Rapid Induction of Histone Hyperacetylation and Cellular Differentiation in Human Breast Tumor Cell Lines following Degradation of Histone Deacetylase-1*

Quinidine inhibits proliferation and promotes cellular differentiation in human breast tumor epithelial cells. Previously we showed quinidine arrested MCF-7 cells in G1 phase of the cell cycle and led to a G1 to G0 transition followed by apoptotic cell death. The present experiments demonstrated that MCF-7, MCF-7ras, T47D, MDA-MB-231, and MDA-MB-435 cells transiently differentiate before undergoing apoptosis in response to quinidine. The cells accumulated lipid droplets, and the cytokeratin 18 cytoskeleton was reorganized. Hyperacetylated histone H4 appeared within 2 h of the addition of quinidine to the medium, and levels were maximal by 24 h. Quinidine-treated MCF-7 cells showed elevated p21 WAF1 , hypophosphorylation and suppression of retinoblastoma protein, and down-regulation of cyclin D1, similar to the cell cycle response observed with cells induced to differentiate by histone deacetylase inhibitors, trichostatin A, and trapoxin. Quinidine did not show evidence for direct inhibition of histone deacetylase enzymatic activity in vitro. HDAC1 was undetectable in MCF-7 cells 30 min after addition of quinidine to the growth medium. The proteasome inhibitors MG-132 and lactacystin completely protected HDAC1 from the action of quinidine. We conclude that quinidine is a breast tumor cell differentiating agent that causes the loss of HDAC1 via a proteasomal sensitive mechanism.

Histone deacetylase (HDAC) 1 proteins comprise a family of related proteins that act in conjunction with histone acetyltransferase proteins to modulate chromatin structure and transcriptional activity via changes in the acetylation status of histones. Histones H3 and H4 are the principal histone targets of HDAC enzymatic activity, and these histones undergo acetylation at lysine residues at multiple sites within the histone tails extending from the histone octamer of the nucleosome core. The association of HDAC proteins with mSin3, N-CoR, or SMRT and other transcriptional repressors has led to the hypothesis that HDAC proteins function as transcriptional corepressors (reviewed in Ref. 1). The spectrum of genes that show alterations in gene transcription rates in response to decreased HDAC activity is quite restricted (2). Yet, small molecule inhibitors of the enzyme histone deacetylase (HDAC) such as trichostatin A (TSA), superoylanilide hydroxamic acid (SAHA), trapoxin, and phenyl butyrate cause major alterations in cellular activity including the induction of cellular differentiation and apoptosis (3)(4)(5). Trichostatin A, SAHA, and trapoxin stimulate histone acetylation by acting as direct inhibitors of HDAC enzyme activity (6). Trichostatin A, SAHA, and trapoxin possess lysine-like side chains and act as chemical analogs of lysine substrates. Molecular models based upon the x-ray crystal structure of an HDAC-like protein indicate that trichostatin A and SAHA can bind within the active site of the HDAC enzyme and interact with a zinc metal ion within the catalytic pocket that is critical for enzymatic activity (7). Trapoxin is an irreversible HDAC enzyme inhibitor (8).
Much remains to be learned about the biochemical events subsequent to HDAC inhibition that lead to cell cycle arrest, cellular differentiation, and apoptosis. However, a spectrum of biological responses characteristic of HDAC inhibitors has emerged, including cell cycle arrest in G 1 , elevated p21 WAF1 expression, hypophosphorylation of retinoblastoma protein (pRb), hyperacetylation of histones, particularly H3 and H4, and apoptosis. Histone hyperacetylation is directly linked to the activation of p21 transcription and is p53-independent (5). This observation provides an important link between HDAC inhibition and cell cycle arrest because p21 WAF1 plays a critical role in causing G 1 cell cycle arrest via inhibition of the G 1 cyclin-dependent kinase family (9). Overexpression of p21 WAF1 has also been associated with apoptosis, but the mechanism of p21 WAF1 induction of apoptosis requires further investigation (10).
Cancer therapy that targets the activity of genes or gene products controlling cell cycle progression, differentiation, and apoptosis is a promising new strategy. Because HDAC inhibitors regulate the cell cycle and cause both cellular differentiation and apoptosis, they comprise an interesting group of compounds with potential for development into a new category of clinically significant anti-tumor agents. Single, key protein targets for "gene-regulatory chemotherapy" are difficult to identify due to the existence of parallel, functionally overlap-ping signaling cascades. For this reason, use of cancer therapies that target multiple intracellular signaling pathways, such as observed with the HDAC inhibitors, is an intriguing approach that addresses the problem of redundancy in growth signaling pathways. In this regard, the HDAC inhibitor phenyl butyrate was recently shown to have clinical anti-tumor activity (11).
Quinidine is a natural product therapeutic agent originally used as an anti-malarial and as an anti-arrhythmic agent. Previous studies with human breast tumor cell lines demonstrated that quinidine (90 M) is an anti-proliferative agent as well. Quinidine arrested cells in early G 1 phase and induced apoptosis by 72-96 h in MCF-7 cells (12), but the biochemical basis for the anti-proliferative effect of quinidine was not well understood. To clarify the molecular mechanisms of the antiproliferative activity of quinidine, we investigated the effects of quinidine on histone acetylation and cell cycle regulatory proteins. In this report, we show that quinidine causes hyperacetylation of histone H4, down-regulation of HDAC1 protein levels, and cellular differentiation in a panel of human breast tumor cell lines. We conclude that quinidine is a novel differentiating agent that stimulates histone hyperacetylation as a result of HDAC1 protein degradation.

MATERIALS AND METHODS
Cell Culture-Permanent cell lines derived from patients with breast carcinomas were used in these studies. MCF-7 cells, passage numbers 40 -55, MCF-7ras (13), T47D, MDA-MB-231, and MDA-MD-435 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) (Bio-Whittaker, Walkersville, MD) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (HyClone Laboratories, Inc., Logan, Utah), 2 mM glutamine, and 40 g/ml gentamicin. Experiments were performed in this medium supplemented with 5% FBS. The cells were maintained at 37°C in a humidified atmosphere of 93% air, 7% CO 2 . After 6 days, cells became about 70 -80% confluent and were passaged at a 1:5 ratio (MCF-7) or at a 1:10 ratio (all others). Normal human mammary epithelial cells (HMEC) were obtained from Clonetics, San Diego, CA, and were grown according to directions of the suppliers. Cells were grown from frozen stocks and used for 1-3 passages. Quinidine, TSA, and all-trans-retinoic acid were purchased from Sigma. The cell-permeant proteasome inhibitors, MG-132 and lactacystin, were purchased from Calbiochem.
Growth Inhibition Assays-Growth inhibition by cell numbers was assayed by plating cells in 35-mm 2 dishes (1-1.5 ϫ 10 5 ) containing DMEM, 5% FBS plus quinidine (90 M). Viable cells were counted using a hemocytometer, and trypan blue (0.02%) exclusion was used as an indicator of viability. Cell growth was also monitored in a 96-well plate format using the One Solution Cell Proliferation Assay (Promega, Madison, WI), which is based upon metabolic bioreduction of a tetrazolium compound (Owen's reagent) to a colored formazan product that absorbs light at 490 nm. The plating density for the 96-well dishes (cells/well) was varied depending upon the relative growth rates of the cell lines as follows: HMEC (2000), MCF-7 (1000), MDA-MB-231 (500), T47D (1500), and MCF-7ras (500). The One Cell Proliferation Assay Reagent was added to each well and incubated for 2 h at 37°C. Absorbance (490 nm) was read using a Molecular Devices PC340 (Sunnyvale, CA).
Microscopic Imaging-Cells were plated (1 ϫ 10 5 ) on sterile coverslips in 35-mm 2 dishes and grown for 96 h in DMEM, 5% FBS supplemented with either 10 M all-trans-retinoic acid (in 0.01% ethanol) or 90 M quinidine (in H 2 0). Control cells were grown in medium containing a final concentration of 0.01% ethanol. The presence of ethanol had no effect upon lipid droplet accumulation compared with cells grown in DMEM, 5% FBS. Cells were fixed in 3.7% formaldehyde/PBS, rinsed in PBS (PBS: 140 mM NaCl, 2 mM KCl, 80 mM Na 2 HPO 4 ⅐7H 2 O, 1.5 mM KH 2 PO 4 , pH 7.0), then treated briefly with 0.4% Triton X-100 in PBS. After rinsing three times in PBS, the cells on coverslips were incubated for 30 min at 37°C with a primary antibody to cytokeratin 18 (1:1 dilution, provided by Dr. Guillaume van Eys, Maastricht University), rinsed, and incubated (30 min/37°C) with Texas Red conjugated secondary antibody (goat anti-mouse IgG, Sigma). Alternatively, cells were incubated with fluorescein-phalloidin (1:200 dilution of a-5 g/0.1 ml solution, Sigma) in the dark for 40 min at room temperature, rinsed, and incubated for 5 min (room temperature) with the fluorescent lipid stain, Nile Red (1:10,000 dilution of a 1 mg/ml acetone solution, Sigma) (14 -15). All coverslips were rinsed in PBS and mounted with Fluoromount-G containing 2.5% N-propyl galate. Images were obtained using a Zeiss Axiovert 100 M confocal microscope (ϫ 63 objective).
Histones were prepared from cells grown at a density of 1 ϫ 10 7 /T-162 flask. To harvest the cells, the flasks were placed on ice, and the growth medium was removed. Following a quick rinse with ice-cold PBS, cells were scraped into 1 ml of ice-cold lysis buffer (10 mM Tris-HCl, 50 mM sodium bisulfite, 1% Triton X-100 (v/v), 10 mM MgCl 2 , 8.6% sucrose, pH 6.5) and nuclei released by Dounce homogenization. The nuclei were collected by centrifugation (3,000 rpm, 10 min, SS-34 rotor) and washed three times with the lysis buffer. Histones were extracted from the crude nuclear pellets using the procedure of Nakajima et al. (16). The pellets were resuspended in 0.1 ml of ice-cold sterile water using a vortex and concentrated H 2 SO 4 to 0.4 N was added. The preparation was incubated at 4°C for 1 h and then centrifuged (17,000 rpm, 10 min, Sorvall SS-34 rotor). The supernatant containing the extracted histones was mixed with 10 ml of acetone, and the precipitate was obtained after an overnight incubation at Ϫ20°C, collected, and airdried. The acid-soluble histone fraction was dissolved in 50 l of H 2 O and stored at Ϫ70°C.
In Vitro HDAC Activity Assay-Quinidine HCl was added to a chicken erythrocyte cellular extract, which contained HDAC activity, at concentrations of 90 M (18). HDAC assays were performed as described in Hendzel et al. (17). Briefly, the cellular extract was incubated with 500 g of acid-soluble histones isolated from [ 3 H]acetate-labeled chicken erythrocytes for 60 min at 37°C. Reactions were terminated by addition of acetic acid/HCl to a final concentration of 0.12/0.72 N. Released [ 3 H]acetate was extracted using ethyl acetate and quantified by scintillation counting. Samples were assayed three times, and the nonenzymatic release of label was subtracted to obtain the reported values.

RESULTS
Hyperacetylation of Histone H4 -Antibodies that recognize acetylated forms of histone H4 have been used as a probe for agents that cause histone hyperacetylation (19). In Western blot experiments, we compared the ability of quinidine to cause hyperacetylation of H4 in MCF-7 cells with that of TSA, an established HDAC inhibitor, known to inhibit proliferation in MCF-7 cells (20). H4 acetylation in response to TSA was rapid (within 0.5 h) (Fig. 1B) and reached a maximum around 12 h. Some level of H4 acetylation persisted in the TSA-treated cells for 48 h. In cells treated with quinidine ( Fig. 1A), detectable H4 acetylation was slightly delayed and could be seen at 2 h but not 1 h of treatment. H4 acetylation was maximal between 12 and 24 h but then sharply fell to an undetectable level at 48 h. Hyperacetylation of H4 was a transient response to both agents. Acetylated H4 is present in MCF-7 cells but under the conditions of Western blotting and immunochemical staining was not detected in control cells (data not shown).
HDAC1 is expressed in MCF-7 cells, and this enzyme contributes to the control of histone deacetylation rates (18). Quinidine caused the rapid disappearance of HDAC1 from MCF-7 cells. HDAC1 protein levels in quinidine-treated cells were reduced after 15-20 min compared with control cells, and HDAC1 protein was undetectable between 30 min and 6 h (Fig.  1, E and F). Partial restoration of HDAC1 protein occurred beginning at 9 h of treatment, but even after 48 h, HDAC1 levels in quinidine-treated cells were still less than control cells. Levels of HDAC1 protein in control MCF-7 cells were relatively constant during this time (Fig. 1C). TSA-treated MCF-7 cells also showed reduced levels of HDAC1 protein as early as 30 min after drug addition, and the reduced HDAC1 protein level was maintained through 48 h (Fig. 1D). The data indicate that loss of HDAC1 protein might contribute to the H4 acetylation response to both TSA and quinidine. However, HDAC1 protein levels were never reduced by TSA below the level of detection as was observed with quinidine. In light of the more extensive H4 acetylation response to TSA than quinidine, we conclude that the direct inhibition of HDAC1 catalytic activity by TSA remains an important component of the H4 acetylation response in vivo. In addition, the time course of the HDAC1 response to quinidine and TSA differ. In response to quinidine, there is initially a more marked decrease in HDAC1 protein levels but a more rapid recovery. TSA treatment caused a sustained reduction in HDAC1 protein levels through 48 h.
To determine if the rapid loss of HDAC protein in the presence of quinidine were mediated through the 26 S proteasome pathway, MCF-7 cells were treated simultaneously for 30 min with quinidine and MG-132 (30 M), an inhibitor of the 26 S proteasome. Cells treated with 90 M quinidine showed a complete loss of HDAC1 protein, which was prevented when MG-132 and quinidine were added simultaneously ( Fig. 2A). Treatment with the solvent, Me 2 SO, or MG-132 in solvent (0.1%) caused a modest reduction in the level of HDAC1 protein.
These reductions in HDAC1 did not elicit a detectable stimulation of H4 acetylation, and we suggest that other HDAC enzymes present in MCF-7 cells, insensitive to Me 2 SO, could compensate for the lost HDAC1 in the maintenance of deacetylated histone H4. This action of quinidine on HDAC1 protein was not reflected in a general decrease in cellular protein content (12), nor were all cell cycle regulatory proteins downregulated in MCF-7 cells in the presence of quinidine (e.g. p21 WAF1 and p53 protein, Fig. 3). Additional studies are required to define the spectrum of proteins affected by quinidine in a proteasome-sensitive manner. Quinidine (90 or 250 M) did not inhibit the activity of the isolated chicken erythrocyte HDAC1 enzyme in vitro (data not shown) suggesting that quinidine caused histone hyperacetylation by eliciting a rapid and transient loss of HDAC1 protein without a direct inhibition of the HDAC enzyme. The suppression of HDAC protein levels in MCF-7 cells was accompanied by a decrease in HDAC enzyme activity in the cell extracts. Histone acetylation and depressed HDAC1 protein levels persisted for approximately 48 h in the presence of quinidine. When MCF-7 cells were exposed to quinidine for 24 h in the presence of either MG-132 or lactacystin, there was no detectable H4 acetylation (Fig. 2B). These results support the idea that quinidine-induced loss of HDAC1 protein is involved in the H4 acetylation response via a proteasomal sensitive pathway. G 1 Phase Cell Cycle Regulatory Profile in MCF-7 Cells-G 1 cell cycle arrest is characteristic of HDAC inhibitors, and reports of alterations in several cell cycle proteins in cells exposed to HDAC inhibitors, particularly the elevation of the p21 WAF1 protein, are numerous (21)(22)(23). It was of interest to determine whether p21 WAF1 and other key cell cycle regulatory proteins such as the retinoblastoma protein (pRb) and the G 1 phase cyclin-dependent kinase activator, cyclin D1, were targets of quinidine action in MCF-7 cells. Western blotting analysis showed that by 12 h the levels of p21 WAF1 were increased in response to quinidine treatment approximately 11-fold, and this elevated level of protein expression persisted through 48 h. A small, less than 2-fold increase in p27 levels was observed in cells exposed to quinidine for 24 -48 h, whereas levels of p16 were unchanged (Fig. 3). Quinidine treatment decreased cyclin D1 and CDK4 protein levels after 12 h of treatment (Fig. 4), indicating that the cyclin-dependent kinase inhibitor, p21 WAF , as well as an important G 1 phase target of p21 WAF1 , the cyclin D1-CDK4 complex, are early targets of quinidine in MCF-7 cells. This profile of activity is consistent with the observed cell cycle arrest of quinidine-treated MCF-7 cells in mid-G 1 phase (12).
In MCF-7 cell extracts probed using anti-pRb antibodies, two separate but closely migrating bands were distinguishable. The upper band contained more highly phosphorylated pRb, and the lower band contained unphosphorylated or hypophosphorylated pRb. Control cells showed a faint pRb signal at 12 h, typical of cells in early G 1 phase, and increased expression of both phosphorylated and unphosphorylated pRb at 24 and 48 h. Quinidine-treated MCF-7 cells had no detectable hyperphosphorylated pRb at any time point examined, and total levels of pRb protein failed to increase with progression through G 1 phase as seen in the control, proliferating cells (Fig. 3). The decrease in pRb phosphorylation level was predictable based on the increase in p21 WAF1 and decreased levels of both cyclin D1 and CDK4 (Fig. 4). In addition, Nakanishi et al. (24) showed that p21 WAF1 can bind pRb protein and block its phosphorylation. However, the actions of quinidine upon p21 WAF1 and cyclin D-CDK4 activity do not explain why the levels of total pRb protein were so low. Reductions in the cellular content of phosphorylated pRb protein in MCF-7 cells by quinidine is an important additional level of cell cycle control that effectively attenuates progression of cells out of G 1 phase and has been reported in other tumor cell lines in response to HDAC inhibition (22). In Fig. 5 we show data suggesting that the 26 S proteasome pathway regulates the total pRb content. MCF-7 cells incubated for 24 h in MG-132 or MG-132 plus quinidine had more total pRb than cells incubated with quinidine alone. Thus, quinidine promoted the loss of both HDAC1 and pRb, and inhibition of the 26 S proteasome pathway restored the levels of both of these proteins to that seen in the untreated cells. We have no direct evidence that quinidine promotes the proteasomal degradation of either protein. We hypothesize that quinidine may direct degradation of HDAC1 by the proteasome or, alternatively, quinidine might stimulate the proteasomal degradation of other regulatory factor(s) that act to maintain HDAC1 and pRb protein levels.
MCF-7 cells express wild-type p53 protein. Normal p53 is a short lived protein that is maintained at low levels, but in response to cell stress or DNA damage, p53 is stabilized and accumulates in the nucleus where it functions as a transcription factor inducing p21 WAF , G 1 cell cycle arrest, and apoptosis (25). Wild-type p53 down-regulates pRb levels in MCF-7 cells (26). Although Saito et al. (22) showed that p53 is not required for pRb down-regulation by HDAC inhibitors in all cell lines, quinidine-treated MCF-7 cells have elevated p53 levels (5-7fold) (Fig. 3). Thus, p53 could contribute to the maintenance of the G 1 cell cycle arrest in MCF-7 by sustaining p21 WAF1 protein levels and suppressing pRb protein levels.
Growth Arrest and Cellular Differentiation in Human Breast Tumor Cell Lines-In contrast to MCF-7 cells, human breast tumor cell lines T47D, MDA-MB-231, and MDA-MB-435 express p53 proteins with distinct point mutations (27). To test for a requirement of p53, this panel of human breast tumor cell lines was exposed to quinidine, and the effects of quinidine on cell growth were compared (Fig. 6). The data shown are viable cell numbers/well, bioreductive metabolism/well, or both. In all four cell lines growth was suppressed in a concentration-dependent fashion between 10 and 90 M quinidine, and maximal growth inhibition was observed at ϳ90 M quinidine (data not shown). These data showed that growth suppression by quinidine is a p53-independent response. It is interesting that quinidine was not overtly cytotoxic in HMEC, a line of normal human mammary cells (28).
Evidence that quinidine elicited cellular differentiation in MCF-7 human breast tumor cells in conjunction with the inhibition of cell growth was obtained using maximally effective concentrations of quinidine or retinoic acid (data not shown). Antibodies directed against cytokeratin 18 (29) were used to probe the organization of the cytoskeleton (Fig. 7). In these studies, all-trans-retinoic acid (10 M) was used to compare the differentiation response (30). Control MCF-7 cells showed expression of cytoplasmic cytokeratin 18 in a disorganized fash- Cells released from confluency were plated into control medium or medium containing 90 M quinidine. Whole cell lysates were prepared 12, 24, or 48 h after plating and assayed by immunoblotting for the cyclin-dependent kinase inhibitors, p21 WAF1 (n ϭ 3), p27 (n ϭ 1), p16 (n ϭ 3), and p53 (n ϭ 3) after electrophoresis of 50 g of protein/lane through 12% SDS-polyacrylamide gels. pRb protein was immunoprecipitated from 500 g of whole cell lysate protein using an antibody that recognizes phosphorylated and non-phosphorylated pRb (56). This entire immunoprecipitate was electrophoresed in a 7.5% SDS-polyacrylamide gel and immunoblotted using this same antibody. Results shown are typical of two independent analyses. ion. Cells that were treated for 96 h with retinoic acid showed an increase in the intensity of the cytokeratin 18 staining and relocalization of cytokeratin 18 throughout the nucleus as well as the cytoplasm. In contrast, cytokeratin 18 staining occurred in a highly organized pattern in MCF-7 cells treated with quinidine for 96 h and the cells adopted a shape and nuclear localization more typical of columnar epithelium.
Lipid droplets are found in the cytoplasm of normal mammary epithelium (31), and cytoplasmic lipid droplet accumulation occurs in a variety of differentiating cell systems. Induction of differentiation in human breast cancer cell lines by oncostatin M (32), the HER-2/neu kinase inhibitor, emodin (33), overexpression of c-e,rbB-2 (34), the vitamin D analog, 1-␣-hydroxyvitamin D5 (35), the HDAC inhibitor, sodium bu-tyrate (36), and retinoic acid (36) is accompanied by the accumulation of cytoplasmic lipid droplets. We utilized a fluorescent stain, Nile Red to monitor lipid droplet formation in mammary tumor cells in response to quinidine. The cells were counterstained with fluorescein-phalloidin that binds actin filaments to assay for changes in the actin cytoskeleton (Fig. 8). lines and increased by retinoic acid and quinidine. Lipid droplet accumulation was more marked in all four cell lines treated with quinidine than with retinoic acid. These data demonstrate that induction of a more differentiated phenotype is a general response of human mammary tumor cells to quinidine.
Hyperacetylation of Histone H4 in Mammary Tumor Cell Lines by Quinidine-To determine whether differentiation and histone acetylation were linked, we investigated the histone H4 acetylation status of quinidine-treated T47D, MDA-MB-231, and MCF-7 ras cells. MCF-7, MCF-7ras, T47D, and MDA-MB-231 cells were incubated for 24 h in the presence or absence of quinidine, and then histones were extracted for immunoblotting. Fig. 9 shows that histone H4 was hyperacetylated in all cell lines treated with quinidine. Control cells contained no hyperacetylated histone H4. DISCUSSION Quinidine-induced histone H4 hyperacetylation in MCF-7 human breast carcinoma cells can be attributed to the rapid elimination of HDAC1 protein, a response that was blocked by MG-132 and lactacystin, two inhibitors of proteasome-mediated proteolysis. HDAC1 protein was undetectable within 30 min after the addition of quinidine to the medium of MCF-7 cells, and hyperacetylated histone H4 appeared between 1 and 2 h. Levels of HDAC1 protein were completely suppressed between 0.5 and 6 h, and during this time H4 acetylation levels increased. H4 acetylation was maintained at 12 and 24 h, despite the partial restoration of HDAC1 protein at these same time points. These data indicate that quinidine-induced reductions in HDAC1 protein levels are unlikely to explain fully the regulation of H4 acetylation state in MCF-7 cells by quinidine. Additional HDAC enzymes or effects upon histone acetylation rates could possibly play a role as well.
An earlier study showed that over this initial 48-h period, 80% of the MCF-7 cell population had shifted into G 0 , a quiescent state marked by the absence of Ki67 antigen immunoreactivity (12). Cellular differentiation manifested as the accumulation of lipid droplets, and a reorganization of the cytokeratin 18 cytoskeleton was evident after this initial 48-h period. Quinidine exhibited all the responses typical of known HDAC inhibitory drugs, with the exception that quinidine had no direct inhibitory effect upon HDAC1 enzymatic activity. We conclude from the current studies that quinidine is a novel differentiating agent that causes histone hyperacetylation, in part, by physical elimination of HDAC1 protein rather than the inhibition of HDAC enzymatic activity.
Histone H4 hyperacetylation and induction of cellular differentiation by quinidine were seen in a panel of human breast tumor cell lines that were selected for study on the basis of their diversity of genetic backgrounds. The differentiation response to quinidine was independent of the estrogen receptor (ER) status. Cell lines representative of ER-positive and ERnegative human breast carcinoma cells were induced to differentiate in the presence of quinidine. The ER status of the estrogen receptor positive cell lines is MCF-7 (ER-␣ and ER-␤), T47D (ER-␣ and ER-␤), and MDA-MB-231 (ER-␤). MDA-MB-435 cells expressed very low levels of ER-␤ and no ER-␣ (37,38). MCF-7 and T47D cells display an epithelial morphology and show similarities with mammary ductal and luminal epithelial cells, respectively (30,39). MDA-MB-231 cells exhibit an elongated cellular morphology that is also typical of MDA-MB-435 cells. Our results demonstrate that quinidine is a differentiation agent in both types of mammary tumor cells.
HDAC inhibitors reverse the transformed phenotype of NIH3T3ras cells, and this property has been used successfully as a screening assay for the identification of new HDAC inhibitors (40,41). Quinidine elicited a more differentiated phenotype in MCF-7ras cells, an MCF-7 cell derivative produced by stable transformation with v-Ha-ras, thus demonstrating that quinidine, like other HDAC inhibitors, can reverse an Ha-rasinduced phenotype.
Quinidine induced differentiation independently of wild-type p53. The ability of quinidine to cause differentiation of p53 mutant cell lines is consistent for a role of histone hyperacetylation in the response. HDAC inhibitors typically induce a p53-independent activation of p21 WAF1 gene expression (5,22). Growing MCF-7 and T47D cells express p21 WAF1 protein in moderate to low levels (42), and quinidine raised p21 WAF1 protein levels in MCF-7 cells approximately 11-fold within 12 h. Although p21 WAF1 was reported to be low to undetectable in MDA-MB-231, p21 WAF1 was detected in Western analyses of both MDA-MB-231 and T47D cells in a p53-independent fashion in response to serum deprivation, adriamycin, etoposide (42,43), and quinidine (data not shown). These data support the idea that the p21 WAF1 gene is present but inactive in growing MDA-MB-231 cells. Since histone hyperacetylation of the p21 WAF1 gene occurs in response to HDAC inhibitors, it might be involved in the pathway of p53-independent activation of p21 WAF1 gene expression (5).
The processes of cellular differentiation and cell cycle progression are interdependent. G 1 arrest is a necessary but insufficient condition for differentiation in numerous cell types including leukemic cells, keratinocytes, colonic epithelium, and muscle cells. In all of these cells, induction of p21 WAF1 protein and G 1 cell cycle arrest occurred prior to differentiation (44 -50) and was generally independent of p53. We hypothesize that the differentiated state can be viewed as a cellular response to G 1 arrest, requiring a change in gene expression profile and suppression of cell death pathways. The response of MCF-7 breast tumor cells to quinidine is consistent with this model.
To begin to understand how quinidine might elicit G 1 arrest of MCF-7 cells, we have focused on the action of quinidine as a potassium channel blocking agent. Quinidine enters cells and inhibits cardiac potassium channels by binding to the intracellular face of the ion pore (51). Although the location of the quinidine-binding site on the ATP-sensitive potassium channel is unknown, quinidine is freely permeable across membranes and inhibits the ATP-sensitive potassium channels whether it is applied to the external or internal surface of a lipid membrane bilayer (52).
In the presence of quinidine, MCF-7 cells accumulate at a position 12 h into G 1 phase (12). This position, defined by cell cycle arrest and release experiments, precedes the lovastatin arrest point by 5-6 h and is clearly distinct from the restriction point described by Pardee (53) near the G 1 /S transition. The present work showed that quinidine treatment caused elevated levels of p53 and p21 WAF1 protein by 12 h (Fig. 3), the point within G 1 where MCF-7 cells arrest in response to quinidine (12). When p53 and p21 WAF1 proteins were assayed before 12 h, p53 was undetectable, and p21 WAF was first detected after 8 h of quinidine treatment (data not shown), suggesting a p53independent induction of p21 WAF1 occurred prior to arrest in G 1 . CDK4 and cyclin D1 protein levels were also reduced, as was CDK4 activity as demonstrated by the abundance of hypophosphorylated pRb protein. Based upon our observations in MCF-7 cells, we conclude that p21 WAF1 protein levels become elevated prior to the G 1 arrest in response to quinidine and could initiate the G 1 arrest. Hypophosphorylated pRb protein is prominent in quinidine-treated MCF-7 cells, and this could act to sustain the G 1 state by preventing the transition into S phase. The G 1 arrest induced by quinidine in MCF-7 cells was correlated with the blockade of ATP-sensitive potassium channels in MCF-7 cells (12,54,55). Direct evidence for the involvement of potassium ions in the G 1 arrest was provided using valinomycin, a potassium-selective ionophore to stimulate a G 1 -S phase transition in the presence of quinidine (12).
In summary, quinidine, a drug that is used therapeutically in the treatment of malarial infections and cardiac arrhythmia, was shown to be useful as an inducer of cellular differentiation in human breast tumor epithelial cells. Quinidine caused histone H4 hyperacetylation and cellular differentiation in human breast tumor cells following the rapid loss of HDAC1 involving a proteasome-dependent pathway. Additional experiments are needed to determine how the action of quinidine upon ATPsensitive potassium channels initiates the molecular events underlying the differentiation response.