Cooperative phosphorylation of the tumor suppressor phosphatase and tensin homologue (PTEN) by casein kinases and glycogen synthase kinase 3beta.

The phosphatase and tensin homologue (PTEN) tumor suppressor is a phosphatidylinositol D3-phosphatase that counteracts the effects of phosphatidylinositol 3-kinase and negatively regulates cell growth and survival. PTEN is itself regulated by phosphorylation on multiple serine and threonine residues in its C terminus. Previous work has implicated casein kinase 2 (CK2) as the kinase responsible for this phosphorylation. Here we showed that CK2 does not phosphorylate all sites in PTEN and that glycogen synthase kinase 3beta (GSK3beta) also participates in PTEN phosphorylation. Although CK2 mainly phosphorylated PTEN at Ser-370 and Ser-385, GSK3beta phosphorylated Ser-362 and Thr-366. More importantly, prior phosphorylation of PTEN at Ser-370 by CK2 strongly increased its phosphorylation at Thr-366 by GSK3beta, suggesting that the two may synergize. Using RNA interference, we showed that GSK3 phosphorylates PTEN in intact cells. Finally, PTEN phosphorylation was affected by insulin-like growth factor in intact cells. We concluded that multiple kinases, including CK2 and GSK3beta, participate in PTEN phosphorylation and that GSK3beta may provide feedback regulation of PTEN.

The phosphatase and tensin homologue (PTEN) tumor suppressor is a phosphatidylinositol D3-phosphatase that counteracts the effects of phosphatidylinositol 3-kinase and negatively regulates cell growth and survival. PTEN is itself regulated by phosphorylation on multiple serine and threonine residues in its C terminus. Previous work has implicated casein kinase 2 (CK2) as the kinase responsible for this phosphorylation. Here we showed that CK2 does not phosphorylate all sites in PTEN and that glycogen synthase kinase 3␤ (GSK3␤) also participates in PTEN phosphorylation. Although CK2 mainly phosphorylated PTEN at Ser-370 and Ser-385, GSK3␤ phosphorylated Ser-362 and Thr-366. More importantly, prior phosphorylation of PTEN at Ser-370 by CK2 strongly increased its phosphorylation at Thr-366 by GSK3␤, suggesting that the two may synergize. Using RNA interference, we showed that GSK3 phosphorylates PTEN in intact cells. Finally, PTEN phosphorylation was affected by insulin-like growth factor in intact cells. We concluded that multiple kinases, including CK2 and GSK3␤, participate in PTEN phosphorylation and that GSK3␤ may provide feedback regulation of PTEN.
Phosphatase and tensin homologue (PTEN) 2 (1-3) is a tumor suppressor that is frequently mutated in human cancers (4 -8). The 55-kDa PTEN protein was originally described as a dual-specificity protein phosphatase, but biochemical studies soon showed that PTEN was a poor protein phosphatase but an efficient phosphoinositide D3-phosphatase (9). In cells, PTEN acts as a tumor suppressor by antagonizing phosphoinositide 3-kinase (PI3K), which activates the Akt Ser/Thr kinase, which in turn activates proliferative and antiapoptotic signaling pathways (10 -14).
Posttranslationally, PTEN is regulated through phosphorylation of a cluster of serine and threonine residues in its C terminus (15)(16)(17)(18)(19)(20)(21). Although not required for the activity of the catalytic domain, phosphorylation of the C-terminal region plays an important role in stabilizing the PTEN protein. In its phosphorylated form, the tail is thought to wrap unto the C2 and catalytic domains of PTEN and thereby block the translocation of PTEN to the cytoplasmic face of the plasma membrane (16,22), thus effectively inhibiting the dephosphorylation of the substrates of PTEN. Tail mutants of PTEN tend to have increased catalytic activity but are rapidly degraded in cells.
Using a mutagenesis approach, Torres and Pulido (17) showed that the C-terminal region of PTEN is constitutively phosphorylated in vivo between residues 369 and 386, mostly on Ser-370 and Ser-385. They also found that the Ser/Thr protein kinase casein kinase 2 (CK2) can phosphorylate these residues in vitro as well as, to a lower extent, Ser-380, Thr-382, and Thr-383 (17). Vazquez et al. (16) also found that all phosphorylation of PTEN occurred in the C-terminal region (residues 354 -403) and identified Ser-370 plus at least one other residue (Ser-380, Thr-382, Thr-383, or Ser-385) as the sites in vivo. They also reported that mutation of Ser-380, Thr-382, or Thr-383 to alanine reduced the half-life and increased the catalytic activity of PTEN (16). Finally, Miller et al. (21) identified Ser-370 and Ser-385 as the major phosphorylation sites in vivo and also detected phosphate on Thr-366. They also found that CK2 readily phosphorylated Ser-370 and Ser-385 in vitro.
Our own group (20) studied PTEN in a different cell type, the T lymphocyte, and we found that PTEN is heavily phosphorylated at Ser-380 and Ser-385 in these cells and that both residues can affect the half-life of PTEN. Other residues were not examined. We have now refined this study using mass spectrometry, phosphospecific antibodies, tryptic peptide mapping, phosphoamino acid analysis, site-directed mutagenesis, and RNA interference of kinases. These studies revealed a more complex regulation of PTEN by several kinases, which may act in concert or in response to different conditions or in different cell types. The participation of glycogen synthase kinase 3␤ (GSK3␤) in PTEN phosphorylation in vivo suggested the possibility of PTEN regulation in a negative feedback loop or by stimuli that activate the PI3K/Akt pathway.

EXPERIMENTAL PROCEDURES
Antibodies-Antibodies to PTEN were from Santa Cruz Biotechnology (Santa Cruz, CA) and Upstate Biotechnology (Lake Placid, NY), anti-CK2␣Ј and anti-GSK3␣ were from Santa Cruz Biotechnology (Santa Cruz, CA), anti-PTEN-phospho-Ser-380, anti-phospho-Thr-Pro, anti-Akt, anti-Akt-phospho-S473, and anti-GSK3␣/␤-phospho-21/9 were from Cell Signaling Technology (Beverly, MA), anti-actin was from Sigma, and anti-CK2␣ was from StressGen Biotechnologies Corp. Plasmids and Proteins-Constructs encoding glutathione S-transferase (GST)-fused PTEN and its phosphorylation site mutants were cloned by standard PCR and recombinant DNA methods. Briefly, frag- ments containing the entire open reading frame of PTEN were subcloned into the pGEX-2T vector (Amersham Biosciences), in-frame with the (GST) polypeptide using EcoRI and XhoI restriction sites flanking the 5Ј and 3Ј ends, respectively. The phosphorylation site mutants of PTEN (S362A/T366A, S370A, S380A, S385A, S380A/385A) were generated using the QuikChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA). All constructs were verified by sequencing. Proteins were expressed and purified according to standard techniques.
For PTEN immunoprecipitation, the clarified lysates were incubated with protein G-Sepharose beads and the PTEN antibody from Santa Cruz Biotechnology for 3 h. Immune complexes were washed three times in lysis buffer and suspended in SDS sample buffer. Proteins were resolved in Novex R 4 -20% Tris-glycine polyacrylamide gels, transferred to nitrocellulose, and incubated with antibodies. The blots were developed by the enhanced chemiluminescence technique (ECL kit, Amersham Biosciences) according to the manufacturer's instructions.
Tryptic peptide mapping was performed as before (23)(24)(25) with the protocol of Luo et al. (26). Phosphoamino acid analysis was performed by complete acid hydrolysis in 1 M HCl, 110°C, and separation in two dimensions in the presence of unlabeled standards.
GST Pull-down Assays-For GST pull-down assays, 293T cell lysates were incubated with GST or GST-PTEN prebound to glutathione-Sepharose beads. The beads were then washed three times with lysis buffer, suspended in SDS sample buffer, and analyzed by immunoblotting.
Mass Spectrometry-Identification of phosphorylated residues by LC-MS/MS was performed on in vitro phosphorylated GST-PTEN and digested with highest grade trypsin. The digest was injected into a highpressure liquid chromatography instrument (LC Packings Inc.), which first separates the peptides on a reverse-phase column from which they elute directly into a quadrupole time-of-flight mass spectrometer (Q-TOF API-US) equipped with a nanoelectrospray ionization source (Waters-Micromass).

Loss of CK2 Has No Impact on Phosphorylation of PTEN at Ser-380 in
Cells-Using a phospho-specific antibody, we recently showed that PTEN is phosphorylated at Ser-380 in Jurkat T cells and in normal human T lymphocytes (20). To determine whether CK2 is responsible for this phosphorylation in cells, we used RNA interference to reduce the cellular levels of the two catalytic subunits of this kinase, CK2␣ and CK2␣Ј. 293T cells (which can be transfected to Ͼ80%) were first transfected with small interfering RNAs (siRNAs) for each kinase separately to verify efficacy and to establish the required concentrations (not shown) and then with both siRNAs together (Fig. 1a). Two days after transfection, these cells showed an ϳ90% reduction in CK2␣ and ϳ70% reduction in CK2␣Ј, whereas actin levels were unchanged. Cell viability and morphology also remained normal. Immunoblotting with the antiphospho-Ser-380 antibody revealed that phosphorylation of PTEN at this site was not affected at all by this dramatic reduction in CK2 levels.
CK2 Phosphorylates PTEN at Ser-370 and Ser-385 but Not Ser-380-The lack of effects of the loss of CK2 on PTEN phosphorylation at Ser-380 must mean either that very low levels of CK2 are enough to carry out a normal phosphorylation at this site or that another kinase is responsible. To first address the former possibility, we phosphorylated PTEN in vitro with CK2 and analyzed the phosphorylation by autoradiography, tryptic peptide mapping, phosphoamino acid analysis, and tandem mass spectrometry. These experiments showed that CK2 readily phosphorylated PTEN (Fig. 1b, lane 2), whereas recombinant Akt did not (Fig. 1b, lane 3), and that this phosphorylation occurred on two distinct peptides (Fig. 1c), both of which contained only phosphoserine (Fig. 1d). The two peptides migrated a very short distance on the thin layer plates, particularly in the second dimension (ascending chromatography), indicating that they are both acidic and very hydrophilic. These properties are found only in the two long C-terminal tryptic peptides corresponding to residues 350 -378 and 379 -402 of PTEN, the latter containing the major phosphorylation sites in PTEN in T cells, Ser-380 and Ser-385 (20). Using PTEN proteins with either or both of these two residues mutated to alanine as substrates showed that CK2 readily phosphorylated all these proteins with some decrease observed only for PTEN proteins with Ser-385 mutated (Fig. 1e). Finally, tandem mass spectrometry (Fig. 1f) detected phosphorylation of two distinct tryptic peptides at residues Ser-370 and Ser-385, respectively (Fig. 1f). In contrast, no peptide containing phosphate at Ser-380 was found. Since the same peptide with phosphate at Ser-385 was readily detected, it seemed that Ser-380 was not phosphorylated at all by CK2.
Together, all these results indicated that CK2 indeed can phosphorylate PTEN very well but that it is not required for phosphorylation of PTEN at Ser-380 in intact cells. Indeed, CK2 did not phosphorylate this between the m/z 1045.44 (y8) and 1114.47 (y9 Ϫ98) ions corresponds to a dehydroalanine residue resulting from the loss of H 3 PO 4 from phosphoserine, therefore establishing the ninth residue (Ser-370) from the C terminus as phosphoserine. This assignment was further confirmed by the observation of the relatively abundant (y10 Ϫ98) and (y12 Ϫ98) ions and the relatively less abundant y9, y10, and y12 ions that contain intact phosphoserine residues. Upper right panel and lower right panel, LC-MS/MS spectra of the m/z 938.39 3ϩ (upper right panel) and 965.05 3ϩ (lower right panel) ions. The 87-Da mass difference between the m/z 2042.8 (y17) and 2129.9 (y18) in the upper spectrum indicated the serine residue (Ser-385) between two aspartic acid residues was unmodified, whereas the 69-Da mass difference between the m/z 2042.8 (y17) and 2112.0 (y18 Ϫ98) in the lower spectrum identifies the corresponding serine residue as phosphorylated. This assignment was supported by the presence of two relatively abundant y19 and y19 Ϫ98 ions in the upper and lower spectra, respectively. residue even in vitro but instead phosphorylates serine residues 370 and 385. We also found that CK2 phosphorylates only serine residues, without any trace of phosphate on threonine residues. We concluded that it is likely that additional kinases participate in the phosphorylation of PTEN at serine and threonine residues in intact cells.
Phosphorylation of PTEN by GSK3␤ and CK1 but Not by c-Akt-Next, we tested a number of other Ser/Thr kinases for their ability to phosphorylate PTEN. As shown in Fig. 1b, recombinant Akt (which readily phosphorylated other proteins in vitro; data not shown) was unable to phosphorylate GST-PTEN. Similarly, cAMP-dependent protein kinase, protein kinase C, and the protein kinases Mek1, Erk2, and PDK1 were all unable to phosphorylate GST-PTEN (not shown). In contrast, GSK3␤ and CK1 incorporated substantial amounts of 32 P into PTEN (Fig. 2a), albeit less than CK2. Tryptic peptide maps of PTEN phosphorylated by GSK3␤ showed a single spot (Fig. 2b), which was both acidic and hydrophilic, suggesting that it corresponds to either of the two C-terminal tryptic peptides of PTEN. Phosphoamino acid analysis revealed that GSK3␤ phosphorylated PTEN on both threonine and serine (Fig. 2c). Tandem mass spectrometry identified the phosphorylated residues as Ser-362 and Thr-366 (Fig. 2d). None of the several other serine or threonine residues contained any phosphate. Indeed, the PTEN-S362A/T366A mutant of PTEN was not phosphorylated at all by GSK3␤ in vitro (Fig. 2e).
The phosphorylation of PTEN by CK1, CK2, or GSK3␤ in vitro was further analyzed with phospho-specific antibodies against PTEN-phospho-Ser-370 and PTEN-phospho-Ser-385 (Fig. 3a, first  and second panels), which confirmed that CK2 readily phosphorylated both sites (lanes 3), whereas CK1 phosphorylated Ser-370 much less than CK2 but Ser-385 a bit better than CK2 (lanes 2). In contrast, GSK3␤ did not phosphorylate either site (Fig. 3a, first and second  panels, lane 4). Phospho-specific antibodies against Ser-362 or Thr-366 are not available, but since Thr-366 is followed by a proline residue, we tested whether this site would react with an anti-phospho-Thr-Pro antibody. Indeed, GSK3 treatment made PTEN readily reactive with this antibody (Fig. 3a, third panel, lane 4). This reactivity was abrogated by mutation of Thr-366 (see below). CK1 was very inefficient in phosphorylating PTEN at Thr-366, whereas CK2 was not able to cause any phosphorylation at this site.
Synergistic Phosphorylation of PTEN by GSK3␤ and CK2-GSK3␤ typically phosphorylates serines or threonines that are located four residues amino-terminal of an already phosphorylated residue, a pattern that fits the notion that Ser-362 becomes a good substrate once Thr-366 has first been phosphorylated. Furthermore, phosphorylation of Ser-370 by CK2 could have the same stimulatory effect on phosphorylation of Thr-366 by GSK3␤. To directly test this possibility, we pretreated GST-PTEN with CK2, or kinase buffer alone, in the presence of unlabeled ATP, pulled down and washed the GST-PTEN protein, and then treated it with GSK3␤ in kinase buffer with [␥-32 P]ATP or with kinase buffer and [␥-32 P]ATP alone. The latter also served as a control to show that residual CK2 did not affect the assay. As shown in Fig. 3b, upper panel, PTEN pretreated with CK2 was severalfold more phosphorylated by GSK3␤ (lane 4) than PTEN pretreated with kinase buffer and ATP without CK2 (lane 2). When this experiment was repeated with GST-PTEN-S370A, the effect of CK2 pretreatment was completely lost (Fig.  3b, lower panel).
Very similar results were obtained with phospho-specific antibodies; although only GSK3b was able to make PTEN reactive with the phospho-Thr-Pro antibody (i.e. phosphorylate Thr-366) (Fig. 3c, lane 4), PTEN treated with both CK2 and GSK3b became much more strongly reactive (lane 5). In contrast, PTEN-T366A/S362A did not react at all with the antibody (lane 6), and the effect of CK2 was completely lost in the PTEN-S370A mutant. Together, these experiments showed that phosphorylation of PTEN by CK2 at Ser-370 efficiently promotes phosphorylation at Thr-366 by GSK3␤.
GSK3␤ Phosphorylates PTEN at Thr-366 in Intact Cells-To determine whether GSK3␤ participates in the phosphorylation of PTEN in intact cells, we reduced the cellular levels of GSK3␤, as well as the closely related GSK3␣, by RNA interference, immunoprecipitated PTEN, and probed it with the anti-phosphothreonine-proline antibody (Fig. 4a). In cells with reduced levels of both GSK3 isoforms, phosphorylation of PTEN at Thr-366 was much reduced, indicating that GSK3 is needed for this phosphorylation in intact cells. In contrast, reactivity with the phospho-Ser-370 and phospho-Ser-385 specific antibodies was not changed, and PTEN and actin levels were unaltered. We concluded that GSK3 indeed phosphorylates at least Thr-366 in intact cells. This notion was further supported by the detection of a small amount of GSK3␤ (as well as CK2) bound to GST-PTEN incubated with cell lysates followed by extensive washing and immunoblotting (data not shown).
Inactivation of GSK3␣/␤ by Insulin-like Growth Factor 1 Stimulation Reduces PTEN Phosphorylation at Thr-366 in Cells-Next, we wanted to learn whether extracellular stimuli that affect the activity of GSK3␣ and ␤ would alter the phosphorylation of PTEN by this kinase in intact cells. Insulin-like growth factor 1 is known to cause a robust activation of c-Akt, which, in turn, phosphorylates GSK3 and inactivates it. Indeed, the addition of this growth factor to our cells caused a sharp increase in phospho-Akt levels (Fig. 5a, first panel) and the appearance of GSK3␣/␤ phosphorylated at Ser-21/Ser-9 (Fig. 5a, third panel). Concomitantly, there was a clear decrease in the reactivity of PTEN with the phospho-Thr antibody. The total levels of Akt and PTEN remained unchanged in these experiments, whereas the levels of GSK3 tended to change, at least in part due to nucleus-to-cytosol translocation of this kinase (27). Nevertheless, these data demonstrated that extracellular stim- uli that activate the PI3K-Akt-GSK3 pathway directly influence PTEN phosphorylation at Thr-366, providing a physiological regulation of this phosphorylation of PTEN.
Phosphorylation of PTEN at Thr-366 Reduces the Activity of PTEN in Cells-Finally, to evaluate whether phosphorylation at Thr-366 has any functional impact on PTEN, we expressed PTEN or the PTEN-T366A mutant in Jurkat T cells, which lack endogenous PTEN, and measured their effects on the phosphorylation of Akt. In these experiments, PTEN-T366A consistently reduced Akt phosphorylation to a higher extent than unmutated PTEN (Fig. 5b). Thus, it appeared that phosphorylation of PTEN at Thr-366 by GSK3 reduces its biological activity.

DISCUSSION
Although our data agreed with the conclusion from previous studies that CK2 plays a role in PTEN phosphorylation (17), we found that CK2 is not the only kinase involved in this important task. Like several other groups (16,17,21), we found that Ser-370 and Ser-385 indeed are phosphorylated in vivo and by CK2 in vitro. However, we saw little phosphorylation of other residues by CK2, and notably, no phosphorylation of threonine residues at all. We also found that CK2␣ and ␣Ј are not required in intact cells for phosphorylation of Ser-380, although overall phosphorylation of PTEN was decreased by CK2 knock-down by RNA interference (data not shown). Thus, CK2 could not be the only kinase . The cells were starved from serum for 1 h prior to lysis. Note that the PTEN mutant reduces Akt phosphorylation more than wild-type PTEN does despite similar levels of expression. c, schematic model of the proposed negative feedback loop. Activation of the PI3K-Akt pathway by insulin-like growth factor leads to phosphorylation and inhibition of GSK3␤, which therefore reduces its phosphorylation of PTEN, leading to increased activity of PTEN to counteract PI3K. involved in PTEN phosphorylation in cells. Indeed, we found that GSK3␤ also phosphorylates PTEN at two sites, Ser-362 and Thr-366, the latter detected by Miller et al. (21) as a site in intact cells. We also detected phosphate at Thr-366 in vivo, and this phosphate decreased upon knock-down of GSK3 or the addition of insulin-like growth factor 1 to the cells. Importantly, GSK3␤ and CK2 phosphorylated non-overlapping sites, and we found that phosphorylation of Ser-370 by CK2 strongly enhances subsequent phosphorylation of Thr-366 by GSK3␤. Thus, these two kinases presumably synergized and could potentially cause rapid changes in the phosphorylation state of PTEN in cells. It also appeared that CK1 can phosphorylate PTEN, particularly at Ser-385, suggesting that this kinase may also play a role in intact cells. Fig. 4b shows a schematic view of the kinases involved in PTEN tail phosphorylation.
We previously reported that PTEN phosphorylation is influenced by D3 phosphorylated inositol lipids (the substrates for PTEN) in what appears to be a negative feedback loop (20). A dilemma in this model was that CK2 is not known to be regulated by these phospholipids directly or indirectly. The introduction of GSK3␤ solves this dilemma (Fig. 5c). This kinase is known to be inhibited by phosphorylation by Akt, suggesting that high levels of D3-phosphoinositides (e.g. in Jurkat T cells) may reduce the phosphorylation at Thr-366 (and perhaps Ser-362). Indeed, catalytically inactive PTEN-C124G expressed in Jurkat T cells was highly phosphorylated on serine but contained nearly undetectable phosphothreonine (20).
Okumura et al. (28) recently showed that the C terminus of PTEN physically interacts with the oncogenic MSP58 protein. This interaction involved Thr-366 of PTEN and was abrogated by a T366A mutation, suggesting that phosphorylation at Thr-366 was necessary for the interaction. By interacting with MSP58, PTEN suppressed MSP58-driven cellular transformation in a manner that did not require the lipid phosphatase activity of PTEN (28). Since we found that GSK3 is responsible for phosphorylation of PTEN at Thr-366, it appeared that GSK3 plays an unexpected role in promoting the tumor suppressor function of PTEN by inducing binding of MSP58.
At present, it remains unclear which kinase phosphorylates Ser-380. Neither CK2 nor GSK3␤ phosphorylated this site in vitro, and reactivity with the phospho-Ser-380 antibody was not affected by knock-down of either kinase. Thus, it appeared that a third kinase is responsible for PTEN phosphorylation at Ser-380. A possibility is CK1, which phosphorylated PTEN in vitro, in contrast to a number of other Ser/Thr kinases, including Akt, PDK1, cAMP-dependent protein kinase, protein kinase C, and mitogen-activated protein (MAP) kinases. Another possibility is a group of Ser/Thr kinases known as the MAST kinases, which contain a PDZ domain and bind to the C-terminal tail of PTEN (29).
Finally, it should be pointed out that PTEN is also regulated both transcriptionally and translationally, and both these mechanisms can contribute to rapid changes in PTEN protein levels and to dysregulation of PTEN in cancer. Understanding how PTEN is regulated should pro-vide new insight into cell signaling mechanisms and may suggest novel approaches to the treatment of PTEN-deficient tumors.