Belinostat and vincristine demonstrate mutually synergistic cytotoxicity associated with mitotic arrest and inhibition of polyploidy in a preclinical model of aggressive diffuse large B cell lymphoma

ABSTRACT

Diffuse Large B-cell lymphoma (DLBCL) is an aggressive malignancy that has a 60 percent 5-year survival rate, highlighting a need for new therapeutic approaches. Histone deacetylase inhibitors (HDACi) are novel therapeutics being clinically-evaluated in combination with a variety of other drugs. However, rational selection of companion therapeutics for HDACi is difficult due to their poorly-understood, cell-type specific mechanisms of action. To address this, we developed a pre-clinical model system of sensitivity and resistance to the HDACi belinostat using DLBCL cell lines. In the current study, we demonstrate that cell lines sensitive to the cytotoxic effects of HDACi undergo early mitotic arrest prior to apoptosis. In contrast, HDACi-resistant cell lines complete mitosis after a short delay and arrest in G1. To force mitotic arrest in HDACi-resistant cell lines, we used low dose vincristine or paclitaxel in combination with belinostat and observed synergistic cytotoxicity. Belinostat curtails vincristine-induced mitotic arrest and triggers a strong apoptotic response associated with downregulated MCL-1 expression and upregulated BIM expression. Resistance to microtubule targeting agents (MTAs) has been associated with their propensity to induce polyploidy and thereby increase the probability of genomic instability that enables cancer progression. Co-treatment with belinostat effectively eliminated a vincristine-induced, actively cycling polyploid cell population. Our study demonstrates that vincristine sensitizes DLBCL cells to the cytotoxic effects of belinostat and that belinostat prevents polyploidy that could cause vincristine resistance. Our findings provide a rationale for using low dose MTAs in conjunction with HDACi as a potential therapeutic strategy for treatment of aggressive DLBCL.

KEYWORDS:

• Cell cycle
• drug resistance
• histone deacetylase inhibitor
• lymphoma
• mitosis
• microtubule
• polyploidy

Abbreviations

DLBCL	=	diffuse large B cell lymphoma
HDAC	=	histone deacetylase
HDACi	=	histone deacetylase inhibitor
GCB	=	germinal center B cell-like
ABC	=	activated B cell-like
NHL	=	non-Hodgkin lymphoma
MTA	=	microtubule targeting agent
DMSO	=	dimethyl sulfoxide
ICC	=	immunocytochemistry
H3S10Ph	=	histone H3 phosphorylated at serine 10
SAC	=	spindle assembly checkpoint
BCL2	=	B cell CLL/lymphoma 2
MYC	=	c-myc
Mt	=	microtubule
PARP	=	poly ADP ribose polymerase
PI	=	propidium iodide
RRR	=	relative risk ratio
PTX	=	paclitaxel

Introduction

Diffuse Large B-cell Lymphoma is the most commonly-diagnosed form of Non-Hodgkin Lymphoma (NHL), affecting approximately 30,000 people each year in the United States.¹ The current standard therapeutic regimen is the anti-CD20 monoclonal antibody Rituximab in conjunction with cyclophosphamide, vincristine, doxorubicin and prednisone, known as R-CHOP. Among all DLBCL cases R-CHOP treatment yields a 60% 5 y survival rate,^2-4 thus new therapeutic strategies are clearly necessary to treat aggressive forms of DLBCL. To this end, histone deacetylase inhibitors (HDACi) are being evaluated in clinical trials for use in treatment of multiple types of NHL and 4 have gained FDA approval for the treatment of advanced peripheral T-cell lymphomas and multiple myeloma.

HDACi are promising cancer chemotherapeutics because they selectively target tumor cells and exhibit tolerable levels of toxicity in humans.^5-7 HDACi are largely ineffective as monotherapy against DLBCL as well as solid tumors, therefore, research focus has shifted to combining HDACi with other therapeutics for treatment of a variety of cancers.^8-10 A confounding issue in identifying effective HDACi-containing drug combinations is that a multitude of mechanisms have been attributed to the anti-cancer effects of HDACi including: alteration of gene expression, activation of pro-apoptotic pathways, induction of tumor cell differentiation, inhibition of angiogenesis, and modulation of cell cycle progression.^5,7,11-14 Taking a mechanistic approach to understand the molecular effects of HDACi specifically in the DLBCL context, we developed a cell-based model of sensitivity and resistance to HDACi. The mechanistic knowledge gained from this system can be used to rationally select therapeutics which can be effectively combined with HDACi. Previously, we identified 2 major responses to the hydroxamate HDACi, belinostat in our model system, including 1) a reversible, cytostatic arrest in G1 and 2) arrest in G2/M that is followed by apoptosis.¹⁵ We showed that the reversible G1 arrest, which we designated as a form of HDACi resistance, is associated with sustained HDACi-induced expression of the cyclin-dependent kinase inhibitors, p21 and p27, as well as their inhibition of the Cyclin E/cdk2 complex through increased association.¹⁵

A major cause of HDACi-induced tumor cell death has been attributed to perturbed mitotic progression.¹⁶ These drugs have been reported to activate the spindle assembly checkpoint (SAC) resulting in an accumulation of cells in pro-metaphase. However, continued exposure to HDACi can cause SAC failure, mitotic catastrophe or slippage, and the induction of pro-apoptotic signaling.^17-19 SAC activation is also a major mechanism attributed to the anti-tumor effectiveness of microtubule targeting therapies such as taxanes and vinca alkaloids.^20,21 These drugs are common in many first line therapeutic regimens including the R-CHOP combination used for most non-Hodgkin's lymphomas. The induction of the SAC by microtubule targeting agents (MTAs) results in an extended mitotic arrest and the onset of senescence or cell death. However, therapeutically targeting mitotic progression can also cause failure of cytokinesis, resulting in a polyploid cell population. Polyploidy is linked to increased genomic instability and the induction of tumor heterogeneity and ultimately drug resistance.^22-25

In this study we have determined that the G2/M arrest we previously reported in HDACi-sensitive DLBCL cell lines is due to accumulation of cells in early mitosis, consistent with SAC activation. In contrast, the HDACi belinostat delays mitotic progression but does not prevent mitotic completion in HDACi-resistant DLBCL cell lines. Based on these findings we hypothesized that combining HDACi with therapeutics that strongly induce mitotic arrest might effectively sensitize resistant cells to the cytotoxic effects of HDACi. We show that combining the HDACi, belinostat, with low doses of the MTAs, vincristine or paclitaxel, synergistically induces apoptosis in resistant DLBCL cell lines, but has little to no effect on a non-transformed B cell-derived cell line. In addition, we find that the mitotic arrest is essential for the cytotoxic synergy between the drugs. Finally, we show that belinostat enhances vincristine-induced cytotoxicity by curtailing vincristine-induced SAC activation and triggering a strong apoptotic response that prevents the generation of polyploid cells. These findings show that HDACi and MTAs are mutually synergistic in inducing cell death in DLBCL cell lines that share molecular characteristics with aggressive forms of DLBCL in a way that reduces the probability of genomic instability.

Results

HDACi-sensitive DLBCL cell lines arrest early in mitosis while HDACi-resistant DLBCL cell lines are able to complete mitosis prior to G1 arrest

In a previous study we documented 2 major responses to the hydroxamate HDACi, belinostat, in DLBCL cell lines.¹⁵ One is characterized by a reversible arrest in the G1 phase of the cell cycle within 24–48 h of treatment without accompanying apoptosis. In contrast, the second response is characterized by G2/M arrest within 24 h and subsequent onset of apoptosis by 48 hours. HDACi have been reported to induce both G2 and M phase arrest.^26-28 To distinguish the 2 phases of the cell cycle we measured global histone H3 serine 10 phosphorylation (H3S10Ph), which occurs once cells enter prophase and begins to decrease in anaphase.²⁹ If belinostat induces early mitotic arrest, cellular levels of H3S10Ph are expected to increase. We show in Fig. 1A that levels of H3S10Ph are markedly increased within 8–12 h after belinostat treatment in the sensitive cell lines, DB and OCI-Ly19. After the initial increase, H3S10Ph levels decrease in both cell lines, consistent with reports that mitotic slippage can occur after HDACi-induced mitotic arrest.^17,19,30,31 Surprisingly, we also observed an increase in H3S10Ph levels in the HDACi-resistant cell lines (SUDHL4 and SUDHL8) with kinetics similar to the HDACi-sensitive lines. To confirm the western blotting results, we examined H3S10Ph levels using immunocytochemistry (ICC) on SUDHL4 and DB cells treated for 16 h with belinostat or DMSO. Fig. 1B,C show that both cell lines are significantly enriched for positively-stained cells after 16 h of belinostat treatment as compared to vehicle (DMSO) treatment.

Figure 1. Belinostat treatment causes accumulation of DLBCL cells in mitosis. (A) Belinostat sensitive (DB, OCI-Ly19) and resistant (SUDHL4, SUDHL8) cells were treated with belinostat for up to 36 h. Belinostat was used at previously-determined IC50 concentrations for each cell line. Equal volumes of cell lysates were subjected to Western blotting with antibodies to either H3S10Ph (mitotic marker) or GAPDH (loading control). (B) Immunohistochemistry using antibody against H3S10Ph for SUDHL4 and DB cells treated with belinostat or DMSO for 16 h. (C) The total number of cells and the number of positive-staining cells were counted in multiple, random fields from 40X and 100X ICC images of cells treated for 16 h with belinostat or DMSO. The results are represented graphically. The unpaired student's t-test was used to calculate significance. ***, p ≤ 0.001. (D) Images (100X magnification) of belinostat-treated, mitotic SUDHL4 or DB cells positively-stained for H3S10Ph.

Although mitotic cells appear to accumulate in SUDHL4 and SUDHL8 cells in response to belinostat, these cell lines eventually arrest in G1.¹⁵ It is possible that the increase in mitotic cells is due to a mitotic delay that is eventually resolved. We therefore examined high magnification images of the H3S10Ph-stained cells for evidence of anaphase mitotic figures, which would indicate that cells are progressing beyond prometaphase, when the spindle assembly checkpoint (SAC) is activated. In the resistant SUDHL4 cell line, anaphase figures are readily apparent at 16 h of belinostat treatment (Fig. 1D). In contrast, we were unable to find any anaphase figures in the belinostat-treated DB cells; all positively-stained cells appear to be arrested prior to metaphase. Altogether our results are consistent with a belinostat-induced transient mitotic delay in the SUDHL4 cells and a pre-metaphase arrest in the DB cells that eventually leads to apoptosis.

Combining belinostat with vincristine results in synergistic cell death in HDACi-resistant DLBCL cell lines

As shown above, belinostat-induced cytotoxicity is associated with early mitotic arrest in DLBCL cell lines. However, at similar submicromolar concentrations belinostat is unable to induce a strong mitotic arrest in the resistant DLBCL cell lines. Studies of the mitotic effects of HDACi in other cell types have established that, in addition to activating the SAC, they cause it to fail prematurely and induce apoptotic signaling^17,19,30,31). We therefore hypothesized that forced mitotic arrest in the resistant cells might sensitize them to belinostat-induced cytotoxicity. To address this, we used the MTA, vincristine, at concentrations low enough to cause near-maximal mitotic arrest and sustained increase in H3S10Ph by 24 h (Fig. 2) but ≤30 % cell death by 48 h (Fig. 3C). Cleavage of poly ADP ribose polymerase (PARP) and pro-Caspase 3 were evaluated in 2 belinostat-resistant cell lines using a 48 h time course of belinostat and vincristine, alone or in combination (Fig. 3A,B). Individually each drug caused low to moderate levels of PARP cleavage and little to no cleavage of caspase 3. The combination of belinostat and vincristine however, resulted in nearly complete PARP cleavage and high levels of Caspase 3 cleavage in both cell lines, indicating efficient onset of apoptosis.

Figure 2. Low dose vincristine treatment results in accumulation of cells in mitosis. (A) SUDHL4 and SUDHL8 cells were treated with vincristine at various doses between 1 and 10 nM for 24 h followed by cell cycle analysis. The percentage of cells in G2/M from 2–3 independent experiments is shown. Curve-fitting was performed using GraphPad Prism software. (B,C) SUDHL4 (B) and SUDHL8 (C) cells were treated with vincristine at concentrations of 5 nM and 3 nM, respectively for up to 24 h. Cell lysates were subjected to immunoblotting with antibodies against H3S10Ph or GAPDH. A representative experiment is shown.

Figure 3. Belinostat combined with low dose microtubule targeting agents act synergistically to induce apoptosis in HDACi-resistant DLBCL cell lines. A,B) SUDHL4 (A) or SUDHL8 (B) cells were treated with belinostat, vincristine, or the combination for up to 48 h. Vincristine concentrations were 5 nM for SUDHL4 and 3 nM for SUDHL8. Cell lysates were subjected to Western blotting with antibodies against PARP (cleaved and uncleaved), Caspase 3 (cleaved and uncleaved), and β-actin. C,D) SUDHL4 or SUDHL8 cells were treated with DMSO, belinostat (Bel), vincristine (VCR), or the combination for 24, 48, and 72 h and subjected to the Annexin V/PI uptake assay. (C) The number of cells staining positive for PtdIns, Annexin V, or both is shown graphically for 3–4 independent replicates. (D) The number of viable cells (negative staining for PI and/or Annexin V) from 3–4 independent replicates is shown graphically. Error bars represent SEM.

As an additional measure of apoptosis, Annexin V/propidium iodide (PI) uptake assays were conducted (Fig. 3C). As expected, belinostat alone caused little cell death by 72 h in both HDACi-resistant cell lines. Low dose vincristine alone caused low to moderate amounts of cytotoxicity. The combination treatment however, caused much more cell death than either drug alone. By 72 h the belinostat/vincristine combination induced apoptosis in greater than 75% of SUDHL4 and SUDHL8 cells as compared to ≤ 10% of dead cells with belinostat alone or ≤ 50% with vincristine alone. Fig. 3D shows the percentage of viable cells (derived from the AnnexinV/PtdIns assays) with each treatment at 24, 48, and 72 h. The relative risk ratio (RRR) calculation using cell viability data can be used to indicate whether a drug combination is synergistic, additive, or antagonistic ³²; RRR values <1.0 are predictive of drug synergy. As described in Materials and Methods, the RRR values for the belinostat/vincristine combination at 72 hours in SUDHL4 and SUDHL8 cells were found to be 0.25 and 0.51, respectively, consistent with synergy between the 2 drugs.

HDACi selectively induce cytotoxicity in transformed cells versus non-transformed cells, but vincristine can be toxic to both. Thus, when combining these drugs it is important to assess potential cytotoxicity in non-transformed cells. GM18564, a non-transformed, Epstein Barr Virus-immortalized human B lymphoblastoid cell line with a 30 h doubling time (not shown), was used to address this issue. Treatment of GM18564 cells with doses of vincristine and belinostat that induced strong cell cycle effects in the experiments with DLBCL cell lines caused no significant change in the percentage of cells in G2/M while exposure to belinostat led to a small increase in the G1 population (Fig. 4A). In addition, the drugs induced little, if any, change in the extent of PARP cleavage (Fig. 4B). These results show that non-transformed B cells are much less sensitive to the effects of these drugs on cell growth and survival. In particular, GM18564 cells are not sensitive at all to the low concentrations of vincristine that elicit strong cell cycle effects in DLBCL cell lines.

Figure 4. Cotreatment with belinostat and vincristine does not cause cytotoxicity in a non-transformed B lymphoblastoid cell line. GM18564 cells were treated with DMSO, belinostat, vincristine, or the combination for 48 h. (A) Cells were collected for cell cycle analysis. The graph shown is a summary of 3 independent replicates. (B) Cell lysates were subjected to Western blotting with antibodies against PARP (cleaved and uncleaved) or GAPDH. A representative blot is shown.

Mitotic arrest is essential for HDACi-induced cytotoxicity in DLBCL cell lines

Vincristine inhibits microtubule (Mt) polymerization while taxanes, such as paclitaxel, stabilize the polymerized state and inhibit depolymerization of Mts. However, both drugs induce mitotic arrest. To determine whether the synergy between belinostat and vincristine is specific to the Mt targeting mechanism of the latter or is dependent on mitotic arrest, we assessed the level of cytotoxicity induced by belinostat in combination with paclitaxel. Using doses of paclitaxel in the low nanomolar range that cause near maximal G2/M arrest (not shown), we examined cytotoxicity by Annexin V/PI assay after 48 h treatment. Paclitaxel alone induced low levels of cytotoxicity (approximately 30%) in SUDHL4 cells (Fig. 5A). However, the combination of belinostat and paclitaxel induced high levels of cytotoxicity (65–75%), similar to what was observed with the vincristine/belinostat combination (Fig. 3C). Analysis of the cell viability data (Fig. 5B) shows that RRR values of < 1.0: 0.52 and 0.50 for 6 and 8 nM paclitaxel plus belinostat, respectively. These results show that the synergy is independent of the mechanism of Mt targeting and suggests that the key event that sensitizes belinostat-resistant DLBCL cells to belinostat-associated cytotoxicity is sustained mitotic arrest.

Figure 5. Mitotic arrest is essential for HDACi-induced cytotoxicity in DLBCL cell lines. (A,B) Co-treatment with paclitaxel and belinostat causes enhanced cytotoxicity in HDACi-resistant DLBCL cell lines. SUDHL4 or SUDHL8 cells treated with DMSO, belinostat, paclitaxel (PTX), or the combination for 48 h. Cells were subjected to the Annexin V/PtdIns uptake assay. (A) The number of cells staining positive for PI, Annexin V, or both is shown graphically for 3–4 independent replicates. (B) The number of viable cells (negative staining for PI and/or Annexin V) from 3–4 independent replicates is shown graphically. Error bars represent SEM. (C-E) Enhanced cytotoxicity induced by the combination of belinostat and vincristine is dependent on cell cycle progression. C.) Experimental design. SUDHL4 cells were treated with vehicle (DMSO) or with belinostat (Bel) and vincristine (VCR) either alone or in combination for 48 h prior to harvest as shown below the timeline. Alternatively cells were treated with belinostat (Bel) for 48 h followed by the addition of vincristine (VCR) or water for an additional 48 h as shown above the timeline. D) Lysates from cells treated as described in (C) were subjected to Western blotting for PARP cleavage and p27, which is a marker of belinostat-induced G1 arrest. E) Annexin V/ Propidium iodide uptake assays were used to examine the induction of cell death by the various treatments described in (C). The graph is a summary of at least 3 independent experiments. Error bars represent SEM. Statistical analysis was carried out using the paired t test comparing the total population of dead and dying cells (PtdIns plus Annexin V plus AnnV/PI) between conditions. ** - p ≤ 0.01, *** - p ≤ 0.001.

To further test the relationship between mitotic arrest and belinostat-induced cytotoxicity, we varied the order of cellular exposure to vincristine and belinostat. In the experiments described above, both vincristine and belinostat were added simultaneously and mitotic arrest was apparent by 24 h (not shown and see Fig. 6B). Since belinostat induces G1 arrest in belinostat-resistant DLBCL cells by 48 h treatment we hypothesized that prior treatment with belinostat would prevent synergy with vincristine. Thus, as described by the timeline shown in Fig. 5C, SUDHL4 cells were either treated with the drugs simultaneously for 48 h (below the line) or pretreated with belinostat for 48 h followed by either water (vehicle) or vincristine for an additional 48 h (above the line). Fig. 5D shows that belinostat treatment for 48 h induced G1 arrest as indicated by increased expression of the cyclin-dependent kinase inhibitor, p27.¹⁵ To measure cell death, we carried out Annexin V/PtdIns assays. As expected, treatment with either drug alone has little effect on PARP cleavage (Fig. 5D) or cell death (Fig. 5E), while simultaneous treatment with both belinostat and vincristine for 48 h strongly triggered both. However, when cells are pre-exposed to belinostat for 48 h prior to vincristine (denoted as Bel then VCR), there was no additional cell death beyond that observed when belinostat was present for 96 h (denoted as Bel then H₂O), and there is no synergy between the drugs (compare bars 4 and 5, Fig. 5D). Altogether these results establish the importance of mitotic arrest in belinostat-induced cytotoxicity in the DLBCL context and indicate that the enhanced cytotoxic effect of the combination is sensitive to the order of drug addition.

Figure 6. Belinostat curtails SAC activation and mitotic arrest induced by vincristine. (A) SUDHL4 and SUDHL8 cells were treated with belinostat, vincristine, or the combination for up to 48 h. Whole cell extracts were subjected to Western blotting with antibodies against BubR1 and α-tubulin. Arrows denote the slower migrating band that represents phosphorylated BubR1. (B) SUDHL4 and SUDHL8 cells were treated with vincristine or vincristine plus belinostat for up to 36 h. Cell lysates were subjected to Western blotting with antibodies against H3S10Ph or GAPDH. Blots representative of at least 3 independent experiments are shown.

Belinostat curtails vincristine-induced SAC activation and causes efficient induction of apoptosis

Previous studies have shown that HDAC activity is required for stable microtubule-kinetochore interactions that are essential for progression beyond prometaphase.^17,19,31 In addition, HDACs appear to be necessary to maintain the SAC once it has been activated.^17,19,30,31 Based on these studies and our results we predict that vincristine induces prolonged activation of the SAC in the HDACi-resistant cell lines and belinostat causes it to fail. To address this hypothesis we examined the phosphorylation of BubR1, an event that indicates SAC activation.^33,34 The mobility of BubR1 in SDS-PAGE gels is slowed by phosphorylation relative to unmodified BubR1. As expected, belinostat alone was not sufficient to induce the accumulation of phosphorylated BubR1 (pBubR1) (Fig. 6A). In contrast, 8 h of exposure to vincristine caused detectable accumulation of pBubR1 in both SUDHL4 and SUDHL8 cells (denoted by the arrows in Fig. 6A). This phosphorylation persisted for another 16–24 hours and then disappears. The combination of the 2 drugs caused an early accumulation of low levels of pBubR1 that was short-lived, suggesting that the strong and persistent SAC activation by vincristine is curtailed in the presence of belinostat.

Mitotic phosphorylation of histone H3 persists through metaphase but begins to decrease in anaphase and is barely detectable by cytokinesis/mitotic exit.²⁹ To further document SAC failure, we measured levels of H3S10Ph in the presence of vincristine alone and the drug combination by Western blotting in SUDHL4 and SUDHL8 cells (Fig. 6B). Vincristine alone caused a prolonged elevation of H3S10Ph consistent with an extended mitotic arrest. Co-treatment with vincristine and belinostat, resulted in a shorter period of H3S10Ph accumulation relative to vincristine alone (Fig. 6B), which could be due to death of cells still in mitosis or mitotic slippage.^20,35-38 In addition, accumulation of pBubR1 was much lower and did not persist (Fig. 6A). These findings indicate that belinostat is capable of attenuating the mitotic arrest induced by vincristine. Total levels of BubR1 became almost undetectable after 40–48 h in the presence of belinostat and VCR. Several reports have shown that BubR1 degradation occurs prior to mitotic exit.^39-41 The decline in total BubR1 occurs in the similar time frame as the decrease in levels of H3S10Ph.

Studies have shown that prolonged mitotic arrest can result in death in mitosis or mitotic slippage. After the latter, cells can undergo multiple fates including cell death by apoptosis, long-term growth arrest or senescence, or continued cycling with a polyploid genome.²⁰ Our findings suggest that belinostat exposure during mitotic arrest triggers a strong apoptotic signal. This contention is further supported by experiments done with the pan-caspase inhibitor, Z-VAD-FMK, which strongly reduced SUDHL4 cell death in the presence of both belinostat and vincristine as determined by Annexin V/PI assays (Fig. 7A). Analysis of the expression of pro- and anti-apoptotic Bcl-2 family members in the presence of belinostat and/or vincristine in SUDHL4 cells showed that neither drug had an impact on BCL2 expression, and that BCL-XL expression was undetectable (not shown). However, belinostat treatment alone down-regulated expression of the pro-apoptotic short from of MCL-1 (lower band, Mcl-1 blot, Fig. 7B) but had little effect on the anti-apoptotic long form (upper band, Mcl-1 blot, Fig. 7B).^42,43 In addition, belinostat up-regulated expression of the BH3-only apoptotic activator, BIM about 2–3-fold. Vincristine treatment had little effect on expression of either MCL-1 or BIM. However, upon treatment with both belinostat and vincristine, both long and short forms of MCL-1 were reduced by at least 50% over 48 h treatment. Importantly, in the period over which MCL-1 expression declined, BIM expression increased and Caspase 3 cleavage commenced (see Figs. 3A and 7B). Thus, in the presence of vincristine, belinostat shifts the balance of pro- and anti-apoptotic Bcl-2 family members toward expression of the former.

Figure 7. Enhanced cytotoxicity induced by the combination of belinostat and vincristine is dependent on apoptotic signaling and correlates with a shift in the balance of anti- and pro-apoptotic factors. (A) SUDHL8 cells were treated for 48 hours with DMSO, belinostat (Bel), vincristine (VCR) or the combination with and without the pan-caspase inhibitor, Z-VAD-FMK. Cell death was measured using Annexin V/ PtdIns uptake assays. The graphs summarize the results of 4 independent experiments. (B) Cells were treated for up to 48 hr with belinostat (Bel), vincristine (VCR), or the combination. Cell lysates were subjected to Western blotting with antibodies against MCL-1 and BIM. GAPDH was used as a loading control. A blot representative of 3–4 independent experiments is shown.

Belinostat prevents vincristine-induced formation of polyploid cells

In the course of cell cycle analyses, we noticed that vincristine treatment caused the generation of a polyploid population of cells in SUDHL8 cells (denoted by the arrow in Fig. 8A). Mitotic slippage without genome division has been documented to occur with other drugs that target mitosis and may be associated with drug resistance and the development of aneuploidy [reviewed in ^23,44-46]. Interestingly, the presence of belinostat largely prevents the vincristine-induced accumulation of polyploid SUDHL8 cells (Fig. 8A). The number of cells in the peak shown in Fig. 8A was quantitated as shown in Fig. 8B. In the presence of vincristine alone, approximately 6% of cells are in this population (light gray bars, minus Z-VAD-FMK). With the belinostat-vincristine combination treatment this population is reduced to less than 1%, similar to the amount observed with DMSO alone. Because we demonstrated that belinostat induces a strong apoptotic signal, we hypothesized that inhibition of apoptosis with the pan-caspase inhibitor, Z-VAD-FMK, would impair the ability of belinostat to reduce the polyploid population, which it did (as seen in Fig. 8B).

Figure 8. Belinostat treatment prevents vincristine-induced polyploidy in SUDHL8 cells. SUDHL8 cells were treated for 48 hours with DMSO, belinostat (Bel), vincristine (VCR), or the combination. (A) Cells were subjected to cell cycle analysis. The arrow denotes a polyploid cell population. (B) Quantitation of the polyploid cell population indicated by the arrow in (A) in SUDHL8 cells treated with and without the pan-caspase inhibitor Z-VAD-FMK. The graph is a summary of 4 independent experiments. Error bars represent SEM. Statistical analysis was carried out using the paired t test comparing the population of polyploid cells with 8N DNA content in the presence and absence of Z-VAD-FMK. ** - p ≤ 0.01, *** - p ≤ 0.001 (C) SUDHL8 cells treated with DMSO, belinostat, vincristine and the combination were subjected to flow-cytometric analysis after staining with Alexa 488-tagged H3S10Ph antibody and propidium iodide. Neither cells with < 2N content nor debris were gated out. The plots shown are representative of 3 independent experiments. Mitotic populations of 4N (left) and 8N (right) cells used for quantitation are denoted as ovals in each panel. (D) Quantitation of polyploid cells (8N) with high H3S10Ph staining as identified in the oval on the right side of each panel shown in (C). Statistical analysis was carried out using the paired t test comparing the total population of polyploid mitotic cells (high H3S10Ph staining) between vincristine and vincristine plus belinostat treatments. ** - p ≤ 0.01, *** - p ≤ 0.001.

Once cells treated with drugs that disrupt mitotic progression have undergone mitotic slippage, several cell fates have been documented, including cell death in interphase, continued cell cycle progression and division, or senescence.^20,35-38 The relative DNA content of the polyploid cell population suggests that some tetraploid cells continued cycling after slippage, replicated their DNA, and may have entered mitosis. To test this we carried out FACs analysis on cells that had been permeabilized and stained with antibodies to H3S10Ph and propidium iodide. In Fig. 8C, the ovals indicate the mitotic populations that show elevated H3S10Ph staining in a representative experiment. The ovals on the left in each plot represent the 4N cell population undergoing mitosis while the ovals on the right represent the portion of the polyploid cell population indicated in Fig. 8A that is mitotic. In the presence of DMSO or belinostat there are few polyploid cells in mitosis (quantitated in Fig. 8D). In the presence of vincristine alone, the number of 4N cells in mitosis is much increased, as expected. In addition, there are cells with elevated H3S10Ph staining in the polyploid population (Fig. 8C,D), indicating that these cells continued cycling after mitotic slippage and entered mitosis. The belinostat-vincristine combination treatment reduces the number of both 4N and polyploid cells with elevated H3S10Ph staining, consistent with belinostat-induced SAC failure and apoptotic signaling. These results show that belinostat enhances vincristine-induced cell death by preventing the survival of tetraploid cells generated by mitotic slippage.

Discussion

Relapsed or refractory DLBCL is associated with a high mortality rate ⁴ and highlights a need for new therapies that can lengthen survival of patients with aggressive forms of DLBCL. In the current study we show that sensitivity to the cytotoxic effects of HDACi in cell lines modeling aggressive DLBCL is associated with mitotic arrest prior to metaphase. In contrast, DLBCL cell lines resistant to the cytotoxic effects of HDACi are able to complete mitosis and arrest reversibly in G1. However, if mitotic arrest is induced with the MTAs vincristine or paclitaxel, the resistant cells become sensitive to the cytotoxic effects of belinostat through failure of mitotic arrest and efficient induction of apoptosis. In addition, we show that vincristine-induced generation of cycling polyploid cells is prevented by co-treatment with belinostat. Thus, the belinostat-vincristine drug combination exhibits mutual cytotoxic synergy in DLBCL cells.

Our previous study described the development of a DLBCL cell line-based model of sensitivity and resistance to the HDACi, belinostat.¹⁵ In the current study we determined that cell lines sensitive to belinostat-induced cytotoxicity undergo mitotic arrest prior to initiation of apoptosis, consistent with reports showing that HDACi can activate the SAC.^17,19 HDAC inhibition causes destabilized interactions between microtubules (Mt) and kinetochores assembled at centromeres.^17,31 A recent study showed that aberrant acetylation of the Mt end-binding protein, EB1, prevents stable chromosome alignment at the metaphase plate.⁴⁷ SAC activation prevents progression to anaphase and causes the accumulation of cells in prometaphase with chromosomes containing high levels of H3S10Ph. We show that belinostat induces significant increases in H3S10Ph in sensitive DLBCL cell lines, DB and OCI-Ly19. The ICC analysis in DB cells shows an accumulation of pre-metaphase H3S10Ph-positive cells and a complete lack of positively-stained cells in anaphase. Altogether these results strongly indicate that belinostat is able to activate the SAC in sensitive DLBCL cell lines.

In contrast, our findings suggest that DLBCL cell lines resistant to the cytotoxic effects of belinostat exhibit a delay in mitotic progression but are ultimately able to complete mitosis and arrest in G1. Establishing a mitotic arrest using vincristine sensitizes the resistant cell lines to the cytotoxic effects of belinostat. The mitotic arrest is essential for this effect since paclitaxel, a second MTA with a distinct mechanism of action, is able to induce mitotic arrest and sensitize the cells to belinostat. In addition, allowing the cells to arrest in G1 via belinostat treatment prior to vincristine exposure prevents mitotic arrest and abolishes the synergistic cytotoxic effect observed upon simultaneous co-treatment with the drugs.

HDAC activity has been shown to be important for maintenance of the SAC [reviewed in ¹⁸]. HDACi-treated cells arrested in prometaphase cannot maintain the arrest and exit mitosis without having entered anaphase or partitioned their genome, suggestive of SAC failure and mitotic slippage.^17,19 Furthermore, extended mitotic arrest caused by treatment with nocodazole is significantly curtailed by HDAC inhibition.^30,48,49 We show very similar results with the clinically-relevant MTA, vincristine. At low doses (≤5nM) this drug maintains elevated levels of H3S10Ph and phosphorylated BubR1 for at least 24 h in the SUDHL4 and SUDHL8 cell lines. However, in the presence of both belinostat and vincristine, elevated H3S10Ph is maintained for only 8–16 h prior to a rapid decline that is concomitant with an increase in caspase 3 and PARP cleavage, indicative of the onset of apoptosis. In addition, the accumulation of phosphorylated BubR1 is greatly decreased, suggesting a reduction in the number of cells in which the SAC is active. Altogether the results suggest that, in the belinostat-resistant cell lines, HDAC inhibition leads to failure of the vincristine-induced SAC and subsequent apoptosis. It is noteworthy that the submicromolar concentrations of belinostat that effectively curtail vincristine-induced mitotic arrest in the resistant cell lines do not efficiently activate the SAC as they do in the sensitive cell lines. This observation suggests that the HDAC targets that mediate SAC activation are distinct from those which cause SAC failure.

By three methods (Annexin V assay, Caspase 3 cleavage, and use of a pan-Caspase inhibitor) we established that the belinostat-vincristine combination treatment induces apoptotic cell death. Belinostat causes upregulation of Bim proteins in the presence or absence of vincristine, but the combination treatment causes a strong down-regulation of MCL-1 protein, thereby generating an environment more conducive to apoptosis. Upregulation of BIM was shown to increase the probability of cell death in mitosis in the presence of reduced MYC expression.⁵⁰ This is relevant to our study because we showed previously that belinostat strongly down-regulates MYC protein expression in DLBCL cell lines.¹⁵ MCL-1 is involved in regulation of both death in mitosis and death after mitotic slippage.^50-52 Mitotic degradation of MCL-1 can increase the probability of death in mitosis when MYC or NOXA is downregulated.^50,52 It can also hasten slippage during mitotic arrest caused by inhibitors of mitosis and increase the probability of apoptosis after slippage.⁵¹ While it is unclear if apoptosis induced by the belinostat-vincristine combination occurs during mitosis or after mitotic slippage, the concomitant downregulation of MCL-1 and upregulation of BIM could contribute to either.

Cancer cells exposed to different classes of mitosis-targeting drugs, including MTAs, respond in 2 main ways to a prolonged mitotic arrest, either dying during mitosis or undergoing mitotic slippage dependent on the cell line or the drug [reviewed in⁵³]. After mitotic slippage, cells have been shown to enter G1 in a tetraploid state after which they may become senescent, die by apoptosis, or progress through the cell cycle, all by mechanisms that are poorly understood [reviewed in⁵⁴]. In the prolonged presence of vincristine, a fraction of SUDHL8 cells slip out of mitosis in a tetraploid state and progress through the cell cycle yielding a population of cells that have 8N DNA content, including some that have entered mitosis. Interestingly, the induction of polyploidy by inhibitors of mitosis is associated with drug resistance,^23,46 and it has been proposed that the continued survival of polyploid cells may give rise to aneuploidy that promotes cancer progression.^46,53-56 Deng et al measured vincristine sensitivity in DLBCL cell lines and found that SUDHL8 cells were relatively resistant while SUDHL4 cells were much more sensitive.⁵⁷ We found that exposure to belinostat largely prevents the vincristine-induced accumulation of polyploid SUDHL8 cells by triggering apoptosis. Thus, HDAC inhibition enhances the cytotoxicity of vincristine by increasing the probability that apoptotic signaling is efficiently activated to reduce the number of cells that undergo mitotic slippage, evade cell death, and continue to cycle with an abnormal number of chromosomes. Our work suggests that HDACi might be used with MTAs and other drugs that inhibit mitotic progression to reduce the probability of resistance. Also, since tumors with aneuploidy tend to be resistant to MTAs, HDACi co-treatment might increase their sensitivity to the MTA-induced cytotoxicity. In fact, a very recent study showed that treatment of paclitaxel-resistant lung cancer cell line with a novel HDACi sensitized the cells to paclitaxel-induced cytotoxicity both in vitro and in vivo.⁵⁸

Synergy between taxanes and HDACi in inducing cytotoxicity has been previously demonstrated in vitro and in xenograft tumor models of ovarian, breast, and prostate cancer.^30,59-68 Belinostat has been found to be synergistic with taxanes in inducing apoptosis in prostate and ovarian cell lines and in clinical samples from ovarian tumors grown in organoid culture.^65,67 Although most of these studies did not explore mechanism, cytotoxicity induced by vorinostat and paclitaxel in breast cancer cell lines was accompanied by increased induction of G2/M arrest.⁶² A study of Hela cells showed that combining the hydroxamate HDACi, trichostatin A, with paclitaxel disrupted the association of BubR1 with kinetochores and reduced clonogenic survival.³⁰ Results were mixed in the few studies that combined vincristine with romidepsin, a Class I-selective HDACi. The combination of romidepsin and vincristine was synergistic in neuroblastoma cell lines ⁶⁹ but antagonistic in several leukemia and lymphoma cell lines, none of which were derived from DLBCL.⁷⁰ Surprisingly, we could not find any studies that combined vincristine with a pan-HDACi of the hydroxamate class in cancer cells.

The use of vincristine in patients is dose-limited due to toxicity. While HDACi are generally well-tolerated in humans, combining them a toxic drug such as vincristine may increase the risk of toxicity to normal cells. Our experiments with a non-transformed, immortalized human B cell-derived cell line demonstrated that exposure to both drugs had little effect on cell cycle distribution and did not induce PARP cleavage, showing that the belinostat/low dose vincristine combination was not cytotoxic. Thus, our results indicate that doses of MTAs lower than those needed for maximal cytotoxicity of tumor cells might be used effectively with HDACi, thereby reducing the potential for toxicity in normal cells. In support, Hwang et al reported that a dose of docetaxel lower than the maximally tolerated dose was effective in inhibiting growth of prostate xenograft tumors when combined with belinostat.⁶⁷ A Phase I study of solid tumors showed that the belinostat/paclitaxel combination is tolerable in humans and thus feasible for further clinical study.⁷¹ Recently approved nanoparticle formulations of vincristine and paclitaxel that increase efficacy and reduce toxicity ^72,73 should make them safer to combine with other drugs such as belinostat. Using a preclinical model of aggressive DLBCL the current study has demonstrated the potential of combining MTAs with belinostat for treatment of relapsed/refractory DLBCL. Our future studies will focus on evaluation of the MTA/belinostat combination in mouse models of DLBCL.

Materials and methods

Cell lines and reagents

DLBCL-derived cell lines were cultured as described previously.¹⁵ RPMI 1640 medium was supplemented with: 1.) 10% FBS and gentamicin for SUDHL4 and DB, 2.) 10% FBS, 1 mM sodium pyruvate, and gentamicin for OCILY19, 3.) 10% FBS, 1 mM glutamine, and gentamicin for SUDHL8, and 4.) 20% FBS and gentamicin for GM18564. All cell lines were maintained at 37°C in a humidified atmosphere containing 5% CO2 at densities between 2.0 × 10⁵ and 1.0 × 10⁶ cells/ml. All cell lines were treated with concentrations of belinostat that inhibited growth by 50% at 24 h as determined previously.¹⁵

Belinostat was provided by Spectrum Pharmaceuticals. Paclitaxel (S1150) was purchased from Selleck Chemicals. Vincristine (1714007) was purchased from Sigma. Antibodies against BubR1 (#4116s), PARP (#9542s), Caspase 3 (#9665s), and α-Tubulin (#2144s) were purchased from Cell Signaling Technologies. Antibodies against phosphorylated Ser10 Histone H3 (#06–570) were purchased from Millipore and antibodies against GAPDH (fl-335) from Santa Cruz Biotechnology.

Cell cycle, apoptosis analysis and quantitation of mitotic populations using flow cytometry

Cells were plated at a density between 2 × 10⁵ and 3 × 10⁵ cells/ml and cultured for 24 hours prior to treatment. For cell cycle analysis, 3 × 10⁶ cells per treatment condition were collected and washed once with cold 1X Dulbecco's modified phosphate-buffered saline (D-PBS) followed by fixation with ice cold 70% ethanol and centrifugation at 700 × g for 3 minutes. Cell pellets were resuspended in 0.5 ml cold PBS to which propidium iodide and RNase A were added to final concentrations of 40 μg/ml and 50 μg/ml, respectively. After incubation (30 minutes, 37°C) samples were analyzed by flow cytometry using a FACScanto II instrument (Becton-Dickinson). Data was analyzed using ModFit LT 3.0 (Verity Software House).

Apoptosis was analyzed using an Annexin V/Propidium iodide uptake assay kit (ENZO Life Sciences) that was supplemented with FITC-conjugated Annexin V (640906) antibody (Biolegend). Approximately 1 × 10⁶ cells per treatment condition were collected and processed according to manufacturer's specifications. Stained cells were analyzed within one hour of processing using the FACSanto II flow cytometer (Becton-Dickinson). All statistical analysis was conducted using Graphpad Prism Version 5.0 software.

For mitotic quantitation through H3S10Ph staining, approximately 1.5×10⁶ cells were pelleted, washed, and then fixed in 0.5 ml 1% formaldehyde in PBS for 10 min at 37°C. Cold methanol (4.5 ml) was added while vortexing for storage overnight at −20°C. Cells were then pelleted and resuspended in 0.5% BSA/PBS to block for 10 min. Cells were incubated with Alexa 488-conjugated anti-phos Ser 10 Histone H3 (Cell Signaling Technologies, #9708) at a 1:10 dilution for 1 h at room temperature in the dark then resuspended in cold PBS to which Propidium iodide and RNase A were added to final concentrations of 40 μg/ml and 50 μg/ml, respectively. After incubation for 30 minutes at 37 °C samples were analyzed by flow cytometry using BD LSRFORTESSA X-20 (BD Biosciences). Data was analyzed using FlowJo Version 9.9.

Western blotting

For generation of whole cell lysates, cells were collected after treatment, counted, washed with cold PBS, and resuspended at a ratio of 100 ul of 2x SDS- PAGE loading buffer [0.08 M Tris pH 6.8, 4% SDS, 20% glycerol, 0.1 M DTT, 0.004% Bromophenol Blue] per 1×10⁶ cells. Equal volumes of each sample were then used for Western blotting. Generation of whole cell extracts was carried out as described previously. Briefly, after washing in PBS, cell pellets were flash frozen in liquid nitrogen and stored −80° C. Cells were resuspended in RIPA buffer [150 mM sodium chloride, 1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 50 mM Tris, pH8.0, protease inhibitor cocktail (Roche) and phosphatase inhibitors (10 mM NaF, 25 mM β-glycerophosphate)] and protein concentration was determined by Bradford assay. Cellular proteins were separated using SDS-PAGE electrophoresis, transferred to 0.45um nitrocellulose, and immunoblotted. Antibody binding was visualized with chemilumiscence reagents (Supersignal West Pico and/or Femto, Pierce) and imaged using a Molecular Imager ChemiDoc XRS system (Biorad). Signals were quantitated using Imagelab version 5.1 software (BioRad).

Immunocytochemistry

Approximately 2.0 × 10⁷ cells per treatment condition were centrifuged at 100xg for 5 minutes. After washing with D-PBS cells were fixed overnight at 4°C in 10% neutral buffered formalin (NBF) which was then replaced with 70% ethanol. Fixed cells were paraffin embedded, sectioned into 3 micron slices, and mounted on slides. The slices were deparaffinized using xylene and bathed in decreasing percentages of ethanol (95%, 80% and 70%). Cells were stained manually according to the LSAB 2 HRP system (Dako). Briefly, antigen retrieval was performed using the Decloaker (Biocare Medical, Concorde, CA) at 95°C for 25 minutes. Cells were incubated overnight with either 1% bovine serum albumin in PBS or antibody (diluted 1:10000 in 0.5% BSA in PBS) against phosphorylated Ser10 Histone H3 followed by 3 × PBS washes. For visualization of reaction, cells were incubated with LSAB 2 HRP system (Dako). After washing 3x with PBS, cells were counter stained with hematoxylin for 3 minutes followed by cover slipping the slides using non-aqueous mounting media. Stained slides were imaged using a Leica DMI6000B inverted microscope with a Leica DCF450 color camera at 40X and 100X magnification. Acquisition was done using Leica LAS-AF 4.0 software (Leica Microsystems, Buffalo Grove, IL).

Disclosure of potential conflicts of interest

C.L.S. received financial support for basic research studies from Spectrum Pharmaceuticals which holds the commercial license for belinostat and provided the drug for this study. However, Spectrum Pharmaceuticals had no influence on the study goals, execution, or preparation of the manuscript.

Acknowledgements

The authors would like to acknowledge the staff of the Flow Cytometry Shared Resource at the Arizona Cancer Center for services rendered. We are grateful to members of the Lymphoma Research Group at the UA for helpful suggestions and critique during the progress of the study. This work was initiated by an award from the Arizona Biomedical Research Commission to C.L.S., and supported through a career development award and developmental project award to C.L.S. [Lymphoma SPORE (1 P50 CA-130805-05, PtdIns - R.I. Fisher)]. A.P.H. received salary support from a cancer biology training grant (T32CA009213) and the Cancer Center Support Grant (P30CA023074). K.B.R. received support from the Undergraduate Research Biology Program at the University of Arizona. Y.Z. received support from a Hyundai Hope on Wheels Young Investigator Award. Salary support for J.T. and M.S. was provided by a SWOG Development Award from the Hope Foundation to M.S. In addition, Spectrum Pharmaceuticals donated belinostat for the study and provided funding to C.L.S.

ORCID

Catharine L. Smith http://orcid.org/0000-0002-9875-4884