ABCC1 and glutathione metabolism limit the efficacy of BCL-2 inhibitors in acute myeloid leukemia

Abstract

The BCL-2 inhibitor Venetoclax is a promising agent for the treatment of acute myeloid leukemia (AML). However, many patients are refractory to Venetoclax, and resistance develops quickly. ATP-binding cassette (ABC) transporters mediate chemotherapy resistance but their role in modulating the activity of targeted small-molecule inhibitors is unclear. Using CRISPR/Cas9 screening, we find that loss of ABCC1 strongly increases the sensitivity of AML cells to Venetoclax. Genetic and pharmacologic ABCC1 inactivation potentiates the anti-leukemic effects of BCL-2 inhibitors and efficiently re-sensitizes Venetoclax-resistant leukemia cells. Conversely, ABCC1 overexpression induces resistance to BCL-2 inhibitors by reducing intracellular drug levels, and high ABCC1 levels predicts poor response to Venetoclax therapy in patients. Consistent with ABCC1-specific export of glutathionylated substrates, inhibition of glutathione metabolism increases the potency of BCL-2 inhibitors. These results identify ABCC1 and glutathione metabolism as mechanisms limiting efficacy of BCL-2 inhibitors, which may pave the way to development of more effective therapies.

Fig. 1. ABCC1 modulates Venetoclax sensitivity in AML cells.

A Schematic representation of the CRISPR/Cas9-based competitive proliferation assay. Each ABC transporter gene is targeted with two sgRNAs in Cas9-expressing MOLM-13 cells. Cells were divided and treated with DMSO or Venetoclax. The percentage of sgRNA-expressing cells (IRFP670⁺) was monitored over time and the area under the curve (AUC) was determined for each sgRNA. The corresponding flow cytometric gating strategy for IRFP670⁺ cells is depicted in Supplementary Fig. 6A. B The AUC after 30 days of treatment of DMSO-treated MOLM-13-Cas9 cells expressing sgRNAs targeting ABC transporters is plotted against the AUC (30 days) of cells treated with 1 nM Venetoclax (left) or 50 nM Cytarabine (right). n = 4 experimental replicates with 2 different sgRNAs per gene. C Competitive proliferation assay of MOLM-13-Cas9 cells treated with 1 nM Venetoclax (left; n = 2 experimental replicates) and HL-60-Cas9 cells treated with 100 nM Venetoclax (right; n = 3 experimental replicates) for 30 days. Percentages of sgRNA/IRFP670+ cells were normalized to day 0 of treatment and to DMSO controls. Data are presented as mean values ± SD. B, C Source data are provided as a Source Data file.

Fig. 2. Pharmacologic inhibition of ABCC1 synergizes with BCL-2-inhibitors.

A BLISS synergy score distribution of MOLM-13 cells treated with Venetoclax at indicated concentrations in combination with Reversan for 5 days, determined using SynergyFinder. Data are represented as the relative deviation from BLISS additivity. n = 2 experimental replicates. B Five-day dose–response curves of Venetoclax in MV4-11 (left) and HL-60 (right) cells, co-treated with either DMSO or 5 μM Reversan. Data are presented as mean values ± SD. n = 3 experimental replicates. C Structure of BH3-mimetics inhibiting the anti-apoptotic proteins BCL-2, BCL-xL and BCL-w. D Heatmap depicting the overall relative deviation from BLISS additivity (BLISS synergy score), determined by SynergyFinder of three different AML cell lines (MOLM-13, KG-1, PL-21) treated with Reversan or MK-571 with indicated drugs for five days. n = 2 experimental replicates. A, B, D Source data are provided as a Source Data file.

Fig. 3. The dual BCL-2/BCL-xL inhibitor AZD-4320 strongly synergizes with ABCC1-inhibition in vitro and in vivo.

A Structure of the dual BCL-2/BCL-xL inhibitor AZD-4320. B Competitive proliferation assay of MOLM-13-Cas9 cells treated with 1 nM AZD-4320 (top) and HL-60-Cas9 cells treated with 20 nM AZD-4320 (bottom) for 29 days. Percentages of sgRNA/IRFP670+ cells were normalized to day 0 of treatment and to DMSO controls. Data are presented as mean values ± SD. n = 3 experimental replicates. C Growth curves of MOLM-13-Cas9 knockout clones (AAVS1.1, ABCC1-KO-1, ABCC1-KO-2) treated with DMSO or 1 nM AZD-4320. Data are presented as mean values ± SD. n = 2 experimental replicates. D Bioluminescence imaging of vehicle- or AZD-4320 treated mice [10 mg/kg] 16 days after transplantation with clonal MOLM-13-Cas9 AAVS1.1 or ABCC1-KO-1 cells. E Signal quantification of the epi-fluorescence in vivo imaging (IRFP670). Boxes represent interquartile ranges, horizontal lines represent the mean, whiskers indicate lower and upper limits. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated; n = 4 (MOLM-13 ABCC1-KO-1 vehicle)/n = 5 (MOLM-13 ABCC1-KO-1 treatment, MOLM-13 AAVS1.1 vehicle and treatment) mice. F Quantification of the percentage of sgRNA/IRFP670+ cells in the bone marrow of mice treated with vehicle or AZD-4320 [10 mg/kg] 16 days after transplantation with clonal MOLM-13-Cas9 AAVS1.1 or ABCC1-KO-1 cells. Boxes represent interquartile ranges, horizontal lines represent the mean, whiskers indicate lower and upper limits. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated; n = 4 mice. The corresponding flow cytometric gating strategy for IRFP670⁺ cells is depicted in Supplementary Fig. 6B. B, C, E, F Source data are provided as a Source Data file.

Fig. 4. Overexpression of ABCC1 promotes resistance to BCL-2 inhibitors by reducing intracellular drug concentrations.

A RT-qPCR analysis of ABCC1 mRNA expression in HL-60 wild-type (WT) and ABCC1-overexpressing (OE) cells. Fold change of ABCC1 expression in ABCC1-OE cells compared to WT. Data are presented as mean values. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated. n = 3 technical replicates, n = 1 experimental replicate. B Representative flow cytometric measurement of intracellular ABCC1 levels in HL-60 WT and ABCC1-OE cells. The corresponding gating strategy is depicted in Supplementary Fig. 6C. C Heatmap of GI₅₀ values [nM] of 5-day viability assays with indicated drugs in HL-60 WT and ABCC1-OE cells. n = 3 experimental replicates. D RT-qPCR analysis of ABCC1 mRNA expression of MOLM-13-Cas9 AAVS1.1 and ABCC1-KO-2 clones, transduced with either empty vector (e.V.) or an ABCC1-overexpressing (OE) construct. Levels were normalized to ACTB levels and to the expression of MOLM-13-Cas9 AAVS1.1 e.V. Data are presented as mean values. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated. n = 2 technical replicates, n = 1 experimental replicate. E Dose–response curves of AZD-4320 in MOLM-13-Cas9 AAVS1.1 or ABCC1-KO-2 clones transduced with either empty vector (e.V.) or an ABCC1-OE construct after 5 days of treatment. Data are presented as mean values ± SD. n = 3 experimental replicates. F Intracellular concentrations of AZD-4320 in HL-60 WT and ABCC1-OE cells determined using LC-MS/MS. The normalized peak area represents the relative compound concentration in the samples. Combined results of hydrophilic interaction liquid chromatography (HILIC) and reversed phase chromatography (RP) are depicted. Boxes represent interquartile ranges, horizontal lines represent the mean, whiskers indicate lower and upper limits. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated. n = 2 experimental replicates. G Median fluorescence intensity (MFI) of Calcein in HL-60 WT and ABCC1-OE cells treated either with DMSO or Reversan [10 µM] for 45 h. The corresponding flow cytometric gating strategy is depicted in Supplementary Fig. 6D. Data are presented as mean values ± SD. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated. n = 3 experimental replicates. H Median fluorescence intensity (MFI) of Calcein normalized to DMSO in HL-60 WT and ABCC1-OE cells treated with DMSO, Venetoclax or AZD-4320 at indicated concentrations for 45 h. Data are presented as mean values ± SD. Significance compared to DMSO was determined with an unpaired Student’s t test with two-tailed P value as indicated. n = 3 experimental replicates. The corresponding flow cytometric gating strategy is depicted in Supplementary Fig. 6D. A, C–H Source data are provided as a Source Data file.

Fig. 5. ABCC1 inhibition reverses Venetoclax-resistance in AML cells.

A Dose–response curves of Venetoclax in MOLM-13, HL-60 and THP-1 cells after 5 days of treatment. Data are presented as mean values ± SD. n = 3 experimental replicates. B Competitive proliferation assay of THP-1-Cas9 cells treated with 500 nM Venetoclax for 38 days. Percentages of sgRNA/IRFP670+ cells were normalized to day 0 of treatment and to DMSO controls. Data are presented as mean values ± SD. n = 3 experimental replicates. The corresponding flow cytometric gating strategy for IRFP670+ cells is depicted in Supplementary Fig. 6A. C Growth curves of THP-1 cells treated with DMSO, 1 µM Venetoclax, 2 µM Reversan or 1 µM Venetoclax in combination with 2 µM Reversan. Data are presented as mean values ± SD. n = 3 experimental replicates. D Ion count ratio of intracellular Venetoclax to Threonine (control amino acid) as determined by LC-MS/MS of Cas9-expressing THP-1 cells transduced with either sgAAVS1.1 or sgABCC1.1. Data are presented as mean values ± SD. n = 2 experimental replicates. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated. E Schematic representation of the generation of Venetoclax-resistant MOLM-13 and MV4-11 cells. Cells were treated up to 2.5 months using increasing concentrations of Venetoclax, until they showed stable growth in media supplemented with 1 μM Venetoclax. F Dose–response curves of Venetoclax in MV4-11 wild-type and Venetoclax-resistant cells co-treated with either DMSO or 5 µM Reversan for 5 days. Data are presented as mean values ± SD. n = 3 experimental replicates. G Growth curves of Venetoclax-resistant MV4-11 treated with DMSO, 1 µM Venetoclax, 2 µM Reversan or 1 µM Venetoclax in combination with 2 µM Reversan. Data are presented as mean values ± SD. n = 3 experimental replicates. A–D, F, G Source data are provided as a Source Data file.

Fig. 6. ABCC1 levels dictate the response of AML to Venetoclax.

A ABCC1 mRNA expression levels in AML patient samples (n = 542) compared to peripheral blood mononuclear cells (PBMCs) from healthy controls (n = 74). Data and statistical analysis were extracted from the Oncomine Platform from the Haferlach Leukemia dataset. Boxes represent interquartile ranges, horizontal lines represent the mean, whiskers indicate lower and upper limits. B ABCC1 mRNA expression levels in AML patient samples (n = 285) compared to bone marrow (n = 5) from healthy controls. Data and statistical analysis were extracted from the Oncomine Platform from the Valk Leukemia dataset. Boxes represent interquartile ranges, horizontal lines represent the mean, whiskers indicate lower and upper limits. C Expression of ABC transporters in AML patient samples (n = 16 individual patient samples) relative to ACTB [%]. Analysis of expression levels of each patient sample was performed in duplicates. Data are presented as mean values ± SD. Detailed patient information is listed in Supplementary Table 1. D Comparison of ABCC1 expression levels in patients with good or poor response to Venetoclax treatment (n = 14 individual patient samples—same as in (C)—2 non-responders excluded). Boxes represent interquartile ranges; horizontal lines represent the median expression; whiskers indicate lower and upper limits of the respective patient cohort. Analysis of expression levels of each patient sample was performed in duplicates. Significance was determined with an unpaired Student’s t test with two-tailed P value as indicated. E Dose–response curves of primary patient-derived AML cells co-treated with either DMSO or Reversan (patient 1 and 2, 5 µM, patient 3 and 4, 2.5 µM) and AZD-4320 for 3 days. Data are presented as mean values ± SD. n = 2 (patient 4)/n = 3 (patient 1–3) samples of the same patient. Gene names refer to the mutational status of the patients (Supplementary Table 1). C–E Source data are provided as a Source Data file.

Fig. 7. Glutathione metabolism modulates the sensitivity to BH3 mimetics.

A Schematic depiction of glutathione metabolism in human cells. The enzyme glutamate-cysteine ligase (GCL) produces γ-glutamylcysteine from glutamate and cysteine. Then, glutathione synthetase (GSS) adds glycine to the C-terminus of γ-glutamylcysteine to form reduced glutathione (GSH), which can be oxidized to GSSG or conjugated to xenobiotic compounds by glutathione-S-transferases (GSTs). The synthesis of GSH can be inhibited using the GCL inhibitor Buthionine sulfoximine (BSO), while GSH-conjugation to drugs can be inhibited using the GSTs inhibitor Ethacrynic acid (EA). Figure was created using bioRender.com. B Growth curves of MOLM-13 cells treated with DMSO, 100 µM BSO, Venetoclax (increasing concentrations 1, 5 and 10 nM) or Venetoclax in combination with 100 µM BSO. Data are presented as mean values ± SD. n = 2 experimental replicates. C Growth curves of Venetoclax-resistant MOLM-13 cells treated with DMSO, 100 µM BSO, 1 µM Venetoclax or 1 µM Venetoclax in combination with 100 µM BSO. Data are presented as mean values ± SD. n = 3 experimental replicates. D Growth curves of MOLM-13-Cas9 knockout clones AAVS1.1 (left), ABCC1-KO-2 (right) treated with DMSO, 100 µM BSO, Venetoclax (increasing concentrations 1, 5 and 10 nM) or Venetoclax in combination with 100 µM BSO. Data are presented as mean values ± SD. n = 2 experimental replicates. E Relative viability of primary patient-derived AML cells treated with Venetoclax (left) or AZD-4320 (right) in combination with DMSO or 100 µM BSO for 3 days. Concentrations used were: patient 1: 9.8 nM Venetoclax and 2.4 nM AZD-4320; patient 2: 39 nM Venetoclax and 39 nM AZD-4320. Data were normalized to DMSO-treated controls and are presented as mean values ± SD. n = 3 experimental replicates. B–E Source data are provided as a Source Data file.