed in a 15-ml conical tube. Homogenization buffer (3 volumes/g tissue) containing 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml aprotinin was added before homogenizing in a Dounce homogenizer and sonicating 5 ϫ 10 s. Protein values were determined by the method of Bradford (26). Samples (100 g protein/lane) were separated on 7.5% SDS-polyacrylamide gels and blotted to NitroPlus.
Total RNA for ribonuclease protection analysis (RPA) was prepared from male Syrian hamster tissues. Tissues were placed in RNA extraction buffer (REB, 4 M guanidine isothiocyanate, 25 mM sodium citrate, 0.1 M ␤-mercaptoethanol) at 10 ml/g tissue and homogenized for 30 s in a Brinkman Polytron. Total RNA was prepared by cesium chloride centrifugation using standard procedures. The total RNA was resuspended in ribonuclease-free water to a concentration of 2 g/l. The ribonuclease protection assay is described below.
Ribonuclease Protection Assay-An antisense RNA probe was prepared from a linearized pSK vector containing a 197-bp HincII/EcoRI fragment (nucleotides 3255-3451) of partial DGK⑀ cDNA with T3 RNA polymerase and [ 32 P]dUTP according to the Ambion MAXIscript T7/T3 in vitro Transcription Kit protocol (Ambion, Inc., Austin, TX). Total RNA (20 g/sample) was processed according to standard protocol or with the Ambion RPA II Kit. Samples were separated on a 5% sequencing gel that was dried and autoradiographed overnight. Hamster probes for a suitable standard (such as glyceraldehyde phosphate dehydrogenase) were not available for these studies. Therefore, results are normalized to total RNA per tissue sample. The relative distributions of DGK message were similar with two independent preparations of tissue RNAs.

Isolation of DGK Clones-
In studies designed to study the hormonal regulation of PKC␣ in the steroid-responsive hamster smooth muscle DDT1 cell line, we determined that glucocorticoids such as triamcinolone acetonide (TAA) did not influence the level of immunoreactive 80-kDa PKC␣. However, TAA treatment did increase the level of a 140-kDa protein (DGK) that cross-reacted with the PKC␣ monoclonal antibody M6 (Fig. 1). In order to isolate the cDNA for this potential PKC␣-related protein, the M6 antibody was used to immunoscreen a glucocorticoid-induced DDT1 Lambda Zap II expression library. Three overlapping clones (pSK4, pSK14, and pSK31) were isolated (Fig. 2). The pSK4 insert (2.6 kb) was selected for further study. Sequence analysis indicated that it contained a 1.3-kb open reading frame and 1.3 kb of 3Ј noncoding sequence. The bacterially expressed sequence was 47 kDa, which correlated with the predicted size of the translated cDNA, and was recognized by M6 (data not shown). However, the deduced amino acid sequence did not have significant homology to PKC␣ or to other protein kinases. Thus, the reason for M6 recognition of this expressed sequence is not due to primary sequence homology and, instead, may be due to some secondary structure common to both proteins.
To obtain additional 5Ј sequence, the library was rescreened with the pSK4 insert. Three larger overlapping clones (pSKA5, A20, and A21, 4.0 kb) were isolated. Since these still appeared to be only partial cDNAs, a 5Ј fragment of pSKA21 (pSKA21a) was used to screen a second DDT1 cDNA library from which clone pSKB24 (4.3 kb) was isolated. 5Ј RACE and inverse PCR were used to generate additional 5Ј sequence. The first 5Ј RACE product (pSK5ЈR1, 0.679 kb) extended 198 bp beyond the 5Ј end of pSKB24. The second 5Ј RACE product (pCRII5ЈR4, 0.452 kb) extended 322 bp further from the 5Ј end of 5ЈR1. The sequence obtained by inverse PCR (pCRII62/26) was identical to 5ЈR4, except that it was 8 base pairs longer. Fig. 2 indicates the spatial relationships of the various DGK clones. The composite sequence is 4.793 kb with an open reading frame of 3.462 kb and a predicted protein size of 127 kDa (Fig. 3). A putative start methionine with a consensus Kozak sequence was identified 25 base pairs after the beginning of the cDNA.
Verification That Clones Are DGK-To verify that the isolated sequences coded for the original 130 -140-kDa protein recognized by the PKC␣ M6 antibody, polyclonal antibodies were prepared to three sequences from different regions of the composite partial DGK cDNA clone. Sequences used for antibody production are shown in Fig. 4A and include: 1) a synthetic 15-amino acid peptide corresponding to the C-terminal deduced amino acid sequence of pSK4 (AJ10), 2) a pSK4-glutathione S-transferase fusion protein (AJ12), and 3) a pSKA21b-histidine-tagged fusion protein (AJ21). All antisera were affinity purified against the cognate protein or peptide. Each of the antibodies recognized a TAA-inducible protein of 130 -140 kDa on immunoblots of DDT1 cell lysates, thus confirming the relationship between the clones and DGK (Fig.  4B). Small differences in DGK recognition among the antibodies were also apparent. Whereas M6, AJ12, and AJ21 clearly recognized a doublet at 140 kDa, AJ10 (the C-terminal antipeptide antibody) recognized only one band. Careful comparison of adjacent lanes stained with M6 and AJ10 indicated that AJ10 recognized the lower (major) band of the doublet. The TAA-mediated increase in DGK protein was accompanied by a TAA-mediated increase in specific mRNA species detected either by Northern blot or ribonuclease protection analysis. 2 Partial and full-length DGK constructs were expressed from baculovirus (BvDGK-p and BvDGK, respectively) in Sf9 cells. The expressed proteins were approximately 120 and 130 kDa, respectively (Fig. 4C). The molecular weights of endogenous hamster brain DGK and recombinant full-length DGK were nearly identical indicating that the putative start methionine indicated in Fig. 3 is at or very near the beginning of the open reading frame.
Sequence Analysis-The composite DGK sequence contains motifs found in other DGK family members including cysteinerich motifs (CRMs) and the putative catalytic subdomains (Fig.  3). DGK is closely related (Ͼ57% homology) to DGK␦ (16) in each of these functional motifs. Unlike other DGK family members, DGK␦ and DGK do not contain identifiable EF hands but do contain an N-terminal pleckstrin homology domain (77% homology). A long intervening sequence between the putative catalytic subdomains is also unique to these DGKs; however, they differ substantially in this intervening sequence (Ͻ38% homology). The C terminus of DGK␦ has homology to the C terminus of the EPH receptor tyrosine kinases. This domain, which is not present in PKC, is thought to be a regulatory domain (16). Thus, despite their similarities in functional motifs, DGK␦ and DGK are distinct sequences that comprise a new subfamily of DGKs.
The spacing of the cysteines in the N terminus of the DGK sequence is characteristic of the general motif HX 10 -12CX 2 CX 12-14 CX 2 CX 4 HX 2 CX 5-7 C found in a variety of signaling molecules including PKCs, Raf kinases, and DGKs (2). 2 T. Klauck, manuscript in preparation. Alignment of the PKC␣, DGK, and c-Raf CRMs shows that the spacing of the cysteines and histidines is highly conserved; however, there is no significant homology outside of these residues (Fig. 5A). In contrast, alignment of the DGK CRMs demonstrates several potentially significant conserved residues (see Consensus in Fig. 5B). In particular, each of the DGK CRM I motifs (except DGK⑀) begins with the sequence GXHXW. An invariant P (or G) occurs two residues before the first cysteine which suggests the importance of secondary structure in this region. CRM 2 motifs also have a conserved W two residues from the beginning and an invariant W between the third and fourth cysteines. The end of CRM 2 is defined by a GX 7 PP sequence that is unique to the DGK family. As noted previously (17), spacing of residues in the CRMs of DGK⑀ is somewhat unique and could potentially be linked to the unique substrate specificity of this isozyme. Thus, the DGK CRMs share sequence similarities that are likely to distinguish their functions from CRMs found in other signaling molecules such as c-Raf, PKCs, unc 13, vav, and n-chimaerin.
All DGK family members contain a second domain of conserved sequence homology that is likely to be important for DGK function and possibly for catalytic activity. In DGK␦ and -, this putative catalytic domain is separated into two subdomains (Fig. 6, A and B). These subdomains are also separated in Drosophila DGK1 (28). Subdomain sequences from each of the DGKs were Ͼ50% identical to the porcine DGK␣ corresponding sequences. Subdomain 1 contains a GXGXXGX 12-14 K motif (at G 473 -K 492 in DGK␣) that is known to participate in ATP binding to protein kinases (10). Although the GXGXXG box is conserved among DGKs, the downstream K, which is essential for ATP binding in protein kinases, is not. Furthermore, recent studies demonstrated that mutations in this region do not affect DGK␣ activity (29). Therefore, this sequence does not appear to be the functional ATP-binding site for DGK phosphotransferase activity. Significant homology between subdomain 1 (but not subdomain 2) and MHCK (30) was also apparent. It should be noted that the GXGXXG motif found in this portion of the MHCK sequence (321-326) does not appear to be the functional ATP-binding domain (which has been mapped to residues 467-473). The homology of MHCK to DGK subdomain 1 may indicate a more general rather than a DGKspecific function.
DGK␣, -␤, and -␥ contain two EF hand motifs that are known to participate in calcium binding and regulation of several proteins, including DGK␣ (19). The absence of EF hand motifs in DGK␦ and -indicate that they belong to a distinct subfamily of calcium-independent DGKs. These results indicate differences in calcium regulation of various DGK activities.
Tissue Distribution of DGK-The distribution of DGK in hamster tissues was characterized at the message level by ribonuclease protection analysis (RPA) and at the protein level by immunoblot analysis. To study the expression of DGK message, a 197-bp antisense RNA probe was made to the 3Ј end of the coding sequence (3.255-3.452 kb) and hybridized to total

140-kDa Diacylglycerol Kinase
RNA samples from various tissues (Fig. 7). This probe corresponds to the final 66 amino acids of hamster DGK that were present in the antigen used to prepare antibody AJ12 used in the immunoblot analysis. Message was detected in every tissue examined with testes being the most abundant. Message levels in brain, lung, spleen, and prostate were also relatively abundant (prostate data not shown).
Immunoblots of total lysates from various hamster tissues were probed with antibody AJ12 (prepared to the final 433 amino acids of DGK) (Fig. 8). These results also demonstrated that DGK is expressed to some degree in most tissues, although abundant message levels (i.e. in the testes) did not always correlate with abundant protein levels. This is potentially due to tissue-specific differences in protein stability or message processing. Of the tissues examined, highest levels were found in brain. Tissue distribution in rats was similar (data not shown). Molecular size heterogeneities among hamster tissues apparent in Fig. 8 were also apparent in immunoblots of rat tissues (data not shown).
DGK Activity-The substantial sequence and domain homologies clearly demonstrate that DGK is related to the DGK family of proteins. To determine if the DGK protein actually had DGK activity, a partial DGK cDNA construct beginning 308 bp from the putative start methionine (3.16-kb open reading frame, 4.112 kb total, see Fig. 3) was cloned into a pBlue-BacHis transfer vector from which a recombinant baculovirus (BvDGK-p) was prepared for expression in insect cells. Ly-sates containing the partial DGK protein had approximately 3-4-fold more DGK activity compared with lysates from cells infected with recombinant PKC␣ baculovirus (Table I). Both the DGK activity and the DGK-immunoreactive protein (Fig. 9) were distributed approximately equally between the soluble and particulate fractions. The partial cDNA used in these expression studies does not contain the pH domain that may influence membrane association. However, we have also noted roughly equal partitioning of DGK between soluble and particulate fractions of cultured cells including fibroblasts and mammary and pituitary epithelial cells. These data show that expression of DGK is associated with increased DGK activity. The specific activity and fold increase are similar to that reported for extracts of COS cells expressing human DGK␣ or -␦ (12,16). It should be noted that DGK activity is dependent upon the type of detergent or lipid included in the assay and on the substrate used (13,31,32). We have not yet explored the optimal assay requirements for DGK activity. The catalytic activity of the partial DGK was extremely unstable as over 50% of the activity was lost after overnight storage at 4, Ϫ20, or at Ϫ70°C in the presence of 10 or 20% glycerol. DISCUSSION In this study we report the isolation of a unique cDNA (DGK) with considerable homology to known DGKs in the CRMs and the two putative catalytic subdomains. Extracts from cells expressing recombinant partial DGK contained significantly more DGK activity than control cells or cells expressing recombinant PKC␣, thus confirming that DGK protein has DGK activity. The positions of the functional domains in DGK family members are summarized in Fig. 10. In general, CRMs are located within the N-terminal half of the proteins, whereas the putative catalytic subdomains are found in the C-terminal regions. Type I DGKs (␣, ␤, ␥) contain EF hands that are associated with the calcium-dependent activities of this type of DGK. Whereas wild type DGK␣ is a calcium-dependent enzyme, EF hand deletion mutants are calcium-inde- pendent (19). Type II DGKs (␦ and ) are not calcium-sensitive enzymes (16). The pleckstrin homology and/or the EPH Cterminal tail homology domains found in these DGKs are likely to play an important role in regulating their activities. DGK⑀ represents a third type of DGK that is distinguished according to its selective hydrolysis of diglycerides containing arachidonate (17). To date, structural motifs that are responsible for restricting the substrate specificity of DGK⑀ have not been identified. Finally, DGK represents a fourth type of DGK that contains two identifiable motifs not found in other mammalian DGKs (18). DGK contains four tandem ankyrin repeats that are also found in Drosophila DGK2 (14). Ankyrin repeats are known to be a general protein recognition motif (33) and could function similarly in DGKs. DGK also contains a sequence homologous to the PKC phosphorylation site on the major PKC Ribonuclease assays were used to estimate DGK message levels in various hamster tissues. A, a 197-bp 32 P-labeled riboprobe was generated from a Hin-dIII/EcoRI fragment of clone 4 (pSK4). B, total RNA (20 g) was isolated from hamster tissues and hybridized to the DGK⑀-specific 197-bp riboprobe as described under "Experimental Procedures." Data are representative of three independent experiments using two separate preparations of RNA from each tissue. Br, brain; Ht, heart; Ki, kidney; Li, liver; Lg, lung; Mu, muscle; Sp, spleen; Ts, testes.

FIG. 8. Tissue distribution of DGK protein.
Hamster tissue homogenates were prepared, and proteins (100 g of protein/lane) were separated by electrophoresis and blotted to NitroPlus. The blot was probed with DGK-specific antibody AJ12. Similar results were obtained in rat tissue distribution studies. Ad, adrenal; Br, brain; Ht, heart; Ki, kidney; Li, liver; Lg, lung; Mu, muscle; Sp, spleen; Ts, testes.

140-kDa Diacylglycerol Kinase
substrate myristoylated alanine-rich C-kinase substrate. In myristoylated alanine-rich C-kinase substrate, this sequence has been reported to be involved in PKC-regulated binding of phosphatidylserine, actin, and calmodulin (34,35). The functions and regulation of this sequence in DGK have not yet been studied. In summary, it is likely that the substantial sequence diversity among DGK family members contributes to unique functions and properties of individual DGKs. Spacing of the cysteines and histidines in the CRMs of DGK and other DGKs is very similar to the conserved 50-amino acid zinc-binding CRMs that have been identified in the PKC family, n-chimaerin and Raf (2,36,37). However, the composition of the intervening residues is unique for each CRM. Functions of individual CRMs have been described for some proteins. For example, phorbol ester/DG binding has been mapped to the CRMs in PKCs (except ) (38 -40). In contrast, the c-Raf CRM does not bind phorbol esters. To date, there is no direct evidence to suggest that the DGK CRMs bind DG or that they are required for catalytic activity. Previous work demonstrated that the porcine DGK␣ is not a high affinity phorbol ester receptor (41). On the other hand, phorbol esters and DG caused redistribution of cellular DGK activity from soluble to particulate fractions in some cells (42)(43)(44). Additional studies are needed to define the DG binding site in DGKs and to determine the function of the DGK CRMs.
The CRM of Raf has been shown to be important for Raf interactions with its upstream effector, Ras (45)(46)(47). Raf-Ras interactions were dependent on a 17-amino acid sequence Nterminal of and including the first eight residues of the single Raf CRM (47)(48)(49). The CRM of Raf also binds phospholipids, which has led to the suggestion that the Raf CRM may be a prototype for a general phospholipid-dependent protein recognition motif. PKC also participates in phospholipid-dependent interactions with other proteins (50,51); however, the role of individual CRMs in these interactions has not yet been studied. It is not yet known if the DGK CRMs mediate DGK interactions with other proteins and/or phospholipids.
The actual site of the catalytic domain in DGK, a lipid kinase, has not yet been defined. Conservation of two homologous subdomains among all known DGKs provides strong evidence that these sequences are essential for DGK activity. These subdomains are contiguous in most mammalian DGKs and in Drosophila DGK2 but are separated in DGK␦, -, and Drosophila DGK1. Surprisingly, a sequence homologous to subdomain 1 is also found in the protein kinase, myosin heavy chain kinase. Therefore, subdomain I is not unique to DGKs and may have a more general function. Although the protein kinase GXGXXGX 12-14 K ATP-binding motif was found in the C terminus of DGK␣, recent studies demonstrate that this sequence is not necessary for DGK activity (29).
Previous work demonstrated that expression of DGK␣, -␤, -␥, -␦, and -⑀ are highly tissue-specific. For example, expression of porcine DGK␣ was limited to the thymus, lymphocytes, and specific regions of the porcine brain (52,53). In contrast, DGK showed a much broader tissue distribution and was highly expressed in brain and testes. In particular, the abundance of message in brain, lung, and spleen distinguish PKC from the more narrowly expressed PKC␦. The broad tissue and cellular distributions suggest that DGK may play a general role in cellular DG homeostasis.
In conclusion, we have identified a novel DGK that is significantly different from previously cloned DGKs based upon its sequence and tissue distribution. Furthermore, glucocorticoid induction of other DGKs has not been noted to date. It is clear that, similar to other signaling molecules such as PKC and phospholipase C, DGK is a heterogeneous family of proteins. There is evidence that individual DGKs may have unique properties and, consequently, unique functions. For example, a number of studies indicate that membrane-associated DGK activities and DGK⑀ (17) preferentially phosphorylate 1-stearoyl 2-arachidonyl DG, a molecular species that is primarily derived from phosphatidylinositol turnover (32, 54 -56). It is probable that other DGKs also have specific roles in DG metabolism and, consequently, in the regulation of DG-dependent biochemical processes. Further studies will be necessary to determine how expression of individual DGKs influences agonist-stimulated and steady state DG levels. Such studies will begin to address the relative importance of different DG pools in DG-dependent cellular processes, such as activation of specific PKC isozymes.