Combination of menin and kinase inhibitors as an effective treatment for leukemia with NUP98 translocations

Abstract

Chromosomal translocations of the nucleoporin 98 (NUP98) gene are found in acute myeloid leukemia (AML) patients leading to very poor outcomes. The oncogenic activity of NUP98 fusion proteins is dependent on the interaction between Mixed Lineage Leukemia 1 and menin. NUP98-rearranged (NUP98-r) leukemia cells also rely on specific kinases, including CDK6 and/or FLT3, suggesting that simultaneous targeting of these kinases and menin could overcome limited sensitivity to single agents. Here, we found that combinations of menin inhibitor, MI-3454, with kinase inhibitors targeting either CDK6 (Palbociclib) or FLT3 (Gilteritinib) strongly enhance the anti-leukemic effect of menin inhibition in NUP98-r leukemia models. We found strong synergistic effects of both combinations on cell growth, colony formation and differentiation in patient samples with NUP98 translocations. These combinations also markedly augmented anti-leukemic efficacy of menin inhibitor in Patient Derived Xenograft models of NUP98-r leukemia. Despite inhibiting two unrelated kinases, when Palbociclib or Gilteritinib were combined with the menin inhibitor, they affected similar pathways relevant to leukemogenesis, including cell cycle regulation, cell proliferation and differentiation. This study provides strong rationale for clinical translation of the combination of menin and kinase inhibitors as novel treatments for NUP98-r leukemia, supporting the unexplored combinations of epigenetic drugs with kinase inhibitors.

Introduction

Chromosomal translocations involving the nucleoporin 98 (NUP98) gene are found in pediatric and adult AML patients, representing one of the most common genetic alterations in pediatric AMLs [1–3]. The presence of NUP98 translocations in leukemia patients confers poor clinical outcome, with 4-year survival rates of ~10-30% [1, 4–6], demonstrating a need for new therapies. NUP98 translocations lead to the expression of NUP98 fusion proteins, which retain the N-terminal domain of NUP98 fused with one of ~30 fusion partners, with NUP98-NSD1 and NUP98-HOXA9 found most frequently in AML [1]. While NUP98 fusions represent potent oncogenes, accompanying mutations (e.g. FLT3) worsen the prognosis [1, 7–9].

Despite a variety of fusion partners, there is a subset of genes commonly upregulated by NUP98 fusions, including HOXA/B cluster genes and MEIS1 [1, 4, 10, 11]. Previous studies demonstrated that NUP98 fusions interact with the Mixed Lineage Leukemia 1 (MLL1 also known as KMT2A) protein complex and colocalize with MLL1 on Hoxa/b cluster genes [10]. Recent work also validated an important role of protein menin, which directly interacts with MLL1, in NUP98-rearranged (NUP98-r) leukemia, and genetic inactivation or small molecule inhibition of menin resulted in anti-leukemic effects in NUP98-r leukemia models [12, 13]. Menin is also an oncogenic co-factor in MLL1-rearranged (MLL1-r) [14–16] and NPM1-mutated (NPM1-mut) [17] leukemias, both are associated with upregulated HOXA/B and MEIS1 genes.

Our group pioneered development of small molecule inhibitors of the menin-MLL1 interaction [18–22], including MI-3454 [23, 24] that is utilized in this study. MI-3454 represents a close structural analog of KO-539 (Ziftomenib), which is currently in phase II clinical trials in AML patients with NPM1 mutations (NCI: NCT04067336) [25, 26]. Interestingly, recent studies report anti-leukemic activity of another menin inhibitor, VTP50469, in animal models of NUP98-r leukemia, but long-lasting remission was not achieved, suggesting a need for combinations [12, 13].

Here, we explored the effect of combining our menin inhibitor MI-3454 with kinase inhibitors targeting either CDK6 or FLT3 to enhance the anti-leukemic effects of menin inhibition in NUP98-r leukemia. Because the role of CDK6 was validated in NUP98-r leukemia [27], and FLT3 mutations frequently co-exist with NUP98 fusions [1], we rationalized that simultaneous inhibition of these kinases and menin could be therapeutically beneficial. We found strong synergistic effects upon combining MI-3454 with either Palbociclib (CDK4/6 inhibitor) or Gilteritinib (FLT3 inhibitor) in pre-clinical models of NUP98-r leukemia, including markedly improved survival over single agents in Patient Derived Xenograft (PDX) models. Despite targeting different kinases, both combinations affected the same pathways relevant to leukemogenesis, supporting similar transcriptional regulation to enhance anti-leukemic effects of menin inhibitor. Since CDK4/6 and FLT3 inhibitors are FDA approved and menin inhibitors are in advanced clinical trials [26], our study provides strong rationale for clinical translation of these combinations as novel treatments for NUP98-r leukemia patients. This work provides an attractive avenue for unexplored combinations of epigenetic inhibitors, such as menin inhibitors, with drugs targeting kinases.

Methods

Cell lines and primary patient samples

Bone marrow cells (BMCs) with NUP98-NSD1, NUP98-HOXA9 or HOXA9/MEIS1 were prepared as described before [18]. Patient samples NUP98_258, NUP98_322 and NUP98_299 were received from Columbia University. NUP98_1055 sample was received from Stem Cell and Xenograft Core at the University of Pennsylvania.

Cell viability assays

BMCs with NUP98-r were plated at 1x10⁴ -1x10⁵ cells/mL and treated with 0.25% DMSO or MI-3454 for 14 days, with media change every 3-4 days before adding MTT reagents and OD readout at 570 nm using PHERAstar BMG.

Growth curve assays and flow cytometry analysis

Human primary AML samples were plated at 2.5x10⁵ cells/mL in MethoCult^™ H4435 medium for up to 12 days after addition of compounds or DMSO. Media was changed at day 6, viable cells were restored to the initial density and re-supplied with compounds. For flow cytometry analysis, 1x10⁵ cells were collected, washed in PBS, and stained with antibodies.

Colony forming assay

Primary cells with NUP98-r were mixed with DMSO or compounds and plated at 10,000-50,000 cells/mL on MethoCult^™ 4435. Colonies were counted at days 10-14, followed by colony replating on fresh methylcellulose medium.

Gene expression studies

RNA was prepared from NUP98-r leukemia cells treated for 7 days using RNeasy Mini Kit (Qiagen) and submitted for RNA sequencing.

PDX experiments

For survival, 6-8 weeks NSGS female mice were transplanted with NUP98_258 or NUP98_322 samples. Treatment was initiated 15-23 days post-transplantation and mice were monitored for leukemia progression by analysis of hCD45+ cells in peripheral blood. Survival or level of hCD45⁺ cells was used to assess treatment efficacy.

Data Sharing.

RNAseq data are available at GEO under accession number GSE246783.

Additional methods can be found in supplemental materials.

Results

MI-3454 demonstrates anti-leukemic effect in NUP98-r leukemia models

We previously demonstrated potent anti-leukemic activity of our menin inhibitor, MI-3454, in MLL1-r and NPM1-mut leukemia models [23, 24]. Here, we first assessed the activity of MI-3454 as a single agent in NUP98-r leukemia models. Since human leukemia cell lines harboring NUP98 translocations are not available, we transformed murine bone marrow cells (BMC) with either NUP98-NSD1 or NUP98-HOXA9. MI-3454 markedly reduced cell growth in both NUP98-r leukemia cell lines (GI₅₀=25-112 nM at day 14 of treatment), but not in the control cell line HM-2 (BMCs transformed with HOXA9 and MEIS1), (Figure 1A,B; supplemental Figure 1A–C). Interestingly, shorter treatment of NUP98-r leukemia cells with MI-3454 had very limited effects (supplemental Figure 1B), consistent with the delayed response in other menin-MLL1-dependent leukemia sub-types [17]. Furthermore, MI-3454 induced apoptosis and differentiation of NUP98-r cells and strongly downregulated leukemia-relevant target genes, including Hoxa9, Meis1, Hoxa10, and Pbx3, supporting on-target activity (Figure 1C–E; supplemental Figures 2A–E and 3A,B).

Figure 1. Effect of MI-3454 menin inhibitor in NUP98-rearranged leukemia models.

(A) Titration curves from MTT cell viability assay performed after 14 days of treatment of murine bone marrow (BM) cells transformed with NUP98-NSD1, NUP98-HOXA9 or HOXA9/MEIS1 (HM-2) with MI-3454. mean ± SD, n = 4 technical replicates were used for each condition. 2-3 independent MTT experiments were performed for each cell line. Representative graphs are shown. GI₅₀ values correspond to MI-3454 concentrations needed for 50% inhibition of cell proliferation. (B) Growth inhibition in murine BM cells with NUP98 translocations treated with MI-3454. mean ± SD, n = 3 technical replicates were used for each condition. Each experiment was repeated 2-3 times with representative graphs shown. (C) Flow cytometry analysis of differentiation marker Gr-1 in murine BM cells with NUP98 translocations upon treatment with MI-3454. mean ± SD, n = 3. (D) Wright-Giemsa-stained cytospins for murine BM cells with NUP98-r after 14 days of treatment with 400 nM of MI-3454. (E) Quantitative RT-PCR performed in murine BM cells with NUP98-NSD1 or NUP98-HOXA9 fusions after 14 days of treatment with MI-3454. Gene expression was normalized to GAPDH and referenced to the DMSO-treated cells. Data represents two independent experiments each performed in triplicates, (mean ± SD, n = 3). (F) Colony counts for methylcellulose colony forming assay performed in primary patient samples with NUP98 translocations upon treatment with MI-3454. (G) Pictures of colonies in NUP98_322 primary sample upon treatment with MI-3454 or DMSO. (H) Wright-Giemsa-stained cytospins for cells from colony forming assay in primary AML samples with NUP98 translocations upon treatment with MI-3454 or DMSO. (I) Quantitative RT-PCR performed in primary AML sample NUP98_258 after 14 days of treatment with MI-3454. Gene expression was normalized to GAPDH and referenced to the DMSO-treated cells. Data represents two independent experiments each performed in triplicates, (mean ± SD, n = 3). (J) Kaplan-Meier survival curves of vehicle or MI-3454 (80 mg/kg, b.i.d., p.o.) treated mice in the NUP98_258 PDX model (n = 8). (K) Flow cytometric quantification of human CD45⁺ cells in peripheral blood (PB) of NUP98_258 PDX mice during mice treatment with MI-3454 (80 mg/kg, b.i.d., p.o.) or vehicle. *P < 0.05; **P< 0.01; ***P < 0.001; ****P < 0.0001; ns - not significant, calculated using two-way Anova with Tukey multiple comparison test (panel B) or Student’s t-test (panels C,E,F,I). Log-rank (Mantel-Cox) test was used to calculate p value in panel J. Technical replicates are shown in panels C,E,F,I.

To further evaluate the activity of MI-3454, we utilized primary samples (NUP98_258 and NUP98_322) from pediatric AML patients harboring NUP98-NSD1 translocations (supplemental Table 1). Importantly, MI-3454 strongly affected colony formation (>50% reduction of colony number at 50 nM of MI-3454), induced differentiation, and downregulated leukemia-associated genes regulated by menin-MLL1 (HOXA9, MEIS1, HOXA5) in both NUP98-r primary samples (Figure 1F–I; supplemental Figure 4A–C).

To assess the in vivo efficacy of MI-3454, we developed a PDX model derived from the NUP98_258 AML sample. Treatment with MI-3454 or vehicle was initiated at 34 days post-transplantation and continued for 53 days with no toxicity observed (supplemental Figure 4D). Leukemia developed rapidly in the vehicle-treated mice, with substantial increases in the level of human CD45+ (hCD45+) cells in peripheral blood (PB) and accelerated mortality with 65-day median survival (Figure 1J, K). In contrast, the level of hCD45+ cells in the MI-3454-treated mice was markedly lower, resulting in significant survival benefit over vehicle-treated mice (Figure 1J,K). Overall, these results demonstrate that the menin inhibitor MI-3454 delays progression of NUP98-r leukemia in vivo, but combinatorial treatments might be needed for more effective eradication of NUP98-r leukemic cells.

MI-3454 synergizes with CDK6 inhibitor in NUP98-r leukemic cells

CDK6 is overexpressed in NUP98-r leukemia cells and was found to be a direct target of NUP98 fusion proteins [27]. Indeed, genetic inactivation or pharmacologic inhibition of CDK6 by Palbociclib (FDA approved CDK4/6 inhibitor) attenuated NUP98-fusion driven leukemogenesis [27]. Thus, we rationalized that combination of menin and CDK6 inhibitors could enhance anti-leukemic activity in NUP98-r leukemia models. To test this hypothesis, we utilized several primary patient samples harboring the most common NUP98 translocations: NUP98-NSD1 (samples: NUP98_258, NUP98_322 and NUP98_299 from pediatric AML patients) and NUP98-HOXA9 (sample NUP98_1055 from adult AML patient) (Supplemental Table 1).

First, we found that simultaneous treatment with MI-3454 and Palbociclib results in more pronounced reduction (>75%) of colony number and size over single agents in all NUP98-r AML samples (Figure 2A–C; supplemental Figure 5A,B). Combination indices calculated using Chou-Talalay method [28] ranged from 0.12 to 0.65, indicating strong synergistic effect (Figure 2A,B). Similarly, combination of MI-3454 with Palbociclib more strongly affected cell growth versus single agents in all NUP98-r samples (Figure 2D,E; supplemental Figure 5C,D), consistent with the enhanced anti-proliferative effects observed upon combining another menin inhibitor (SNDX-50469) and CDK6 inhibitor (Abemaciclib) in the MLL1-rearranged or NPM1-mutated leukemia cells [29]. Furthermore, combinatorial treatment with MI-3454 and Palbociclib induced more pronounced cell differentiation (increased cell size, high cytoplasmic-to-nucleus ratio, vacuolated cytoplasm, increased expression of CD11b and CD14 differentiation markers) over single agents (Figure 3A–C; supplemental Figure 6A–D). These results indicate that combinatorial treatment of NUP98-r cells with menin and CDK6 inhibitors is more effective relative to single agents in inducing cell growth arrest and differentiation, supporting stronger reduction of the oncogenic potential of NUP98 fusions.

Figure 2. Effect of combinatorial treatment with menin and CDK6 inhibitors in primary patient samples with NUP98 translocations.

(A, B) Colony counts for methylcellulose colony forming assay performed in primary patient samples with NUP98 translocations: NUP98-NSD1 (A) or NUP98-HOXA9 (B) upon treatment with MI-3454, Palbociclib and their combination. n = 2. Combination indexes (CIs) calculated using Compusyn software. CI values <1 indicate synergy. Compound concentrations in nanomolar (nM). Technical replicates are shown. (C) Pictures of colonies in NUP98_258 and NUP98_322 primary samples upon treatment with DMSO, MI-3454, Palbociclib or their combinations. Concentrations in combinatorial treatments are the same as for single agents. (D,E) Growth inhibition in primary patient samples with NUP98 translocations (NUP98-NSD1 (D) and NUP98-HOXA9 (E)) upon treatment with DMSO, MI-3454, Palbociclib or their combinations. Two independent experiments each performed in triplicates were performed with representative graphs shown. mean ± SD, n = 3 technical replicates used at each condition. *P < 0.05; **P< 0.01; ***P < 0.001; ****P < 0.0001; ns - not significant, calculated using one-way Anova with Tukey multiple comparison test (panels A,B) or two-way Anova with Tukey multiple comparison test (panels D,E). Pal: Palbociclib, MI: MI-3454.

Figure 3. Combination of menin and CDK6 inhibitors induces differentiation.

(A) Wright-Giemsa-stained cytospins for primary AML samples with NUP98 translocations upon treatment with DMSO, MI-3454, Palbociclib or their combination. Samples were collected at the end point of colony assay in the corresponding primary samples. (B, C). Flow cytometry analysis of differentiation markers CD11b (B) and CD14 (C) upon 7 days (for NUP98_1055) or 12 days (for NUP98_258 and NUP98_322) of treatment of the NUP98-r primary patient samples with DMSO, MI-3454, Palbociclib or their combination. mean ± SD, n = 3 technical replicates used at each condition. *P < 0.05; **P< 0.01; ***P < 0.001; ****P < 0.0001, calculated using one-way Anova with Tukey multiple comparison test (panels B,C). Concentrations of compounds in nanomolar (nM). Pal: Palbociclib, MI: MI-3454. Technical replicates are shown in panels B,C.

FLT3 inhibition enhances anti-leukemic activity of menin inhibitor

Activating mutations in the FLT3 kinase, including FLT3-ITD (Internal Tandem Duplications, ITD), are frequently found in patients with NUP98 rearrangements and lead to poor prognosis, with up to 92% of AML patients with NUP98-NSD1 harboring FLT3-ITD [1]. Indeed, majority of samples utilized in this study harbor FLT3-ITD (Supplemental Table 1). Thus, we rationalized that FLT3 inhibitors could enhance the anti-leukemic activity of MI-3454 in these models.

First, we found that while single agents MI-3454 or Gilteritinib (FDA approved FLT3 inhibitor used in AML patients) had limited effect on colony formation, their combination resulted in >75% reduction in colony number, with remaining colonies showing decreased size and dispersed morphology (Figure 4A–C; supplemental Figure 7A–C). Similarly, we observed more pronounced cell growth inhibition for this combination versus single agents in all NUP98-r AML samples (Figure 4D; supplemental Figure 7D), in agreement with recently reported synergistic suppression of cell proliferation by Revumenib (another menin inhibitor) combined with Gilteritinib in NUP98-NSD1/FLT3-ITD+ AML sample [13]. In addition, combination of MI-3454 with Gilteritinib induced more pronounced differentiation over single agents, as reflected by increased expression of the CD14 differentiation marker and cell morphology changes (Figure 4E,F; supplemental Figure 7E). Overall, these results demonstrate that FLT3 inhibitors can enhance the anti-leukemic activity of menin inhibitors in NUP98-r leukemia cells harboring FLT3-ITD, supporting the therapeutic potential of this combination.

Figure 4. Combinatorial effect of menin and FLT3 inhibitors in primary AML samples with NUP98 translocations.

(A, B) Colony counts for methylcellulose colony forming assay performed in primary patient samples with NUP98 translocations: NUP98-NSD1 (A) or NUP98-HOXA9 (B) upon treatment with DMSO, MI-3454, Gilteritinib and their combination. mean ± SD, n = 2 technical replicates used at each condition. Concentrations of compounds in nanomolar (nM). (C) Pictures of colonies in NUP98_322 and NUP98_1055 primary cells upon treatment with DMSO, MI-3454, Gilteritinib and their combination. (D) Growth inhibition in primary patient samples with NUP98 translocations after: 12 days (in NUP98_258), 8 days (in NUP98_322) or 7 days (in NUP98_1055) of treatment with DMSO, MI-3454, Gilteritinib and their combination. mean ± SD, n = 3. (E) Flow cytometry analysis of CD14 differentiation marker after 12 days of treatment of the NUP98_258 sample with DMSO, MI-3454, Gilteritinib and their combination. mean ± SD, n = 3 technical replicates used at each condition. Concentrations of compounds in nanomolar (nM). (F) Wright-Giemsa-stained cytospins for NUP98_1055 primary AML sample upon 7 days of treatment with DMSO, MI-3454, Palbociclib or their combination. *P < 0.05; **P< 0.01; ***P < 0.001; ****P < 0.0001; ns - not significant, calculated using one-way Anova with Tukey multiple comparison test (panels A, D). Gilt: Gilteritinib, MI: MI-3454. Technical replicates are shown in panels A,B,D.

Combination of menin and CDK4/6 inhibitors improves survival of NUP98-rearranged leukemic mice

To assess the in vivo effect of combining MI-3454 with Palbociclib, we developed PDX models in NSGS mice utilizing two NUP98-r AML samples (NUP98_258 and NUP98_322). Treatment of NUP98_258 PDX mice with vehicle, MI-3454, Palbociclib, or their combination was initiated 15 days after transplantation and continued for 48 days without substantial toxicity (supplemental Figure 8A). Vehicle-treated mice developed overt leukemia with prominent splenomegaly, high level of hCD45+ cells in PB, bone marrow (BM) and spleen (48-88% at the endpoint), and a median survival of 70 days (Figure 5A,B; supplemental Figure 8B–E). Mice treated with a 60 mg/kg dose of MI-3454 did not show significant reduction of the level of hCD45+ cells or survival benefit (Figure 5A,B). On the other hand, treatment with Palbociclib reduced the level of hCD45+ cells initially (<1% in PB), resulting in a median survival of 130 days (Figure 5A,B). Remarkably, the level of hCD45+ cells in the PB of mice treated with the combination was substantially lower than in any other group (<5% at day 130 post-transplantation), despite withdrawing the treatment at day 59 (Figure 5A,B). Accordingly, the combination of MI-3454 with Palbociclib led to markedly increased survival (median survival of 180 days), representing >250% survival benefit versus vehicle-treated mice (Figure 5B), with several mice remaining alive for >220 days post-transplantation.

Figure 5. In vivo activity of MI-3454 in combination with CDK6 and FLT3 inhibitors in PDX models of NUP98-r leukemia.

(A) Flow cytometric quantification of human CD45⁺ cells in peripheral blood (PB) of NUP98_258 PDX mice during mice treatment with vehicle, MI-3454 (60 mg/kg, b.i.d., p.o.), Palbociclib (50 mg/kg, q.d., p.o.) or their combination. Mean ± SD, n = 5-7. (B) Kaplan-Meier survival curves of vehicle, MI-3454, Palbociclib or their combination treated mice in the NUP98_258 PDX model (n = 7-8 per group). Treatment regimen as in panel A. (C) Kaplan-Meier survival curves of vehicle, MI-3454 (60 mg/kg, b.i.d., p.o.), Palbociclib (50 mg/kg, q.d., p.o.) or their combination treated mice in the NUP98_322 PDX model (n = 6-7 per group). (D) Flow cytometric quantification of human CD45⁺ cells in bone marrow samples isolated from NUP98_322 PDX mice after 47 days of treatment (samples collected at day 70 post-transplantation) with vehicle, MI-3454, Palbociclib and their combination. Mean ± SD, n = 4. Treatment regimen as in panel C. (E) Wright-Giemsa-stained cytospins for bone marrow samples isolated from NUP98_322 PDX mice after 47 days of treatment (day 70 post-transplantation). Treatment regimen as in panel C. (F) Kaplan-Meier survival curves of NUP98_322 PDX mice treated with vehicle, MI-3454 (80 mg/kg, b.i.d., p.o.), Gilteritinib (30 mg/kg, q.d., p.o.) or their combination. Mean ± SD, n = 8 mice per group. (G) Flow cytometric quantification of human CD45⁺ cells in peripheral blood of NUP98_258 PDX mice during mice treatment with vehicle, MI-3454 (80 mg/kg, b.i.d., p.o.), Gilteritinib (30 mg/kg, q.d., p.o.) or their combination. Mean ± SD, n = 8 mice per group. ***P < 0.001; ns - not significant, calculated using one-way Anova with Tukey multiple comparison test (panel D). Log-rank (Mantel-Cox) test was used to calculate p values (referenced to the vehicle-treated mice) in panels B, C and F.

In the NUP98-322 PDX, treatment of mice with MI-3454, Palbociclib or their combination was initiated at day 23 post-transplantation and continued for 49 days. The median survival for the vehicle-treated mice was 91 days, while both single agents slightly increased survival (115 and 126 days for MI-3454 and Palbociclib, respectively) (Figure 5C). Importantly, combination of MI-3454 and Palbociclib substantially improved mice survival (median survival of 175 days) over vehicle-treated mice (Figure 5C). To further explore the in vivo response of the NUP98_322 PDX model to this combination, we sacrificed four mice from each group after 47 days of treatment and observed markedly increased hCD45+ cell infiltration in BM of vehicle-treated mice (16-40% in BM) over single agents (<27% for MI-3454 and <12% for Palbociclib), (Figure 5D; supplemental Figure 9A). Notably, mice treated with the combination of MI-3454 and Palbociclib showed near complete reduction of hCD45+ cells in BM (<0.6%), a more pronounced differentiating phenotype of BM cells, and reduced patient-derived cell infiltration in PB over other treatment groups (Figure 5D,E; supplemental Figure 9B), in agreement with survival studies (Figure 5C). Additionally, blood parameters were not negatively affected by the combination, supporting that it is well tolerated in mice (supplemental Figure 9C,D). Taken together, these data demonstrate that combining menin and CDK6 inhibitors has a superior anti-leukemic effect over single agents, supporting potential clinical value of this combination in NUP98-r leukemia.

Combination of MI-3454 with Gilteritinib reduces leukemia burden of NUP98-r leukemia

Based on the pronounced anti-leukemic effect of the combination of menin and FLT3 inhibitors in NUP98-r AML samples (Figure 4), we aimed to assess the in vivo effect of this combination in NUP98_258 PDX model (harboring NUP98-NSD1 and FLT3-ITD). Treatment of mice with vehicle, single agents MI-3454 and Gilteritinib, or their combination was initiated 22 days post-transplantation and continued for 47 days. The control mice developed terminal leukemia within ~2 months, with a median survival of 67 days and high level of hCD45+ cells in PB (>70%) (Figure 5F,G). Treatment with single agents significantly reduced the level of circulating hCD45+ cells, leading to ~20% increased survival for both agents over vehicle-treated mice (Figure 5F,G). Importantly, combination of MI-3454 with Gilteritinib resulted in markedly reduced level of hCD45+ cells in PB as compared to the vehicle- or single agent-treated mice, although it increased after discontinuing the treatment (Figure 5G). Consequently, a median survival of 105 days for the combination cohort represents a substantial survival benefit over vehicle- or single agent-treated mice (Figure 5F). Overall, these data demonstrate much stronger reduction of leukemia burden by the combination of menin and FLT3 inhibitors over single agents in NUP98-r leukemia with FLT3 activating mutations, although continuous treatment might be required to further improve the outcome.

Combination of menin and CDK6 inhibitors affects cell cycle and differentiation pathways

To explore the mechanism of enhanced anti-leukemic activity when combining MI-3454 and Palbociclib, we performed RNA-sequencing (RNA-seq) studies in the NUP98_1055 AML sample upon 7 days of treatment with DMSO, single agents or their combination. Differential analysis (relative to DMSO) revealed that the combination results in >2-fold increase in the number of affected genes over single agents, with a total of 464 down- and 807 up-regulated genes, supporting modulation of new pathways not affected by the single agents (Figure 6A, supplemental Figure 10A; supplemental Table 2). Importantly, combination of MI-3454 with Palbociclib also led to more pronounced changes of genes affected by single agents, suggesting that these compounds can potentiate their transcriptional activity, Figure 6A,B. This was further validated by K-means clustering [30] of differentially expressed genes, which identified five clusters of genes grouped by similar expression profiles (Figure 6B; supplemental Table 3). Clusters 1 and 2 comprise genes more strongly upregulated by the combination, and based on gene ontology analysis, cluster 1 includes genes associated with differentiation and myeloid activation (e.g. CD14, MMP9, CCL2), while cluster 2 comprises immune response genes (e.g. IFIT1/2, CXCR1, IL7R) (Figure 6B; supplemental Table 3,4). In contrast, clusters 3 and 4 contain genes more strongly downregulated by the combination over single agent treatment (Figure 6B). Cluster 3 comprises genes associated with proliferation and cell cycle regulation, including KIF2C, CCNA2, CCNB1/2, CDK1/2 (supplemental Table 4), in agreement with strong anti-proliferative effect of the combination (Figure 2D,E). Cluster 4 also involves genes associated with cell proliferation and leukemogenesis, including MEIS1, FLT3, MEF2C (supplemental Table 4). The remaining cluster 5 involves genes similarly up-regulated by MI-3454 and the combination, including a sub-set of differentiation genes, suggesting that differentiation is driven by menin inhibition.

Figure 6. Global gene expression studies for combination of MI-3454 with kinase inhibitors in NUP98_1055 primary AML sample.

A,D. Comparison of differentially expressed (DE) genes (adjusted p < 0.05; fold-change > |1.5|) from RNA-seq studies in NUP98_1055 primary AML sample after 7 days of treatment with DMSO, MI-3454 (250 nM), Palbociclib (75 nM) (A) or Gilteritinib (150 nM) (D) or their combinations (n=3 samples per treatment group). VennDiagrams show the overlap of upregulated DE genes (top) and downregulated DE genes (bottom) relative to DMSO. B,E. Heat maps of DE genes after k-means clustering of DE genes using a computationally determined 5 clusters (for combination with Palbociclib, B) or 4 clusters (for combination with Gilteritinib, E) Representative pathways within each cluster determined using GOrilla gene ontology analysis are listed. Red indicates high expression, blue indicates low expression. C, F. Summary of fast gene set enrichment analysis (fgsea) results for gene sets strongly affected by the combination of MI-3454 with Palbociclib (C) or Gilteritinib (F). Each bubble represents a gene set. Size of bubbles on the plot indicate the level of significance and y-axis indicates the normalized enrichment score (NES) for the gene sets. Bubbles are colored based on manual annotation of the category of gene set (cell cycle, cancer markers, epigenetic/transcriptional regulation).

Gene Set Enrichment Analysis (GSEA) revealed that the combination of MI-3454 with Palbociclib is more effective than single agents in suppressing gene expression programs relevant to cell cycle and cancer markers (Figure 6C; supplemental Table 5). For example, strong downregulation enrichment was observed for pediatric cancer markers gene sets (Whiteford et al.) [31], with c-MYC target genes (e.g. CDCA7) [32] and cell cycle genes (e.g. CCNA2, CDK1, CCNB2) within the top downregulated genes (supplemental Figure 10B,C). Similarly, the cell cycle gene sets were strongly downregulated (Reactome) [33] upon combinatorial treatment, with a similar subset of cell cycle regulation genes (e.g. CCNA2/B2, CDK1) found in the top 20 differentially expressed leading edge genes (supplemental Figure 10D,E). Furthermore, gene sets associated with epigenetic regulation (methylation or acetylation) or chromatin modification were also more strongly affected by the combination (Figure 6C; Supplemental Table 5).

We have also performed RNA-seq in the NUP98-258 AML sample harboring NUP98-NSD1 translocation (supplemental Table 6,7,8) and identified three clusters of genes with similar expression profiles (supplemental Figure 11A–D). Importantly, the overall pattern of transcriptional changes induced by the combination of MI-3454 and Palbociclib was similar between NUP98_258 and NUP98_1055 samples, despite harboring different NUP98 fusions. Genes associated with differentiation (MMP1/8/10) were more strongly upregulated, while cell cycle regulation (CCNB2, CDK6) and self-renewal (e.g. MEF2C) genes were more strongly downregulated by the combination. GSEA revealed strong enrichment for cell cycle regulation, cancer markers, and hematopoietic stem cell gene sets (supplemental Figure 12A,B; supplemental Table 9). Overall, our data suggest that similar pathways are modulated by the combination of menin and CDK6 inhibitors in NUP98-r leukemia cells harboring different fusions.

Transcriptional regulation induced by the combination of menin and FLT3 inhibitors

To gain insight into enhanced anti-leukemic effect upon combining MI-3454 with Gilteritinib, we performed RNA-seq studies in NUP98_1055 AML sample after 7 days of treatment (supplemental Table 2). Differential analysis (relative to DMSO) revealed >2-fold increase in the number of genes affected by the combination over single agents, Figure 6D; supplemental Figure 13A. The combination of MI-3454 with Gilteritinib led to stronger modulation of the majority of genes affected by single agents, which is similar to what we observed for the combination of MI-3454 with Palbociclib (see above and Figure 6A). Detailed analysis of differentially expressed genes showed four clusters, with clusters 1 and 2 comprising genes, respectively, more strongly upregulated or downregulated by MI-3454/Gilteritinib combination, (Figure 6E; supplemental Table 10,11). Upregulated genes in cluster 1 comprise differentiation (e.g. MMP2, CEBPE) and myeloid activation (ELANE, CXCL1) genes. The downregulated genes in cluster 2 are relevant to cell proliferation and leukemogenesis (e.g. MEIS1, FLT3, MECOM) as well as to cell cycle (e.g. CCNA2/B2, CCND3). The remaining clusters 3 and 4 contain additional differentiation and immune response genes, which are similarly up-regulated by one of the single agents and combination, suggesting that these effects are less driven by the combination (Figure 6E; supplemental Table 10,11).

GSEA analysis revealed that the combination of MI-3454 and Gilteritinib suppressed pathways relevant to cell cycle, cancer markers, and transcriptional regulation (Figure 6F). For example, strong enrichment was found for cell cycle gene sets (with CDK1, CCNA2 and CCNB2 within the top downregulated genes), pediatric cancer markers [31], and gene sets comprising targets of Hoxa9 and Meis1 [34] (supplemental Figure 13B–D). Thus, gene expression changes are consistent with strong anti-leukemic effects observed when MI-3454 was combined with Gilteritinib in NUP98-rearranged AML models (Figure 4 and 5F,G).

Combination of MI-3454 with different kinase inhibitors affects similar pathways

Global gene expression changes induced by the combinations of MI-3454 with either CDK6 or FLT3 inhibitors in the NUP98-r leukemia cells suggest similarities between affected genes and pathways (see above). To further investigate these effects, we performed clustering of differentially expressed genes examining all treatment groups (all single agents and two combinations), resulting in five clusters (Figure 7A, supplemental Table 12). Strikingly, the combinations induce more pronounced gene expression changes compared to single agents for the majority of genes that are differentially expressed by any treatment (Figure 7A). Clusters 1 and 4 are enriched for genes more strongly downregulated by combinations compared to their respective single agents (Figure 7A). These clusters include genes associated with cell cycle (e.g. CCNA2, CCNB2, CDK1), signal transduction and cell proliferation (e.g. MEIS1, MEF2C, FLT3), Figure 7A, supporting that synergistic effects of combining menin with CDK6 or FLT3 inhibitors results from more effective suppression of these genes by the combinations. Clusters 2, 3 and 5, which are upregulated by the two distinct combinations, are enriched in genes linked to differentiation (e.g. ITGAM, CD14) or immune response (Figure 7A). Importantly, gene expression analysis revealed that the combinations of MI-3454 with either Palbociclib or Gilteritinib affect similar pathways relevant to leukemogenesis, thus enhancing the anti-leukemic effects of menin inhibitor.

Figure 7. Comparison of genes and pathways affected by combinations of menin and kinase inhibitors.

A. Heat maps of DE genes after k-means clustering of DE genes using computationally determined 5 clusters from RNA-seq studies in NUP98_1055 primary AML sample after 7 days of treatment with DMSO, MI-3454, Palbociclib or Gilteritinib or their combinations (n=3 samples per treatment group). Compound concentrations as in Figure 6. Representative enriched pathways (from GOrilla enrichment analysis tool) and selected genes within each cluster are listed. B. Overlap of significantly enriched gene sets (padj<0.05) from fgsea analyses from RNA-seq studies in NUP98_1055 primary AML sample. Top: VennDiagrams showing overlap of altered gene sets after single agent treatments, bottom: VennDiagrams showing overlap of altered gene sets after treatment with combinations relative to DMSO. C,E. Representative gene set enrichment plots from fgsea analyses for each treatment conditions relative to DMSO for the gene sets from MsigDB: proliferation associated genes (Benporath et al.)[42] C) and cell cycle genes (Reactome[33]), E); N indicates number of genes from each set. Padj = adjusted p-value; NES = normalized enrichment score. D,F. The heatmaps showing genes comprising the top 20 genes from the leading edge of the gene set enrichment plots. Gene sets are indicated at the top of heatmaps. Red indicates high expression, blue indicates low expression.

Comparison of gene sets affected by single agents or combinations indicates substantial overlap between the combinations of MI-3454/Palbociclib and MI-3454/Gilteritinib (77%), while the overlap for genes affected by individual kinase inhibitors was smaller (~50 %) (Figure 7B; Supplemental Table 13). Further analysis clearly shows that pathways associated with proliferation and cell cycle regulation are more strongly enriched by both combinations than single agents (Figure 7C–F). For example, gene sets relevant to cell cycle regulation are not significantly affected by MI-3454 or Gilteritinib, but show strong downregulation for the combination (Figure 7E,F). Furthermore, similar gene sets are enriched when combining MI-3454 with Palbociclib, Figure 7E,F. Taken together, even though Palbociclib and Gilteritinib inhibit two unrelated kinases, when combined with MI-3454 they show similar gene expression patterns consistent with stronger suppression of genes and pathways relevant to cell cycle and cell proliferation. Consequently, these combinations exert more effective anti-leukemic activity over single agents in NUP98-r leukemia models.

Discussion

Menin is a well validated target in various leukemia sub-types, including leukemia with MLL1 translocations [14–16], NPM1 mutations [17] or NUP98 translocations [12]. This is exemplified by promising results from phase I clinical studies with menin inhibitors Ziftomenib [25] and Revumenib [35] in refractory and relapsed AML patients with MLL1-r or NPM1-mut. However, 50% of AML patients did not respond to menin inhibitors, and a subset of patients (~38%) treated with Revumenib developed mutations in menin, mediating clinical resistance [36]. Thus, combinatorial treatments might be required to achieve better clinical efficacy and to overcome resistance.

The menin-MLL1 complex has also been implicated in leukemogenesis mediated by NUP98 fusion proteins [12]. Besides, specific kinases, such as CDK6 (overexpressed in NUP98-r leukemia) [27] or FLT3 (mutated in NUP98-r AMLs) [1], are involved in leukemogenesis by NUP98 fusions. The co-dependency of NUP98-r leukemia cells on the menin-MLL1 complex and CDK6 or FLT3 kinases provided the rationale to evaluate combinatorial effects of menin and kinase inhibitors. Our studies revealed a strong synergetic effect when combining either Palbociclib or Gilteritinib with MI-3454 in NUP98-r primary AML samples. Specifically, colony formation, cell growth, apoptosis, and differentiation of NUP98-r leukemia cells were more strongly affected by both combinations as compared to the single agents. Mechanistically, both combinations induced stronger downregulation of cell cycle (e.g. CCNA2, CCNB1/2, CDK1/2) or proliferation (e.g. MEIS1, MECOM, FLT3) genes, and upregulation of differentiation genes (e.g. MMPs) over single agents, providing an insight into transcriptional regulation induced by the combinations. Importantly, both combinations augmented anti-leukemic efficacy in PDX models of NUP98-r leukemia over single agents, resulting in strong leukemia suppression.

Recent work by Rasouli et al. [13] demonstrated a synergistic reduction of cell viability in NUP98-r primary AML cells expressing FLT3-ITD induced by combining the menin inhibitor Revumenib with Gilteritinib, in agreement with our findings. These results are also consistent with a strong synergistic anti-leukemic activity that we and others observed upon combining menin and FLT3 inhibitors in AML models harboring MLL1 translocations or NPM1 mutations [24, 29, 37]. Similarly, synergistic in vitro lethality in the MLL1-r and NPM1-mut AML cells was previously shown upon combining SNDX-50469 menin inhibitor with CDK6 inhibitor Abemaciclib [29], in line with our findings in NUP98-r leukemia models reported here. Thus, the current study validates and expands upon previous results by demonstrating synergistic suppression of cell growth and by examining the overlapping phenotypic and transcriptomic programs affected by the combination of menin and kinase inhibitors in NUP98-r AML models. Overall, our results and complementary data from other groups [13, 24, 29] support that enhanced anti-leukemic effects observed upon combining menin and kinase (FLT3 or CDK6) inhibitors extend beyond NUP98-r AMLs, namely to leukemias with MLL1 translocations or NPM1 mutations. All these results support clinical translation of such combinations to various leukemia sub-types that rely on the menin-MLL1 interaction.

When directly compared, combination of MI-3454 with either Palbociclib or Gilteritinib affected similar pathways despite inhibition of different kinases. Therefore, our work provides an attractive avenue for relatively unexplored combinations of epigenetic drugs, such as menin inhibitors, with drugs targeting kinases. In addition to clinical use of FLT3 inhibitors in AML [38], inhibitors targeting other kinases, including PI3K/AKT/mTOR, JAK2, CHK1, MAPK, CDKs, AXL, VEGFR, BTK, SYK and SRC, were introduced to clinical studies in AML [39, 40]. Many of these kinases regulate cell cycle and proliferation pathways, thus their combinations with menin inhibitors could exhibit enhanced activity in NUP98-r or other leukemias. For example, we previously demonstrated that combination of MI-3454 with FLT3 inhibitors is very effective in MLL1-r and NPM1-mut leukemia [24]. Thus, combining epigenetic drugs with kinase inhibitors could represent a more general strategy to effectively eradicate leukemic cells and overcome limited sensitivity to single agents.

With encouraging clinical data for menin inhibitors (Ziftomenib and Revumenib, both are in phase II clinical trials) in adult AML patients with MLL1-r and/or NPM1 mutated leukemia [25, 26, 35], there is an opportunity to expand these trials to patients with other leukemia sub-types. Our studies report combinatorial treatments of menin inhibitor with CDK6 or FLT3 inhibitors in NUP98-r leukemia models, supporting that both combinations might represent rational therapeutic strategies for leukemia patients with NUP98 translocations as well as other leukemia sub-types (e.g. with MLL1 translocations or NPM1 mutations) that warrant clinical investigation. Since Palbociclib and Gilteritinib were granted FDA approvals, their combination with menin inhibitors might be readily tested in patients. Our studies also suggest that menin inhibitors may synergize with inhibitors of other kinases to further expand these drugs to various leukemia sub-types, including high HOX leukemias [41].

Data Availability Statement

The datasets generated and analyzed during the current study are available from the corresponding authors on reasonable request.