Gene Cloning and Characterization of CDP-diacylglycerol Synthase from Rat Brain*

A cDNA encoded a 462-amino acid protein, which showed CDP-diacylglycerol synthase (CDS) activity was cloned for the first time as the vertebrate enzyme mol- ecule from rat brain cDNA library. The deduced molecular mass of this rat CDS was 53 kDa, and putative primary structure included several possible membrane-spanning regions. At the amino acid sequence level, rat CDS shared 55.5%, 31.7%, and 20.9% identity with al-ready known Drosophila , Saccharomyces cerevisiae , and Escherichia coli CDS, respectively. This rat CDS pre-ferred 1-stearoyl-2-arachidonoyl phosphatidic acid as a substrate, and its activity was strongly inhibited by phosphatidylglycerol 4,5-bisphosphate. By immunoblot- ting analysis of COS cells overexpressed with the epitope-tagged for rat CDS, a 60-kDa band was detected. By epitope-tag immunocytochemistry, the CDS protein was mainly localized in close association with the mem- brane of the endoplasmic reticulum of the transfected cells. The intense mRNA expression of CDS was local- ized in the cerebellar Purkinje cells, the pineal body, and the inner segment of photoreceptor cells. Addition- ally, very intense expression was detected in postmitotic spermatocytes and spermatids. -glycero-3-phos-phate (di-C18:0-PA), 1,2-dioleoyl- sn -glycero-3-phosphate (di-C18:1-PA), and 1-stearoyl-2-arachidonoyl- sn -glycero-3-phosphate (C18:0/C20:4-PA) as long chain PAs. To analyze the sensitivity of CDS activity to phosphoinositide, PI, PIP (phosphatidylinositol 4-monophosphate), and PIP 2 (Sigma) were added to the reaction mixture at concentration of 2, 5, 10, and 20 mol%. The reaction was continued for 10 min at 30 °C and was stopped with 100 (cid:109) l of 1 N HCl. A two-phase system was created by adding 250 (cid:109) l of chloroform/methanol (1:1, v/v), and the mixtures were thoroughly vortexed and then centrifuged. The lower organic phase was washed with 100 (cid:109) l of chloroform, 1 N HCl (1:1, v/v) and was analyzed by thin layer chromatography which was developed in a solution con- taining chloroform/methanol/water/acetic acid (50:28:8:4, v/v/v/v). 1,2-Diacyl- sn -glycero-3-diphosphocytidine (CDP-DAG) from egg yolk leci- thin (Sigma) was used as a marker. The bands of CDP-DAG detected by autoradiography was scraped with a sharp spatula and collected for liquid scintillation counting.

Phosphoinositide cycle mediates one of the intracellular signal transduction pathways in eukaryotic cells and produces a class of second messengers that are involved in cell growth (1,2), differentiation (3,4), the action of hormones and neurotransmitters (5), and sensory perception (6 -8). Triggering of the cell surface receptors, such as G-protein-coupled receptors and tyrosine kinase receptors, initiates the cycle by activating phospholipase C (PLC), 1 resulting in the hydrolysis of phos-phatidylinositol 4,5-bisphosphate (PIP 2 ) into two second messengers, diacylglycerol (DAG) and inositol 1,4,5-triphosphate (IP 3 ). DAG is subsequently phosphorylated by DAG kinase to synthesize phosphatidic acid (PA), a presumed novel messenger (9 -11). In this cycle, CDP-diacylglycerol synthase (CDS) (CTP-phosphatidate cytidylyltransferase) catalyzes the converting process from PA to CDP-DAG, the precursor to phosphatidylinositol (PI), which is phosphorylated to synthesize PIP 2 eventually.
The importance of CDS in the signal transduction has been strengthened in a recent study by Wu et al. (12), showing that overexpression of a photoreceptor cell-specific isoform of CDS in Drosophila increases the amplitude of the light response. Its mutants (cds mutants) cannot sustain a light-activated current and undergo light-dependent retinal degeneration, which can be suppressed by a mutation in PLC. The difference in phototransduction mechanism between vertebrate and invertebrate is well known, such as the primary second messenger role of cGMP in vertebrate photoreceptors versus that of IP 3 in invertebrate. However, recent investigations have shown unique distribution of phosphoinositide-specific PLC (13)(14)(15), protein kinase C (16,17), and IP 3 receptor (18) in vertebrate rod outer segment, and that light may enhance the activity of PLC (19 -22) and protein kinase C (23), and phosphoinositide synthesis (24). In addition, it has been reported that cytoplasmic Ca 2ϩ concentration mediates light adaptation in vertebrate photoreceptors (25,26). Thus it is suggested that in vertebrates as well as invertebrates the phosphoinositide cycle may play a role in the phototransduction signaling. It is therefore possible that CDS is also important in the signal transduction mechanism of vertebrate retina and neural cells.
As a step toward understanding the possible functional significance of this enzyme in vertebrate cellular signaling, the present study was attempted to perform molecular cloning of a CDS molecule from rat brain and to clarify its enzymatic feature and tissue and cell localization. Our result shows that a newly identified CDS prefers arachidonate-containing PA as a substrate, suggesting strongly a role for CDS in the phosphoinositide synthesis. It is also shown that the activity of CDS is inhibited by PIP 2 , suggesting that polyphosphoinositides regulate their own synthesis through CDS activity. In addition, we show the detailed localization at mRNA level in the retina and brain, but unexpectedly the highest expression of its mRNA is detected in the testis.

EXPERIMENTAL PROCEDURES
Polymerase Chain Reaction (PCR)-Total RNA was extracted from adult Wistar rat brain by acid guanidinium thiocyanate-phenol-chloroform extraction (27). Poly(A) ϩ RNA was isolated by chromatography on an oligo(dT)-cellulose column. First-strand cDNA was prepared using First-Strand cDNA Synthesis kit (Pharmacia Biotech Inc.).
Primers used for PCR were composed of two degenerate oligonucleotides, which were made according to the amino acid sequences of conserved regions between Drosophila CDS (12) and E. coli CDS (28): the regions corresponded to the amino acid sequences TW(E/Q)GFIGG (aa 276 -283 for 5Ј primer) and IPGHGGI(M/L) (aa 391-398 for 3Ј primer) (amino acid numbers represent those of Drosophila CDS). The sequences for the primers were designed according to the mammalian codon usage: GGAATTCAC(C/A)TGG(G/C)A(G/A)GG(C/A)TT(C/T)AT-(C/T)GG(C/A/G)GG for 5Ј primer and CGGGATCCA(T/G)(G/A)AT(G/ T)CC(G/T)CC(G/A)TG(G/T)CC(G/A)GG(G/A)AT for 3Ј primer. The 5Ј ends of 5Ј and 3Ј primers were designed to contain an EcoRI site and a BamHI site, respectively, for subsequent cleavage of subcloned cDNA fragments. PCR amplification was performed by using the first-strand cDNA and Ampli-Taq DNA polymerase according to following schedule: 94°C for 30 s, 55°C for 45 s, and 72°C for 2 min, for 30 cycles, followed by further incubation at 72°C for 7 min. Subcloned cDNA fragments were sequenced by the model 373 DNA autosequencer (Applied Biosystems).
cDNA Cloning-A cDNA library of adult rat brain on postnatal day 49 was constructed as described previously (29). Clones (4 ϫ 10 6 ) derived from the cDNA library were screened by hybridization in the presence of 50% formamide at 42°C with the subcloned PCR fragments labeled with [ 32 P]dCTP. Washing conditions were carried out in 0.1 ϫ SSC (SSC: 0.15 M NaCl, 0.015 M trisodium citrate), 0.1% sodium dodecyl sulfate (SDS) at 42°C. Among five clones isolated, one clone (pCDS4) containing the longest cDNA insert (3.5 kb) was selected for further sequence analysis on both strands as described above.
Immunoblotting and Immunocytochemistry by Epitope-tagging-An epitope-tag composed of eight amino acids (FLAG marker peptide, Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys; Eastman Kodak Corp.) was fused to the newly identified molecule by cloning the 24 base pairs of FLAG coding sequence up-or downstream to the coding region of the novel cDNA. The FLAG epitope-tagged molecules were expressed in COS-7 cells using the expression vector, pSRE, by the DEAE-dextran method (30 -32). Total lysate of the overexpressed COS-7 cells was prepared as described above. Total lysate boiled for 5 min in Laemmli's sample buffer was subjected to 10% SDS-polyacrylamide gel electrophoresis (35). The separated proteins were then electrophoretically transferred to a nitrocellulose membrane (0.45-m pore size). After blocking the nonspecific binding sites in the 5% skim milk (w/v) in phosphatebuffered saline, 0.01% Tween 20, the membrane was incubated for 1 h at room temperature with antibody against FLAG and then treated with peroxidase-conjugated anti-mouse IgG antibody for 1 h. The immunoreactive bands were detected using a chemiluminescence detection kit (ECL Western blotting detection kit, Amersham).
Transfected cells were fixed with 4% paraformaldehyde, 0.1 M so-dium phosphate buffer (pH 7.2), 0.2% Triton X-100, or fixed with the same fixative without the detergent and freeze-thawed after immersion in 40% sucrose. The cells were incubated with the anti-FLAG antibody (Kodak) and sites of the antigen-antibody reaction were visualized using the avidin-biotinylated peroxidase complex (ABC) system (Vector Laboratories) with 3,3Ј-diaminobenzidine tetrahydrochloride as a substrate. Some of the freeze-thawed specimens, after immunoreaction, were postfixed with 1% OsO 4 , treated with 0.5% uranyl acetate, and embedded in Epon 812. Ultrathin sections were examined under electron microscope. Northern Blot Analysis-Total RNAs were extracted from adult rat brains, eyeballs, and several other tissues as described above. Each of the total RNA samples (30 g/lane) was denatured with formamide and size-separated by formaline/agarose gel electrophoresis. The RNAs were transferred and fixed to nylon membranes (Nytran, Schleicher & Schuell) and hybridized with a probe corresponding to the 3Ј non-coding sequences (nucleotides 1423-1875) labeled with [ 32 P]dCTP. Conditions for hybridization and washing were performed as described previously (29).
In Situ Hybridization Histochemistry-Fresh frozen blocks of adult rat brains, eyeballs, and testes were sectioned at 10 or 25 m thickness on a cryostat. The sections were mounted on silane-coated glass slides, fixed with 4% paraformaldehyde, 0.1 M sodium phosphate buffer (pH 7.2), pretreated and hybridized as described previously (29). Two kinds of probes were used in the experiment: a cDNA probe corresponding to nucleotides 526 -1311 labeled with [␣- 35

RESULTS
The nucleotide and deduced amino acid sequences of the composite CDS cDNA are presented in Fig. 1. The putative initiation codon was preceded by in-frame stop codons. The predicted open reading frame encoded a protein of 462 amino acids with the deduced molecular mass of 53 kDa. The deduced amino acid sequence of this rat CDS shared 55.5%, 31.7% and 20.9% identity to that of Drosophila (12), Saccharomyces cerevisiae (36), and Escherichia coli (28), respectively (Fig. 2a). Near the carboxyl terminus, domains sharing more than 80% identity to each other were contained in all molecules. Several protein kinase-recognition sequence motifs were identified such as those recognized by myosin-I heavy chain kinase (aa 345-348), multifunctional Ca 2ϩ /calmodulin-dependent protein kinase II (aa 52-57), cGMP-dependent protein kinase (aa 312-318), proline-dependent protein kinase (aa 265-268 and 271-274), casein kinase I (aa 76 -80), and glycogen synthase kinase 3 (aa 312-318) (37). Hydrophobicity analysis by the method of Kyte and Doolittle (38) revealed that this novel molecule, like the other three CDSs, was very hydrophobic and appeared to contain several possible membrane-spanning regions, as judged by the fact that at least 19 sequential residues have an average hydrophilic value of greater than ϩ1.6 (Fig. 2b). The amino terminus of this molecule was hydrophilic, similar to those of CDSs of Drosophila and S. cerevisiae.
By measurement of the CDS activity of this novel molecule, the total lysate from COS-7 cells transfected with the full length of the present cDNA showed about 5-7 times higher activity toward PAs than the control total lysate (e.g. 184 Ϯ 9.3; 28 Ϯ 1.5 pmol/min/mg total protein under the assay condition at 200 M C18:0/C20:4-PA). The enzymatic activity of this molecule was the highest toward 1-stearoyl-2-arachidonoyl-PA (C18:0/20:4-PA) among the others, whereas little or no activity was detected toward PAs containing saturated fatty acyl groups in both of the sn-1 and -2 positions (di-C12:0-PA, di-C18:0-PA) (Fig. 3). From double-reciprocal (Lineweaver-Burk) plots, values for the apparent K m toward C18:0/C20:4-PA, di-C18:1-PA, and PA from egg yolk lecithin were 102, 114, and 138 M, respectively; and for the apparent V max were 268, 259, and 198 pmol/min/mg total protein, respectively. Values for apparent V max /K m (specificity constant) calculated from Eadie-Hofstee plots were 2.51, 2.28, and 1.48, respectively. When the utilization of CTP and dCTP for the activation of phospholipid intermediates was compared, dCTP was incorporated into phospholipid precursors at almost the same rate as CTP. In the analysis of inhibitory effects of PI, PIP, and PIP 2 on the CDS activity of the present molecule, the activity decreased by 50% at 2 mol% PIP 2 and by 80% at 10 mol% and more (Fig. 4), under the assay condition at 1000 M C18:0/C20:4-PA (the present molecule showed the maximum activity at 200 -1000 M C18: 0/C20:4-PA). In contrast, the activity decreased by at most 40% at 5-20 mol% PIP. Addition of PI at various concentrations from 2 to 20 mol% produced little effects on CDS activity. Similar results were obtained in the experiments for C18:0/ C20:4-PA and di-C18:1-PA at 100 or 500 M (data not shown).
In the immunoblotting of epitope (FLAG)-tagged CDS with the antibody against the FLAG tag, a single immunoreaction band was observed at size of 60 kDa (Fig. 5).
In the immunocytochemistry (Fig. 6a) of the transfected COS-7 cells with the antibody against FLAG tag, immunoreactive cells accounted for approximately 20 -40% of the total cell population and they were randomly dispersed in each culture dish. The immnoreactivity was localized widely throughout the cells in forms of delicate networks composed of fine dots with a tendency of condensation in one pole of juxtanuclear cytoplasm, and they were also deposited along the nuclear rim. In immunoelectron microscopy (Fig. 6b), the immunoreactivity was detected along the membranes of endoplasmic reticulum, vesicles and vacuoles, and nuclear envelopes. No immunoreactivity was detected in any cells when the transfection was made of the cDNA without the FLAG tag.
In Northern blot analysis (Fig. 7) of adult rat on postnatal day 49, the hybridization bands were detected very strongly in testis at sizes of 3.5 and 2.2 kb, suggesting a possibility of alternative splicing. The hybridization band was strongly detected in eyeball and brain at a size of 3.5 kb. Among other tissues, a weak band of 3.5 kb was detected in kidney, small intestine, and placenta, whereas a faint band of 3.5 kb was seen in thymus and lung after long exposure. Any hybridization bands could not be detected in heart, liver, and ovary. By in situ hybridization histochemical analysis of the adult brain (Fig. 8), the most intense expression signals were detected in the cerebellar Purkinje cells and intense expression was found in the pineal body. Moderate to low expression was seen in layers II-VI of the cerebral cortex, in the hippocampal pyramidal cell layer and subiculum, and the olfactory mitral cells. Low expression was seen in almost all neurons in the fore-, mid-and hindbrains with much lower expression in the caudate putamen. Low expression was also seen in the choroid plexuses. No significant expression was detected in the white matter including the corpus callosum and cerebellar medulla or in the glia limitans. In the retina, positive expression signals were confined to the inner segment of the photoreceptor cells (Fig. 9).
In the adult testis, the hybridization signals were densely deposited in most of the adluminal compartment of the seminiferous tubule composed of postmitotic spermatocytes and spermatids, while no significant signals were detected in the apical compartment composed of spermatozoa or in the basal compartment composed of proliferating spermatogonia. No significant signals were found in the interstitial cell clusters (Fig. 10).
In the control experiment in which several random sections of brain, retina, and testis were hybridized with the labeled oligonucleotide probe in the presence of 100-fold excess amounts of the unlabeled probe, no significant expression signals were detected in any regions of the tissue sections. DISCUSSION We report here for the first time the molecular cloning of vertebrate CDS and clarify its primary structure, enzymatic features, and localization in detail. The primary structure of this rat CDS shared 55.5%, 31.7%, and 20.9% identity to that of Drosophila (12), S. cerevisiae (36), and E. coli (28), respectively (Fig. 2a). The high hydrophobicity and the inclusion of several putative membrane-spanning regions in the primary structure of the present molecule and the localization associated with the endoplasmic reticulum and nuclear envelopes are well consistent with available biochemical data of the mammalian CDS so far reported (34, 39 -41). The deduced molecular size of 53 kDa is smaller than the size of 60 kDa detected by epitope-tagged immunoblotting. This discrepancy may be ascribed to the posttranslational modification of CDS in vivo. From the sequence, it is clear that there are several potential serine/threonine phosphorylation sites, suggesting the possibility that phosphorylation by serine/threonine kinases may be involved in regulation of this enzyme activity.
The present study shows the substrate specificity of rat CDS toward 1-stearoyl-2-arachidonoyl PA (C18:0/C20:4-PA) among several PAs (Fig. 3). In mammalian cells, CDP-DAG is used to produce phosphatidylinositol, phosphatidylglycerol, and cardiolipin, and CDP-DAG is therefore an important regulatory branching point in the phospholipid metabolism. Since the PI predominantly consist of the 1-stearoyl-2-arachidonoyl species (42), the present substrate specificity of rat CDS suggests strongly that this enzyme molecule selectively participates in the phosphoinositide cycle, but not in the synthesis of phosphatidylglycerol and cardiolipin.
The present study also clarifies the marked inhibition of rat CDS activity in vitro by PIP 2 (Fig. 4), a presumed end product of the phosphoinositide cycle. This represents another new example of the potential for acidic phospholipids including phosphoinositides to modulate the in vitro activities of several membrane enzymes related to the signal transduction. PI-4phosphate 5-kinase appears to be inhibited by PIP 2 and be activated by PA (43)(44)(45), casein kinase I appears to be inhibited by PIP 2 (46), phosphoinositide-specific PLC-␥1 appears to be activated by PA (47), and several protein kinase C isoform are activated by PIP 2 (phosphatidylinositol 3,4-bisphosphate as well as phosphatidylinositol 4,5-bisphosphate) and phosphatidylinositol 3,4,5-trisphosphate (48 -50). This PIP 2 -induced enzymatic inhibition and the substrate specificity of rat CDS suggest the presence of a mechanism for PIP 2 to regulate its own synthesis through the phosphoinositide cycle by feedback.
The PIP 2 -induced inhibition has recently been demonstrated to occur on arachidonoyl-DAG kinase, which catalyzes the synthesis of PA from sn-1-acyl-2-arachidonoyl-DAG selectively by phosphorylation (51). Considering the fact that DAG kinase and CDS are sequentially involved in the phosphoinositide resynthesis, the finding that both arachidonoyl-specific DAG and CDS are inhibited by PIP 2 strongly suggests that PIP 2 is synthesized on demand as opposed to being stored, and the conversion process from DAG to CDP-DAG through PA, the initial steps of resynthesis of phosphoinositides, is tightly regulated in the phosphoinositide-mediated signaling cascades. In addition, similar to the CDS molecule, the arachidonoyl-DAG kinase has been shown to be expressed also highly in the testis as well as the brain and to be an integral membrane protein (52). It is thus suggested that these two arachidonoyl-specific enzyme molecules are located in the same subcellular membranes and are regulated by the lipid microenvironment within the membranes in a paired fashion in the two consecutive reactions of the phosphoinositide cycle. The detailed cellular  and subcellular localization of the arachidonoyl-DAG kinase in the testis as well as the brain and retina, and its comparison with that of this CDS may be informative in this regard. The functional significance of the most intensive expression of rat CDS mRNA in testis, especially in postmitotic spermatocytes and spermatids remains to be elucidated.
The localization of this CDS mRNA in the inner segment of rat photoreceptor cells (Fig. 9) corresponds well to the presence of the Drosophila homologue in the photoreceptor cells, suggesting some similar functional significance between the two homologous molecules of far remote animal species. As cited in the Introduction, it has clearly been demonstrated that phosphoinositide-specific PLC isoforms cloned from mammalian retina are distributed in the photoreceptor cells (13)(14)(15) and that light induces the decrease in PIP 2 and increase in IP 3 in the photoreceptor cells (19 -22), and light also enhanced synthesis of PI by activation of phosphoinositide cycle in these cells (24). The existence of IP 3 receptor in rod outer segment has also been shown (18), and Ca 2ϩ -mediated light adaptation has been demonstrated in vertebrate photoreceptor cells (25,26). In addition, protein kinase C, an enzyme activated by DAG, is reportedly present in the rod outer segment (16,17), and its activity appears to be stimulated by light (23). Protein kinase C is known to phosphorylate several rod outer segment proteins, including rhodopsin (53) and phosphodiesterase ␥ (54). Therefore, despite the known difference in the primary second messenger species, cGMP versus IP 3 , between vertebrate and invertebrate phototransduction, our finding implies the possibility that some additional roles for the phototransduction are played by this CDS molecule through the phosphoinositide cycle in rat photoreceptor cells. The intense gene expression of this CDS in the pineal organ (Fig. 8b), a phylogenically homologous organ to the retina, is in favor of this possibility. The heterogeneous but wide expression localization of CDS mRNA in the gray matter of the brain (Fig. 8a) is also instructive for the functional significance of CDS in the neuronal signal transduction. The intense expression of CDS mRNA in the cerebellar Purkinje cells and moderate expression in the hippocampal pyramidal cells are consistent with the gene expression pattern of the receptor for IP 3 , one of the two second messengers produced by phosphoinositide cycle (55).
The molecular identification of the first mammalian CDS completed by the present study would accelerate further elucidation of the functional significance of this enzyme in the phosphoinositide-related signal pathway.