Phosphorylation of Protein Phosphatase 1 by Cyclin-dependent Protein Kinase 5 during Nerve Growth Factor-induced PC12 Cell Differentiation*

The transcription factor Egr-1 activates cyclin-dependent protein kinase 5 (Cdk5) during nerve growth factor (NGF)-induced differentiation of PC12 cells into neurons (Harada, T. Morooka, T., Ogawa, S., and Nishida, E. (2001) Nat. Cell Biol. 3, 453-459). The downstream target of Cdk5 in the Egr-1/Cdk5 pathway is not clear. In this study, we observed that phosphorylation of protein phosphatase 1 (PP1) on Thr320 is reduced in brain extracts from Egr-1-/- mice, indicating that a kinase downstream of Egr-1 phosphorylates PP1. In HEK 293 cells co-transfected with PP1 and Cdk5, Cdk5 phosphorylates PP1. In vitro, Cdk5 purified from bovine brain phosphorylates bacterially expressed recombinant PP1. In NGF-treated PC12 cells, inhibition of Cdk5 by olomoucine or silencing Cdk5 expression by small interfering RNA strategy, suppresses PP1 phosphorylation. Silencing Cdk5 expression by small interfering RNA also blocks NGF-induced neurite outgrowth. Overexpression of PP1 (wild type) promotes NGF-induced differentiation of PC12 cells, whereas that of PP1 (T320A) has no effect. Our data indicate that PP1 is a downstream target of the NGF/Egr-1/Cdk5 pathway during NGF-induced differentiation of PC12 cells and suggest that PP1 phosphorylation promotes neuronal differentiation.

Cyclin-dependent protein kinase 5 (Cdk5) is a heterodimer of a catalytic Cdk5 and a regulatory p25 subunit (10 -13). The p25 subunit is a proteolytic fragment of p35 protein, and both p35 and p25 activate the catalytic activity of Cdk5 subunit (13). Cdk5 is inactive in dividing cells but becomes progressively more active in differentiating cells (14,15) and is predominantly expressed in terminally differentiated neurons (10 -13). Cdk5 is involved in brain development, neuronal differentiation, and cell signaling and regulates microtubule dynamics. When PC12 cells are treated with nerve growth factor (NGF), they differentiate into neurons. The transcription factor Egr-1 (early growth factor-1) is induced by NGF and is essential for the differentiation (16,17). NGF activates Cdk5 via Egr-1 (15,18), and inhibition of Cdk5 activity blocks NGF-induced neurite outgrowth (15). These observations indicate that Cdk5 is a component of the Egr-1-dependent signaling pathway and that an Egr-1-driven increase in Cdk5 activity is required for differentiation of PC12 cells to neurons.
In this study, we observed that PP1 phosphorylation is reduced in the brain extracts of Egr-1 Ϫ/Ϫ mice, suggesting that a kinase that acts downstream of Egr-1 phosphorylates PP1 in the brain. This prompted us to investigate whether Cdk5 phosphorylates PP1. Herein, we show that PP1 is phosphorylated by Cdk5 in vitro, in transfected mammalian cells and in differentiating PC12 cells. We also demonstrate that in NGF-exposed PC12 cells, blocking PP1 phosphorylation by inhibiting Cdk5 activity, blocks neurite outgrowth. Overexpression of PP1 (WT) but not PP1 (T320A) promotes differentiation. Our data indicate that Cdk5 phosphorylates PP1 in vivo and suggest that PP1 phosphorylation has a role in neuronal differentiation.

EXPERIMENTAL PROCEDURES
Animals-Experimental mouse studies were conducted in adult male wild type (ϩ/ϩ) C57Bl/6 or Egr-1-deficient (Ϫ/Ϫ) mice (n ϭ 3). Egr-1 Ϫ/Ϫ mice have a C57Bl/6 genetic background. The originating parental mice lacking Egr-1 (19) were a gift from Dr. Jeffrey Milbrandt (University of St. Louis, St. Louis, MO). All of the experiments were performed according to the regulations of the Canadian Council of Animal Care and the Animal Care Committee of the Lady Davis Institute for Medical Research. The mice were genotyped by PCR using DNA obtained by tail docking and amplification using genespecific primers and Taq DNA polymerase. Primer pairs were neo-929 (5Ј-CTCGTGCTTTACGGTATCGC-3Ј), Egr-1-1263 (5Ј-AACCGGCCCAGCAAGACACC-3Ј), and Egr-1-1677 (5Ј-GGGCACAGGGGATGGGAATG-3Ј). A 414-bp DNA was amplified from C57Bl/6 mice using the 1263 ϩ 1677 primer combination, and no DNA was amplified using the neo929 ϩ 1677 primers. A 520-bp DNA fragment was amplified from Egr-1 Ϫ/Ϫ mice neo929 ϩ 1677 primers, and no DNA was amplified using the 1263 ϩ 1677 primers. The mice were killed at 3 months of age by cervical dislocation, and their brains were removed immediately. Each brain was quickly homogenized in an equal amount (w/v) of cold extraction buffer (50 mM Hepes, pH 7.2, 1 mM EDTA, 1 mM EGTA, 0.1 mM dithiothreitol, 50 mM ␤-glycerol phosphate, 0.2 mM okadaic acid, 10 mM NaF, and 1 g/ml each of leupeptin, pepstatin, and aprotinin) using a Down's glass homogenizer. The samples were centrifuged at 14,000 ϫ g for 20 min using a bench top centrifuge. The resulting supernatants were used to generate Fig. 1.
Transfection of siRNA-siRNA-Cdk5 (SMARTpool containing four pooled SMARTselected siRNA duplex) and a nonspecific control were purchased from Upstate Cell Signaling Solution. PC12 cells were transfected with siRNA using siMPORTER TM (Upstate Cell Signaling Solution) reagent following the manufacturer's instructions. NGF was added 6 h after transfection. The cells were harvested at the indicated time points.
Neurite Outgrowth Assay and Immunocytochemical Staining-NGF-treated cells grown on cover slips were fixed at room temperature with 4% paraformaldehyde in PBS for 30 min, washed with PBS containing 0.1% Triton X-100, and then permeabilized by incubation with PBS containing 1% bovine serum albumin and 0.1% Triton X-100 for 60 min. The cells were then incubated with monoclonal anti-Myc antibody (1:500 in PBS containing 1% bovine serum albumin and 0.1% Triton X-100) for ϳ12 h at 4°C. The washed cells were incubated with Alexa Fluor 488 goat anti-mouse IgG (Molecular Probes) for 1 h at room temperature, washed, and finally visualized and photographed using a Nikon fluorescent microscope. Neurite length was measured on the photographs using a micrometer (26,27). Only those neurites that were equal or longer than the cell diameter were considered to be neurites, and cells that contained one or more neurites were regarded as having neurite outgrowth. Cells that had one or more neurites with the length twice or more than twice the diameter of the cell body were considered differentiated according to previously described criteria (28). More than 100 cells were quantitated in each experimental group.
In Vitro Phosphorylation-Phosphorylation of PP1 by Cdk5 was carried out at 30°C in a reaction mixture containing 0.5 mg/ml of PP1, 1 mM EDTA, 1 mM dithiothreitol, 10 mM MgCl 2 , 10 mM NaF, 50 mM ␤-glycerol phosphate, 0.2 M okadaic acid, 1 mM ATP, and 400 units/ml of Cdk5. The reaction was initiated by adding an aliquot of Cdk5 into the mixture containing the assay components. At the indicated time points, the aliquots were withdrawn and mixed with an equal volume of SDS/ PAGE sample buffer, boiled, and analyzed by Western blot analysis using anti-PP1 or anti-Thr(P) 320 antibody.
Cdk5 activity in mouse brain extract was assayed as described previously (20) using a synthetic peptide substrate of Cdk5 (KTPKKAKKPKTPKKAKKL) (29). The protein concentration of each brain extract was adjusted to 2 mg/ml by dilution with extraction buffer. The final concentrations of the assay components were: 25 mM Hepes, pH 7.2, 0.1 mM EDTA, 0.1 mM EGTA, 0.1 mM dithiothreitol, 10 mM NaF, 50 mM ␤-glycerol phosphate, 20 nM okadaic acid, 10 mM MgCl 2 , 0.5 mM [␥-32 P]ATP, and 50 M peptide substrate. The reaction was initiated by the addition of 10 l of brain extract to 40 l of the assay mixture. After 30 min at 30°C, the reaction was stopped by adding 10 l of 50% trichloroacetic acid. The vials were incubated at 4°C for 10 min and then centrifuged for 5 min using a bench top centrifuge at 10,000 rpm. The supernatant (20 l each) was withdrawn and analyzed for the amount of radioactivity incorporated into the substrate using a phosphocellulose filter paper assay (20). Cdk5 specific activity is expressed as the nanomoles of phosphate transferred to the peptide substrate/min/mg protein in the brain extract.
Egr-1 is a transcription factor (16,17) and so may regulate PP1 phosphorylation by controlling the expression of a kinase that phosphorylates PP1. Thr 320 of PP1 is a proline-directed phosphorylation site recognized by cyclin-dependent kinases: Cdk1 and Cdk2 (3)(4)(5)(6). Cdk1 and Cdk2 are not expressed significantly in adult brain, whereas Cdk5 shares a very similar substrate specificity with Cdk1 and Cdk2 (30) and is the major cyclin-dependent kinase in the neurons of mammalian brain (10 -12). Previous studies showed that Egr-1 activates Cdk5 in neurons by increasing the expression of p35 subunit of Cdk5 (15,18). If Egr-1 activates brain Cdk5, then Egr-1 Ϫ/Ϫ brains would be expected to have reduced p35 level and Cdk5 activity. To test this possibility, we analyzed the levels of Cdk5, p35, and p25 by Western blotting and measured Cdk5 activity in Egr-1 ϩ/ϩ and Egr-1 Ϫ/Ϫ brain extracts. Because p35 is proteolyzed to p25, we determined the level of p35 based on the levels of p25 and p35. The Cdk5 level was similar, but the sum of the levels of p35 and p25 was ϳ2.7-fold more in Egr-1 ϩ/ϩ than in Egr-1 Ϫ/Ϫ brain extracts (Fig. 1, A and B). The Cdk5 specific activity in Egr-1 Ϫ/Ϫ brain extracts was ϳ38% less than that of Egr-1 ϩ/ϩ (Fig. 1C). Thus, as expected, both p35/p25 level and Cdk5 activity are reduced in Egr-1 Ϫ/Ϫ brains. The observation of reduced PP1 phosphorylation coupled to reduced Cdk5 activity in Egr-1 Ϫ/Ϫ brain suggests that Cdk5 may be a kinase that phosphorylates PP1 in the brain. FIGURE 1. PP1 phosphorylation is reduced in Egr-1 ؊/؊ mouse brain. Three Egr-1 ϩ/ϩ and three Egr-1 Ϫ/Ϫ mice brains were homogenized under identical conditions. Each homogenate was analyzed by Western blot analysis using the indicated antibodies or subjected to Cdk5 activity assay. Based on the blot band intensity, the relative amount of indicated proteins and phosphorylated PP1 in each sample was determined. A, Western blots. B, relative amounts. C, Cdk5 specific activity. To determine the relative amount, the band intensity value of indicated protein in each sample of corresponding blot was normalized against the corresponding actin band intensity value of that sample. To calculate the relative amount of phosphorylated PP1, the phosphorylated PP1 band intensity value of each sample was normalized against the band intensity value of PP1 in that sample. Cdk5 specific activity was determined as described under "Experimental Procedures." The values are the averages of three determinations. IB, immunoblot.
We next co-transfected Myc-PP1 (WT) with Xpress-Cdk5 and FLAG-p25 (Xpress-Cdk5/FLAG-p25) in HEK 293 cells. Controls included co-transfection of Myc-PP1 (T320A) with Xpress-Cdk5/FLAG-p25 and Myc-PP1 (WT) with Cdk5 mutant, HA-Cdk5 (DN). HA-Cdk5 (DN) with Asp 144 mutated to Asn, although reported to contain no detectable kinase activity when co-transfected with p35 in C33A cervical carcinoma cells (13), displays kinase activity ϳ10% of Cdk5 (WT) when co-transfected with FLAG-p25 in HEK 293 cells (data not shown). Transfected cells were lysed and analyzed by immunoblot analysis for the expression of the various transfected genes and PP1 phosphorylation. The relative amount of phosphorylated PP1 was calculated based on the band intensities of PP1 and phosphorylated PP1.
As shown in Fig. 3A, in cells co-transfected with Myc-PP1 (WT) and Xpress-Cdk5/FLAG-p25, Myc-PP1 was phosphorylated ϳ3.5-fold more than basal level (compare lanes 2 and 3 and see F). In cells co-transfected with Myc-PP1 (WT) and HA-Cdk5 (DN)/p25-FLAG, Myc-PP1 phosphorylation was slightly above basal level (compare lanes 2 and 6 and see F). These data indicate that PP1 is highly phosphorylated by Cdk5 and very modestly by Cdk5 (DN). To extend and confirm these results, we treated HEK 293 cells, co-transfected with Myc-PP1 and Xpress-Cdk5/FLAG-p25, with the Cdk5 inhibitor olomoucine (21,31) or vehicle. Treated cells were then analyzed for PP1 phosphorylation by Western blot analysis as described above. PP1 was highly phosphorylated in vehicle-treated cells; however, PP1 phosphorylation progressively diminished with increasing olomoucine concentration. When the concentration of olomoucine reached 200 M, PP1 phosphorylation was almost at basal level (supplemental Fig. S1). Based on these results, we conclude that Cdk5 phosphorylates PP1 on Thr 320 in HEK 293 cells transfected with Myc-PP1 and Xpress-Cdk5/FLAG-p25.
The p25 regulatory subunit of Cdk5 is formed by the proteolytic cleavage of p35 protein (13). Although p25 activates Cdk5, p35 is the natural activator of Cdk5 (11,12). To evaluate whether Cdk5/p35 phosphorylates PP1, we transfected Myc-PP1 and Xpress-Cdk5/FLAG-p35 in HEK 293 cells. Transfected cells were lysed and analyzed for expression of the transfected genes and Myc-PP1 phosphorylation as described above. The relative amount of phosphorylated Myc-PP1 in cells transfected with Myc-PP1 and Xpress-Cdk5/FLAG-p35 was ϳ3-fold more than in those cells transfected with Myc-PP1 alone (data not included). These results determined that Cdk5/ p35, like Cdk5/p25, phosphorylates PP1.
In Vitro Phosphorylation-To test whether Cdk5 phosphorylates PP1 directly, we performed in vitro kinase assays. Bacterially expressed recombinant PP1 was incubated with purified Cdk5 in the presence of ATP/Mg 2ϩ for various time points. Incubated samples were analyzed for PP1 phosphorylation by immunoblot analyses using anti-Thr(P) 320 antibody (Fig. 4A).
No immunoreactive band against anti-Thr(P) 320 was observed in samples containing Cdk5 alone incubated with ATP/ Mg 2ϩ (lane 1), and PP1 did not become phosphorylated when incubated alone with ATP/Mg 2ϩ (lane 2). In samples containing PP1, Cdk5, and ATP/Mg 2ϩ , PP1 phosphorylation increased with increasing incubation time (lanes 3-7). These observations demonstrate that Cdk5 phosphorylates PP1 in vitro. The  Cdk5 used in this experiment was purified from bovine brain extract. To rule out the possibility that a contaminant kinase in our Cdk5 preparation may have phosphorylated PP1, we performed in vitro kinase assay as described above for 30 min in the presence of increasing concentrations of the Cdk5 inhibitor olomoucine (Fig. 4C). Incubation with 50 M olomoucine suppressed ϳ60% of PP1 phosphorylation (lane 5). When the concentration of olomoucine was increased to 100 M, PP1 phosphorylation was almost completely suppressed (lane 6). Taken together, these observations indicate that Cdk5 phosphorylates PP1 in vitro.
PP1 Associates with Cdk5-To characterize the interaction between Cdk5 and PP1, we co-transfected Xpress-Cdk5 with Myc-PP1 (WT) or Myc-PP1 (T320A) in HEK 293 cells. Transfected cells were lysed, and each lysate was subjected to immunoprecipitation using either anti-Myc antibody or control IgG. Each immune complex was then immunoblotted against anti-Xpress or anti-Myc antibody. As shown in Fig. 5A, Xpress-Cdk5 specifically co-immunoprecipitated with Myc-PP1 (WT) (lane 3). These data are consistent with our previous report (21) and indicates that PP1 binds with Cdk5. Xpress-Cdk5 also coimmunoprecipitated with Myc-PP1 (T320A) (lane 4), indicating that mutation of Thr 320 to Ala does not significantly interfere with the binding of PP1 to Cdk5. Because PP1 is phosphorylated on Thr 320 in HEK 293 cells (Fig. 3, lane 2), these data also indicate that blocking Thr 320 phosphorylation does not prevent PP1 from binding to Cdk5.
PP1 Is Phosphorylated during Neuronal Differentiation-PC12 cells treated with NGF differentiate into neurons (32). In these cells, NGF activates Egr-1, which in turn activates Cdk5 by inducing the expression of p35 (15). Cdk5 activity is required for PC12 cell differentiation (15). It is possible that NGF-activated Cdk5 phosphorylates PP1, and this phosphorylation may be an important event for neuronal differentiation. To test this idea, we treated PC12 cells with NGF for various time points and then analyzed the treated cells.
Cdk5 Is Involved in PP1 Phosphorylation in NGF-induced Differentiating PC12 Cells-Consistent with a previous report (14), Western blot analysis determined that Cdk1 and Cdk2 proteins were highly expressed in PC12 cells exposed to NGF for 0 day but became undetectable on day 3 of NGF exposure (data not shown). These data suggest that PP1 may have been phosphorylated by Cdk1 and/or Cdk2 at day 0 and 1 but that these kinases are less likely to be responsible for significant PP1 phosphorylation with increasing time of differentiation.
To evaluate whether Cdk5 is involved, PC12 cells were exposed to NGF for 3 days and then treated with Cdk5 inhibitor olomoucine. Treated cells were lysed and analyzed for PP1 phosphorylation. As shown in Fig. 7, olomoucine inhibited PP1 phosphorylation in a dose-dependent manner (Fig. 7, B and C) without affecting the level of PP1 (Fig. 7, A and C). These data indicate that a kinase sensitive to olomoucine phosphorylates  PP1 in NGF-exposed PC12 cells. Olomoucine, in addition to Cdk5, also inhibits Cdk1 and Cdk2 (31). Therefore, to confirm whether Cdk5 phosphorylates PP1, we suppressed Cdk5 expression by siRNA strategy. We transfected PC12 cells with siRNA-Cdk5 or siRNA-control. Transfected cells were then exposed to NGF for various time points and analyzed by immunoblot analysis. The levels of Cdk5, PP1, and phosphorylated PP1 were quantitated and compared (Fig. 7, D-G).
Cdk5 expression was not affected by siRNA-Cdk5 in day 0 and day 1 cells because the relative amount of Cdk5 was similar in cells transfected with siRNA-Cdk5 and siRNA-control (Fig.  7G). By day 2, siRNA-Cdk5 suppressed Cdk5 expression by ϳ80% (Fig. 7, D, lane 6, and G). PP1 expression was similar at all time points (Fig. 7, E and G), indicating that suppression of Cdk5 expression does not affect PP1 level. PP1 phosphorylation on the other hand was similar in siRNA-Cdk5 and siRNA-control transfected day 0 cells (compare lanes 1 and 2 in Fig. 7F). However, on day 1, PP1 phosphorylation was suppressed by ϳ30%, and by day 2, the relative amount of phosphorylated PP1 was ϳ 1 ⁄ 10 that of day 0 (Fig. 7G). Thus, silencing ϳ80% Cdk5 expression suppressed ϳ90% PP1 phosphorylation in PC12 cells treated with NGF for 2 days. Based on these results, we concluded that Cdk5 phosphorylates PP1 during NGF-induced differentiation of PC12 cells.
Effect of Silencing Cdk5 Expression on PC12 Cell Differentiation-To determine the role of PP1 phosphorylation, we suppressed PP1 phosphorylation by suppressing Cdk5 expression and quantitated neurite outgrowth. We transfected PC12 cells with siRNA-control and siRNA-Cdk5 and exposed the cells to NGF. The relative amount of phosphorylated PP1 in cells transfected with siRNA-control remained unchanged (ϳ0.78 versus ϳ0.79) after 1 and 2 days of NGF treatment (data not shown). The relative number of cells displaying neurite outgrowth in these cells is 18% at day 1 and increased to ϳ36% at day 2 (Fig. 8). However, in cells transfected with siRNA-Cdk5, the relative amount of phosphorylated PP1 is reduced (ϳ0.72 versus ϳ0.30) (data not shown), and the number of cells showing neurite outgrowth did not change significantly (ϳ20 and ϳ15%) (Fig. 8) at days 1 and 2, respectively. Thus, from day 1 to day 2, siRNA-control transfected cells maintained a constant level of phosphorylated PP1 and doubled the number of cells displaying neurite outgrowth. Cells transfected with siRNA-Cdk5 maintained PP1 phosphorylation at day 1 and contained a level of neurite outgrowth similar to that observed in cell transfected with siRNA-control. By day 2, cells transfected with siRNA-Cdk5 had significantly suppressed amounts of phosphorylated PP1 and did not show the expected increase in the number of cells displaying neurite outgrowth. These data demonstrate that suppression of Cdk5 expression by siRNA-Cdk5 blocks PP1 phosphorylation and neurite outgrowth.
Effect of Overexpression of PP1 (WT) and PP1 (T320A) on NGF-induced Differentiating PC12 Cells-Cdk5 in addition to PP1 also phosphorylates a number of cellular proteins (11). Suppressing Cdk5 expression by siRNA will also suppress phosphorylation of these other targets. Hence, the failure of cells transfected with siRNA-Cdk5 to advance to differentiation, could be due to suppressed phosphorylation of Cdk5 targets other than PP1. Therefore, to more specifically evaluate the role of PP1 phosphorylation, we transfected PC12 cells with Myc-PP1 (WT), Myc-PP1 (T320A), and Myc-vector control. Transfected cells were treated with NGF to activate cellular Cdk5. In Myc-PP1 (WT) transfected cells, activated Cdk5 would be expected to phosphorylate Myc-PP1 (WT) and increase the cellular level of phosphorylated PP1. In Myc-PP1 (T320A) and Myc-vector transfected control cells, the cellular level of phosphorylated PP1 would be expected to remain at similar but basal level. We then determined the number of cells committed to differentiation by measuring neurite outgrowth. The cells were considered fully differentiated when they displayed one or more neurites equal or longer than twice the diameter of the cell body (28). Because multipolar neurons that predominate mammalian brain contain multiple neurites (a single axon and one or more dendrites that emerge from all parts of the cell body), we also analyzed the number of neurites/cell.
As shown in Fig. 9 (A and B), ϳ64 and ϳ52% of Myc-vector and Myc-PP1 (T320A) transfected cells displayed neurites, respectively. In Myc-PP1 (WT) transfected cells, this number rose to ϳ79%. Only ϳ15% each of Myc-vector and Myc-PP1 (T320A) transfected cells displayed fully differentiated phenotype (Fig. 9, A and C). For Myc-PP1 (WT) transfected cells, ϳ27% of cells were fully differentiated. Furthermore, among FIGURE 6. Phosphorylation of PP1 is sustained during PC12 differentiation. PC12 cells were treated with NGF for the indicated time points then lysed. Protein (100 g) from each lysate was analyzed by Western blot using the indicated antibodies. Based on the blot band intensities, the relative amount of PP1 and phosphorylated PP1 for each sample was calculated. A and B, Western blots. C, relative amounts. To determine relative amount of PP1, the band intensity value of each sample in A representing total PP1 is expressed as the fraction of the band intensity value of day 0 sample in A. The values are the averages of three independent determinations. The differences at various time points are not significant (p Ͼ 0.05). To determine the relative amount of phosphorylated PP1, the band intensity value of each sample in B representing phosphorylated PP1 was normalized against the band intensity value of total PP1 of the sample in A. The resulting value of each sample was then expressed as the fraction of the resulting value of sample corresponding to day 0. The values are the averages of three independent determinations. The differences at various time points are not significant (p Ͼ 0.05). IB, immunoblot.
neurite containing Myc-vector transfected cells, ϳ52 and ϳ33% contained one and two neurites, respectively, and ϳ15% contained three or more neurites (Fig. 9D). Among neurite con-taining Myc-PP1 (T320A) transfected cells, ϳ47 and 32% contained one or two neurites, respectively, and ϳ21% advanced to three or more neuritic stage. In contrast, ϳ35 and 27% of Myc-PP1 (WT) transfected cells had one and two neurites, respectively, and ϳ38% advanced to higher differentiation stage containing three or more neurites. Thus, overexpression of Myc-PP1 (T320A) did not affect PC12 differentiation. Overexpression of Myc-PP1 (WT), however, significantly promoted neurite outgrowth, neurite extension, and the number of neurites per cell. These observations indicate that overexpression of PP1 (WT) promotes neuronal differentiation.

DISCUSSION
Cdk1 and Cdk2 phosphorylate and inactivate PP1 in dividing cells, and this phosphorylation plays an important role in cell cycle regulation (3)(4)(5)(6)(7)(8)(9). To examine PP1 phosphorylation by Cdk5 in the brain and in NGF-treated PC12 cells, we used an antibody that specifically cross-reacts with PP1 phosphorylated on Thr 320 . We first confirmed the specificity of the antibody by using PP1 (T320A) mutant. Our subsequent results show that PP1 is highly phosphorylated when cotransfected with Cdk5 in HEK 293 cells (Fig. 3). In NGF-treated PC12 cells, silencing Cdk5 expression or suppressing Cdk5 activity significantly reduced PP1 phosphorylation (Fig. 7). Finally, in vitro purified Cdk5 phosphorylated bacterially expressed recombinant PP1 (Fig. 4). Taken together, our data indicate that Cdk5 phosphorylates PP1 in vitro and in vivo. It is interesting that in HEK 293 cells, the basal phosphorylation of PP1 is insensitive to Cdk inhibitor olomoucine (supplemental Fig. S1). These data suggest that PP1 is also phosphorylated by a kinase(s) other than Cdk1, Cdk2, and Cdk5 in vivo. Morfini et al. (33) demonstrated that in rat cortical neurons PP1 was activated when Cdk5 was inactivated, indicating that Cdk5 suppresses PP1 activity in vivo. However, the authors also reported that Cdk5 did not FIGURE 7. PP1 phosphorylation is inhibited by olomoucine and Cdk5 targeted siRNA. A-C, inhibition by olomoucine. PC12 cells exposed to NGF for 72 h were treated with indicated concentration of olomoucine for 1 h. Treated cells were lysed and then analyzed by Western blot analysis. Based on band intensities of the blots, relative amount of PP1 and phosphorylated PP1 in various samples was determined as in Fig. 6. D-G, inhibition by siRNA-Cdk5. PC12 cells transfected with siRNA-Cdk5 or siRNA-control and treated with NGF for the indicated time points were analyzed by Western blot. Based on blot band intensities, relative amounts of Cdk5, PP1, and phosphorylated PP1 were calculated. To calculate relative amount of Cdk5, the band intensity value of each sample in D transfected with siRNA-Cdk5 (or siRNA-control) is expressed as the fraction of the band intensity value of the sample in D transfected with siRNA-Cdk5 (or siRNA-control) for 0 day. The relative amount of PP1 for indicated sample was also calculated as described above for Cdk5 except that the band intensity values of blot E representing total PP1 were used. To calculate the relative amount of phosphorylated PP1, the band intensity value of each sample in the blot F representing phosphorylated PP1 was normalized against the band intensity value of that sample in E representing total PP1. The resulting value of each sample is then expressed as the fraction of the resulting value of corresponding sample representing cells treated with NGF for 0 day. IB, immunoblot. MARCH 2, 2007 • VOLUME 282 • NUMBER 9 phosphorylate PP1 in vitro and hence that Cdk5 inactivated PP1 by a mechanism that did not require PP1 phosphorylation. This study contradicts our in vitro phosphorylation data presented here.

PP1 Phosphorylation by Cdk5
Morfini et al. (33) did not fully describe their in vitro phosphorylation procedure. Therefore, we are unable to determine the cause of the disparity between their results and our data. There are unique problems associated with in vitro phosphorylation of PP1 (4,6). PP1 autodephosphorylates (4), and hence, its phosphorylation can only be detected in the presence of high concentration of PP1 inhibitors. Furthermore, PP1 phosphorylation greatly depends upon the purity of the kinase used. For example, highly purified Cdk1/cyclin A phosphorylates PP1, but when less purified Cdk1/cyclin A is used, PP1 is not phosphorylated significantly (4). It is possible that Morfini et al. (33) may not have included sufficiently high doses of PP1 inhibitors and/or used a Cdk5 that was not purified in their phosphorylation mixture.
PC12 cells are an excellent and widely used model for studying neuronal differentiation (32). In these cells, binding of NGF to its cell surface receptor leads to the activation MAP kinase. Activated MAP kinase phosphorylates ternary complex factor, which together with serum response factor induces Egr-1 (16,17). Indeed Egr-1 was first identified as a transcription factor increased within hours after NGF treatment of PC12 cells (17). The MAP kinase/Egr-1/Cdk5 pathway is one of the major signaling events triggered by NGF and leads to neurite extension. Inhibition of Cdk5 activity blocks neurite outgrowth (15), but the targets of Cdk5 in this pathway are not completely known.
In the current study, we observed that PP1 is phosphorylated in PC12 cells exposed to NGF (Fig.  6). When Cdk5 expression is silenced, PP1 phosphorylation is significantly suppressed (Fig. 7G). In vitro and in transfected HEK 293 cells, Cdk5 phosphorylates PP1 (Figs. 3 and 4). Our data demonstrate that PP1 is phosphorylated by Cdk5 in NGF-exposed differentiating PC12 cells and suggest that PP1 is downstream of Cdk5 in the NGF/ MAP kinase/Egr-1/Cdk5 pathway. In dividing PC12 cells, Cdk5 is inactive and is only activated when cells are treated with NGF (14,15). Therefore, one would expect PP1 phosphorylation to surge concomitant with Cdk5 activation. However, as shown in Fig. 6, PP1 is phosphorylated in dividing cells (day 0) and remains phosphorylated at the same level after cells are exposed to NGF and throughout the differentiation period. This observation indicates that PP1 phosphorylation is sustained at a constant level during the transition of dividing PC12 cells to neuronal differentiated PC12 cells.  Withdrawal from the cell cycle is a prerequisite for differentiation, and it has been suggested that terminal mitosis and differentiation must be coupled for successful differentiation (34). Cdk1 and Cdk2, which are active in dividing cells, are inactivated early in the neuronal differentiation period (14). Cdk5 on the other hand is inactive in dividing cells but becomes progressively more active in differentiating cells (14,15). It is possible that when PC12 cells are exposed to NGF, PP1 is phosphorylated initially by Cdk1 and/or Cdk2. With increasing time of NGF exposure, PP1 may be less and less phosphorylated by Cdk1 and Cdk2 and more and more by Cdk5. Thus, a constant level of phosphorylated PP1 is maintained, and this may be important to couple terminal mitosis to neuronal differentiation. NGF-exposed PC12 cells with suppressed Cdk5 expression and PP1 phosphorylation displayed significantly impaired differentiation as exemplified by a reduced number of cells with neurite outgrowth when compared with control cells with normal levels of Cdk5 and phosphorylated PP1 (Fig. 8). This observation suggests that PP1 phosphorylation may be required for neurite outgrowth. However, Cdk5 phosphorylates a number of cellular proteins. Therefore, to more clearly understand the importance of PP1 phosphorylation, we transfected PP1 (WT) or PP1 (T320A) into PC12 cells. PP1 (WT) enhanced neurite outgrowth, increased the number of neurites/cell, and promoted neurite extension to the fully differentiated phenotype (Fig. 9). PP1 (T320A) on the other hand, did not influence significantly any of the differentiation characteristics analyzed in this study. Because the main difference between the PP1 (WT) and PP1 (T320A) is that only former is phosphorylated, these data suggest that PP1 phosphorylation promotes neuronal differentiation. Many proteins involved in the regulation of cytoskeletal dynamics and membrane trafficking are phosphorylated in vivo; and some of these proteins are PP1 targets (35). It is possible that PP1 phosphorylation may keep PP1 inactive and allow proper cytoskeletal remodeling and/or membrane trafficking required for differentiation (36).
Cdk5 requires the p35 subunit for its activity (11). In PC12 cells, Egr-1 induces the expression of p35 and thereby activates Cdk5 (15). Expression of NAB2, a co-repressor of Egr-1, almost completely abolishes NGF-induced p35 expression (15). These data indicate that Egr-1 controls p35 expression in NGF-treated PC12 cells. In the current study, we find that the level of p35/p25 protein in the Egr-1 Ϫ/Ϫ brain extract is ϳ2.7-fold less than in the Egr-1 ϩ/ϩ mice brains (Fig. 1). Our results indicate that in adult rat brain Egr-1 promotes p35 expression but that p35 expression is also regulated independent of Egr-1.
In NGF-exposed PC12 cells, suppression of ϳ80% of Cdk5 expression, inhibits ϳ90% of PP1 phosphorylation (Fig. 7G), indicating that Cdk5 is the major PP1 kinase in these cells. Egr-1 Ϫ/Ϫ brains, which have ϳ38% less Cdk5 activity than Egr-1 ϩ/ϩ brains, also display ϳ50% less PP1 phosphorylation than Egr-1 ϩ/ϩ brains (Fig. 1). This correlation between Cdk5 activity and PP1 phosphorylation suggests that Cdk5 may be a major PP1 kinase in the adult mouse brain also. It would be interesting to validate this suggestion by analyzing PP1 phosphorylation in adult mouse brains deficient in Cdk5 activity. However, Cdk5 null mice suffer perinatal death (37). The p35 null mice survive to adulthood (38), but in these mice p39, an isoform of p35 (39), compensates for the loss of p35 function, and Cdk5 remains active (38). It should be noted that the level of p35/p25 in Egr-1 Ϫ/Ϫ brains is ϳ36% of Egr-1 ϩ/ϩ brains, but Cdk5 activity in Egr-1 Ϫ/Ϫ brains is ϳ62% of that in Egr-1 ϩ/ϩ brains (Fig. 1). These data indicate that Cdk5 activity does not correlate with the level of p35 in Egr-1 Ϫ/Ϫ brains. It is possible that deletion of Egr-1 Ϫ/Ϫ gene suppresses p35 expression and the brain in turn enhances the expression of p39 to compensate the loss of p35. Analysis of p39 expression in Egr-1 Ϫ/Ϫ and Egr-1 ϩ/ϩ brains will substantiate this hypothesis.
Previous studies showed that inactivation of PP1 is involved in the induction and maintenance of long term potentiation, a form of synaptic plasticity that contributes to learning and memory (35, 40 -42). In vitro, PP1 is inactivated by Thr 320 phosphorylation (3)(4)(5)(6). In the brain, Egr-1 concentration increases with neuronal activity, and Egr-1 is required for the induction of long term potentiation (43)(44)(45). PP1 phosphorylation is reduced in Egr-1 Ϫ/Ϫ mouse brains (this study), and Cdk5, which phosphorylates PP1 in response to Egr-1 activation, is involved in the regulation of neuronal plasticity (46 -48). These observations together suggest that PP1 phosphorylation may play a role in the induction and maintenance of long term potentiation.