Phosphatidylcholine Biosynthesis via CTP:Phosphocholine Cytidylyltransferase β2 Facilitates Neurite Outgrowth and Branching*

Hallmarks of neuronal differentiation are neurite sprouting, extension, and branching. We previously showed that increased expression of CTP:phosphocholine cytidylyltransferase β2 (CTβ2), an isoform of a key phosphatidylcholine (PC) biosynthetic enzyme, accompanies neurite outgrowth (Carter, J. M., Waite, K. A., Campenot, R. B., Vance, J. E., and Vance, D. E. (2003) J. Biol. Chem. 278, 44988–44994). CTβ2 mRNA is highly expressed in the brain. We show that CTβ2 is abundant in axons of rat sympathetic neurons and retinal ganglion cells. We used RNA silencing to decrease CTβ2 expression in PC12 cells differentiated by nerve growth factor. In CTβ2-silenced cells, numbers of primary and secondary neurites were markedly reduced, suggesting that CTβ2 facilitates neurite outgrowth and branching. However, the length of individual neurites was significantly increased, and the total amount of neuronal membrane was unchanged. Neurite branching of PC12 cells is known to be inhibited by activation of Akt and promoted by the Akt inhibitor LY294002. Our experiments showed that LY294002 increases neurite sprouting and branching in control PC12 cells but not in CTβ2-deficient cells. CTβ2 was not phosphorylated in vitro by Akt. However, inhibition of Cdk5 by roscovitine blocked CTβ2 phosphorylation and reduced neurite outgrowth and branching. These results highlight the importance of CTβ2 in neurons for promoting neurite outgrowth and branching and represent the first identification of a lipid biosynthetic enzyme that facilitates these functions.

In response to nerve growth factor (NGF), 6 rat pheochromocytoma (PC12) cells stop proliferating and differentiate into sympathetic neuron-like cells (1). The morphological hallmark of neuronal differentiation is neurite sprouting and elongation and subsequent maturation of neurites into axons and dendrites. These processes increase the demand for membrane components. Accordingly, the biosynthesis of the predominant membrane phospholipid, phosphatidylcholine (PC), is accelerated during neurite outgrowth in response to NGF (2,3).
PC is synthesized in neurons by the CDP-choline pathway (4), in which the rate-limiting reaction is catalyzed by CTP: phosphocholine cytidylyltransferase (CT) (5). Three CT isoforms, encoded by two genes, have been identified in rodents. CT␣ is encoded by the Pcyt1a gene, whereas CT␤2 and CT␤3 are derived from the Pcyt1b gene (6,7). CT␣ and CT␤2 share considerable sequence identity, but CT␣ contains a nuclear localization signal (8), whereas CT␤2 does not (7). Immunofluorescence studies have found that CT␣ is primarily located in the nucleus and that CT␤2 resides in the cytosol and on the endoplasmic reticulum (9). The subcellular localization of CT␤3 has not yet been reported. Although CT␣ is expressed in all tissues, CT␤2 and CT␤3 mRNAs are predominantly expressed in the brain (6). In neurons, PC is synthesized not only in cell bodies but also in distal axons (4, 10 -12). During neurite outgrowth of PC12 cells and Neuro2a cells, CT␤2 expression and CT activity are increased, whereas CT␣ expression is unchanged, indicating a link between CT␤2 expression and neurite outgrowth (3).
Neurites are categorized by their location relative to the cell body. Primary neurites project directly from the cell body. Secondary neurites (i.e. branches) project from primary neurites. The sprouting of primary neurites and branch formation are distinct processes that are independently regulated (13). During neuronal differentiation, NGF activates phosphatidylinositol-3-kinase, which activates Akt (14). Neurite extension and branching are promoted in some populations of neurons by activation of Akt, whereas in other neurons activation of Akt impairs neurite elongation and branching (15). In PC12 cells, Akt activation inhibits neurite branching (16).
On the basis of these findings, we hypothesized that neurons require CT␤2 in axons to provide a localized source of PC for neurite outgrowth, extension, and/or branching. We show that CT␤2 is abundant in PC12 cells and in distal axons of primary cultures of neurons. RNA silencing of CT␤2 in PC12 cells reduced the number of primary neurites and markedly reduced the number of branches but increased the length of individual neurites. Moreover, stimulation of neurite branching in PC12 cells, a process that is normally induced by inhibition of Akt, was abrogated in CT␤2-deficient cells. These data provide important new insights into the role of CT␤2 and PC biosynthesis in neurite sprouting and branch formation.

EXPERIMENTAL PROCEDURES
Cell Culture-Rat pheochromocytoma cells (PC12) were obtained from the American Type Cell Culture Collection. Cells were maintained in F12-K medium supplemented with 15% heat-inactivated horse serum and 2.5% fetal bovine serum at 37°C in a humidified atmosphere containing 5% CO 2 . For differentiation and transfection experiments, cells were seeded on collagen-coated 35-mm dishes at a density of 2 ϫ 10 5 cells/ dish. The cells were incubated overnight and then treated with medium containing 0.5% horse serum and 50 ng/ml 2.5 S NGF (Alomone Laboratories, Jerusalem, Israel), and cells were grown for up to 5 days.
Compartmented Cultures of Rat Sympathetic Neurons and Retinal Ganglion Cells-Sympathetic neurons from the superior cervical ganglia of 1-day-old rats were isolated as previously described (17). Briefly, following dissection, the ganglia were mechanically and enzymatically dissociated and plated into the center compartment of compartmented culture dishes. Initially, all compartments contained 2.5 S NGF (100 ng/ml). NGF was withdrawn from the center compartment (containing cell bodies and proximal axons) after 7 days. The side compartments (containing distal axons alone) contained 100 ng/ml NGF throughout the experiments. Retinal ganglion cells were isolated from 1-day-old rats and cultured in compartmented culture dishes as previously described (18). Proximal axons/cell bodies, distal axons of sympathetic neurons, and distal axons of retinal ganglion cells were harvested from 2-week-old cultures (18,19).
Antibodies-Anti-M rabbit polyclonal antibodies that recognize CT␤2 were generated against the conserved membrane domain of rat liver CT (amino acids 256 -288) and were a generous gift from Dr. R. Cornell (Simon Fraser University, Vancouver, Canada). Anti-human CT␤2 rabbit polyclonal antibodies were a generous gift from Dr. S. Jackowski (St. Jude Children's Research Hospital, Memphis, TN). The CT␤2-specific antibody was raised against a peptide corresponding to amino acids 347-365 of CT␤2 (7). The mouse anti-tubulin monoclonal antibodies were from Sigma, and the anti-phospho-Akt (Ser-473) and anti-Akt antibodies were from Cell Signaling Technology (Beverly, MA). The Cdk5 (cylcin-dependent kinase 5) antibody was purchased from Chemicon International, Inc. (Temecula, CA).
Transfection of PC12 Cells and Measurement of Neurite Length and Branching-PC12 cells were plated at a density of 2 ϫ 10 5 cells/35-mm dish for 12 h, after which the medium was changed to F12-K medium without serum. Cells were transfected with 2.5 g of DNA/35-mm dish. For RNA silencing, cells were co-transfected with a 3.7-kb mammalian expression vector encoding recombinant green fluorescent protein (GFP) (phrGFP) (Stratagene, Cedar Creek, TX) and pSILENCER 4.0 encoding rat CT␤2. PC12 cells were transfected with equimolar amounts of phrGFP and either the silencing construct (pSiCT␤2) or pSILENCER 4.0 containing a scrambled, noncod-ing insert. For CT␤2 overexpression studies, cells were transfected with a cDNA encoding a hemagglutinin (HA)-tagged CT␤2 (pCI[CT␤2-HA]) and phrGFP. The cells were transfected in the presence of Lipofectamine 2000 according to the manufacturer's protocol (Invitrogen). Briefly, pSiCT␤2, pSILENCER 4.0 containing the scrambled insert, or pCI[CT␤2-HA] and phrGFP were added in a 3:1 ratio to Lipofectamine 2000 for 20 min to allow formation of DNA complexes, which were then applied to PC12 cells. After 12 h, the cells were given serum-free F12-K medium containing 50 ng/ml NGF to promote differentiation and neurite outgrowth. After 2, 3, and 4 days, cells were viewed by fluorescence microscopy. Cells that had been co-transfected with phr-GFP and pSiCT␤2 were identified by their fluorescence at 520 nm.
Cells bearing at least one neurite longer than 20 pixels were considered to be differentiated, and all neurites and branches were counted and/or measured. At each time point, at least 20 random fields of view were sampled.
Immunoprecipitation and Immunoblotting-Two dishes of PC12 cells or six dishes of sympathetic neurons and retinal ganglion cells were rinsed with ice-cold phosphate-buffered saline and harvested into 1 ml of immunoprecipitation lysis buffer containing 50 mM Tris-HCl (pH 7.4), 1% Triton X-100, and a protease inhibitor mixture (Sigma). Cell lysates were centrifuged at 10,000 ϫ g for 15 min at 4°C, and then supernatants were incubated for 1 h with 3 l of anti-M antibodies on a rotating shaker at 4°C. A 50% slurry of Protein A-Sepharose beads (40 l) was added, and the mixtures were incubated for 1 h at 4°C. Protein A-Sepharose bead conjugates containing immunoprecipitated CT␤2 were washed three times with buffer containing 50 mM Tris-HCl (pH 7.4) and 0.1% Triton X-100. Proteins were eluted from the Protein A-Sepharose beads using SDS-PAGE sample buffer (62.5 mM Tris-HCl (pH 8.3), 10% (v/v) glycerol, 1% SDS, and 0.04% bromphenol blue) and heated at 70°C for 3 min. Immunoprecipitated proteins were separated by electrophoresis on 10% polyacrylamide gels containing 0.1% SDS and then transferred to Immobilon-P transfer membranes (Millipore, Cambridge, Canada). Ponceau S staining was used to compare protein loading in lanes of the gel. The membranes were blocked for 2 h with 5% skimmed milk in Tris-buffered saline (20 mM Tris-HCl (pH 7.5) containing 500 mM NaCl and 0.1% Tween 20) and then incubated overnight with anti-M (membrane domain) (dilution 1:1,000), anti-CT␤2 (dilution 1:1,000), or mouse anti-tubulin (dilution 1:1,000) antibodies. All blots were washed for 1 h and subsequently incubated with anti-rabbit, or anti-mouse, IgG conjugated to horseradish peroxidase (1:2,500 dilution) for 1 h. Immunoreactivity was detected using Westdura or Westfemto ECL reagent (Amersham Biosciences).
In Vitro Measurement of CT Activity-PC12 cells from two 35-mm dishes were collected in 1 ml of homogenization buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, and 100 g/ml each of leupeptin and aprotinin. The cells were sonicated for 20 s at 4°C. Cell lysates were centrifuged at 7,000 ϫ g for 5 min to remove nuclei and unbroken cells. Aliquots of the supernatant were used for measurement of CT activity and immunoblot analysis. CT activity was determined in the presence of PC/oleate vesicles by monitoring the conversion of phospho-[ 3 H]choline to CDP-[ 3 H]choline (20).
Incubation of PC12 Cells with [ 3 H]Choline-PC12 cells were transfected with control plasmid or phrGFP ϩ pSiCT␤2. Six h later, the medium was changed to serum-free medium that contained 50 ng/ml NGF. After 48 h, cells were incubated with [ 3 H]choline (5 Ci/35-mm dish) for 3 h. Cells were harvested, the amount of cellular protein was determined, and radioactivity in PC was measured.
In Vivo Phosphorylation of CT␤2-PC12 cells were incubated with 50 ng/ml NGF for 4 days, after which the cells were incubated for 48 h with LY294002 or for 1 h with 25 M roscovitine or an equivalent volume of vehicle (dimethyl sulfoxide) in the presence of 100 M [ 32 P]orthophosphate (Amersham Biosciences). Cells were rinsed with ice-cold phosphate-buffered saline and harvested into immunoprecipitation lysis buffer. The cell lysates were then centrifuged at 10,000 ϫ g for 15 min at 4°C, after which supernatants were incubated with 3 l of anti-M antibody on a rotating shaker at 4°C. After 1 h, 40 l of a 50% slurry of protein A-Sepharose beads were added, and the mixture was incubated for an additional 1 h at 4°C. Protein A-Sepharose bead conjugates containing immunoprecipitated CT␤2 were washed three times. CT␤2 was eluted from the beads by heating for 3 min in SDS-PAGE sample buffer at 70°C. Proteins were separated by electrophoresis on 10% polyacrylamide gels containing 0.1% SDS. The gel was dried overnight and then exposed to a PhosphorImager screen (Eastman Kodak Co.) for up to 5 days. Quantification of [ 32 P]orthophosphate-labeled CT protein was performed with a Bio-Rad PhosphorImager.
In Vitro Phosphorylation Assays-PC12 cells were harvested into 1 ml of immunoprecipitation buffer. CT␤2 was immunoprecipitated using anti-M antibody as described above. Tubulin was immunoprecipitated using anti-tubulin antibodies. Cdk5 was immunoprecipitated using 10 l of anti-Cdk5 antibodies. Following immunoprecipitation, Cdk5bound protein A-Sepharose beads were warmed at 30°C for 10 min. CT␤2 protein was eluted from the beads with 40 l of buffer containing 50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , and 2 mM dithiothreitol and boiled for 5 min. The eluted CT␤2 protein was added to Cdk5-protein A-Sepharose beads in the presence of 1 mCi of [ 32 P]ATP and incubated for 15 min at 30°C. The kinase reaction was terminated by heating at 100°C for 3 min in the presence of 5ϫ SDS-sample buffer. Proteins were separated by electrophoresis on 10% polyacrylamide gels containing 0.1% SDS. The gels were dried overnight and then exposed to a PhosphorImager screen (Eastman Kodak Co.) for up to 3 days.
Phospholipid Extraction and Quantification-Cell lysates were centrifuged at 7,000 ϫ g for 5 min to remove nuclei and unbroken cells. Phospholipids were extracted with chloroform/ methanol (2:1, v/v) and separated by thin layer chromatography on silica gel G60 plates with chloroform/methanol/acetic acid/ formic acid/water (70:30:12:4:1) as developing solvent. Phospholipids were visualized by exposure of the plate to iodine vapor and identified by comparison with phospholipid standards. The relevant spots were scraped from the plate, and amounts of phospholipids were determined by phosphorus analysis (21).
CT␤2-silencing RNA Oligonucleotides-An oligonucleotide specific for CT␤2 mRNA (ACA GGT ATC CCA AAA TCCC) was designed using a sequence finder algorithm (Ambion, Austin, TX) and synthesized at the core facility (Department of Biochemistry, University of Alberta). The sequence showed no homology to other sequences in the NCBI data base; nor was the siCT␤2 oligonucleotide homologous to any other mRNA listed in GenBank TM . The siCT␤2 oligonucleotide was inserted into pSILENCER 4.1CMV (Ambion, Austin, TX), and the plasmid was named pSiCT␤2. Insertion of the cDNA was confirmed by sequencing.
Cloning and Hemagglutinin Modification of CT␤2 cDNA-A cDNA encoding the entire open reading frame of CT␤2 was cloned from a cDNA population generated from PC12 cell mRNA using the following primers: 5Ј-GCCATGCCAGTAG-TTACCACT-3Ј (forward) and 3Ј-GCTAAGGTTTGTGTGG-GTTGTC-5Ј (reverse). The amplicon was inserted into TOPO 2.1 vector (Invitrogen) and used as a template for the addition of an HA tag to the 3Ј-end of the open reading frame. The following sequence encoded the HA tag: 3Ј-TCAAGCATAATCTG-GAACATCATATGGATACTTCTCATCCTCATCCCCCT-CACTCAT-5Ј. The amplicon was cloned into the mammalian expression vector pCI (Invitrogen) and named pCI[CT␤2-HA]. Expression of CT␤2-HA protein was confirmed by immunoblotting of proteins from lysates of cells using anti-HA and anti-CT␤2 antibodies.
Other Methods-Protein concentrations were measured using the Bio-Rad protein assay with bovine serum albumin as a standard. LY294002 and roscovitine were purchased from Sigma and dissolved in dimethyl sulfoxide.

CT␤2 Is Present in Axons of Sympathetic Neurons and Retinal
Ganglion Cells-In light of our previous observation that NGF increases the expression of CT␤2 in PC12 cells (3), we reasoned that the subcellular localization of CT isoforms within neurons might provide insight into their contributions to PC biosynthesis and neurite growth. Our laboratory has previously demonstrated that PC biosynthetic enzyme activities, including CT, are present in distal axons of rat sympathetic neurons (4). The finding that in most cell types CT␣ is predominantly nuclear, whereas CT␤ is cytoplasmic, suggested that CT␤2, rather than CT␣, is a functionally important CT isoform in axons. We therefore immunoprecipitated CT from PC12 cells, rat sympathetic neurons, and retinal ganglion cells using an antibody (anti-M) raised against the membrane domain of rat liver CT. We confirmed that CT␤2 was immunoprecipitated by the anti-M antibody by immunoblotting the immunoprecipitated proteins from retinal ganglion cells with anti-CT␤2 antibodies ( Fig. 1). Based on these immunoblots, and consistent with the molecular mass of CT␤2 (43 kDa), we identified the upper band in Fig. 1 as CT␤2. Immunoblotting of the immunoprecipitated proteins in the absence of primary antibody identified the other major band in Fig. 1 as the IgG heavy chain of the anti-M antibody. CT␤2 was also detected in PC12 cells and rat sympathetic neurons ( Fig. 1, right-hand panels).
To determine whether or not axons contain CT␤2, we immunoprecipitated CT from distal axons of compartmented cultures of sympathetic neurons and retinal ganglion cells. In this culture system, cell bodies and proximal axons are physically separated from distal axons by a silicone grease barrier and a Teflon divider (17). Thus, distal axons can be harvested independently from cell bodies/proximal axons. As shown in Fig. 1 (left), immunoblotting of anti-M-immunoprecipitated proteins with anti-M or anti-CT␤2 antibodies revealed that CT␤2 is abundant in both distal axons and cell bodies/proximal axons of sympathetic neurons and also in distal axons of retinal ganglion cells. These immunoblotting experiments demonstrate that CT␤2 is present in PC12 cells as well as in distal axons of rat sympathetic neurons and rat retinal ganglion cells.
CT␤2 Facilitates Sprouting of Neurites and Branch Formation-Since CT␤2 is present in axons ( Fig. 1), and since the amount of CT␤2, but not CT␣, is increased during neurite outgrowth and/or extension (3), we considered that CT␤2 might regulate neurite growth. We therefore hypothesized that suppression of CT␤2 expression in PC12 cells would impair neurite growth. To test this hypothesis, we used a RNA silencing strategy to "knock down" expression of CT␤2 in PC12 cells. PC12 cells were co-transfected with a GFP-encoding mammalian expression vector (phrGFP) and either a vector encoding small interfering RNA targeted to CT␤2 mRNA (pSiCT␤2) or a vector containing a scrambled insert (designated "control"). The transfected cells were incubated with NGF to induce differentiation and neurite outgrowth. We estimated that transfection efficiency was ϳ30% according to the proportion of cells that expressed GFP in several random fields of view. In pSiCT␤2-transfected cultures, the amount of CT␤2 protein, as detected by immunoblotting, was reduced compared with that in control cells ( Fig. 2A). Immunoblotting of tubulin demonstrated that protein synthesis in general was not suppressed ( Fig. 2A). We also measured the in vitro activity of CT in cellular homogenates. This activity represents the combined activities of CT␣ and CT␤. In pSiCT␤2-transfected cells, CT activity was 2.9 nmol/min/mg protein, whereas CT activity of control cells was 4.3 nmol/min/mg protein (Fig. 2B). A trend (p Ͻ 0.09) toward reduced CT activity (ϳ30%) in CT␤2-silenced cells was observed. However, since CT activity reflects the activities of both CT␣ (the predicted major isoform of CT) and CT␤, and since RNA interference-silencing of CT␤2 was operative in only ϳ30% of the cells, it is likely that CT␤2 activity was reduced in the SiCT␤2-transfected cells.
Consistent with the lack of reduction of total CT activity in SiCT␤2-transfected cells, the incorporation of [ 3 H]choline into PC in PC12 cells over a 3-h time period was not significantly reduced (4.69 ϫ 10 6 Ϯ 0.37 ϫ 10 6 dpm/mg of protein for control and 4.41 ϫ 10 6 Ϯ 0.19 ϫ 10 6 dpm/mg of protein for SiCT␤2-transfected cells).
To determine if suppression of CT␤2 expression altered the morphology of NGF-treated PC12 cells, the cells were co-transfected with phrGFP and pSiCT␤2, and the morphologies of cells transfected with pSiCT␤2 and those transfected with the empty vector were compared (Fig. 3). Transfected cells were identified by the presence of GFP fluorescence. Within 2 days of NGF treatment, the number of neurites (primary neurites plus branches) per cell was markedly less in CT␤2-deficient cells than in cells transfected with pSILENCER containing a scrambled insert or in control cells transfected with empty vector. After 4 days of NGF treatment, differences in the number of neurites per cell, the degree of neurite branching, and neurite length between SiCT␤2-transfected cells and control cells were more pronounced (Fig. 3, A and B versus C and D). Moreover, the neurites of siCT␤2-expressing cells grew in a more linear fashion, with fewer points of attachment to the collagen substratum and fewer directional changes.  JANUARY 4, 2008 • VOLUME 283 • NUMBER 1

JOURNAL OF BIOLOGICAL CHEMISTRY 205
To quantify differences in numbers of primary neurites and branches, we counted the total number of primary neurites and branches per cell (Fig. 4A) in control and CT␤2-deficient cells over a 5-day period of differentiation. After 2 days of NGF treatment, control cells contained 34% more neurites per cell than did CT␤2-deficient cells (Fig. 4B) (3.09 Ϯ 0.06 versus 2.03 Ϯ 0.07 (p Ͻ 0.05)). During NGF treatment, control cells steadily produced primary neurites and branches. After 5 days of NGF treatment, control cells contained 5.41 Ϯ 0.35 neurites/cell. In contrast, CT␤2-deficient cells had only marginally increased the number of neurites between day 2 (2.03 Ϯ 0.07 neurites/ cell) and day 5 (2.60 Ϯ 0.27 neurites/cell). As a result, after 5 days of exposure to NGF, CT␤2-deficient cells contained fewer than half the neurites of control cells (2.60 Ϯ 0.27 versus 5.41 Ϯ 0.35 neurites/cell (p Ͻ 0.003)) (Fig. 4B).
Since primary neurites and branches are distinct neurite populations (Fig. 4A) that are regulated by separate signaling pathways (22), we determined if a decrease in the amount of CT␤2 affected each of these neurite populations. Within 2 days of NGF treatment, the extent of neurite sprouting was significantly less in CT␤2-deficient cells than in control cells (2.52 Ϯ 0.09 versus 1.83 Ϯ 0.01 neurites/cell (p Ͻ 0.01)) (Fig. 4C). The difference in the number of primary neurites was still apparent after 4 days of NGF-induced differentiation (control versus siCT␤2-expressing cells: 3.48 Ϯ 0.12 versus 2.46 Ϯ 0.18 neurites/cell (p Ͻ 0.01)) (Fig. 4C). Furthermore, CT␤2-deficient cells had far fewer branches than did control cells after 2 and 4 days of NGF treatment (Fig. 4D). Thus, CT␤2 deficiency impairs neurite sprouting and branching.
CT␤2 Is Not Required for Neurite Extension-When a primary neurite has sprouted from the cell body, either the neurite continues to elongate or neurite extension pauses for branch formation. Since CT␤2-deficient PC12 cells contained fewer neurites (both primary neurites and branches) than did control cells (Figs. 3 and 4), we measured the lengths of primary neurites and branches to determine whether or not CT␤2 is required for neurite extension. In control cells, mean neurite length (including both primary neurites and branches) did not increase between days 2 and 5: 84 Ϯ 1.5 pixels at day 2 and 100 Ϯ 37 pixels at day 5 (Fig. 5A). Typically, once a neurite attained a length of ϳ80 pixels, it had produced at least one branch, and further elongation of the primary neurite was limited. However, extension of primary neurites of CT␤2-deficient cells continued throughout the period of NGF treatment. Consequently, by day 5, neurites of CT␤2-deficient cells were 2.5 times longer than those of control cells (245 Ϯ 1.4 versus 100 Ϯ 37 pixels (p Ͻ 0.05)). In addition to the striking increase in length and lack of branches, neurites of pSiCT␤2-transfected cells also differed morphologically from neurites of control cells. Compared with neurites of control cells, neurites of CT␤2-deficient cells extended in straight lines with far fewer points of attachment to the collagen-coated dishes or regions of localized thickening within the neurite (Fig. 3).
A potential explanation for why silencing of CT␤2 expression in PC12 cells decreased the number of neurites is that the synthesis and availability of PC were reduced. To investigate this possibility, we calculated the total length of all neurites (i.e. the sum of the length of all primary neurites and branches) in control cells and CT␤2-deficient cells. After 2 and 4 days of NGF treatment, the total length of neurites was not different between control and CT␤2-deficient cells (day 2, 257 Ϯ 5 pixels/cell versus 315 Ϯ 41 pixels/cell, respectively; day 4, 580 Ϯ 89 pixels/cell and 713 Ϯ 86 pixels/cell, respectively) (Fig. 5B). Thus, the total length of all neurites per cell and, presumably,  the total amount of membrane were the same in control and siCT␤2-expressing cells, implying that reduction in the amount of CT␤2 does not impair neurite elongation.
Overexpression of CT␤2 in PC12 Cells Does Not Stimulate Neurite Sprouting or Branch Formation-Since siRNA-mediated reduction of CT␤2 expression in PC12 cells significantly decreased total neurite number, we investigated whether or not overexpression of CT␤2 increased neurite sprouting and branch formation. We therefore generated a cDNA encoding HA-tagged CT␤2 (CT␤2-HA) and cloned it into a mammalian expression vector. Expression of CT␤2-HA protein in PC12 cells was verified by immunoblotting of lysates of CT␤2-HAtransfected cells with anti-HA antibodies. Fig. 6A shows that a 43-kDa, HA-containing protein corresponding to the predicted size of CT␤2-HA was expressed in CT␤2-HA-transfected cells. Accordingly, CT activity in lysates of CT␤2-HA-transfected cells was significantly (30%) higher than in control cells (1.30 Ϯ 0.09 nmol/min/mg protein versus 1.00 Ϯ 0.01 nmol/min/mg protein, respectively (p Ͻ 0.05)) (Fig. 6B). To determine if CT␤2 overexpression altered neurite morphology, we co-transfected PC12 cells with a GFP-encoding vector to select and visualize cells that expressed CT␤2-HA. After 2 days of NGF treatment, the cells were visualized by fluorescence microscopy, and numbers of primary neurites and branches were determined. CT␤2-HA-expressing cells contained 4.9 Ϯ 0.6 neurites/cell, whereas control cells contained 3.7 Ϯ 0.4 neurites/cell (Fig. 6C). Although this difference is not statistically significant (p Ͻ 0.15), there is a trend toward an increased number of neurites in CT␤2-HA-expressing cells. Furthermore, enhanced expression of CT␤2 did not significantly increase the number of primary neurites (3.05 Ϯ 0.25 neurites/cell in control cells versus 3.57 Ϯ 0.16 neurites/cell (p Ͻ 0.15) in CT␤2-HA-transfected cells). Despite a strong trend toward increased number of branches, overexpression of CT␤2 did not significantly increase the number of branches (0.66 Ϯ 0.31/cell in control cells versus 1.38 Ϯ 0.30/cell in CT␤2-HA-transfected cells (p Ͻ 0.17)). Thus, whereas suppression of CT␤2 expression dramatically reduced the number of primary neurites and branches (Fig. 4), overexpression of CT␤2 did not significantly increase the number of primary neurites or branches (Fig. 6). Since silencing of CT␤2 expression did not reduce total neurite length (Fig. 5B), we hypothesized that the amounts of membrane phospholipids in PC12 cells would be unaffected by alterations in CT␤2 expression. Indeed, the mass (nmol of phospholipid/mg of cell protein) of PC, phosphatidylethanolamine, sphingomyelin, lysophosphatidylcholine, and phosphatidylserine was not significantly altered by overexpression of CT␤2 (CT␤2-HA-transfected cells) or in cells in which CT␤2 expression was reduced (i.e. pSiCT␤2-transfected cells (Fig. 7). Thus, the total amount of phospholipids in PC12 cells was not altered by either a decrease or an increase in the level of CT␤2. Taken together, these data demonstrate that CT␤2 is required for normal neurite branching and sprouting but not for neurite extension.
CT␤2 Is Required for LY294002-dependent Stimulation of Neurite Outgrowth-We investigated the mechanism by which reduction of CT␤2 impairs neurite sprouting and branching. Activation of Akt has been proposed to inhibit neurite branching in PC12 cells, since treatment of the cells with LY294002, an inhibitor of phosphatidylinositol 3-kinase, abolished Akt activation and promoted neurite branching (16). Because our data suggested that CT␤2 plays a role in regulating neurite outgrowth and branch formation, we hypothesized that CT␤2 might be a downstream target of the phosphatidylinositol-3kinase/Akt signaling cascade. To test this hypothesis, we compared the effect of LY294002 on neurite outgrowth and branch formation in control PC12 cells and in PC12 cells in which CT␤2 expression had been reduced. Immunoblotting of cellular proteins with anti-phospho-Akt antibodies confirmed that LY294002 (10 M) reduced the phosphorylation of Akt (Fig.  8A). Consistent with a previous report (16), PC12 cells incubated with LY294002 contained more neurites than did untreated cells (5.4 Ϯ 0.2 neurites/cell for LY294002-treated cells versus 3.1 Ϯ 0.06 neurites/cell for control cells (p Ͻ 0.01)) (Fig. 8B). In sharp contrast, however, decreased expression of CT␤2 in pSiCT␤2-transfected cells blocked the increase in the total number of neurites that was observed in control cells treated with LY294002 (Fig. 8B); CT␤2-deficient cells in the absence of LY294002 contained 2.05 Ϯ 0.16 neurites/cell, and LY294002-treated siCT␤2-transfected cells contained 1.95 Ϯ 0.02 neurites/cell.
We also counted the number of primary neurites and branches individually to determine if LY294002 specifically affected one of these neurite subpopulations. In control cells, LY294002 increased the number of primary neurites/cell by 75% (p Ͻ 0.01) (Fig. 8C) and increased the number of branches by 81% (from 0.57 Ϯ 0.10 to 1.03 Ϯ 0.12 (p Ͻ 0.05)) (Fig. 8D). In CT␤2-deficient cells, on the other hand, LY294002 did not increase the number of primary neurites (Fig. 8C). Untreated CT␤2-deficient cells contained 2.05 Ϯ 0.16 primary neurites/ cell, whereas LY294002-treated CT␤2-deficient cells contained 1.96 Ϯ 0.02 primary neurites/cell. Whereas CT␤2-deficient cells contained far fewer branches/cell than did control cells (0.19 Ϯ 0.03 versus 0.57 Ϯ 0.10) (Fig. 8D), LY294002 increased the number of branches by 63% (from 0.19 Ϯ 0.03 to 0.31 Ϯ 0.01 branches/cell (p Ͻ 0.05)). Nevertheless, the number of branches remained much lower than in control cells. Thus,  these data strengthen the conclusion that CT2 is important for neurite branching.
Phosphorylation of CT␤2-More than one CT isoform contributes to CT activity in PC12 cells. We therefore assessed whether or not treatment with LY294002 directly affected the phosphorylation status of CT␤2. Since the phosphatidylinositol 3-kinase inhibitor, LY294002, did not stimulate neurite sprouting and only slightly increased branching in CT␤2-deficient cells, we speculated that CT␤2 might be a downstream target of the phosphatidylinositol 3-kinase/Akt signaling pathway. CT␤2 is heavily phosphorylated in vivo (7). However, studies on the kinases and phosphatases that act on CT␤2 have not been reported. In HeLa cells, both insulin and epidermal growth factor can stimulate CT phosphorylation (23), and HeLa cells express both CT␣ and CT␤ mRNAs (7). Thus, it is likely that CT␤2 is phosphorylated in response to growth factors, including NGF.
From the amino acid sequence of CT␤2, we identified a putative Akt phosphorylation motif (RSRSPS) within the carboxyl terminus of CT␤2. To determine if CT␤2 was a substrate for phosphorylation by Akt, we incubated differentiated PC12 cells with LY294002 or an equivalent volume of vehicle for 48 h. The cells were then incubated with [ 32 P]orthophosphate for 1 h, and CT␤2 was immunoprecipitated and separated by SDS-PAGE. As shown in Fig. 8A, treatment of PC12 cells with LY294002 reduced the phosphorylation of Akt but did not significantly alter the 32 P labeling of CT␤2 (Fig. 9). Thus, in NGF-differentiated PC12 cells, CT␤2 is unlikely to be directly regulated by Akt phosphorylation.
Another candidate kinase for phosphorylation of CT␤2 is Cdk5 (24). Cdks comprise a family of proline-directed threonine and serine kinases that orchestrate transitions through the cell cycle. Cdk5 has 60% sequence identity to other Cdks and is the sole isoform that is highly enriched in brain (25). Cdk5 is required for axon growth (26) and is activated during regeneration of facial nerve axons after injury (27). CT␤2 has a putative Cdk5 phosphorylation consensus motif (SPSR) within its carboxyl terminus. We therefore investigated whether or not CT␤2 was directly phosphorylated by Cdk5. CT␤2 was immunoprecipitated from homogenates of PC12 cells and used as a substrate for Cdk5 phosphorylation in an in vitro kinase assay. We used tubulin as a negative control, because it lacks a Cdk5 phosphorylation motif and is not known to be phosphorylated by Cdk5. As shown in Fig. 10A, CT␤2 was phosphorylated in vitro by Cdk5, whereas tubulin was not.
Since Cdk5 phosphorylates CT␤2 in vitro, we next determined if Cdk5 also phosphorylated CT␤2 in intact cells. If CT␤2 were phosphorylated by Cdk5, incubation of PC12 cells with roscovitine, a Cdk inhibitor, would be expected to diminish Cdk5-dependent 32 P labeling of CT␤2. To test this hypothesis, PC12 cells were incubated with and without roscovitine for 1 h prior to labeling of the cells with [ 32 P]orthophosphate. CT␤2 was immunoprecipitated from cell lysates with anti-M antibodies, and the incorporation of [ 32 P] into CT␤2 was assessed. Fig. 10B shows that less 32 P was incorporated into CT␤2 in PC12 cells treated with roscovitine than in untreated cells. These observations suggest that CT␤2 is a substrate for Cdk5. In addition, roscovitine treatment of PC12 cells markedly reduced neurite branching (Fig. 10C), suggesting a link between reduced phosphorylation and activity of CT␤2 and inhibition of neurite branching. At the concentration of roscovitine used for these experiments (25 M for 48 h), PC12 cell viability was unaffected. For example, for six fields counted for each condition, the number of cells per field was the same for control cells (39.0 Ϯ 12.6) as for roscovitine-treated cells (37.4 Ϯ 9.3).

DISCUSSION
We investigated the role of CT␤2, an isoform of CT (the rate-limiting enzyme in PC biosynthesis) in neurite sprouting, extension, and branching of PC12 cells. We show that CT␤2 is abundant in PC12 cells, as well as in axons of rat sympathetic neurons and rat retinal ganglion cells. When the amount of CT␤2 protein was reduced in PC12 cells by RNA silencing, neurite branching, the total number of neurites per cell, and the number of primary neurites per cell were markedly diminished, whereas the total length of neurites was unchanged. These data demonstrate that CT␤2 plays a role in neurite sprouting and branching but not in neurite elongation. Our studies also indicate that CT␤2 is required for the stimulation of neurite outgrowth that occurs when the phosphatidylinositol-3-kinase/ FIGURE 9. CT␤2 is not phosphorylated by Akt. PC12 cells were incubated overnight with 50 ng/ml NGF and then for 48 h with 50 ng/ml NGF and 10 M LY294002 (LY294002) or an equivalent volume of vehicle (Control). The cells were then labeled with 100 mM [ 32 P]orthophosphate for 1 h. CT␤2 was immunoprecipitated from cell lysates and separated by SDS-PAGE, and the gel was exposed to a phosphor imager screen (top). The same gel was stained with Coomassie Blue and shows IgG at ϳ50 kDa, indicating equivalent protein loading on the gel (middle). Shown is one experiment representative of two independent experiments with similar results. Band intensities were measured by densitometry (bottom) and are expressed as arbitrary units relative to abundance of the IgG band and compared with values from untreated control cells. Values are averages from two independent experiments with error bars representing variation from the mean.
Akt signaling pathway is inactivated. We show that CT␤2 is phosphorylated by Cdk5 and that inhibition of Cdk5 by roscovitine reduces neurite branching, consistent with a link between phosphorylation and activity of CT␤2 and neurite branching.
CT␤2 Is Abundant in Axons and Facilitates Neurite Sprouting and Branch Formation-Phospholipids are not synthesized exclusively in cell bodies of neurons (12), and enzymes of the Kennedy pathway for PC biosynthesis have been detected in distal axons of rat sympathetic neurons (4) and in rat brain synaptosomes (10). Indeed, at least 50% of PC in axonal membranes is biosynthesized within axons (28). Moreover, when distal axons, but not cell bodies, of sympathetic neurons are treated with inhibitors of PC biosynthesis, axonal growth is significantly impaired (28). The biosynthesis of phosphatidylethanolamine, phosphatidylserine, and sphingomyelin has also been detected in axons of rat sympathetic neurons (4). Furthermore, the presence of GM3 synthase, a ganglioside-specific sialyltransferase, has been reported in axons of hippocampal neurons (29). Interestingly, however, not all lipid classes are made in distal axons, since cholesterol biosynthesis appears to be restricted to cell bodies/proximal axons (4).
The most pronounced expression of CT␤2 mRNA in the mouse is in the brain (30). However, since ϳ90% of the cells in the brain are glial cells, not neurons (reviewed in Ref. 31), we do not know the relative level of expression of CT␤2 in neurons compared with other types of cells in the brain. We now show that CT␤2 is abundant in PC12 cells and in distal axons of sympathetic neurons (peripheral neurons), and retinal ganglion cells (central nervous system neurons). Based on these findings, we proposed that axonal synthesis of PC via CT␤2 would be important for some aspects of neurite growth, such as sprouting, branching, and/or extension. We, therefore, suppressed CT␤2 expression in differentiating PC12 cells. Silencing of CT␤2 expression strikingly altered neurite morphology and reduced neurite sprouting and branch formation. In contrast, the total combined length of all neurites was not decreased. As an additional test of whether or not CT␤2 regulates neurite sprouting and branching, we expressed HA-tagged CT␤2 in differentiating PC12 cells. The amount of CT␤2 protein and CT activity were increased, but the number of neurites was not significantly increased. Thus, either the residual amount of CT␤2 and CT␣ in PC12 cells can together produce sufficient PC for neurite sprouting and branching or, alternatively, CT␤2 is not the sole requirement for these processes. Several reports have identified proteins in PC12 cells, including the microtubule-binding protein, raspotlin, and the signaling molecule, Raf, that are required for proper branching of neurites (32). Like CT␤2, rapostlin promotes branch formation, and in its absence PC12 cells produce fewer branches.
CT␤2 Is Required for LY294002-stimulated Neurite Outgrowth-Successful neuritogenesis requires a delicate balance between neurite extension and branch formation (33). In CT␤2-deficient PC12 cells, we observed a dramatic decrease in neurite branching but an increase in the length of individual primary neurites. Time lapse recording of axon extension in sensory neurons has shown that axon growth pauses for up to several hours at the site of future branch points (34). During this pause, microtubules become splayed and disorganized at the future branch points. Eventually, the microtubular network reorganizes and short microtubules project into the developing neurite (reviewed in Ref. 33). When an axon or neurite is unable to recognize signals for pausing and reorganization for branch formation, uninterrupted growth of the neurites results in production of abnormally long neurites. This phenomenon has also been observed in PC12 cells that lack Akt, a serine/threonine kinase known to promote neurite growth and inhibit branch formation. PC12 cells lacking Akt contain significantly fewer neurites than do control cells that contain Akt, but the neurites extend greater distances than in control cells that contain Akt (16). Thus, inactivation of Akt appears to facilitate neurite formation and/or branching.
LY294002 is a phosphatidylinositol 3-kinase inhibitor that prevents the activation of Akt (16). Incubation of control PC12 cells with LY294002 increased the number of neurites per cell, both primary neurites and branches. However, in CT␤2-deficient PC12 cells, LY294002 failed to increase the number of primary neurites (i.e. sprouting). Thus, CT␤2 is required for the FIGURE 10. CT␤2 is phosphorylated by Cdk5. A, PC12 cells were differentiated in the presence of NGF for 4 days, and an in vitro kinase assay was performed. CT␤2, tubulin, and Cdk5 were immunoprecipitated from cell lysates with anti-M, anti-tubulin, and anti-Cdk5 antibodies, respectively. Immunoprecipitated Cdk5 was incubated with either tubulin or CT␤2 in the presence of [ 32 P]ATP for 15 min. Proteins were separated by SDS-PAGE and exposed to a phosphor imager screen for 3 days. B, PC12 cells were incubated overnight with 50 ng/ml NGF and then for 48 h with 50 ng/ml NGF and 25 M roscovitine or an equivalent amount of vehicle. Cells were transferred for 1 h to phosphate-free medium to which 100 mM [ 32 P]orthophosphate had been added. CT␤2 was immunoprecipitated. Equal amounts of proteins from cell lysates were separated by SDS-PAGE and then exposed to a phosphor imager screen for 5 days. Results in A and B are representative of three independent experiments with similar results. C, PC12 cells were incubated overnight with 50 ng/ml NGF and then for 48 h with 50 ng/ml NGF and 25 M roscovitine or an equivalent volume of vehicle. Cells were viewed by phase-contrast microscopy. At least six fields of view were examined for each condition and showed similar results. promotion of sprouting induced by inhibition of the phosphatidylinositol 3-kinase/Akt signaling pathway.
Within its carboxyl terminus, CT␤2 contains a putative Akt phosphorylation motif, RSRSPS. We therefore determined whether or not CT␤2 is phosphorylated by Akt in intact cells. LY294002 treatment of PC12 cells did not alter the 32 P labeling of CT␤2 (Fig. 9B), suggesting that CT␤2 is not directly regulated by phosphorylation via Akt.
NGF is known to activate many signaling pathways in neurons (35). Cdk5 and its co-activator, p35, are abundant in the brain and are among the proteins activated by NGF (36); these proteins have been strongly implicated in axon growth and recovery after injury (26,27), and Cdk5 appears to be involved in efficient cytoskeletal remodeling. Since CT␤2 contains a Cdk5 phosphorylation consensus motif (SPSR) within its carboxyl terminus, we hypothesized that Cdk5 directly phosphorylates CT␤2. Indeed, we demonstrated that Cdk5 can phosphorylate CT␤2 in vitro and in intact cells (Fig. 10). In addition, short term treatment of PC12 cells with roscovitine, a Cdk inhibitor, markedly decreased CT␤2 phosphorylation and neurite branching in intact cells. CT␤2 and Cdk5 (38) are both present within axons. Several lines of evidence suggest that Cdk5 promotes axon extension and impairs axon branching. In Cdk5-null mice, motor axons exhibit extensive anomalous branching patterns (37). Furthermore, inhibition of Cdk5 with roscovitine attenuates axon elongation (38). Thus, we propose that when Cdk5 is activated, CT␤2 is phosphorylated and activated, and consequently, neurite outgrowth and branching are promoted.
Phosphatidylcholine Synthesis and Neurite Formation-To our knowledge, we have identified the first lipid biosynthetic enzyme that facilitates neurite sprouting and branch formation.
Marszalek et al. (39) reported that overexpression of a fatty acid-metabolizing enzyme, acyl-CoA synthetase-2, significantly increases neurite extension in PC12 cells. However, the numbers of neurites and branches were unchanged. It is possible that CT␤2 provides a locally synthesized pool of PC in axons at future branch points. Consistent with this idea, when axons of Aplysia sensory neurons were injured, the axonal membrane, rather than the cell body, provided lipid for resealing the membrane and for the rapid growth following axotomy (40).
In addition to providing new membrane material for neurite sprouting and branching, CT␤2 might replace PC that has been degraded in axons. Phospholipase D hydrolyzes PC (41) and is present in PC12 cells (42). Neurite outgrowth in PC12 cells requires activation of phospholipase D via the MAP kinase signaling cascade (43). In our previous study (2), we demonstrated that inhibition of the mitogen-activated protein kinase pathway by U0126 decreased CT activity, consistent with a role for mitogen-activated protein kinase in activation of CT␤2. Thus, CT␤2 might replenish the pool of PC that has been hydrolyzed by phospholipase D. Alternatively, or in addition, CT␤2 might synthesize PC that provides phosphocholine for sphingomyelin synthesis. Sphingomyelin can be made in distal axons of sympathetic neurons (4) and is a component of detergent-insoluble glycolipid complexes in axonal membranes (44). Last, we cannot rule out a role for CT␤2 in neurite outgrowth and/or branching that is independent of its catalytic role in PC synthesis.
The CDP-choline pathway for PC biosynthesis is, under most metabolic conditions, including in NGF-stimulated rat sympathetic neurons (5), regulated by CT activity. In PC12 cells, PC biosynthesis has also been reported to be regulated by the availability of diacylglycerol and by the activity of cholinephosphotransferase, the enzyme that catalyzes the final step of the CDPcholine pathway (2). Furthermore, cytidine and uridine (the product of cytidine deamination) can enhance PC biosynthesis in PC12 cells (45) and rat brain slices (46). Interestingly, treatment of PC12 cells with uridine increased CDP-choline levels and enhanced neurite outgrowth and branch formation in a dose-dependent manner (47). This enhancement of neurite outgrowth was attributed, at least in part, to an increased amount of CDP-choline (47). Thus, it appears that both the amount of CT␤2 protein and substrate availability can regulate PC biosynthesis during PC12 cell differentiation.
In summary, we provide evidence that CT␤2 facilitates neurite outgrowth and branching. Since CT␤2 appears to be involved in neurite branch formation and is abundant in the brain, we predict that a complete deficiency of CT␤2 might result in subtle neurological defects, perhaps in learning and memory or motor coordination, rather than gross neuroanatomical deficits.