Steroid-free combination of 5-azacytidine and venetoclax for the treatment of multiple myeloma

Abstract

Multiple myeloma (MM) is an incurable plasma cell malignancy that, despite an unprecedented increase in overall survival, lacks truly risk-adapted or targeted treatments. A proportion of patients with MM depend on BCL-2 for survival, and, recently, the BCL-2 antagonist venetoclax has shown clinical efficacy and safety in t(11;14) and BCL-2 overexpressing MM. However, only a small proportion of MM patients rely on BCL-2 (approx. 20%), and there is a need to broaden the patient population outside of t(11;14) that can be treated with venetoclax. Therefore, we took an unbiased screening approach and screened epigenetic modifiers to enhance venetoclax sensitivity in 2 non-BCL-2 dependent MM cell lines. The demethylase inhibitor 5-azacytidine was one of the lead hits from the screen, and the enhanced cell killing of the combination was confirmed in additional MM cell lines. Using dynamic BH3 profiling and immunoprecipitations, we identified the potential mechanism of synergy is due to increased NOXA expression, through the integrated stress response. Knockdown of PMAIP1 or PKR partially rescues cell death of the venetoclax and 5-azacytidine combination treatment. The addition of a steroid to the combination treatment did not enhance the cell death, and, interestingly, we found enhanced death of the immune cells with steroid addition, suggesting that a steroid-sparing regimen may be more beneficial in MM. Lastly, we show for the first time in primary MM patient samples that 5-azacytidine enhances the response to venetoclax ex vivo across diverse anti-apoptotic dependencies (BCL-2 or MCL-1) and diverse cytogenetic backgrounds. Overall, our data identify 5-azacytidine and venetoclax as an effective treatment combination, which could be a tolerable steroid-sparing regimen, particularly for elderly MM patients.

Introduction

The median age of onset of multiple myeloma (MM), an aggressive cancer of terminally differentiated plasma cells, is approximately 69 years of age.¹ While triplet and quadruplet combinations of novel therapies, including immunomodulatory agents and cellular therapies, have improved the therapeutic landscape for MM, outcomes remain suboptimal. Survival rates vary by age and cytogenetic status.² Patients eligible for autologous stem cell transplant (ASCT) have a median overall survival (OS) of eight years,³ while for patients >75 years, this falls to five years.⁴ Therefore, there is an urgent need to identify tolerable targeted agents for the treatment of elderly MM patients.

Reliance on the anti-apoptotic proteins BCL-2, MCL-1 and BCL-XL has emerged as a vulnerability for MM cells.⁵ The anti-apoptotic BCL-2 proteins (BCL-2, MCL-1, BCL-XL, BFL-1) function by binding and inhibiting the activity of pro-apoptotic proteins (BIM, BAX, etc.).^6,7 Due to their function in inhibiting cell death, the anti-apoptotic proteins are attractive therapeutic targets. Previously, it was thought that the pro-survival protein MCL-1 was the main anti-apoptotic protein maintaining the viability of MM cells.⁸ However, selective MCL-1 inhibitors (AMG-176, S68345) are associated with potential cardiac toxicity, therefore limiting their clinical utility.^9,10 The selective BCL-2 inhibitor, ABT-199 / venetoclax (ven), is the first BH3 mimetic to be approved by the Food and Drug Administration (FDA) for the treatment of chronic lymphocytic leukemia (CLL) and acute myeloid leukemia (AML) in combination with a demethylating agent.^11-13

BCL-2 dependence in MM was initially identified in a subset of patients, characterized by the presence of a translocation in cyclin D1 (CCND1) t(11;14).¹⁴ This was confirmed across a panel of MM cell lines and primary MM samples; sensitivity to Ven was associated with t(11;14) and correlated with BH3 profiling.¹⁵ In the phase III BELLINI trial evaluating the combination of Ven, dexamethasone (Dex) and bortezomib in relapsed/refractory (R/R) MM, the patients with t(11;14) translocation were particularly sensitive.¹⁶ However, a proportion of patients without the t(11;14) translocation were also sensitive to Ven, specifically those with high expression of BCL-2.¹⁷ This highlights the need to identify combination treatments that could induce BCL-2 dependence in MM and broaden the patient population that could be treated with Ven, as only 16-24% of patients have a t(11;14) translocation.¹⁸ Venetoclax is FDA-approved for the treatment of AML in combination with the demethylating agent 5-azacytidine (5-Aza).¹³ 5-Aza is a cytosine analog that is incorporated into both DNA and RNA, where it acts as an epigenetic modifier inhibiting DNA methyltransferases, resulting in hypomethylation.¹⁹ In AML, Ven in combination with 5-Aza is well tolerated in elderly patients who were ineligible for intensive induction therapy.^13,20 Previously, it was shown that methylation regulated BCL-2 expression in mixed lineage leukemia (MLL).²¹ Whole exome sequencing analysis of MM revealed mutations in a series of epigenetic modifier genes, including UTX, MLL, MLL2, and HOX9.²² This led us to hypothesize that epigenetic modifiers may induce BCL-2 dependence in MM. Our aim was to perform an unbiased screening approach to identify epigenetic modifiers that could induce BCL-2 dependence in MM cells, thereby sensitizing MM cells to Ven, to expand the cohort of patients that could be treated with Ven.

Methods

For a more detailed description of the materials and methods used in the present study, see the Online Supplementary Appendix.

CD138+ cell isolation from primary multiple myeloma bone marrow samples

Primary MM samples were attained by informed consent from patients from Beaumont Hospital, Dublin, Ireland. Ethical approval was granted from the Beaumont Hospital and the RCSI ethics committee (Study N 19/32). For detailed methods of isolation, see the Online Supplementary Appendix.

Cell culture

Multiple myeloma cell lines (JJN3, KMS18, RPMI-8226, H929, MM1S, KMS27 and U266) were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum; KMS-12BM is supplemented with 20% fetal bovine serum (Gibco, Invitrogen, Carlsbad, CA, USA), 1% L-Glutamine (2 mM) (Gibco, Invitrogen, Carlsbad, CA, USA), and 1% v/v penicillin/ streptomycin (50 units/mL) (Gibco, Invitrogen, Carlsbad, CA, USA). Primary MM bone marrow stroma cells (MM-BMSC) were cultured in DMEM medium supplemented with 10% fetal bovine serum (Gibco, Invitrogen, Carlsbad, CA, USA), 1% L-Glutamine (2 mM) (Gibco, Invitrogen, Carlsbad, CA, USA), and 1% v/v penicillin/streptomycin (50 units/mL) (Gibco, Invitrogen, Carlsbad, CA, USA).

Dynamic BH3 profiling

For BH3 profiling, JJN3 cells were seeded at 3x10⁶ cells in a T25 flask for 24 hours (hr), cells were treated with 3 μM 5-Aza for 20 hr. Cells were resuspended in 100 μM of buffer (1% FBS 0.4% EDTA [2 mM] in PBS, final pH 7.4) and incubated with a panel of BH3 peptides for 60 min at 21^oC (Table 1). Details of the methods are provided in the Online Supplementary Appendix.

Statistical analysis

GraphPad Prism 9.0 software was used for all statistical analyses. Dose-response curves and IC₅₀ values were calculated using linear regression curve fit (Log inhibitor vs. normalized response, variable slope). Unless otherwise stated, results are expressed as the mean + Standard Error of the Mean (SEM) of 3 independent experiments. Webb’s fractional product method was used to calculate the synergistic activity of pairs of drugs / inhibitors denoted A and B on cell viability using the following equation:

Table 1.List of BH3 peptides and interactions with BCL-2 proteins.

where CI >1.0 indicates antagonistic activity

CI = 1.0 indicates additive interactions

CI <0.8 indicates strong synergy

Results

Epigenetic modifier screen to induce BCL-2 dependence and sensitivity to venetoclax

Taking an unbiased screening approach, we screened 21 epigenetic modifiers in 2 non-BCL-2 dependent MM cell lines to try and induce BCL-2 dependency and enhance sensitivity to Ven. We confirmed that JJN3 mainly relied on MCL-1 for survival, as they were most sensitive to AMG-176, a BH3 mimetic that selectively inhibits MCL-1.²³ In the BH3 profile, the A12 BH3 peptide that antagonizes MCL-1²⁴ caused a large reduction in mitochondrial membrane potential (Figure 1A).²⁵ The KMS18 cell line was mixed BCL-2- / BCL-XL-dependent, as evidenced by sensitivity to ABT-263 that antagonizes BCL-2, BCL-XL, and BCL-w,²⁶ and the BH3 peptides BAD and HRK induced loss of mitochondrial membrane potential (Figure 1B).²⁷ Table 2 contains the list of the epigenetic compounds that were screened in both the KMS18 and JJN3 cell lines using CellTiter-Glo®. The classes of epigenetic modifiers included histone deacetylase inhibitors (HDACi), histone methyltransferase inhibitors (HMTi), DNA methyltransferase inhibitors (DNMTi), histone acetylase inhibitors (HATi), and bromodomain and extra terminal inhibitors (BETi) (Table 2). The 21 epigenetic modifier drugs were screened alone or in combination with Ven in the 2 MM cell lines, at 2 doses: 100 and 500 nM (Figure 1C, D). Based on the initial screen, the top 12 hits were assessed for an ability to induce cell death, as assessed by Annexin V / propium iodide (PI), in combination with Ven in the JJN3 cell line (Online Supplementary Figure S2) and the KMS-18 cell line (Online Supplementary Figure S3). The hits were ranked based on changes in the area under the curve (AUC) ratio of the epigenetic modifier alone versus the epigenetic modifier and Ven (Figure 1E). The lead hits from the cell death analysis in both the JJN3 and KMS18 cell lines were the histone methyltransferase inhibitors (JIB-04 and GSK-J4), the DNA methyltransferase inhibitor (5-aza), and the HDAC inhibitor (panobinostat) (Figure 1F, G). However, as evidenced by the dose-response curves (Online Supplementary Figures S2, S3), the top 4 hits all showed similar enhanced cell death and changes in IC₅₀ values upon combination treatment.

Figure 1.Screening epigenetic modifiers in combination with venetoclax in multiple myeloma cell lines. Area under the curve (AUC) of the dose response to BH3 mimetics ABT-199, ABT-263, WEHI-539, and AMG-176 is plotted following 24-hour (hr) treatment as measured by Annexin V / propidium iodide (PI) staining. BH3 profiles of (A) JJN3 and (B) KMS18. Mitochondrial membrane potential was measured using JC-1 following exposure to BH3 peptides and BH3 mimetics over 180 minutes (min). Median ± Standard Error of Mean of 3 independent experiments is shown. (C) JJN3 and (D) KMS18 cells were treated for 24 hr with epigenetic (Epi) modifiers ± venetoclax (Ven). Cells were treated with 2 doses of epigenetic modifier alone, 100 nM and 500 nM, or in combination with 1 μM Ven JJN3 cells or 500 nM Ven in KMS18 cells for 24 hr. Cell viability was assessed following treatment using Cell Titer Glo^® and data were graphed using GraphPad prism. The 12 epigenetic modifier hits from (F) JJN3 and (G) KMS18 were assessed for cell death by Annexin V / PI staining. Cells were treated with 30 nM, 100 nM, 300 nM, 1 μM and 3 μM of each epigenetic modifier alone, and 1 μM Ven in JJN3 cells and 500 nM Ven in KMS18 cells for 24 hr. The dose-response curves were graphed in Graphpad. Ratio of AUC values, as shown in (E), were calculated and graphed for each drug in combination with Ven.

Table 2.List of epigenetic modifiers.

Figure 2.Assessing the effect of bone marrow stroma cells on the combination of 5-azacytidine and venetoclax in multiple myeloma cell lines. (A) Cell viability was assessed in JJN3 cells using Annexin V / propidium iodide (PI) staining following 24-hour (hr) treatment with dose response of venetoclax (Ven) ± 5-azacytidine (5-Aza) 3 μM. (B) Live cell confocal imaging of JJN3 cells following treatment with DMSO, 3 μM 5-Aza ± 1 μM Ven for 24 hr. Mitochondria were stained with PKMO and DNA stained with Picogreen; 5 images were taken per treatment condition, with images representative of 3 independent experiments. (C) The combination treatment was assessed in a panel of multiple myeloma (MM) cell lines by measuring Annexin V / PI positivity by flow cytometry following 24 hr of treatment. Sensitivity was expressed as IC₅₀ μM for Ven alone or Ven + 5-Aza as determined by GraphPad. (D) Workflow of the co-culture experiments. HS-5 cells or MM bone marrow stromal cells (BMSC) from an MM patient sample were seeded for 24 hr. MM cells were co-cultured for 24 hr. and then treated with drugs for 24 hr before assessment by flow cytometry. (E) JJN3 cells were seeded alone (orange) or co-cultured with HS-5 cells for 24 hr (maroon). Cells were then treated with either 5-Aza (1 μM or 3 μM) alone or combined with 1 μM Ven (shown by the plus symbol) for 24 hrs. % cell death of tumor cells was assessed by Annexin V / PI staining on the flow cytometer. (F) The same experiment was carried out for the MM1S cells, seeded alone (purple) or co-cultured with HS-5 (maroon) for 24 hr before treatment. Cells were then treated with either 5-Aza (1 μM or 3 μM) alone or combined with 1 μM Ven (shown by the plus symbol) for 24 hr. (G) The same experiment as described for (E) was carried out with JJN3 cells co-cultured with MM-BMSC cells. Median ± Standard Error of Mean from 3 independent experiments is shown. Two-way ANOVA with Bonferroni multiple comparison test was used to calculate P values. ns: not significant; *P≤0.05, **P≤0.01, ***P≤0.001.

5-azacytidine enhanced venetoclax-induced cell death even in the presence of bone marrow stromal cells

While the lead hits from the screen were the histone demethylase inhibitors, those compounds have only been assessed pre-clinically. Excitingly, the combination of 5-Aza and Ven has not been previously tested in MM; it has been shown to have acceptable tolerability in elderly AML patients and is FDA-approved as a combination treatment.¹³ As shown in the JJN3 cells, while Ven caused very little cell death alone (IC₅₀ 14.2 μM), when combined with 5-Aza, there is a log fold change in IC₅₀ value (IC₅₀ 0.4 μM) (Figure 2A). This is also evident from live cell imaging of the JJN3 cells, with enhanced cell killing in the combination treatment and loss of mitochondrial staining with PK Mito Orange (PKMO)²⁸ (Figure 2B). Interestingly, although Ven did not induce cell death, it did cause a phenotypic change to the mitochondrial structure, as determined by PKMO staining. Next, we assessed the combination of Ven + 5-Aza in a panel of 5 additional MM cell lines (Online Supplementary Figure S4A-F). Remarkably, across the panel of cell lines tested, 5-Aza significantly altered the IC₅₀ value when combined with Ven in all 5 cell lines (Figure 2C). The KMS27 cell line was most sensitive to Ven treatment alone, probably because this cell line has a t(11;14) translocation, but the addition of 5-Aza reduced the IC₅₀ by half. The combination index score for synergy is represented as a heatmap in Online Supplementary Figure S4G, H. We also assessed the reverse combination measuring the effect of Ven treatment on a 5-Aza dose-response curve, with 4/5 of the cell lines showing major changes in IC₅₀ values upon the addition of Ven (Online Supplementary Figure S4J-N), with combination index scores of synergy graphed as a heatmap (Online Supplementary Figure S4 O). The complex interplay between MM cells and the bone marrow microenvironment has been shown to influence cell behavior, particularly MM survival and drug resistance.²⁹

To assess the impact of bone marrow stroma cells on the 5-Aza and Ven combination, we performed co-culture experiments using HS-5, a bone marrow stroma cell line, and primary MM bone marrow stromal cells (MM-BMSC) (Figure 2D). Surprisingly, co-culture of JJN3 cells with HS-5 cells conferred significant increase in cell death with 5-Aza and Ven, compared to JJN3 cells alone (Figure 2E). Of interest, when MM1S cells were co-cultured with HS-5 cells, they were slightly less sensitive to the 5-Aza and Ven combination (Figure 2F). Next, we co-cultured JJN3 cells with MM-BMSC isolated from a MM patient bone marrow sample (Figure 2G). The JJN3 cells were still sensitive to 5-Aza and Ven combination when co-cultured with the MM-BMSC. In summary, while the MM1S cells benefit from a slight protective effect and the JJN3 cells are slightly sensitized following co-culture with the HS-5 cells, 5-Aza and Ven is still an effective treatment combination in the presence of bone marrow stroma cells across multiple cell lines with diverse anti-apoptotic dependencies.

5-azacytidine induces NOXA through the integrated stress response antagonizing MCL-1, sensitizing multiple myeloma cells to venetoclax treatment

To understand the mechanism of action of how 5-Aza may enhance sensitivity to Ven in MM cell lines, a series of approaches were taken. Firstly, we used dynamic BH3 profiling to determine changes in BH3 peptide-induced cytochrome c release from mitochondria, following 5-Aza treatment (Figure 3A). As is clear from the BH3 profile, there was a statistically significant increase in the amount of cytochrome c released following the addition of the A12 BH3 peptide, that selectively binds MCL-1, following treatment with 5-Aza. In addition, direct Ven treatment, as part of the BH3 profile on permeabilized cells, caused a statistically significant increase in the release of cytochrome c, following 5-Aza treatment in the JJN3 cells (Figure 3A). Next, we confirmed the dynamic BH3 profile by assessing the pro-apoptotic proteins that might be interacting with MCL-1 using immunoprecipitation. Following 5-Aza treatment, there was an increase in the amount of both BIM EL and NOXA bound to MCL-1 (Figure 3B). Previously, it was shown in AML cells, that 5-Aza induced NOXA by the integrated stress response.³⁰ Indeed, we found a similar induction of NOXA following 5-Aza treatment in the JJN3 cells, which was even greater in the combination of 5-Aza and Ven (Figure 3C). We also assessed if this induction of NOXA was dependent on the integrated stress response. Pre-treatment with ISRIB, which blocks the integrated stress response by targeting eIF1B,³¹ also blocked the induction of the NOXA protein following the combination treatment (Figure 3C). Ven alone induces a small amount of NOXA protein, but it enhances the amount of NOXA induced by 5-Aza (Figure 3C). We confirmed that ISRIB did not alter eIF2α phosphorylation but prevented the induction of activating transcription factor 4 (ATF-4) by 5-Aza and 5-Aza + Ven (Figure 3D). Confirming the importance of the integrated stress response in the cell death induced by Ven + 5-Aza, there was a significant reduction in cell death upon pre-treatment with ISRIB (Figure 3E). There is evidence in the literature that 5-Aza can induce dsRNA, through reactivation of endogenous retrovirus.³² Therefore, we assessed the amount of dsRNA in JJN3 cells treated with 5-Aza, Ven or the combination (Figure 4A). While all treatments induced some dsRNA, the combination of Ven + 5-Aza induced the most significant amount, as assessed by pixel count (Figure 4B). Next, we assessed if the knockdown of the dsRNA sensor PKR, which can activate the integrated stress response, could protect from the cell death induced by the combination treatment. Knockdown of PKR significantly reduced the cell death induced by the combination treatment (Figure 4C). The densitometry on PKR knockdown is shown in Online Supplementary Figure S5C. Validating the importance of NOXA in the cell death mechanism, siRNA knockdown of PMAIP1 (NOXA gene), also significantly protected against the combination induced cell death (Figure 4D). Similarly, densitometry for NOXA knockdown is shown in Online Supplementary Figure S5D. A proposed model is shown in Figure 4E. Combined, the data suggest the mechanism of synergy following 5-Aza + Ven is due, at least in part, to an increase in dsRNA, activating PKR, activating the integrated stress response, with an increase in eIF2α phosphorylation, inducing NOXA protein that antagonizes MCL-1. By antagonizing MCL-1, the cells are more sensitive to Ven (Figure 4E).

Figure 3.5-azacytidine increases NOXA expression through the integrated stress response antagonizing MCL-1. (A) The effect of 5-azacytidine (5-Aza) 3 μM treatment on JJN3 cells after 24 hours (hr) was assessed by dynamic BH3 profiling (see Methods section). (B) JJN3 cells were treated with 5-Aza 3 [xM for 24 hr. The association of NOXA and BIM with MCL-1 was determined by immunoprecipitation and western blot analysis. (C) JJN3 cells were treated with 3 μM 5-Aza; venetoclax (Ven) 1 [xM; (3 μM 5-Aza + Ven1 μM); (3 μM 5-Aza + Ven1 μM + 500 nM ISRIB) for 24 hr and the expression of BCL-2, MCL-1, NOXA, DNMT1 and β-actin were determined by western blot analysis. (D) Cells were treated as in (C) for 6 hr and expression of p-eIF2α, total eIF2α, ATF4, and p-actin was measured by western blot. (E) Cell viability was assessed in JJN3 cells by Annexin V / propidium iodide (PI) staining following treatment with 3 μM 5-Aza; (3 μM 5-Aza + Ven1 μM); (3 μM 5-Aza + Ven1 μM + 500 nM ISRIB) for 24 hr. Mean of 3 independent experiments ± Standard Deviation is shown. **P≤0.01, ***P≤0.001, ****P≤0.0001

Figure 4.5-azacytidine increases dsRNA and knockdown of PKR protects against the combination treatment. (A) JJN3 cells were treated with 3 [xM 5-azacytidine (5-Aza), 1 [xM venetoclax (Ven) and (3 μM 5-Aza + 1 [xM Ven), for 24 hours (hr) before confocal imaging with DAPI stain for nucleus, MitoTrackerTM (deep red FM) for mitochondria and dsRNA for double-stranded RNA. Images are representative of 3 independent experiments. (B) Pixel count for dsRNA is graphed for the 3 independent experiments for the treatments described in (A). (C) Cell viability was assessed using Annexin V / propidium iodide (PI) staining by flow cytometry following control or siRNA to PKR with the following treatments 1 μM Ven; 3 μM 5-Aza; (3 μM 5-Aza + 1 μM Ven). Western blot shows the expression of PKR, with actin as a loading control. (D) Cell viability was assessed using Annexin V / PI staining by flow cytometry following control or siRNA to NOXA with the treatments listed in (D). Western blot shows the expression of NOXA, with actin as a loading control. (E) Induction of dsRNA activating PKR, knockdown of PKR, NOXA with siRNA or pre-treatment with ISRIB all inhibited the cell death induced by 5-Aza + Ven. *P≤0.05 **P≤0.01.

Figure 5.Dexamethasone does not enhance venetoclax + 5-azacytidine cell death and induces T-cell death. (A) JJN3 cells were treated with 1 μM venetoclax (Ven); 3 μM 5-azacytidine (5-Aza); 1 μM dexamethasone (Dex); (1 μM Ven + 1 μM Dex); (1 μM Ven + 3 μM 5-Aza); (1 [xM Ven + 3 [xM 5-Aza + 1 [xM Dex) for 24 hours (hr). Cell viability was measured using Annexin V/ propidium iodide (PI) staining. Mean of 3 independent repeats + Standard Deviation is shown. (B) MMIS cells were treated with 1 [xM Ven; 3 μM 5-Aza; 1 uM Dex; (1 μM Ven + 1 μM Dex); (1 μM Ven + 3 μM 5-Aza); (1 μM Ven + 3 μM 5-Aza + 1 μM Dex) for 24 hours (hr). Cell viability was assessed as in (A). (C) KMS12BM were treated with 0.5 μM Ven; 3 μM 5-Aza; 1 μM Dex; (0.5 μM Ven + 3 μM 5-Aza); (0.5 μM Ven + 3 μM 5-Aza + 1 [xM Dex) for 24 hr. Cell viability was assessed and graphed as in (A). (D) Gating strategy used for identifying CD3 T cells. (E) Cell viability was assessed in CD3⁺ T cells isolated from peripheral blood mononuclear cells using Annexin V / PI staining following 16-hr treatment with 3 μM 5-Aza; 300 nM Ven; 1 μM Dex; (300 nM Ven + 3 μM 5-Aza); (1 μM Dex + 300 nM Ven). Mean of 3 independent experiments is shown. (F) Cell viability was assessed in CD3⁺ T cells following treatment with Ven at 100 nM and 300 nM with and without 1 [xM Dex. Graphed are the mean of 2 independent experiments ± SD. (G, H) Intracellular BCL-2 staining was measured in T cells following 16 hr treatment with 300 nM, 1 [xM and 3 [xM Dex. One-way ANOVA was performed to determine P values. ns: not significant; *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001.

Assessing if a steroid enhances the cell death induced by 5-azacytidine and venetoclax in multiple myeloma and immune cells

It is standard practice for a steroid, such as Dex, to be added to a combination treatment regimen in MM.³³ Therefore, we assessed the combination of Ven, 5-Aza and Dex in a panel of MM cells. Interestingly, we did not find any significant increase in cell death upon the addition of 1 μM Dex to the combination of Ven and 5-Aza, across 3 different MM cell lines (Figure 5A-C). Indeed, the combination of Ven and 5-Aza caused significantly more cell death than Dex (1 μM) and Ven in both the JJN3 and MM1S cell lines (Figure 5A, B). Next, we measured the effect of the combination treatment on immune cells, Ven has been shown to induce cell death in B³⁴ and T cells.³⁵ Using peripheral blood mononuclear cells from healthy donors, we treated the cells ex vivo for 16 hr and measured cell death of CD3⁺ T cells by annexin V/PI staining (Figure 5D). In the CD3⁺ T cells, 5-Aza caused a minimal amount of cell death alone. Ven induced some cell death in CD3⁺ T cells, as was found previously in mouse CD8⁺ T cells,³⁵ and additive cell death was seen with the combination treatment of 5-Aza and Ven (Figure 5E). The combination of Dex and Ven caused slightly more cell death, which was significant, than Ven and 5-Aza (Figure 5E). Dex combined with Ven caused a dose-dependent increase in T-cell death (Figure 5F). Potentially, the reason for this enhanced cell death is due to Dex inducing a BCL-2 expression in T cells. There was a dose-dependent increase in BCL-2 expression following Dex treatment, as measured by intercellular BCL-2 staining by flow cytometry (Figure 5G, H). Our data potentially suggest that the addition of a steroid may induce more cell death in the immune cells while having little effect on the tumor cell killing, at least ex vivo. Based on this result, there is a potential to use a steroid-sparing regimen to test the combination of Ven/5-Aza in the clinic. This steroid effect on T cells may also have implications for the efficacy of many other therapies in MM.

Venetoclax and 5-azacytidine show enhanced killing of primary multiple myeloma samples irrespective of stage or cytogenetics

As the combination treatment had only been assessed in MM cell lines, we next wanted to confirm our findings that the combination of 5-Aza and Ven was effective in heterogeneous MM patient samples. Following isolation of CD138⁺ patients’ cells, the cells were seeded and treated with either increasing doses of BH3 mimetics for 16 hr or the combination of 5-Aza and Ven and cell death was measured using Annexin V/PI staining on the flow cytometer (Figure 6A, Online Supplementary Figures S6A-K, S7A-K). Of the 11 primary MM bone marrow samples analyzed ex vivo (Figure 6A), 8 were non-t(11;14) patients (Online Supplementary Table S1). The combination of 5-Aza and Ven showed enhanced cell killing in 6 out of the 8 non-t(11;14) (Figure 6B) and 2 out of 3 t(11;14) samples (Figure 6C), with 6 samples showing a CI score of less than 1 (Online Supplementary Figure S6A-K). Patient cytogenetics were not known at the time of Ven treatment. The dose of Ven used induced substantial cell death alone in the t(11;14) patient samples, with a statistical difference in the ex vivo sensitivity of non-t(11;14) samples versus the t(11;14) samples (Figure 6D). These data highlight the use of BH3 mimetic profiling as a biomarker to identify patients who may respond to Ven. There was a diverse anti-apoptotic dependence across the MM patient sample (Online Supplementary Figure S6). Four of the primary samples CD138⁺ patient cells were more BCL-2 dependent, 4 were more MCL-1 dependent, and 3 samples were a mix, with the BCL-XL dependent samples showing the lowest cell death following combination treatment (Figure 6E). Lastly, there was no statistical difference in the combination treatment of Ven and 5-Aza in upfront versus relapsed patient samples (Figure 6F) or in patients with normal or altered cytogenetics (Figure 6G). Importantly, here we have shown for the first time that 5-Aza and Ven are effective at killing primary MM samples ex vivo, over-coming high-risk cytogenetic features and multiple lines of prior therapies.

Discussion

Multiple myeloma is a disease with a diverse genetic background and, to date, the development of molecularly targeted therapies has proven elusive. Our aim was to take an unbiased screening approach to identify epigenetic modifiers that could induce BCL-2 dependence in MM cells to broaden the patient population that could benefit from treatment with Ven.¹⁵ The DNA methyltransferase inhibitor 5-Aza emerged as one of the top hits from our screen. Ven in combination with 5-Aza has never been tested in MM, even though it is an approved combination regimen in AML.³⁶

Figure 6.Venetoclax and 5-azacytidine in combination enhance cell death in primary multiple myeloma samples ex vivo. Cell viability was assessed using Annexin V / propidium iodide (PI) staining following 16-hour (hr) treatment. (A) Workflow of isolating CD138⁺ cells from patient (Pt) bone marrow samples and treating them ex vivo. (B) Summary of the response across the 8 non-t(11;14) patients to 3 μM 5-azacytidine (5-Aza), 300 nM venetoclax (Ven), or combination. (C) Summary of the response across 3 t(11;14) patients to 5-Aza, Ven or in combination. (D) Response of non-t(11;14) versus t(11;14) patients to 300 nM Ven following 16-hr treatment. Summary of response to 5-Aza and Ven in 11 patients across (E) anti-apoptotic dependencies, (F) diagnosis, and (G) cytogenetics; ns: not significant; **P≤0.01, ***P≤0.001.

We identified a mechanism of synergism by which 5-Aza in combination with Ven induces NOXA protein inhibiting MCL-1, sensitizing MM cells to Ven. The induction of NOXA is reliant on the integrated stress response, with evidence of phosphorylation of eIF2-α. A similar mechanism was identified for 5-Aza in AML, ATF4 was responsible for inducing NOXA expression.³⁰ Importantly, we demonstrated that PMAIP1 (NOXA) knockdown partially rescued the cell death induced by the combination treatment, as did inhibition of the integrated stress response with ISRIB. We also showed for the first time, the translational potential of 5-Aza and Ven in combination, which demonstrated strong cell killing effect in primary CD138⁺ MM cells ex vivo, irrespective of cytogenetic background.

In the case of AML, either 5-Aza or decitabine were both effective at enhancing the response to Ven treatment.^13,20 In contrast, in our study, while 5-Aza enhanced the cell death induced by Ven, decitabine did not cause a similar enhanced cell death. Decitabine is a deoxyribonucleotide and is incorporated into DNA, whereas 5-Aza is a ribonucleoside and is incorporated into both DNA and RNA.^19,37 Potentially, this difference could explain why 5-Aza enhances cell death in MM and may suggest the integration into mRNA may lead to activation of the integrated stress response. A series of studies have also shown evidence that 5-Aza acts through activation of endogenous retroviral elements (ERV), inducing viral mimicry and activating the dsRNA sensing pathways (e.g., MDA5, TLR3).^32,38 We showed that the combination treatment induced dsRNA in the cells, and that knockdown of the ds-RNA sensor PKR partially rescued the cell death induced by the combination. We did not investigate if the dsRNA also activates RIG-1 to signal through the mitochondrial antiviral-signaling protein (MAVS), to induce type 1 interferons and proinflammatory cytokines.^32,39 The combination of Ven + 5-Aza works very effectively in the clinic in AML.^13,36 Activation of the dsRNA pathway by the combination treatment, as shown in our data, could also activate an immune response; this could potentially play an important role in the mechanism of action in vivo. However, this was not assessed in this study. Steroid utilization in MM is standard practice.⁴⁰ Therefore, we assessed if the addition of the steroid Dex enhanced the cell death induced by the combination treatment (Ven + 5-Aza). We did not detect any enhanced killing in 3 different MM cell lines, with diverse anti-apoptotic dependence, following the addition of Dex to the combination. The combination of 5-Aza + Ven was more effective at killing than Dex + Ven, which was previously identified as an effective combination in MM.^41,42 Potentially, our data may help to explain the latest phase III CANOVA trial data with Ven and Dex in t(11;14) patients, which did not reach a significant endpoint in comparison to pomalidomide + Dex, despite evidence of longer progression-free survival.⁴³ Next, we assessed the combination on immune cell killing ex vivo on T cells, as it was previously identified that mature T cells express high BCL-2^{2 7}, ^{4 4} and are sensitive to Ven treatment alone.^35,45

We found a similar induction of cell death in human CD3 T cells with Ven treatment alone, and the addition of 5-Aza did enhance cell killing. However, Dex + Ven caused significantly more death of the T cells than 5-Aza + Ven, and Dex induced BCL-2 expression in the CD3 cells. These data would suggest that a steroid has little therapeutic benefit, as it did not enhance the MM cell killing; but it may enhance the killing of immune cells, potentially limiting the capacity of immune cells to kill tumor cells.¹⁷ In a randomized trial comparing lenalidomide with high-dose versus low-dose Dex, the low-dose Dex regimen was associated with better short-term overall survival.⁴⁶ Indeed, a high-dose Dex regimen was associated with toxicities, particularly in frail, elderly patients. In a recent abstract, the results of a phase III trial comparing daratumumab and lenalidomide without longterm Dex to lenalidomide and Dex were described.⁴⁷ This trial showed that the overall response rates were higher in the Dex-sparing regimen. This study highlighted that there is a case to not add Dex to MM regimens, particularly in frail, elderly patients with MM, in the situation where the patient is receiving highly effective therapy.

To assess the translational relevance of 5-Aza and Ven, we screened the combination in a panel of primary MM CD138⁺ cells from MM patients, with different cytogenetic backgrounds. In total, 11 patient samples were assessed: 4 were MCL-1 dependent, 3 were mixed anti-apoptotic dependent, and 4 samples were BCL-2 dependent. The 2 samples that did not respond to the combination treatment had a higher sensitivity to the BCL-XL antagonist. Given the mechanism of the synergy described, induction of the NOXA protein binding to MCL-1 could potentially explain the lack of response in these 2 patient samples. Of the 4 samples that were BCL-2 dependent, it was later revealed by cytogenetics that 3 had t(11;14) translocation. Gomez-Bougie et al. looked at the anti-apoptotic dependence in MM samples.⁴⁸ They saw that BCL-2 dependence was mostly found in the t(11;14) subgroup and in 35% of the samples there was a co-dependency on either BCL-2/MCL-1 or BCL-XL/MCL-1. While we only assessed a small cohort of samples, the results corroborated previous studies, confirming that there are diverse anti-apoptotic dependencies in MM primary samples.²⁵ In our study, we used patient-specific sensitivity measurements by carrying out ex vivo functional cell death measurements of primary MM samples treated with BH3 mimetics to determine the anti-apoptotic dependence. Our group has previously identified BCL-2 dependence in a patient with secondary plasma cell leukemia.⁴⁹ BH3 profiling can also be used to identify resistance mechanisms to Ven, including switching of anti-apoptotic dependence.^50-53 An important study by Gupta et al. used ex vivo assessment of MM patients’ samples to Ven as an indicator of PFS.⁵⁴ As we have also shown here, assays such as precision testing and BH3 profiling are dynamic and give a functional measurement of cellular response to treatment; as such, they can be useful for identifying sensitive patients a priori.

Collectively, our data identify 5-Aza and Ven as an effective therapeutic combination for the treatment of MM, including in patients with non-t(11;14) disease with high-risk cytogenetics, such as 17p and 1q amplification. This combination has the potential to be effective across a broad patient population and should be considered in a steroid-free regimen for assessment in a clinical trial setting.

Footnotes

• Received June 29, 2023
• Accepted March 8, 2024

Correspondence

^#SGl and TNC contributed equally as senior authors

T. Ní Chonghaile tnichonghaile@rcsi.com

Disclosures

TNC received research funding from AbbVie. JQ has received Honoraria from Janssen and Takeda, and travel funding from Roche. MOD has no competing interest but has employment and equity in ONK Therapeutics. All the other authors have no conflicts of interest to disclose.

Funding

The authors would like to thank the following funding agencies for their support: the Leukemia Research Foundation (LRF-RCSI-2020) and Science Foundation Ireland (19/FFP/6461) (to TNC) and Irish Research Council (IRCLA/2022/2822) Breakthrough Cancer Research (BCR-RCSI-2020) (to LF).