Additive effects of vorinostat and MLN8237 in pediatric leukemia, medulloblastoma, and neuroblastoma cell lines

Summary

Purpose

Histone deacetylase (HDAC) inhibitors, such as vorinostat, decrease Aurora kinase activity by a variety of mechanisms. Vorinostat and MLN8237, a selective Aurora A kinase inhibitor, disrupt the spindle assembly and the mitotic checkpoint at different points, suggesting that the combination could have increased antitumor activity. The purpose of this study was to determine the cytotoxicity of vorinostat and MLN8237 in pediatric tumor cell lines.

Methods

Cell survival was measured after 72 h of drug treatment using a modified methyl tetrazolium assay. For drug combination experiments, cells were exposed to medium alone (controls), single drug alone, or to different concentrations of the combination of the two drugs, for a total of 36 concentration pairs per plate. The interaction of the drug combination was analyzed using the universal response surface approach.

Results

The cells express the target of MLN8237, Aurora A. For each cell line, the single agent IC₅₀ for MLN8237 and for vorinostat was in the clinically relevant range. Both drugs inhibited cell survival in a concentration-dependent fashion. At concentrations of MLN8237 exceeding approximately 1 μM, there was a paradoxical increase in viability signal in all three lines that may be explained by inhibition of Aurora B kinase. The combination of MLN8237 and vorinostat showed additive cytotoxicity in all three cell lines and nearly abrogated the paradoxical increase in survival noted at high single-agent MLN8237 concentrations.

Conclusion

MLN8237 and vorinostat are active in vitro against cancer cell lines. These results provide important preclinical support for the development of future clinical studies of MLN8237and vorinostat.

Introduction

Despite major progress in the survival of children with pediatric malignancies, refractory and relapsed cancer remains a therapeutic challenge. Development of agents with new mechanisms of action is a critical strategy to overcome chemoresistance. Vorinostat (suberoylanilide hydroxamic acid, SAHA, Zolinza®) is a pan-histone deacetylase (HDAC) inhibitor that inhibits class I and class II enzymes [1]. HDAC inhibitors induce cell differentiation, cell cycle arrest, and apoptosis, and inhibit migration, invasion and angiogenesis in many cancer models [2, 3]. Treatment with HDAC inhibitors results in an acetylated form of chromatin that is associated with active gene expression [4, 5]. HDAC inhibitors also directly interfere with mitotic progression, probably through inhibition of HDAC3 activity [6, 7]. Numerous transcription factors involved in controlling oncogenic processes show increased acetylation in response to HDAC inhibitors [1, 5]. The mechanism involved in the antineoplastic effect of vorinostat and other HDAC inhibitors is not well understood.

In the Pediatric Preclinical Testing Program (PPTP), vorinostat inhibited growth of all tested cell lines including leukemia, lymphoma, neuroblastoma, rhabdomyosarcoma, rhabdoid tumor, Ewing sarcoma, and glioblastoma [8]. However, the drug concentrations that were active in vitro were generally beyond the clinically achievable levels (1–2 μM) [8]. Similarly, in vivo, vorinostat induced differences in event free survival (EFS) in pediatric solid tumor xeno-grafts, but no objective responses were observed [8]. Although vorinostat may not be effective as a single agent in pediatric solid tumors, HDAC inhibitors do appear to have the potential to exert additive or synergistic effects if other pathways are also inhibited. Additionally, vorinostat has been well-tolerated in children, with a maximum tolerated dose of 230 mg/m²/day for 4 days every 3 weeks as a single agent [9].

MLN8237 is a reversible small molecule inhibitor of the Aurora A serine/threonine kinase [10]. The Aurora family of kinases includes Aurora A, B, and C. Aurora A and B are expressed in all actively dividing cells, while Aurora C appears to be restricted to testicular tissue [11]. Aurora A kinase is critical in centrosome duplication, bipolar spindle assembly, and cell entry into mitosis [11]. The gene encoding Aurora A is located on chromosome 20q13.2, which is frequently amplified in tumors, and overexpression of Aurora A kinase results in the transformation of normal cells, supporting the role of Aurora A as an oncogene [12–14]. Treatment of tumor cells with Aurora A kinase inhibitors results in centrosome maturation and separation, G2-M transition, formation of mitotic spindle poles and spindles, and alignment of chromosomes to the metaphase plate [15–19]. Aurora B kinase inhibition is associated with an accumulation of tetraploid cells as cytokinesis, but not cell cycle progression, is inhibited [12]. In the PPTP, MLN8237 showed the highest in vitro activity in leukemia cell lines [20]. In vivo, MLN8237 induced significant differences in EFS in 80 % of solid tumor models and all six leukemia models. The in vivo activity observed against the neuroblastoma panel exceeded that observed for conventional chemotherapy [20]. MLN8237 is currently being studied in a phase I/II trial by the Children's Oncology Group; the recommended phase 2 dose in children has been established as 80 mg/m²/dose once daily for 7 of 21 days.

HDAC inhibitors interact with Aurora kinases on a number of levels. They have been shown to decrease expression of Aurora A and Aurora B kinases in lung cancer cells [21, 22]. Additionally, by inducing acetylation of the chaperone heat shock protein 90 (hsp90), HDAC inhibitors disrupt the association of hsp90 with Aurora kinases and deplete Aurora kinase activity [23]. HDAC inhibitors inactivate the spindle assembly checkpoint (SAC) in a transcription-independent fashion. This inhibition permits cells to enter anaphase prematurely, resulting in accumulation of aneuploid cells that eventually undergo apoptosis [24]. It is unclear whether Aurora A has a direct or indirect role in maintaining the SAC. If the intersection of HDAC inhibitors and Aurora A kinase inhibitors occurs at mitotsis, then the combination of HDAC inhibitors with Aurora kinase inhibitors might disrupt the SAC at two different points, resulting in mitotic catastrophe and synergistic activity.

Methods

Drugs

Vorinostat (SAHA, suberoylanilide hydroxamic acid) was obtained from Cayman Chemical (Ann Arbor, MI) and MLN8237 was obtained from Millennium Pharmaceuticals, Inc. (Cambridge, MA). Vorinostat was dissolved in dimethyl sulfoxide (DMSO) to an initial concentration of 75.7 mM. MLN8237 was dissolved in DMSO to an initial concentration of 10 mM. Each drug was then diluted in phosphate buffered saline to the final concentrations used in each experiment.

Cell line and culture

Three human pediatric cell lines were used in the experiments: IMR-32, a neuroblastoma line; MOLT-4, a T-cell leukemia line; and Daoy, a medulloblastoma line. MOLT-4 and Daoy were cultured in RPMI-1640 medium with L-glutamine (Invitrogen Corp., Grand Island, NY) with heat inactivated 10 % fetal bovine serum (Invitrogen Corp.) in 5 % CO₂ at 37°C. IMR-32 was cultured using alpha-MEM medium (Invitrogen Corp.) with heat inactivated 10 % fetal bovine serum (Invitrogen Corp.) in 5 % CO₂ at 37°C.

Cytotoxicity assays

A modified methyl tetrazolium assay (MTT; 3-(4,5 dimethyl-2-thiazolyl)-2,5-diphenyltetrazolium bromide) was used to determine the sensitivity of the cell lines to vorinostat and MLN8237 [25, 26]. For single drug experiments, 135 μl of 1×10⁵ cells/ml of each cell line was plated into 96-well microtiter plates using an automated drug delivery system (Biomek Automated Laboratory Workstation, Beckman Coulter, Inc., Fuller, CA). All cells were growing exponentially at the time of drug addition. The effect of evaporation in the wells was minimized by using only the inner wells for analyses. Twenty-four hours after plating, a single drug at specified concentrations was added to each well. Replicates of six wells were used at each concentration. Two replicate plates were used for each experiment. After 72 h of drug exposure, 15 μl of MTT (5 mg/ml) was added to each well, the plates were incubated for 4 h at 37°C, the medium was replaced with 150 μl of DMSO to solubilize the formazan precipitate and the optical density (OD) was measured at 550 nm using a microplate spectro-photometer (Anthos Labtec Instruments, Salzberg, Austria). Cell survival was calculated by subtracting the background OD of media alone and then dividing the OD of test wells by the OD of the control (untreated) wells.

For drug combinations, similar experiments were performed except that in each 96-well plate, cells were exposed to medium alone (controls), single drugs alone (vorinostat only, MLN8237 only) or varying concentrations of the combination of two drugs (vorinostat and MLN8237), for a total of 36 concentration pairs per plate. All combination drug concentrations were tested in four replicate plates.

Analysis of drug effects

The average percent cell survivals for each drug alone were calculated and plotted against the drug concentrations. The estimated concentrations at which cell growth was inhibited by 10 %, 25 %, 75 %, and 90 % (i.e., the IC₁₀, IC₂₅, etc.) from the single drug experiments were used to choose the range of concentrations of each drug for the combination experiments. The universal response surface approach (URSA) [27] as implemented in ADAPT II [28] was used to analyze the cytotoxic effects of the combination of vorinostat and MLN8237. The following equation was fitted to experimental data using nonlinear regression:

$$\begin{array}{cc}\hfill 1=& \frac{\mathit{Da}}{{\mathit{IC}}_{50a}{\left(\frac{E}{100-E}\right)}^{\frac{1}{\mathit{ma}}}}+\frac{\mathit{Db}}{{\mathit{IC}}_{50b}{\left(\frac{E}{100-E}\right)}^{\frac{1}{\mathit{mb}}}}\hfill \\ \hfill & +\frac{\alpha \cdot \mathit{Da}\cdot \mathit{Db}}{{\mathit{IC}}_{50a}\cdot {\mathit{IC}}_{50b}\cdot {\left(\frac{E}{100-E}\right)}^{\frac{1}{2\mathit{ma}}}\cdot {\left(\frac{E}{100-E}\right)}^{\frac{1}{2\mathit{mb}}}}\hfill \end{array}$$

where Da is the concentration of drug a, Db is the concentration of drug b, IC₅₀ is the median effective drug concentration, E is the measured effect (fraction of cells surviving), and ma or mb is the slope parameter of the individual drug's concentration-effect curve [29]. When α is positive, Loewe synergy is indicated, whereas a negative α reflects Loewe antagonism. The interaction is considered additive if the 95 % confidence interval (CI) for α includes zero [27, 29].

Immunoblotting

Pediatric cancer cells (IMR-32, MOLT-4, Daoy) were exposed to MLN8237 (0.5 μM) or vorinostat (1 μM) for 6 h. Cells treated with vorinostat were lysed in M-PER Mammalian Protein Extraction Reagent (Thermo Scientific, Rockford, IL), rinsed with phosphate buffered saline three times, and core histones were extracted (Supplementary Material 1). Cells treated with MLN8237 were lysed with Aurora A lysis buffer (see Supplemental Material 2). Proteins or core histones were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred onto a nitrocellulose membrane (Invitrogen, Carlsbad, CA), and blocked in Odyssey blocking buffer (Li-cor, Lincoln, NE) for 1 h at room temperature. Immunoblots were incubated overnight (4°C) with monoclonal anti-Aurora A (1:500; Cell Signaling, Danvers, MA), anti-phospho-histone H3 (Ser10) (1:1,000; Cell Signaling), β-actin (1:10,000; Sigma-Aldrich, St. Louis, MO), anti-acetylH4 (1:1,000; EMD Millipore, Billerica, MA), anti-acetylH3 (1:1,000; Millipore) or anti-histone H3 (1:2,000; Abcam, Cambridge, MA) antibodies diluted in Odyssey blocking buffer with 0.1 % Tween-20. Bound primary antibodies were incubated for 1 h at room temperature with IRDye800 conjugated affinity purified anti-rabbit or anti-mouse secondary antibody (1:2,000; Rockland, Gilbertsville, PA), and the signal was visualized on an Odyssey infrared imaging system. Relative band density was analyzed using ImageQuant software (GE Healthcare, Uppsala, Sweden). Aurora A and phospho-histone H3 levels were normalized to β-actin levels. AcetylH4 and acetylH3 levels were normalized to total histone H3 levels.

Results

Expression of target proteins

All three cell lines express Aurora A kinase protein (Fig. 1). MLN8237 does not inhibit histone H3 phosphorylation at ser10 in Daoy, IMR-32, or MOLT-4 cells, indicating active Aurora B kinase in the MLN8237-treated cells (Fig. 2). Histone acetylation increased after treatment with vorinostat (Fig. 3a and b).

Fig. 1.

Expression of Aurora A kinase protein in pediatric cancer cells. Total protein extracts (30 μg) from untreated Daoy, IMR-32, and MOLT-4 cells were analyzed by Western immunoblot. β-actin was used as loading control

Fig. 2.

Effect of MLN8237 on Aurora B kinase. Cells were treated with 0.5 μM MLN8237 for 24 h, and total protein extracts (30 μg) were analyzed by Western immunoblot. β-actin was used as loading control

Fig. 3.

Vorinostat increases histone acetylation. a Cells were treated with 1 μM vorinostat and core histone extracts were analyzed by Western immunoblot. Total H3 was used as loading control. b Quantification of histone H3 and H4 acetylation levels compared to total H3 loading control

Single drug activity

The single toxicity curves for vorinostat and MLN8237 for the leukemia, medulloblastoma, and neuroblastoma cell lines are illustrated in Figs. 4 and 5. The IC₅₀s of vorinostat for the various cell lines were 0.5 μM for MOLT-4, 0.7 μM for IMR-32, and 2.7 μM for Daoy. MLN8237 demonstrated a narrow range of IC₅₀s; 0.02 μM for MOLT-4, 0.03 μM for IMR-32, and 0.04 μM for Daoy. There was no precipitation of vorinostat or MLN8237 at the highest concentration. Both vorinostat and MLN8237 inhibited cell survival in a concentration-dependent manner. Treatment with vorinostat resulted in near complete inhibition of cell growth in MOLT-4 and IMR-32 cells at concentrations >1 μM. In the Daoy medulloblastoma cell line, treatment with vorinostat concentrations of 10 μM were needed to completely inhibit cell growth. Maximum inhibition of cell growth in all three lines occurred with 0.05–0.5 μM MLN8237. However, at concentrations of MLN8237 exceeding approximately 1 μM, there was a paradoxical increase in apparent survival in all three lines, most pronounced for Daoy.

Fig. 4.

Cytotoxicity of vorinostat against Daoy (Black diamond), IMR-32 (Black square) and MOLT-4 (Black circle) cell lines

Fig. 5.

Cytotoxicity of MLN8237 against Daoy (Black diamond), IMR-32 (Black square) and MOLT-4 (Black circle) cell lines

Cytotoxicity of drug combinations

The concentration-effect curves for leukemia (Fig. 6), medulloblastoma (Fig. 7), and neuroblastoma (Fig. 8) cell lines are illustrated with the different concentrations of vorinostat and MLN8237. Table 1 shows the results of the synergy analysis. Vorinostat and MLN8237 are additive (α = 0) in the leukemia cell line MOLT-4, medulloblastoma cell line Daoy, and neuroblastoma cell line IMR-32. Antagonism was not observed in any cell line. The paradoxical increase in cell survival seen with MLN8237 was not seen with the combination.

Fig. 6.

Cytotoxicity of MLN8237+Vorinostat on MOLT-4 Leukemia Cells

Fig. 7.

Cytotoxicity of MLN8237+Vorinostat on Daoy Medulloblastoma Cells

Fig. 8.

Cytotoxicity of MLN8237+Vorinostat on IMR-32 Neuroblastoma Cells

Table 1.

Cytotoxic effects of the combination of vorinostat and MLN8237

Cell line	Drug	IC50 (μM)	m	alpha	95 % CI
MOLT-4	Vorinostat	0.59	1.59	0.001	(−0.26, 0.26)
	MLN8237	0.02	5.48
Daoy	Vorinostat	1.45	1.75	0.002	(−0.52, 0.53)
	MLN8237	<0.01	0.18
IMR-32	Vorinostat	0.62	1.96	0.003	(−0.24, 0.25)
	MLN8237	0.02	3.28

IC₅₀: the concentration at which cell survival is inhibited by 50 %; m: the slope parameter that indicates sigmoidicity of dose-response curve; 95 % CI: the 95 % confidence interval around estimate of alpha

Discussion

Although the PPTP published preclinical data on vorinostat and MLN8237 as single agents [8, 20], no information is available about the combination of vorinostat and MLN8237 in pediatric tumor models. Our experiments demonstrate that Aurora A kinase, the target of MLN8237, is expressed in pediatric medulloblastoma, neuroblastoma, and leukemia cells. MLN8237 is a selective inhibitor of Aurora A kinase. Aurora B kinase activity remains intact, as shown by the increased phosphorylation of histone H3 at Ser10 in treated cell lines. Additionally, treatment of the cells with vorinostat resulted in the expected increase in histone acetylation, which is the chromatin conformation associated with active gene transcription. Our experiments show that the cytotoxic effects of the combination of vorinostat and MLN8237 are additive for MOLT-4, Daoy, and IMR-32. For MLN8237, the single agent IC₅₀s in medulloblastoma, neuroblastoma, and leukemia (<0.04 μM) are well below the average steady state concentrations achieved in adults (~1.7 μM) [30] and children (3.7 μM) [31] at the recommended phase 2 dose of MLN8237. For vorinostat, the single-agent IC₅₀s were in or near the clinically achievable range (1–2 μM) [32].

MLN8237 at high concentrations loses selectivity for Aurora A kinase and inhibits Aurora B kinase as well [10]. Therefore, the apparent paradoxical increase in cell survival at high MLN8237 concentrations in our studies may actually result from “off target” Aurora B inhibition [33]. Aurora B inhibition leads to a defective spindle assembly checkpoint, resulting in accelerated mitotic slippage. The G1 tetraploids that derive subsequent to mitotic slippage can remain viable, and depending on various underlying genetic factors, may undergo additional rounds of DNA synthesis leading to endoreduplication. These tetraploid cells could remain metabolically active during the period of time assayed in the experiments performed here and, therefore, could appear viable using the MTT assay. Interestingly, the addition of vorinostat abrogated this effect suggesting that vorinostat may be a beneficial combination partner for either Aurora A or Aurora B kinase inhibitors.

The dose limiting toxicities of vorinostat in children were neutropenia, thrombocytopenia, and hypokalemia [9], while the dose limiting toxicities of MLN8237 included mucositis, mood alteration, neutropenia, and thrombocytopenia [31]. Thus the combination of vorinostat and MLN8237 may results in additive myelosuppression. The data from this study will be helpful in designing the dose escalation plan for future clinical trials.

In summary, we have shown that the combination of vorinostat and MLN8237 are additive in vitro against a variety of pediatric malignancies at concentrations that are readily obtained after single agent administration in the clinical setting. In light of the high activity that was seen for MLN8237 by the PPTP, these results provide important preclinical support for the development of future clinical studies combining vorinostat and MLN8237, and we are currently developing a phase 1 trial of this combination in children.