Nuclear Localization of Enzymatically Active Green Fluorescent Protein-CTP:Phosphocholine Cytidylyltransferase α Fusion Protein Is Independent of Cell Cycle Conditions and Cell Types*

To address the recent controversy about the subcellular localization of CTP:phosphocholine cytidylyltransferase α (CTα), this study was designed to visualize green fluorescent protein (GFP)·CTα fusion proteins directly and continuously under different conditions of cell cycling and in various cell lines. The GFP·CTα fusion proteins were enzymatically active and capable of rescuing mutant cells with a temperature-sensitive CT. The expressed GFP·CTα fusion protein was localized to the nucleus in all cell lines and required the N-terminal nuclear targeting sequence. Serum depletion/replenishment did not cause shuttling of CTα between the nucleus and cytoplasm. Moreover, the subcellular localization of CTα was examined continuously through all stages of the cell cycle in synchronized cells. No shuttling of CTα between the nucleus and cytoplasm was observed at any stage of the cell cycle. Stimulation of cells with oleate had no effect on the localization of CTα. The GFP·CTα lacking the nuclear targeting sequence stayed exclusively in the cytoplasm. Regardless of their localization, the GFP·CTα fusion proteins were equally active for phosphatidylcholine synthesis and mutant rescue. We conclude that the nuclear localization of CTα is a biological event independent of cell cycle in most mammalian cells and is unrelated to activation of phosphatidylcholine synthesis.

To address the recent controversy about the subcellular localization of CTP:phosphocholine cytidylyltransferase ␣ (CT␣), this study was designed to visualize green fluorescent protein (GFP)⅐CT␣ fusion proteins directly and continuously under different conditions of cell cycling and in various cell lines. The GFP⅐CT␣ fusion proteins were enzymatically active and capable of rescuing mutant cells with a temperature-sensitive CT. The expressed GFP⅐CT␣ fusion protein was localized to the nucleus in all cell lines and required the N-terminal nuclear targeting sequence. Serum depletion/replenishment did not cause shuttling of CT␣ between the nucleus and cytoplasm. Moreover, the subcellular localization of CT␣ was examined continuously through all stages of the cell cycle in synchronized cells. No shuttling of CT␣ between the nucleus and cytoplasm was observed at any stage of the cell cycle. Stimulation of cells with oleate had no effect on the localization of CT␣. The GFP⅐CT␣ lacking the nuclear targeting sequence stayed exclusively in the cytoplasm. Regardless of their localization, the GFP⅐CT␣ fusion proteins were equally active for phosphatidylcholine synthesis and mutant rescue. We conclude that the nuclear localization of CT␣ is a biological event independent of cell cycle in most mammalian cells and is unrelated to activation of phosphatidylcholine synthesis.
CTP:phosphocholine cytidylyltransferase (CT) 1 is the regulatory enzyme of the CDP-choline pathway for de novo synthesis of PC (1). CT catalyzes the second reaction between choline kinase and cholinephosphotransferase. The CDP-choline pathway is present in all mammalian cells and is essential for cell survival (2). Complete inactivation of CT by a temperaturesensitive mutation leads to apoptosis (3). CT␣ was first purified by Weinhold and co-workers in 1986 (4), and its cDNA was first cloned from rat liver by Cornell and co-workers in 1990 (5). PC synthesis is nearly normal even when over 90% of CT is inactivated (2), suggesting that most cellular CT remains inactive for PC synthesis. However, nearly all cellular CT␣ has been found localized in the nucleus (6) although the major site of PC synthesis was found associated with the endoplasmic reticulum (7)(8)(9). Because of this physical separation of CT␣ localization and PC synthesis, the precise localization of CT␣ and universality of CT␣ localization in the nucleus have become interesting topics for investigation.
Since the nuclear localization of CT␣ was first reported by Kent and co-workers (6), the subcellular localization of CT␣ has become increasingly controversial. A study by Houweling et. al (10) demonstrated that significant staining of CT was detected in both the cytoplasm and nucleus of primary rat hepatocytes and of rat liver thin slices. Adding to this complexity, two more isoforms of CT (CT␤1 and -␤2) without the N-terminal nuclear targeting signal were found in animal tissues, including liver, and were present in the cytoplasm (11). It is not known how the participation of CT␣ in the CDP-choline pathway is regulated, because the amino acid sequences responsible for several previously suspected mechanisms of CT␣ activation can be deleted without affecting its involvement in the CDP-choline pathway (12)(13)(14). In a recent study by Cornell and co-workers (15), a novel mechanism for CT␣ activation was proposed. Based on the observations that activation of PC synthesis during the G 0 to G 1 transition was accompanied by a translocation of CT␣ from the nucleus to the cytoplasm in cultured IIC9 cells, it was proposed that this translocation was a mechanism for CT␣ activation. Yet, this proposed mechanism of translocation from the nucleus to the cytoplasm might not be necessary if two other isoforms of CT are already present in the cytoplasm. Therefore, it inevitably becomes important to determine the specific localization of CT␣ in other mammalian cells and to verify whether the observation of such a cell cycle-dependent translocation between the nucleus and the cytoplasm could be extended to other mammalian cells.
In this study, we devised a new strategy of monitoring the subcellular localization of CT␣ directly and continuously in live mammalian cells at various stages of the cell cycle. This study was designed to address the following questions. 1) Is the nuclear localization of CT␣ a universal event and present in other mammalian cells? The subcellular localization of CT␣ was previously investigated in four types of mammalian cells: Chinese hamster ovary cells (CHO-K1 and MT58) (6,12,16), human hepatocarcinoma cells (HepG2) (6), mouse fibroblast cells (NIH3T3, L-cells, and IIC9) (6), and primary rat liver cells (6,10). Our current study extended observations into six additional cell lines; the study found no evidence of translocation but did find constitutive localization of CT␣ in the nucleus of all cell lines tested. 2) Is the CT␣ translocation between the nucleus and the cytoplasm in IIC9 cells during serum depletion/ replenishment (15) also present in other mammalian cells? We examined CT␣ localization in Chinese hamster ovary cells un-der similar conditions and found no evidence of translocation. 3) Is the subcellular localization of CT␣ changed during the synchronized movement of cells through all stages of the cell cycle? Because the subcellular localization of CT␣ in continuous stages of the cell cycle has never been addressed before, we examined the localization of CT␣ in synchronized cells at different stages of the cell cycle in Chinese hamster ovary cells. Additionally, the use of GFP⅐CT␣ fusion proteins for direct visualization in cells circumvented specificity problems inherent with indirect immunofluorescence detection. In this report, we describe the findings of this study.
Cell Culture-Rat primary hepatocytes were obtained by collagenase perfusion (17). Hepatocytes were cultured on collagen-coated culture dishes and incubated in DMEM with 10% FBS, 10 g/ml insulin, and 10 mM Hepes buffer at 37°C overnight prior to transfection. K1 and MT58 cells were cultured in F12 medium with 10% FBS at 33°C. HepG2 and sheep choroid plexus cells were cultured in Eagle's minimal essential medium with 10% FBS at 37°C. MCF-10A cells were cultured in DMEM/F12 with 10% FBS at 37°C. Other cells were cultured in DMEM with 10% FBS at 37°C.
CT⅐GFP Fusion Constructs-The cDNA for Aequorea victoria GFP (Life Technologies, Inc.) was cloned into the mammalian expression vector pCDNA3. For construction of GFP⅐CT-pCDNA3, the GFP-specific 5Ј and 3Ј primers with EcoN1 sequences were used to generate a polymerase chain reaction fragment of GFP with removal of the start and stop codons and addition of an EcoN1 restriction site to each end of the GFP coding region (EcoN1-GFP-EcoN1). Full-length cDNA for rat liver CT␣ in the mammalian expression vector pCDNA3 (CT-pCDNA3) was digested with EcoN1, which cuts at a unique site between residues 58 and 59 of CT␣, and treated with calf intestine alkaline phosphatase. The treated CT␣-pCDNA3 was ligated to the EcoN1-GFP-EcoN1 fragment, and correct orientation of the GFP insertion was confirmed with restriction mapping. This construction places GFP behind the nuclear targeting region (residues 8 -28) (12) of CT␣ but in front of the catalytic domain (residues 72-236) (19) of CT␣. For construction of GFP⅐CT ⌬N -pCDNA3, the GFP⅐CT-pCDNA3 construct was digested with XhoI restriction enzyme to obtain a template for removal of the nuclear targeting region (residues 1-40) by polymerase chain reaction. 5Ј-TCTAGAATGTTACGG CAGCCAGCTCCTTTTTCT primer and a 3Ј primer, both containing XbaI sequences, were used to remove nucleotides 1-120 of CT and to amplify the product. The XbaI restriction sites were used to clone the construct into the pCDNA3 vector.
Transfection of Cells-MT58 cells were transfected with 10 g of GFP-pCDNA3, GFP⅐CT-pCDNA3, and GFP⅐CT ⌬N -pCDNA3 plasmids by calcium phosphate precipitation as described (18). G418-resistant colonies were selected with 500 g/ml G418 and maintained in 200 g/ml G418.
Labeling of Cellular PC with [methyl-3 H]Choline Chloride-MT58, MT58 GFP, MT58 GFP⅐CT, and MT58 GFP⅐CT ⌬N cells were seeded into 6-well plates (3 ϫ 10 5 cells/well) at 33°C and incubated in F12 medium at 40°C for 4 h. Medium was replaced with 1 Ci [methyl-3 H]choline chloride in fresh F12 medium and incubated for 24 h at 40°C. Cells were washed twice with phosphate-buffered saline and harvested. Lipids were extracted by the method of Bligh and Dyer (20). The lipid extracts were dried under nitrogen and dissolved in scintillation mixture. Radioactivity counts were determined by scintillation counting.
Western Blot-Total proteins (50 g) from each transfected cell line were separated by SDS-polyacrylamide gel electrophoresis (21) and transferred to nitrocellulose membranes (22). The membranes were probed with anti-full-length GFP antibody, washed extensively, and probed with goat anti-rabbit IgG horseradish peroxidase-conjugated antibody. GFP protein was visualized by a reaction with Supersignal chemiluminescent substrate (Pierce) and exposure to x-ray films.
Fluorescence Microscopy-Cells transfected with GFP-pCDNA3, GFP⅐CT-pCDNA3, and GFP⅐CT ⌬N -pCDNA3 plasmids were observed after 48 h under a Zeiss Axioplan-2 epifluorescence microscope equipped with a fluorescence filter. Digital images of cells were recorded using a Spot camera.
Determination of Cell Rescue of MT58 at 40°C-MT58 cells and transfected MT58 cells were seeded into 12-well plates and incubated in F12 medium at 33°C for 12 h and shifted to 40°C. Cells were harvested at desired time points and counted.
Serum Depletion/Replenishment-MT58 cells (3 ϫ 10 5 ) stably transfected with GFP⅐CT-pCDNA3 were seeded into a 35-mm culture dish and incubated in serum-free Ham's F12 medium for 36 h at 33°C. The medium was replaced with Ham's F12 medium with 10% fetal bovine serum and incubated at 33°C. The localization of GFP⅐CT fusion protein was examined with a fluorescence microscope at 2-h intervals. Cells were analyzed by flow cytometry.
Synchronization of MT58 GFP⅐CT Cells-Cells (7 ϫ 10 6 ) were plated into a 150-mm culture dish in Ham's F-12 medium with 10% fetal calf serum and incubated for 1 h at 33°C. The cell monolayer was washed and then incubated in Ham's F-12 medium with 0.25 g/ml nocodazole for 16 h at 33°C. The cells blocked at the G 2 /M border were either unattached or lightly attached to the dish. These cells were collected by gentle washing with the culture medium, transferred to a centrifuge tube, and spun at 1000 rpm for 5 min. The cell pellet was washed twice with medium, and the cells were resuspended in medium. 3 ϫ 10 5 cells were plated in 35-mm culture dishes and incubated at 33°C for 1 h. The medium was replaced with medium containing 2 mM hydroxyurea and further incubated for 9 h. The medium was replaced with fresh F12 and incubated for various time periods, after which the cells were visualized by microscopy or harvested for flow cytometric analysis.
Flow Cytometry-Cells were prepared for flow cytometric analysis by fixing cells (5 ϫ 10 5 ) in ethanol and suspending cells in 50 g/ml propidium iodide, 0.6% Nonidet P40, and 37 g/ml RNase. The stained cells were analyzed on a Coulter XL flow cytometer, which excites the cells at 488 nm and measures the red fluorescence per cell, which can be equated to the DNA content per nucleus.

RESULTS
The Expressed Fusion Proteins Were Intact-We made three constructs for transfections as diagrammed in Fig. 1. In the first construct, expression of GFP alone was used as a control for determination of transfection efficiency and for subcellular distribution of a protein without a specific targeting signal. GFP was distributed evenly throughout cellular organelles. In the second construct, the GFP coding region was inserted inframe between the nuclear targeting signal and the rest of CT␣ (between residues 58 and 59); this construct was designated GFP⅐CT. In the third construct, the nuclear targeting sequence of CT␣ was removed from the second construct; this third construct was designated GFP⅐CT ⌬N . The expression plasmids were introduced into the cultured MT58 and K1 cells to verify whether the fusion proteins were intact. Transfections were FIG. 1. Construction of GFP⅐CT fusion proteins. For construction of GFP⅐CT, A. victoria green fluorescent protein coding region was inserted after the nuclear targeting signal of rat CT (between residues 58 and 59). For construction of GFP⅐CT ⌬N , residues 1-40, which contain the nuclear targeting sequence (residues 8 -28), were removed from GFP⅐CT. GFP alone was used as control. All three constructs were inserted into the pCDNA3 vector for expression in mammalian cell lines.
achieved by calcium phosphate precipitation, and the stably transfected cells with green fluorescence were selected in the presence of 500 g/ml G418. Specific localization of the fusion proteins depended completely on the fusion proteins remaining intact after being expressed. Otherwise the localization of green fluorescence would not represent the localization of the fusion protein but rather of degradation fragments of the fusion protein. To verify the expression of the constructed fusion proteins, MT58 cells transfected with the plasmids were probed with specific antibodies against GFP by Western blot analysis. Fig. 2 depicts the presence of fusion proteins in the stably transfected cells. The GFP⅐CT construct produced a fusion protein with an apparent molecular mass (70 kDa) of the predicted fusion protein between CT (43 kDa) and GFP (27 kDa). The GFP⅐CT ⌬N construct produced a smaller fusion product corresponding to the removal of the 40 amino acid residues of the CT␣ nuclear targeting signal. There are two bands in MT58 cells that reacted nonspecifically to the GFP antibody and were not related to GFP. The control GFP construct produced a large amount of GFP protein at the predicted molecular mass. Because the efficiency of expressing GFP alone is much higher than that of larger fusion constructs, we had to reduce the amount of total protein loaded on the gel to 1/10 of the other lanes to achieve comparable exposures. Therefore, the nonspecific bands in MT58 cells did not show up in this lane. Nevertheless, both fusion proteins remained intact in the stably transfected cells, suggesting that the localization of green fluorescence represents the localization of the intact fusion protein.
CT⅐GFP Fusion Proteins Were Enzymatically Active-We then determined whether the fusion proteins were enzymatically active for PC synthesis. The stably transfected MT58 cells were shifted to 40°C to inactivate the endogenous CT activity. Cells were then labeled with 3 H-choline for 24 h, the organic phase of the labeled cells was extracted, and radioactive counts in each sample were determined. The only way for watersoluble 3 H-choline to get into the organic phase was to complete all three steps of the CDP-choline pathway. In a separate experiment, we determined that PC accounted for 96% of the radioactivity in the organic phase. Thus, the incorporation of 3 H-choline into the organic phase was a direct representation of the activity of the CDP-choline pathway and CT activity in MT58 cells at 40°C. Results of this experiment showed that both fusion constructs of GFP⅐CT␣ were enzymatically active (Fig. 3). We also determined by counting cells after incubation at 40°C that the fusion proteins were capable of rescuing MT58 cells at the non-permissive temperature. The results showed that both fusion proteins, but not GFP alone, restored the growth of MT58 cells at 40°C, with generation times similar to that of CT expression in MT58 cells (data not shown).
In summary, the GFP⅐CT␣ fusion proteins with or without the nuclear targeting sequence behaved indistinguishably from CT␣ alone in terms of enzyme activity and ability to rescue MT58 cells at 40°C. Thus, the localization of the fusion proteins should represent the localization of CT␣ very closely. These constructs were then expressed in cells from several selected origins to determine directly the localization of the fusion proteins in live cells. This design minimized the effects of nonspecificity often seen with antibody staining and cell fixation and represented cellular localization closer to physiological conditions.
Endogenous CT Did Not Affect the Localization of CT Fusion Proteins-After confirming that the fusion protein did retain the expected enzymatic activity and cellular function, we monitored the transfected MT58 cells by fluorescence microscopy. The main purpose of this design was to further confirm that the fusion constructs would behave similarly to the observation of CT␣ nuclear localization made originally by Kent and co-workers (12). A similar nuclear localization in MT58 cells would validate that the GFP⅐CT␣ fusion construct retains critical determinants for nuclear targeting and that its localization in other cell types is a close reflection of native CT␣ in the physiological state. GFP⅐CT fusion proteins were indeed localized exclusively to the nucleus of MT58 cells (Fig. 4A) at both permissive and nonpermissive temperatures. Upon removal of the N-terminal nuclear targeting sequence, GFP⅐CT ⌬N revealed a clear pattern of nuclear exclusion (Fig. 4A). These results indicated that the fusion constructs displayed patterns of subcellular localization identical to that of native CT␣ reported previously. Because all cell lines included in our intended survey contain endogenous CT, it was important to determine whether the nuclear localization of overexpressed fusion protein is affected by endogenous CT. We expressed the fusion constructs in K1 cells in which the level of endogenous CT is at least 20-fold higher than that of MT58 at the permissive temperature. Fig. 4B demonstrates that the patterns of the fusion proteins in K1 cells were identical to those of MT58 cells (Fig. 4A) at various levels of expression. This result suggests that the mechanism for nuclear retention is not saturable and that the localization of the expressed fusion protein was not affected by the endogenous CT.
Cell Line-independent Localization of CT in the Nucleus-One of the advantages of using GFP⅐CT fusion was that the localization could be visualized directly in live cells. This visualization required no prior selection of the transfected cells. Primary hepatocytes are quiescent cells. However GFP⅐CT fusion protein provided an opportunity for the nuclear or nonnuclear localization of the overexpressed protein to be directly detected in the transfected primary cells. Forty-eight h after transfection, GFP⅐CT fusion protein was detected exclusively in the nucleus (Fig. 4D). The nuclear localization of GFP⅐CT fusion protein was seen in all transfected hepatocytes. In contrast, GFP⅐CT ⌬N fusion protein was exclusively localized to the cytoplasm of all transfected hepatocytes. This suggested that the nuclear localization of GFP⅐CT fusion protein in hepatocytes was also directed by the N-terminal nuclear targeting sequence of CT␣. A similar nuclear localization of GFP⅐CT fusion protein was also detected in a wide range of cell lines from different species and tissues. These cell lines included rat hepatoma cells (RH7777, Fig. 4C), baby hamster kidney cells (Fig. 4E), human hepatoma cells (Hep G2, Fig. 4F), sheep choroid plexus (Fig. 4G), rat embryo fibroblast (Fig. 4H), human lung cells (MRC-9, Fig. 4I), and human mammary gland (MCF-10A, Fig. 4J). The exclusively cytoplasmic localization of GFP⅐CT ⌬N fusion protein was observed in all cell lines except rat embryo fibroblast, in which the fusion protein was evenly distributed throughout the cells.

Serum Depletion/Replenishment Caused neither Translocation of CT nor Synchronized Entry of Cells to G 1 or S Phase-
The cell cycle-associated regulation of PC synthesis (23) and the recent finding of CT translocation from the nucleus to the cytoplasm (15) raised an intriguing possibility that CT may translocate across the nuclear membrane in a cell-cycle dependent manner. Our experimental design of using GFP⅐CT␣ fusion constructs allowed us to monitor any potential movement of CT during the cell cycle. To determine whether there was a translocation of GFP⅐CT fusion protein from the nucleus to the cytoplasm during the G 0 to G 1 transition, we repeated the serum deprivation/replenishment conditions similar to those described by Cornell and co-workers (15). The stably transfected MT58 cells with GFP⅐CT fusion plasmid were deprived of serum for 36 h at the permissive temperature. Upon replenishment with 10% serum, the localization of GFP⅐CT fusion protein was examined with a fluorescence microscope continuously at 2-hour intervals. At all time points, GFP⅐CT fusion protein was detected exclusively in the nucleus (Fig. 5C). Because serum deprivation/replenishment has been used traditionally to prepare cells for synchronized entry into the G 1 phase of the cell cycle, we examined the populations of cells in each phase of the cell cycle by propidium iodide staining and flow cytometric analysis. Cells stayed at the G 0 /G 1 position induced by serum depletion and did not move synchronously into S phase upon serum replenishment (Fig. 5D), suggesting that reentry of G 0 cells into the cell cycle was not synchronized by serum replenishment.
Nuclear Localization of CT in the Synchronized Population of Cycling Cells-The failure of serum depletion/replenishment to synchronize cells prompted us to examine the localization of GFP⅐CT fusion protein in cells synchronized at different phases of the cell cycle by a different method. To achieve a highly synchronized movement of cells through the cell cycle, we used a combination of nocodazole and hydroxyurea treatments, which are capable of synchronizing cells at the G 2 /M boundary (25) and the late G 1 /S boundary (24), respectively. The cycling cells were collected by first blocking cells at the G 2 /M boundary with nocodazole, a specific and reversible inhibitor of mitotic spindle assembly (26). The blocked cells were easily collected because mitotic cells were loosely attached to the culture surface. These cells were then plated onto culture dishes, and the entry into S phase was blocked by 1 mM hydroxyurea, a specific and reversible inhibitor of DNA polymerase (24). In the presence of hydroxyurea, all cells were at the border of late G 1 /S (Fig. 5A, top panel). Removal of hydroxyurea triggered the synchronous entry of the cells into the cell cycle at S phase. The combination of nocodazole and hydroxyurea provided very synchronous cell populations in all four major phases of the cell cycle (Fig. 5B). A cell cycle analysis program (ModFit) determined that each phase contained over 70% synchrony (data not shown). Upon continuous examination, GFP⅐CT fusion protein was localized to the nucleus in all the cells of all phases of the cell cycle (Fig. 5A).

DISCUSSION
The current study clearly demonstrates that CT␣ localization in the nucleus is apparently a cell type-and cell cycleindependent event for mammalian cells. We have also demonstrated for the first time, without the influence of other isoforms of CT, that localization of CT␣ in primary hepatocytes is also nuclear. It is also the first localization of CT␣ during all phases of the cell cycle of mammalian cells. This nuclear localization was dependent solely on the permanent N-terminal nuclear targeting sequence of CT␣. The mechanism of CT␣ nuclear targeting seemed unsaturable even in the presence of a high level of exogenous CT␣. It also was clear that CT␣ stayed in the nucleus during all stages of the cell cycle except M phase, when the nuclear membrane is completely disintegrated (27). The complete disintegration of the nuclear membrane would allow a protein of any size without a specific targeting mechanism to be evenly distributed in the cells. The N-terminal nuclear targeting sequence-dependent nuclear localization of CT␣ also excluded the possibility that CT moved through the nuclear pores passively. The nuclear localization of CT␣ was apparently universal to all the cell types tested in the current study, because CT␣ was uniformly nuclear even in the asynchronized populations of cells. The subcellular localization of CT seems to have no apparent effect on the synthesis of PC. More importantly, the recently reported translocation of CT between the nucleus and the cytoplasm was not observed in any cell type or at any stages of the cell cycle in our current study. We propose two possible explanations for the difference between our current report and the previous report from Cornell and co-workers (15). 1) The serum-induced translocation of CT is an event specific only to IIC9 cells and not a universal event in mammalian cells. 2) The indirect detection of CT in IIC9 cells using antibodies was not specific to CT␣.
A major difference between our current findings and previous findings was the CT localization in primary hepatocytes. The study by Houweling et al. (10) demonstrated that significant staining of CT was detected in both the cytoplasm and the nucleus of primary rat hepatocytes. According to the recent findings of CT isoforms (11,28), we offer an explanation for why CT was detected in both the nucleus and the cytoplasm in primary hepatocytes. The antibody used in the previous studies was raised against a conserved region of CT ( 164 DFVAHD-DIPYSSAG). This conserved region is also shared by other isoforms of CT (CT␤1 and CT) that are known to be present specifically in the cytoplasm (28). Therefore, the detection of CT using this antibody was not specific to CT␣, and this antibody is capable of recognizing cytoplasmic isoforms of CT, resulting in the detection of both nuclear and cytoplasmic CT.
Another difference between our current study and previous studies is that treatment of cells with oleate did not change the subcellular localization of CT␣ fusion protein (data not shown). Addition of oleate to HeLa cells has been shown to induce translocation to the nuclear membrane (29). Currently, it is not clear if the failure of GFP⅐CT␣ fusion protein to bind nuclear membranes was affected by the fusion construction with GFP. Nevertheless, the failure of GFP⅐CT␣ fusion protein to bind to nuclear membranes upon oleate stimulation had no effect on its ability to synthesize PC and rescue MT58 cells at the nonpermissive temperature.
Previous constructs of CT␣ lacking the N-terminal nuclear targeting sequence resulted in CT␣ localization in both the nucleus and the cytoplasm (12). Our protein fusion design of GFP⅐CT ⌬N was the first active CT␣ construct localized exclusively in the cytoplasm. Such an unexpected specificity for the cytoplasm allowed us to determine whether the nuclear or cytoplasmic localization of CT␣ would have different impacts on the ability of CT␣ to synthesize PC and rescue mutant cells lacking CT. The answer was clear that both nuclear CT and cytoplasmic CT were equally effective for PC synthesis and mutant rescue. This observation is consistent with the fact that the molecular weight of CDP-choline is well below the molecular weight cut-off of the nuclear membrane pores. Therefore, the movement of CDP-choline across the nuclear membrane should not be restricted. No matter where CDP-choline is synthesized, it is readily available for PC synthesis. It is not clear why certain cells may have CT present both in the nucleus and in the cytoplasm. The presence of isoforms of CT in the cytoplasmic compartments has diminished the possibility that CT␣ is required to translocate from the nucleus to the cytoplasmic compartments for PC synthesis.
As previously mentioned, the endoplasmic reticulum is a primary site for PC synthesis. PC synthesis via the CDPcholine pathway has been observed to have the channeling effect (30), in which PC can only be labeled by the initial substrate choline but not by cholinephosphate or CDP-choline. A possible interpretation of channeling is that the three enzymes of this pathway form a complex that is only accessible by choline. According to this hypothesis, all three enzymes of the CDP-choline pathway are expected to localize to the primary site for PC synthesis. The question of why mammalian cells need multiple isoforms of CT specifically localized to various compartments, whereas the presence of a single isoform anywhere in the cell is sufficient for its role for PC synthesis, remains intriguing and unanswered.