A Domain with Homology to Neuronal Calcium Sensors Is Required for Calcium-dependent Activation of Diacylglycerol Kinase α*

Diacylglycerol kinases (DGKs) phosphorylate diacylglycerol produced during stimulus-induced phosphoinositide turnover and attenuate protein kinase C activation. Diacylglycerol kinase α is an 82-kDa DGK isoform that is activated in vitro by Ca2+. The DGKα regulatory region includes tandem C1 protein kinase C homology domains and Ca2+-binding EF hand motifs. It also contains an N-terminal recoverin homology (RVH) domain that is related to the N termini of the recoverin family of neuronal calcium sensors. To probe the structural basis of Ca2+ regulation, we expressed a series of DGKα deletions spanning its regulatory domain in COS-1 cells. Deletion of the RVH domain resulted in loss of Ca2+-dependent activation. Further deletion of the EF hands resulted in a constitutively active enzyme, suggesting that sequences in or near the EF hands are sufficient for autoinhibition. Binding of Ca2+ to the EF hands protected sites within both the RVH domain and EF hands from trypsin cleavage and increased the phenyl-Sepharose binding of a recombinant DGKα fragment that included both the RVH domain and EF hands. These observations suggested that Ca2+ elicits a concerted conformational change of these two domains. A cationic amphiphile, octadecyltrimethylammonium chloride, also activated DGKα. As with Ca2+, this activation required the RVH domain. However, this agent did not protect the EF hands and RVH domain from trypsin cleavage. These findings indicate that the EF hands and RVH domain act as a functional unit during Ca2+-induced DGKα activation.

Hydrolysis of phosphatidylinositol 4,5-bisphosphate is a common mechanism of stimulus transduction (1). Diacylglycerol (DAG) 1 released in this reaction activates protein kinase C (PKC) and is then rapidly metabolized back to phosphatidyli-nositol in a series of reactions initiated by a diacylglycerol kinase (DGK). As such, DGKs attenuate DAG-mediated PKC activation (2). Recent studies indicate that DGKs are also activated by mechanisms independent of phosphoinositide turnover (3,4). Diacylglycerol kinases catalyze the ATP-dependent phosphorylation of sn-1,2-diacylglycerol to form phosphatidic acid (PA), which is also a lipid mediator (5,6). Several DGK isoforms have been cloned (7). All these sequences share a homologous catalytic domain and two or three C1 protein kinase C homology domains (7)(8)(9). Some DGKs contain EF hands, which are Ca 2ϩ -binding sites (7). These DGKs also have a domain at their N termini with homology to the recoverin family of neuronal calcium sensors (Fig. 1). We term this the recoverin homology (RVH) domain. In S-modulin, the frog orthologue of recoverin, this domain associates with the EF hands to mediate Ca 2ϩ -dependent inhibition of rhodopsin kinase (10).
The varied structures of DGK regulatory domains suggest divergent mechanisms of regulation. Several studies have shown variation among DGKs with regard to activation by phospholipids, sphingosine, or Ca 2ϩ (11)(12)(13)(14)(15). Kanoh and coworkers (16 -18) have studied DGK␣, a Ca 2ϩ -activated isoform highly expressed in oligodendrocytes and thymocytes. They have shown that Ca 2ϩ binds the EF hand region of the enzyme and that deletion of the EF hands results in constitutive enzyme activation (19,20). We have now examined a series of DGK␣ mutants in which the RVH and EF hand domains are sequentially deleted. Our results indicate that the N-terminal RVH domain is required for Ca 2ϩ to activate this enzyme. In contrast to the constitutive activation seen with deletion of the EF hands, DGKs with deletions involving only the RVH domain expressed activity similar to that of wild-type enzyme in the absence of Ca 2ϩ . Sites within both the EF hands and RVH domain were protected from trypsin proteolysis by Ca 2ϩ , indicating that both domains participate in a Ca 2ϩ -induced conformational change. A cationic amphiphile, octadecyltrimethylammonium chloride, markedly stimulated DGK␣ activity in vitro. This effect, like Ca 2ϩ -dependent activation, was dependent on the RVH domain. The DGK␣ RVH domain does not itself bind Ca 2ϩ . However, it does appear to function together with the EF hands to couple Ca 2ϩ binding to release of EF handmediated autoinhibition of DGK␣.
Expression of Diacylglycerol Kinases in COS-1 Cells-Fragments containing DGK␣ and the mutant sequences were subcloned into pCDNA3 (Invitrogen). To facilitate quantitation of expression, a FLAGepitope tag was introduced at C terminus of DGK␣. Primers (s-nhe1, 5Ј-CGAATTCTGGACATCGGCTAGCCAAGT-3Ј, NheI site underlined; and a-flag1, 5Ј-GGCTGCAGTCAGCCTGGTCCCTTGTCGTCATCGTC-TTTGTAGTCCCCTCCGCACAGAAGCCAAAG-3Ј, stop codon underlined) were designed to amplify a FLAG-tagged DGK␣ fragment. PCR amplification was performed at 94°C for 45 s, 58°C for 1 min, and 72°C for 1 min for a total of 25 cycles with a final elongation step at 72°C for 10 min. The amplified fragment was cloned into pBluescript KSϩ. This construct was digested with NheI and XhoI, and the NheI-XhoI fragment was inserted into DGK␣-pCDNA3 in place of a 460-base pair NheI-XhoI fragment at the DGK C terminus, positions 2031-2491. The FLAG epitope was similarly inserted into the truncated DGK␣ constructs.
COS-1 cells were cultured in high glucose Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (HyClone), 100 units/ml penicillin, 100 g/ml streptomycin, 250 ng/ml amphotericin B. Transfection of COS-1 cells was performed by the calcium-phosphate method (22). After 48 h, cells were harvested and lysed by sonication in ice-cold 20 mM Tris⅐HCl, pH 7.4, FIG. 1. Alignment of DGK␣ N termini with neuronal calcium sensors. The first 214 amino acids of DGK␣ were aligned with other Ca 2ϩ -activated DGKs and with neuronal calcium sensors using ClustalX. Numbering is according to the DGK␣ sequence. Neuronal calcium sensors are subclassified into visinins (recoverins), frequenins, VILIPs, and other NCS proteins (42). Sequences of a few other four EF hand proteins are also shown for comparison. Only some of the sequences are shown for regions corresponding to the DGK␣ EF hands. Amino acids conserved in both neuronal calcium sensors and DGKs are shaded. Residues 8 -78 comprise the RVH domain. EF hands conforming to the Prosite consensus are boxed. The first zinc-coordinating amino acid of the DGK␣ C1a domain (His 205 ) is indicated. Guanylate cyclase-activating proteins and calcineurin B subunits did not align as well with the RVH domain and had weaker overall homology. 250 mM sucrose, 100 mM NaCl, 2.5 mM EGTA, 1 mM MgCl 2 , 1 mM DTT, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM PMSF, 50 M ATP, and 0.02% Triton X-100. After removal of undisrupted cells by brief centrifugation, the lysates were centrifuged at 100,000 ϫ g (Beckman TL-100) for 20 min at 4°C to pellet membranes. The resultant supernatants were rapidly frozen in a dry ice/ethanol bath and stored at Ϫ70°C until assayed. For immunodetection, 5-10 g of the 100,000 ϫ g supernatants of lysates from COS-1 cells transiently expressing DGK␣ or the truncation mutants was applied to SDS-PAGE. Proteins were electrophoretically transferred to nitrocellulose membranes (Schleicher & Schuell). After blocking in TBS-T buffer containing 5% non-fat dry milk for 1 h, the membrane was incubated with anti-FLAG M2 antibody (1:2000 in TBS-T) for 1 h. Membranes were washed with TBS-T and incubated with horseradish peroxidase-conjugated sheep anti-mouse IgG (1:5000 dilution in TBS-T) for 30 min. After washing 4 times in TBS-T buffer, the horseradish peroxidase conjugates were detected by chemiluminescence.
The GST-fused DGK␣-pGEX-4T-3 constructs were expressed in E. coli strain BL21(DE3). Tranformed cells were grown at 37°C in LB medium supplemented with 100 g/ml ampicillin to an A 600 nm of 0.8 and induced with 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside for 4 h at 37°C or overnight at 25°C. Cells were harvested and lysed in 20 mM Tris⅐HCl, pH 7.5, 250 mM sucrose, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.5% Triton X-100, 10 g/ml leupeptin, 10 g/ml aprotinin, and 1 mM PMSF by two passes through a French press. The lysates were clarified by centrifugation at 10,000 ϫ g for 20 min. The GST fusions were then purified to apparent homogeneity by glutathione-agarose affinity chromatography. When removal of the GST moiety was desired, the recombinant protein adhering to the glutathione-agarose beads was incubated with 2 IU thrombin per mg of protein at 4°C overnight. The supernatant was then collected and incubated with p-aminobenzamidine-agarose at 1:100, v/v ratio for 1 h to remove the thrombin. The resultant supernatant was dialyzed against 50 mM Tris⅐HCl, pH 7.5, 100 mM NaCl, 250 mM sucrose, and 1 mM DTT. Aliquots were rapidly frozen and stored at Ϫ80°C.
Diacylglycerol Kinase Assays-The standard DGK assay contained in volume of 200 l the following: 1 mM sodium deoxycholate, 50 mM triethanolamine⅐HCl, pH 7.5, 100 mM NaCl, 1 mM MgCl 2 , 1 mM EGTA, 0.1 mM [␥-32 P]ATP (100 cpm/pmol), 20 M sn-1-palmitoyl-2-oleoyl-glycerol (16:0, 18:1 DAG), 1 mM DTT, and enzyme (11,23). In a typical reaction, an appropriate volume of DAG stock solution (10 -20 mM in CHCl 3 ) was evaporated under a stream of nitrogen in a 16 ϫ 100-mm glass test tube. To the DAG droplet were added the following: 50 l of 4ϫ assay buffer, 50 l of 4ϫ detergent, DTT, water, and enzyme to a final volume of 180 l. Stock solutions of 4ϫ sodium deoxycholate, 10ϫ [␥-32 P]ATP, and 4ϫ aqueous buffer were as described previously (21). Reactions were initiated by adding 20 l of 1 mM [␥-32 P]ATP. Reactions were allowed to proceed for 10 min at 25°C and terminated by the addition of 3.0 ml of CHCl 3 /ethanol (2:1 v/v) containing 1.0 mg dihexadecylphosphate and 1.0 mg of sorbitan trioleate. The organic phase was washed 3 times with 2.0 ml of 1.0% HClO 4 and 0.1% H 3 PO 4 in H 2 O/ ethanol (4:1, v/v). The volume of the final organic phase was 2.25 ml. Cerenkov counting 1.2 ml of this organic phase determined incorporation of 32 P into PA. For some assays, mixed micelles of octyl glucoside and phosphatidylserine (PS) were employed instead of deoxycholate. In these assays, the total concentration of micelle components, octyl glucoside ϩ PS ϩ DAG, was maintained at 25 mM. Total octyl glucoside added to the assays was the sum of micellar and monomeric octyl glucoside, which was calculated as described (24). For purposes of these calculations, the critical micelle concentration of octyl glucoside was assumed to be 25 mM. The DAG concentration in these assays was 0.5 mM (2 mol %). Other assay components were unchanged. When Ca 2ϩ was added to assays, the buffer contained 1 mM EDTA instead of EGTA, and the free Mg 2ϩ was maintained at 1 mM. The total Mg 2ϩ and Ca 2ϩ added were calculated using published stability constants to give the desired levels of free cations (25). Other assays employing Triton and OTAC or Triton and C16SB instead of deoxycholate have been described previously (23). All data reported are averages of at least duplicate determinations that agreed within 10% in all cases. Moreover, all results are representative of two or more independent experiments performed with completely independent enzyme preparations.
Ca 2ϩ Overlay Analyses-Calcium binding was assessed by the 45 Ca overlay method of Maruyama et al. (26). 1 g of recombinant proteins or immunoprecipitated DGK␣ truncation mutants were separated by SDS-PAGE and transferred to nitrocellulose. The membrane was washed free of Ca 2ϩ in 75 ml of 5 mM EGTA, pH 7.0. It was rinsed three times with 10 mM imidazole HCl, pH 6.8, 60 mM KCl, and 5 mM MgCl 2 , and then incubated for 15 min at 25°C in 30 ml of the same buffer supplemented with 250 nM 45 CaCl 2 (1 Ci/ml). The membrane was then rinsed twice with 45% ethanol, blotted dry, and exposed to x-ray film. The nitrocellulose was stained with Amido Black to verify protein transfer. To prepare immunoprecipitates of the DGK␣ truncation mutants for these experiments, COS-1 cells transfected with DGK␣ or the truncation mutants were extracted with buffer containing 20 mM Tris⅐HCl, pH 7.4, 250 mM sucrose, 100 mM NaCl, 5 mM EGTA, 1 mM NaF, 1 mM MgCl 2 , 1 mM DTT, 50 M ATP, 1% Triton X-100, 10 g/ml leupeptin, 10 g/ml aprotinin, and 1 mM PMSF. The extract was precleared with protein A/G-agarose beads to eliminate nonspecific binding. Anti-FLAG antibody was then added together with fresh protein A/G-agarose beads, and the mixture was incubated at 4°C overnight. The immune complexes were collected by centrifugation. The precipitate was washed with the above buffer containing only 0.1% Triton X-100, and the bound proteins were eluted into SDS reducing buffer.
Limited Proteolysis of DGK␣:2-202-Recombinant GST-DGK␣:2-202, which includes the RVH and EF hand domains, was expressed in E. coli, and the GST moiety removed as described above. Limited trypsin proteolysis was performed as described by Rudnicka-Nawrot et al. (27). The reaction contained, in a volume 150 l, 0.6 mg/ml (90 g) of DGK␣:2-202, 50 mM Tris⅐HCl, pH 7.5, 100 mM NaCl, 1 mM DTT, and either 0.1 mM CaCl 2 or 0.1 mM EGTA. The reaction was initiated by adding 1 g of trypsin and incubating at 37°C. Samples (18 l) were withdrawn at various times and quenched by adding 2 l of 1 mM N ␣ -p-tosyl-L-lysine chloromethyl ketone. The products were analyzed by SDS-PAGE. To examine the effect of OTAC on proteolysis, the reaction mixture was supplemented with 0.4 mM Triton X-100, 0.4 mM Triton X-114, and 0.2 mM OTAC (20 mol %), and the trypsin concentration was halved. The Triton mixture was added together with the OTAC to simulate the conditions under which maximal activity stimulation is observed (11,23). Triton X-100 and Triton X-114 were also included in the control reactions.
Peptides from selected time points of the trypsin digests were analyzed by capillary liquid chromatography-mass spectrometry using an ABI 140D solvent delivery system. Samples (5 l, 2.7 g of protein) were injected directly into 300-m inner diameter fused silica capillar-ies packed with C18 resin (Vydac). The columns were eluted with a 60-min, 2-95% gradient of acetonitrile in H 2 O containing 0.2% isopropyl alcohol, 0.1% acetic acid, and 0.001% trifluoroacetic acid. The solvent flow rate was 7 l/min. The column was eluted directly into the electrospray ionization source of a Finnigan LCQ mass spectrometer. Nitrogen was used as the sheath gas at 35 pounds/square inch, and no auxiliary gas was used. Electrospray ionization was conducted with a spray voltage of 4.8 kV, a capillary voltage of 26 V, and a capillary temperature of 200°C. Spectra were scanned over an m/z range of 200 -2000. Base peak ions were trapped using the quadrupole ion trap and further analyzed both with a high resolution scan performed at an isolation width of 3 m/z and with collision-induced dissociation scans at a collision energy of 40.0. Sequences of all eluting peptides were confirmed on the collision-induced dissociation scans. Ion currents at m/z ratios corresponding to the peptide fragments were integrated over the entire peak using Finnigan Bioworks software provided with the mass spectrometer.
Other Methods-Calcium-dependent binding of recombinant DGK␣: 2-202 (RVH domain ϩ EF hands, GST cleaved off) and DGK␣:99-202 (EF hands only, GST cleaved off) to phenyl-Sepharose was assayed by a modification of the procedure of Landar et al. (28). Briefly, 240 g of recombinant polypeptide in a total volume of 0.5 ml was loaded onto 1 ml of phenyl-Sepharose columns in 50 mM Tris⅐HCl, pH 7.5, 500 mM NaCl, 1 mM DTT, and either 10 mM CaCl 2 or 10 mM EGTA. The columns were eluted with 4 ml of the same buffer, and 0.5-ml fractions were collected.
Purification of DGK activities from testis cytosol, salivary cytosol, NIH 3T3 cells (two activities), thymus cytosol, and testis membranes has been described (23). Calcium stimulated the activities from testis cytosol, salivary cytosol, and one of the 3T3 activities (23). Methods used for expression of DGK activities in Saccharomyces cerevisiae have also been described (23). Protein concentrations of enzyme preparations were determined by the Bradford method (29). Thin layer chromatography of DAG and PA was as described previously (21). Concentrations of phospholipids used in DGK assays were confirmed by determination of organic phosphate (30). Protein sequences with homology to the DGK␣ N terminus were identified by PSI BLAST searches and aligned  Assays were performed on 100,000 ϫ g supernatants of lysates from COS-1 cells transiently expressing the DGK constructs. These were assayed for DGK activity by the deoxycholate method and by the octyl glucoside/phosphatidylserine method with both 10 and 20 mol % PS. All activities are corrected by subtracting the background DGK activity, determined under identical assay conditions, of a lysate of COS-1 cells transfected in the same experiment with the pCDNA3 vector. Expression levels of DGK␣ and its mutants in COS-1 cell were determined by densitometry of immunoblots using anti-FLAG M2 antibody (Kodak). In all cases, the majority of the anti-FLAG immunoreactivity was in the 100,000 ϫ g supernatant. Immunoreactivity associated with the pellets was estimated from immunoblots as follows: WT DGK␣, Ͻ10%; DGK␣ ⌬40, 15%; DGK␣ ⌬87, 15%; DGK␣ ⌬196, 25% (data not shown). Details of the methods are given under "Experimental Procedures." All data are averages of duplicate determinations which, in all cases, agreed within 10%. Similar results were observed in two additional independent experiments. WT, wild type. using ClustalX (31,32). Structural features of proteins with homology to DGKs were visualized using RasMol 2.6 (33).

Expression of DGK␣ and Its Truncation Mutants in COS-1
Cells-Regulatory domains of Ca 2ϩ -activated DGKs contain several conserved regions. These include EF hands and tandem C1 PKC homology domains (8,9). EF hand-containing DGKs also contain a 70-amino acid conserved sequence at their N termini (34,35). Motif searches revealed that this domain is related to the recoverin family of neuronal calcium sensors (Fig. 1). We thus refer to this region as the RVH domain. The homology between DGKs and neuronal calcium sensors extends through the EF hands (Fig. 1).
To investigate the role of the RVH domain in DGK␣ regulation, a series of N-terminal deletion mutants was prepared (Fig. 2). DGK␣ ⌬40 lacks the first half of the RVH domain and DGK␣ ⌬87 lacks the entire RVH region. DGK␣ ⌬196 lacks both the RVH domain and the EF hands. A FLAG epitope attached to the C termini facilitated detection and quantification of protein expression. These mutants were expressed in COS-1 cells. Cytosol from COS-1 cells expressing DGK␣ or the deletion mutants showed a marked increase in DGK activity as compared with control cells transfected with vector only. The mutant activities were stable in cell lysates but were only partially recoverable from DEAE-cellulose or Mono Q columns. The 100,000 ϫ g supernatants of COS-1 lysates were thus used for all studies. The presence of protein products with the expected molecular masses was verified by immunoblotting (Fig.  2). Densitometry of the immunoblots indicated that DGK␣ ⌬40, DGK␣ ⌬87, and DGK␣ ⌬196 were expressed at 1/4th, 1/4th, and 1/8th the level of wild-type DGK␣. COS-1 cells express an endogenous DGK activity. Transfected cells all expressed activities at least 20-fold over this background. After subtraction of the COS-1 background and normalization for expression, the specific activities of DGK␣ ⌬40 and DGK␣ ⌬87, which lack RVH domains, were not significantly different from wild-type enzyme (Table I). However, deletion of the EF hands (DGK␣ ⌬196) resulted in an 18-fold increase in specific activity. Similar results were obtained in both the deoxycholate and octyl glucoside/PS assays. All of the constructs were activated 5-10fold as the surface concentration of PS in octyl glucoside micelles was increased from 10 to 20 mol % ( Table I). As discussed below, the ⌬40, ⌬87, and ⌬196 mutant activities were also activated by OTAC and C16SB similarly to several Ca 2ϩ -independent DGKs. To obtain further evidence that the mutants express DGK activity, we also expressed the truncated DGKs in S. cerevisiae. All of the mutants expressed DGK activity in the yeast, and in all cases, the phosphatidate product co-migrated with authentic PA on thin layer chromatograms (data not shown). As yeast do not possess an endogenous DGK, expression of activity in this background provides additional confirmation that the mutant DGKs are catalytically competent (23). These results indicate that all the truncated DGKs express a functional catalytic domain. Activities expressed by full-length DGK␣ and DGK␣ ⌬196 with the native C termini were identical to those expressed by the corresponding epitopetagged constructs, indicating that the FLAG tag does not appreciably alter DGK␣ activity.
Activation of DGK␣ Truncation Mutants by Ca 2ϩ -We examined Ca 2ϩ activation of full-length DGK␣ and the deletion mutants. Wild-type enzyme showed significant stimulation by Ca 2ϩ , close to maximum activity being achieved with 5 M cation (Fig. 3). However, none of the mutants, including DGK␣ ⌬40 and ⌬87, which retain the EF hands, showed any Ca 2ϩdependent activation. To exclude the possibility that deletion of RVH domain disrupted the folding of EF hands, we immunoprecipitated DGK␣ ⌬40 and ⌬87 with anti-FLAG antibody and performed Ca 2ϩ overlay assays on these immunoprecipitates (Fig. 4). Our results clearly showed that deletion of RVH domain did not affect Ca 2ϩ binding. As described below, deletion of the RVH domain also had no effect on Ca 2ϩ binding to regulatory domain sequences expressed in E. coli. Thus, whereas Ca 2ϩ binding occurs at the EF hands, the RVH domains are also required for Ca 2ϩ -dependent activation. The activities of the ⌬40 and ⌬87 mutants were similar to that of wild-type DGK␣ in the absence of Ca 2ϩ . Additional deletion of the EF hands increased the activity to a level equal to that of fully Ca 2ϩ -activated wild-type enzyme (Fig. 3). This suggests that sequences in or near the EF hands inhibit the catalytic domain and that in full-length DGK␣, Ca 2ϩ relieves this inhibition.
A Ca 2ϩ -induced Conformational Change Involves Both the EF Hands and the RVH Domain-Binding of Ca 2ϩ to recoverin elicits a 45°rotation of the region corresponding to the DGK␣ RVH domain relative to the C-terminal EF hands (36). The loss of Ca 2ϩ -dependent activation in the RVH-deleted mutants may reflect loss of a similar conformational change during DGK␣ activation. We expressed the RVH domain, EF hands, and C1 domains of DGK␣ as recombinant proteins in E. coli (Fig. 5). Overlay assays indicated that DGK␣:2-202, DGK␣:99-202, DGK␣:2-336, and DGK␣:99-336, all of which include the EF hands, bind Ca 2ϩ (Fig. 5). The RVH domain alone did not bind Ca 2ϩ and was not required for Ca 2ϩ binding (Fig. 5). To probe for a Ca 2ϩ -induced conformational change, we performed limited trypsin proteolysis of DGK␣:2-202 (DGK␣-RVHϩEF) with and without Ca 2ϩ . Aliquots of the reaction were stopped at various times and analyzed by SDS-PAGE (Fig. 6). Calcium protected DGK␣:2-202 from proteolysis (Fig. 6). In the absence of Ca 2ϩ (0.1 mM EGTA), the polypeptide was completely digested within 30 min, whereas in 0.1 mM Ca 2ϩ , appreciable full-length polypeptide remained after 4 h. These results are not due to an effect of Ca 2ϩ on trypsin. High concentrations of Ca 2ϩ (Ͼ1 mM) protect trypsin from autolysis and modestly stimulate its activity, but these effects should be negligible in the 0.1 mM Ca 2ϩ used for these experiments (37,38). Moreover, if Ca 2ϩ were stimulating trypsin activity, the protection we observed would be even more significant.
To identify those sites particularly susceptible to trypsin, peptide fragments in the 5-min and 4-h digests were analyzed by capillary liquid chromatography-mass spectrometry. Table  II shows integrated ion currents of the peptides identified in each digest. In the 5-min digests, Ca 2ϩ had only small effects on the appearance of most peptides derived from sequences between the N terminus and Lys 58 . However, cleavage at Lys 25 was inhibited by Ca 2ϩ . Calcium markedly inhibited cleavage at sites from Lys 89 /Arg 90 to the C terminus. In the 4-h digests, Ca 2ϩ reduced the level of most fragments, which is consistent with the fact that appreciable full-length polypeptide was still present on SDS-PAGE. As was seen in the 5-min digests, the levels of peptides derived from cleavage at Lys 89 /Arg 90 and beyond were much more dramatically reduced by Ca 2ϩ . However, Ca 2ϩ significantly increased the levels of three peptides in this region. One of these, Leu 120 -Asn 202 , includes both EF hands, whereas Asn 126 -Asn 202 and Met 144 -Asn 202 encompass the second EF hand only. Protection of these fragments from further digestion by Ca 2ϩ suggests that both EF hands of DGK␣ coordinate Ca 2ϩ , as previously concluded by Yamada et  Fig. 6. Samples of the digests from the 5-min and 4-h time points were analyzed by capillary high pressure liquid chromatography/mass spectrometry as described under "Experimental Procedures." The table shows the fragments identified (sequence numbering as in Fig. 1), the predicted and observed masses, and the integrated ion currents observed in full mass spectrometry mode. al. (39). Overall, Ca 2ϩ inhibited tryptic cleavage of sites within the RVH domain (Lys 25 ), in the loop between the RVH domain and the EF hands (Lys 89 /Arg 90 ), and within the EF hands (multiple sites). Calcium-induced conformational changes of EF hand proteins, including neuronal calcium sensors, can result in altered binding to hydrophobic resins (28,40,41). We thus examined whether Ca 2ϩ modulated the binding of DGK␣-RVH ϩ EF to phenyl-Sepharose. As shown in Fig. 7, inclusion of Ca 2ϩ caused DGK␣-RVH ϩ EF to be retained by the resin. The peak elution in EGTA was between 1 and 1.5 ml, whereas in Ca 2ϩ it was between 2 and 3 ml. The EF hands alone (DGK␣:99-202) bound tightly to phenyl-Sepharose, both with and without Ca 2ϩ , and were not eluted under the conditions used for this experiment (data not shown). This suggests exposure of a hydrophobic surface upon loss of the RVH domain. Overall, these observations are consistent with a Ca 2ϩ -induced conformational change involving both the RVH domain and the EF hands.
Activation of DGK␣ by OTAC and C16SB-We have previously shown that DGK␣ is markedly stimulated by the cationic amphiphile, OTAC (11). This agent stimulated three other Ca 2ϩ -dependent DGK activities from testis cytosol, salivary cytosol, and NIH 3T3 cells to a similar degree (data not shown). Several Ca 2ϩ -independent DGK activities, including an arachidonoyl-DGK from testis membranes, and cytosolic activities from NIH 3T3 cells and thymus cytosol were only modestly (2-6-fold) activated by OTAC (11,23). Wild-type DGK␣ activity expressed in COS-1 cells was activated 109-fold by OTAC (Table III). Similar stimulation was seen with the zwitterionic amphiphile, C16SB, although 15-fold higher concentrations were required to achieve the same activation. In lysates of COS-1 cells expressing the truncated DGK activities, OTAC and C16SB had much smaller effects (Table III). Activation of the mutants was comparable to that seen with Ca 2ϩ -independ-ent DGKs. In the presence of maximally activating concentrations of OTAC, Ca 2ϩ does not further activate wild-type DGK␣ (23). These results suggest that OTAC and C16SB have two effects on DGKs, a nonspecific stimulation seen with all isoforms and an additional stimulation seen only with Ca 2ϩ -activated DGKs. This latter effect, like Ca 2ϩ -dependent activation, required the RVH domains. OTAC and C16SB thus appear to act through RVH and EF hand domains to mimic Ca 2ϩ -dependent activation of DGK␣. We examined whether OTAC, like Ca 2ϩ , protects DGK␣:2-202 from trypsin proteolysis. In the absence of Ca 2ϩ , OTAC modestly accelerated the proteolysis of DGK␣:2-202 (data not shown). In the presence of OTAC, Ca 2ϩ no longer protected DGK␣:2-202 (Fig. 8). Triton alone slightly increased proteolysis, but the rate was greatly accelerated by OTAC. Inclusion of OTAC/Triton or C16SB/Triton in the binding and wash buffers of overlay assays had no effect on 45 Ca 2ϩ binding (data not shown). Overall, these results suggest that C16SB and OTAC mimic Ca 2ϩ -dependent DGK␣ activation by disrupting EF hand-mediated autoinhibition but do not compete with Ca 2ϩ binding or mimic the Ca 2ϩ -induced conformational change. Loss of OTAC and C16SB stimulation of DGK␣ activity with deletion of the RVH domain provides independent evidence of a role for this domain in enzyme activation. DISCUSSION Our results indicate that Ca 2ϩ -dependent diacylglycerol kinase ␣ activation requires an N-terminal RVH domain with homology to the recoverin family of neuronal calcium sensors. Several amino acids within the N-terminal region of recoverin that corresponds to the DGK RVH domain contribute to Ca 2ϩ -  dependent inhibition of rhodopsin kinase (10). Neuronal calcium sensors derive from an ancestral 4 EF hand protein but have a unique structure (42). In contrast to the dumbbellshaped structure of calmodulin and troponin C, neuronal calcium sensors are folded into compact globular structures (36,(43)(44)(45). Neuronal calcium sensors also have short helices at the N and C termini and between EF3 and EF4 that are not present in calmodulin ( Fig. 1; positions 8 -17, 147-153, and 195-200) (36,43). Guanylyl cyclase-activating proteins and calcineurin B subunits are related to neuronal calcium sensors but do not align as well in the regions corresponding to the RVH domain (Fig. 1). Consistent with this, the N-terminal helix of guanylyl cyclase-activating protein-2 has a different orientation than that of recoverin (44). Alignment of the first 204 amino acids of DGK␣ with recoverin suggests that many of its unique features are also present in DGKs (Fig. 1). Nonpolar amino acids involved in the interaction between the N-terminal region of recoverin and its EF hands have homologues in DGKs ( Fig. 1; Y70, H73, V74, F78, I100, A101, M104, L120, Y121, I137, and M144, numbering from the figure). Our observation that the DGK␣ EF hands bind phenyl-Sepharose much more tightly than the combined RVH ϩ EF hand polypeptide may reflect exposure of many of these residues upon loss of the RVH domain. These similarities suggest that the DGK N-terminal region resembles recoverin. Neuronal calcium sensors are myristoylated at their N termini. In Ca 2ϩ -free recoverin, the myristoyl moiety is tucked into a hydrophobic pocket formed by several nonpolar residues ( Fig. 1; L13, L27, W30, Y31, F53, I56, Y57, F60, F61, Y98, V99, L102, W116, and L120, numbering from the figure) (36). These positions are also nonpolar in DGK␣ and other EF hand-containing DGKs, suggesting that, although DGKs are not myristoylated, this pocket is conserved. Binding of Ca 2ϩ to recoverin results in a 45°rotation of the region corresponding to the DGK␣ RVH domain relative to EF hands 3 and 4 (36,46). The two domains pivot about Gly 108 , which is conserved in DGKs. Calcium binding to recoverin also ejects the myristate and N-terminal helix (36,46). DGKs lack the consensus for N-terminal myristoylation but align with neuronal calcium sensors in the region corresponding to the N-terminal helix (Fig. 1, positions 8 -17). In DGK␣, Ca 2ϩ protected Lys 25 from trypsin cleavage. The corresponding amino acid in recoverin is located in a loop between the first and second helices, far from the C-terminal EF hands, and becomes less exposed as a result of the Ca 2ϩ -induced conformational transition. In recoverin, sequences corresponding to the ancestral EF1 and EF4 are disabled, and it is binding of Ca 2ϩ to EF3 that elicits the conformational change (46). The two EF hands of DGKs correspond to the ancestral EF3 and EF4. Overall, these similarities suggest that Ca 2ϩ activation of DGK␣ involves a conformational transition of the N-terminal region analogous to that of recoverin. Consistent with a coupled conformational transition, Ca 2ϩ protected sites within both the RVH domain and the EF hands from trypsin cleavage. Deletion of the EF hands from DGK␣ resulted in a constitutively active enzyme, suggesting that Ca 2ϩ activation releases an autoinhibitory effect exerted by this region. The presumed conformational change of the RVH domain and EF hands may cover a region of the EF hands involved in autoinhibition and/or contribute to Ca 2ϩ -induced unfolding of the entire enzyme. Surprisingly, OTAC, a cationic amphiphile, and C16SB, a zwitterionic amphiphile, also activated DGK␣. As with Ca 2ϩdependent activation, this effect required the presence of the RVH domain. However, OTAC did not protect the RVH/EF hands polypeptide from trypsin proteolysis, indicating that it does not elicit the same conformational change as Ca 2ϩ . This result provides additional evidence that the RVH domain plays a critical role in DGK␣ activation. Overall, our results indicate that functional coupling of the DGK␣ EF hands to its RVH domain is required to couple Ca 2ϩ binding to release of the catalytic domain from EF hand-mediated autoinhibition.