Nuclear translocation of phospholipase C-delta1 is linked to the cell cycle and nuclear phosphatidylinositol 4,5-bisphosphate.

Nuclear phosphoinositides, especially phosphatidylinositol 4,5-bisphosphate, fluctuate throughout the cell cycle and are linked to proliferation and differentiation. Here we report that phospholipase C-delta(1) accumulates in the nucleus at the G(1)/S boundary and in G(0) phases of the cell cycle. Furthermore, as wild-type protein accumulated in the nucleus, nuclear phosphatidylinositol 4,5-bisphosphate levels were elevated 3-5-fold, whereas total levels were decreased compared with asynchronous cultures. To test whether phosphatidylinositol 4,5-bisphosphate binding is important during this process, we introduced a R40D point mutation within the pleckstrin homology domain of phospholipase C-delta(1), which disables high affinity phosphatidylinositol 4,5-bisphosphate binding, and found that nuclear translocation was significantly reduced at G(1)/S and in G(0). These results demonstrate a cell cycle-dependent compartmentalization of phospholipase C-delta(1) and support the idea that relative levels of phosphoinositides modulate the portioning of phosphoinositide-binding proteins between the nucleus and other compartments.

Nuclear phosphoinositides, especially phosphatidylinositol 4,5-bisphosphate, fluctuate throughout the cell cycle and are linked to proliferation and differentiation. Here we report that phospholipase C-␦ 1 accumulates in the nucleus at the G 1 /S boundary and in G 0 phases of the cell cycle. Furthermore, as wild-type protein accumulated in the nucleus, nuclear phosphatidylinositol 4,5bisphosphate levels were elevated 3-5-fold, whereas total levels were decreased compared with asynchronous cultures. To test whether phosphatidylinositol 4,5bisphosphate binding is important during this process, we introduced a R40D point mutation within the pleckstrin homology domain of phospholipase C-␦ 1 , which disables high affinity phosphatidylinositol 4,5-bisphosphate binding, and found that nuclear translocation was significantly reduced at G 1 /S and in G 0 . These results demonstrate a cell cycle-dependent compartmentalization of phospholipase C-␦ 1 and support the idea that relative levels of phosphoinositides modulate the portioning of phosphoinositide-binding proteins between the nucleus and other compartments.
A distinct phosphoinositide cycle is present in nucleus, and growing evidence suggests its metabolism is important for DNA repair, mRNA export, and gene transcription (1)(2)(3)(4)(5). Moreover, changes in nuclear phosphoinositide levels are correlated to cell cycle progression and independently regulated from the phosphoinositide cycle at the plasma membrane (6,7). Because phospholipase C (PLC) 1 is a key regulator of phosphoinositide metabolism at the plasma membrane, understanding what controls the localization of this enzyme to the nucleus and how its activity there affects nuclear phosphoinositide metabolism is important.
In mammals, PLC is a 13-member family of phosphodiesterases (␤, ␥, ␦, ⑀, and subtypes) essential to a wide range of cellular responses, including exocytosis, endocytosis, gene transcription, cytoskeletal remodeling, and membrane trafficking (8 -11). PLC-mediated hydrolysis of phosphatidylinositol 4,5bisphosphate (PIP 2 ) generates inositol 1,4,5-trisphosphate and diacylglycerol critical second messengers that mobilize cellular calcium and activate protein kinase C, respectively (8). PLC␦ 1 has been shown to be activated by capacitative calcium entry (12), transglutaminase II (13), a form of Rho GTPase-activating protein (14), free fatty acids such as arachidonic acid (15), and phosphatidylserine (16). Upon activation, PLC␦ 1 binds the inositol head group of PIP 2 with high affinity and specificity through its non-catalytic PH domain (17)(18)(19). This interaction facilitates association with the plasma membrane, particularly with ruffles, where PIP 2 is enriched (20). When a point mutation in the PH domain (R40D) is introduced, however, the ability of PLC␦ 1 to target the plasma membrane is greatly reduced, resulting in a higher cytosolic concentration (18).
Several PLC isoforms are detected in the nuclear compartment, primarily due to alternative spliced variation (21). PLC␦ 1 , however, is the only isoform that has been demonstrated to continuously cycle between the nucleus and cytoplasm (22,23). When Madin-Darby canine kidney cells are treated with leptomycin B (LMB), which inhibits CRM1-dependent nuclear export, transiently expressed GFP-PLC␦ 1 accumulates in the nucleus, consistent with nucleocytoplasmic shuttling (22). PLC␦ 1 is also detected in nuclei of rat hepatoma cells (24) and astrocytes (25). It is unclear, however, what regulates the steady-state distribution of this enzyme between the nuclear and extranuclear compartments.
Yamaga et al. (22) have indicated that the R40A point mutation within the pleckstrin homology domain of PLC␦ 1 results in faster and greater nuclear accumulation under conditions in which nuclear export is blocked, possibly due to a greater fraction available to engage the import apparatus. Although mechanisms that regulate nuclear import, such as dimerization and phosphorylation, have not been ruled out, it is likely that a PIP 2 -mediated sequestration at the plasma membrane influences the steady-state distribution of PLC␦ 1 between the nuclear and cytoplasmic compartments.
The idea that phospholipids influence compartmentalization of PI-binding proteins is growing and evident in the recent work with Tub, which binds PIP 2 through its tubby domain (26), and ING2, which binds PI(5)P through its plant homeodomain finger (27). Upon G-protein-linked hydrolysis of PIP 2 , Tub translocates to the nucleus (26), whereas ING2 accumulates in the nucleus as the levels of nuclear PI(5)P increase (27). We set out to test whether high affinity PIP 2 binding plays a role in targeting PLC␦ 1 to the nuclear compartment and whether this could be related to the cell cycle.
Here we establish that PLC␦ 1 accumulates in the nucleus in G 0 and at the G 1 /S-phase boundary and that its accumulation is significantly decreased by the R40D point mutation in its PH domain, which abrogates high affinity PIP 2 -binding. Furthermore, nuclear accumulation of PLC␦ 1 correlates with increased total nuclear levels of PIP 2 , consistent with a general mecha-nism by which phosphoinositides can influence the nuclear accumulation of proteins that bind these lipids.

EXPERIMENTAL PROCEDURES
Synchrony, Flow Cytometry, and Cell Cycle Analysis-NIH-3T3 fibroblasts (American Type Culture Collection) were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum, 1 mM penicillin and streptomycin, 1 mM non-essential amino acids, and 1 mM sodium pyruvate (all supplements were from Invitrogen) at 37°C in a 10% CO 2 humidified incubator. To synchronize cells to G 0 , adherent cells were washed three times and maintained in a medium with reduced serum (0.5% fetal bovine serum) for 30 h, a technique that causes many cell types to exit the cell cycle and become quiescent (28 -30). To synchronize cells to the G 1 /S boundary, adherent fibroblasts were twice blocked with 2 mM thymidine (6,31). Briefly, cultures growing at an exponential rate were washed with Dulbecco's modified Eagle's medium and incubated in growth medium supplemented with 2 mM thymidine for 12-15 h, washed three times with growth medium, and incubated for an additional 8 -10 h without thymidine. Cells were then treated again with growth medium supplemented with 2 mM thymidine for an additional 12-15 h. To release them from thymidine block, cells were washed three times in growth medium. To analyze DNA content, cells were washed with phosphate-buffered saline (PBS) (Invitrogen), harvested with trypsin/EDTA, washed again, and fixed in 3.7% formaldehyde with 0.1% BSA in PBS at 4°C overnight. Cells were permeabilized in PBS containing 0.1% BSA and 0.1% Triton X-100 for 5 min, washed with PBS, and incubated in 50 mM citrate buffer with 50 g/ml propidium iodide and 10 g of RNase A for 30 min at 37°C. After washing the cells again, cell cycle analysis was immediately performed using a fluorescence-activated cell sorter (FAC-Scan; BD Biosciences), and the data were analyzed using SyncWizard.

myo-[ 3 H]Inositol Labeling, Lipid Extraction, and Thin
Layer Chromatography-Cells were cultured in growth media supplemented with 2-8 Ci/ml myo-[ 3 H]inositol (PerkinElmer Life Sciences) in 60-mm dishes for 48 -72 h to reach isotopic steady state. Cultures were then washed with Dulbecco's modified Eagle's medium and taken through the above synchronization procedures in the presence of radiolabel. Whole cell levels of phosphoinositides were measured essentially as described previously (32). Briefly, synchronized cells were rinsed twice in ice-cold PBS with 1 mM calcium chloride and 0.5 mM magnesium chloride (PBS-Ca), followed by ice-cold PBS. Monolayers were scraped into 750 l of methanol:0.1 M HCl (v/v, 1:1), placed in silianized borosilicate glass tubes, and mixed vigorously for 30 s. 500 l of chloroform was then added to each sample, which was mixed for 30 s, and then placed on a rocker at room temperature for 15 min. The aqueous and organic phases were separated by centrifugation for 5 min at 1000 ϫ g. The upper aqueous phase was removed and saved for inositol polyphosphate analysis. The lower phase was back extracted twice with 500 l each of methanol:0.1 M EDTA (v/v, 1:0.9) and then dried under nitrogen and dissolved in chloroform:methanol (v/v, 1:1). Lipid extracts were applied to pre-scored Linear KD silica plates (Whatman) that were treated with 40% methanol, 1% potassium oxalate, and 1 mM EGTA in water and heat-activated. Purified PI, PI(4)P, and PIP 2 (10 g of each) were used as standards. The solvent system used to separate the lipids was chloroform:methanol:water:concentrated ammonium hydroxide (v/v/v/v, 60:47:11.3:2), similar to that described in Ref. 33. Areas corresponding to migration of PI, PI(4)P, and PIP 2 standards were scraped into vials containing 100 l of methanol:10% Nonidet P-40 (v/v, 1:1). 4 ml of scintillation fluid (EcoLite) was added to each vial and counted the following day in a liquid scintillation spectrometer. The cpm values were normalized to total lipid phase-extractable phosphorus and specific activity of the radiolabeled medium.
Subcellular Fractionation-Cells were lysed by osmotic swelling, similar to a previously described protocol (7). In 60-mm dishes, NIH-3T3 fibroblasts were briefly washed with ice-cold PBS-Ca. Cells were then treated briefly with PBS containing 1 mM EDTA, centrifuged at 600 ϫ g for 5 min at 4°C, and promptly resuspended in 500 l of pre-chilled resuspension buffer (10 mM NaCl, 1.5 mM MgCl 2 , 10 mM Tris-HCl, pH 7.4) on ice for 7 min. The degree of cytoplasmic swelling was assessed by phase-contrast microscopy. The swollen cells were then transferred to a Dounce homogenizer and lysed by 20 strokes of the glass pestle. Separated nuclei were then layered onto a sucrose cushion (320 mM sucrose, 7.7 mM MgCl 2 , 2.1 mM EGTA, and 0.1 mM phenylmethylsulfonyl fluoride) and centrifuged at 300 ϫ g for 3 min at 4°C. The supernatant was removed, and nuclei were washed twice with 1 ml of resuspension buffer. 750 l of methanol:0.1 M HCl was added to the nuclear pellet and taken through the lipid extraction procedure described above. The degree of contamination was assessed using standard compartmental markers for the cytoplasm (lactate dehydrogenase) and endoplasmic reticulum (cytochrome-c oxidase) according to the manufacturer's protocol (Sigma).
DNA Constructs and Cell Transfection-Full-length human PLC␦ 1 cDNA was inserted into pEGFP-N1, a cytomegalovirus-driven mammalian expression vector that carries the gene for enhanced green fluorescent protein (EGFP; Clontech). This catalytically active chimera, designated PLC-␦ 1 -EGFP, served as a template for the production of various mutant PLC-␦ 1 -EGFP fusion proteins, as described previously (34). Briefly, a catalytically inactive (H356A) form and a double mutant (R40D/H356A) that fails to bind PIP 2 with high affinity were constructed using a Pfu mutagenesis approach (Stratagene). The mutants were designated PLC␦ 1 H356AEGFP and PLC␦ 1 R40DH356AEGFP, respectively. We also utilized PH(␦ 1 )EGFP (residues 1-126 of full-length PLC␦ 1 fused to EGFP) and its corresponding R40D point mutant (20). Point mutations have been verified via DNA sequencing, in vitro binding, and enzymatic activity assays (35). For transient expression of all PLC␦ 1 EGFP chimeric proteins, NIH-3T3 cells were plated on #1.0 borosilicate chambered glass coverslips (Nalge Nunc International) coated with fibronectin (50 g/ml; Sigma) and allowed to adhere overnight. The following day, cells were transfected with FuGENE 6 reagent according to manufacturer's protocol (Roche Applied Science).
In some experiments, cells were fixed with 3.7% formaldehyde for 5 min and incubated in 4Ј,6-diamidino-2-phenylindole (DAPI; 5 g/ml) for 5 min to clearly identify the nuclear compartment. Images were captured with an AxioCam 330mA 12-bit charge-coupled device camera (Zeiss) and viewed with Carl Zeiss Axovision 3.1 software. To avoid bias, samples were coded, and the investigator was blinded to the experimental conditions. Captured Z-stacks were deconvolved using a constrained iterative point spread function according to recommended Niquest criteria and analyzed. Images were further analyzed in Life Science Resources software. The degree of nuclear accumulation was determined by analysis of intensity profiles in both the nuclear and cytoplasmic compartments. The formula used to determine the average nuclear to cytoplasmic (N:C avg ) pixel intensity ratio is ⌺[(N avg Ϫ B avg )/ (C avg Ϫ B avg )]/n, where N avg ϭ average nuclear pixel intensity excluding nucleoli, C avg ϭ average cytoplasmic pixel intensity excluding perinuclear regions and vacuoles, B avg ϭ average background pixel intensity, and n ϭ the number of cells sampled. A comparable method was used to determine the ratio of plasma membrane (P avg ) to cytoplasm pixel intensity for analysis of PH(␦ 1 )EGFP compartmentalization in deconvolved images. The P avg , however, was determined by sampling a stretch of plasma membrane no less than 10 m in length. The method provides an average of plasma membrane intensities, which vary significantly along the membrane surface.
Fixation and Indirect Immunofluorescence-Monolayers were rinsed once in PBS-Ca and then fixed with 3.7% formaldehyde solution (Fisher-Scientific) in PBS for 10 min at room temperature. Samples were then washed three times in PBS for 5 min and permeabilized with 0.2% Triton X-100 (Rohm & Haas Co.) in PBS for 5 min at room temperature. The detergent was replaced with blocking solution (PBS, 5% BSA (Sigma), 5% goat serum (Pierce)) for 30 min at room temperature. The blocking solution was then replaced with primary antibody solution (1:200 dilution of anti-PLC␦ 1 clone S-11-2 (05-343; Upstate) or 1:200 dilution of anti-PIP 2 (Assay Design) in PBS with 1% BSA and 1% goat serum) and placed in an incubator at 37°C for 1 h. In some experiments, anti-PIP 2 was pre-incubated with vesicles containing 20 M PI(4)P or PIP 2 (vesicles prepared as a 70% phosphatidylcholine, 20% phosphatidylserine, 10% phosphatidylinositol mix) for 1 h. Samples were washed three times in 1 ml of PBS for 7 min and then incubated at 37°C for 1 h in goat anti-mouse IgG Alexa488 secondary antibody (Molecular Probes) or goat anti-mouse IgG (HϩL) conjugate Texas Red (Molecular Probes) diluted 1:3000 in PBS with 1% BSA and 1% goat serum. Each well was then washed three times in PBS for 7 min. Fixed cells were visualized by epifluorescence microscopy (Olympus IMT-2 inverted microscope with 100-watt Mercury arc lamp), and images were taken with Nikon Plan Fluor ϫ40 oil objective (numerical aperture, 1.3) and Olympix AstroCam (Life Science Resources). Images were processed and analyzed with Esprit imaging software (Life Science Resources). Alternatively, Z-stacks were obtained and deconvolved on the Zeiss microscope as described above.
Western Blot Analysis-C6 glioma cells were grown to 50% confluence and fed normal growth media or blocked to the G 1 /S boundary as described above. After synchronization was complete and verified by fluorescence-activated cell-sorting analysis, an equal number of cells were harvested and washed in PBS-Ca. Nuclei were isolated as described above, without the use of a sucrose cushion. After several washes, the nuclear pellets were lysed with a high-salt extraction buffer (50 mM Tris-HCl, pH 8.0, 450 mM NaCl, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 1% Nonidet P-40, 1% mammalian protease inhibitor) on ice for 15 min and centrifuged for 5 min at 14,000 ϫ g; the supernatant was removed and designated the nuclear fraction. Total volume of nuclear fraction (ϳ1 ϫ 10 6 cells) was loaded into each lane and compared with 10% of the cytoplasmic fraction. Lysates were transferred to a polyvinylidene difluoride membrane (Bio-Rad), probed with the antibody described above (1:200 in PBS with 3% dried, fat-free milk), and detected using enhanced chemiluminescence (Amersham Biosciences).
Statistical Analysis-All statistical analysis was preformed in GraphPad Prism. Nuclear to cytoplasm ratios were pooled from duplicate wells from multiple experiments. To determine whether the distribution of pooled data passed normality, we used the Kolomogorov-Smirnov test. To determine significance of the differences among mean values, we used a one-way analysis of variance and Newman-Kuels post test.

RESULTS
PLC␦ 1 , fused to EGFP, was expressed in NIH-3T3 mouse fibroblasts. These large cells have a relatively flat morphology that facilitates localization studies of this type. To prevent hydrolysis of cellular phosphoinositides, the enzyme was inactivated by H356A mutation. In our initial work, we observed that 6% of cells transiently expressing PLC␦ 1 H356AEGFP had a greater degree of fluorescence associated with the nuclear compartment than the cytoplasmic compartment. Because PLC␦ 1 is a nucleocytoplasmic shuttling protein, we reasoned that nuclear accumulation could be attributed to a brief stage of the cell cycle, in which the steady-state distribution has shifted toward the nuclear compartment. Because this protein is well known to bind strongly to PIP 2 , we also hypothesized that a change in nuclear or extranuclear PIP 2 could influence this redistribution. We first set out to determine whether and how nuclear PLC␦ 1 levels fluctuate during the cell cycle of NIH-3T3 fibroblasts and to measure, in parallel, changes in total cellular and nuclear PIP 2 concentrations.
Cell Cycle Synchrony and NIH-3T3 Cells-Subconfluent NIH-3T3 fibroblasts were synchronized by serum withdrawal or double thymidine block. When asynchronous cultures growing at logarithmic rates are withdrawn from serum, they exit the cell cycle at G 1 and become quiescent (G 0 ) (28 -30). NIH-3T3 fibroblasts have asynchronous distributions of DNA content (Fig. 1A) that indicate distinct G 1 (54.2%), S (34.7%), and G 2 /M (11.2%) populations. Reducing the serum to 0.5% for 30 h (Fig. 1B) resulted in cells with more G 1 DNA content (81.7%) and fewer cells in S (14.8%) or G 2 /M (3.6%). Because G 0 and G 1 FIG. 1. Flow cytometric analysis of DNA content of NIH-3T3 fibroblasts. Propidium iodide content was measured via flow cytometry during (A) asynchronous growth (10% serum), (B) synchronous G 0 (0.5% serum, 30 h), (C) G 1 /S boundary (double 2 mM thymidine block, 10% serum), or (D) release from G 1 /S checkpoint for 3 h. E is a growth curve of NIH-3T3 fibroblasts cultured in media containing 10% serum (OE) or 0.5% serum (f), further indicating that A represents a logarithmic population and B represents a more quiescent population.
DNA content are equivalent with fluorescence-activated cellsorting analysis, we assayed cell proliferation under logarithmic growth conditions (doubling time of 17 h; Fig. 1E) and compared that to growth in low-serum medium (doubling time of 29 h). These data show that reduced serum (0.5%) for 30 h enriched the number of quiescent NIH-3T3 cells in these cultures. 2 mM double thymidine block (Fig. 1C) was used to synchronize cultures to the G 1 /S checkpoint (6). 97.2% of cells had S-phase DNA content after a 2 mM double thymidine block (G 1 , 1.7%; G 2 /M, 1.1%). As others have shown, washing out the thymidine releases cells from G 1 /S block (31); as early as 3 h after release, 93.2% of cells were in G 2 /M as indicated by increased DNA content (Fig. 1D). These data demonstrate that these treatments synchronize NIH-3T3 cultures to a predicted DNA content, blocking them at well-defined points in the cell cycle.
Nuclear Accumulation of PLC␦ 1 is Cell Cycle-dependent, Occurring at the G 0 and the G 1 /S Checkpoint in NIH-3T3 Fibroblasts and at the G 1 /S Checkpoint in C6 Glioma Cells-To explain the degree of variance in nuclear PLC␦ 1 levels in asynchronous cell populations, we hypothesized that the enzyme accumulated in the nucleus in a cell cycle-dependent manner. We found that when cells were blocked at G 0 or G 1 /S, the ratios of nuclear to cytoplasmic PLC␦ 1 H356AEGFP intensity were enhanced by 1.4-and 2.0-fold, respectively, indicating that PLC␦ 1 accumulated in the nucleus during these stages of the cell cycle (Fig. 2, B and C) as a result of translocation, when compared with asynchronous cultures. Treatment with leptomycin B, a toxin used to block CRM1-driven nuclear export (Fig. 2, A and D), showed similar results (36,37).
To determine whether transient expression of PLC␦ 1 H356AEGFP led to cell cycle synchrony, we analyzed DNA content by fluorescence-activated cell-sorting analysis. We found that cell cycle partitioning was similar to that of non-transfected cells, suggesting that the catalytically inactive mutant did not disrupt cell cycle progression in these fibroblasts (data not shown). Furthermore, synchronized NIH-3T3 cells expressing PLC␦ 1 H356AEGFP were also released from thymidine block and observed over time; 4 h later, PLC␦ 1 returned to the cytoplasmic compartment (data not shown), suggesting that cell synchrony did not lead to irreversible accumulation of our chimera in the nucleus.
Although using enhanced green fluorescence protein (EGFP) offers many advantages in studying living cells, this tag may influence function or localization of the protein of interest (38). With this in mind, we used chimeric proteins engineered with the EGFP tag attached to the C terminus, well removed from the N-terminal PH domain, which is critical for high affinity PIP 2 interactions (18). Furthermore, we determined that no significant relationship existed between total intensity and the nuclear/cytoplasmic intensity ratio in any condition tested, indicating that nuclear accumulation was not due simply to overexpression of this protein (linear correlation coefficients ranged from 0.06 to 0.35). Nonetheless, it was still possible that addition of the EGFP tag may have influenced protein distribution; therefore, we used indirect immunofluorescence of endogenous PLC␦ 1 to determine whether or not this agreed with our transient expression results. Rat C6 glioma cells were used in these experiments because they express much higher levels of PLC␦ 1 compared with NIH-3T3 fibroblasts. High nuclear levels of endogenous PLC␦ 1 are also observed in C6 glioma cells syn- chronized to the G 1 /S boundary when compared with asynchronous cultures (Fig. 2, E and F). Moreover, when rat C6 glioma cells are synchronized to the G 1 /S boundary, a similar increase in nuclear PLC␦ 1 is clearly observed when compared with asynchronous nuclei (Fig. 2G) by Western blot analysis.
Levels of Total PIP 2 Decrease at Both G 0 and G 1 /S-Yamaga et al. (22) reported that R40D mutation enhanced the degree to which this protein accumulated in the nucleus when export is blocked. Therefore, we reasoned that a decrease in total PIP 2 at the plasma membrane in G 0 and at the G 1 /S boundary could account for a shift of bound PLC␦ 1 to free PLC␦ 1 , increasing the amount able to engage the import apparatus. Although previous investigators have found that total cellular levels of phosphoinositides remain relatively unchanged throughout the HeLa cell cycle (6), we decided to determine whether these levels were altered in NIH-3T3 cells during cell cycle synchrony. We compared asynchronous growing cell cultures that were labeled with myo-[ 3 H]inositol to cells synchronized to G 0 and G 1 /S. The levels of total PIP 2 were reduced by 3.5-fold in cells synchronized to G 0 compared with asynchronous cultures (Fig. 3). Furthermore, the levels of total PIP 2 decreased by 2.8-fold at the G 1 /S boundary (Fig. 3). Similar decreases in PIP were seen at these stages as well. The levels of inositol monophosphate and inositol polyphosphates, however, remained nearly constant (data not shown).

Fraction of Available PIP 2 at the Plasma Membrane Decreases during Quiescence but Not at the G 1 /S-phase Checkpoint in NIH-3T3 Cells as Indicated by PH(␦ 1 )EGFP-
In the typical mammalian cell, PIP 2 represents ϳ1% of the total phosphoinositide pool (39), of which the majority is at the plasma membrane. Although the effective concentration of PIP 2 at the plasma membrane is estimated to be 20 M (39), it is unclear what fraction is free to bind PLC␦ 1 or other proteins. Furthermore, it is unclear whether change in the available fraction corresponds to changes observed with myo-[ 3 H]inositol labeling of total levels. To determine whether free PIP 2 levels decrease at G 0 and G 1 /S phase concomitantly with total levels, we used PH(␦ 1 )EGFP, a well-known indicator of PIP 2 at the plasma membrane (17,20,40,41). We compared mean plasma membrane pixel intensity (P avg ) to cytoplasmic pixel intensity (C avg ) ratios (Fig. 4A). The concentration of free PIP 2 at the plasma membrane did not change due to G 1 /S synchrony; however, plasma membrane intensities were significantly reduced during G 0 by 1.35-fold. Fig. 4B illustrates the relative frequency histograms of binned P avg to C avg intensity ratio of logarithmic (black bar), G 0 (white bar), or G 1 /S-phase blocked cells (checkered bar) expressing PH(␦ 1 )EGFP, indicating that available levels of plasma membrane PIP 2 were significantly decreased during quiescence but not at the G 1 /S checkpoint.
Nuclear PIP 2 Is Increased during G 0 and G 1 /S-Although a decrease in plasma membrane PIP 2 could increase the fraction of PLC␦ 1 available for nuclear import, the apparently compensating increase in the available PIP 2 found at the G 1 /S boundary would not, by itself, explain the observed shift to the nucleus. In HeLa cells it has been shown that total nuclear phosphoinositides are elevated at the G 1 /S boundary and decrease as early as 2 h after release from G 1 /S block by Ͼ2-fold and that these levels subsequently recover throughout S-phase progression, although no net change is observed in the total cellular PI pool in these cells (6). On the other hand, nuclear phosphoinositide levels remain constant in MEL cells, whereas turnover of phosphoinositides in this compartment is elevated at a similar stage of the cell cycle (7). We hypothesized that an increase in total nuclear PIP 2 could result in a greater fraction of bound PLC␦ 1 in the nucleus, accounting for accumulation of PLC␦ 1 at G 0 and G 1 /S. This could occur even in the absence of any changes in the intrinsic rates of transport at the nuclear pore. To determine whether nuclear phosphoinositides were changed at these stages of the cell cycle, we isolated nuclei from synchronized cultures grown in the presence of myo-[ 3 H]inositol and measured their levels of 3 H-labeled lipids.

FIG. 4. Available PIP 2 levels at the plasma membrane in NIH-3T3 fibroblasts.
Living cells expressing PH(␦ 1 )EGFP were imaged, and Z-stacks were captured and deconvolved. A is a relative frequency histogram of binned plasma membrane pixel intensity to average cytoplasmic pixel intensity ratio of asynchronous logarithmic (black bars), G 0 (white bars), or G 1 /S boundary blocked cells (checkered bars). The average plasma membrane intensity was determined by averaging the intensity along a 10-m stretch of plasma membrane and within the neighboring cytoplasm and used to determine the ratio of fluorescence in these compartments, respectively. These data passed normality using the Kolomogorov-Smirnov test. B, comparison of mean plasma membrane to mean cytoplasmic intensity ratios from three independent experiments in cells that transiently express PH(␦ 1 )EGFP (under logarithmic growth, black bar (n ϭ 26); quiescent conditions, white bar (n ϭ 31); G 1 /S-phase block, checkered bar (n ϭ 21)). The shift in average plasma membrane pixel intensities to cytoplasmic pixel intensities in response to synchronization was compared with asynchronous cultures. Bars indicate S.E. ***, p Ͻ 0.001. Nuclei were prepared by Dounce homogenization and centrifugation through a sucrose cushion, similar to a previously described protocol (7). Isolated nuclei were examined by phase microscopy and determined to be relatively free of adherent perinuclear material (data not shown). Levels of contaminating cytosolic and endoplasmic reticulum markers were 6.3 Ϯ 2.3% (mean Ϯ S.D.) of lactate dehydrogenase and 6.5 Ϯ 1% of cytochrome-c oxidase. Comparable results have been reported for nuclei similarly prepared from HeLa cells (6). We found that nuclear PIP 2 levels were increased at G 0 by 2.4-fold and at G 1 /S by 5.3-fold when compared with asynchronous logarithmic cell cultures (Fig. 5). PIP levels underwent similar changes. Although it is possible that changes in the levels of rare PIP 2 isomers led to the 2.4 -5.3-fold increase in nuclear PIP 2 levels, it seems unlikely because Ͼ98% of the PIP 2 in the whole cell is the 4,5-isomer (39), of which 5-10% is nuclear as estimated from our results. Other isomers such as PI(3,5)P 2 and PI(3,4)P 2 appear to be much rarer than this (32).
Accumulation of Transiently Expressed PLC␦ 1 H356AEGFP at the G 0 and G 1 /S-phase Checkpoint Is Enhanced by Its Ability to Bind PIP 2 -Given that total nuclear PIP 2 levels increase significantly during G 0 and G 1 /S and that these increases correlate with nuclear accumulation of PLC␦ 1 , we decided to directly test the role of PIP 2 binding during this process by introducing a R40D point mutation in the PH domain. The double mutant PLC␦ 1 R40DH356AEGFP was expressed at similar levels compared with PLC␦ 1 H356AEGFP protein under asynchronous conditions; however, eliminating the ability of PLC␦ 1 to bind PIP 2 markedly reduced PLC␦ 1 accumulation during G 0 and the G 1 /S-phase checkpoint. Fig. 6, A and B, shows relative frequency histograms of the nuclear to cytoplasmic intensity ratios that illustrate a wide variation in compartmentalization of PLC in asynchronous cultures and how nuclear accumulation is enhanced in G 0 and G 1 /S phases. The R40D mutation diminished this shift in nuclear accumulation at G 0 , suggesting that the ability of PLC␦ 1 to bind PIP 2 influenced PLC␦ 1 accumulation at these points in the cell cycle. Fig.  6C illustrates that the average nuclear to cytoplasmic ratio increased by 1.3-and 1.5-fold during G 0 and G 1 /S, respectively, when compared with asynchronous cultures. Introduction of the R40D mutation suppresses nuclear redistribution during the cell cycle. Fig. 6D illustrates that PLC harboring the R40D mutation is not significantly increased during G 0 . Some nuclear accumulation at G 1 /S is still observed in those cells expressing the R40D mutant. The results of analysis using epifluorescent images are presented, but deconvolved images were also analyzed, and the results were similar (data not shown).
Because a broad range of nuclear to cytoplasmic ratios existed, we compared the relative frequency of cells having N:C avg Ͼ 1. Only 13% of asynchronous cells analyzed had an N:C avg Ͼ 1 (Fig. 6A). This was increased 4.1-and 6.2-fold when cells were synchronized to G 0 and G 1 /S phase, respectively. On the other hand, in Fig. 6C, the R40D mutant had 26% of cells with an N:C avg Ͼ 1 under asynchronous conditions. This was increased only 1.3-and 1.9-fold at G 0 and G 1 /S phase, respectively, further indicating that the ability of PLC␦ 1 to bind PIP 2 played an important role in nuclear accumulation during G 0 and at the G 1 /S checkpoint.
Co-localization of PLC␦ 1 to a Detergent-resistant Fraction of Nuclear PIP 2 -A fraction of PIP 2 recovered in isolated nuclei is resistant to detergent treatments (42) and presumably associated with matrix proteins (43). This fraction of PIP 2 has been localized to RNA splicing machinery in nuclear speckles (44 -46). Here, we demonstrate that a monoclonal antibody directed toward PIP 2 is specific for nuclear PIP 2 in NIH-3T3 fibroblasts (Fig. 7AϪC). Cells were fixed with freshly prepared paraformaldehyde, permeabilized with detergent (Triton X-100), and incubated in primary antibody alone (Fig. 7A) or with primary FIG. 6. Nuclear accumulation of PLC␦ 1 H356AEGFP and its R40D mutant as a function of cell cycle. Transfected cells were visualized via fluorescence microscopy, and images were analyzed as indicated under "Experimental Procedures." A and B are relative frequency histograms of binned nuclear to cytoplasmic intensity ratios of asynchronous (black bars), quiescent G 0 (white bars), or G 1 /S-phase blocked cells (checkered bars) expressing (A) PLC␦ 1 H356AEGFP or (B) PLC␦ 1 R40DH356AEGFP. C, under normal serum conditions, PLC␦ 1 H356AEGFP is primarily cytosolic (black bar, n ϭ 121), as indicated by a nuclear to cytoplasmic ratio of Ͻ1. During quiescence (white bar, n ϭ 74) and G 1 /S-phase blockage (checkered bar, n ϭ 61), cells have significantly higher nuclear intensity. D, in parallel, under normal serum conditions, PLC␦ 1 R40DH356AEGFP is primarily in the cytoplasm (black bar, n ϭ 121), as indicated by a nuclear to cytoplasmic ratio of Ͻ1. Nuclear accumulation during G 0 (white bar, n ϭ 73) and G 1 /S-phase checkpoint (checkered bar, n ϭ 56) is significantly different from that of cells growing at a logarithmic rate; however, it is significantly less than PLC␦ 1 H356AEGFP. The data above are pooled from three independent experiments. ***, p Ͻ 0.001; **, p Ͻ 0.01; *, p Ͻ 0.05. antibody that was pre-incubated with vesicles containing 20 M PI(4)P (Fig. 7B) or PIP 2 (Fig. 7C). Pre-incubation with excess PIP 2 -loaded vesicles (Fig. 7C) greatly diminished fluorescence, whereas pre-incubation with PI(4)P did not (Fig. 7B). To determine whether the detergent-resistant fraction changed with cell cycle synchrony, we compared asynchronous fibroblasts (Fig. 7D) with cultures synchronized to the G 1 /S boundary (Fig.  7E) via epifluorescence microscopy. Analysis of the average intensity associated with nuclei was similar despite synchrony, suggesting that the detergent-resistant fraction of nuclear PIP 2 did not change significantly and that the detergent extracted the bulk of the PIP 2 increased by double thymidine block. PIP 2 and transiently expressed PLC␦ 1 H356AEGFP had a non-uniform distribution throughout the nuclear compartment and partially co-localize (Fig. 7, FϪI). Analysis of pixel intensities for the green and red channels, however, indicated a poor correlation between the concentration of detected PIP 2 antibody and PLC␦ 1 H356AEGFP (r 2 , ϳ0.12). Fig. 7F illustrates the degree of DAPI staining in a single plane near the middle of the nucleus. These epifluorescence images were deconvolved to remove out-offocus light. We found that PLC␦ 1 H356AEGFP was mostly present in areas that stain lightly with DAPI (Fig. 7G, arrow i) and excluded from areas that stain intensely (Fig. 7G, arrow ii). PIP 2 had a very similar distribution throughout the nucleus (Fig. 7H). The overlay reveals that PIP 2 , but not PLC␦ 1 H356AEGFP (Fig. 7I, arrow iii), was present as well in some areas that had a moderate amount of DAPI staining. DISCUSSION Our results show that PLC␦ 1 accumulates in the nucleus in G 0 and at the G 1 /S boundary, suggesting that this protein plays a role in the cell cycle. We also found that nuclear redistribution of PLC␦ 1 correlates with increased nuclear PIP 2 levels of ϳ5.3and 2.4-fold at the G 1 /S boundary and at G 0 , respectively. On the other hand, whole cell PIP 2 significantly decreased at these stages by 3.5-and 2.8-fold, respectively. To directly test whether PIP 2 binding plays a role in compartmentalization of PLC␦ 1 , we introduced a mutation in its pleckstrin homology domain (R40D) that disrupts high affinity PIP 2 binding and found that nuclear accumulation of the mutated enzyme was significantly less than that of its wild-type counterpart. These results demonstrate that nuclear phosphoinositide and nuclear PLC␦ 1 levels change throughout the cell cycle and suggest a direct role for PIP 2 in compartmentalization of PLC␦ 1 at G 0 and the G 1 /S boundary. PLC␦ 1 has been found previously in the nuclear compartment in culture tissue cell lines, but until now, an alternatively spliced form of PLC␦ 4 was the only ␦ isoform found to accumulate in the nucleus in a cell cycle-dependent manner (47). Nonetheless, Yamaga et al. (22) have shown that PLC␦ 1 is a FIG. 7. Detergent-resistant fraction of nuclear PIP 2 and its localization with PLC␦ 1 H356AEGFP in NIH-3T3 cells. Epifluorescence images using a monoclonal antibody directed against PIP 2 (Assay Designs) in NIH-3T3 fibroblasts incubated with (A) with no vesicles, (B) PI(4)P vesicles, or (C) PIP 2 vesicles. As a qualitative assessment, this detergent-resistant fraction of PIP 2 was compared with asynchronous cells (D), and no significant change was observed after cells were synchronized to the G 1 /S boundary (E). FϪI show deconvolved images of a single plane near the center of the nucleus with DAPI alone (F, blue; arrow i indicates lightly stained/decondensed chromatin; arrow ii indicates intensely stained/condensed chromatin), transiently expressed PLC␦ 1 H356AEGFP (G, green, merged with DAPI), PIP 2 monoclonal antibody (H, red, merged with DAPI), or an overlay of all three channels (arrow iii indicates areas of moderately stained chromatin; PIP 2 , but not PLC␦ 1 H356AEGFP, localizes to these areas). PLC␦ 1 and PIP 2 have a similar non-uniform distribution throughout the nuclear compartment and display partial co-localization (yellow color in the overlay). White scale bars, 5 m.
protein that shuttles between the cytoplasmic and nuclear compartments; a nuclear export signal in the EF hand domain (residues 164 -177) is similar to the known leucine-rich consensus export sequence LLLXLLXXLXLX. In Madin-Darby canine kidney cells, point mutation at the putative export sequence L172A/I174A resulted in significant nuclear accumulation, whereas the wild-type protein was mainly localized to the cytoplasm and plasma membrane and did not significantly accumulate in the nucleus (22). It was also demonstrated that nuclear export of PLC␦ 1 was CRM1-dependent because treatment of cells with LMB, a toxin that inhibits CRM1-dependent nuclear export, resulted in dramatically increased levels of this protein in the nucleus (22). Using a series of deletion mutants, this group also identified residues (Lys 432 -Lys 487 ) of PLC␦ 1 that were necessary for nuclear import (23). They found that point mutations K432A and K434A, in combination, prevented PLC␦ 1 from accumulating in the nucleus despite the cell treatment with LMB, indicating that these residues were critical for import (23). Thus, PLC␦ 1 is capable of continuously shuttling between the nucleus and the cytoplasm. This was not related to the GFP tag or overexpression of this protein because endogenous PLC␦ 1 accumulates in the nucleus of NRK cells treated with LMB as well (22). Consistent with these results, using indirect immunofluorescence and Western blotting, we found that endogenous PLC␦ 1 accumulates in the nucleus at the G 1 /S boundary in C6 glioma cells.
Our results are most succinctly explained by high affinity PIP 2 binding of PLC␦ 1 that sequesters this protein in the nucleus when nuclear PIP 2 levels are elevated. Moreover, the reduction in whole cell PIP 2 and elevation of nuclear levels of this lipid are consistent with an increase in the nuclear to total PIP 2 ratio up to 20-fold (Fig. 8A). Given constant intrinsic rates of import and export, this dramatic change in the PIP 2 ratio can readily explain the accumulation of this enzyme at steady state during these phases of the cell cycle (Fig. 8B).
Our results support the idea that bound PLC tracks the available PIP 2 levels as they rise or fall in each compartment as a function of cell cycle progression. Because bound PLC falls, more free PLC is available for transport, and vice versa (Fig.  8B). Even if the intrinsic PLC transport rate ratio remains constant, changes in available PIP 2 levels will produce new steady-state concentrations of total PLC (bound ϩ free PLC) in each compartment. Thus, by increasing the ratio of nuclear to extranuclear PIP 2 levels, an increase in total PLC in the nucleus can occur. Whether the actual transport rates of our protein change during the cell cycle, however, has yet to be addressed.
Nuclear accumulation of GFP-(R40A)PLC␦ 1 , bearing a point mutation that greatly diminishes PIP 2 interactions, is more rapid and extensive than that of wild-type GFP-PLC␦ 1 , under conditions in which nuclear export is blocked (22). This is readily explained by our model: the levels of the R40A mutant in two compartments may simply reflect the effective rate of transport into the nucleus, which is faster when more PLC is free in the cytosol. Under our conditions that do not block export, the intrinsic transport rates appear to give rise to a A, this simple model helps explain the shift of total [PLC] to the nuclear compartment at the G 1 /S boundary and in G 0 as the ratio of nuclear/total PIP 2 increases. B, PLC B and PLC F are the concentrations of free and bound PLC in each compartment. The nuclear transport process can be described as the overall ratio of the intrinsic rates of transport of PLC into and out of the nucleus. PLC B tracks the available PIP 2 levels as they rise or fall. As PLC B falls, more PLC F is available for transport and vice versa. Even if the intrinsic PLC transport rate ratio remains constant, changes in available PIP 2 levels will produce new steadystate concentrations of total PLC (PLC B ϩ PLC F ) in each compartment. Thus, by reducing extranuclear PIP 2 levels and increasing nuclear PIP 2 levels, a metabolic enzyme operating on this substrate can be shifted to this compartment. This model only makes the reasonable assumption that PLC does not saturate the available PIP 2 or the intrinsic transport processes. steady-state nuclear to cytoplasmic ratio of the R40D mutation PLC, independent of high affinity PIP 2 binding. It is possible that an unknown nuclear binding partner of PLC␦ 1 could account for the nuclear accumulation we observe at the G 1 /S boundary; nonetheless, it is clear that the enzyme with an intact PH domain has a greater propensity to partition to the nuclear compartment during these stages of the cell cycle compared with PLC␦ 1 that cannot bind PIP 2 .
Homozygous deletion of PLC␦ 1 in mouse clearly indicates that this protein is not essential for overall cell division and growth; however, aberrant organization and expression of specific cell type markers and spontaneous skin tumors suggest that it plays a role in the balance between differentiation and proliferation in certain cell types (48). Furthermore, rescue experiments suggest that PLC␦ 1 activity is downstream of PLC␥ 1 activity and important for sustained calcium signaling during calcium-induced keratinocyte differentiation. Whether this relates to cytosolic or nuclear PLC activity is unknown. Because both calcium mobilization and phosphoinositide metabolism are closely linked to the cell cycle, it is possible that nuclear PLC␦ 1 plays a role in generating local second messengers in the nucleus and could thereby regulate a host of nuclear processes. Indeed, yeast Plcp1, a homolog to PLC␦ 1 , is found in the nucleus and localizes to centromeric loci at the G 2 /M checkpoint, affecting centromeric chromatin structure (49). Furthermore, mRNA export in yeast is dependent on Plcp1 activity (50), likely due to generation of inositol 1,4,5-trisphosphate by the yeast PLC, which is subsequently metabolized to inositol hexakisphosphate, a regulator of mRNA export (51). Currently, inositol 1,4,5-trisphosphate-mediated nuclear calcium mobilization is an area of great interest, and the emerging role of phosphoinositides in nuclear biology has received increasing attention. PIP 2 has been localized to most subcompartments of the nucleus, including the envelope, nucleoli, and interchromatin granule clusters (43), whereas a detergent-resistant PIP 2 component is associated with the nuclear matrix (1) and nuclear speckles (44 -46). Many enzymes that regulate PI metabolism co-localize to nuclear PIP 2 in these subcompartments (44 -46). Likewise, in fixed and detergent-permeabilized specimens, we find PLC␦ 1 and PIP 2 throughout the nuclear matrix and excluded from areas that stain intensely with DAPI, with partial co-localization by indirect immunofluorescence. PIP 2 is thought to bind actin-related proteins that are part of the Brg-or Brm-associated factor complex (52). In vitro studies demonstrate that PIP 2 is sufficient to target this complex to chromatin (52). Furthermore, PIP 2 binds to the C-terminal tails of histones H1 and H3 and counteracts histone H1-mediated basal transcription by RNA polymerase II (52)(53)(54). Clearly, the presence of nuclear phosphoinositides and their metabolism may influence a wide range of nuclear processes.
Metabolism of phosphoinositides can also influence nuclear accumulation of other PI-binding proteins, such as ING2 and Tub. Recent work demonstrates that overexpression of PIKII␤ decreases nuclear PI(5)P by converting it to PIP 2 , thereby decreasing nuclear accumulation of ING2 (27). ING2 is a protein that copurifies with components important to chromatin remodeling (55), is up-regulated in response to DNA damage (56), and is thought to be important for apoptosis (27). On the other hand, Tub is bound to PIP 2 in the plasma membrane via its tubby domain; upon PLC␤-mediated hydrolysis of this pool of PIP 2 , Tub is released and translocates to the nucleus (26). Although it has not been reported that Tub interacts with nuclear PIP 2 , a potential mechanism to target specific subcompartments within the nucleus may involve its tubby domain. Thus, a rise in the ratio of nuclear to extranuclear polyphos-phoinositides could cause the accumulation of various regulatory proteins in the nucleus. Recruitment of polyphosphoinositide-binding proteins such as PLC might influence gene regulation, mRNA processing, and DNA synthesis by modulating the levels of lipids and generating local second messengers; This could provide a simple mechanism to coordinate nuclear activities with cell surface events linked to polyphosphoinositide metabolism.