Diacylglycerol Kinase ε, but Not ζ, Selectively Removes Polyunsaturated Diacylglycerol, Inducing Altered Protein Kinase C Distribution in Vivo *

Porcine aortic endothelial cells have previously been shown to contain particularly high basal levels of polyunsaturated diacylglycerol (DAG) together with a very high degree of membrane-associated protein kinase C (PKC), which is largely insensitive to further activation (Pettitt, T. R., Martin, A., Horton, T., Liossis, C., Lord, J. M., and Wakelam, M. J. O. (1997) J. Biol. Chem. 272, 17354–17359). To investigate the possibility that the high polyunsaturated DAG levels were constitutively activating PKC, we transfected porcine aortic endothelial cells with two different forms of human diacylglycerol kinase, ε and ζ. In vitro, the former is specific for polyunsaturated structures, whereas the latter shows no apparent selectivity. Overexpression of DAGKε specifically reduced the level of polyunsaturated DAG in the transfected cells while having little effect on the more saturated structures. It also caused the redistribution of PKCα and ε from the membrane to the cytosol. Overexpression of DAGKζ caused a general reduction in DAG levels but had little effect on PKC distribution. These results for the first time show that DAGKε specifically phosphorylates polyunsaturated DAGin vivo and that in so doing it regulates PKC localization and activity. This provides support for the proposal that it is the polyunsaturated DAGs that function as messengers and convincing evidence for DAGKε being a physiological terminator of DAG second messenger signaling.

stimulation also induces polyunsaturated DAG generation at much longer times, where it correlates with progression hrough the cell cycle (2).
The best-characterized roles of these DAGs are as lipid second messengers, activating both the classical and novel PKC families (3,4). However, although all DAG species can activate PKC to some extent in vitro, we have previously proposed that the polyunsaturated forms are the true messengers under physiological conditions, whereas the more saturated DAGs have little or no activity (5,6).
An intracellular messenger needs a mechanism of generation that can be rapidly activated in response to the appropriate stimulus. However, it also requires an equally rapid and specific mechanism of removal to terminate the signal. A major route for the removal of DAG is through phosphorylation catalyzed by diacylglycerol kinase (DAGK) to form phosphatidic acid (PA) (7). This enzyme exists as at least eight different isoforms that, on the basis of their structure, have been separated into five classes. Each shows a slightly different tissue distribution. Only one, DAGK⑀, a type III enzyme found at high levels in testis but also present in other tissues has been shown to exhibit differential selectivity in vitro, where it preferentially phosphorylates polyunsaturated DAG, particularly sn-1acyl-2-arachidonyl glycerol (8). These observations suggest that DAGK⑀ could be responsible for terminating the polyunsaturated DAG intracellular messenger signal; however, this has not been shown in vivo. Although PA is a putative messenger itself and is required for progression from G 1 into S phase in interleukin-2-stimulated mouse CTLL-2 cells (9), polyunsaturated PA formed by DAGK⑀ is probably inactive as an intracellular signal (5) as it is rapidly utilized to reform phosphoinositide.
To determine if distinct DAGK isoforms have distinct in vivo substrates, we transfected porcine aortic endothelial (PAE) cells with human DAGK⑀ and DAGK, then examined the effect on DAG composition and PKC distribution. Cells transfected with DAGK⑀ have significantly reduced polyunsaturated DAG as compared with the very high levels seen in the parental cells (10), and there is a concomitant redistribution of PKC from the membrane to the cytosol. Transfection with DAGK, a form with no known species selectivity (7,11), did not alter the DAG species profile and had little effect on PKC distribution.

EXPERIMENTAL PROCEDURES
Materials-All solvents were of Analytical or HPLC grade from Fisher. Lipid standards were purchased from Avanti Polar Lipids Inc., Alabaster, AL. Other chemicals were from Sigma-Aldrich. Cell culture reagents were purchased from Life Technologies, Inc. The cDNA for DAGK⑀ was cloned into pcDNA1/Neo (Invitrogen), whereas that for DAGK was cloned into pcDNA3.1/Zeo(ϩ) (Invitrogen); they were generously provided by Drs. M. K. Topham and S. M. Prescott, Eccles Institute of Human Genetics, University of Utah, Salt Lake City, UT. DAGK antibodies raised in rabbit as described previously (11) were also * The work was funded by The Wellcome Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. PAE cells were cultured in Ham's F-12 nutrient mixture containing Glutamax, 10% fetal calf serum, penicillin, and streptomycin and used upon reaching 70 -80% confluency. Most cells were serum-starved for 24 h before use. Transfection was by electroporation. Geneticin was (100 g/ml) used to select clones expressing human DAGK⑀, whereas zeocin (100 g/ml) was used to select clones expressing human DAGK.
Cell Stimulation and Lipid Extraction-Cells were washed twice with phosphate-buffered saline then incubated for 15 min in Ham's F-12 containing 20 mM HEPES, pH 7.4, and 0.1% bovine serum albumin (fraction V; Sigma) at 37°C before stimulation with 10 M sn-1-18 :1n-9,2-lysophosphatidic acid (LPA) where described. Incubations were terminated by aspiration of the media followed by the addition of ice-cold methanol and 1 g of 1,2-12:0/12.0 DAG internal standard. Lipids were extracted, and diradylglycerols were derivatized and analyzed by HPLC as described previously (12,13). Data are expressed as means Ϯ S.D., where n ϭ 3-6.
Inositol Phosphate Accumulation Assay-Cells were labeled with myo-[2-3 H]inositol (0.5 Ci/ml) in inositol-free medium for 48 h and stimulated in the presence of 20 mM LiCl with vehicle or LPA (10 M), and the inositol phosphates were analyzed as described previously (14).
Western Blotting of Diacylglycerol Kinase and Protein Kinase C Isozymes-PAE cells stimulated with LPA (10 M) for 20 s and 5 min where described were washed in ice-cold phosphate-buffered saline and harvested in 20 mM HEPES, pH 7.4, containing 1 mM phenylmethylsulfonyl fluoride and 20 g/ml leupeptin. The samples were frozen at Ϫ70°C overnight, thawed, and sonicated for 20 s in a bath sonicator, and membranes and cytosol were separated by centrifugation (300,000 ϫ g, 20 min). The membranes were washed once. The cytosolic proteins were concentrated by precipitation with 12.5% trichloroacetic acid (4°C, 1 h) and centrifugation (13,000 ϫ g, 10 min), and the pellet was washed with 200 l of ice-cold acetone. The cytosolic and membrane proteins were solubilized by boiling in 75 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 4 M urea, 5% ␤-mercaptoethanol, bromphenol blue for 5 min, then cell fraction equivalents were separated by electrophoresis on 7.5% SDS-polyacrylamide minigels. After transfer onto polyvinylidene difluoride membranes (Immobilon-P; Millipore), the PKC isozymes were probed using mouse monoclonal antibodies to PKC␣ (1 in 2500 dilution) and PKC⑀ (1 in 2500) and detected using enhanced chemiluminescence. DAGK isozymes were probed with rabbit polyclonal antibodies to DAGK⑀ (1 in 2000) and DAGK (1 in 1000).
PKC Immunoprecipitation-Cells grown in 10-cm dishes were washed twice with phosphate-buffered saline, lysed for 15 min at 4°C with 1 ml of ice-cold hypertonic lysis buffer (20 mM Tris, pH 7.6, 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol, 0.5% Nonidet P-40, protease inhibitor mixture; (Roche Molecular Biochemicals), scraped into Eppendorf tubes, sonicated for 5 min in a sonicating bath cooled with ice, centrifuged (13000 ϫ g, 5 min), and the cleared supernatant was transferred to clean tubes. Each sample was mixed for 1 h at 4°C with 5 l of mouse monoclonal PKC antibody followed by a further 30 min with 10 l of goat anti-mouse IgG antibody (Sigma). Each sample was then split into two equal fractions and mixed with 50 l of 20% Protein G-Sepharose beads (Amersham Pharmacia Biotech) for 1 h at 4°C. After centrifugation (13,000 ϫ g, 5 min), the immunoprecipitate pellet was washed four times with 500 l of kinase buffer (50 mM HEPES, pH 7.2, 100 mM NaCl, 75 mM KCl, 20 mM ␤-glycerophosphate, 10 mM MgCl 2 , 2.5 mM CaCl 2 , 1 mM sodium orthovanadate, 1 mM dithiothreitol) and then resuspended in 20 l of kinase buffer.
Protein Kinase C Assay-To the immunoprecipitate was added 5 l of PKC substrate (either histone type IIIS or neurogranin fragment, amino acids 28 -43; 1 mg/ml; Sigma) in kinase buffer, 10 l of lipid micelle mix (500 M phosphatidylserine, 100 M mixed DAG, 0.5% Triton X-100 in kinase buffer), and 5 l of kinase buffer (or 5 M tetradecanoyl phorbol acetate in kinase buffer for the positive controls). 10 l of 100 M ATP containing 0.5 Ci of [ 32 P]ATP was added to start the reaction. After incubation (1 h, 30°C), the reaction was stopped with 20 l of 5% acetic acid. The phosphorylated substrate was separated by electrophoresis on 17.5% SDS-polyacrylamide gels for histone, whereas with the neurogranin fragment each sample was spotted onto P81 cation exchange paper (Whatman), air dried, washed 4 ϫ 10 min with 500 ml of 0.5% phosphoric acid and washed once with 95% ethanol. The 32 P-phosphorylated substrate was detected and quantified using a Molecular Dynamics PhosphorImager.

Expression of DAGK⑀ and DAGK in parental PAE cells and DAGK-transfected clones was examined by Western blotting.
Although the parental cells express both enzymes, the clones showed enhanced levels with 2-4-fold increases in DAGK⑀ (64 kDa) and 5-10-fold increases in DAGK (115 kDa; Fig. 1). Preparation of cytosolic and membrane fractions revealed that these enzymes were restricted to the membrane fraction both in resting and LPA-stimulated cells (data not shown). The identities of the immunoreactive bands at approximately 90 and 100 kDa in the DAGK blot remain unclear but may be cleavage products of the parent 115-kDa protein.
The parental cells grew faster than the DAGK clones with doubling times of 21, 26, and 36 h (based on cell numbers after different culture times) for parental, DAGK⑀ clone 7, and DAGK clone 7 cells, respectively. The growth of both the parental and transfected PAE cells was largely insensitive to serum starvation for 48 h (starved from approximately 30% confluency), with only a 20% increase in cell-doubling time. The DAGK clones have a greatly increased propensity to align into ordered swirls when fully confluent (Fig. 2). The degree of order varied between clones but appeared to correlate with the level of DAGK expression; those expressing more enzyme showed greater alignment. The DAGK clones always showed this arrangement at confluency. The DAGK⑀ clones were more variable, sometimes showing ordered alignment into swirls and sometimes not, whereas the parental cells only occasionally showed a partial ordering. This variability with the parental and DAGK⑀ clones may relate to differences in the relative concentrations of cytokines and growth factors in different batches of cell culture serum and how these affect the DAG and/or PA signaling pathways modulated by DAGK.
PAE cells contain high levels of polyunsaturated DAG under resting conditions, and these remain largely insensitive to further stimulation (10). If DAGK⑀ had the same specificity in vivo as in vitro, overexpression of active enzyme would be expected to reduce these levels. Thus, total lipid extracts were prepared, derivatized with 3,5-dinitrobenzoyl chloride, and separated by HPLC to resolve the different DAG molecular species. Fig. 3 shows a comparison of DAG profiles from one DAGK⑀ clone and the parental cells, both grown in 10% serum. Polyunsaturated species whose levels are greatly reduced in the clone, including sn-1-stearoyl,2-eicosatrienoyl glycerol (18:0/20:3n-9; peak 16) and sn-1-stearoyl,2-arachidonyl glycerol (18:0/20:4n-6; peak 12), are marked with an asterisk. Other DAGK⑀ clones gave similar results with greatly reduced polyunsaturated DAG levels but little change in more saturated species, as summarized in Table I. Overexpression of DAGK⑀ reduced total DAG mass by approximately 40% (3.4 Ϯ 0.5 compared with 5.4 Ϯ 0.8 nmol/10 7 cells). This is almost entirely due to the loss of polyunsaturated structures, particularly 18:0/20:3n-9 and 18:0/20: 4n-6, which together account for approximately 1.6nmol/10 7 cells in the parental cells but only 0.1nmol/10 7 cells in the DAGK⑀ clones. Serum starvation for 24 h had little effect on the DAG profiles but did reduce total mass by about 65% (from 3.9 Ϯ 0.5 to 1.4 Ϯ 0.2 nmol/10 7 cells in DAGK⑀ clone 8). The DAGK clones showed very little change in the DAG species profiles relative to the parental cells (data not shown). However, the basal DAG levels were approximately halved relative to the parental cells when grown either in 10% serum (2.7 Ϯ 0.5 nmol/10 7 cells compared with 5.4 Ϯ 0.8 nmol/10 7 cells) or serum-free (1.8 Ϯ 0.4 nmol/10 7 cells compared with 3.6 Ϯ 0.5 nmol/10 7 cells).
We next investigated whether the reduced basal polyunsaturated DAG levels in the DAGK⑀ clones had reestablished the potential for agonist-induced activation of phosphoinositide PLC as measured by rapid polyunsaturated DAG production and inositol 1,4,5-trisphosphate release. LPA stimulation for 25 s, 1 min, or 5 min caused no consistent inositol phosphate generation and had no significant effect on polyunsaturated DAG levels (data not shown). However, after 5 min, LPA did cause a small elevation in more saturated species, particularly 18:1n-9/18:1n-9, which increased from 247 pmol/10 7 cells to 455 pmol/10 7 cells. These were probably generated through the phospholipase D/phosphatidic acid phosphohydrolase pathway since we previously reported that butan-1-ol completely blocked the formation of these structures in LPA-stimulated PAE cells (10). Taken together, the data indicate that LPA is unable to stimulate an increased phosphoinositide PLC activity in these cells, even if basal polyunsaturated DAG levels are greatly reduced.
We have shown previously that essentially all the PKC␣ and approximately 50% of the PKC⑀ was associated with the membrane fraction in resting PAE cells (10). We hypothesized that this was due to the high basal levels of polyunsaturated DAG, in effect, constitutively activating the enzyme and driving it onto the membrane. Thus by reducing the polyunsaturated DAG levels, more PKC should be found in the cytosol. Therefore we compared the PKC distribution in the DAGK⑀ clones with that in the parental cells, both cultured for the last 24 h under either serum-free or 10% serum conditions. When cultured in serum-free conditions, the clones with their reduced levels of polyunsaturated DAG, but not the parental cells, showed PKC redistribution to the cytosol (Fig. 4). PKC␣ was approximately 95% membrane associated in the parentals compared with about 50% in the clones, whereas for PKC⑀ the membrane association was approximately 60% compared with about 40%. For the parental cells and the clones cultured in 10% serum, nearly all the PKC␣ was on the membrane, whereas for PKC⑀ about 60% was on the membrane. Serumstarved parental cells show essentially the same PKC distribution as that seen when grown in 10% serum. Stimulation with LPA for 20 s and 5 min had no obvious effect on PKC distribution (data not shown). Total PKC␣ expression in the DAGK clones was similar to that seen in the parental cells; however, PKC⑀ expression was elevated in the DAGK⑀ clones and probably also the DAGK clones, although there was some variability between individual clones. The DAGK clones showed a PKC distribution almost identical to that in the parental cells under both serum conditions (Fig. 5).
To investigate if DAGK transfection altered PKC kinase activities, PKC␣ and ⑀ were immunoprecipitated from total lysates, and their abilities to phosphorylate histone or a PKC substrate peptide (amino acids 28 to 43 from neurogranin) were examined in the presence or absence of the phorbol ester, tetradecanoyl phorbol acetate (TPA). Although somewhat variable, PKC␣ activity was generally similar in the parental and

FIG. 3. Basal DAG species profiles from parental cells and a DAGK⑀ clone.
Cells were grown in 10% serum. Lipids were extracted, dinitrobenzoyl-derivatized, and separated into molecular species by reverse-phase HPLC. Other DAGK⑀ clones give a similar DAG profile. See Table I for peak identities. *, peaks greatly reduced by overexpression of DAGK⑀. DAGK clones, both, under basal and phorbol ester-stimulated conditions (Fig. 6A). For PKC⑀, basal activity was highest in the DAGK clones and lowest in the parental cells (Fig. 6B).
Although not statistically significant, the data also suggested a small increase in stimulatable PKC⑀ activity, particularly in the DAGK clones. Since 500 nM TPA is believed to maximally activate all DAG-sensitive PKCs in vitro, the activity seen in the presence of phorbol ester is taken to correlate with total PKC content in the immunoprecipitates. Thus, in Fig. 6, C and D, the basal PKC activity is expressed as a percentage of maximal activity (ϩ TPA). This gives an approximation to the relative PKC specific activities under resting conditions. DAGK overexpression had essentially no effect on basal PKC␣ activ-ity, which remained at approximately 50% maximal. In contrast, PKC⑀ activity was substantially increased, particularly in the DAGK clone, where the protein kinase was almost fully activated under basal conditions. DISCUSSION DAGK is a ubiquitous enzyme with roles in phospholipid synthesis and cell signaling. Most studies to date have reported the purification, structural identification and, in vitro characterization of novel isoforms. The data presented here provide the first demonstration that DAGK can modulate DAG levels in cells. In addition, the results show the selective modulation of DAG species levels by a particular isoform, DAGK⑀, but not by  Fig. 3, which are significantly reduced in DAGK-transfected PAE cells (p Ͻ 0.01).

FIG. 4. Effect of DAGK⑀ overexpression on PKC distribution.
Cytosol and membrane fractions were prepared from cells grown for the last 24 h either in serum-free or 10% serum conditions, separated on 7.5% SDS-polyacrylamide gels, and analyzed by Western blotting. Data are representative of several independent experiments.

FIG. 5. Effect of DAGK overexpression on PKC distribution.
Cytosol and membrane fractions were prepared from cells grown for the last 24 h either in serum-free or 10% serum conditions, separated on 7.5% SDS-polyacrylamide gels, and analyzed by Western blotting.
DAGK. This demonstrates that the substrate specificity suggested by in vitro studies (8,11) does indeed occur in vivo. The selective removal of polyunsaturated DAG by DAGK⑀ is accompanied by regulation of the cellular distribution of PKC␣ and ⑀. This provides support for the proposition that polyunsaturated DAG, presumably derived through a phospholipase C pathway, is the true signaling form of this lipid, whereas the more saturated species derived through a phospholipase D/phosphatidic acid phosphohydrolase pathway are largely inactive.
DAGK⑀ clones grown in 10% serum showed a similar PKC distribution to the parental cells; however, when grown for 24 h in serum-free media, the PKC redistributed to the cytosol (Fig.  4). Growth in the absence of serum greatly reduced the total DAG mass (1.4 nmol/10 7 cells as compared with 3.9 nmol/10 7 cells) but had no effect upon the DAG species profiles. Thus the reduction in polyunsaturated DAG mass must be sufficient to cause PKC redistribution. In DAGK⑀ clone 8, 16:0/20:4n-6, 18:0/20:4n-6, and 18:0/20:3n-9 together represented 71 pmol/ 10 7 cells under serum-free conditions as compared with 252 pmol/10 7 cells when grown in 10% serum. In the parental cells, although total DAG mass dropped after a 24-h serum-free incubation (3.6 nmol/10 7 cells as compared with 5.4 nmol/10 7 cells), the total polyunsaturated DAG mass remained greater at 917 pmol/10 7 cells than that seen in the DAGK⑀ clones grown with 10% serum. In these parental cells the PKC distribution under serum-free conditions remained similar to that seen in 10% serum, with most enzyme at the membrane (Fig.  4). Thus the key regulator of PKC redistribution is the absolute mass of polyunsaturated DAG in a cell rather than its proportion relative to more saturated species. However, it should also be noted that different subcellular localizations might contribute to the differential effect of the two DAGK forms on PKC distribution. If DAGK and the PKCs are localized to different compartments, then DAGK modulation of local DAG levels may not affect the PKC found elsewhere in the cell. In contrast, co-localization would be expected to have a larger effect on PKC activation.
The activity data for PKC␣ (Fig. 6, A and C) support the idea that although different amounts of PKC may be at the membrane under basal conditions, to maintain homeostasis the cell only needs a certain level of kinase activity and, thus, modulates it accordingly, resulting in similar activities for the parental and DAGK-transfected cells. The activity in the presence of phorbol ester indicates that the three cell lines have similar total levels of PKC␣. In contrast, the data for PKC⑀ suggest that there is more enzyme in the DAGK clones (Fig. 6B). This agrees with the Western blot data where more PKC⑀ was detected in the transfected cells (Figs. 4 and 5). The increased kinase activity in the immunoprecipitates under basal conditions may partly reflect these elevated enzyme levels, although the PKC⑀-specific activity appeared to be higher than in the parental cells, particularly for the DAGK clones (Fig. 6D). However, since the enzyme had to be immunoprecipitated and then assayed in vitro, introducing a variety of nonphysiological changes, we would argue that the PKC redistribution data are a more reliable measure of PKC activation in vivo, although not necessarily of kinase activity. In keeping with this, the increase in PKC⑀ protein observed in the DAGK transfectants presumably reflects a reduction in the membrane polyunsaturated DAG concentration reducing the down-regulation of enzyme levels.
The molecular size of DAGK calculated from its nucleotide sequence is 103.9 kDa, but as reported previously, it runs slightly slower than this on SDS-polyacrylamide gels (11). COS-7 cells transfected with the DAGK gene expressed a doublet of 114 and 117 kDa as did nontransfected human glioblastoma-derived A-172 cells (11); however, both parental and transfected PAE cells expressed a band of approximately 115 kDa together with variable amounts of possible cleavage products at approximately 90 and 100 kDa (Fig. 1). Tissuespecific splice variants may be common since a 130-kDa form has been found in human skeletal tissue, although its properties appear identical to those of the 114-kDa form (15).
The higher DAGK protein expression achieved with the DAGK clones (5-10-fold above parental) as compared with the DAGK⑀ clones (2-4 fold) suggests that the latter isozyme is probably more toxic to the cell, causing cell death when present at slightly elevated levels. This is a common problem with overexpressing lipases or other lipid-metabolizing enzymes, since only small changes can cause major membrane damage and resultant cell death. This sensitivity is likely to be greatest for enzymes involved in lipid signaling. To maintain membrane integrity and capacity for signaling, PIP 2 , the substrate for PLC-catalyzed polyunsaturated DAG messenger generation, must be rapidly resynthesized. The first step, catalyzed by DAGK, terminates the polyunsaturated DAG signal by forming polyunsaturated PA. This is the preferred substrate for at least one form of CDP-DAG synthase (16), generating CDP-DAG (or CMP-PA), which is then utilized by phosphatidylinositol synthase for phosphoinositide synthesis (17). Sequential phosphorylation then regenerates PIP 2 . Both DAGK⑀ (18) and rat brain CDP-DAG synthase (16) are strongly inhibited by PIP 2 in vitro, providing a potential negative-feedback mechanism to regulate phosphoinositide generation. Overexpression of DAGK⑀ would alter this balance since enhanced removal of polyunsaturated DAG would reduce the PKC activation, which negatively regulates PLC activity (13), potentially leading to sustained PIP 2 hydrolysis. This would deplete PIP 2 levels, releasing PIP 2 inhibition of CDP-DAG synthase and DAGK⑀ as the cell attempts to replenish phosphoinositide. As a result the cell may become locked into an energy-draining futile cycle, which if uncontrolled, would ultimately kill it.
DAGK overexpression, particularly DAGK, induced a no- FIG. 6. Effect of DAGK overexpression on PKC activity. PKC␣ and PKC⑀ immunoprecipitates from total cell lysates were assayed for kinase activity in the presence and absence of TPA. In panels A and B data are expressed as arbitrary PhosphorImager densiometric units (volume)/500 g of total lysate protein; mean Ϯ S.D. (n ϭ 3). Since 500 nM TPA is believed to maximally activate all DAG-sensitive PKCs in vitro, the activity seen under these conditions is taken to correlate with total amount of PKC present in the immunoprecipitates. Thus, in panels C and D, the basal PKC activity is expressed as a percentage of maximal activity obtained in the presence of TPA. ticeable change in cell morphology at confluency (Fig. 2) possibly through effects on membrane fluidity caused by the reduction in total diradylglycerol levels. However, the observation that DAGK is found at high levels within the nucleus and has a myristoylated alanine-rich C kinase substrate homology region that can act as a nuclear localization sequence (15,19,20) suggests that its effects maybe exerted primarily within this organelle. Phosphorylation of this domain by PKC␣ or PKC␥ has been shown to cause DAGK translocation out of the nucleus and away from its presumed substrate, nuclear DAG (20). Blocking this translocation by PKC down-regulation or overexpression of DAGK in the nucleus increased the cell cycle time (20). A nuclear localization would fit with our observations that DAGK is associated almost entirely with the membrane fraction from PAE cell lysates, since this would contain ruptured nuclei; also, that the cell doubling time was longest for the DAGK clones. Recent work with mouse DAGK, which shows a 95% homology with the human form, found that expression is regulated both temporally and spatially during embryonic development and correlates with the development of sensory neurones and regions undergoing apoptosis (21), suggesting a role for this enzyme in gene regulation. Thus its overexpression may lead to an abnormal expression of other genes and hence morphological change, possibly of the type observed here. Overexpression of DAGK⑀ may partially mimic the effects of DAGK, resulting in the more variable morphology seen at confluency in the DAGK⑀ clones.
In conclusion DAGK⑀, but not DAGK, modulates PKC␣ and ⑀ activation by specifically regulating the endogenous levels of polyunsaturated-signaling DAG and, by generating polyunsaturated PA, initiates the resynthesis of the parent phosphoinositide.