Isoform-specific and Protein Kinase C-mediated Regulation of CTP:Phosphoethanolamine Cytidylyltransferase Phosphorylation*

Background: Post-translational regulation of CTP:phosphoethanolamine cytidylyltransferase (Pcyt2) is largely unexplored. Results: Pcyt2 isoforms are characteristically phosphorylated within a newly identified central linker segment not present in other cytidylyltransferases. Conclusion: Pcyt2 phosphorylation influences cell growth and is regulated by conventional PKCs and serum deficiency. Significance: The isoform-specific phosphorylation reveals a post-translational regulation of Pcyt2 that is distinct from other cytidylyltransferases. CTP:phosphoethanolamine cytidylyltransferase (Pcyt2) is the main regulatory enzyme for de novo biosynthesis of phosphatidylethanolamine by the CDP-ethanolamine pathway. There are two isoforms of Pcyt2, -α and -β; however, very little is known about their specific roles in this important metabolic pathway. We previously demonstrated increased phosphatidylethanolamine biosynthesis subsequent to elevated activity and phosphorylation of Pcyt2α and -β in MCF-7 breast cancer cells grown under conditions of serum deficiency. Mass spectroscopy analyses of Pcyt2 provided evidence for isoform-specific as well as shared phosphorylations. Pcyt2β was specifically phosphorylated at the end of the first cytidylyltransferase domain. Pcyt2α was phosphorylated within the α-specific motif that is spliced out in Pcyt2β and on two PKC consensus serine residues, Ser-215 and Ser-223. Single and double mutations of PKC consensus sites reduced Pcyt2α phosphorylation, activity, and phosphatidylethanolamine synthesis by 50–90%. The phosphorylation and activity of endogenous Pcyt2 were dramatically increased with phorbol esters and reduced by specific PKC inhibitors. In vitro translated Pcyt2α was phosphorylated by PKCα, PKCβI, and PKCβII. Pcyt2α Ser-215 was also directly phosphorylated with PKCα. Mapping of the Pcyt2α- and -β-phosphorylated sites to the solved structure of a human Pcyt2β showed that they clustered within and flanking the central linker region that connects the two catalytic domains and is a novel regulatory segment not present in other cytidylyltransferases. This study is the first to demonstrate differences in phosphorylation between Pcyt2 isoforms and to uncover the role of the PKC-regulated phosphorylation.

CTP:phosphoethanolamine cytidylyltransferase (Pcyt2) is the main regulatory enzyme for de novo biosynthesis of phosphatidylethanolamine by the CDP-ethanolamine pathway. There are two isoforms of Pcyt2, -␣ and -␤; however, very little is known about their specific roles in this important metabolic pathway. We previously demonstrated increased phosphatidylethanolamine biosynthesis subsequent to elevated activity and phosphorylation of Pcyt2␣ and -␤ in MCF-7 breast cancer cells grown under conditions of serum deficiency. Mass spectroscopy analyses of Pcyt2 provided evidence for isoform-specific as well as shared phosphorylations. Pcyt2␤ was specifically phosphorylated at the end of the first cytidylyltransferase domain. Pcyt2␣ was phosphorylated within the ␣-specific motif that is spliced out in Pcyt2␤ and on two PKC consensus serine residues, Ser-215 and Ser-223. Single and double mutations of PKC consensus sites reduced Pcyt2␣ phosphorylation, activity, and phosphatidylethanolamine synthesis by 50 -90%. The phosphorylation and activity of endogenous Pcyt2 were dramatically increased with phorbol esters and reduced by specific PKC inhibitors. In vitro translated Pcyt2␣ was phosphorylated by PKC␣, PKC␤I, and PKC␤II. Pcyt2␣ Ser-215 was also directly phosphorylated with PKC␣. Mapping of the Pcyt2␣-and -␤-phosphorylated sites to the solved structure of a human Pcyt2␤ showed that they clustered within and flanking the central linker region that connects the two catalytic domains and is a novel regulatory segment not present in other cytidylyltransferases. This study is the first to demonstrate differences in phosphorylation between Pcyt2 isoforms and to uncover the role of the PKC-regulated phosphorylation.
Phosphatidylethanolamine (PE) 3 and its alkenylacyl derivatives (PE plasmalogens) are essential membrane phospholipids implicated in the regulation of various cellular processes, including cell proliferation, signaling, autophagy, and apoptosis (1)(2)(3)(4)(5). PE phospholipids are synthesized de novo from ethanolamine and diacylglycerol/alkylacylglycerols through the CDPethanolamine-Kennedy pathway and by decarboxylation of phosphatidylserine in the mitochondria. The de novo Kennedy pathway is quantitatively the most important route for the biosynthesis of PE in mammalian cells (6). In this pathway, ethanolamine is first phosphorylated by ethanolamine kinase to phosphoethanolamine (P-Etn), which is then converted to CDP-ethanolamine (CDP-Etn) by CTP:phosphoethanolamine cytidylyltransferase (Pcyt2; also known as ET (7,8) or ECT (9)). In the final step, CDPethanolamine:1,2-diacylglycerol ethanolamine phosphotransferase transfers P-Etn from CDP-Etn to DAG/alkylacylglycerol to produce PE/plasmalogens. We have recently reviewed the role of Pcyt2 in phospholipid homeostasis (10).
Recent studies using Pcyt2-deficient mouse models established a direct link between PE synthesis and triglyceride metabolism (11)(12)(13). Heterozygous Pcyt2 deletion (Pcyt2 Ϫ/ϩ ) reduced the metabolic flux through the CDP-ethanolamine pathway and generated an excess of DAG. The unused DAG was redirected toward triglyceride synthesis, which also stimulated fatty acid synthesis and resulted in excessive accumulation of fat in multiple tissues. Consequently, Pcyt2 ϩ/Ϫ mice developed metabolic syndrome, characterized by adult-onset obesity, liver steatosis, hypertriglyceridemia, and insulin resistance (11). Liver-conditional knock-out mice (l-Pcyt2 Ϫ/Ϫ ), however, relied on mitochondrial phosphatidylserine decarboxylation to form liver PE. They also developed severe fatty liver phenotypes; however, they did not develop insulin resistance (13). * This work was supported by grants from the Canadian Institutes of Health Complete Pcyt2 knock-out (Pcyt2 Ϫ/Ϫ ) mice die early during embryogenesis (2), showing that Pcyt2 and the CDP-ethanolamine pathway are essential for animal development. Pcyt2 is activated during muscle cell differentiation (14), and the importance of Pcyt2 for growth and development has also been demonstrated in several genome-wide studies (15)(16)(17)(18)(19). Pcyt2 is coexpressed with genes important for ES self-renewal and differentiation (15), and preimplantation embryos in culture conditions overexpress Pcyt2 (16). Pcyt2 is a target gene of histone deacetylase activity and was down-regulated by histone deacetylase and microRNA transcriptional repression during oligodendrocyte differentiation (17). Pcyt2 was up-regulated by NGF-initiated but not by IL6-initiated PC12 cell differentiation (18), and it is a candidate gene in the regulation of mouse brain and retinal development (19).
The importance of PE and Pcyt2 activity for tumor growth has also been well documented (1, 20 -23). Breast cancer cells have reduced Pcyt2 activity and PE/plasmalogen levels relative to noncancerous cells (24), and a limited Pcyt2 activity has also been reported in lymphomatous liver (20) and ras-transformed 3T3 fibroblasts (21). Genome models of cancer metabolism predict Pcyt2 as a synthetic lethal gene in pairing with PS synthase-2 and an important drug target (22). Transcription factor TFAP2C plays an important role in breast cancer biology, and Pcyt2 is induced by TFAP2C in estrogen-responsive breast cancer cells MCF-7 (23). Pcyt2 is regulated by the early growth response protein 1 (EGR1) and NFB (1). Pcyt2 is, however, still less expressed in MCF-7 cells compared with the nontumorigenic MCF-10A cells because of the reduced activity of EGR1 and increased activity of NFB (24).
Unlike other members of the cytidylyltransferase family, Pcyt2 monomers contain two cytidylyltransferase catalytic folds, which we refer to as N-cat (N-terminal domain) and C-cat (C-terminal domain; Fig. 1). Mammals express two functional Pcyt2 proteins, the longer Pcyt2␣ and the shorter Pcyt2␤, produced by alternative splicing of exon 7 (25). Exon 7 encodes an 18-amino acid peptide in Pcyt2␣ that is contained within the ϳ75-residue linker region that joins the two catalytic domains and may contribute to distinct kinetic properties of the two isoforms (26). The longer isoform, Pcyt2␣, is more active at higher substrate concentrations (P-Etn and CTP), whereas Pcyt2␤ is more active at lower substrate concentrations (26). The crystal structure of human Pcyt2␤ has been solved by x-ray diffraction (Protein Data Bank code 3elb).
There is no information available regarding the post-translational regulation of Pcyt2. We have recently reported an unexpected stimulation of Pcyt2 activity and phosphorylation in MCF-7 cells when grown under conditions of serum deprivation (1). In this study, we identified the isoform-specific phosphorylation sites and mapped them to the inter-domain linker and flanking regions. Mainly nonphosphorylated in the presence of serum, Pcyt2␣ became phosphorylated in the linker segment, including the ␣-specific motif, and in the C-cat domain in the absence of serum. Pcyt2␤ was uniquely phosphorylated in the N-cat and proximal linker region and shared most of the phosphorylation with Pcyt2␣ within the C-cat domain. We show that the Pcyt2␣ isoform is regulated by protein kinase C (PKC) phosphorylation at two conserved sites within the C-terminal domain. Mutation analysis and phosphorylation studies establish that conventional PKCs are activators of Pcyt2 and contribute to the elevated Pcyt2 activity and PE synthesis required for stimulated cancer cell growth. To our knowledge, this is the first study to identify the phosphorylation sites and types of regulation of Pcyt2 by phosphorylation that is fundamentally different from other cytidylyltransferases.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-MCF-7 human breast cancer cells were transfected with plasmids containing His 6 -tagged Pcyt2␣ and Pcyt2␤ as described (1). The Pcyt2 expression plasmids, pcDNA3.1ET␣ and pcDNA3.1ET␤, were produced from the full-length clones BC003473 for Pcyt2␣ and BU557590 for Pcyt2␤ and made in-frame with the V5/His 6 tag peptide sequence at the C terminus. Approximately 60% confluent MCF-7 cells grown in a 100-mm dish were transfected with 24 g of the pcDNA3.1ET␣ or pcDNA3.1ET␤ using Lipofectamine (Invitrogen), and the cells were grown 24 h in serum-deficient minimum Eagle's medium or in the same media supplemented with 10% fetal bovine serum.
Pcyt2␣ and Pcyt2␤ Analysis with MALDI-Qq-TOF Mass Spectroscopy-Purification of His-tagged Pcyt2␣ and -␤ proteins overexpressed in MCF-7 cells was performed by nickel column chromatography as described (1). Purified His-tagged Pcyt2␣ and -␤ proteins were digested overnight at 37°C both with trypsin and AspN in 50 mM ammonium bicarbonate, pH 8.2. The phosphopeptides from digested samples were enriched through specific interactions of phosphate groups with immobilized SwellGel gallium(III) disc, according to the phosphopeptide isolation kit (Pierce, catalog no. 89853). In brief, after adjusting the pH to 2-3, digested samples were added to SwellGel discs and bound phosphopeptides, eluted with 50 l of 100 mM ammonium bicarbonate (acidified to pH Ͻ2.5 by acetic acid), and further purified using zip-tips (Millipore). Pcyt2␣ and -␤ phosphopeptide analysis was conducted using LCQ DECA XP ion-trap (Thermo Finnigan, San Jose, CA) and MALDI-Qq-TOF mass spectrometers (MDS SCIEX, Toronto, Canada) as described (26). The enriched phosphopeptide mixtures were separated on a Vydac reversed-phase column (Grace Vydac, Hesperia, CA) using 5% dilution of a mixture of 99.9% acetonitrile ϩ 0.1% acetic acid, at the flow-rate of 10 l/min for 5 min. The eluted peptides were then detected by MALDI-Qq-TOF mass spectrometer equipped with a low flow metal needle assembly and operating in a data-dependent mode (27). Peptide sequencing data were analyzed by Sequest software (Lisle, IL). Raw spectra were compared against the theoretical tandem mass spectra in the Sequest database to which the Pcyt2␣ and -␤ protein sequences were added on our initiative because they were not present in the database prior to this analysis (27).
Effect of Pcyt2␣ Mutations on Enzyme Activity and Cell Growth-The enzyme activity of the cells transfected with plasmids containing V5 and His-tagged versions of the wild-type Pcyt2␣ or ␣S215A, ␣S223A, and ␣Ser215A,S223A Pcyt2␣ mutants was assayed as described (1). We used 50 g of the cell lysate in 100 l of the reaction mixture of 10 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 5 mM dithiothreitol, 2 mM CTP, 0.01 Ci of [ 14 C]P-Etn, 1 mM unlabeled P-Etn. The [ 14 C]CDP-Etn product formed after 15 min was separated from the [ 14 C]P-Etn substrate on silica gel G plates by methanol, 0.5% NaCl, ammonia (50:50:5 v/v/v), visualized by 0.1% ninhydrin in butanol, and analyzed by scintillation counting. Protein content was determined using the bicinchoninic acid protein assay (Pierce).
Furthermore, the induction of cell survival by autophagy was examined. Content and lipidation of the autophagy-specific marker LC3 was measured by Western blotting as described previously (5).
Treatments with Phorbol Esters and PKC Inhibitors-Serumstarved MCF-7 cells were treated with 25 nM phorbol 12-myristate 13-acetate (PMA, Sigma) for 0, 0.5, 1, and 2 h. At the end of treatment periods, cells were homogenized in a lysis buffer and centrifuged to clear the lysates, and supernatants were subjected to immunoprecipitation using Pcyt2 (␣ϩ␤) antibody (1). The precipitated proteins were separated on SDS-PAGE, transferred onto PVDF membrane, and probed with Ser(P) and Pcyt2 (␣ϩ␤) antibodies. The cell lysates were also used for determination of total Pcyt2 activity as described above. In a second set of experiments, serum-starved cells were treated for 2 h with 25 nM PMA alone or together with the PKC inhibitors, Gö6976 (50 nM) or myristoylated EGF-R fragment 651-658, (50 nM) (both from Calbiochem). The Pcyt2 activity and the analyses using anti-Pcyt2 (␣ϩ␤) and anti-Ser(P) were performed as above.
Radiolabeling with [ 14 C]Ethanolamine-MCF-7 cells (10 6 cells/60-mm dish) were grown as described under "Cell Culture and Transfections." Twenty four hours after transfection with plasmids containing His-tagged Pcyt2␣, S215A, S223A, or S215A,S223A mutants, the cells were labeled for 1 h with [ 14 C]ethanolamine (0.1 Ci/dish) in the presence of 50 M unlabeled ethanolamine. Radiolabeled lipid-and water-soluble fractions were separated by the method of Bligh and Dyer (1). Total ethanolamine incorporation into PE was determined by evaporation of the entire chloroform phase under nitrogen, followed by the separation of PE from lyso-PE in a TLC system of chloroform/methanol/ammonia (65: 35:5). Spots corresponding to PE were collected and subjected to liquid scintillation analysis.
In Vitro Transcription and Translation of Pcyt2-To obtain a nonphosphorylated Pcyt2 protein, an untagged Pcyt2␣ cDNA was prepared and translated in vitro using the PURExpress in vitro protein synthesis kit (New England Biolabs). This system produces a protein product devoid of post-translational modifications. The untagged Pcyt2␣ cDNA template was generated by PCR using the pcDNA.ET␣ plasmid. We amplified the entire open reading frame and the required flanking sequences, including the ribosomal binding site, but deleted the V5 and His tags using Pcyt2-specific primers (forward primer, 5Ј-taagaaggagatataccaatgatccggaacgggcacgg-3Ј-3Ј; reverse primer, 5Ј-tattcattagtcaatctcccctccaggcggcgcc-3Ј). In the second round of PCR, we added the T7 promoter at the 5Ј-end of Pcyt2 using the universal primer (5Ј-gaaattaatacgactcactatagggagaccacaacggtttccctctagaaataattttgtttaactttaagaaggagatatacca-3Ј) and the Pcyt2 reverse primer. Pcyt2 cDNA was purified using a gel extraction kit (Sigma) and 250 ng of the cDNA incubated with mixture I (PURExpress) at 37°C for 1 h to generate Pcyt2␣ protein. The mixture was concentrated (60 min at 1,500 ϫ g at 4°C) using a Microcon YM-100 spin concentrator (Millipore Corp.) and mixed with 0.25 volumes of nickel-nitrilotriacetic acid-agarose beads (Qiagen) (mixture II) and incubated at 4°C for 45 min to remove the translation assay components. Mixture II was then applied to a micro-spin column (Bio-Rad), and the unbound Pcyt2␣ protein was released by centrifugation (2 min at 1500 ϫ g at 4°C). Pcyt2␣ proteins from multiple reactions were combined and pooled samples applied in the in vitro PKC phosphorylation assays below.
In Vitro Pcyt2 Phosphorylation with PKC␣, PKC␤I, and PKC␤II-Pcyt2␣ free from tags and post-translational modifications was subjected to phosphorylation with [␥-32 P]ATP and either PKC␣, PKC␤I, or PKC␤II (Enzo Life Sciences) (28). The reaction was carried out at 30°C for 10 min in 100 l of mixture containing 20 mM HEPES, pH 7.4, 10 mM MgCl 2 , 0.1 mM CaCl 2 , 100 M ATP, 100 g/ml phosphatidylserine, 20 g/ml diacylglycerol, 0.03% Triton X-100, 1 l of [␥-32 P]ATP, and different aliquots (20 or 40 l) of Pcyt2␣ protein. Reactions were initiated by adding 0.4 g of PKC and terminated after 10 min by adding 20 l of 1% BSA and 1 ml of 10% TCA. After 5 min of incubation on ice, the aliquots (40 l) from each reaction were transferred to a nitrocellulose membrane on an aspirator funnel and washed three times with 2 ml of 5% TCA. Membranes were dried under vacuum and the ␥-32 P was incorporated into Pcyt2␣ protein determined by counting. The assays were performed in triplicate, and the background radioactivity for the reactions without Pcyt2␣ or PKC served as controls (28).
Statistical Analysis-All measurements were expressed as means Ϯ S.D. from at least three independent experiments. GraphPad Prism 3.0 software (San Diego) was used to perform one-or two-way analysis of variance and Student's paired t test. Differences were considered statistically significant at p Յ 0.05. Densitometry analysis was performed by Scion Image software (Scion Inc).

Pcyt2␣ and Pcyt2␤ Catalytic Domains and Phosphorylation
Sites- Fig. 1 provides a simplified representation of the Pcyt2␣ and -␤ structures showing the position of the two catalytic (CT) domains. The two splice isoforms differ only in the length of the linker connecting the two CT domains as follows: 74 residues for the ␣ isoform and 56 residues for Pcyt2␤. Putative phosphorylation sites were deduced by the Scan-Prosite proteomic server (ExPASy-Swiss Institute of Bioinformatics). Details on how the specific phosphorylations shown in Fig. 1 were characterized are described below.
Pcyt2␣ and Pcyt2␤ Phosphorylation Is Distinct and Serumdependent-In response to serum deficiency, Pcyt2 is activated by phosphorylation, leading to elevated synthesis of PE by the CDP-ethanolamine-Kennedy pathway (1). Contributions of separate isoforms to the increased Pcyt2 activity have not been investigated, and the mechanism behind the increased enzyme phosphorylation in serum-deficient MCF-7 cells remains unknown. In addition, because the predicted phosphorylation sites are conserved, the contribution of each isoform to total activity was difficult to establish. Therefore, we transfected MCF-7 cells separately with His-tagged Pcyt2␣ and Pcyt2␤ (1), cultured the cells under different serum conditions, and purified overexpressed proteins using a nickel-agarose column. The purified Pcyt2 proteins were digested with trypsin, and the phosphopeptides were enriched by column chromatography (27) and investigated by mass spectroscopy (MS). The final phosphorylation results obtained from multiple transfections and purifications are summarized in Table 1 and Fig. 1. The initial analysis of the MS data revealed several general features of Pcyt2 phosphorylations as follows. (i) Maximal phosphorylation of both isoforms was observed in the absence of serum as predicted by previous data (1). (ii) The general pattern of phosphorylation was very distinctive for Pcyt2␣ and -␤. (iii) Phosphorylation was mostly at serine residues as indicated previously (1). (iv) Additional phosphorylations were identified at multiple threonine residues, specifically in Pcyt2␤ (Thr-146/ Thr-147, and Thr-157), and one phosphorylation at a tyrosine residue in Pcyt2␣ (Tyr-308).
Pcyt2␤ Is Uniquely Phosphorylated within the First Catalytic Domain-Strikingly distinct phosphorylation was established at the N-terminal catalytic domain (Fig. 1). Pcyt2␣ was not phosphorylated in the N-Cat domain under any experimental conditions. Pcyt2␤, however, was phosphorylated at multiple residues within the N-Cat domain as follows: 26 -152 and in the linker segment immediately following the N-Cat domain. All Pcyt2␤ N-terminal phosphorylations were identified under serum-deficient conditions (Table 1), when the enzyme activity was elevated (1). Pcyt2␤ was specifically phosphorylated at ␤Ser-145 in the CKII consensus motif ( 141 TQGVpSpTpTDLVGR 152 ) and at three threonine residues ␤Thr-146, -147 and -157 ( Fig. 2 and Table 1). Most importantly, part of the 141-152 peptide belongs to the cytidylyltransferase signature motif 140 RTQGVSTT 147 , which interacts with the enzyme substrate   Pcyt2␣ Is Uniquely Phosphorylated at the ␣-Specific Motif-Generally, Pcyt2␣ was not phosphorylated in the presence of serum, with the exception of ␣Ser-300 located in the C-cat domain, which was phosphorylated under both growth condi-tions ( 296 CRYVpSEVVIGAPYS 309 ; Table 1). The corresponding serine in Pcyt2␤, Ser-282, was phosphorylated only in the absence of serum. Independent evidence for the phosphorylation of the ␣Ser-300 / ␤Ser-282 exists in the PhosphoSitePlus database (Ser(P)-282) and likely represents a constitutive phosphorylation site. In the absence of serum, Pcyt2␣ was specifically phosphorylated at two serine residues ␣Ser-192 and ␣Ser-195 within the ␣-specific motif ( 192 SpSEVpSSQCPG 201 in Fig.  3A). The ScanProSite analyses did not identify a kinase that could be responsible for this phosphorylation.  MARCH 28, 2014 • VOLUME 289 • NUMBER 13

JOURNAL OF BIOLOGICAL CHEMISTRY 9057
Phosphorylation of Pcyt2␣ and -␤ within the Second Catalytic Domain-Unlike the N-cat domain, the C-cat domain was phosphorylated in both Pcyt2 isoforms with some specific vari-ations. Phosphorylations at ␣Ser-215 and ␣Ser-223, which are part of PKC consensus sites 215 SQK and 223 SGK, were identified only under serum-deficient conditions when total phosphorylation and activity were enhanced (1). Pcyt2␤ was not phosphorylated at the corresponding ␤Ser-197, but it was phosphorylated at the second PKC site ␤Ser-205 under both growth conditions. Representative spectra identifying those identical phosphorylations (␣Ser-223 and ␤Ser-205) are in Fig.  3, B and C. An additional common phosphorylation was detected at ␣Ser-209/␤Ser-191 (Table 1), however, the specific kinase(s) for this site could not be identified.
From the composite of MS data shown in Table 1, we can conclude that most phosphorylations were associated with serum-deficient conditions when the enzyme was more active. Pcyt2 isoforms were phosphorylated at specific and mutual sites and therefore could experience both distinct and identical types of regulation. The most noticeable phosphorylation differences were present within the N-cat domain, which was exclusively phosphorylated in Pcyt2␤. Pcyt2␣, however, was phosphorylated at the ␣-specific motif spliced out in Pcyt2␤. Both isoforms were phosphorylated within the second catalytic domain and shared the phosphorylation at one PKC site (␣Ser-223/␤Ser-205).
Phorbol Esters and PKC Inhibitors Regulate Pcyt2 Activity and Phosphorylation-We next determined whether the phosphorylation and activity of endogenous Pcyt2 could be modified by a PKC activator, PMA, and PKC-specific inhibitors, Gö6976 and EGF-R fragment 651-658. Serum-deficient MCF-7 cells were treated with PMA for 0 -2 h followed by immunoprecipitation with Pcyt2 (␣ϩ␤) antibody and probing with Ser(P)-and Pcyt2 (␣ϩ␤) antibodies (Fig. 4, A and C) or used for determination of Pcyt2 activity (Fig. 4, B and D). Serine phosphorylation of Pcyt2(␣ϩ␤) was increased by PMA addition, and both PKC inhibitors reduced the PMA-induced phosphorylation (Fig. 4, A and C). At the same time, Pcyt2 activity was elevated by PMA treatments, and Gö6976 or EGF-R attenuated the PMA-induced Pcyt2 activity (Fig. 4, B and D). These results established that endogenous Pcyt2 proteins were subjects of activation by phosphorylation that is regulated by PKC.
Pcyt2␣ Is Phosphorylated with PKC␣, PKC ␤I, and PKC ␤II-We next tested whether different PKC isozymes could directly phosphorylate Pcyt2␣. To avoid the interference from other types of phosphorylation, we produced an untagged Pcyt2␣ protein by an in vitro transcription-translation system that generates a protein product devoid of post-translational modifications, including phosphorylation. The translated Pcyt2␣ protein was pooled from multiple reactions and used at two different concentrations in PKC phosphorylation assays with [ 32 P]ATP and either PKC␣, PKC␤I, or PKC␤II (Fig. 4E). All three PKC isoforms increased the incorporation of 32 P into nascent Pcyt2␣ protein, and the incorporation was proportional to the concentration of Pcyt2␣. Kinase-catalyzed phosphorylation was ranked in the order PKC␣ Ͼ PKC␤II Ͼ PKC␤I. These results established that Pcyt2␣ could be directly phosphorylated by all three conventional PKCs but is the best substrate for PKC␣.
To further demonstrate that Pcyt2␣ and Pcyt2␤ are substrates for PKC␣, the isoforms were overexpressed in serum-treated cells (conditions established to produce minimal Pcyt2 phosphoryla-  (C and B), which is also part of the PKC consensus site ( Table 1). The specific ions (m/z) of the phosphorylated residues are indicated. Spectra are representative of at least three transfections and protein purifications. tion), purified, and subjected to phosphorylation with PKC␣ using unlabeled ATP. The PKC␣-phosphorylated proteins were than characterized by MS. As indicated in Fig. 4, F and G, the PKC␣specific phosphorylation was detected on one of the predicted sites, ␣Ser-215. Neither ␣Ser-223 nor the same sites in Pcyt2␤ were phosphorylated by PKC␣ under the same conditions (data not shown). This site could be the target for PKC␤.
Mutations of Both PKC Sites Reduced Pcyt2␣ Phosphorylation and Activity-To probe whether the phosphorylation at the PKC-directed sites has a regulatory function, we created  1 g (hatched bars) and 0.2 g (open bars). The 32 P activity incorporated into Pcyt2␣ was compared with assays without Pcyt2 or PKC added. F and G, overexpressed V5/His-tagged Pcyt2␣ was purified from serum-treated MCF-7 cells (conditions established to produce minimal phosphorylation) and was subjected to the PKC␣ assays as above, however, using unlabeled ATP. The phosphorylated peptides were separated and characterized by MALDI-Qq-TOF tandem mass spectroscopy. Shown are the two m/z fragmentations for PKC␣-treated Pcyt2␣. Both patterns identify Pcyt2␣ phosphorylation at Ser-215. single (␣S215A and ␣S223A) and double (␣S215A,S223A) Pcyt2␣ mutants and performed transfections under serum-deficient conditions (conditions established to produce maximal phosphorylation). The overexpressed proteins were immunoprecipitated with Pcyt2 tag V5 antibody and probed with anti-Ser(P). As shown in Fig. 5A, the single mutations ␣S215A and ␣S223A similarly reduced Ser phosphorylation, with the double mutant ␣S215A,S223A showing an almost complete loss of phosphorylation. The activities of ␣S215A, ␣S223A, and ␣S215A,S223A mutants were reduced to ϳ50, 8, and 36% of the wild-type Pcyt2␣, respectively (Fig. 5B). In addition, as shown for the double mutant in Fig. 5C, the hypophosphorylation of the PKC double-site mutant could not be further modified with PMA or EGF-R (Fig. 5C). When the effect of various PMA concentrations on endogenous and overexpressed Pcyt2␣ was examined (Fig. 5D), PMA strongly stimulated endogenous activity with maximal 7-fold activation at 100 nM PMA in serum-deficient conditions. The activity of the overexpressed Pcyt2␣ was ϳ4-fold above the endogenous level and showed no further activation with PMA (Fig. 5D). This is not surprising because the overexpressed Pcyt2␣ was already highly phosphorylated under serum-deficient conditions, including both PKC sites, as established by multiple MS analyses described earlier.
The effect of Pcyt2␣ mutations on PE synthesis by the CDPethanolamine-Kennedy pathway is shown in Fig. 5E. Transfected cells were radiolabeled with [ 14 C]ethanolamine, and its incorporation into newly synthesized PE was determined. Data showed that the reduction in PE synthesis (Fig. 5E) was mirrored by the reduction in Pcyt2 activity (Fig. 5B). Therefore, we concluded that the mutation of ␣Ser-215 and ␣Ser-223 significantly impacted the enzyme phosphorylation and activity, showing that both sites are important for regulation by serum deficiency. Interestingly, the double mutant showed a maximal reduction in phosphorylation that, however, did not further suppress the enzyme activity and PE synthesis, suggesting that the two PKC sites were not synergistically regulated and are likely independent.
Pcyt2␣ Mutants Do Not Support Cell Growth-Regulation of Pcyt2 activity by PKC␣, also an important marker of breast cancer cell proliferation (29), suggests that the stimulation of Pcyt2␣ and PE synthesis under serum deficiency (1) is important in supporting the cancer cell growth and survival. To test if the Pcyt2␣ mutants could influence the MCF-7 cell growth, the cells were transfected with wild-type or the phosphorylation site mutants described above, and cell growth was monitored under serum-deficient conditions. We have previously shown that the endogenous Pcyt2 was elevated and supported the MCF-7 cell proliferation under serum-deficient conditions (1). The overexpressed Pcyt2␣ increased the growth further by additional stimulation of membrane PE synthesis (1). Overexpressed mutated proteins did not support the cell growth and were toxic at longer times (Fig. 5F). The observed effect on cell proliferation was concomitant with the decrease in PE synthesis under the same conditions (Fig. 5E).
Under conditions of serum/nutrient deprivation, cell growth is typically supported by the process of autophagy (30,31). Hence, we further tested contribution of autophagy to the observed changes in PE synthesis and cell proliferation. We have previously demonstrated that Pcyt2 could increase the supply of PE for LC3-I lipidation during induction of autophagy (5,31), which could stimulate cell growth (Fig. 5F). The main event in the autophagy is a covalent attachment of PE to the autophagy-specific marker LC3-I (lipidation) to form the PEbound LC3-II. Therefore, we tested if the overexpressed Pcyt2␣ increased the MCF7 cell growth (Fig. 5F) by stimulating the LC3-I lipidation. We monitored the content of LC3-I (PE-free form) and LC3-II (PE-bound form) by Western blotting (Fig. 5,  G and H). However, in mock-transfected cells, the amount of LC3-I and LC3-II was reduced in the absence of serum. Although the ratio of LC3-II to LC3-I in the presence of serum was ϳ2.60, the same ratio was unchanged in serum-deprived cells (ϳ1.09) (Fig. 5G), showing that the autophagy was generally impaired and therefore did not contribute to the observed changes in cell growth in serum deficiency (Fig. 6B). Hence, it is not surprising that the overexpression of Pcyt2␣ did not increase the LC3 lipidation either after 24 h (LC3II/LC3I ratio in mock versus Pcyt2␣ was 1.14:1.06) or 48 h (LC3II/LC3I ratio in mock versus Pcyt2␣ was 0.98:1.01) (Fig. 5H). We concluded that the increased Pcyt2 activity observed in serum-deficient cells (1) stimulated the cell growth by increasing the membrane PE formation (1) without additional effects on cell renewal by autophagy.
Identification of a Centrally Located Pcyt2 Phosphorylation Region-To determine where the common and isoform-specific phosphorylation sites are located in the Pcyt2 tertiary structure, we mapped them onto the solved crystal structure of the human Pcyt2␤, available in the Protein Data Bank under code 3elb. The human (NP_002852) and murine (EDL34775.1) Pcyt2␤ sequences are 92% identical, and all of the phosphorylation sites we have identified are conserved between the two species. A comparison of the sequences of murine Pcyt2␣ and human Pcyt2␤ is shown in Fig. 6A along with the phosphorylation sites we have identified by MS and the secondary structure motifs obtained from the solved structure (3elb). As shown in Fig. 6B, Pcyt2 is composed of two nearly identical ␣/␤ Rossman folds, each composed of five ␤-strands (␤1-␤5) and six ␣-helices. Other cytidylyltransferases, including CTP:phosphocholine cytidylyltransferase (Pcyt1/CCT) or CTP:phosphoglycerol cytidylyltransferase (GCT), contain one ␣/␤ Rossman fold, and one catalytic domain. Two monomers of CCT or GCT associate to form a dimer that is analogous to the Pcyt2 monomer with its two catalytic domains (9,32,33). Fig. 6A also highlights the linker segment that connects the two catalytic folds. In the solved structure of human Pcyt2␤, 30 residues of this 57-residue segment were unresolved, as indicated by the broken line in Fig. 6B. This unresolved region includes the ␣-specific 18-mer motif. From residues 187 to 212, the Pcyt2␤ structure is ordered, including a short helix (␣F) that makes contact with the C-cat domain. Our phospho-site mapping indicated that all the phosphorylated sites cluster within the linker area or at sites proximal to it in the tertiary structure (residues ␤Ser-145, ␤Thr-146, and ␤Thr-147 in N-cat and residues ␤Ser-197, ␤Ser-205, and ␤Ser-282 in C-cat (Fig. 6B).  A, sequence alignment of murine Pcyt2␣ (UniProt Q922E4) and human Pcyt2␤ (UniProt Q99447), is displayed. The structural elements for the human ␤ isoform were obtained from Protein Data Bank code 3elb. N-and C-domains contain identical Rossman folds of six helices (␣A to ␣E) and five ␤-sheets (␤1 to ␤5); (h are ␣-helix residues and s are ␤-strand residues). The area unresolved in the Pcyt2␤ crystal is inside the boxed sequence, indicated by a dashed line, and includes the ␣-specific peptide (gray box). The identified phosphorylated residues are shown for mPcyt2␣ in red and for mPcyt2␤ in blue. B, phosphorylation sites mapped onto the solved structure of Pcyt2␤. The image of human Pcyt2␤ (Protein Data Bank code 3elb) was prepared using PyMOL. The N-terminal catalytic domain is blue, and the C-terminal catalytic domain is in gray. A CMP ligand in stick representation occupies the active site of the C-cat domain. The linker region connecting the two domains is highlighted red. The side chains of the phosphorylated residues are shown as spheres (carbon, green; oxygen, red); the numbering is for Pcyt2␤ (Pcy2␣ numbering in parentheses). The broken line denotes the portion of the linker that is unresolved in 3elb and is not to scale. The unresolved segment contains residues 157-186 in Pcyt2␤ and residues 157-203 in Pcyt2␣; these include identified phosphorylation sites at residues 157, 162, 167, and 168 in Pcyt2␤ and residues 192 and 195 in Pcyt2␣.

DISCUSSION
For optimal PE synthesis by the CDP-ethanolamine pathway, Pcyt2 catalytic function must be precisely controlled. However, relatively little is known about the regulation of the activity of Pcyt2 in cells. Studies described herein examine the regulation of Pcyt2 by phosphorylation and for the first time demonstrate distinct phosphorylation properties of two Pcyt2 isoforms. We established the isoform-specific and PKC-mediated phosphorylations that can regulate Pcyt2␣ activity. We identified a Pcyt2-specific phosphorylation region. Findings presented in this study clearly distinguish the regulation of Pcyt2 from the regulation of structurally and functionally similar CTP-phosphocholine cytidylyltransferase (Pcyt1, CCT) from the CDPcholine branch of the Kennedy pathway (32,33).
Pcyt2␣ Is a Physiological Target of PKC-The PKC family of serine/threonine kinases has been strongly implicated in the regulation of PE synthesis and cancer cell growth. DAG and phorbol esters increase the rate of PE synthesis (7), and based on the metabolite pool sizes and the enzyme activity, the regulation of the CDP-ethanolamine pathway was predominantly mediated at the level of Pcyt2 (8,34,35). PMA stimulation of PE synthesis is accomplished through stimulation of Pcyt2 activity as well as by increases in the cellular content of the substrates ethanolamine (7) and DAG (8). PMA-resistant cells have an impaired PE synthesis and Pcyt2 activity (34,35), and phorbol esters target two different PKC isozymes in MCF-7 cells, PKC-␤I/II for the stimulation of PE synthesis and PKC␣ for the stimulation of PE degradation (8). PKC␣ has been strongly implicated in breast cancer development (29). We provided evidence that part of the cancer cell defense mechanism against unfavorable growth conditions of serum deficiency includes the stimulation of Pcyt2␣ activity by phosphorylation and increased membrane PE synthesis (1).
Here, we demonstrated that the increased phosphorylation and activity of endogenous Pcyt2 was strongly attributed to the increased activity of PKC. PKC activator PMA increased although PKC inhibitors decreased the enzyme phosphorylation and activity in a time-and dose-dependent manner. Our data established that Pcyt2␣ is a physiological target of conventional PKCs. The MS analysis identified two PKC consensus site phosphorylations in Pcyt2␣ at ␣Ser-215 and ␣Ser-223. These residues were specifically phosphorylated under serumdeficient conditions when the enzyme was more active. In vitro analysis showed that Pcyt2␣ was a better substrate for PKC␣ than for PKC␤I and PKC␤II, and the phosphorylation of ␣Ser-215 with PKC␣ was confirmed by MS. Phosphorylation at ␣Ser-223 appeared to be an independent target of phosphorylation by other PKCs because this site was phosphorylated in serum deficiency. but it was not found to be phosphorylated by PKC␣. Importantly, a regulatory role for PKC phosphorylation at the consensus sites was suggested by the inhibition of PE synthesis and/or enzymatic activity when phosphorylation was blocked by mutations at these sites.
How Could Phosphorylation in the Linker Segment Influence Enzyme Activity?-Four phosphorylation sites identified in the linker region were found within the structurally unresolved portion; thus we do not know how their phosphorylation status could impact catalytic activity. The sites in the ordered portion of the linker include the PKC-directed sites ␣Ser-215 and ␣Ser-223 (␤Ser-205). These sites are located at the start and at the end of the helical segment (␣F) of the linker. Phosphorylation could affect the ordering of this helix, which is situated near the interface between the two catalytic folds, making contacts with helices ␣A and ␣C of N-cat and helix ␣C of C-cat. This in turn could impact on cooperativity between active sites (36).
Pcyt2␤ Phosphorylation Pattern Includes the Signature Motif RTQGVSTT-In contrast to Pcy2␣, Pcyt2␤ was not directly phosphorylated with PKC␣, and the consensus ␤Ser-197 (corresponding to ␣Ser-215) and ␤Ser-205 (corresponding to ␣Ser-223) phosphorylation was not regulated by serum deficiency as strongly as in Pcyt2␣. The lack of activation of these sites in Pcyt2␤ possibly contributed to the observed reduced activity of Pcyt2␤ relative to Pcyt2␣ established earlier (1,26). Although the N-terminal domain of Pcyt2␣ was not found phosphorylated under any experimental conditions, Pcyt2␤ was specifically phosphorylated at three N-cat sites located within a cytidylyltransferase family signature motif 139 RTQGVSTT 147 . Residues 145-147 are at the start of helix ␣E. In GCT and CCT crystal structures, this portion of helix ␣E made several contacts with CTP-or CDP-choline (9,32). Helix ␣E has been suggested as a potential target for regulation in other cytidylyltransferases (32,33). Therefore, Pcyt2␤ Ser-145, Thr-146, and Tyr-147 phosphorylations could directly modify ␣E helix contributions to the active site and the progress of the reaction. The Pcyt2␤ phosphorylation sites in helix ␣E are potential targets of CK2-mediated phosphorylation. CK2 is, however, constitutively active in different cellular compartments and has been frequently associated with regulation of nuclear proteins. Both Pcyt2 proteins predominantly localize in the cytoplasm, and it is not clear how CK2 would specifically phosphorylate Pcyt2␤ under serum-deficient conditions. Therefore, a goal of future studies could be to better understand the specific role of the CK2 signaling in ␣E phosphorylation and Pcyt2␤ function.
To conclude, our studies provide novel characterization of Pcyt2 phosphorylation. Both isoforms undergo elevated phosphorylation in serum deprivation, suggesting that Pcyt2 regulation is dependent on cell growth and nutritional conditions. The existence of PKC-regulated phosphorylation of Pcyt2␣ and the specific phosphorylation of important Pcyt2␤ cytidylyltransferase motifs suggest that different mechanisms of regulation control the activity of individual isoforms under serumdeficient conditions. Our data support the notion that the majority of phosphorylations are restricted to the central linker region and flanking segments. Phosphorylation in the linker area is generally accompanied by stimulation of enzyme activity. As the linker is a distinct structural feature of Pcyt2, it may have evolved as a discrete regulatory region to control the enzyme's catalytic function by phosphorylation.