Abstract
Histone deacetylase inhibitors (HDACi) are compounds that target the epigenome and cause tumor cell-selective apoptosis. A large number of these agents that have different chemical structures and can target multiple HDACs are being testing in clinical trials and vorinostat is now an approved drug for the treatment of cutaneous T-cell lymphoma. Although these agents are showing promise for the treatment of hematologic malignancies, it is possible that different drugs may have different mechanistic, biological, and therapeutic activities. When comparing an HDACi belonging to the hydroxamic acid class of compounds (vorinostat) with a cyclic tetrapeptide (romidepsin), we showed that these agents regulate the expression of a common set of cellular genes, but certain genes specifically responded to each agent. Using the Eμ-myc mouse model of B-cell lymphoma, we showed previously that overexpression of the prosurvival proteins Bcl-2 and Bcl-XL inhibited the apoptotic and therapeutic activities of the vorinostat. Herein, we compared and contrasted the apoptotic-inducing activities of the hydroxamic acid oxamflatin with romidepsin. Like vorinostat, oxamflatin was unable to kill lymphomas overexpressing Bcl-2 and Bcl-XL, indicating that these proteins can generally protect cells against this class of HDACi. In contrast, romidepsin was able to induce apoptosis in lymphomas overexpressing Bcl-2 with delayed kinetics of cell death and could mediate therapeutic responses against these lymphomas. However, romidepsin was inactive when Bcl-XL was overexpressed. These data provide strong support that HDACi of different chemical classes may have subtle yet potentially important differences in their molecular and biological activities. [Mol Cancer Ther 2008;7(5):1066–79]
- histone deacetylase inhibitor
- romidepsin
- apoptosis
- Bcl-2 family therapy
Introduction
Histone deacetylase inhibitors (HDACi) are an exciting new class of anticancer agents that are showing great promise in clinical trials especially for the treatment of hematologic malignancies (1–3). At least 12 different HDACi have been or are presently being tested in clinical trials (2, 3). In general, these agents are well tolerated, `and even in early phase trials, clinical responses have been observed in patients with leukemias and lymphomas, and even some solid tumors such as mesothelioma, prostate carcinoma, and non-small cell lung carcinoma have responded to treatment (2, 3). HDACi can elicit a range of biological responses including induction of cell death via apoptosis, autophagy, or necrosis; inhibition of cell proliferation by activating cell cycle progression at G1-S or S-G2-M checkpoints; induction of cellular differentiation; modulation of immune responses; and suppression of angiogenesis (1). It is possible that more than one of these effects may be required for the observed therapeutic responses to HDACi. The canonical view of HDACi activity was that these agents inhibited the activity of class I (HDACs 1-3 and 8), class II (HDACs 4-7, 9, and 10), and class IV (HDAC11) HDACs causing chromatin remodeling and altered gene expression, resulting in the various biological effects that are deleterious to tumor cell growth and survival (4). However, it was soon realized that the HDACs can target substrates other than histones and that the post-translational modification of other cellular proteins via acetylation may indeed play an important role in the biological activities of HDACi (5).
A range of naturally occurring or synthetic HDACi have been tested for their anticancer activities in preclinical studies. Six broad classes of HDACi have been defined based on the chemical structures (1). These include the short-chain fatty acids (butyrate and valproic acid), hydroxamates [suberoylanilide hydroxamic acid (vorinostat), trichostatin A, LBH589, oxamflatin, and PXD101], benzamides (MS-275, CI-994, and MGCD-0103), cyclic tetrapeptides [depsipeptide (romidepsin), trapoxin A, and apicidin], electrophilic ketones (trifluoromethylketone), and miscellaneous (depudecin). It is apparent that not all HDACi have the same specificity or affinity for the 11 different target HDACs (1). Accordingly, some HDACi such as vorinostat, LBH589, and PXD101 are broadly acting and likely inhibit most if not all class I, II, and IV HDACs (1). By comparison, other HDACi are more selective inhibitors. For example, romidepsin is reported to more potently inhibit HDACs 1 and 2 than HDACs 4 and 6 (6), whereas tubacin selectively inhibits HDAC6 and does not mediate histone hyperacetylation (7).
It is not yet known if targeting a single specific HDAC is sufficient to mediate a given antitumor response and/or if this will depend on the cell type. Moreover, as crystal structures for all mammalian class I, II, and IV HDACS are not available, molecular modeling to predict the selective inhibitory activity of a given compound against target HDACs has not been possible. As such, empirical screens are necessary at this stage to compare and contrast the activities of different HDACi. We have established a model system using primary B-cell lymphomas from Eμ-myc transgenic mice to identify the proteins and pathways necessary for anticancer agents to induce apoptosis and elicit therapeutic responses (8). Using this system, we determined that the HDACi vorinostat mediated apoptosis of Eμ-myc lymphomas via the intrinsic apoptotic pathway and used the BH3-only proteins Bim and Bid to mediate this effect (8). Moreover, we showed that overexpression of the prosurvival protein Bcl-2, which inhibits activation of the intrinsic apoptotic pathway, completely suppressed the apoptotic and therapeutic activities of vorinostat. Herein, we utilize this system to compare and contrast the activities of structurally diverse HDACi. Consistent with our previously published results using vorinostat (8), we showed that overexpression of Bcl-2 inhibited the apoptotic activity of the hydroxamate-based HDACi oxamflatin. In contrast, we found that the cyclic tetrapeptide romidepsin had the unique ability to kill Eμ-myc lymphomas that overexpress Bcl-2, although with extended kinetics compared with that seen with Eμ-myc lymphomas. Moreover, unlike vorinostat, romidepsin was able to induce a therapeutic benefit against Eμ-myc lymphomas overexpressing Bcl-2. However, we also determined that the expression of endogenous Bcl-XL could determine relative resistance of Eμ-myc/Bcl-2 lymphomas to romidepsin, and romidepsin was unable to kill Eμ-myc cells that were transduced to overexpress Bcl-XL. We conclude that HDACi that belong to different structural classes can have subtle yet potentially important differences in activity that may be exploited clinically to stratify patients for treatment using these different agents.
Materials and Methods
Eμ-myc Lymphomas, Cell Culture, and Reagents
Eμ-myc, Eμ-myc/Bcl-2, and Eμ-myc/Bcl-XL lymphomas were developed as described previously (8) and cultured in six-well plates (Greiner Bio-One) in high-glucose DMEM supplemented with 10% FCS, penicillin/streptomycin, 0.1 mmol/L l-asparagine, and 50 μmol/L 2-mercaptoethanol. HDACi were dissolved in DMSO for the preparation of stock solutions (10 mmol/L). Romidepsin was kindly provided by Gloucester Pharmaceuticals and oxamflatin was produced by Dr. Anthony Dear. Tetramethylrhodamine ethyl ester was purchased from Molecular Probes.
Western Blot Analysis
Eμ-myc lymphoma cells were lysed in lysis buffer [0.15 mol/L NaCl, 10 mmol/L Tris-HCl (pH 7.4), 5 mmol/L EDTA, 1% Triton X-100] supplemented with protease inhibitors (leupeptin, pepstatin, and phenylmethylsulfonyl fluoride; Sigma-Aldrich) as described previously (8). Proteins (30-50 μg) were separated on 10% or 15% SDS-polyacrylamide gels electroblotted onto Immobilon-P nylon membranes (Millipore). Membranes were incubated with the following antibodies: anti-mouse Bcl-2 (BD PharMingen), anti-mouse Bcl-XL (BD PharMingen), anti-mouse Bcl-w (Chemicon Australia), anti-mouse Mcl-1 (Rockland), anti-mouse A1 (Sapphire Biosciences), anti-Flag tag (Sigma-Aldrich), anti-acetylated histone H3 and anti-acetylated histone H4 (Upstate Biosystems), anti-β actin (Sigma-Aldrich), and anti-tubulin (Sigma-Aldrich) overnight at 4°C followed by subsequent incubation with horseradish peroxidase–conjugated secondary antibodies (DAKO). Immunoreactive bands were visualized by enhanced chemiluminescence (Amersham).
In vitro Cell Death Analysis
Eμ-myc lymphoma cells (5 × 105/mL) were incubated in the presence of the indicated compounds for 20 h in 1 mL cell culture medium in 24-well plates (Greiner Bio-One). Viability of cells as measured by trypan blue exclusion assay, propidium iodide uptake, Annexin V staining, cell cycle analysis, or tetramethylrhodamine ethyl ester staining were done as described (8).
Mice
C57BL/6 mice (6-8 weeks old) were used for in vivo apoptosis assays and therapy studies. The Peter MacCallum Cancer Centre Animal Ethics Committee approved all mouse protocols used in this study. PCR-based genotyping and Western blotting analysis were used to validate lymphoma genotypes (data not shown).
In vivo Apoptosis and Therapy Assays
For in vivo apoptosis assays, C57BL/6 mice were injected with Eμ-myc lymphomas (5 × 105 cells per animal) and after 10 to 15 days on which lymph nodes became well-palpable romidepsin (5.6 mg/kg) was administered i.v. After the indicated time points, mice were sacrificed and cells were harvested from brachial lymph nodes for fluorescence-activated cell sorting–based assays to measure apoptotic signaling (8).
To assess therapeutic efficacy of romidepsin, C57BL/6 mice were injected with Eμ-myc lymphomas of the indicated genotypes i.v. (5 × 105 cells per animal). Peripheral WBC counts were then monitored until they exceeded 13 × 103/μL (Sysmex Hematology Analyzer K-1000) and romidepsin was administered at 5.6 mg/kg i.v. once every 4 days for a total of four injections. Previously, we had determined that this regimen represented the maximum tolerated dose in lymphoma-bearing mice (data not shown). Mice in the control cohort received the corresponding amount of vehicle. Cohorts consisted of 8 to 11 mice each, 2 to 3 independently derived lymphomas per genotype. Peripheral WBC counts and body weights were recorded weekly. On signs of major distress or when lymphomas were relapsing as indicated by enlarged brachioaxial lymph nodes, mice were euthanized and a necropsy was done. For analysis of therapeutic efficacy, tumor-induced mortality “events” were recorded. Kaplan-Meier analysis was done and comparisons made using the log-rank (Mantel-Cox) test (MedCalc software version 8.0.2.0).
Results
Romidepsin Can Overcome the Antiapoptotic Effects of Bcl-2 In vitro
We and others have determined previously that overexpression of prosurvival Bcl-2 proteins can inhibit the apoptotic activity of HDACi in vitro using human tumor cell lines (9–13). Moreover, using Eμ-myc lymphomas that had been retrovirally transduced with control vector (MSCV) or with a vector encoding murine Bcl-2, we showed that overexpression of Bcl-2 conferred resistance to the apoptotic and therapeutic effects of vorinostat (8). Three different primary Eμ-myc lymphomas overexpressing Bcl-2 and control vector-transduced Eμ-myc cells were tested for sensitivity to the HDACi oxamflatin and romidepsin. Both agents could effectively kill Eμ-myc but not Eμ-myc/Bcl-2 lymphomas in a 24-h dose-response assay as assessed by outer cell membrane damage (Fig. 1A-F ) and loss of mitochondrial membrane potential (Fig. 1B).
Eμ-myc lymphomas overexpressing Bcl2 are resistant to romidepsin in vitro in short-term assays. 4242Eμ-myc, 4242Eμ-myc/Bcl-2, 229Eμ-myc, 229Eμ-myc/Bcl-2, 226Eμ-myc, and 226Eμ-myc/Bcl-2 lymphomas were incubated with the indicated concentrations of romidepsin or oxamflatin for 24 h. Cell viability was assessed by (A) propidium iodide staining and (B) loss of MOMP. Bars, SE of at least three independent experiments.
To determine if the HDACi-inhibitory effects of Bcl-2 were long lasting, we did a time course experiment using doses of oxamflatin (0.1 μmol/L) and romidepsin (3.0 nmol/L) that were sufficient to kill Eμ-myc lymphomas in 24 h. Overexpression of Bcl-2 conferred resistance to oxamflatin even following 72 h of continuous exposure of the cells to this HDACi (Fig. 2A ). In contrast, romidepsin could kill two of the four Eμ-myc/Bcl-2 lymphomas (4242Eμ-myc/Bcl-2 and 229Eμ-myc/Bcl-2) over time, whereas another two independently derived Eμ-myc/Bcl-2 lymphomas (102Eμ-myc/Bcl-2 and 226Eμ-myc/Bcl-2) remained relatively insensitive to romidepsin. The primary function of prosurvival Bcl-2 proteins is to inhibit the activity of Bak and Bax proteins and thereby protect the mitochondrial membrane from damage (14). We therefore wished to determine if the induction of apoptosis mediated by romidepsin in 4242Eμ-myc/Bcl-2 and 229Eμ-myc/Bcl-2 lymphomas was via perturbation of the mitochondrial membrane or through some other mechanism. We quantitated HDACi-induced mitochondrial outer membrane permeabilization (MOMP) by staining with tetramethylrhodamine ethyl ester. Consistent with the data shown in Fig. 2A, oxamflatin and romidepsin induced robust MOMP in all four Eμ-myc/MSCV lymphomas following 24-h treatment that increased over time (Fig. 2B). However, oxamflatin did not mediate any substantial change in MOMP in any of the Eμ-myc lymphomas that overexpress Bcl-2. In contrast and consistent with the data shown in Fig. 2A, romidepsin induced MOMP in 4242Eμ-myc/Bcl-2 and 229Eμ-myc/Bcl-2 and this effect was greatly attenuated or completely lost in the 226Eμ-myc/Bcl-2 and 102Eμ-myc/Bcl-2 lymphomas.
Romidepsin can kill Eμ-myc/Bcl2 lymphomas over time. 4242Eμ-myc, 4242Eμ-myc/Bcl-2, 229Eμ-myc, 229Eμ-myc/Bcl-2, 226Eμ-myc, 226Eμ-myc/Bcl-2, 102Eμ-myc, and 102Eμ-myc/Bcl-2 lymphomas were incubated for up to 72 h with the concentration of HDACi required to kill ∼70% of Eμ-myc lymphomas following 24-h treatment (3 nmol/L romidepsin or 0.1 μmol/L oxamflatin). Cell viability was assessed by (A) propidium iodide staining and (B) loss of MOMP. Bars, SE of at least three independent experiments. C, 4242Eμ-myc cells were treated with 3.0 nmol/L romidepsin, 0.1 μmol/L oxamflatin or vehicle (lanes 7-9) for 2 h (lanes 1, 4, and 7), 8 h (lanes 2, 5, and 8), and 24 h (lanes 3, 6, and 9). Whole-cell lysates were used for Western blot analysis using antibodies specific for acetylated histones H3 and H4. Blots were reprobed with anti-tubulin polyclonal antibody to assess protein loading. D, 4242Eμ-myc/Bcl-2 and 226Eμ-myc/Bcl-2 cells were treated with 3.0 nmol/L romidepsin for 2 h (lanes 1 and 4), 8 h (lanes 2 and 5), and 24 h or vehicle for 24 h (lanes 7 and 8). Whole-cell lysates were used for Western blot analysis using antibodies specific for acetylated histones H3 and H4. Blots were reprobed with anti-β actin polyclonal antibody to assess protein loading.
We have shown previously that cells resistant to HDACi-induced apoptosis through overexpression of prosurvival Bcl-2 proteins undergo arrest in either the G1 (8) or G2-M phases of the cell cycle following HDACi treatment depending on the cell type (10, 12). We therefore assessed the cell cycle profiles of Eμ-myc/Bcl-2 lymphomas treated with oxamflatin and romidepsin over 3 days. Treatment of 226Eμ-myc/Bcl-2 (Supplementary Table S1)5 or 102Eμ-myc/Bcl-2 (data not shown) lymphomas with oxamflatin or romidepsin over 3 days resulted in a decrease in the percentage of cells in S phase and increase of cells in G1 (Supplementary Table S1).5 Using loss of 2n DNA content (sub-G1) as readout for DNA fragmentation and thus apoptosis, neither oxamflatin nor romidepsin induced substantial cell death even following 3 days of continuous exposure to replenished agent. Similar results were seen when 4242Eμ-myc/Bcl-2 (Supplementary Table S1)5 and 229Eμ-myc/Bcl-2 (data not shown) lymphomas were treated with oxamflatin. In contrast, treatment of 4242Eμ-myc/Bcl-2 (Supplementary Table S1)5 and 229Eμ-myc/Bcl-2 (data not shown) lymphomas with romidepsin resulted in an increase in the percentage of cells showing DNA fragmentation indicative of apoptosis. Taken together, these data show that overexpression of Bcl-2 robustly inhibits the apoptotic activities of the hydroxamate-based HDACi oxamflatin. In contrast, two of the Eμ-myc/Bcl-2 lymphomas that were completely resistant to oxamflatin-induced apoptosis were sensitive to romidepsin-mediated apoptosis following >24-h exposure to drug.
To ensure that romidepsin and oxamflatin induced equivalent histone hyperacetylation at doses of each compound that could kill Eμ-myc lymphomas, we did Western blot analysis to assess the acetylation status of histones H3 and H4. As shown in Fig. 2C, treatment of 4242Eμ-myc lymphomas with 3.0 nmol/L romidepsin and 0.1 mmol/L oxamflatin induced equivalent acetylation of histones H3 and H4 over a 24-h time course. Moreover, addition of 3.0 nmol/L romidepsin to 4242Eμ-myc/Bcl-2 and 226Eμ-myc/Bcl-2 lymphomas resulted in an equivalent increase in histone acetylation in a time-dependent manner. These data indicate that the differential sensitivity of 4242Eμ-myc/Bcl-2 and 226Eμ-myc/Bcl-2 to romidepsin is not related to variations in HDAC-inhibitory activity of the compound in lymphomas that are relatively resistant or sensitive to romidepsin-induced apoptosis.
Apoptotic and Therapeutic Activity of Romidepsin against Eμ-myc and Eμ-myc/Bcl-2 Lymphomas In vivo
Our in vitro data indicated that romidepsin was capable of rapidly killing Eμ-myc lymphomas and could kill 229Eμ-myc/Bcl-2 and 4242Eμ-myc/Bcl-2 lymphomas over time but could not kill 226Eμ-myc/Bcl-2 or 102Eμ-myc/Bcl-2 lymphomas. We therefore sought to determine if similar results were observed in vivo. For this purpose, we did in vivo apoptosis assays that involved treatment of lymphoma-bearing mice with romidepsin, harvesting of tumors over time, and assessment of apoptosis using fluorescence-activated cell sorting–based assays. All four Eμ-myc lymphomas grown in the lymph nodes of C57BL/6 mice were sensitive to romidepsin with an increase in apoptotic cells over background detected at 8 to 12 h following addition of romidepsin (Fig. 3A-D ). The percentage of apoptosis increased over the 24-h time course using readouts for outer cell membrane damage and DNA fragmentation (Fig. 3A-D). Consistent with the results seen in vitro, all four Eμ-myc/Bcl-2 lymphomas were resistant to romidepsin-induced apoptosis 24 h after exposure to the HDACi (Fig. 3E-H). The 226Eμ-myc/Bcl-2 and 102Eμ-myc/Bcl-2 lymphomas remained insensitive to romidepsin-induced apoptosis in vivo, even at the 36 and 48 h time points, respectively (Fig. 3G and H). However, consistent with our in vitro data, 4242Eμ-myc/Bcl-2 (Fig. 3E) and 229Eμ-myc/Bcl-2 (Fig. 3F) lymphomas did undergo apoptosis at later time points following exposure to romidepsin, although as with the in vitro assays the level of apoptosis achieved in these Bcl-2-overexpressing lymphomas at most time points was substantially less than that observed in the parental Eμ-myc lymphomas.
Romidepsin can kill Eμ-myc/Bcl2 lymphomas in vivo. C57BL/6 mice bearing (A) 4242Eμ-myc, (B) 229Eμ-myc, (C) 226Eμ-myc, (D) 102Eμ-myc, (E) 4242Eμ-myc/Bcl-2, (F) 229Eμ-myc/Bcl-2, (G) 226Eμ-myc/Bcl-2, and (H) 102Eμ-myc/Bcl-2 lymphomas were injected with romidepsin (5.6 mg/kg i.v.) or vehicle. Lymphoma cells were harvested at the time points (hours) indicated following romidepsin treatment or 24 h following vehicle treatment (v). Apoptosis was measured by either Fluorogold staining for outer cell membrane permeabilization (gray columns) or DNA fragmentation (white columns).
We next assessed the therapeutic effects of romidepsin against Eμ-myc and Eμ-myc/Bcl-2 lymphomas to determine if the induction of apoptosis by romidepsin in vivo translated into a therapeutic benefit. For therapy experiments, Eμ-myc lymphomas were transplanted into C57BL/6 mice and treatment with romidepsin or vehicle commenced when WBC counts in the peripheral blood reached a pathologic threshold (>13 × 103/μL). The survival of mice bearing Eμ-myc lymphomas treated with romidepsin was significantly extended compared with vehicle-treated mice (Fig. 4A-D ). Interestingly, romidepsin also significantly extended the survival of mice bearing 229Eμ-myc/Bcl-2 and 4242Eμ-myc/Bcl-2 lymphomas but provided little or no therapeutic benefit in mice bearing 102Eμ-myc/Bcl-2 or 226Eμ-myc/Bcl-2 lymphomas (Fig. 4E-H).
Therapeutic effect of romidepsin in vivo. C57BL/6 mice (10 mice per group) bearing (A) 4242Eμ-myc, (B) 229Eμ-myc, (C) 226Eμ-myc, (D) 102Eμ-myc, (E) 4242Eμ-myc/Bcl-2, (F) 229Eμ-myc/Bcl-2, (G) 226Eμ-myc/Bcl-2, and (H) 102Eμ-myc/Bcl-2, lymphomas were treated with romidepsin or vehicle. Therapy commenced after WBC counts reached ≥13 × 103/μL. Therapy consisted of either 5.6 mg/kg romidepsin (injected i.v. every 4 d for a total of four doses) or vehicle. Kaplan-Meier survival curves of vehicle-treated mice (dashed line) and romidepsin-treated mice (solid line) are shown. Median survival and P values for the different lymphomas were as follows: 4242Eμ-myc, median survival vehicle 19 d, median survival romidepsin 28 d, P < 0.0003; 4242Eμ-myc/Bcl-2, median survival vehicle 12 d, median survival romidepsin 22.5 d, P < 0.0001; 229Eμ-myc, median survival vehicle 20 days, median survival romidepsin 30 d, P < 0.0001; 229Eμ-myc/Bcl-2, median survival vehicle 18 d, median survival romidepsin 30 d, P < 0.0001; 226Eμ-myc, median survival vehicle 15 d, median survival romidepsin 19.5 d, P < 0.0001; 226Eμ-myc/Bcl-2, median survival vehicle 16 d, median survival romidepsin 16 d, P=0.86; 102Eμ-myc, median survival vehicle 14 d, median survival romidepsin 22 d, P < 0.0001; 102Eμ-myc/Bcl-2, median survival vehicle 11 d, median survival romidepsin 14.5 d, P < 0.07.
Enhanced Expression of Bcl-XL in 226Eμ-myc/Bcl-2 and 102Eμ-myc/Bcl-2 Lymphomas Correlates with Resistance to Romidepsin-Induced Apoptosis
To determine why 226Eμ-myc/Bcl-2 and 102Eμ-myc/Bcl-2 lymphomas remain resistant to romidepsin-induced apoptosis compared with 229Eμ-myc/Bcl-2 and 4242Eμ-myc/Bcl-2 lymphomas, we assessed the expression of prosurvival Bcl-2 proteins in the cells. All cells overexpressed approximately equivalent amounts of exogenous Bcl-2 (Fig. 5A ). We next assessed the endogenous expression of prosurvival Bcl-2 family members in these lymphomas (Fig. 5B). The expression of Bcl-w, Mcl-1, and A1 was approximately equivalent in all Eμ-myc/Bcl-2 lymphomas. In contrast, the levels of Bcl-XL were significantly higher in 226Eμ-myc/Bcl-2 and 102Eμ-myc/Bcl-2 lymphomas compared with 229Eμ-myc/Bcl-2 and 4242Eμ-myc/Bcl-2 lymphomas.
Expression of exogenous Bcl-2 and endogenous prosurvival Bcl-2 family proteins in Eμ-myc and Eμ-myc/Bcl2 lymphomas. A, expression of exogenous Bcl-2 was detected by Western blot using whole-cell lysates from 4242Eμ-myc, 4242Eμ-myc/Bcl-2, 229Eμ-myc, 229Eμ-myc/Bcl-2, 226Eμ-myc, 226Eμ-myc/Bcl-2, 102Eμ-myc, and 102Eμ-myc/Bcl-2 lymphomas. Blots were reprobed with anti-tubulin polyclonal antibody to assess protein loading. B, expression of endogenous Bcl-XL, Mcl-1, Bcl-w, and A1 was detected by Western blot using whole-cell lysates from 4242Eμ-myc/Bcl-2, 229Eμ-myc/Bcl-2, 226Eμ-myc/Bcl-2, and 102Eμ-myc/Bcl-2 lymphomas. Blots were reprobed with anti-tubulin polyclonal antibody to assess protein loading.
To determine if increased expression of Bcl-XL could confer resistance to romidepsin, we developed 4242Eμ-myc/Bcl-XL lymphomas and tested these for sensitivity to HDACi. Treatment of 4242Eμ-myc and 4242Eμ-myc/Bcl-XL lymphomas with increasing concentrations of romidepsin or oxamflatin over 24 h resulted in dose-dependent loss of plasma membrane integrity and mitochondrial function in 4242Eμ-myc lymphomas, whereas 4242Eμ-myc/Bcl-XL lymphomas were unaffected (Fig. 6A and B ). Moreover, cell cycle analysis revealed that DNA fragmentation occurred in Eμ-myc lymphomas in response to increasing doses of oxamflatin and romidepsin, whereas Eμ-myc/Bcl-XL lymphomas arrested in the G1 phase of the cell cycle (data not shown). Similar results were seen using 102Eμ-myc/Bcl-XL and 229Eμ-myc/Bcl-XL lymphomas (data not shown).
Eμ-myc lymphomas overexpressing Bcl-XL are resistant to romidepsin and oxamflatin in vitro. 4242Eμ-myc and 4242Eμ-myc/Bcl-XL were incubated with the indicated concentrations of (A) romidepsin or (B) oxamflatin for 24 h or with (C) 3 nmol/L romidepsin or (D) 0.1 μmol/L oxamflatin for up to 72 h. Cell viability was assessed by propidium iodide staining and by loss of MOMP. Bars, SE of at least three independent experiments.
Treatment of 4242Eμ-myc/Bcl-XL, lymphomas with romidepsin or oxamflatin over a 72-h time course resulted in little or no outer cell membrane permeabilization nor any significant decrease in mitochondrial membrane potential (Fig. 6C and D). In contrast, parental Eμ-myc lymphomas were effectively killed by romidepsin and oxamflatin within the first 24 h (Fig. 6C and D). Similar results were observed using 102Eμ-myc/Bcl-XL and 229Eμ-myc/Bcl-XL lymphomas (data not shown).
Discussion
HDACi are showing promise in the clinic as anticancer agents, particularly for the treatment of hematologic malignancies (1–3). A large number of HDACi have been synthesized or purified from natural sources and there are clear differences in the chemical composition of the different compounds and their relative affinities for different class I, II, and IV HDACs (1, 15, 16). Although HDACi have variously been shown to induce tumor cell apoptosis, inhibit cell cycle progression, augment antitumor immune responses, or suppress angiogenesis depending on the tumor cell type, concentration of HDACi used, or cellular context of the experiment, very few comparative studies aimed at identifying different molecular or biological activities of structurally diverse HDACi have been done to identify subtle yet potentially important unique activities of the different compounds.
Herein, we have directly compared the apoptotic activities of the cyclic tetrapeptide romidepsin with the hydroxamic acid oxamflatin using primary lymphomas derived from Eμ-myc transgenic mice. The ability to genetically manipulate Eμ-myc lymphomas through the use of retroviral gene transduction and perform both in vitro and in vivo studies on independently derived primary tumors makes this an ideal preclinical system to identify proteins and molecular pathways that may be important for HDACi activity. Using this system, we showed previously that the hydroxamic acid vorinostat mediated Eμ-myc lymphoma cell apoptosis via the intrinsic apoptotic pathway and overexpression of either Bcl-2 or Bcl-XL rendered Eμ-myc cells resistant to the apoptotic and therapeutic activities of vorinostat (8). We showed herein that overexpression of Bcl-2 similarly protects Eμ-myc lymphomas against apoptosis induced by oxamflatin even following 72-h continuous exposure to the compound. In contrast, we identified Eμ-myc lymphomas overexpressing Bcl-2 that were resistant to apoptosis induced by oxamflatin (shown here) or vorinostat (8) but were sensitive to romidepsin-induced apoptosis following extended (that is, >24 h) exposure to the compound. Importantly, we showed that these lymphomas were sensitive to romidepsin in vivo and the compound retained therapeutic efficacy against these cells. However, we also identified Eμ-myc/Bcl-2 lymphomas that were insensitive to romidepsin-induced apoptosis in vitro and in vivo. One apparent difference between the romidepsin-resistant and romidepsin-sensitive Eμ-myc/Bcl-2 lymphomas was the relative expression of Bcl-XL and we showed that overexpression of Bcl-XL conferred resistance to romidepsin in all lymphomas tested. Although Bcl-2 and Bcl-XL are thought to function in a very similar manner to inhibit the proapoptotic activities of Bax and Bak and thereby protect the mitochondrial outer membrane from damage (14), it appears that these proteins may have distinct roles in certain physiologic or pathologic contexts. It has been shown that Bcl-XL may indeed be up to 10 times more effective than Bcl-2 in inhibiting apoptosis mediated by chemotherapeutic drugs such as doxorubicin and etoposide (17). Moreover, knockout of Bcl-2 has no effect on the latency of myc-induced lymphomagenesis and the severity of the disease was unaltered indicating that other prosurvival Bcl-2 family members play an important role in regulating myc-induced apoptosis (18). Indeed, it appears that Bcl-X is critical to regulate survival of immature B cells, whereas Bcl-2 may play a more important role in regulating apoptosis in mature B cells (18, 19).
Exactly how romidepsin can overcome the protective activities of Bcl-2, whereas other compounds such as vorinostat and oxamflatin cannot remains to be determined. Our data showing that romidepsin can induce MOMP in the sensitive Eμ-myc/Bcl-2 lymphomas provides evidence that the compound is still mediating its effects via the intrinsic apoptotic pathway rather than by an alternative or additional pathway. Moreover, as the romidepsin-resistant Eμ-myc/Bcl-2 lymphomas and the Eμ-myc/Bcl-XL lymphomas show no loss of MOMP following exposure to romidepsin, we conclude that the ability to protect the mitochondria from damage mediated by romidepsin is the key event in mediating tumor cell resistance to the compound. Bcl-2 can inhibit the apoptotic activity of many different anticancer agents that have diverse mechanisms of action (20). Interestingly, like romidepsin, aphidicolin has been shown to kill Bcl-2-overexpressing cells following prolonged exposure concomitant with increased DNA fragmentation and outer cell membrane permeabilization (21).
Our studies provide further evidence that structurally different HDACi may have subtle yet potentially important differences in their molecular and biological activities. We have shown previously that tumor cell apoptosis induced by romidepsin is inhibited by the multidrug resistance protein P-glycoprotein, whereas vorinostat and oxamflatin are unaffected by P-glycoprotein expression (12). Moreover, we showed that the cleavage and activation of the BH3-only protein Bid, which is a critical mediator of HDACi-induced apoptosis (8, 10), occurs in a caspase-independent manner following treatment with oxamflatin or vorinostat but is reliant on caspase activity in response to romidepsin (12). Finally, gene expression profiling has revealed that romidepsin and vorinostat equivalently regulate the expression of large numbers of common genes; however, vorinostat-specific and romidepsin-specific genes were identified (22), further showing that these agents do indeed possess some unique molecular activities. Interestingly, we have found that overexpression of Bcl-2 confers resistance to apoptosis mediated by other hydroxamic-based HDACi (LAQ824 and LBH589), the short-chain fatty acid valproic acid and the benzamide MS-275,6 indicating that the ability of romidepsin to overcome the effects of Bcl-2 may be unique to this compound. Whether other cyclic tetrapeptides such as trapoxin A or apicidin can also kill lymphomas overexpressing Bcl-2 remains to be determined.
In summary, we have shown that, unlike oxamflatin or vorinostat, romidepsin can overcome the antiapoptotic effects of Bcl-2 in vitro and this translates into a therapeutic effect in vivo. As vorinostat is already a Food and Drug Administration–approved drug and romidepsin is being tested clinically to achieve Food and Drug Administration approval (2, 23), identifying the molecular events that may inhibit the therapeutic efficacy of each compound may be important for the future stratification of patients being treated with one or other compound. Our data indicate that, in situations where Bcl-2 is overexpressed in a particular tumor, romidepsin may be a more preferred treatment option than vorinostat. In contrast, if P-glycoprotein is overexpressed in a particular tumor, then vorinostat would likely be more efficacious than romidepsin, which is a substrate for P-glycoprotein. However, our data also indicate that the relative expression of a single antiapoptotic protein such as Bcl-2 may not provide a true indication of the relative sensitivity of a tumor to romidepsin. Indeed, we showed that the relative expression of Bcl-XL inversely correlated with the apoptotic and therapeutic activities of romidepsin and Bcl-XL appeared to be a far more potent inhibitor of romidepsin-mediated apoptosis than Bcl-2. Clearly, romidepsin and hydroxamate-based compounds such as vorinostat and oxamflatin have different relative affinities for individual HDACs, with romidepsin being more selective for class I HDACs, whereas vorinostat and related compounds are considered pan-HDACi (1). It remains to be determined if the subtle yet potentially important antitumor effects of structurally diverse HDACi relate to differences in their target specificity resulting in some specific gene regulatory activities or other inherent differences in the activities of the compounds that could be unrelated to HDAC-inhibitory activity.
Disclosure of Potential Conflicts of Interest
R.W. Johnstone: Novartis and Merck grant support. The other authors disclosed no potential conflicts of interest.
Acknowledgments
We thank David Huang, Andreas Strasser, Clare Scott, Jerry Adams, Suzanne Cory, and Alan Harris (Walter and Eliza Hall Institute) for helpful advice, Eμ-myc mice, and eDNAs for Bcl-2 and Bcl-XL and Drs. Chris Clarke and Nigel Waterhouse (Peter MacCallum Cancer Centre) for helpful advice and input. Depsipeptide was kindly provided by Gloucester Pharmaceuticals.
Footnotes
↵5 Supplementary material for this article is available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
↵6 Unpublished observations.
Grant support: R.W. Johnstone is a Pfizer Australia research fellow and is supported by National Health and Medical Research Council program grant 251608, Cancer Council Victoria, and Leukemia Foundation of Australia.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Accepted February 21, 2008.
- Received November 4, 2007.
- Revision received February 5, 2008.
- American Association for Cancer Research