PCAF Modulates PTEN Activity*♦

The PTEN protein has a single catalytic domain possessing both lipid phosphoinositol and protein phosphatase activities. The lipid phosphoinositol phosphatase activity is essential for PTEN to block the cell cycle in the G1 phase and thereby to suppress tumor formation and progression (Cantley, L. C., and Neel, B. G. (1999) Proc. Natl. Acad. Sci. U. S. A. 96, 4240-4245), although the mechanisms governing PTEN activity under normal and neoplastic growth conditions remain unclear. Here, we report that PTEN interacts physically and functionally with PCAF, a histone acetyltransferase that regulates gene transcription through interaction with p300/CBP and various sequencespecific transcription factors (Nakatani, Y. (2001) Genes Cells 6, 79-86). Expression of PCAF results in increased acetylation of lysine residues (Lys125 and Lys128) within the catalytic cleft of PTEN, a structure essential for phosphatidylinositol 3,4,5-trisphosphate specificity (Lee, J. O., Yang, H., Georgescu, M. M., Di Cristofano, A., Maehama, T., Shi, Y., Dixon, J. E., Pandolfi, P., and Pavletich, N. P. (1999) Cell 99, 323-334). The acetylation of PTEN caused by PCAF expression depends on the presence of growth factors. Reduction of endogenous PCAF activity using shRNA results in a loss of PTEN acetylation in response to growth factors and restores the ability of PTEN to down-regulate phosphatidylinositol 3-kinase signaling and to induce G1 cell cycle arrest. The retention of phosphatidylinositol 3-kinase/AKT signaling and cell cycle regulatory activities of acetylationresistant PTEN K125R and K128R mutants in the presence of enforced PCAF expression suggest a causal relationship. Together, these findings indicate a mechanism of PTEN regulation that forges a link between distinct cancer-relevant pathways central to the control of growth factor signaling and gene expression.

The PTEN (phosphatase and tensin homolog deleted on chromosome 10) tumor suppressor gene is frequently mutated in diverse tumor types including those of endometrium, breast, prostate, lung, and brain (4). Germ line mutations of PTEN cause Cowden disease, an autosomal dominant hamartoma syndrome with an increased risk of developing tumors in a variety of tissues (4,5). PTEN has been shown to play a crucial role in the regulation of cell migration, growth, and apoptosis (6 -8). These functions are mediated by its lipid phosphatase activity, which is specific for the 3Ј position of PI(3,4,5)P 3 2 and phosphatidylinositol 3,4-bisphosphate (9), leading to inhibition of PI3kinase signaling and subsequent inactivation of the Akt pathway. PTEN has also been shown to dephosphorylate tyrosine residues on FAK and Shc proteins (10).
The regulation of PTEN in normal and pathophysiological settings is an area of active investigation. Structurally, the PTEN protein consists of an amino-terminal phosphatase domain and a carboxyl-terminal domain, which is subdivided into C2, phosphorylation, and PDZ binding domains. The C2 domain lends itself to regulation of PTEN function by contributing to membrane localization, and the phosphorylation domain controls protein stability as governed by casein kinase II-directed phosphorylation (11)(12)(13). Given the constitutive nature of casein kinase II activity, other more tightly controlled mechanisms of regulation of PTEN function would be predicted. To address this issue, we searched for PTEN-interacting proteins by analyzing cellular protein complexes containing PTEN using tandem affinity purification coupled with mass spectrometry. One candidate was the histone acetyltransferase, PCAF (p300/CBP-associated factor), which associated with PTEN at endogenous levels in cells and whose expression caused acetylation of PTEN, inhibition of PTEN regulation of PI3K signaling, and inhibition of PTEN-regulated cell cycle arrest.
recommended procedure. Pten Ϫ/Ϫ MEFs were infected with PTEN-and PCAF-coding retroviruses produced in 293T cells. Various PTEN mutants were constructed in enhanced GFP (EGFP) and MSCV vectors using PCR-based mutagenesis with Pfu Turbo DNA polymerase. Mammalian expression vectors for wild-type and mutant PCAF were described previously (14,15).
Immunoblotting and Coimmunoprecipitation-293T cells were transfected with the appropriate combination of plasmids using Lipofectamine 2000 or treated with 50 ng/ml of EGF or 10% serum for the indicated time points, then harvested and extracted in lysis buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, and 0.5% Nonidet P-40. Lysates were cleared by centrifugation at 15,000 ϫ g for 20 min at 4°C and analyzed by SDS-PAGE. For immunoprecipitation, lysates were incubated with anti-FLAG or anti-GFP antibody-conjugated beads (Sigma and Covance, respectively) for 4 h at 4°C. The immunoprecipitates were washed four times with lysis buffer and then subjected to immunoblotting.
Microscopy for Localization-Images were acquired on a DeltaVision deconvolution microscope (Applied Precision) equipped with a Coolsnap CCD camera (Roper Scientific). 293T cells seeded on coverslips were transfected with the appropriate combination of plasmids and subsequently fixed with 0.4% paraformaldehyde in phosphate-buffered saline followed by microscopic analysis.
Suppression of PCAF Expression by shRNA-Three shRNA plasmids for PCAF-directed knockdown were constructed in the pSuper vector and contained the following sequences: FIGURE 1. PTEN interacts with PCAF. a, 293T cell lysates were precipitated with anti-PTEN or control IgG antibody followed by immunoblotting for coprecipitated PCAF. b, similarly, lysates were precipitated with anti-PCAF followed by immunoblotting for coprecipitated PTEN. Efficiency of immunoprecipitation was assessed by re-probing each blot with antibodies used for precipitation. c, schematic of GFP-PTEN fusion proteins constructed for PCAF interaction domain mapping. d, left panel, 293T cells were transfected with GFP-PTEN and FLAG-PCAF expression constructs, followed by anti-FLAG immunoprecipitation and detection of GFP-PTEN associated proteins by anti-GFP immunoblotting. Efficiency of precipitation was assessed by reprobing with anti-FLAG. Right panel, whole cell lysates were immunblotted with anti-FLAG and anti-GFP to confirm exogenous protein expression. e, lysates from 293T cells transfected with empty vector or FLAG-PCAF and GFP-PTEN constructs were precipitated with anti-FLAG and associated GFP-PTEN proteins were detected by anti-GFP immunoblotting. Efficiency of precipitation was assessed as described for d.IP, immunoprecipitation; WB, Western blotting. CAGACTCCTGGGAATAGTA, ACCTGTGGTTGAAGGC-TCT, and TTGCGCTTTCTTCACGGCG. These constructs were mixed and transfected into 293T cells with Lipofectamine 2000.
Cell Cycle Analysis-Pten Ϫ/Ϫ MEFs were infected with Ecotropic virus containing MSCV-hyg-PTEN and pBabe-PCAF and selected with hygromycin and puromycin, respectively. Cells were washed once with phosphate-buffered saline, fixed in 70% ethanol, stained with propidium iodide (25 g/m1) (Sigma), and incubated for 30 min at 37°C with RNase A (20 g/ml). The DNA content of the cells was then evaluated by flow cytometry (FACSort, Becton Dickinson). Linear red fluorescence (FL2) was analyzed.

RESULTS
PTEN and PCAF Interact in Vivo-To identify PTEN interacting proteins, tandem affinity purification coupled with mass spectrometry was used to isolate and analyze PTEN-containing complexes. FLAG-hemagglutinin-NH 2 -terminal-tagged PTEN was expressed in 293T cells and purified by successive FLAG and hemagglutinin immunoprecipitations. Subsequently PTEN-associated proteins were analyzed by mass spectrometry, resulting in the identification of the histone acetyltransferase, PCAF, as a member of the complexes. The validity of the candidacy of PCAF was also suggested by a yeast two-hybrid screen of human and mouse cDNA libraries with the carboxyl-terminal region of PTEN (amino acids 186 -401) that encompasses the regulatory phosphorylation domain. One candidate cDNA recovered, KIAA1343, exhibits structural similarity to the transcriptional adapter protein, hADA2, a molecule associated with PCAF in vivo (17,18). Since both of these approaches pointed toward PCAF, we sought to determine whether PTEN and PCAF interacted in vivo in 293T cells, which express wild-type PTEN and PCAF, using reciprocal coimmunoprecipitation. These experiments showed that anti-PTEN immunoprecipitates contained PCAF and, conversely, extracts exposed to anti-PCAF antibody also coprecipitated PTEN (Fig. 1, a and b), demonstrating that PTEN and PCAF interact physically at endogenous levels.
To define the PCAF interaction domain in PTEN, various PTEN deletion constructs (Fig. 1c) were coexpressed together with FLAGtagged PCAF (FLAG-PCAF) in 293T cells. FLAG-PCAF interacted with full-length, N202 (amino acids cell lysates were precipitated with anti-PTEN or control IgG antibody followed by immunoblotting for acetylated PTEN with anti-acetyl-lysine antibody. b, lysates treated with the histone deacetylase inhibitor, trichostatin A (TSA), were immunoprecipitated with anti-acetyl-lysine antibody or mouse IgG followed by anti-PTEN immunoblotting. c, lysates of 293T cells cotransfected with GFP-PTEN or GFP-PTEN186C and wild-type or HAT domain mutant PCAF constructs (⌬579 -608 and ⌬609 -624) were precipitated with anti-GFP, and acetylated GFP-PTEN was detected by anti-acetyl-lysine immunoblotting. The filter was reprobed with anti-GFP to determine immunoprecipitation efficiency. 1-202), and 186C (amino acids 186 -403) GFP-tagged PTEN proteins but not with 331C (amino acids 331-403) (Fig. 1d). Deletion of the region shared by N202 and 186C (⌬186 -202) generated a protein that failed to interact with PCAF (Fig. 1e) and lacks lipid phosphatase activity (data not shown). PTEN is predominantly localized to the cytoplasm in most situations (7) but is capable of moving into the nucleus under some conditions (19). To determine whether PTEN and PCAF occupy cellular compartments within which they could physiologically interact, we performed deconvolution microscopy of PTEN and PCAF that were differentially labeled (Fig. 2). As expected, PCAF localized to the nucleus in the presence of either wildtype PTEN or PTEN⌬186 -202. In the presence of PCAF, wildtype PTEN also was predominantly nuclear, while PTEN⌬186 -202, which does not interact with PCAF, was predominantly cytoplasmic. Thus, PCAF binding may provide a means of moving PTEN from the cytoplasm to the nucleus. We conclude that PTEN interacts with the PCAF complex in vivo through the PTEN hinge region, a structure that separates the amino-terminal catalytic domain from the carboxyl-terminal regulatory domain (3).
PTEN Acetylation Increases in Response to PCAF-We next determined whether the expression of PCAF results in the acetylation of PTEN. First, endogenous PTEN was immunoprecipitated and then immunoblotted with an anti-acetyl-lysine antibody, revealing that PTEN is acetylated at steady state (Fig.  3a). Correspondingly, immunobloting detected PTEN in antiacetyl-lysine immunoprecipitates (Fig. 3b). Furthermore, the acetylation level of endogenous PTEN increased upon exposure of cells to tricostatin A, an inhibitor of Class 1 and 2 HDACs (20) (Fig. 3b), leading us to conclude that the acetylation of PTEN in vivo is a dynamic process. We then tested whether PCAF expression affects the level of PTEN acetylation. Enforced PCAF expression resulted in enhanced acetylation of wild-type PTEN (Fig. 3c) but not the PTEN 186C truncation mutant, a finding consistent with the possibility that one or more of the lysine residues from positions 1-186 were acetylated in response to PCAF expression.
Acetylation of PTEN Lysines 125 and 128 in Response to PCAF Regulates Enzymatic Activity-An interspecies comparison of PTEN amino acid sequences shows that lysine residues 125 and 128 are evolutionarily conserved (with the exception of the substitution of arginine for lysine at the first position in Saccharomyces cerevisiae) (Fig. 4a). To determine whether these lysine residues were the targets of acetylation caused by PCAF expression, we constructed mutant PTEN alleles in which non-acetylatable arginines were substituted for lysines 125 and 128 (K125R, K128R, and K125/ 8R). A K80R allele was also generated since mutation at this position has been found in a human glioblastoma tumor (5) and a K144R allele, which is not evolutionally conserved, was constructed to serve as a potential negative control. When coexpressed with PCAF, the K125R and K128R alleles of PTEN exhibited reduced acetylation, as compared with wildtype or other arginine substitution mutants of PTEN (Fig.  4b). That these modifications can be directly mediated by PCAF was demonstrated by showing that PCAF directly acetylated PTEN, but not a K125/8Q mutant, in an in vitro acetylation assay (Fig. 4c). These data indicate that lysines 125 and 128 of PTEN are preferentially acetylated by PCAF; it is also possible that other lysine residues are sequentially modified by PCAF following the initial acetylation of Lys 125 and Lys 128 .
PCAF has an established role in regulating gene transcription by acetylating histones and various transcription factors, including p53 (2). Of relevance to this report is the observation that PCAF-related histone acetyltransferases, such as p300/ CBP, have also been shown to modify and regulate the function of cytosolic proteins such as importin-␣7 (21). These observations led us to conduct a series of experiments designed to assess the biochemical and functional impact of acetylation on PTEN activity. Since PCAF caused the acetylation of two lysine residues within the catalytic site of PTEN, we postulated that PCAF might alter the lipid phosphatase activity of PTEN that is responsible for regulating the activation of Akt. Consistent with this idea, an in vitro lipid phosphatase assay showed that PCAF inhibited 45% of PTEN activity but only 19% of acetylationresistant K125R activity (Fig. 4d). FIGURE 5. The ability of PTEN to down-regulate Akt activation is inhibited by acetylation in response to PCAF. a, lysates of 293T cells, transfected with PTEN and PCAF plasmid combinations and cultured under 2% serum condition, were immunoprecipitated with anti-Akt antibody followed by p-Akt and total Akt immunoblotting. b, 293T cells were transfected with GFP-tagged wild-type PTEN, and PTEN lysine 125 and 128 arginine (Arg) or glutamine (Gln) mutants (to mimic acetylation) and cultured under 2% serum. Lysates were immunoprecipitated with anti-Akt followed by p-Akt(Ser 473 ) immunoblotting. The efficiency of immunoprecipitation was assessed by reprobing with anti-Akt. c, PTEN null MEFs were infected with retroviruses expressing PTEN and various PTEN mutants and p-Akt and total Akt were detected by direct immunoblotting of whole cell lysates. SEPTEMBER 8, 2006 • VOLUME 281 • NUMBER 36

JOURNAL OF BIOLOGICAL CHEMISTRY 26565
Akt Activation Is Regulated by PTEN Acetylation in Response to PCAF-When PCAF was coexpressed with wild-type PTEN in 293T cells, we observed a marked increase in Akt phosphorylation (Fig. 5a). This allowed an assessment of the ability of various PTEN mutants to modulate the phosphorylation levels of endogenous Akt in cells grown under conditions of growth factor exposure (Fig. 5, b and c). Consistent with results described above, expression of wild-type PTEN and the K125R and K128R mutants led to a decrease in the level of phosphorylated Akt in 293T cells and in Pten Ϫ/Ϫ MEFs, while other PTEN mutants were compromised for this activity (Fig. 5, b and  c). Neither the K125/128Q mutant, which mimics acetylated PTEN, nor PCAF alone were able to decrease phospho-Akt levels (Fig. 5, b and c). Taken together, these results indicate that PCAF-mediated acetylation of PTEN inhibits its activity in the regulation of the PI3K/AKT pathway.
The Ability of PTEN to Regulate G 1 Arrest Is Dependent on PCAF-PTEN inhibition of Akt activation has been reported to elicit cell cycle arrest (8) and to modulate cellular responses to growth factor stimulation (9). This led us to test whether PCAF could inhibit these PTEN-mediated activities. First, wild-type PTEN or the K125R and K128R mutants were expressed in Pten Ϫ/Ϫ MEFs in the absence of PCAF (Fig. 6). Each resulted in similar G 1 cell cycle arrests (78.0, 73.9, and 70.0% G 1 , respectively, compared with 56.2% G 1 for empty vector-infected cells). Alleviation of G 1 arrest occurred when PCAF was coexpressed with wild-type PTEN (54.3% G 1 ). This relief of the arrest was not observed when PCAF was coexpressed with the acetylation-resistant, lipid phosphatase-competent PTEN mutants, K125R and K128R ( Fig. 6; 73.5 and 73.9% G 1 , respectively). The double K125/128Q mutant, was unable to cause a G 1 arrest (54.9% and 59.4% G 1 , respectively).
Growth Factor-driven Cellular Akt Activation Is Modulated by the Presence of PTEN and PCAF-We then analyzed PTEN acetylation status when 293T cells were treated with serum or EGF (Fig. 7a). PTEN acetylation and its interaction with PCAF were enhanced with serum and growth factor treatment. Correspondingly, when PCAF is reduced by shRNA knockdown, PTEN is poorly acetylated (Fig. 7b). Moreover, this diminution  of acetylation corresponds to an increased ability of PTEN to down-regulate PI3K signaling, as reflected by a reduction in Akt phosphorylation (Fig. 7c). These data indicate that PTEN function is inhibited when it is acetylated by PCAF and that this promotes G 1 /S transition through activation of the PI3K pathway.

DISCUSSION
The results presented here are consistent with PCAF-mediated acetylation providing regulated inactivation of PTEN catalytic activity. One possible mechanism for this is that modification of lysine residues within the catalytic pocket may interfere with substrate binding. However, Lys 125 and Lys 128 are located in the flexible catalytic domain P-loop structure (3), which suggests that the addition of small acetyl groups would not physically impede interaction with substrate. Alternatively, acetylation of these lysine residues might cause the loss of a positive charge in the catalytic cleft known to be required for selectivity of the negatively charged PI(3,4,5)P 3 substrate, as well as its orientation within the cleft (3).
It is interesting to note that the yeast two hybrid screen identified KIAA1343 as a PTEN-interacting partner. This raises the possibility that it might act as a bridge between PTEN and PCAF and that its modulation might affect that interaction. While this has not been directly tested, KIAA1343 did not appear in the tandem affinity purification/mass spectrometry complexes. Moreover, 293T cells express very little KIAA1343 and they are responsive to manipulations of the PTEN and PCAF interaction. Whether this points the way toward cell type specificity of the complexes, or was the fortunate consequence of the presence in the yeast two hybrid assay of the endogenous PCAF homolog, GCN5, is an area requiring investigation.
Phosphorylation of p300/CBP by cyclin E/cdk2 has been shown to stimulate its HAT activity at the G 1 /S boundary (22). It is possible that PCAF may be similarly activated by cell cycle responses to extracellular cues, such as growth factor signaling, and then it in turn regulates PTEN activity by acetylation. This is consistent with our observations that PCAF knockdown reduced phospho-Akt levels when cells were stimulated with growth factor or serum, and that PCAF decreases the percent of cells in the G 1 phase in PTEN-reconstituted Pten Ϫ/Ϫ MEFs. Recent investigations have shown that nuclear-localized PTEN coincides with G 0 -G 1 phases of the cell cycle, while cytosoliclocalized PTEN is responsible for regulating apoptosis (23)(24)(25). Our results extend these observations and suggest that PCAFmediated acetylation of nuclear PTEN, as initiated by growth stimulation, results in the attenuation of PTEN phosphatase activity and promotion of cell cycle progression.
It has been reported that expression of HATs, including PCAF, correlates with cell transformation as the expression of PCAF is up-regulated in c-Myc-transduced endothelial cells (26) and GCN5L2, which belongs to the PCAF family of HATs, is overexpressed in H-RAS-transformed fibroblasts (27). High levels of p300/CBP protein in clinical biopsies of prostate cancer correlate with large tumor size and forced reduction of p300 expression in prostate cancer cell lines results in reduced proliferation (28). Mutation of CBP and other HATs can occur in various genetic diseases (29 -31) and primary tumors (32), although PCAF has not been found to be mutated during astrocytic tumor progression (33). Another possibility is that mutations causing an increase in PCAF protein expression or activity occur in cancers with less frequently occurring incidences of PTEN mutation. Alternatively, mutation of other histone acetylases that function together with PCAF may contribute to PCAF-dependent pathology. For example, chromosomal translocation of p300 and CBP has been shown to cause hematological malignancies due to defects in cell cycle progression and apoptosis (32).
Thus, regulation of enzymatic activity by acetylation may be a novel and dynamic mechanism for regulating cellular proliferation. Here, we describe an instance in support of this notion by demonstrating inhibition of the lipid phosphatase activity of the PTEN tumor suppressor by PCAF, through specific acetylation of evolutionarily conserved, catalytic domain lysine residues required for selectivity, and engagement of PI(3,4,5)P 3 .