Molecular Cloning and Characterization of the Human Diacylglycerol Kinase β (DGKβ) Gene

Diacylglycerol kinases are key modulators of levels of diacylglycerol, a second messenger involved in a variety of cellular responses to extracellular stimuli. A number of diacylglycerol kinases encoded by separate genes are present in mammalian genomes. We have cloned cDNAs encoding several isoforms of the human homologue of the rat diacylglycerol kinase β gene and characterized two such isoforms that differ at their carboxyl terminus through alternative splicing and the usage of different polyadenylation signals. Quantitative analysis of gene expression in a panel of human tissue cDNAs revealed that transcripts corresponding to both isoforms are co-expressed in central nervous system tissues and in the uterus, with one variant being expressed at relatively higher levels. As green fluorescent protein fusions, the two isoforms displayed localization to different subcellular compartments, with one variant being associated with the plasma membrane, while the other isoform was predominantly localized within the cytoplasm. Differences were also observed in their subcellular localization in response to phorbol ester stimulation. Enzymatic assays demonstrated that the two isoforms display comparable diacylglycerol kinase activities. Therefore, the human diacylglycerol kinase β gene can generate several enzyme isoforms, which can display different expression levels and subcellular localization but similar enzymatic activities in vitro.

Diacylglycerol (DG) 1 represents a key signaling intermediate downstream of Gq/phospholipase C-␤ coupled receptors. It is synthesized by phospholipase C-␤ from phosphoinositides as a response to a variety of molecules involved in intercellular communication, including hormones, neurotransmitters, and growth factors. The best known target of DG action is protein kinase C (PKC), whose activity plays a central role in the control of proliferation and differentiation of many different cell types (1). In addition, DG can also modulate the activity of intracellular proteins such as members of the Rho and Ras families, thus potentially affecting other cellular functions such as cytoskeletal organization (2). The control of steady-state cellular levels of DG is therefore crucial to normal cellular physiology. The signaling properties of DG are terminated by its conversion to phosphatidic acid (itself a second messenger, reviewed in Ref. 3, through the action of diacylglycerol kinases (DGKs), a class of evolutionary conserved enzymes presently counting nine mammalian subtypes encoded by separate genes (␣, ␤, ␥, ␦, ⑀, , , , ; reviewed in Ref. 4). The cloning of DGK genes from several mammalian species (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19) has allowed an analysis of the domain architecture of these proteins. Structurally, mammalian DGK subtypes are characterized by the presence of conserved domains, common to all subtypes and of additional subtype-specific functional domains, which allow their grouping into five separate classes (3,4). The functional domains shared by all DGK subtypes comprise the catalytic (kinase) domain and a cysteine-rich domain with homology to the C1A and C1B motifs of PKC. By analogy with the C1A and C1B motifs in PKC, the cysteine-rich domains of DGK have been proposed to bind DG and present it to the catalytic domain (4,20). However, a functional analysis of DGK␣, DGK␤, and DGK␥ demonstrated that the catalytic domain in these proteins lies in the COOH-terminal region, outside the zinc finger domain (21). The subtype-specific functional domains (present in different DGK classes) include motifs of known function (e.g. calcium binding motifs, pleckstrin homology domains, myristoylated alanine-rich C kinase substrate (MARCKS) phosphorylation site domains) as well as domains showing significant sequence conservation but whose function remains unknown (e.g. the C1 domain of Class I DGKs) (4). Very recently, plant DGK isoforms generated from the same gene through alternative splicing have been characterized which differ at their COOH terminus for the presence of a calmodulin-binding domain (22); however, the presence of calmodulin-binding domains in mammalian DGK enzymes has not been described to date. The precise function of the subtype-specific protein motifs is, in many instances, still unclear. However, the diversity and complexity of DGK enzymes indicates the potential for regulation of DGK activity at several levels, including subcellular localization (e.g. see Refs. 12,19,and 23), substrate specificity (e.g. see Refs. 11,15,and 16), and modulation of kinase activity (e.g. see Refs. 24 -26).
While the expression of DGK genes in non-central nervous system tissues is relatively limited (except perhaps for the genes encoding DGK⑀ and DGK), the brain is a major site of DGK gene expression (4,27,28), implicating functional diversification of DGK enzymes in this organ. Recently, a detailed analysis has been carried out of DGK expression in rat brain, with special reference to DGK␣, DGK␤, and DGK␥ (28), showing localized expression in distinct brain regions for these DGK subtypes. DGK␤ is a member of class I DGKs. Like the other members of this subgroup (the DGK␣ and DGK␥ subtypes), it is characterized by the presence of a calcium binding (EF-hand) motif and an additional 70-amino acid domain of uncharacterized function, in addition to the kinase domain and the cysteine-rich domain motifs shared with all DGKs. To date, the only mammalian DGK␤ gene for which a full-length cDNA is available has been cloned from the rat (8). In that study, expression of rat DGK␤ was found to be predominant in the adult brain, with lower expression levels in spleen, adrenal, small intestine, and heart. In the rat brain, DGK␤ expression localizes to the caudate putamen, the nucleus accumbens, the olfactory bulb and tubercle, and the hippocampal pyramidal cell layer (28). The enzymatic properties of DGK␤ and its expression pattern make it an important modulator of PKC activation in response to signaling downstream of G proteincoupled receptors, in brain regions representing important glutamatergic, dopaminergic, acetylcholinergic, and serotonergic terminal fields (29). Given the important role proposed for PKC in the control of mood, learning, and memory (e.g. see Ref. 30), modulation of the DG pool in these brain areas by DGK␤ may be relevant to cognitive and emotional aspects of central nervous system function.
We have previously reported the identification and characterization of the human DGK␤ locus using bioinformatics tools and presented RT-PCR evidence supporting the expression of mRNAs from this locus and the existence of alternative splicing events capable of generating a number of isoforms (31). In particular, one of the transcripts arising from the human DGK␤ locus corresponds to an EST annotated in GenBank TM as differentially expressed in bipolar disorder patients (31), which can encode a DGK␤ protein displaying a COOH-terminal truncation downstream of the catalytic domain. This EST potentially implicates the control of DGK␤ splicing in mood disorders, although nothing is known of the properties of the encoded protein variant and of the relative abundance of its transcripts. Here we report the cloning of cDNAs encoding multiple DGK␤ isoforms from human brain, confirming that alternative splicing of the DGK␤ locus yields a number of transcripts encoding different DGK␤ isoforms. We have then focused our analysis on the alternative splicing event potentially associated with mood disorders, which can generate DGK␤ isoforms differing at their COOH terminus, and show that the encoded proteins display comparable enzymatic activities but different expression levels, responses to increased substrate availability and subcellular localization. The potential implications of these findings for DG signaling in the context of normal and aberrant neuronal physiology are discussed.
Tissue Culture, RNA Isolation, and Northern Blotting-Total RNA was extracted from rat tissues as described (33). Neuro2-A, PC12, and NB-OK-1 cells were cultured as described previously (34). Polyadenylated RNA was extracted from subconfluent monolayers employing a modified oligo(dT)-cellulose binding protocol (31), and 3 g of each sample were loaded on denaturing gels and subjected to Northern blot analysis according to standard procedures (31). Blots were exposed on Kodak XAR-5 films at Ϫ80°C with intensifying screens for 10 days. Human polyadenylated RNA from various human tissues was purchased from CLONTECH (Palo Alto, CA).
In Vitro Translation-The cDNAs for hDGK␤ STD and hDGK␤ SV3Ј (GenBank TM accession numbers AX032742 and AX032745, respectively) were subcloned into expression vector, pcDNA3.1/V5/His-TOPO. The resulting constructs were used as templates for an in vitro transcription reaction coupled to translation reaction (TNT kit, Promega, Madison, WI). Control reactions contained no template or a luciferase containing plasmid vector as template. 7.5% SDS-PAGE coupled to Western blot analysis using anti-V5 antibody as a probe was used to identify the protein produced in TNT reactions.
Enzymatic Assays-The cDNAs for hDGK␤ STD and hDGK␤ SV3Ј in pcDNA3.1/V5/His-TOPO or the vector alone were transfected into COS cells by use of LipofectAMINE (Invitrogen). After 3 days the cells were harvested and lysed by sonication in lysis buffer (8). Immunoblotting was carried out with anti-epitope tag antibody (anti-V5 antibody). The amounts of the expressed proteins (hDGK␤ STD and hDGK␤ SV3Ј) were equalized by means of densitometric indication of the corresponding bands, and the protein concentrations were also adjusted. Equal amount of each lysate was used to measure DG kinase activity in the presence of calcium (0.1 mM) by the octylglucoside mixed-micelle assay toward 1-stearoyl-2-linoleoyl-sn-glycerol (18:0/18:2 DG (Sigma)) as described previously (8,9,12).
Subcellular Localization Studies in HEK-293T Cells-Mammalian expression vectors encoding human DGK␤ variants were constructed in pEGFP-C1 (CLONTECH). Full-length DGK␤ STD and DGK␤ SV3Ј were fused to a green fluorescent protein (GFP) sequence attached at their NH 2 termini using a combination of cDNA fragments and engineered PCR fragments that ensured reading frame conservation.
All PCR-based constructs were verified by dideoxy chain termination sequencing with dye terminators using an ABI PRISM 6700 Automated nucleic acid work station (PE Biosystems).
Living cells transfected with GFP constructs were observed with a confocal laser scanning microscope (Axioplan 2, LSM 510, Carl Zeiss, Göttingen, Germany) at 24 or 48 h after transfection.
In translocation experiments, the culture medium was replaced after transfection with medium without fetal bovine serum for 1 day before tetradecanoylphorbol 13-acetate (TPA) addition. TPA was purchased from Sigma.

RESULTS
Cloning of Human DGK␤ cDNAs-We have previously reported (31) the identification of the human DGK␤ locus, characterized its genomic structure, and determined through bioinformatics and RT-PCR approaches that the locus is transcribed as a complex series of mRNAs due to alternative splicing and the use of different polyadenylation signals. This analysis predicted the existence of three alternatively spliced exons internal to the coding sequence and of a polyadenylation signal in the intron immediately downstream of the penultimate coding exon of the gene. A series of RT-PCR studies, which employed primers spanning each of the three putative alternatively spliced exons or were specific for the alternative 3Ј regions of DGK␤, confirmed the existence of these alternative splicing events (31). These studies were carried out on different preparations of human tissue cDNA, prepared from independent batches of commercially obtained mRNA. In particular, the existence of an alternatively spliced 3Ј exon can result in the generation of transcripts predicted to encode two human DGK␤ variants. One of such variants is fully homologous to the predicted rat DGK␤ amino acid sequence (and is thus referred to as "standard human DGK␤ isoform," or DGK␤ STD), and the other represents a truncated variant lacking part of the coding sequence present within the last exon (and is thus referred to as "splice variant 3Ј human DGK␤ isoform," or DGK␤ SV3Ј). The existence of three internal alternatively spliced exons results in further complexity in the predicted coding capacity of the human DGK␤ locus, potentially resulting in DGK␤ STD or SV3Ј proteins with alternative amino acid regions. To physically clone human DGK␤ cDNAs, we elected to amplify by proofreading PCR the entire predicted coding sequences from cDNA derived from fetal and adult human brain, based on the nucleotide sequence of the human DGK␤ genomic locus and the predicted rat and human DGK␤ coding sequences (31). To obtain cDNAs for the predicted variants, we employed a single forward primer anchored 5Ј to the predicted DGK␤ initiation codon and two alternative reverse primers anchored 3Ј to the stop codons predicted for the STD and SV3Ј DGK␤ variants. The resulting PCR products were cloned and subjected to DNA sequencing. This analysis confirmed the cloning of three DGK␤ STD cDNAs and five DGK␤ SV3Ј cDNAs. Collectively, these cDNA clones represented 8 of the 16 DGK␤ isoforms potentially generated from the human DGK␤ locus (Fig. 1) and confirmed the predicted cDNA structure and splicing events previously reported by us (31). To further confirm the identification of the human homologue of the previously reported rat DGK␤, a Northern blot comprising a panel of polyadenylated RNAs from adult rat tissues comparable with the one reported in Ref. 8 was probed with one of the obtained human DGK␤ cDNA clones. The results ( Fig. 2A) revealed a single transcript of ϳ6.5 kb present exclusively in the brain. In their report, Goto and co-workers (8) observed DGK␤ transcripts expression in the heart and adrenal; our inability to detect DGK␤ expression in these rat tissues employing a human probe may be due to lower sensitivity of our Northern blotting assay, differences in mRNA quality, or in the use of a probe from a different species (human versus rat). To address this discrepancy, we performed a quantitative RT-PCR (TaqMan) analysis on a panel of cDNAs prepared from various rat tissues, using a set of primers and probe specific for the published rat DGK␤ sequence (Fig. 2B). The results confirmed that, among the adult rat tissues sampled in our experiments, the brain appears to be the only organ expressing DGK␤ transcripts, in accordance with the Northern blotting data obtained using a human DGK␤ probe (Fig. 1). The apparent size of the transcript and the predominance of expression in brain tissue are in accordance with the data reported for the rat DGK␤ gene (8).
The cloning of human DGK␤ cDNAs with different structures indicates that the human locus can generate transcripts potentially encoding several protein isoforms, which differ in the presence or absence of four amino acid insertions of 7, 12, 25, and 35 residues, respectively (see Fig. 1). Of particular interest are DGK␤ variants displaying a deletion of the COOHterminal 35 amino acids, comprising a region 3Ј of the catalytic

FIG. 1. Cloning and sequence analysis of multiple human DGK␤ cDNAs.
A, agarose gel electrophoresis of the products resulting from RT-PCR amplification of human DGK␤ cDNAs from fetal (HFB) and adult brain (HAB) cDNA. DNA products of approximately 2.4 kb were obtained from each cDNA source. B, domain architecture of human DGK␤ proteins. The conserved functional domains are indicated, as well as the protein regions relevant to the alternative splicing and the alternative polyadenylation events (stippled areas). Due to the existence of these events the human DGK␤ locus can potentially generate up to 16 different isoforms. C, summary and classification of the obtained full-length human DGK␤ cDNAs, demonstrating the existence of transcripts capable of encoding multiple DGK␤ isoforms. The isoforms indicated with an asterisk refer to the DGK␤ STD and SV3Ј isoforms analyzed in this manuscript.
domain. In fact, deletion of this domain is associated with a human DGK␤ EST (dbEST Id: 1371986 GenBank TM accession number AF019352) annotated as differentially expressed in mood disorder patients (31) and suggests that DGK␤ variants bearing this truncation may be overrepresented in at least some cases of bipolar affective disorder. Given the importance of DG levels for PKC activity, the role of PKC in modulating aspects of cognitive and emotional behavior, and the expression of DGK␤ in relevant areas of the rat central nervous system (8,27,28), further studies on DGK␤ isoforms were focused on the comparative analysis of the expression, function, and subcellular localization of DGK␤ proteins lacking the COOH-terminal coding exon.
Quantitative Expression Patterns of DGK␤ STD and DGK␤ SV3Ј-To determine the pattern of expression of DGK␤ transcripts potentially encoding the DGK␤ STD and DGK␤ SV3Ј isotypes, a series of quantitative RT-PCR (TaqMan) studies were carried out on a panel of cDNAs synthesized from human tissue samples, including several distinct brain regions (e.g. see Ref. 34). Primers and probes were designed to recognize transcripts encoding either the DGK␤ STD or the DGK␤ SV3Ј isoforms, independently of the presence or absence of the internal alternatively spliced exons. The results (Fig. 3) indicate that the human DGK␤ locus is expressed predominantly in the central nervous system and uterus, with much lower levels of expression in other tissues. Increased DGK␤ transcript levels in adult brain relatively to fetal brain indicate that DGK␤ expression may be developmentally regulated. Within the adult brain, the amygdala, caudate nucleus, and hippocampus express the highest levels of DGK␤ transcripts, in agreement with data previously reported for the rat DGK␤ gene (8,27,28). Although the tissue-specific profiles of expression for transcripts encoding the COOH terminus variants are essentially superimposable, important differences are observed in the relative levels of expression. In fact, transcripts encoding DGK␤ STD proteins are expressed in the adult brain at levels about 10 -100-fold higher than transcripts encoding DGK␤ SV3Ј isotypes (Fig. 3). An important observation is the expression of DGK␤ transcripts in the uterus, suggesting a role for this enzyme in aspects of uterine function. The apparent discrepancy between the absolute and ␤-actin-normalized expression profiles of DGK␤ SV3Ј transcripts may be due to the low levels of expression of DGK SV3Ј mRNAs in normal human tissue.
Enzymatic Properties of DGK␤ STD and DGK␤ SV3Ј Isotypes-The existence of transcripts capable of encoding DGK␤ isotypes differing at their COOH terminus may indicate that the encoded proteins differ in their enzymatic activities. To address this aspect of DGK␤ function, experiments were carried out in which the enzymatic activities of DGK␤ STD and SV3Ј isotypes (comprising the three internal alternatively spliced exons) were assayed as previously described for the rat DGK␤ enzyme (8). First, an in vitro translation experiment was set up to determine whether the DGK␤ STD and SV3Ј cDNAs can be translated into proteins. The results (Fig. 4A) indicated that DGK␤ STD and SV3Ј cDNAs can direct the synthesis of proteins with an apparent size in accordance with the predicted molecular mass of 90.5 and 87.0 kDa, encoded by their respective open reading frames. Second, a series of DG kinase assays were carried out to investigate the catalytic properties of the two isoforms. These enzymatic assays demonstrated that the DGK␤ STD and SV3Ј display comparable DG kinase activites under in vitro conditions (Fig. 4B). Therefore, the COOH-terminal region missing in the DGK␤ SV3Ј does not appear to influence the kinase activity of the enzyme, despite its vicinity to the protein's catalytic domain.
Differential Localization and Pharmacological Responses of DGK␤ STD and DGK␤ SV3Ј Proteins-An important aspect of the regulation of DGK activity is its subcellular localization. Although many DGKs are predominantly cytosolic, some DGK isoforms (notably the ␣, , , and isotypes) have been reported to be present in the nucleus (reviewed in Ref. 4) where they are thought to modulate levels of nuclear DG. Nuclear DGK activity has been associated with the control of cell division and may even affect mRNA synthesis and splicing (reviewed in Ref. 4). When overexpressed in COS-7 cells, the rat DGK␤ enzyme was found exclusively in the membrane fraction (8). The discovery and cloning of several isoforms of the human DGK␤ enzyme (Ref. 31 and this work) prompted an investigation into the possibility of enzyme localization to different subcellular compartments. To this end, translational fusions were constructed between the GFP and STD or SV3Ј variants. GFP-DGK␤ fusions were then overexpressed in HEK-293T cells and their subcellular distribution investigated under standard culture conditions using confocal microscopy. The results (Fig. 5A) indicated that the DGK␤ STD variant localized predominantly to the plasma membrane (in a manner consistent with previous observations on rat DGK␤; Ref. 8), while the SV3Ј variant was predominantly cytosolic. Notably, neither variant was found in the nucleus. GFP fusion of DGK␤ STD was associated with the plasma membrane only in growing cells. Indeed, quiescent cells cultured without serum did not show plasma membrane localization of DGK␤ STD (Fig. 5B, upper left panel). Next, the response of GFP-DGK␤ isoform fusions to treatment with the phorbol ester TPA in serum-starved HEK-293T cells was investigated (Fig. 5B). Membrane translocation in response to phorbol esters is a well established phenomenon for various DG-binding proteins such as DGKs (35) PKCs (1) and Munc 13-1 (37), although the exact mechanism of TPA-induced translocation remains to be clarified. Quiescent HEK-293T cells transfected with the DGK␤ STD isoform-GFP fusion showed a predominantly cytosolic distribution of the fusion protein.
When challenged with TPA, a rapid (10 min) redistribution of the protein to the membrane was observed (Fig. 5B). The possible presence of a membrane-targeting signal in the COOH-terminal tail of DGK␤ STD was investigated using various protein motif data bases and software tools such as Pfam (38), PRINTS (39) ProDom (40), Prosite (41), PSORT II (42), and SMART (36). However, no membrane-targeting signal was discovered in DGK␤ STD. In contrast, the distribution of the DGK␤ SV3Ј isoform-GFP fusion was identical in cycling and quiescent HEK-293T cells, with no significant effects of TPA on its subcellular distribution. In conclusion, these results suggest that the DGK␤ STD and SV3Ј isoforms differ in their subcellular localization and in their response to increased substrate concentration. DISCUSSION We report for the first time the identification and functional characterization of cDNAs encoding different isoforms of the human DGK␤ gene. The results reported here support the data previously reported using bioinformatics tools (31) and confirm that the human DGK␤ gene is transcribed as a complex series of mRNAs, as a result of alternative splicing and differential polyadenylation signal usage. This is only the second analysis concerning a mammalian DGK␤ gene and its encoded proteins and the first report providing evidence for the existence of multiple DGK␤ isoforms with different properties. The isolation of fulllength cDNAs with the potential to encode different DGK␤ isoforms suggests the existence of a degree of diversification of human DGK␤ activity, which may involve the modulation of isoform expression, enzymatic activity, and/or of subcellular localization. The predicted human DGK␤ isoforms differ in the presence of three internal, alternatively spliced coding exons and a COOH-terminal region, which is differentially present due to alternative polyadenylation signal usage (31). As a result, a total of 16 different isoforms can be generated from the human DGK␤ locus. From fetal and adult human brain, we have cloned and sequenced cDNAs encoding eight of these, which presumably represent the most abundantly transcribed DGK␤ mRNA species. Evidence for the existence of transcripts capable of encoding the remaining DGK␤ isoforms has been described elsewhere (31). The lack of isoform-specific DGK␤ antibodies renders a confirmation of the existence of different DGK␤ isoforms in vivo rather difficult, but the cloning and functional data presented here and the expression data reported previously by us indicate that DGK␤ transcripts capable of encoding different isoforms are at least transcribed and that these transcripts can be translated in vitro and in vivo (as V5 or GFP fusions) to yield protein products of the expected size.
We focused our analysis on a comparison of the properties of human DGK␤ isoforms differing for the presence of the COOHterminal portion of the gene (that is, the DGK␤ STD and SV3Ј isoforms). In fact, a human DGK␤ EST (dbEST Id: 1371986; GenBank TM accession number AF019352) is annotated in Gen-Bank TM as being differentially expressed in bipolar disorder patients and corresponds to the 3Ј region of human DGK␤ transcripts encoding the isoforms lacking the COOH-terminal region. This EST does not extend sufficiently toward the 5Ј end of the coding sequence to determine the representation of the other alternatively spliced exons and is therefore uninformative toward a possible association between mood disorders and DGK␤ alternative splicing events other than that influencing the COOH-terminal coding sequence. Transcripts encoding different protein isoforms of the same gene may differ in their prevalence and/or tissue distribution. Our TaqMan analysis of the expression of mRNAs encoding the DGK␤ STD and SV3Ј isoforms indicates that the two DGK␤ transcript classes are expressed at substantially different levels (with transcripts encoding the STD variants being present at 10 -100-fold higher levels than those encoding the SV3Ј variants), but shows an essentially superimposable pattern of tissue distribution. Therefore, in normal human tissue transcripts encoding the SV3Ј isoforms are expressed at far lower levels than those encoding the STD isoforms. It would be extremely interesting to confirm the possible differential expression of the SV3Ј isoforms in tissues from bipolar disorder patients. To understand the physiological significance of the existence of DGK␤ isoforms differing at their COOH terminus, we have attempted a functional comparison of the DGK␤ STD and SV3Ј isoforms and investigated their diacylglycerol kinase activities in vitro and their subcellular localization in vivo. The results indicate that, despite showing similar enzymatic activity, the two isoforms may display differences in their subcellular localization and in their response to increased substrate availability. These differences may bear important consequences for the activity of the enzyme in vivo. In fact, it has been shown that DGK activity does not occur ubiquitously within the cell. Experiments have shown that membrane translocation has been described to be a necessary (but not sufficient) step for DGK activation and that DGK activity seems to be localized at membrane sites where DG is produced upon receptor-mediated phospholipase C-␤ activation, rather than acting on DG randomly produced at membrane sites (reviewed in Ref. 4). In line with this view, DGK␤ SV3Ј isoforms may not be active in vivo, despite showing a relatively normal kinase activity in vitro, due to their inability to associate with the cell membrane. It would therefore be interesting to determine whether these isoforms can show diacylglycerol kinase activity in intact cells in vivo. Taken together, our data suggest that an altered balance in the relative level of STD and SV3Ј DGK␤ transcripts (such as may be the case in some mood disorder syndromes) may lead to abnormal DG turnover within neurons in the amygdala, caudate nucleus, and hippocampus, regions that are known to be involved in the control of emotional and cognitive behavior.
Although the functional significance for the existence of multiple DGK␤ isoforms remains unknown, it is likely that the alternatively spliced exons encoding the 7, 12, 25, and 35 amino acid domains will influence regulatory aspects of the protein's kinase activity and/or subcellular localization (as shown here for the 35 amino acid COOH-terminal region) or even a coupling with other intracellular signaling pathways. For instance, a recent study demonstrated that an alternatively spliced COOH-terminal domain of tomato DGK functions as a calmodulin-binding domain, thus providing a coupling between calcium and phospholipid signaling in this plant (22). The identification of multiple DGK␤ isoforms described here may spur interest on the possible existence of multiple isoforms of other DGKs. Furthermore, it sheds a novel light on the functional and regulatory aspects of an important enzyme modulating the levels of DG and phosphatidic acid, two key intracellular messengers, in neuronal cells.