Role of the Diacylglycerol Kinase α-Conserved Domains in Membrane Targeting in Intact T Cells*

Diacylglycerol kinase (DGK) phosphorylates diacylglycerol to phosphatidic acid, modifying the cellular levels of these two lipid mediators. Ten DGK isoforms, grouped into five subtypes, are found in higher organisms. All contain a conserved C-terminal domain and at least two cysteine-rich motifs of unknown function. DGKα is a type I enzyme that acts as a negative modulator of diacylglycerol-based signals during T cell activation. Here we studied the functional role of the DGKα domains using mutational analysis to investigate membrane binding in intact cells. We show that the two atypical C1 domains are essential for plasma membrane targeting of the protein in intact cells but unnecessary for catalytic activity. We also identify the C-terminal sequence of the protein as essential for membrane binding in a phosphatidic acid-dependent manner. Finally we demonstrate that, in the absence of the calcium binding domain, receptor-dependent translocation of the truncated protein is regulated by phosphorylation of Tyr335. This functional study provides new insight into the role of the so-called conserved domains of this lipid kinase family and demonstrates the existence of additional domains that confer specific plasma membrane localization to this particular isoform.

The transient generation of lipids at the cell membrane by the concerted action of lipases, kinases, and phosphatases is a very effective mechanism for the localization and activation of signaling molecules. The majority of lipid-modifying enzymes are cytosolic proteins that must in turn translocate to the membrane to modify their substrates. Characterization of the mechanism that governs membrane translocation of these proteins is, thus, essential to assess precisely their role in the regulation of cell responses.
The diacylglycerol kinases (DGK) 5 are a family of enzymes that phosphorylate diacylglycerol (DAG) to produce phospha-tidic acid (PA); this modulates the levels of these two lipids, both of which have recognized second messenger functions. Early experiments showed that DGK activity occupies a central position in phospholipid synthesis; this enzyme attracted additional interest when it was shown to participate in intracellular signaling (1). DGK activity is found in organisms from bacteria to mammals, although the protein identified as a DGK in bacteria is an integral membrane protein (2) and has little structural homology with the DGK characterized in multicellular organisms (3). Ten DGK isoforms have been identified in mammals and grouped into five subtypes based on the presence of various domains in their primary sequences, which define distinct regulatory motifs.
Translocation from one subcellular compartment to another appears to be a general theme in the regulation of DGK proteins. Many of the early data regarding DGK activity regulation refer to changes in DGK activity or localization based on enzymatic studies without specifying the isoform concerned (4,5). Characterization of the DGK isoforms has broadened the concept that these proteins regulate their actions by translocating from one cell compartment to another and by forming part of different protein complexes (6 -11). The non-conserved regulatory domains appear to govern subtype-specific DGK translocation. The characterization of these domains has allowed an initial outline of the signaling pathways responsible for the subcellular relocalization of the various DGK isoforms (9,12,13).
In addition to their distinct regulatory motifs, all DGK family members have a conserved catalytic domain, proposed to harbor the ATP-binding site. Precise characterization of the DGK␣ ATP-binding site has remained elusive; all DGK have a conserved GGDGXXG motif that is essential for activity that is hypothesized to fulfill this function (14). Members of this lipid kinase family share another conserved structure, a cysteinerich motif (CRM), present at least twice in all DGK. The homology of DGK CRM domains to the phorbol ester/DAG binding, C1-type motifs of protein kinase C originally led to the idea that these motifs were responsible for DAG binding in DGK. Sequence analysis nonetheless indicates that, with the exception of the first C1 domains in DGK␤ and DGK␥, the DGK C1 domains lack the key residues that define this motif as a canonical C1-like, phorbol ester binding domain (15). The participation of the conserved C1 domains in DGK enzyme activity, thus, remains a matter of debate. Some studies suggest that these conserved motifs are required (16), whereas others found them dispensable for DAG phosphorylation (17).
DGK␣ is a type I DGK expressed abundantly in the cytosol of T lymphocytes that translocates to the membrane during T cell activation (18). DGK␣ acts as a negative modulator of the signals that determine T lymphocyte activation (8). T cells from DGK␣-deficient mice have enhanced responses (19), further confirming the relevance of this enzyme in regulating immune function. Structural-functional analyses showed that the regulatory N-terminal domain of the enzyme, containing a recoverin-like domain and a tandem of EF hand motifs, acts as a negative regulator of enzyme activation and receptor-induced membrane translocation (8,20). Translocation of wild type (wt) DGK␣ is a very rapid and transient event, whereas a catalytically inactive mutant form remains at the membrane for a longer period of time (18). This suggests feedback regulation based on PA generation, which guarantees transient DGK␣ localization at the membrane. Deletion of the DGK␣ N-terminal domain does not alter enzyme activity but results in constitutive localization of the enzyme at the plasma membrane in intact T cells. This suggests that the Ca 2ϩ regulatory domain, characteristic of type I enzymes, is responsible for this PA negative feedback.
Studies in primary T cells and T cell lines demonstrate that DGK␣ translocates to the plasma membrane after stimulation of endogenous T cell receptor (TCR) and/or ectopically expressed muscarinic type I receptor (8,18). Analysis of other cell types also suggests that the plasma membrane is the main site of DGK␣ activity (13,21). It is still not known, however, how DGK␣ is targeted specifically to this subcellular compartment. We used the DGK␣ construct that lacks the Ca 2ϩ regulatory domain to further analyze the structural determinants responsible for its constitutive localization at the plasma membrane. Using mutational analysis and studies in intact cells, we show that the two DGK␣ C1 domains, although not necessary for catalytic activity, are critical for plasma membrane targeting. We identify the sequence C-terminal to a conserved Prorich region also necessary for membrane binding through a still unidentified mechanism. Finally, we confirm that Tyr 335 phosphorylation is an absolute requirement for DGK␣ membrane localization in a receptor-dependent manner. With this study, we complete a detailed mapping of the diverse DGK␣ motifs and provide new insight into the functional role of the domains conserved in most members of this lipid kinase family.

EXPERIMENTAL PROCEDURES
Cell Culture-The J-HM1-2.2 cell line was generated by stable transfection of the human muscarinic subtype-1 receptor in the Jurkat T cell line (22). J-HM1-2.2 cells were maintained in RPMI 1640 medium (Invitrogen); COS-7 and Jurkat cells were purchased from the ATCC and cultured in Dulbecco's modified Eagle's medium (BioWhittaker). All media were supplemented with 10% heat-inactivated fetal calf serum (Invitrogen).
Plasmids and Transfection-A cDNA fragment encoding mouse DGK␣ was subcloned into the EcoRI site of the mammalian expression vector pcDNA 3.1 containing a 5Ј-hemagglutinin tag in the molecular cloning site (a kind gift of Dr. A. Zaballos, Centro Nacional de Biotecnología, Madrid, Spain). The GFP-DGK␣ and GFP-DGKa⌬EF constructs have been described (8). GFP-DGKa⌬EFKD (kinase-dead) and GFP-DGKa⌬EFKDY335F were generated by direct mutagenesis of Gly 433 to Ala (KD) or Tyr 335 to Phe (Y335F) using GFP-DGKa⌬EF as a template. GFP-DGKa⌬CT mutants were generated by generating specific DNA restriction sites.
DGK Assay-Transfected cells were frozen on dry ice and lysed by sonication in DGK assay buffer (50 mM Tris HCl, pH 7.4, 100 mM NaCl, 0.25 M saccharose, 1 mM dithiothreitol, 1 mM EDTA, 5 mM EGTA, and the protease inhibitors phenylmethylsulfonyl fluoride, leupeptin, and aprotinin). Cell lysates were centrifuged (20,000 ϫ g), and the supernatant was used for the DGK reaction as described (23). Protein (20 g) from whole lysates or immunoprecipitates of tagged proteins was used as enzyme source. Standard phosphorylation assays were performed (10 min, 37°C; final volume, 50 l). The reaction was terminated by adding 100 l of 1 M HCl, and lipids were extracted with 20 l of CHCl 3 /MeOH (1:1, v/v). Organic layers were recovered, dried, dissolved in 20 l of CHCl 3 , and applied to silica gel TLC plates with dioleoyl-PA as a standard. Plates were developed with a CHCl 3 , MeOH, 4 M NH 4 OH solvent system (9:7:2, v/v/v). Dried plates were autoradiographed, and bands corresponding to PA were quantified by autoradiogram scanning.
Fractionation Studies-T lymphocytes were harvested, washed in PBS, resuspended in TES buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.25 M succors) with protein inhibitors, and lysed by 10 passages of the cell suspension through a 25-gauge needle. Nuclei were removed by centrifugation. Supernatants were centrifuged (100,000 ϫ g, 1 h), and pellets were resuspended in TES buffer supplemented with 1% Nonidet P-40. Supernatant and pellets were resolved in SDS-PAGE and analyzed in Western blot with the indicated antibodies.
Analysis of CD69 Cell Surface Expression-Transfected cells were stimulated with anti-CD3/anti-CD28 antibodies. CD69 expression on the cell surface was analyzed 6 h after stimulation gating for GFP-positive cells) using a phosphatidylethanolamine-conjugated anti-human CD69 monoclonal antibody (Pharmingen). Immunofluorescence intensity of the cells was determined by flow cytometry (FACSCalibur, BD Biosciences).
Immunofluorescence Microscopy-T cells were harvested 36 h after electroporation with plasmids, washed, and allowed to attach to poly-L-lysine-coated coverslips (1 h, room temper-ature). In the case of TCR stimulation, Jurkat cells were harvested 24 h post-transfection, washed once, and resuspended in HEPES-balanced solution. Cells were transferred to chambered coverslips coated with anti-CD3/CD28 monoclonal antibody (final concentration 5 g/ml, 4°C, overnight). All subsequent incubations and washes used a 0.5% bovine serum albumin, 0.1% Triton X-100 in PBS buffer; antibodies were incubated (37°C, 1 h) in a humidified chamber. Nuclei were labeled with Topro-3 (Molecular Probes). Fluorescence was analyzed on a Leica confocal microscope (TCS-NT) with the associated software. For inhibitor treatment, the DGK inhibitor R59949 (10 M; Calbiochem) was added after the first frame, and images were recorded every 11 s.
Stimulation with Antibody-coated Microspheres-Antibodies were adsorbed to microspheres by mixing 1 g of antibody (1:1 CD3:CD28) in PBS with 0.5 ϫ 10 6 microspheres in a final volume of 1 ml and incubated (1.5 h at room temperature) with continuous mixing; 1.5 ml of 1% bovine serum albumin in PBS was added, and mixing was continued (30 min). Microspheres were washed three times with PBS and suspended in PBS for the addition to cells. For stimulation, cells were mixed with antibody-coated microspheres at a 2:1 cell/bead proportion and plated on poly-DL-lysine-coated chamber slides. Images were captured each 15 s by confocal microscopy and processed using ImageJ software.

Identification of the DGK␣ Minimal Catalytic
Core-DGK␣ has a highly conserved catalytic domain and a variable N-terminal domain shown to exert a negative regulatory role for enzyme activity and subcellular localization. The exact identity of the "minimal catalytic domain" needed to phosphorylate DAG nonetheless remains unclear. To study enzymatic activity of the catalytic domain, we engineered a protein lacking the regulatory N terminus and both C1 domains (⌬327DGK␣) (Fig.  1A). Expression of the truncated protein was confirmed by Western blot, and enzymatic activity was determined in cell lysates and immunoprecipitates of transfected COS cells (Fig.  1B). The expression level of this construct was reproducibly much lower than that of the wt enzyme (Fig. 1B). However, enzymatic assay on immunopellets revealed that the DGK␣⌬ 327 mutant retains significant activity, confirming that the C1 domains are not directly related to the catalytic activity.
C1 Domains Are Essential for DGK␣ Membrane Targeting-In addition to the C-terminal conserved domain, all DGK family members have at least two C1 domains, which belong to the general HX 10 -12 CX 2 CX 12-14 CX 2 CX 4 HX 2 CX 5-7 C motif found in more than 200 distinct signaling proteins (15). Structural analysis showed that in addition to Cys and His, other residues are required for phorbol ester and DAG binding (24). Based on the presence of these additional residues, C1 domains can be classified as typical or atypical (25). Sequence analysis indicated that, with the exception of the first C1 domains in DGK␤ and DGK␥, DGK C1 domains lack one, two, or all three residues (Pro, Gly, and Gln) that define a C1 as a typical phorbol ester binding (and probably DAG binding) domain ( Fig. 2A).
To study the role of C1 domains in DGK␣ cell membrane localization, we used a Jurkat T cell line variant in which TCR-like responses, including DGK␣ translocation, are triggered by carbachol stimulation of an ectopically expressed muscarinic type I receptor (8,26). We engineered point mutations of the conserved His in the first and the second C1 in the full-length GFP-DGK␣ sequence. Membrane translocation was then examined by confocal microscopy after carbachol stimulation of the cells. Mutation of the conserved His residues in either C1 prevented protein translocation in response to receptor stimulation (Fig. 2B). Alignment of DGK C1 sequences showed a conserved Trp after the first His, with the exception of the first and second C1 of DGK, which had the Phe generally found in C1 domains in other proteins ( Fig. 2A). Because the Trp after the first His appears to be characteristic of DGK, we analyzed the role of this residue in DGK␣. A conservative mutation of this Trp to Phe did not alter receptor-regulated translocation. Mutation to a non-conservative Gly abolished enzyme translocation after receptor stimulation, indicating the need for an aromatic residue (Fig. 2B). These results point to the DGK␣ C1 region as essential for membrane localization of the protein through lipid and/or protein interactions.
Deletion of the N-terminal Ca 2ϩ binding domain results in a protein (GFP-DGK␣⌬EF) constitutively located at the plasma membrane when expressed in Jurkat cells (8). We next analyzed whether the C1 domains were also required for constitutive membrane localization of this mutant. As for the wt protein, mutation of Trp to Phe had no effect on enzyme localization, whereas mutation to Gly induced complete protein dissociation from the membrane (Fig. 2B, bottom). These results concur with those described for the full-length construct and show that failure of the enzyme to translocate after receptor stimulation was not due to a delay in translocation kinetics. Finally, a construct expressing the C1 domains alone fused to enhanced GFP did not localize to the plasma membrane (not shown), suggesting that this region, although necessary, is not sufficient for DGK␣ plasma membrane localization.
The previous experiments suggest that the C1 region, whereas necessary, is not sufficient for plasma membrane localization. We next decided to perform additional mutations/truncation on the GFP-DGK␣⌬EF mutant to investigate the nature of the requirements for its constitutive membrane localization. A summary of the different mutants used in the next experiments is shown in Fig. 3.

Deletion of the DGK␣ C-terminal Region Prevents Constitutive
Membrane Association of GFP-DGK␣⌬EF-Functional studies with the DGK␤ isoform have identified an alternative spliced isoform that lacks the last 35 amino acids. Deletion of this C-terminal region greatly alters plasma membrane localization of this isoform in vivo (27). Sequence analysis of the C-terminal sequence of type I enzymes indicated conservation of a Pro-rich sequence followed by an amino acid stretch with no obvious conservation among distinct subtypes (Fig.  4A). To analyze the relevance of the C-terminal sequence in DGK␣ stability at the membrane, we generated a new construct in which the last 13 amino acids of the GFP-DGK␣⌬EF construct were deleted (GFP-DGK␣⌬EF⌬CT). When expressed in Jurkat T cells, we found that the protein showed the appropriate molecular weight, but DGK activity was much lower than that detected for GFP-DGK␣⌬EF when assayed in vitro (Fig. 4B). Confocal analysis in intact cells established that deletion of the C-terminal sequence disrupted constitutive membrane localization of the GFP-DGK␣⌬EF construct at the membrane (Fig. 4C). This suggests that, as previously shown for DGK␤, this region might represent an important site for membrane interaction. Fractionation analysis confirmed that deletion of the C-terminal domain affects cytosol/membrane distribution (Fig. 4D). We have previously demonstrated that the GFP-DGK␣⌬EF construct acts as a constitutive active DGK␣ mutant, decreasing the DAG levels generated during stimulation of the TCR. As a result, cells expressing this mutant show decreased membrane expression of the activation marker CD69 in response to TCR triggering (8). To further confirm that deletion of the C terminus affected membrane localization of DGK␣ and consequently enzyme function, we determined CD69 expression after TCR stimulation in cells expressing GFP-DGK␣⌬EF or GFP-DGK␣⌬EF⌬CT. As is shown, the attenuation on CD69 expression observed in GFP-DGK␣⌬EF expressing cells is lost in cells expressing GFP-DGK␣⌬EF⌬CT (Fig. 4E).
PA Generation Stabilizes GFP-DGK␣ in the Absence of the N-terminal Region-The GFP-DGK␣⌬EF deletion mutant shows higher activity in vitro than the wt enzyme and is no longer stimulated by Ca 2ϩ (8,28). We reasoned that lack of plasma membrane localization of GFP-DGK␣⌬EF⌬CT could be related to its diminished enzymatic activity. To further confirm the requirement for enzyme activity in the constitutive plasma membrane localization of the GFP-DGK␣⌬EF mutant, we generated a kinase-dead version of this construct (DGK␣⌬EFKD) by mutation of Gly 433 to Ala in the conserved GGDG sequence. All DGK are characterized by a conserved GGDGXXG motif that was first identified in the Drosophila melanogaster DGK (14) and which is essential for catalysis (8,12). As previously shown for the wild type protein bearing the same mutation (8), the GFP-DGK␣⌬EFKD did not phosphorylate DAG in vitro (not shown). Again, plasma membrane localization was observed for the GFP-DGK␣⌬EF mutant (Fig.  5A, left). As predicted, this localization was lost for the GFP-DGK␣⌬EFKD (Fig. 5A, right), confirming that DGK␣ activity is needed for plasma membrane stabilization in the absence of the Ca 2ϩ regulatory motifs.
Lack of enzyme activity does not prevent receptor-dependent translocation of DGK␣ (8). We next analyzed whether GFP-DGK␣⌬EFKD could relocate to the plasma membrane in response to external stimuli. The construct was ectopically expressed in Jurkat T cells, and localization was examined after triggering of the TCR and the co-stimulatory molecule CD28. Plating the cells onto anti-CD3/anti-CD28-coated plates induced the characteristic spreading of the cells that was accompanied by rapid relocalization of GFP-DGK␣⌬EFKD to the plasma membrane (Fig. 5B). Subcellular fractionation of unstimulated cells showed some redistribution of the GFP-DGK␣⌬EFKD with the membrane fraction. This probably reflects partial localization of this mutant at internal membranes, at organelles difficult to identify by confocal analysis of intact cells. Fractionation of stimulated cells further confirmed the complete relocalization of the mutant to the membrane in response to TCR triggering (Fig. 5C). These experiments demonstrate that, whereas lack of enzyme activity impairs stabilization of GFP-DGK␣⌬EFKD at the plasma membrane, receptorderived signals are able to relocate the truncated enzyme to the site of substrate generation. The C-terminal region was again indispensable for TCR-dependent translocation since GFP-DGK␣⌬EF⌬CT, which lacks this region, remained in the cytosol under the same conditions (Fig. 5D).

Phosphorylation of Tyr 335 Is Required for TCR-dependent Membrane Translocation-
The previous experiments demonstrate that the C1 domains together with the C-terminal region are essential for DGK␣ binding to the plasma membrane. The GFP-DGK␣⌬EFKD mutant translocates to the membrane in response to TCR triggering, although it does not contain Ca 2ϩ regulatory elements. This suggests that, in addition to Ca 2ϩ generation, supplementary signals are needed to promote receptor-dependent DGK␣ translocation to the membrane.
Phosphorylation of the Tyr 335 residue, found in the hinge region between the C1 domains and the conserved catalytic region of DGK␣, has been proposed to be necessary for stimulusdependent translocation to the membrane of ectopically expressed DGK␣ (21). We, therefore, tested whether this residue was also important for either constitutive localization of GFP-DGK␣⌬EF or receptor-dependent translocation of GFP-DGK␣⌬EFKD. To this end new mutants were generated where Tyr 335 was mutated to Phe both in the GFP-DGK␣⌬EF and the GFP-DGK␣⌬EF mutants. Replacement of Tyr 335 by Phe did not impair enzyme activity or constitutive membrane localization of the GFP-DGK␣⌬EF mutant (Fig. 6A). This suggests that, once the protein is located at the membrane, phosphorylation of this Tyr is no longer required for membrane stabilization. On the contrary, the GFP-DGK␣⌬EFKDY335F mutant was unable to translocate to the membrane after TCR triggering (Fig. 6B) showing that whereas phosphorylation of this residue is not needed for enzyme stabilization at the membrane, it is essential for receptor-dependent DGK␣ translocation. To further evaluate if translocation to the membrane was exclusively related to TCR dependent signals, Jurkat T cells were stimulated by incubation with anti-CD3/C28-coated microspheres. As was observed, the GFP-DGK␣⌬EFKD mutant relocated to the membrane with apparently exclusion from the T cell synapse (Fig. 6C). Again Tyr 335 mutation to Phe fully prevented relocalization of the protein, confirming the role of this residue for TCR-dependent regulation of DGK␣.

DISCUSSION
The DGK enzymes have recently emerged as an important family of lipid kinases that participate in a number of biological processes. Most of these enzymes are found in the cytosol and must relocate to the membrane, where their substrates are found.
Like the other two type I DGK, DGK␣ contains Ca 2ϩ regulatory elements and is activated in vitro in response to Ca 2ϩ and anionic lipids (28). Studies in intact lymphocytes have shown that the N-terminal region acts as negative regulatory domain preventing membrane localization of DGK␣ in the absence of receptor-triggered signals (8). Removal of the Ca 2ϩ regulatory elements generates a truncated form of DGK␣ that presents constitutive localization at the plasma membrane (8). Based on these previous observations, we have performed additional analyses to explore in depth the requisites for DGK␣ binding to membranes in intact cells.
Our studies reveal that impairment of enzyme activity by mutation of the conserved Gly 433 in the catalytic region disrupts constitutive membrane localization of the truncated mutant. Mutation of the DGK␣ C1 domains and truncation of the C-terminal region also prevent constitutive membrane binding. Whereas mutants lacking enzyme activity relocated to the membrane in response to receptor stimulation, intact C1 domains and the C-terminal region were absolutely necessary for membrane localization, suggesting independent functions.
DGK␣ contains two C1 domains that, according to the classification by Hurley and Misra (25), are atypical and do not bind DAG. This is a common feature for the C1 domains of all DGK family members, with the exception of the first C1 in DGK␤ and DGK␥. As expected phorbol 12-myristate,13-acetate cannot bind to the DGK␣ C1 domains (17) and does not induce enzyme translocation when added to intact cells (8,13). Our results indicate that the C1 domains are not necessary for DGK␣ kinase activity and suggest that previous reports of very low (17) or no C-terminal domain activity (29) were probably due to differences in assay conditions. These studies also concur with reports showing that a mutant devoid of the C1 domain retains activity and can be activated by phosphatidylserine (PS) (20). We show that, although they are atypical, the DGK␣ C1 domains represent an essential module in plasma membrane interaction through protein and/or lipid binding. Our experiments agree with previous reports using the DGK and DGK isoforms (9,30) and confirm that the essential role of the C1 domains for membrane localization is a general characteristic of the DGK family. The precise function of C1 domains in DGK membrane binding is not known, although they have been identified as sites of interactions with lipids (31) and proteins (7), as shown for atypical C1 domains in other proteins such as Raf (32). The upper panel shows an autoradiogram of the TLC plate. One-tenth of the sample was analyzed for protein expression by SDS-PAGE and Western blot (WB, lower panel). One representative experiment is shown of three performed with similar results. C, the DGK␣ mutants were transfected into Jurkat T cells, and plasma membrane localization was assessed by confocal microscopy of live cells. The figure shows single confocal frames. D, transfected Jurkat T cells were lysed, and subcellular fractions were purified as indicated under "Experimental Procedures." Cytosol (C) and membrane (M) fractions were analyzed by SDSPAGE, and the presence of protein was assessed with anti-GFP antibodies. Expression of transferrin receptor (TfR) and IB␣ was determined as control of the purity of the fraction. E, transfected Jurkat T cells were stimulated with antiCD3/ antiCD28 antibodies, and CD69 expression in the GFP-positive population was determined by FACS analysis. Experiments are representative of three independent experiments. NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48

Functional Mapping of DGK␣ Conserved Domains
Our analysis reveals that DGK␣ C-terminal domain plays a role in membrane binding similar to that reported for DGK␤ (27). Ectopically expressed GFP-DGK␤ localizes at the plasma membrane of exponentially growing HEK293 cells, whereas it is cytosolic in quiescent cells (27). Phorbol 12-myristate,13-acetate treatment of cells induces membrane translocation of GFP-DGK␤ by virtue of its typical C1 domain. Characterization of a spliced form with a truncated C terminus has shown that this region prevents phorbol 12-myristate,13-acetate-induced translocation even in the presence of an intact C1 domain. Our results confirm the relevance of the DGK␣ C-terminal region for membrane binding and suggest a general mechanism that is conserved in type I isoforms.
The role of the DGK␣ C-terminal region in membrane binding is unknown at the moment, although our experiments show that truncation of this region impairs catalytic activity. Two other lipid kinases ceramide kinase and sphingosine kinase share with the DGKs the conserved DGK catalytic region. Both enzyme families contain an SGDG motif reminiscent of the GGDGXXG signature (33,34). Mutation of the second Gly in the sphingosine kinase SGDG motif abolishes kinase activity as does mutation of the second Gly residue in the DGK GGDG motif (35). Recent studies comparing the catalytic domains of DGK with those of ceramide kinase and sphingosine kinase have characterized several conserved Asp residues extended along the DGK catalytic region that are essential for enzyme activity (36). One of these residues, Asp 697 , lies near the deleted C-terminal region (Fig. 3), suggesting that the integrity of the catalytic site extends along the entire C-terminal region. Interestingly, D697N mutation impairs Ca 2ϩ activation of DGK␣, confirming interaction of the N-terminal regulatory domain with the C-terminal region.
Structural function analysis of sphingosine kinase have revealed that its C-terminal domain is also important for membrane targeting, suggesting a common feature of proteins containing a conserved DGK catalytic domain (37). These studies by Delon et al. (37) identified the binding lipid at the C-terminal region of sphingosine kinase as PA. At this point we cannot conclude that PA binding to the C-terminal domain is responsible for membrane localization. The analysis of the DGK␣ C-terminal domain showed little conservation in this region among the type I isoforms, except for two Pro residues followed by a sequence that is highly conserved within each subtype. The presence of positively charged residues (Lys and Arg) in the DGK␤ C-terminal domain (27) suggested a PA-interacting region. No such sequence is present in DGK␣, except for a single Arg. Functional analysis of the PA binding domain in other proteins such as Raf has nevertheless shown that a single Lys can be responsible for PA recognition (38).
Mutation of Gly 433 in the conserved GGDG motif of the catalytic region impairs enzyme activity and disrupts constitutive membrane localization of the truncated mutant. This, again, strongly suggest a role for PA in stabilizing DGK␣ at the membrane in the absence of the Ca 2ϩ regulatory domain. However, other possibilities such as existence of auxiliary proteins that will bind only to the active DGK␣ must also be considered.
In intact cells, DGK␣ bearing a deletion of the N-terminal region is found exclusively at the plasma membrane, which is greatly enriched in PS (39). Exposure of additional sites for PS recognition as a result of N-terminal domain truncation would be in accordance with previous reports showing that DGK␣ is activated in vitro by PS only in the presence of Ca 2ϩ (28). Thus, a conformation change after Ca 2ϩ binding would be required to expose recognition motifs for this lipid. Truncation of the N-terminal region including the C1 domains does not abolish PS-dependent enzyme activation (36), confirming that PS recognition sites are found in the catalytic region. There is limited structural information on the stereospecific recognition of the PS head group by its effector proteins. Crystallization of the protein kinase C␣ C2 domain bound to a PS molecule revealed direct interaction of an Asn in the side chain with the Ser in the PS head group, whereas a Thr in the side chain interacts with other PS polar moieties (40). The DGK␣ C-terminal domain contains polar residues (Asn, Ser, and Thr) that are conserved among DGK␣ orthologues, which could provide the residues required for PS interaction. It is, thus, tempting to speculate that the lack of membrane stabilization of the C-terminal-truncated mutant is due to the deletion of a region that allows PS and/or PA-mediated binding.
Lack of kinase activity in the DGK␣⌬EF mutant eliminates constitutive plasma membrane localization, although TCR triggering can relocate this mutant to the membrane confirming that kinase activity is not required for receptor-dependent DGK␣ translocation. Tyr phosphorylation of DGK␣ is proposed to be an important mechanism for its receptor-dependent membrane localization (8). Tyr 335 in the human sequence modulates vitamin E-dependent membrane relocalization of DGK␣ (21). Our results here confirm the importance of this residue in receptor-dependent translocation, suggesting that its phosphorylation induces a DGK␣ conformation that favors membrane binding. Again, either mutation of the C1 domain or truncation of the C-terminal region impairs membrane localization in response to external stimuli, further confirming that these two regions represent important membrane binding domains.
This structural-functional analysis of the DGK␣ conserved motifs allows us to draw important conclusions regarding enzyme structure. DGK␣ activity requires its location at the membrane, where its substrate lies. This is the role of the C1 domains, which is necessary for membrane localization in all DGK family members. In contrast, the C-terminal domain appears to be essential only for type I enzymes. The simplest explanation for the lack of membrane localization of the GFP-DGK␣⌬EFKD mutant is that PA generation contributes to membrane binding in the absence of the N-terminal domain. Alternatively, the mutation might alter the conformation of the catalytic domain, diminishing enzyme affinity for the membrane. This would be compensated by Tyr 335 phosphorylation, which induces a conformation that promotes C1 and C-terminal-mediated membrane interactions. Based on our functional analysis of enzyme domains and other available data, we propose a detailed model of DGK␣ activation after receptor engagement (Fig. 7). In the absence of receptor triggering, the DGK␣ N-terminal Ca 2ϩ regulatory domain interacts with the C-termi-    NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48 nal half of the protein, masking all regions required for membrane binding. This closed conformation maintains the enzyme inactive in the cytosol. After receptor stimulation, a Ca 2ϩ -dependent conformational change takes place favoring Tyr 335 phosphorylation. This in turn induces an open, active conformation that allows enzyme interaction with the membrane. The C1 domains and the C-terminal region represent binding sites for membrane PS, facilitating DGK␣ localization to the plasma membrane, where it phosphorylates DAG. Finally, the elevated PA concentration and dephosphorylation of DGK␣ restore the "closed" conformation, and the enzyme returns to the cytosol. This model of DGK␣ membrane targeting and activation would provide exquisite control of DAG membrane levels, responsible in turn for activation/localization of DAG-regulated proteins. Our study provides new insights into the mechanisms by which membrane recruitment of this important enzyme is regulated and establishes a basis for additional systematic mechanistic and functional studies.