Vorinostat increases carboplatin and paclitaxel activity in non-small cell lung cancer cells

Abstract

We observed a 53% response rate in non-small cell lung cancer (NSCLC) patients treated with vorinostat plus paclitaxel/carboplatin in a Phase I trial. Studies were undertaken to investigate the mechanism (s) underlying this activity. Growth inhibition was assessed in NSCLC cells by MTT assay after 72 h of continuous drug exposure. Vorinostat (1 µM) inhibited growth by: 17±7% in A549, 28±6% in 128-88T, 39±8% in Calu1, and 41±7% in 201T cells. Vorinostat addition to carboplatin or paclitaxel led to significantly greater growth inhibition than chemotherapy alone in all 4 cell lines. Vorinostat (1 µM) synergistically increased the growth inhibitory effects of carboplatin/paclitaxel in 128-88T cells. When colony formation was measured after drug withdrawal, vorinostat significantly increased the effects of carboplatin but not paclitaxel. The % colony formation was: control 100%; 1 µM vorinostat 83% ± 10%; 5 µM carboplatin, 41% ± 11%; carboplatin/vorinostat, 8% ± 4%; 2 nM paclitaxel, 53% ± 11%; paclitaxel/vorinostat 46% ± 21%. In A549 and 128-88T, vorinostat potentiated carboplatin induction of gamma-H2AX (a DNA damage marker) and increased α-tubulin acetylation (a marker for stabilized mictrotubules). In A549, combination of vorinostat with paclitaxel resulted in a synergistic increase in α-tubulin acetylation, which reversed upon drug wash-out. We conclude that vorinostat interacts favorably with carboplatin and paclitaxel in NSCLC cells, which may explain the provocative response observed in our clinical trial. This likely involves a vorinostat-mediated irreversible increase in DNA damage in the case of carboplatin and a reversible increase in microtubule stability in the case of paclitaxel.

Introduction

Histone deacetylases (HDACs) represent an important new therapeutic target in cancer. HDACs remove acetyl groups from core histones and thereby compact chromatin structure and suppress gene expression. Genes that are repressed via HDAC action include cell cycle inhibitors, differentiation factors, and apoptosis inducers^{1, 2}. Several HDAC inhibitors are in pre-clinical and clinical development³. Among these, vorinostat (suberoylanilide hydroxamic acid, SAHA), shows single-agent anti-proliferative activity in a variety of pre-clinical tumor models ^4–7 and is in clinical use for the treatment of cutaneous T-cell lymphoma⁸. Indicative of its potential value in lung cancer, vorinostat significantly inhibits lung tumor development in mice exposed to the tobacco-specific carcinogen 4- (methylnitrosamino)-1- (3-pyridyl)-1-butanone⁹, significantly inhibits the growth of NSCLC cell lines in vitro¹⁰, and significantly suppresses the growth of A549 lung cancer xenografts⁷.

Based on several clinical and pre-clinical considerations, we conducted a phase I trial of orally administered vorinostat in combination with carboplatin/paclitaxel in advanced solid malignancies¹¹. In our study, 10 of 19 advanced NSCLC patients experienced a partial response and 4 had stable disease. The 53% response rate observed in these patients compares favorably with the 20–30% response rate which is observed historically for the carboplatin/paclitaxel regimen^{12, 13}. Based on these promising results and the observed safety of the vorinostat/carboplatin/paclitaxel combination, a randomized, placebo-controlled phase II clinical trial has been initiated for patients with previously untreated advanced stage NSCLC. In parallel, pre-clinical studies were undertaken to determine mechanism for enhancement of carboplatin/paclitaxel by vorinostat and develop potential markers of response that may be used to facilitate patient slection in future clinical trials.

The mechanism of action of HDAC inhibitors was initially ascribed to their effect on histone protein acetylation and the consequent changes in chromatin structure and gene expression¹. However, it is becoming evident that their mechanism of action is more complex, especially with the demonstration that HDAC inhibitors alter the function of both nuclear and cytoplasmic HDACs³, and that HDACs modulate the acetylation status and function of a number of non-histone proteins including p53, α-tubulin, and the molecular chaperone hsp90^{3, 14}.

How might vorinostat enhance the cytotoxicity of carboplatin/paclitaxel in NSCLC cells? Treatment of cells with carboplatin results in the formation of platinum:DNA adducts. The recognition and processing of platinum adducts triggers both p53-dependent and p53-independent signaling pathways that lead to cell cycle arrest and apoptosis induction¹⁵. Vorinostat may increase carboplatin cytotoxicity by relaxing the structure of chromatin and increasing the accessibility of carboplatin to its DNA target. Consistent with this possibility, vorinostat increasesthe cytotoxicity of a variety of anti-cancer drugs that target DNA, including cisplatin¹⁶.

Vorinostat may increase paclitaxel cytotoxicity by inhibiting HDAC6 activity. HDAC6 is a cytoplasmic HDAC that regulates tubulin acetylation¹⁷. Pharmacological inhibition of HDAC6 activity in cultured cells leads to an increase in tubulin acetylation and microtubule stabilization¹⁸. Since paclitaxel binds preferentially to polymerized tubulin¹⁹, vorinostat, by inhibiting HDAC6, may decrease the concentration of paclitaxel required for cytotoxic effects. Studies by Marcus et al. confirm that taxane binding is increased in cells in which HDAC6 is inhibited²⁰. The need for a biologic and mechanistic understanding rather than empiric observation is the rationale for performing the experiments described here so as to elucidate the molecular basis of the positive interaction between vorinostat and carboplatin/paclitaxel obtained in our phase I clinical trial.

Materials and Methods

Cell Lines

The human NSCLC cell lines A549 (bronchioloalveolar), 128.88T (squamous cell carcinoma), 201T (adenocarcinoma), and Calu1 (epidermoid carcinoma) were employed. 128.88T and 201T cells were developed and generously provided by Dr. Jill Siegfried, University of Pittsburgh^{21, 22}. Calu1 and A549 cells were obtained from ATCC (Manassas, VA). The cell lines were maintained in Basal Medium Eagle (BME) (GIBCO, Invitrogen, Carlsbad, CA) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (HyClone Laboratories, Inc., Logan, UT), 2 mM L-glutamine (GIBCO, Invitrogen) and 100 U/ml penicillin-streptomycin (GIBCO, Invitrogen). Cells were incubated at 37°C in a humidified atmosphere containing 5% CO₂.

Chemicals and Reagents

Paclitaxel and carboplatin were purchased from Sigma-Aldrich (Saint Louis, MO USA). Paclitaxel was dissolved in sterile DMSO to create a 50 mg/mL (12 mM) stock solution. Carboplatin was dissolved in sterile H₂O to produce a 10 mg/mL (27 mM) stock solution. Vorinostat (generous provision from Merck & Co., Inc. and the National Cancer Institute) was dissolved in DMSO to produce a 50 mM stock solution. Paclitaxel, carboplatin, and vorinostat stock solutions were divided into individual aliquots and stored at −20°C. Individual aliquots were diluted in complete tissue culture medium immediately prior to use in an experiment.

Antibodies

The antibodies used in these studies included anti-poly (ADP-ribose) polymerase (PARP) (BD Bioscience Pharmingen #556362, San Diego, CA), anti-α-tubulin (Sigma-Aldrich #T6199), anti-acetylated α-tubulin (Sigma-Aldrich #T7451), anti-phospho-H2AX^Ser139 (γ-H2AX) (Millipore Biotechnology, #07–164, Billerica, MA, USA), H2AX (Upstate Biotchenology, Temecula, CA), anti-p21 horseradish peroxidase (HRP) conjugate (Santa Cruz Biotechnology #sc-397 HRP, Santa Cruz, CA), and anti-Bax HRP conjugate (Santa Cruz Biotechnology, #sc- 493). Actin was detected using the anti-actin antibody Ab-1 (Oncogene Research, San Diego, CA). Anti-rabbit and anti-mouse HRP-conjugated secondary antibodies were from Amersham Life Sciences (Arlington Heights, IL) and Promega Corporation (Madison, WI), respectively.

MTT assay

NSCLC cells were plated into 96-well tissue culture plates (2×10³ cells per well) in a volume of 0.1 mL complete growth medium. Cells were allowed to attach overnight at 37°C. The next day, cells were treated with vehicle (DMSO at 0.01%) or increasing concentrations of a chemotherapy drug or drug combination. Each treatment condition was tested in 3–6 replicate wells per experiment. After 72 h, cell growth was assessed using the MTT Cell Proliferation Assay kit (ATCC) according to the manufacturer’s recommendations. Absorbance was measured at 570 nm using a MRX microplate reader (DYNEX technologies, Chantilly, VA). The effect of treatment on cell growth was calculated for each drug or drug combination using the equation: % Cells Remaining = 100[ (O.D. 570 nm _{treatment-blank} /O.D. 570 nm _{control-blank})]. In some experiments, % Growth Inhibition was also calculated using the equation: % Growth Inhibition = 100 - % Cells Remaining. Data analysis was performed using GraphPad Prism 4 software (San Diego, CA). The IC₅₀ concentration for single-agent vorinostat, carboplatin, paclitaxel and their combinations was derived by fitting the dose-growth inhibition data to Hill’s equation [ (growth inhibition)_max* (drug conc.)^H]/[ (growth inhibition)₅₀⁺ (baseline inhibition)^H] using Adapt-it software.

To determine the nature of interaction between vorinostat and carboplatin/paclitaxel, the MTT dose-response data were analyzed using a program written in the MATLAB® environment (©2007, The MathWorks, Inc. Natick, MA). We developed this program as a modification of the Webb’s fraction method. The logistic dose response curve is described in eq. 1:

$$r\left(d\right)=\frac{1-h}{1+{\left(\frac{d}{c}\right)}^b}+h$$ eq.1

h, c, and b are parameters of the equation specific to each drug. h describes the horizontal asymptote, c is the d-value when r (d=c) = (1+h)/2, and , b describes the slope of the line at d=c. The d is the variable that specifies the dose of the drug. The parameters are determined by fitting the equation to the single-drug MTT data. The parameters were obtained by using the MATLAB® function fmincon; the weighted sum of the squares (the sum of the difference between the line and the data points weighted by the standard deviation of the associated data point) is optimized by varying the parameter set that describes the line. Webb describes an additive relationship as the product of the responses of the monotherapy (eq. 2).

$$r_{1,2}\left(d_1,d_2\right)=r_1\left(d_1\right)r_2\left(d_2\right)=\left(\frac{1-h_1}{1+{\left(\frac{d_1}{c_1}\right)}^{b_1}}+h_1\right)\left(\frac{1-h_2}{1+{\left(\frac{d_2}{c_2}\right)}^{b_2}}+h_2\right)$$ eq.2

r_1,2 represents an additive surface. We then compared the experimental value to the predicted value. If the experimental response is greater than the calculated additive response then the combination is antagonistic; if the experimental value is less than the additive response then the combination is synergistic; if they are equal then the treatment is additive as shown in the relationships below (eq. 3)

$$\begin{array}{l}{r}_{1,2}^{\text{exptl}}\left(d_1,d_2\right)<r_{1,2}^{\text{calc},\text{add}}\left(d_1,d_2\right)\Rightarrow \text{synergy}\hfill \\ r_{1,2}^{\text{exptl}}\left(d_1,d_2\right)=r_{1,2}^{\text{calc},\text{add}}\left(d_1,d_2\right)\Rightarrow \text{additivity}\hfill \\ r_{1,2}^{\text{exptl}}\left(d_1,d_2\right)>r_{1,2}^{\text{calc},\text{add}}\left(d_1,d_2\right)\Rightarrow \text{antagonism}\hfill \end{array}$$ eq.3

Trypan blue dye exclusion assay

Trypan blue and colony formation assays were performed with a slight modification of the method described by Chaudhry et al²³. A549 cells were seeded in 100 mm² culture plates at 1–2 × 10⁶ cells per plate. Cells were allowed to attach overnight and thereafter treated with vehicle (0.01% DMSO) or different concentrations of carboplatin and/or paclitaxel in the absence or presence of vorinostat. After 72 h of continuous exposure to treatment agents, cells were harvested using trypsin digestion, washed with PBS and resuspended in fresh tissue culture medium. A 10 µL aliquot of the cell suspension was mixed with 90 µL of 0.4% (v/v) trypan blue. The proportion of viable cell (cells capable of excluding trypan blue stain) for each treatment condition was determined within 5 minutes of staining by manual counting with a hemocytometer. Each determination was done in triplicate. The percent cell viability was determined by normalizing the mean viable cell count for each treatment condition to the mean viable cell count for the vehicle-treated control cells.

Colony formation assay

To determine the ability of viable cells to proliferate post therapy, one hundred viable cells per treatment group (from the trypan blue studies) were seeded per well into 6-well plates (in triplicate). The medium was replaced once weekly until plates were ready for colony count (typically 10 to 14 days after seeding). Cells were fixed with 70% methanol for 5 minutes, twice, followed by staining with 0.1% crystal violet for 5 min. The stain was washed off under running tap water and the plates were allowed to dry. The number of distinctly stained colonies (containing at least 50 cells per colony) was counted using a colony counter (Fisher Scientific) and the results averaged for each treatment group.

Measurement of H2AX phosphorylation by flow cytometry

H2AX phosphorylation was measured using a commercially available assay kit from Upstate Biotechnology. A549 cells (2.0 × 10⁶) were seeded into 100 mm tissue culture dishes in complete tissue culture medium and allowed to attach overnight. The cells were then treated with vehicle, vorinostat (1 µM), carboplatin (10 µM or 40 µM) or the combination for 24 h. The cells were recovered by trypsinization and washed with sterile PBS. The cells were fixed, permeabilized, and stained with either a negative control mouse IgG-FITC conjugate or the anti-phospho-H2AX (Ser¹³⁹)-FITC conjugate exactly as described by the manufacturer. The stained cells were analyzed on a Coulter Epics XL flow cytometer using Expo 32 ADC software (Coulter Corporation). A minimum of 5,000 events were analyzed per treatment group.

Immunoblot analysis

Cells were seeded into T25 flasks in complete growth medium at a density of 0.4 × 10⁵ to 0.6 × 10⁵ cells/ml. Within 48 h of seeding, the culture medium was replaced with fresh complete growth medium containing either vehicle (0.01% DMSO) or chemotherapy agents as indicated in the respective figures. At the designated times, a cell scraper was used to collect cells directly into the treatment medium. Cells were collected by centrifugation (129 × g for 6 min at 4° C) and washed once in ice-cold PBS. The resulting pellets were stored at −80°C. To prepare whole cell extracts (WCE), cell pellets were allowed to thaw on ice. Pellets were then resuspended for 30 min in lysis buffer [1% Triton X-100, 0.1% SDS, 50 mM Tris, and 150 mM NaCl], containing 1× protease inhibitor cocktail (BD Biosciences Pharmingen) and 50× phosphatase inhibitor cocktail (Calbiochem, San Diego, CA). Cell lysates were clarified by centrifugation at 15,000 × g for 10 min at 4°C.

For analysis of γ-H2AX and total H2AX expression, histone extraction was performed with a slight modification of the abcam online protocol²⁴. Harvested cells were washed in ice cold PBS twice and then resuspended in TEB lysis buffer (PBS containing 0.5% Triton X 100 (v/v), 2 mM phenylmethylsulfonyl fluoride, 0.02% (w/v) sodium azide) for 10 min on ice. The resulting lysate was centrifuged at 400 × g for 10 min. The nuclear pellet was resuspended in TEB buffer and centrifuged again. Nuclear proteins were extracted with 0.2N HCL overnight at 4°C. The supernatant was recovered following centrifugation at 400 × g for 10 min. WCE and the acid-extracted nuclear protein were quantitated using the BCA Protein Assay Kit (Pierce Laboratories, Rockford, IL) according to manufacturer’s directions.

Proteins were resolved on Criterion pre-cast SDS-polyacrylamide gels (BIORAD, Hercules, CA) under denaturing conditions and were electrophoretically transferred to PVDF membranes (NEN Life Science Products, Boston, MA). Membranes were blocked for a minimum of 1 h in a 5% w/v solution of nonfat milk in TBST (10 mM Tris, pH 7.6, 150 mM NaCl, and 0.05% Tween 20) followed by incubation with primary antibody for 1 h at room temperature (p21, Bax, H2AX, actin, α-tubulin or acetylated α-tubulin) or overnight at 4°C (PARP or γ-H2AX). The blots were washed in TBST and subsequently incubated with secondary antibody conjugated with HRP for 1 h. The blots were again washed and immunoreactive complexes detected using Renaissance western blot chemiluminescence reagents (NEN Life Science Products). In some cases, densitometry was used to quantify protein expression. Computer images of each immunoblot film were created using an hp scanjet 3970. The resulting images were analyzed using Un-Scan-It gel analysis software (Silk Scientific, Inc., Orem, Utah).

Statistics

Statistical analysis was performed using Graph Pad Prism software statistical package, version 4. Comparisons of cytotoxicity assay results were made using a two-sided Student’s t test, where indicated. A p value of < 0.05 was considered statistically significant.

Results

Vorinostat inhibits the growth of NSCLC cells

To ascertain the growth inhibitory effects of single-agent vorinostat, 4 cell lines derived from different histologic subtypes of NSCLC were cultured for 72 h with vehicle or increasing concentrations of drug. At the conclusion of the incubation period, growth inhibition was measured by MTT assay (Figure 1A). The resulting dose-response data was used to determine the vorinostat concentration which inhibits growth by 50% (IC₅₀) using the Hill’s equation. The IC₅₀ concentrations were 1.94 µM (95% CI: 1.47–2.4 µM) in A549 cells, 1.69 µM (95% CI: 1.52–1.87 µM) in 128-88T cells, 1.29 µM (95% CI: 0.96–1.62 µM) in 201T cells, and 1.21 µM (95% CI: 0.87–1.55 µM) in Calu 1 cells. In our phase I clinical trial, the recommended phase II dose of 400 mg vorinostat results in a maximum serum concentration of 1.81 ± 0.7 µM¹¹. The administration of a clinically achievable (1 µM) concentration of vorinostat resulted in significant growth inhibition in each cell line (Fig. 1B).

Figure 1. Vorinostat inhibits the growth of NSCLC cells.

The indicated NSCLC cell lines were seeded into individual wells (2×10³ cells/well) in 96-well plates in complete growth medium (BME + 10% FBS). The next day, cells received either vehicle (DMSO, 0.01% final concentration) or vorinostat. (A) Vorinostat dose response curves. Each point represents the mean ± SD for six replicate wells within a single experiment. The percentage of cells remaining at each concentration of vorinostat was calculated as described in Methods. Vehicle controls were set at 100% cells remaining in each experiment. (B) Growth inhibition at 1 µM vorinostat. Growth inhibition was measured by MTT assay, as in (A). Bars indicate the mean growth inhibition (±SD) for 2 to 5 independent experiments per cell line.

Vorinostat interacts with carboplatin and/or paclitaxel in NSCLC cells

The 53% partial response rate to vorinostat/carboplatin/paclitaxel combination therapy observed in our phase I trial is higher than the 20–30% historical response rate observed for carboplatin/paclitaxel in patients with advanced NSCLC. This is suggestive of vorinostat enhancement of carboplatin/paclitaxel activity. To experimentally determine the nature of the interaction between vorinostat and carboplatin/paclitaxel in NSCLC cells, a modification of Webb’s fraction method was used to assess combination effects. For the purpose of this analysis, the carboplatin/paclitaxel combination was treated as a single drug. 128-88T cells were treated with vehicle or increasing concentrations of vorinostat alone, carboplatin/paclitaxel, or the combination of vorinostat plus carboplatin/paclitaxel. Growth inhibition was measured after 72 h by MTT assay. The dose response curves for vorinostat alone and the carboplatin/paclitaxel combination are presented in Figs. 2A and 2B, respectively. The response curves obtained for combining vorinostat (at 3 different dose levels) with carboplatin/paclitaxel are presented in Fig. 2C. The interaction between vorinostat and carboplatin/paclitaxel was dose-dependent. It was additive at 0.5 µM vorinostat, synergistic at 1.0 µM vorinostat, and antagonistic at 2.5 µM vorinostat (Fig. 2D).

Figure 2. Evaluation of the interaction between vorinostat and carboplatin/paclitaxel in 128-88T cells.

128-88T cells were seeded into 96-well plates as indicated in Fig. 1. The next day, cells received vehicle alone, vorinostat alone (0–2.5 µM), carboplatin/paclitaxel (at various dose ratios), or vorinostat plus carboplatin/paclitaxel. Carboplatin (CP) and paclitaxel (PT) were treated as a single drug by combining them at a fixed ratio (dose ratio 1= 40 µM carboplatin, 4 nM paclitaxel) and subjecting the mixture to 2-fold serial dilution in complete growth medium. Dose response curves for vorinostat alone and the carboplatin/paclitaxel combination are presented in panels A and B, respectively. Each point represents the mean percentage of cells remaining (± SD) for three replicate wells. (C) Predicted and experimentally determined response curves for the combination of vorinostat plus carboplatin/paclitaxel. Predicted values for an additive interaction between vorinostat and carboplatin/paclitaxel (open circles) were calculated using the following parameters for the vorinostat and carboplatin/paclitaxel logistic dose response curves: vorinostat b = 2.076, c = 2.027, h = 0.000; carboplatin/paclitaxel b = 1.323, c = 0.554, h = 0.210. Open squares indicate the experimentally determined response values. (D) Cross-sectional representation of response curves for the combination of vorinostat with carboplatin/paclitaxel at specified vorinostat concentrations. Dashed lines indicate the expected response if the interaction between the two drugs is additive. Open squares indicate the experimentally derived responses. Points below the line, on the line, or above the line indicate drug synergy, additivity, or antagonism respectively.

The synergy observed between a clinically achievable concentration of vorinostat (1.0 µM) and carboplatin/paclitaxel may be attributable to vorinostat enhancement of carboplatin activity, paclitaxel activity, or both. To determine which chemotherapy drug was affected, we examined the consequences of combining a fixed dose of vorinostat with either carboplatin or paclitaxel in the 4 different NSCLC cell lines. Carboplatin inhibited the growth of each of the cell lines in a dose-dependent manner, and vorinostat addition significantly increased growth inhibition over the entire range of carboplatin concentrations tested (Fig. 3A). In the presence of vorinostat, the IC₅₀ concentration of carboplatin was reduced by 4- to 10-fold. In A549 cells, combination of vorinostat with carboplatin resulted in greater than additive growth inhibition, suggestive of a synergistic drug interaction in this cell line.

Figure 3. Growth inhibition is significantly increased by combination of vorinostat with carboplatin or paclitaxel in NSCLC cells.

(A) The indicated cell lines were seeded into 96-well plates and allowed to attach overnight. The next day, cells were treated with vehicle (control), vorinostat alone (1 µM), carboplatin alone (open circles) or vorinostat plus carboplatin (closed circles). Growth inhibition was measured after 72 h by MTT assay. Vehicle controls were set at 100% cells remaining. Each point represents the mean ± SD for six replicate determinations within a single experiment. The % growth inhibition measured for vorinostat alone within each experiment is presented in the right corner of each panel. The concentration of carboplatin that inhibits growth by 50% (IC50) when administered alone or in combination with vorinostat is presented adjacent to the corresponding dose-response curve. Asterisks indicate vorinostat/carboplatin combinations in which growth inhibition is significantly greater (p < 0.05) than that measured for carboplatin alone. (B) Growth inhibition was measured as in (A) after 72 h of continuous exposure to vehicle (0.01% DMSO), vorinostat (1 µM), paclitaxel (open squares), or vorinostat plus paclitaxel (closed squares). The paclitaxel IC50, when administered alone or in combination with vorinostat, is presented adjacent to the corresponding dose-response curve. Asterisks indicate vorinostat/paclitaxel combinations in which growth inhibition is significantly greater (p < 0.05) than that measured for paclitaxel alone. Similar results were obtained in a second experiment.

Paclitaxel also inhibited the growth of each of the NSCLC cell lines, with IC₅₀ values ranging from 5 – 10 nM (Fig. 3B), consistent with previously published reports for paclitaxel activity in NSCLC cells²⁵. In contrast to carboplatin, the ability of vorinostat to increase paclitaxel activity was dependent on paclitaxel concentration: At paclitaxel concentrations < 8 nM, growth inhibition was significantly increased by the addition of vorinostat but not at concentrations > 8 nM (Fig. 3B). Vorinostat enhancement of paclitaxel activity was also schedule dependent: vorinostat increased growth inhibition in 128-88T cells when added prior to or simultaneous with paclitaxel but had no effect if added after paclitaxel (data not shown).

We also used trypan blue dye exclusion and colony formation assays to assess the effect of vorinostat on the cytotoxicity of carboplatin or paclitaxel in A549 cells. The ability of cells to exclude trypan blue was measured after 72 h of continuous drug exposure. As expected from our MTT assay results, carboplatin mediated a dose-dependent reduction in cell viability. A further significant reduction in viability was achieved by combining carboplatin with vorinostat (Fig. 4A). Vorinostat also increased carboplatin activity in clonogenic assays, in which a fixed number of viable cells were plated after drug withdrawal, and colony formation was measured 10–14 d later (Fig. 4B). The ability of vorinostat to potentiate carboplatin activity even after drug removal (statistically significant at 5 µM carboplatin) suggests that the effects of vorinostat on carboplatin activity are irreversible. A distinctly different behavior was observed in the presence of paclitaxel. As measured using trypan blue dye exclusion, vorinostat significantly increased growth inhibition in the presence of 2 nM paclitaxel but had no effect when combined with paclitaxel at concentrations ≥ 4 nM (Fig. 4C). Although single-agent paclitaxel mediated a dose-dependent reduction in colony formation, colony formation was not further suppressed by vorinostat addition at any of the paclitaxel concentrations tested (Fig 4D). The ability of vorinostat to increase paclitaxel activity upon continuous drug exposure (detected in MTT and trypan blue assays) but not in clonogenic assays (after drug removal) suggests that the effects of vorinostat on paclitaxel activity are reversible.

Figure 4. Effects of vorinostat on carboplatin or paclitaxel activity assessed by trypan blue dye exclusion assay and colony formation assays.

(A, C) For trypan blue assays, cells were exposed to vehicle or the indicated treatment agents. Vorinostat (V) was used at a fixed concentration (1 µM). Carboplatin (C) and paclitaxel (P) were used at the concentrations shown. The percentage of viable cells remaining after 72 h continuous drug exposure was determined as described in Methods. (B, D) Washed, viable cells remaining from the trypan blue assay were seeded into triplicate wells (100 per well) in the absence of further drug treatment to assess colony formation potential. The number of distinct colonies formed after 10–14 days was assessed by crystal violet staining as described in Methods. The number of colonies that formed for control (vehicle treated) cells was set at 100% colonies. No colonies were detected for the 10C/V, 20C, or 20C/V treatment groups. Asterisks denote statistically significant differences (** p < .005; * p < 0.05). Results are representative of two experiments.

Vorinostat increases carboplatin-induced DNA damage in A549 and 128-88T cells

We hypothesized that the ability of vorinostat to mediate chromatin relaxation would allow greater access of carboplatin to the chromosome, leading to increased DNA damage and cytotoxicity. There is a correlation between the amount of phosphorylated H2AX (γ-H2AX) detected in cells and the amount of double strand DNA breaks resulting from the action of DNA damaging agents^{26, 27}. Therefore, to investigate whether the increased cytotoxicity observed with the carboplatin plus vorinostat combination was associated with increased DNA damage, we examined γ-H2AX expression in 128-88T cells by immunoblot. Carboplatin induced γ-H2AX expression, and this was potentiated by vorinostat (Fig. 5A). Quantitative evaluation revealed that vorinostat reduced by approximately 4-fold the amount of carboplatin needed to achieve a given level of γ-H2AX expression. For example, γ-H2AX expression was equivalent in cells treated with 20 µM carboplatin and cells treated with 5 µM carboplatin plus vorinostat. Vorinostat enhancement of γ-H2AX expression occurred without a change in total levels of H2AX protein (data not shown). The ability of vorinostat to augment carboplatin induction of γ-H2AX expression was also observed in A549 cells using a quantitative flow cytometry assay (Table 1). A549 cells were treated with vehicle, vorinostat, carboplatin, or the combination of vorinostat plus carboplatin for 24 h. The cells were fixed, permeabilized, and stained using either a negative control antibody (mouse IgG-FITC) or a FITC-conjugated γ-H2AX antibody. No staining was observed using the negative control (data not shown). However, when the γ-H2AX antibody was used, carboplatin but not vorinostat treatment resulted in an increase in positive cells. The percentage of γ-H2AX-positive cells was synergistically increased by combination of vorinostat with carboplatin. The vorinostat-mediated increase occurred at two different carboplatin dose levels and was observed 24 h post-treatment, prior to any morphological signs of apoptosis.

Figure 5. Identification of molecular markers of vorinostat/carboplatin and vorinostat/paclitaxel combination in NSCLC cells.

(A) 128-88T cells were treated with vehicle (controls), vorinostat (V, 1 µM), carboplatin (C, 2.5 µM to 160 µM) or the combination of vorinostat plus carboplatin. After 48 h, proteins were acid extracted and analyzed by immunoblot for γ-H2AX. Blots were re-probed for total H2AX protein (not shown). The ratio of phosphorylated H2AX to total H2AX protein expression was determined by densitometry of immunoblot films and is presented in the bar graph (open bars: carboplatin; black bars: vorinostat plus carboplatin). (B) A549 cells were treated with vehicle (controls), vorinostat (V, 1 µM), carboplatin (C, 2.5 µM to 20 µM) or the combination of vorinostat plus carboplatin. After 48 h, WCE were prepared and analyzed by immunoblot for PARP, p21, or Bax. Blots were re-probed with α-tubulin or actin as control for protein quantitation and loading. (C) A549 cells were treated with vehicle, vorinostat alone (V, 1 µM), paclitaxel (P, 0.5 nM to 4 nM) or the combination of vorinostat plus paclitaxel. After 48 h, WCE were prepared and analyzed for PARP, p21, acetylated tubulin (ac tub), and total tubulin (tub) protein expression by immunoblot. (D) A549 cells were treated with vehicle, paclitaxel (4 nM), vorinostat (1 µM), or the vorinostat/paclitaxel combination for 24 h. The cells were washed and then left to grow in drug-free medium for the times indicated. Protein lysates were prepared and analyzed for acetylated α-tubulin and total α-tubulin.

Table 1.

H2AX phosphorylation is synergistically increased in A549 cells by combination of vorinostat with carboplatin

	% γ-H2AX positive cells
Treatment	No vorinostat	Plus vorinostat
Vehicle	1.3	1.5
10 µM carboplatin	2.5	22.1
40 µM carboplatin	17.3	39.5

A549 cells were treated with vehicle, vorinostat (1µM), carboplatin (10 µM or 40 µM), or the combination of vorinostat plus carboplatin for 24 h. After treatment, the cells were washed, fixed and stained using either a negative control antibody or a FITC-conjugated antibody specific for γ-H2A.X, as described in Methods. The expression of γ-H2AX was quantitated by flow cytometry. At least 5,000 events per treatment group were analyzed.

To further investigate the effect of vorinostat on carboplatin signaling pathways, A549 (p53 wt) cells were treated with vehicle, vorinostat alone, carboplatin alone, or the combination of vorinostat plus carboplatin for 48 h. Whole cell extracts were prepared and analyzed by immunoblot for PARP cleavage (as an indicator of apoptosis), and p21 and Bax protein expression (as downstream targets of the p53-dependent DNA damage response) (Fig. 5B). We elected to use A549 cells for our initial molecular studies because we observed a greater than additive interaction between carboplatin and vorinostat in these cells in both MTT (Fig. 3A) and H2AX phosphorylation assays (Table 1). Suggestive of vorinostat potentiation of carboplatin-induced apoptosis, there was a modest reduction in the level of full-length PARP protein in A549 cells treated with the vorinostat/carboplatin combination compared to either agent alone. The expression of p21 was induced both by vorinostat and by carboplatin. Generally, the p21 expression level was greater in cells treated with the vorinostat/carboplatin combination compared to carboplatin alone. Across multiple experiments, Bax levels were comparable between cells treated with carboplatin alone and those treated with vorinostat plus carboplatin. Together the above data demonstrates that vorinostat potentiates carboplatin-induced DNA damage and p21 induction.

Vorinostat potentiates paclitaxel-induced apoptosis and reversibly increases α-tubulin acetylation in NSCLC cells

To determine if there is also a molecular interaction between vorinostat and paclitaxel in A549 cells, WCE were prepared after 48 h of treatment with vehicle, paclitaxel alone, vorinostat alone, or the combination of vorinostat plus paclitaxel and analyzed by immunoblot. There was no significant PARP loss when cells were treated with vorinostat (1µM) or paclitaxel as single agents (except with 4 nM paclitaxel). However, combination of vorinostat with 2 nM paclitaxel resulted in greater PARP loss than paclitaxel alone (Figure 5C). Notably, this is the same concentration of paclitaxel at which addition of vorinostat significantly increased growth inhibition in both MTT and trypan blue assays. Apoptosis induction in A549 cells was confirmed using annexin binding assays, where the % annexin-positive cells 24 h post-treatment was: vehicle (1.5%); vorinostat (4.2%); paclitaxel (8.3%); vorinostat plus paclitaxel (16%). Therefore, conditions exist in which apoptosis is increased by combination of vorinostat with paclitaxel.

The cytotoxic effect of paclitaxel results from its ability to stabilize microtubules leading to a G2/M cell cycle arrest, followed by mitotic slippage and eventual apoptosis²⁸. By inhibiting cytoplasmic HDAC6 (an enzyme that catalyzes microtubule deacetylation and disassembly), vorinostat may also promote microtubule stabilization¹⁸. To determine whether vorinostat and paclitaxel co-regulate microtubule stability, which could account for the increased cytotoxicity, A549 cells were treated with vehicle, vorinostat, paclitaxel, or the combination of vorinostat/paclitaxel for 48 h. WCE were prepared and analyzed by immunoblot for acetylated α-tubulin, a surrogate marker for stabilized microtubules (Figure 5C). A synergistic increase in the level of acetylated α-tubulin was observed in cells treated with the combination of vorinostat and paclitaxel. Vorinostat also increased α-tubulin acetylation in 128-88T cells, although a synergistic rise in tubulin acetylation was not observed in these cells upon combination with paclitaxel (Supplemental Figure 1).

Given the reversibility of the interaction between vorinostat and paclitaxel observed in colony formation assays, we hypothesized that tubulin acetylation would also be reversible if modulation of this molecular target plays an essential role in the drug interaction. To test this, A549 cells were treated with vehicle, paclitaxel, vorinostat or the vorinostat/paclitaxel combination for 24 h, after which the cells were washed and allowed to grow in drug-free medium. Cell harvest proceeded at time 0 h, 4 h, 24 h, and 48 h post drug removal. As expected, there was a dramatic increase in acetylated tubulin expression in cells treated with vorinostat or the vorinostat/paclitaxel combination immediately after drug removal at time 0 h. However, as the time post drug wash-out increased, there was a progressive loss of acetylated tubulin. The effect of vorinostat alone was diminished 4 h post drug removal and reached basal (control) levels within 24 h, demonstrating that vorinostat reversibly regulates tubulin acetylation in NSCLC cells (Figure 5D).

We subsequently determined whether the vorinostat-mediated increases in γ-H2AX expression and α-tubulin acetylation we observed (using only the two-drug combinations vorinostat/carboplatin or vorinostat/paclitaxel) persisted in cells treated with the vorinostat/paclitaxel/carboplatin combination that was employed in our phase I clinical trial. Positive results would support evaluation of these proteins as potential biomarkers of drug activity. 128-88T cells were exposed continuously to vehicle, each single agent, each two-drug combination, or the combination of vorinostat/carboplatin/paclitaxel for various times. Protein lysates were prepared and analyzed by immunoblot for changes in expression of PARP, γ-H2AX, or acetylated α-tubulin. The expression of γ-H2AX was not distinctly different between vehicle and drug-treated cells at 4 h and 24 h (data not shown). Beginning at 48 h post-treatment, H2AX phosphorylation increased when carboplatin was combined with vorinostat. The potentiating effect of vorinostat on carboplatin-induced H2AX activation persisted in cells treated with the 3-drug combination (Fig. 6). Notably, H2AX phosphorylation increased in vorinostat/carboplatin-treated cells prior to the onset of apoptosis, which was detected 72 h post-treatment by loss of full-length PARP and the appearance of the 89 kDa PARP cleavage fragment (Fig. 6). Cells treated with 1 nM paclitaxel alone showed no induction of acetylated α-tubulin, whereas vorinostat treatment increased acetylated α-tubulin expression at all time points examined. The level of acetylated α-tubulin was generally similar in cells treated with vorinostat, vorinostat/paclitaxel, vorinostat/carboplatin, or the vorinostat/carboplatin/paclitaxel combination (Fig. 6). At the drug doses and time points examined in this study, neither vorinostat alone (1µM), paclitaxel alone (1 nM), nor the vorinostat/paclitaxel combination induced apoptosis in 128-88T cells.

Figure 6. Changes in γ-H2AX expression and tubulin acetylation persist in 128-88T cells treated with vorinostat/carboplatin/paclitaxel.

(A) 128-88T cells were continuously exposed to vehicle, vorinostat (V, 1 µM), carboplatin (C, 10 µM), paclitaxel (P, 1 nM), or the indicated two-drug or three-drug combinations. Protein lysates prepared from cells harvested at 48 h and 72 h post-treatment were analyzed by immunoblot for PARP, γ-H2AX, total H2AX, acetylated α-tubulin, and total α-tubulin. PARP_SS: In order to show the 89kD PARP cleavage product (black arrow) the PARP blot was re-developed using SuperSignal™ detection reagent from Pierce. Densitometry was used to quantify the expression of each protein under each treatment condition. Quantitative data are presented as the ratio of PARP/total α-tubulin, γ-H2AX/total H2AX, and acetylated α-tubulin/total α-tubulin.

Discussion

These in vitro studies were conducted in order to explain the apparent vorinostat-mediated enhancement of the clinical activity of carboplatin/paclitaxel in NSCLC observed in our recent phase I clinical trial¹¹. We demonstrate that combination of a clinically achievable concentration of vorinostat with carboplatin/paclitaxel results in synergistic growth inhibition in 128-88T lung cancer cells in vitro. Furthermore, we found that addition of vorinostat to either carboplatin or paclitaxel resulted in a dramatic decrease in their IC₅₀ concentrations (4–10 fold decrease for carboplatin and 2–5 fold decrease for paclitaxel) and resulted in significantly greater growth inhibition than either chemotherapy agent alone in 4 different NSCLC cell lines. We also determined that vorinostat increases carboplatin-induced H2AX phosphorylation (a marker of DNA damage) and increases α-tubulin acetylation (a marker of stabilized microtubules) in 2 different NSCLC cell lines. We conclude from these cell and molecular assays that vorinostat can augment the activity of both carboplatin and paclitaxel in NSCLC.

The ability of vorinostat to increase the cytotoxicity of paclitaxel in NSCLC cells is consistent with prior results obtained using ovarian cancer and thyroid cancer models^{29, 30}. We propose that vorinostat increases paclitaxel cytotoxicity by promoting microtubule stabilization (as detected by increased α-tubulin acetylation) (Fig. 5B). Stabilized (polymerized) microtubules are expected to be a better substrate for paclitaxel binding¹⁹. Interestingly, we found that the ability of vorinostat to interact favorably with paclitaxel depends upon the concentration of paclitaxel studied. For example, at a final concentration of 2 nM paclitaxel, the addition of vorinostat results in a statistically significant increase in growth inhibition in MTT assays and trypan blue assays and increased PARP cleavage (Fig. 3B, Fig. 4C and Fig. 5C). However, at a final concentration of 4 nM paclitaxel, vorinostat has no consistent beneficial effect. Vorinostat may potentiate paclitaxel activity at lower concentrations but not at higher concentrations of paclitaxel if at the higher concentrations, the mechanism of paclitaxel- induced toxicity is no longer based on its ability to modulate microtubule dynamics. In support of this hypothesis, Torres et al. showed that the mechanism -by which paclitaxel induces cell death is concentration dependent. At low concentrations, paclitaxel induces cell death primarily through its microtubule stabilizing effect leading to G2/M arrest and p21 and p53 protein induction. In contrast, at higher concentrations, paclitaxel induces cell death via a Raf-1 dependent pathway³¹.

Support for the idea that vorinostat modification of microtubule dynamics is important for its favorable interaction with low concentration of paclitaxel is provided by prior studies which showed that combination of HDAC inhibitors, PXD101 and Trichostatin A, with docetaxel and paclitaxel respectively, led to enhanced tubulin acetylation, which correlated with increased cytotoxicity in vitro and in vivo^{32, 33}. We extend these findings by showing that acetylated α-tubulin expression induced by combination of vorinostat with paclitaxel is rapidly reversible (and correlates with a reversible effect on cell growth). This finding may be particularly relevant in the clinical setting where vorinostat is undergoing extensive evaluation but with different doses and schedules of administration. Perhaps, the optimal dosing strategy will be one that ensures prolonged duration of exposure rather than the highest plasma concentration when vorinostat is combined with microtubule targeting agents.

The formation of intrastrand DNA cross-links induced by carboplatin is believed to be central to its cytotoxicity³⁴. Although carboplatin does not directly induce double-strand DNA breaks, such breaks may form as a downstream consequence of nucleotide excision repair processes or by the collapse of stalled replication forks³⁵. Using phosphorylation of histone H2AX protein as a surrogate marker to detect these lesions^{26, 27}, we observed an increase in DNA damage when vorinostat was combined with carboplatin in 2 different cell lines (Fig. 5A, Fig. 6 and Table 1). H2AX phosphorylation may increase because vorinostat relaxes the chromatin structure, which facilitates access of carboplatin to the DNA resulting in the generation of more lesions. Alternatively, HDAC inhibition may prevent the repair of a fixed number of carboplatin-induced DNA lesions (lesion persistence). In our hands, repeated attempts to directly quantify DNA-platinum adducts with flameless atomic absorption spectroscopy were unsuccessful, perhaps because of the weak platinating activity of carboplatin, which is about 10- to 100 fold less than cisplatin^{36, 37}. Although our study is unable to clearly distinguish between these two possibilities, HDAC inhibitors were shown previously to influence lesion persistence. Compared to irradiation alone, exposure to vorinostat and other HDAC inhibitors prolonged the duration of γ-H2AX foci in irradiated prostate, melanoma and NSCLC cell lines^38–40. This was associated with a reduction in DNA repair enzyme activity and reduced non-homologous end joining repair³⁹. Although we are the first to demonstrate that vorinostat increases carboplatin induction of γ-H2AX in NSCLC cells, Qian et al. showed that the combination of PXD101, an HDAC inhibitor, with carboplatin resulted in increased expression of γ-H2AX in ovarian cancer cells³².

We conclude that vorinostat interacts favorably with both carboplatin and paclitaxel in NSCLC cells. The molecular basis of this interaction probably involves an irreversible increase in DNA damage in the case of carboplatin as reflected by increased expression of γ-H2AX, while a reversible increase in acetylation of non-histone proteins such as tubulin may be central to the interaction with paclitaxel. We speculate that the ability of vorinostat to interact with both chemotherapy drugs may explain the provocative response rate observed in advanced NSCLC patients enrolled in our phase I clinical trial of vorinostat/carboplatin/paclitaxel. We next plan to analyze clinical specimens in order to determine whether a direct relationship exists between vorinostat modulation of H2AX phosphorylation and α-tubulin acetylation and vorinostat enhancement of chemotherapy responses.

Abbreviations

BME

Basal Medium Eagles

BSA

bovine serum albumin

DMSO

histone deacetylase

HDAC

horseradish peroxidase

HRP

(3- (4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, MTT

NSCLC

non-small cell lung cancer

PARP

poly (ADP-ribose) polymerase

WCE

whole cell extracts