Histone Acetylation-independent Effect of Histone Deacetylase Inhibitors on Akt through the Reshuffling of Protein Phosphatase 1 Complexes*

Despite advances in understanding the role of histone deacetylases (HDACs) in tumorigenesis, the mechanism by which HDAC inhibitors mediate antineoplastic effects remains elusive. Modifications of the histone code alone are not sufficient to account for the antitumor effect of HDAC inhibitors. The present study demonstrates a novel histone acetylation-independent mechanism by which HDAC inhibitors cause Akt dephosphorylation in U87MG glioblastoma and PC-3 prostate cancer cells by disrupting HDAC-protein phosphatase 1 (PP1) complexes. Of four HDAC inhibitors examined, trichostatin A (TSA) and HDAC42 exhibit the highest activity in down-regulating phospho-Akt, followed by suberoylanilide hydroxamic acid, whereas MS-275 shows only a marginal effect at 5 μm. This differential potency parallels the respective activities in inducing tubulin acetylation, a non-histone substrate for HDAC6. Evidence indicates that this Akt dephosphorylation is not mediated through deactivation of upstream kinases or activation of downstream phosphatases. However, the effect of TSA on phospho-Akt can be rescued by PP1 inhibition but not that of protein phosphatase 2A. Immunochemical analyses reveal that TSA blocks specific interactions of PP1 with HDACs 1 and 6, resulting in increased PP1-Akt association. Moreover, we used isozyme-specific small interfering RNAs to confirm the role of HDACs 1 and 6 as key mediators in facilitating Akt dephosphorylation. The selective action of HDAC inhibitors on HDAC-PP1 complexes represents the first example of modulating specific PP1 interactions by small molecule agents. From a clinical perspective, identification of this PP1-facilitated dephosphorylation mechanism underscores the potential use of HDAC inhibitors in lowering the apoptosis threshold for other therapeutic agents through Akt down-regulation.

Histone deacetylase (HDAC) 2 is recognized as one of the promising targets for cancer treatment because many HDAC inhibitors have entered clinical trials in both solid and liquid tumors (1)(2)(3). Nevertheless, the mechanisms underlying the antiproliferative effects of HDAC inhibitors remain elusive. Although they have been shown to activate the transcription of a defined set of genes through chromatin remodeling (4), increasing evidence suggests that modifications of the epigenetic histone code may not represent the primary cause for HDAC inhibitormediated growth inhibition and apoptosis in cancer cells (5,6). To date, at least two distinct histone acetylation-independent mechanisms have been described for the action of HDAC inhibitors on cellular targets. First, because certain HDAC members can mediate the deacetylation of non-histone proteins, their inhibition interferes with signaling processes in which these proteins are involved independently of the activity of HDAC inhibitors in transcriptional activation. For example, HDAC3 regulates NFB signaling in the nucleus by deacetylation the Rel-A subunit (7), and HDAC6 modulates microtubule acetylation and chemotactic cell motility by acting as a tubulin deacetylase (8 -12). Moreover, Class I HDACs function as a STAT3 deacetylase (13), and the mammalian Class III HDACs SirT1 and SirT2 can deacetylate p53 and tubulin, respectively (14,15). Second, HDACs 1 and 6 have been shown to form complexes with protein phosphatase 1 (PP1) (16,17), of which the combined deacetylase/phosphatase activities underlie the ability of HDAC1 to modulate transcriptional activity of cAMP-responsive element-binding protein (16) and that of HDAC6 to regulate microtubule dynamics (17). Also noteworthy is that the binding of HDACs to PP1 was highly specific because no association between HDACs and protein phosphatase 2A (PP2A) was noted (17).
Moreover, data from this and other laboratories (18 -20) show that HDAC inhibitors could facilitate the dephosphorylation of Akt and other signaling kinases, although the causative mechanism remains undefined. In this report, we used TSA to demonstrate that HDAC inhibitors facilitate dephosphorylation of Akt by altering the dynamics of HDAC-PP1 complexes. We provide evidence that HDAC inhibitors selectively target HDACs 1 and 6 to disrupt the respective HDAC-PP1 complexes, resulting in increased association of PP1 with Akt. This PP1-facilitated kinase dephosphorylation underscores the diverse functions of HDAC inhibitors in mediating antineoplastic activities at different cellular levels.

EXPERIMENTAL PROCEDURES
Cell Culture-U87MG human glioblastoma cells were kindly provided by Dr. Ing-Ming Chiu (The Ohio State University, Columbus, OH), and PC-3 prostate cancer cells were purchased from the American Type Culture Collection (Manassas, VA). Because both U87MG and PC-3 cells are PTEN-null, they exhibit constitutively active Akt. These cancer cells were cultured in 10% fetal bovine serum (FBS)-supplemented RPMI 1640 medium containing 100 units/ml penicillin and 100 g/ml streptomycin (Invitrogen).
Immunoblotting-U87MG and PC-3 cells treated with various concentrations of HDAC inhibitors in 10% FBS supplemented RPMI 1640 medium for 48 h were collected and lysed by Nonidet P-40 isotonic lysis buffer (50 mM Tris-HCl, pH 7.5, 120 mM NaCl, 1% (v/v) Nonidet P-40, 1 mM EDTA, 50 mM NaF, 40 mM ␤-glycerophosphate, and 1 g/ml each of aprotinin, pepstatin, and leupeptin). Protein concentrations of the lysates were determined by using a Bradford protein assay kit (Bio-Rad). Equivalent amounts of proteins from each lysate were resolved by SDS-PAGE and then transferred onto Immobilon-nitrocellulose membranes (Millipore, Bellerica, MA) in a semidry transfer cell. The transblotted membrane was washed twice with Tris-buffered saline (TBS) containing 0.1% Tween 20 (TBST). After blocking with TBST containing 5% nonfat milk for 40 min, the membrane was incubated with the appropriate primary antibody in TBST containing 1% nonfat milk at 4°C overnight. All of the primary antibodies were diluted 1: 2000 in 1% nonfat milk-containing TBST. After treatment with the primary antibody, the membrane was washed two times with TBST for a total of 20 min, followed by goat anti-rabbit or anti-mouse IgG-horseradish peroxidase conjugates (diluted 1:3000) for 1 h at room temperature and washed three times with TBST for a total of 1 h. The immunoblots were visualized by ECL chemiluminescence (Amersham Biosciences).
Cell Viability Assay-The effect of the HDAC inhibitors on cell viability was assessed by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay in 6 -12 replicates. U87MG and PC-3 cells were seeded and incubated in 96-well, flat-bottomed plates in 10% FBS-supplemented RPMI 1640 medium 24 h before drug treatment. The cells were exposed to individual HDAC inhibitors at the indicated concentrations in 10% FBS-supplemented RPMI 1640 medium at 37°C in 5% CO 2 for 48 h. The medium was removed and replaced by 200 l of 0.5 mg/ml of MTT in RPMI 1640 medium, and the cells were incubated in the CO 2 incubator at 37°C for 2 h. Supernatants were removed from the wells, and the reduced MTT dye was solubilized with 200 l/well Me 2 SO. Absorbance was determined on a plate reader at 570 nm.
Flow Cytometry/Apoptosis Assays-After drug treatment as described above, floating cells were collected, and adherent cells were harvested by scraping. The combined cells were fixed in ice-cold 70% ethanol and storage at Ϫ20°C. The cells were centrifuged, washed with PBS, and incubated in 0.2 M phosphate citrate buffer, pH 7.8 at 37°C to extract low molecular weight DNA. The samples were stained with propidium iodide solution (10 mg/ml) in the presence of RNase A (100 mg/ml) for 30 min at room temperature. Cell cycle phase distributions were determined on a FACScort flow cytometer and analyzed by the ModFitLT V3.0 program.
Affinity Purification of HDAC-PP1 Complexes-U87MG cells were treated with TSA at the indicated concentrations in 10% FBS-supplemented RPMI 1640 medium for 48 h. Control cells received the vehicle treatment of 0.1% (v/v) dimethyl sulfoxide. The cells were washed by phosphate-buffered saline and lysed by the aforementioned Nonidet P-40 isotonic lysis buffer. Cell extract was incubated with microcystin-LR-Sepharose (Upstate Biotechnology, Inc.) at 4°C overnight on a rotator. After a brief centrifugation, Sepharose beads were collected and washed with the lysis buffer four times. The bound proteins were eluted by 50 l of SDS sample buffer and subjected to SDS-PAGE followed by immunoblotting with appropriate antibodies.
Co-immunoprecipitation of PP1-Akt Complexes-U87MG cells were treated with various concentrations of TSA for 48 h and lysed by the aforementioned Nonidet P-40 isotonic lysis buffer with a mixture of protease inhibitors. After centrifugation at 13000 ϫ g for 15 min, the supernatants were collected and incubated with protein A-Sepharose beads (Sigma) for 15 min to eliminate nonspecific binding. The mixture was centrifuged at 1000 ϫ g for 5 min, and the supernatants were exposed to Akt or PP1 antibodies in the presence of protein A-Sepharose beads at 4°C for 2 h. After a brief centrifugation, protein A-Sepharose beads were collected, washed with the aforementioned lysis buffer four times, suspended in 2ϫ SDS sample buffer, and subjected to Western blot analysis with antibodies against PP1 or Akt.
Subcellular Fractionation-The nuclear and cytosolic fractions of U87MG cells were prepared by using a nuclear/cytosol fraction kit (MBL Co., Watertown, MA) according to the manufacturer's instructions. In brief, the cells were cultured to 50% confluency in T-75 flasks, treated with TSA for 48 h, and collected by centrifugation. The pelleted cells were resuspended in 0.2 ml of cytosol extraction buffer A mix containing dithiothreitol and a mixture of protease inhibitors and were mixed vigorously on a vortex mixer. The suspension was then incubated on ice for 10 min, mixed with 11 l of cytosol extraction buffer B, and incubated on ice for 1 min. The lysates were centrifuged at 16,000 ϫ g for 5 min. The supernatant, representing the cytoplasmic fraction, was transferred to a prechilled tube, and the pellet was resuspended in 100 l of nuclear extraction buffer mix. The suspension was mixed vigorously on a vortex mixer for 15 s, incubated on ice for 40 min, and centrifuged at 16,000 ϫ g for 10 min to collect the nuclear extract in the supernatant fraction.
Ser/Thr Phosphatase Activity-The Ser/Thr phosphatase activity was determined by using a nonradioactive, malachite green-based Ser/ Thr phosphatase assay kit (Upstate Biotechnology, Inc.) according to the manufacturer's instructions. In brief, 5 g of U87MG cell extracts in the aforementioned lysis buffer were incubated with 175 M of the phosphopeptide substrate (K-R-pT-I-R-R) in the phosphatase buffer (20 mM MOPS, pH7.5, 60 mM 2-mercaptoethanol, 0.1 M NaCl, and 0.1 mg/ml serum albumin) with a total volume of 25 l. The reaction was incubated in a 96-well plate at room temperature for 10 min, terminated by adding the malachite green solution, and allowed to stand for 15 min to permit color development. Absorption at 650 nm was measured in a microplate reader, and the serine/threonine phosphatase activity was calculated using a standard curve based on free phosphate generated by a standard solution. All of the experiments were performed in triplicate.
Immunoprecipitated Phosphatidylinositol 3-Kinase (PI3K) Kinase Assay-PI3K activity was determined by the use of a PI3K enzymelinked immunosorbent assay kit from Echelon Biosciences Inc. (Salt Lake City, UT) according to the manufacturer's instruction. U87MG cells were cultured to 50% confluency in 10-cm culture dishes, and the medium was removed after TSA treatment. The drug-treated cells were washed with 10 ml/dish of ice-cold buffer A (137 mM NaCl, 10 mM Tris-HCl, pH 7.4, 1 mM CaCl 2 , 1 mM MgCl 2 , and 0.1 mM sodium orthovanadate) three times, incubated with 1 ml of ice-cold lysis buffer (buffer A plus 1% Nonidet P-40 and 1 mM phenylmethylsulfonyl fluoride) on ice for 20 min, scraped from the dish, transferred to microcentrifuge tubes, and centrifuged. The supernatant was collected, and the protein concentrations were determined by the Bradford assay (Bio-Rad). Equal amounts of proteins were treated with 5 l of anti-PI3K antibodies (Upstate Biotechnology, Inc.) at 4°C for 1 h, to which was added 60 l of a 50% slurry of protein A-agarose beads in PBS, followed by incubation at 4°C for 1 h with mixing. The immunocomplex was collected by brief centrifugation, washed with lysis buffer three times, and incubated for 2 h at room temperature with 10 l of 10 M L-␣phosphatidyl-D-myo-inositol 4,5-bisphosphate, 5 l of 10ϫ reaction buffer (40 mM MgCl 2 , 200 mM Tris, pH 7.4, 100 mM NaCl, 250 M ATP), and 35 l of distilled water. The reaction was terminated by the addition of 2.5 l of 100 mM EDTA. After brief centrifugation, 50 l of each reaction mixture was transferred to a 96-well plate, and 50 l of diluted L-␣-phosphatidyl-D-myo-inositol 3,4,5-trisphosphate (PIP 3 ) detector was added to each well and incubated at room temperature for 1 h. 100 l of reaction mixture from each well was transferred to the corresponding well in the detection plate and incubated at room temperature for 1 h in the dark. After three washes with 300 l of TBS with Tween 20 (0.05% v/v), 100 l of the secondary detection reagent was added to each well followed by incubation for 1 h in the dark and three more washes with 300 l of TBS with Tween 20 (0.05% v/v). 100 l of tetramethyl benzidine solution was added to each well and incubated for ϳ20 -30 min to allow color development. The reaction was stopped by the addition of 100 l of stop solution (0.5 M H 2 SO 4 ) to each well. Absorbance at 450 nm was read, and the PI3K activity in each sample was calculated using a standard curve generated by using a PIP 3 standard solution. All of the experiments were performed in triplicate.
Autoradiographic Determination of Phosphoinositide Formation-U87MG cells, cultured in T25 flasks, were labeled with 1 mCi/ml of [ 32 P]orthophosphate (HCl-free; PerkinElmer Life Sciences) in Dulbecco's modified Eagle's phosphate-free and serum-free medium (Invitrogen) for 4 h, washed three times with buffer A (30 mM Hepes, pH 7.2, 110 mM NaCl, 10 mM KCl, 1 mM MgCl 2 , and 10 mM glucose), and treated with various concentrations of TSA in 10% FBS-supplemented RMPI 1640 medium for 48 h. The cells were extracted with 3 ml of chloroform/methanol (1:2, v/v), followed by 4 ml of chloroform, 2.4 M HCl (1:1, v/v), and 1 ml of chloroform four times. The combined organic phase was dried under a stream of nitrogen and resuspended in 90 l of chloroform for TLC analysis by using a 20 ϫ 20-cm silica gel 60 TLC plate (EM Science) impregnated with 1% potassium oxalate in 50% ethanol. The TLC plate was developed in chloroform/acetone/methanol/ acetic acid/water (80:30:26:24: (23). Radioactive spots were detected by autoradiography using Kodak X-Omat film, and total PIP 3 was quantified by both densitometry and an AMBIS B scanning system (San Diego, CA), both of which showed comparable results.
Kinase Assay for Phosphoinositide-dependent Kinase-1 (PDK-1)-The PDK-1 kinase activity was performed using a PDK-1 kinase assay kit (Upstate Biotechnology, Inc.) according to a described procedure (24). This cell-free assay is based on the ability of recombinant PDK-1 to activate its downstream kinase serum-and glucocorticoid-regulated kinase, which in turn phosphorylates the Akt/serum-and glucocorti-  NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 coid-regulated kinase-specific peptide substrate RPRAATF with [␥-32 P]ATP. The 32 P-phosphorylated peptide was then separated from residual [␥-32 P]ATP by P81 phosphocellulose paper and quantitated by a scintillation counter after three washes with 0.75% phosphoric acid.

Cellular Effect of Histone Deacetylase Inhibitors on Akt
Isozyme-specific Knockdown of HDACs with siRNA-Isozyme-specific siRNAs were used to attenuate the expression of HDACs 1 and 6 in U87MG cells using reagents obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Transfection of U87MG cells with these siRNA was carried out according to the vendor's instructions. In brief, the cells were cultured to 40 -50% confluency in 10% FBS-supplemented antibiotic-free RPMI 1640 medium. The HDAC1 and HDAC6 siRNAs were transfected to the cells by the siRNA transfection reagent (Santa Cruz, CA) to the final concentration of 200 nM. A scramble siRNA was used in parallel experiments as a control. The transfected cells were incubated at 37°C, and the extents of siRNA-mediated down-regulation of HDAC1 and HDAC6 expression were monitored by immunoblotting analysis at different time intervals. It was found that 30-h exposure to siRNA caused more than 80% repression of HDAC expression, and thus it was used through this study.
Immunocytochemical Analysis of TSA-mediated Co-localization of PP1 and Akt-U87MG cells were treated with 0.5 M TSA in 10% FBSsupplemented RPMI 1640 medium for different time intervals, washed with Dulbecco's PBS, fixed with 4% paraformaldehyde for 30 min at room temperature, and then washed with PBS. For double staining of phospho-Akt and PP1, the cells were permeabilized with 0.1% Triton X-100 in 1% FBS-containing PBS, treated with rabbit anti-phospho-Ser 473 -Akt (1:500 dilution) at 4°C for 24 h, and washed with PBS. Subsequently, the cells were treated with mouse anti-PP1 (1:100 dilution) for 8 h at room temperature and washed with PBS. For fluorescent microscopy, Alexa Fluor 488 goat anti-mouse and anti-rabbit IgG (Molecular Probes, Inc.) were used for conjugating PP1 and phospho-Akt, respectively. The nuclear counterstaining was performed using a 4Ј,6Ј-diamino-2-phenylindole-containing mounting medium (Vector, CA) prior to examination. Images of immunocytochemically labeled samples were observed using a Zeiss confocal microscope (LSM510) with an argon laser and a helium-neon laser and appropriate filters (excitation wavelength, 488 nm for PP1, 633 nm for p-Akt, and 543 nm for 4Ј,6Ј-diamino-2-phenylindole).

Differential Effects of HDAC Inhibitors on Akt Dephosphorylation-
To shed light onto the causative relationship between HDAC inhibition and Akt deactivation, we assessed the effects of four different HDAC inhibitors, including HDAC42, SAHA, MS-275, and TSA, on various HDAC-related biomarkers (histone H3 acetylation, p21 expression, and tubulin acetylation) vis à vis Akt phosphorylation state in U87MG glioblastoma and PC-3 prostate cancer cells. Of the four HDAC inhibitors examined, HDAC42 belongs to a novel class of phenylbutyrate-based HDAC inhibitors (21) and has an IC 50 of 30 nM in inhibiting HDAC activity in nuclear extracts. The reported IC 50 values for TSA, SAHA, and MS-275 are 5-15 nM, 120 nM, and 4.8 M, respectively (2).
Western blot analysis shows that exposure of U87MG and PC-3 cells to these inhibitors led to substantial increases in histone H3 acetylation and p21 WAF/CIP1 expression (Fig. 1A). However, these four inhibitors behaved differently with regard to ␣-tubulin acetylation, indicating differences in the respective activities in inhibiting the ␣-tubulin deacetylase HDAC6 (8). Although HDAC42 and TSA produced robust hyperacetylation of ␣-tubulin at submicromolar concentrations, SAHA was effective at low micromolar concentrations, and MS-275 was totally ineffective, even at 5 M, in inhibiting tubulin deacetylation. Moreover, examination of Akt phosphorylation at both Thr 308 and Ser 473 indicates that HDAC42 and TSA at submicromolar concentrations were able to substantially reduce phospho-Akt levels. A similar effect on phospho-Akt was not achieved with SAHA until the concentration reached 5 M. In contrast, the repressing effect of MS-275 on Akt phosphorylation, even at 5 M, was marginal. It is noteworthy that there existed an inverse relationship between the levels of acetylated ␣-tubulin and phospho-Akt, providing a potential link between HDAC6 inhibition and Akt dephosphorylation. This premise was further supported by the differential effect of selective siRNA-mediated knockdown of individual HDAC isozymes on phospho-Akt levels (see below). Moreover, the time course of TSA-mediated Akt dephosphorylation lagged behind that of p21 overexpression and H3 hyperacetylation by at least 12 h (Fig. 1B), suggesting that this Akt effect might not be related to these two biomarkers.
Selective Dephosphorylation of Signaling Kinases-Pursuant to the above finding, a question emerged with regard to the specificity of HDAC inhibitor-mediated kinase dephosphorylation. Accordingly, we examined the impact of two representative HDAC inhibitors, TSA and SAHA, on the phosphorylation status of ERKs, p38, and JNK MAP kinases versus Akt in U87MG cells ( Fig. 2A). As noted, both TSA and SAHA caused a modest decrease in phospho-ERK1/2 levels despite their respective activities in Akt dephosphorylation, whereas those of phospho-p38 and phospho-JNK remained unaffected. These findings indicate that there existed a certain degree of specificity in HDAC inhibitor-mediated kinase dephosphorylation, suggesting the involvement of a unique signaling mechanism.
Differential Effects of HDAC Inhibitors on Cell Proliferation and Cell Cycle-Together, these findings suggest that these four HDAC inhibitors exhibited distinct profiles regarding pharmacological targets, which might underlie differences in their antitumor activities. For example, although these agents were able to up-regulate p21 expression and histone acetylation, two hallmark features in association with intracellular HDAC inhibition, at submicromolar concentrations, their ability to suppress cell proliferation varied by almost an order of magnitude (Fig. 2B). Although TSA and HDAC42 were effective in suppressing proliferation in U87MG and PC-3 cells at submicromolar concentrations, it would require at least 2.5-5 M for SAHA and MS-275 to attain a similar antiproliferative effect. This differential potency paralleled the respective activities in causing Akt dephosphorylation, suggesting a putative link between Akt down-regulation and the antitumor activities of HDAC42, TSA, and, to a lesser extent, SAHA. Moreover, flow cytometric analyses indicate that the effect of these HDAC inhibitors on growth inhibition was largely attributable to cell cycle arrest in lieu of apoptosis. Analyses of cell cycle distribution in drug-treated U87MG and PC-3 cells shows that HDAC42 and TSA induced dose-dependent accumulation of cells in the G 2 /M phase, accompanied by comparative decreases in the G 0 /G 1 and, in particular, S fractions (TABLE ONE). The sub-G 0 /G 1 fraction, an indication of apoptosis, was not observed until the concentrations of HDAC42 and TSA reached 5 and 0.5 M, respectively. Meanwhile, the cell cycle-arresting behavior of SAHA in these two cell lines differed. It caused dose-dependent increases in the G 2 /M population in U87MG cells, but the effect on cell cycle distribution in PC-3 cells changed from G 1 arrest to G 2 /M arrest as the concentration increased. In contrast, MS-275 induced dose-dependent G 1 cell cycle arrest in both cell lines.

TSA-mediated Akt Dephosphorylation Is Not Caused by Changes in the Expression Level of Proteins Involved in Phospho-Akt Regulation-
Mechanistically, this Akt dephosphorylation might be mediated through the deactivation of upstream kinases or the activation of downstream phosphatase. To discern the role of transcriptional activation in this drug action, we assessed the expression levels of a series of signaling proteins related to the regulation of Akt signaling pathways in TSAtreated U87MG cells, which included the p85 regulatory and p110 catalytic subunits of PI3K, PDK-1, Akt, the negative Akt modulators CTMP (25) and TRB3 (26), PP1, and PP2A. As shown in Fig. 3A, TSA exposure did not alter the expression level of any of these signaling proteins, excluding the involvement of transcriptional activation in altering the status of Akt phosphorylation.
Furthermore, three lines of evidence argued against the possibility that TSA-mediated Akt deactivation was caused by a decrease in PI3K or PDK-1 kinase activity (Fig. 3B). First, immunoprecipitated PI3K kinase activity in U87MG cells treated with different doses of TSA remained the same as that of the Me 2 SO control (Fig. 3B, left panel, open  bars). Second, TSA at different doses exhibited no appreciable inhibitory effects on the kinase activity of recombinant PDK-1 (Fig. 3B, left  panel, gray bars). Third, autoradiographic analysis of 32 P-labeled phospholipids demonstrated that the levels of PIP 3 , a PI3K lipid product, and other inositol lipids in U87MG cells were unaffected by 0.5 M TSA treatment (Fig. 3B, right panel).   NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46

Cellular Effect of Histone Deacetylase Inhibitors on Akt
Inhibition of PP1 Prevents TSA-mediated Akt Dephosphorylation-In the literature, a number of small molecules such as C 2 -ceramide, 4-hydroxynonenal, FTY720, and N-ethylmaleimide have been reported to facilitate Akt dephosphorylation by activating PP2A or PP2A-like phosphatase (27)(28)(29)(30)(31). Consequently, we investigated a potential link between TSA-mediated Akt down-regulation and protein phosphatase activation by examining the effect of tautomycin, calyculin A, and okadaic acid on phospho-Akt levels in TSA-treated U87MG cells. These three compounds exhibit distinct specificity in protein phosphatase inhibition, i.e. tautomycin is a highly specific PP1 inhibitor (32), okadaic acid at low doses (Յ100 nM) is selective for PP2A (33,34), and calyculin A lacks selectivity between PP1 and PP2A. Fig. 4 demonstrates that although none of these inhibitors affected the ability of TSA to inhibit histone deacetylation, their differential effects on the TSA-mediated dephosphorylation of Akt was noteworthy. As shown, calyculin A (50 nM) and tautomycin (5 M) could completely abrogate the effect of TSA on phospho-Akt, whereas okadaic acid lacked appreciable protective activity against the dephosphorylation. Because PP1 represents a common target for calyculin A and tautomycin, this finding suggests the involvement of PP1 in TSA-facilitated kinase dephosphorylation.
TSA Disrupts HDAC-PP1 Complexes, Resulting in Increased PP1-Akt Associations-The link between HDAC inhibition and PP1-mediated kinase dephosphorylation could be attributed to two plausible mechanisms. First, TSA treatment might lead to increased PP1 activity. Second, in light of the reported HDAC-PP1 complex formation (16,17), HDAC inhibitors might cause disruption of these complexes, which would free PP1 to interact with Akt. Of these two mechanisms, the former was refuted by lack of appreciable increases in either the PP1 expression level (Fig. 3A) or the overall Ser/Thr protein phosphatase activity after TSA exposure (data not shown). To evaluate the second possibility, we investigated the expression profile of HDAC isozymes in U87MG cells. Among the seven different isozymes examined, U87MG cells expressed HDACs 1, 2, 3, 5, and 6, but the expression of HDACs 4 and 7 was undetectable. In addition, TSA treatment did not alter the expression level of any of these HDACs (data not shown). To assess the impact of TSA on the dynamics of HDAC-PP1 complexes, we exposed the lysates of drug-treated U87MG cells to microcystin affinity beads to purify PP1-associated complexes (16). Western blot analysis of the affinity-purified PP1 complexes indicates that HDACs 1, 3, and 6 could be pulled-down by the affinity beads. Moreover, the levels of affinity beadbound HDACs 1 and 6 decreased in response to TSA in a dose-dependent manner (Fig. 5A). The level of PP1-associated HDAC3, however, remained unaltered by the TSA treatment.
Based on the report that the PP1-binding domain of HDAC6 encompassed the catalytic motif (17), binding of the HDAC inhibitor to the catalytic domain might hinder the binding of PP1 to HDAC6. Presumably, HDAC inhibitors could also sequester HDAC1, but not HDAC3, from PP1 association through the same mode of mechanism. In contrast, the PP1 inhibitor tautomycin did not cause disturbance of the HDAC-PP1 complex (data not shown), suggesting no overlap between the HDAC-binding domain and the catalytic site of PP1. Also noteworthy is that TSA-mediated dissociation of HDAC-PP1 complexes did not significantly change the relative distribution of PP1 or PP2A in the nucleus or cytoplasm (Fig. 5B).  In addition, two lines of evidence indicate that the TSA-mediated disruption of HDAC-PP1 complexes was accompanied by a dose-dependent increase in PP1-Akt associations. First, immunoprecipitation of PP1 in the lysates of TSA-treated cells followed by Western blotting of Akt, or vice versa, demonstrates a positive correlation between TSA doses and amounts of Akt co-immunoprecipitated with PP1 (Fig. 6A). Second, immunocytochemical examinations show that exposure of U87MG cells to TSA for 4 h led to co-localization of PP1 and phospho-Akt, which, however, was not noted with cells receiving Me 2 SO vehicle (Fig. 6B). The physical interaction between PP1 and Akt led to partial and complete dephosphorylation at 8 and 24 h post-treatment, respectively.
Validation of the Involvement of HDACs 1 and 6 in Akt Dephosphorylation-Together, the above findings suggest that TSA caused PP1-mediated Akt dephosphorylation by sequestering HDACs 1 and 6. To validate the role of these two isozymes in Akt deactivation, we used isozyme-specific siRNAs to selectively knock down the expression of HDACs 1 and 6 and that of a scrambled siRNA as a negative control (Fig.  7A). As shown, repressed expression of HDACs 1 and 6 both led to decreased Akt phosphorylation, whereas control siRNA transfection had no appreciable effect on phospho-Akt levels (Fig. 7B).

DISCUSSION
The present study demonstrates a novel histone acetylation-independent mechanism by which HDAC inhibitors mediate the dephosphorylation of Akt through the disruption of HDAC-PP1 complexes. Together with the previous findings of the functional roles of HDAC-PP1 pairs in the modulation of cAMP-responsive element-binding protein phosphorylation (16) and tubulin acetylation (17), this histoneindependent mechanism provides a potential basis to account for the antineoplastic activities of these agents in growth inhibition and apoptosis induction.
Of the four inhibitors examined, HDAC42 and TSA were most effective in facilitating Akt dephosphorylation, followed by SAHA, whereas MS-275 exhibited a marginal effect at therapeutically attainable concentrations (Յ5 M). In light of the pivotal role of Akt in cell proliferation, this differential activity in Akt down-regulation might be attributed to differences in the antitumor activities among these four HDAC inhibitors. TSA and HDAC42 was potent in suppressing the proliferation of U87MG and PC-3 cells, in part, because of their ability to downregulate Akt signaling. It is noteworthy that the activity of these agents to suppress Akt phosphorylation paralleled the respective potency in inducing ␣-tubulin acetylation, a biomarker for HDAC6 (8). This find-ing suggests a role for HDAC6 inhibition in Akt dephosphorylation. Consequently, MS-275 was ineffective in causing Akt deactivation as a result of its inability to inhibit HDAC6.
The present study further demonstrated a mechanistic link between PP1 and TSA-mediated Akt dephosphorylation. Our immunochemical study showed that TSA blocks specific associations of PP1 with HDACs 1 and 6, suggesting the involvement of both isozymes in regulating the FIGURE 6. TSA treatment leads to increased PP1-Akt associations. A, immunoprecipitation (IP) of PP1 in lysates of TSA-treated cells, followed by immunoblotting (IB) of Akt (left panel) or vice versa (right panel). U87MG cells were treated with TSA at the indicated concentrations, and the cell lysates were exposed to anti-Akt or anti-PP1 in the presence of protein A-Sepharose beads. The immunocomplex was immunoblotted with anti-PP1 or anti-Akt. B, immunocytochemical examinations of PP1 and phospho-Akt in U87MG cells treated with Me 2 SO vehicle or 500 nM TSA for the indicated times. FIGURE 5. A, TSA selectively disrupts the association of PP1 with HDACs 1 and 6. U87MG cells were treated with TSA at the indicated concentrations for 48 h. The cell lysates were exposed to microcystin affinity beads to purify PP1-associated complexes. The affinity-purified complexes were probed with individual HDAC isozyme-specific antibodies by Western blot analysis (WB). B, TSA does not change the relative abundance of PP1 or PP2A in the nucleus or cytoplasm in drug-treated U87MG cells. ␣-Tubulin and nucleolin are used as internal references for the cytoplasmic and nuclear fractions, respectively. dynamics of PP1 complexes. Subsequent co-immunoprecipitation and immunocytochemical assays revealed that disruption of HDAC-PP1 complexes leads to increased association of PP1 with Akt, resulting in Akt deactivation. Moreover, we used isozyme-specific siRNAs to confirm the role of HDACs 1 and 6 as key mediators in facilitating Akt dephosphorylation. The effect of reduced abundance of HDACs 1 or 6 on Akt deactivation mimicked that of HDAC inhibitors. We reasoned that both siRNAs and HDAC inhibitors could destabilize PP1-HDAC complexes, resulting in increased association of PP1 with Akt. With regard to ERK1/2, PP1 has been shown to act on their upstream kinases including Raf, MEK1/2, and MEK kinases (35). Consequently, HDAC inhibitor-mediated ERK dephosphorylation might be attributable to the deactivation of these upstream kinases.
Mechanistically, the HDAC inhibitor-mediated reorganization of PP1 complexes represents a unique mode of Akt regulation independent of PI3K-mediated activation. Previously, drug-facilitated PP1-dependent Akt dephosphorylation was reported in ErbB2-overexpressing breast cancer cells treated with the heat shock protein 90 (Hsp90) inhibitor geldanamycin or the ErbB inhibitor ZD1839 (36), however, through a distinct mechanism. In contrast to the reshuffling of PP1 complexes, the effect of geldanamycin and ZD1839 on Akt deactivation was attributable to the dephosphorylating activation of PP1 as a result of ErbB inhibition. In the present study, no increase in Ser/Thr protein phosphatase activity was observed in TSA-treated cells. In addition, it was reported that inhibition of HDAC6 resulted in the hyperacetylation and loss of the chaperone activity of Hsp90 (37). Because Hsp90 forms intracellular complexes with Akt (38), inhibition of Hsp90 function provides a plausible mechanism for down-regulating Akt kinase activity through the promotion of Akt degradation (39,40). However, involvement of Hsp90 in HDAC inhibitor-induced Akt dephosphorylation in U87MG and PC-3 cells is refuted by the finding that the Akt level remained unaltered.
Recently, PP1 has emerged as an important therapeutic target in light of its regulatory role in a plethora of cellular functions (41). However, targeting PP1 proves elusive because it forms complexes with over 50 established or putative regulatory subunits in different cellular compartments. The selective action of HDAC inhibitors on cellular HDAC-PP1 complexes represents the first example of modulating specific PP1-protein interactions by small molecule agents. From a clinical perspective, identification of this PP1-facilitated dephosphorylation mechanism underlies the potential use of HDAC inhibitors in lowering the apoptosis threshold for other therapeutic agents through the downregulation of Akt and ERK signaling. This therapeutic strategy is illustrated by the ability of HDAC inhibitors to sensitize cancer cells to the apoptotic effects of the Bcl-Abl kinase inhibitor imatinib (42), the Her-2 antibody trastuzumab (19), the receptor tyrosine kinase FLT-3 inhibitor PKC412 (20), the purine analogue fludaribine (43), and the Hsp90 antagonist 17-allylamino-demethoxy geldanamycin (44,45).
In conclusion, the ability of HDAC inhibitors to deactivate Akt through the reorganization of PP1 complexes underlines the complexity of the pharmacological functions of these agents. In light of the clinical application of HDAC inhibitors, a better understanding of this novel histone-independent mechanism will allow the design of more effective strategies to optimize the use of the agents in cancer treatment and/or prevention.