Blockade of BCL-2 Proteins Efficiently Induces Apoptosis in Progenitor

ORIGINAL ARTICLE Blockade of BCL-2 proteins efﬁciently induces apoptosis in progenitor cells of high-risk myelodysplastic syndromes patients

S Jilg1, V Reidel1, C Müller-Thomas1, J König1, J Schauwecker2, U Höckendorf1, C Huberle1, O Gorka3, B Schmidt4, R Burgkart2, J Ruland3, H-J Kolb1, C Peschel1, RAJ Oostendorp1, KS Götze1 and PJ Jost1

Deregulated apoptosis is an identifying feature of myelodysplastic syndromes (MDS). Whereas apoptosis is increased in the bone marrow (BM) of low-risk MDS patients, progression to high-risk MDS correlates with an acquired resistance to apoptosis and an aberrant expression of BCL-2 proteins. To overcome the acquired apoptotic resistance in high-risk MDS, we investigated the induction of apoptosis by inhibition of pro-survival BCL-2 proteins using the BCL-2/-XL/-W inhibitor ABT-737 or the BCL-2-selective inhibitor ABT-199. We characterized a cohort of 124 primary human BM samples from MDS/secondary acute myeloid leukemia (sAML) patients and 57 healthy, age-matched controls. Inhibition of anti-apoptotic BCL-2 proteins was speciﬁcally toxic for BM cells from high-risk MDS and sAML patients, whereas low-risk MDS or healthy controls remained unaffected. Notably, ABT-737 or ABT-199 treatment was capable of targeting the MDS stem/progenitor compartment in high-risk MDS/sAML samples as shown by the reduction in CD34+ cells and the decreased colony-forming capacity. Elevated expression of MCL-1 conveyed resistance against both compounds. Protection by stromal cells only partially inhibited induction of apoptosis. Collectively, our data show that the apoptotic resistance observed in high-risk MDS/sAML cells can be overcome by the ABT-737 or ABT-199 treatment and implies that BH3 mimetics might delay disease progression in higher-risk MDS or sAML patients.

Leukemia (2016) 30, 112–123; doi:10.1038/leu.2015.179

INTRODUCTION positive cells when directly compared with smears from low-risk 13 Myelodysplastic syndromes (MDS) are clonal disorders of hema- MDS patients. These findings supported the notion that an topoietic stem cells mostly observed in older patients.1–4 Owing to acquired apoptotic resistance in the malignant MDS clone 3,5–11 the progressive cytopenias associated with MDS, patients often contributes to disease progression. suffer from fatigue, infections and bleeding. In contrast to the Pro- and anti-apoptotic members of the BCL-2 family tightly hypocellularity observed in the peripheral blood, bone marrow control the intrinsic apoptotic signaling pathway. The pro- (BM) cellularity is often normal or even elevated.2 This disparity apoptotic BCL-2 members can be separated into two classes, between BM cellularity and cytopenia has been attributed to an which include the BAX/BAK-like proteins and the BH3-only increased level of apoptosis in the BM compartment.2 Interest- proteins. The BAX/BAK-like proteins form pores in the outer ingly, the level of cell death in BM cells differs substantially mitochondrial membrane to release pro-apoptotic factors such as 14 between different clinical MDS risk categories.2,3 This is exempli- cytochrome c. The BH3-only proteins exhibit their pro-apoptotic fied by the fact that the level of apoptosis in low-risk MDS is function by either sequestering the pro-survival BCL-2 family substantially higher than the levels of apoptosis observed in high- members away from BAX or BAK or, alternatively, by directly risk MDS/secondary acute myeloid leukemia (sAML) or in healthy activating them.15,16 Together, this results in the activation of the control BM cells.5–11 This illustrates that MDS cells acquire an caspases-3 and -7 and subsequently in cell death.17 In contrast, apoptotic resistance upon disease progression, which coincides pro-survival proteins such as the BCL-2 family members BCL-2, with the appearance of an elevated number of myeloid blasts and MCL-1 or BCL-XL or, alternatively, the inhibitor of apoptosis an increased likelihood of disease acceleration into sAML.2,12 proteins, protect cells from apoptosis by blocking the activation of – Previous work has shown that the balance between pro- and caspases.14 16,18 anti-apoptotic BCL-2 family proteins in the BM of MDS patients is The functional role of BCL-2 proteins to apoptotic resistance in deregulated. Using flow cytometry for quantification of protein MDS can be studied using pro-apoptotic BH3-mimetic com- expression, the ratio between the expression levels of BCL-2 pounds. BH3 mimetics bind into the BH3 groove of pro-survival and BCL-XL was compared with the levels of pro-apoptotic BAX BCL-2 proteins thereby displacing them from their inhibitory – and BAD in primary MDS samples. In higher-risk MDS patients, binding to BAX or BAK.19 22 We utilized the BH3 mimetics ABT-737 the balance was shifted toward the anti-apoptotic BCL-2 and the BCL-2-specific compound ABT-199. ABT-737 binds with family members. In addition, elevated BCL-2 expression was high affinity (Ki ⩽ 1nM) to BCL-2, BCL-XL and BCL-W, but not to also observed by immunocytochemistry on BM smears from MCL-1 or A1.19–22 The BCL-2-selective compound termed ABT-199 high-risk MDS patients, which showed significantly more BCL-2- has a sub-nanomolar affinity for BCL-2 (Ki o0.010 nM) and binds

1III. Medizinische Klinik für Hämatologie und Internistische Onkologie, Klinikum rechts der Isar, Technische Universität München, Munich, Germany; 2Klinik für Orthopädie und Sportorthopädie, Klinikum rechts der Isar, Technische Universität München, Munich, Germany; 3Institut für Klinische Chemie und Pathobiochemie, Klinikum rechts der Isar, Technische Universität München, Munich, Germany and 4Gemeinschaftspraxis Hämato-Onkologie Pasing, Munich, Germany. Correspondence: Dr PJ Jost, III. Medical Department for Hematology and Oncology, Klinikum rechts der Isar Technical University of Munich, Ismaninger Strasse 22, Munich 81675, Germany. E-mail: [email protected] Received 15 January 2015; revised 24 June 2015; accepted 25 June 2015; accepted article preview online 8 July 2015; advance online publication, 28 July 2015 Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 113 18 less avidly to BCL-XL (Ki =48nM) or BCL-W (Ki = 245 nM). It has no Numbers of erythroid progenitor colonies (Burst-forming units-erythroid 23 measurable binding to MCL-1 (Ki4444 nM). ABT-199 currently or colony-forming units for the granulocytic-macrophagic lineage, holds great promise for the treatment of B-cell neoplasias24,25 and and multi-potential granulocytic-erythroid-macrophagic-megakaryocytic initial data also suggest efficacy in de novo AML.26,27 lineage) were assessed after 14 days. Transmitted light photographs were Here, we studied the effect of ABT-737 and ABT-199 on BM cells obtained on a Keyence BIOREVO BZ-900 microscope. from a large cohort of primary human MDS patients and compared it with a cohort of healthy age-matched controls. We Flow cytometry found that ABT-737 and ABT-199 efficiently killed primary stem/ BMMNCs were stained with Annexin V-FITC in AnnexinV staining solution progenitor cells as well as more differentiated BM cells from high- (0.1M HEPES/NaOH, pH 7.4, 1.4M NaCl 0.9%, 25 mM CaCl2), followed by risk MDS/sAML patients, whereas low-risk MDS or healthy controls staining with fluorescently labeled antibodies against CD34 (clone 4H11), remained unaffected. CD45 (clone 2D1) or isotype control (clone P3.6.2.8.1). Dead cells were excluded by 7-aminoactinomycin D (7AAD) staining. For intracellular staining, cells were stained against CD34, followed by fixation in PATIENTS AND METHODS 2% paraformaldehyde, permeabilization using perm/wash buffer (BD Bioscience, Franklin Lakes, NJ, USA) and subsequent staining with Patient samples and cell lines fluorescently labeled antibodies against BCL-2 (clone Bcl-2/100, BD Human BM samples were collected according to the institutional guide- Bioscience), BCL-XL (clone 54H11, Cell Signalling, Cambridge, UK), MCL-1 lines and in concordance with the Declaration of Helsinki. Written informed (clone 19C4, WEHI, Melbourne, VIC, Australia) or respective isotype controls consent was obtained from each patient. The investigation was approved (Cat.: 556357, BD Bioscience; clone DA1E, Cell Signaling; clone eBRG1). by the Local Ethics Committee of the University Hospital of the Technical Dead cells were excluded by Fixable Viability Dye staining. If not otherwise University in Munich. sAML was defined as ⩾ 20% of blasts in the BM and a stated, reagents and antibodies were purchased from eBioscience. Flow history of MDS. All other MDS samples or sAML were classified according to analysis was performed on a BD FACS Canto II (BD Bioscience) and data the International Prognostic Scoring System (IPSS), the revised WHO were analyzed using FlowJo software (TreeStar Inc., Ashland, OR, USA). classification-based Prognostic Scoring System (r-WPSS), the World Health Organization (WHO) 2008 classification or the cytogenetic risk score according to Schanz et al.28 Samples were obtained when clinically Gene expression analysis required from patients either before or during treatment and irrespective Gene expression analysis was performed using the Human Genome U133 of the therapeutic regimen. Control samples were obtained from human Plus 2.0 Array from Affymetrix (Santa Clara, CA, USA). The Affymetrix femoral heads discarded after implantation of total endoprosthesis of the normalization method was used. All expression measurements of each hip joint from 57 hematologically healthy age-matched donors. The array are divided by the median (calculated across all calls (present, mixed embryonic liver-derived stromal cell line EL08-1D2 was used as a stromal and absent)) and plotted on a logarithmic scale to normalize the data and support cell line as described previously.29–31 show a log2 median-centered intensity blot. BMMNCs of 66 MDS patients (refractory anemia: n = 3, refractory anemia with ringed sideroblasts (RARS): n = 6, refractory cytopenia with multilinear dysplasia (RCMD): n =7, Cell isolation and culture refractory cytopenia with multilinear dysplasia and ringed sideroblasts Mononuclear cells from primary human BM samples were isolated via (RCMD-RS) n = 6, RAEB-1 (refractory anemia with blast excess 5-9%): n = 23, density-gradient centrifugation using the Biocoll Separation Solution RAEB-2 (refractory anemia with blast excess X10%): n = 21) and peripheral ’ (Biochrom AG, Berlin, Germany) following the manufacturer s instructions. blood mononuclear cells from 110 healthy controls were analyzed. To test + fi + CD34 cells were puri ed via positive selection using the CD34 for any significant differences of gene expression between the MDS MicroBeads kit (Miltenyi Biotec, Bergisch Gladbach, Germany) and purity groups, ‘early’, ‘intermediate’, ‘late’ and healthy control one-way analysis of was confirmed to be at least 95%. BM mononuclear cells (BMMNCs) were 5 variances (ANOVA) were used with post-hoc pairwise comparisons in case cultured at a density of 5 × 10 cells/ml in serum-free media consisting of of significance. All reported P-values are two-sided, with a significance level fi Iscove's Modi ed Dulbecco's Medium (IMDM) with L-alanyl-L-glutamine of 0.05 and have not been adjusted for multiple testing (unpaired (IMDM GlutaMAX) with 20% BIT 9500 serum substitute (1% (w/v) bovine Student's t-test). Pairwise differences are presented with 95% confidence μ μ serum albumin, 10 g/ml insulin, 200 g/ml iron-saturated transferrin; intervals (CIs). Statistical analyses were performed using SPSS version StemCell Technologies, Vancouver, BC, Canada) and enriched with 19.0.0 (IBM Corporation, Armonk, NY, USA). recombinant human stem cell factor (100 ng/ml), FMS-related tyrosine kinase-3 ligand (100 ng/ml), thrombopoetin (10 ng/ml), interleukin-6 (5 ng/ml), interleukin-3 (10 ng/ml; all from R&D Systems, Minneapolis, Statistical analysis MN, USA), β-mercaptoethanol (10 μM; Gibco, Carlsbad, CA, USA) and low- To test for any significant differences in apoptosis induction after inhibitor density lipoproteins (4 μg/ml; Sigma-Aldrich, St Louis, MO, USA). To assess treatment between the different MDS/sAML risk categories one-way the impact of stromal cells, samples were seeded on the murine embryonic ANOVA was used with post-hoc pairwise comparisons in case of liver-derived cell line EL08-1D2 at a density of 5 × 105 cells/ml. EL08-1D2 significance. Pairwise differences are presented with 95% CIs. Comparing were cultured on 0.1% gelatin-coated 24-well plates in long-term two samples, the unpaired Student's t-test was used to test for any culture medium (StemCell Technologies) with 35% essential medium with significant differences between treated samples and control. All reported L-alanyl-L-glutamine (alpha-MEM GlutaMAX, Gibco), 15% FCS (StemCell P-values are two-sided, with a significance level of 0.05 and have not been Technologies), horse serum (5%; StemCell Technologies), penicillin adjusted for multiple testing. For the correlation of protein expression by and streptomycin (1%; Gibco), β-mercaptoethanol (10 mM; Merck, mean fluorescence intensity and MDS cell survival, the strength of the Darmstadt, Germany). association between mean fluorescence intensity ratio and viability under inhibitor treatment was calculated by the Spearman rank correlation and Inhibitor the functional relationship was described by linear regression analysis. Statistical analyses were performed using GraphPadPrism version 5.01 ABT-737 (Active Biochem, Maplewood, NJ, USA) and ABT-199 (AbbVie, (Graphpad Software, Inc., San Diego, CA, USA). North Chicago, IL, USA) were dissolved in dimethyl sulfoxide (DMSO) and used in a final concentration of 1 μM or as stated otherwise. DMSO was used at 0.001% as vehicle control. RESULTS Induction of apoptosis by ABT-737 or ABT-199 efficiently kills high- Colony formation assay risk MDS/sAML cells but not low-risk/intermediate-risk MDS cells Hematopoietic progenitors were assessed after treatment with ABT-737 Differential expression of pro-survival BCL-2 family members at (1 μM), ABT-199 (1 μM) or DMSO (0.001%) for 72 h in cytokine-supplemented, serum-free culture with or without stromal support. 1 × 104 BMMNCs distinct clinical stages of MDS have prompted the idea that the were plated in duplicates in methylcellulose medium supplemented clinical progression to higher-risk disease is accompanied by an 2,9,13 with an optimal cytokine mix according to the manufacturer's acquired resistance to apoptosis. To test whether differences protocols (MethoCult H4435 enriched; StemCell Technologies). in expression of critical pro- or anti-apoptotic BCL-2 proteins

CD34+

0h 72h untreated DMSO ABT-199 ABT-737

5 8.37 2.99 5 7.35 0.39 5 10.9 4.67 5 13.9 2.77 10 10 10 10 100

104 104 104 104 80 60 DMSO 3 3 3 3 10 10 10 10 40 ABT-199

low-risk 20 ABT-737 102 102 102 102 viable cells (%) 0 0 85.4 3.20 0 90.6 1.64 0 80.1 4.29 0 79.3 4.03 0244872 time (h) 0 102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105

5 2.93 0.96 5 5 12.5 2.52 5 15.1 4.20 10 10 0.92 0.38 10 10 100 80 104 104 104 104 60 DMSO 103 103 103 103 40 ABT-199 20 ABT-737 7-AAD viable cells (%)

int.-risk 2 2 2 2 10 10 10 10 0 0244872 0 93.7 2.39 0 0 79.8 5.16 0 75.6 5.09 94.5 4.18 time (h) 0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105

5 1.37 6.39 5 0.094 0.19 5 1.68 21.8 5 2.01 28.2 100 10 10 10 10 DMSO 80 ABT-199 104 104 104 104 60 ABT-737 40 103 103 103 103 20 viable cells (%) 102 102 102 102 0 high-risk/sAML0 int.-risk 244872 low-risk 0 0 0 0 time (h) high-risk/sAML 86.1 6.12 95.8 3.88 27.6 48.9 15.3 54.4

0102 103 104 105 0102 103 104 105 0102 103 104 105 0102 103 104 105

Annexin V Figure 1. ABT-737 or ABT-199 kill stem/progenitor cells from high-risk MDS/sAML patients in a time-dependent manner. (a) Viability of BMMNC measured by ﬂow cytometry using Annexin V and 7AAD and gating on CD34+ cells from individual representative MDS patients diagnosed with low-risk, intermediate-risk (Int.-risk) MDS or high-risk MDS/sAML after treatment with ABT-199 (1 μM), ABT-737 (1 μM) or DMSO for the indicated time points. (b) Viability of cells treated with ABT-199 or ABT-737 as in a represented as percentage of viable cells after inhibitor treatment and viable cells after vehicle treatment for the indicated time points. Samples are classiﬁed using r-WPSS.

might explain the differences in apoptotic susceptibility, we The clonal hierarchy within cancers often defines a stem cell-like analyzed the gene expression data from a large cohort of MDS population and a more differentiated cellular population, which patient samples and respective controls.32 The gene expression has recently also been detected in MDS.38 To differentiate analysis showed a significantly decreased expression of the BCL-2 between MDS-propagating cells and a more differentiated cellular + inhibiting BH3-only protein BIM in all seven BIM probe sets in late subpopulation, we gated on CD34 cells by flow cytometry, the stages of MDS (RAEB-2) when compared with early stages of cellular subpopulation harboring, at least in part, the MDS- 9,10,38,39 MDS (refractory anemia, RARS, RCMD, RCMD-RS; Supplementary propagating cells. Of note, all samples were used directly Figure 1). BIM is a pro-apoptotic member of the BCL-2 family that after extraction from the BM to retain optimal survival properties binds and inhibits all pro-survival BCL-2 family members and and to prevent apoptosis induction by prior deep-freezing. potently induces apoptosis.15 The reduced expression of pro- Consistent with our hypothesis, we observed a dose-dependent apoptotic BIM in higher-risk MDS supports the hypothesis that an induction of cell death in three representative high-risk/sAML acquired resistance to apoptosis develops upon disease progres- samples (Supplementary Figure 2A). Owing to the high level of heterogeneity of human MDS and in accordance with previous sion. Of note, we did not observe clear differences in the publications on myeloid malignancies,40 we found inter-individual expression levels of any of the pro-survival BCL-2 family members, differences in drug sensitivity.27 Sample two and three proved to which might be explained by the regulation of these proteins on be highly sensitive to ABT-737, with ~ 50% of cells undergoing the transcriptional or post-translational level.33–35 cell death at doses as low as 10 nM. Sample one was more resistant The reduced BIM expression in higher-risk MDS samples showing a drop in viability at higher doses (Supplementary suggested that re-activation of apoptosis by BH3 mimetics such Figure 2A). fi as ABT-737 or ABT-199 might be speci cally effective in higher-risk ABT-737 also showed a time-dependent toxicity mostly indu- MDS samples. To functionally test the apoptotic susceptibility of cing cell death within the first 24 h of treatment but effectively MDS cells, we obtained BMMNCs from MDS patients, who killing all cells after 72 h in a representative high-risk/sAML sample underwent BM aspiration for diagnosis, disease monitoring or (Figures 1a and b, high-risk/sAML). Interestingly, stem/progenitor evaluation of treatment response. BM aspirates of 104 patients cells from representative low-risk or intermediate-risk MDS with MDS and 20 patients with sAML were analyzed. MDS patients patients remained largely unaffected by treatment with ABT-737 were classified according to the WHO classification,36 the IPSS12 or (Figures 1a and b, low-risk and intermediate-risk). This supported r-WPSS37 scores and the cytogenetic risk score according to the notion that ABT-737 was specifically toxic for cells from Schanz et al.28 (Supplementary Table 1). patients that had experienced disease acceleration, a clinical

Leukemia (2016) 112 – 123 © 2016 Macmillan Publishers Limited Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 115 situation that has previously been associated with an acquired utilized femoral heads discarded after hip replacement surgery resistance to apoptosis.2,3 from 57 age-matched donors and purified CD34+ progenitor cells A similar pattern of cell death induction was observed in when required (Supplementary Table 1B). samples treated with the BCL-2-selective inhibitor ABT-199 when In contrast to the toxic effect observed in high-risk MDS cells, directly compared with ABT-737. A representative low-risk and ABT-737 (Figures 2a and b) or ABT-199 (Figures 2c and d) did not intermediate-risk samples remained unaffected by treatment with induce apoptosis in healthy CD34+-purified stem/progenitor cells. ABT-199 for 72 h (Figures 1a and b, low-risk and intermediate-risk), Moreover, the bulk of more differentiated BM cells extracted from whereas progenitor cells from a high-risk/sAML sample rapidly age-matched controls also remained unaffected by treatment with underwent apoptosis (Figures 1a and b, high-risk/sAML). ABT-737 (Supplementary Figure 2B,C) or treatment with ABT-199 As previously described, we identified an elevated number of (Supplementary Figure 2D and E). This showed that the healthy apoptotic cells (7AAD+) in low-risk samples when compared with hematopoietic compartment of elderly patients remained unaffected intermediate-risk samples at baseline (Figure 1a).2,8,9,13 However, by BH3-mimetic compounds and suggested that BH3 mimetics because of the variable sample quality, the differences between might be safely employed in MDS patients. The data also support intermediate-risk and high-risk samples were less prominent the notion that pro-apoptotic cues present specifically in MDS (Figure 1a, 0 h). Irrespective of the level of apoptosis at baseline, cells from higher-risk MDS patients predispose to apoptosis the response to ABT-199 or ABT-737 treatment was strikingly induction by BH3-mimetic compounds. different between the clinical MDS risk groups. Reactivation of apoptosis overcomes the acquired apoptotic Healthy age-matched hematopoietic stem/progenitor cells remain resistance in high-risk MDS/sAML unaffected by BCL-2 inhibition High-risk MDS/sAML cells largely resist spontaneous cell It is important to preserve the healthy remaining hematopoiesis in death.2,5–11As BH3 mimetics are capable of reactivating mitochon- MDS patients to avoid additional cytopenia. To exclude toxic drial apoptosis, we tested whether such compounds override the effects of ABT-737 or ABT-199 on the healthy hematopoiesis, we apoptotic resistance observed in higher-risk MDS using a large monitored apoptosis induction in BM hematopoietic cells of cohort of primary patient samples (Supplementary Table 1).22,41,42 healthy, age-matched controls (Supplementary Table 1B). We The molecular, cytogenetic and clinical heterogeneity of MDS is utilized age-matched controls to mimic the patient population exceptionally high, making it difficult to conclude cellular accurately, as the hematopoietic capacity of the BM changes outcomes from only limited patient samples or even cell line substantially in elderly patients. To obtain control samples, we models. We therefore tested the induction of apoptosis in primary

CD34+ CD34+

0h 72h 100 untreated DMSO ABT-737 80 105 6.04 5.05 105 6.4 1.64 105 12.2 2.65

104 104 104 60

103 103 103 40 ABT-737 7AAD 2 2 2 DMSO 10 10 10 viable cells (%) 20 0 87.2 1.75 0 86.9 5.02 0 80.2 4.97 0 2 3 4 5 2 3 4 5 2 3 4 5 0 10 10 10 10 0 10 10 10 10 0 10 10 10 10 0 244872 Annexin V time (h)

CD34+ CD34+

0h 72h 100 untreated DMSO ABT-199

5 11.4 0.29 5 15.4 0.70 5 22.6 0.76 80 10 10 10

104 104 104 60

103 103 103 40

7AAD ABT 199 2 2 2 viable cells (%) 20 10 10 10 DMSO 0 88.3 0.05 0 83.8 0.096 0 76.6 0.03 0 0 102 103 104 105 0 102 103 104 105 0 102 103 104 105 0244872 time (h) Annexin V Figure 2. Stem/progenitors from healthy age-matched control bone marrow remain unaffected by BCL-2 family inhibition. (a) BMMNCs of an + individual healthy age-matched donor after enrichment for CD34 cells to a purity of 495% were subjected to ABT-737 (1 μM). Apoptosis was measured after 72 h. (b) Shown is the viability over time of BMMNC from an individual healthy donor after treatment with ABT-737 (1 μM)asin a. Viable cells are represented as percentage of viable cells at the start of the experiment and compared with vehicle control. (c) BMMNCs of an individual healthy age-matched donor treated with ABT-199 (1 μM)asina.(d) Shown is the viability over time of BMMNC after treatment with ABT-199 (1 μM)asinc. Viable cells are represented as percentage of viable cells at the start of the experiment and compared with vehicle control.

© 2016 Macmillan Publishers Limited Leukemia (2016) 112 – 123 Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 116 ** *** *** * *** *** *** ** *** ** *** ** *** *** ** *** *** * *** * *** * *** *

100 100 100 100 80 80 80 80 60 60 60 60

40 40 IPSS 40 40 r-WPSS 20 20 20 20 viable cells (%) viable cells (%) viable cells (%) ABT-737 ABT-199 viable cells (%) ABT-737 ABT-199 0 0 0 0

low (7) high (1) low (5) high (4) sAML (5) interm. (9) sAML (10) healthy (7) interm. (7) healthy (7) healthy (15) healthy (15) interm.2 (4) interm.2 (4) (very) low (8) (very) low (8) interm.1 (12) interm.1 (11) (very) high (12) (very) high (10)

* *** * ** *** * *** ** * *** * * *** *** *** *** * *** *** * **

100 100 100 100 80 80 80 80 60 60 60 60

WHO 40 40 40 40 category risk score

Cytogenetic 20 20 viable cells (%)

20 20 viable cells (%) viable cells (%) ABT-737 viable cells (%) ABT-199 ABT-737 ABT-199 0 0 0 0

sAML (5) good (27) good (23) sAML (10) RARS (4)RCMD (8) interm. (7) healthy (7) interm. (4) RCMD (11) RAEB-2 (7) healthy (7) RAEB-2 (3) healthy (15) healthy (15) RAEB-1 (11) RAEB-1 (10) very poor (5) very poor (3) Figure 3. Toxicity of ABT-737 or ABT-199 treatment correlates with MDS disease progression to elevated clinical risk groups. (a) Viable BMMNCs of 15 healthy donors and 8 patients with r-WPSS-based classification as (very) low-risk MDS, 9 intermediate-risk MDS, 12 (very) high- risk MDS and 10 patients with sAML were treated for 72 h with ABT-737 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.001) and post-hoc pairwise t-tests as shown *Po0.05, **Po0.005 and ***Po0.0005), mean differences and 95% CI are listed in Supplementary Figure 7. (b) Viable BMMNC of 7 healthy donors and 8 patients with r-WPSS-based classification as (very) low-risk MDS, 7 intermediate-risk MDS, 10 (very) high-risk MDS and 5 patients with sAML were treated for 72 h with ABT-199 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.0001) and post-hoc pairwise t-tests as shown (*Po0.05, **Po0.005 and ***Po0.0005). Mean differences and 95% CI are listed in Supplementary Figure 7. (c) BMMNCs from 15 healthy donors, 7 patient with IPSS-based classification as low-risk, 12 patient with intermediate- 1 risk, 4 patients with intermediate-2 risk and 1 patient with high-risk according to IPSS were treated with ABT-737 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.0001) and post-hoc pairwise t-tests as shown (*Po0.05, **Po0.005 and ***Po0.0005). Mean differences and 95% CI are listed in Supplementary Figure 7. (d) BMMNCs from 7 healthy donors, 5 patient with IPSS-based classification as low-risk, 11 patient with intermediate-1 risk, 4 patients with intermediate-2 risk and 4 patients with high-risk were treated with ABT-199 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.0036) and post-hoc pairwise t-tests as shown (*Po0.05, **Po0.005 and ***Po0.0005), mean differences and 95% CI are listed in Supplementary Figure 7. (e) Viable BMMNCs from 15 healthy donors and samples from 11 patients with WHO classification-based RCMD (refractory cytopenia with multilinear dysplasia), 11 patients RAEB-1 (refractory anemia with blast excess 5-9%), 7 patients RAEB-2 (refractory anemia with blast excess ⩾ 10%) and 10 patients with sAML were treated with ABT-737 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.0001) and post-hoc pairwise t-tests as shown (*Po0.05, **Po0.005 and ***Po0.0005), mean differences and 95% CI are listed in Supplementary Figure 7. (f) Viable BMMNCs from 7 healthy donors and samples from 4 patient with WHO classification- based RARS, 8 patient with RCMD, 10 patients RAEB-1, 3 patients RAEB-2 and 5 patients with sAML were treated with ABT-199 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.0018) and post-hoc pairwise t-tests as shown (*Po0.05, **Po0.005 and ***Po0.0005), mean differences and 95% CI of differences are listed in Supplementary Figure 7. (g) BMMNCs from 15 healthy donors, 27 patients with good prognosis, 7 patients with intermediate prognosis and 5 patients with very poor prognosis according to the cytogenetic risk score by Schanz 28 et al. were treated with ABT-737 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.0001) and post-hoc pairwise t-tests as shown (*Po0.05, **Po0.005 and ***Po0.0005), mean differences and 95% CI are listed in Supplementary Figure 7. (h) BMMNCs from 7 healthy donors, 23 patients with good prognosis, 4 patients with intermediate prognosis and 3 patients with (very) poor prognosis according to the 28 cytogenetic risk score by Schanz et al. were treated with ABT-199 (1 μM) or vehicle control (DMSO). One-way ANOVA (Po0.0541). Mean differences and 95% CI are listed in Supplementary Figure 7. All panels: each circle represents the ratio between viable cells after inhibitor treatment and viable cells after vehicle treatment for 72 h. Cells were gated for CD34 expression and stained for viability as in Figure 1. Shown is the mean and error bars denote standard deviation.

MDS cells from patients diagnosed with different clinical MDS risk apoptosis induction afforded by ABT-737 was similar in all samples categories after treatment with ABT-737 or ABT-199. We subjected within an individual risk category, which pointed toward a risk 39 primary BMMNC samples from MDS/sAML patients to ABT-737 group-specific effect. This is exemplified by the fact that treatment and directly compared this cohort with 15 healthy significant differences in cell death induction were observed even control subjects. between closely related risk groups such as (very) low-risk and When compared with healthy controls, we found that induction intermediate-risk patients or intermediate-risk and high-risk of apoptosis was significantly increased in the stem/progenitor patients (Figure 3a). population from r-WPSS-classified patients with intermediate-risk In the 24 primary MDS patient samples that were sufficiently MDS, (very) high-risk MDS or sAML (Figure 3a). Of note, the level of annotated to calculate the IPSS, we found significant differences in

Leukemia (2016) 112 – 123 © 2016 Macmillan Publishers Limited Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 117 cell death induction when comparing healthy control samples to categorized high-risk/sAML samples significantly better than intermediate-risk 1 or intermediate-risk 2 samples. We also healthy or low-risk samples (Figure 3b). observed significant differences in cell death induction when A similar elevation in cell death was observed in high-risk comparing apoptosis in low-risk samples to intermediate-risk 1 samples compared with healthy controls after ABT-199 treatment samples or intermediate-risk 2 samples (Figure 3c). when using the IPSS (Figure 3d). High-risk samples (RAEB-1 and The same pattern was observed when patients were classified sAML) also underwent apoptosis more readily than healthy according to the WHO 2008 classification.36 This classification is controls/RARS samples when the WHO classification was utilized based primarily on histomorphological parameters and does not (Figure 3f). include clinical parameters. Significant differences for induction of In summary, the level of apoptosis observed in CD34+ stem/ apoptosis in CD34+ cells were detected between all MDS progenitor cells from MDS patients after ABT-737 and to a slightly categories except RAEB-1 vs RAEB-2 (Figure 3e). lower degree also after ABT-199 treatment correlated with an The substantial heterogeneity of human MDS is partly explained elevated clinical risk category. These data suggested that pro- by cytogenetic aberrations that are increasingly used for prog- apoptotic intervention is a powerful tool to overcome apoptotic nostication. We therefore subjected our cohort to the cytogenetic resistance in high-risk/sAML cells, whereas healthy or low-risk cells risk score published by Schanz et al.28 Using this risk score, we remained largely unaffected. tested whether cytogenetic aberrations preclude the treatment effect of ABT-737 or ABT-199. We observed significant differences ABT-737 or ABT-199 specifically target colony-forming stem/ comparing apoptosis in samples from healthy controls to the level progenitor cells from high-risk MDS/sAML patients of apoptosis observed in good prognosis samples, intermediate MDS is a clonal disorder driven by the transformation of a prognosis samples or very poor prognosis samples when treated hematopoietic stem cell.38,43 It is therefore critical to target the with ABT-737 (Figure 3g) and a trend toward significant stem/progenitor compartment to combat the disease. Hence, we differences in ABT-199-treated samples (one-way ANOVA interrogated whether ABT-737- or ABT-199-induced killing of P = 0.0541, Figure 3h). CD34+ cells had any impact on the colony-forming capacity. We Clinically used scoring systems use composite clinical variables subjected primary MDS samples to ABT-737 and/or ABT-199 to define different risk categories. To better understand which treatment for 72 h before cultivation in growth factor- individual parameter might differentiate between resistant and supplemented methylcellulose to monitor colony formation. Of sensitive MDS samples, we also grouped our cohort by the note, the cells were treated with the compounds for 72 h in liquid presence of cytopenia as the sole parameter. In line with our data culture, whereas the cultivation on methylcellulose was carried on from IPSS or r-WPSS-grouped samples and in line with the notion for 14 days without the drugs. that higher-risk samples are more sensitive, we found that In line with our fluorescence-activated cell sorting-based data, samples from patients with two or three cytopenias underwent the number of colonies in healthy controls was not affected apoptosis significantly better than healthy controls or patients by ABT-737 treatment as exemplified by the numbers of with only one cytopenia (Supplementary Figure 6). The blast count individual colony types (Figure 4a). ABT-199 also proved as single parameter is best represented by the WHO category non-toxic to any colony type in healthy age-matched control (Figure 3e). samples (Figure 4b). Similarly, colony numbers in samples from The significant differences observed between lower-risk sam- (very) low-risk (Figures 4a and b, low-risk) or intermediate-risk ples compared with higher-risk samples in all classification MDS patients (Figures 4a and b, intermediate-risk) remained systems available emphasized the uniformity of the apoptotic largely unaffected by treatment with either compound when response for each individual risk group. This argues that the level compared with vehicle control. In contrast, ABT-737 and of pro-apoptotic cues (also referred to as ‘priming for cell death’) ABT-199 treatment substantially decreased colony formation in increases as the disease progresses over time, rendering BH3 high-riskMDS/sAMLsamples(Figures 4a and b). This further mimetics effective only in higher-risk samples. supported our previous finding that BH3-mimetic compounds To understand whether ABT-737 also influences the cellular are selectively toxic for progenitor cells from higher-risk survival of more differentiated cellular subpopulations, we MDS/sAML patients. measured the induction of apoptosis in the bulk of BMMNC. We In addition, we treated a small cohort with both compounds in found a slightly reduced propensity of ABT-737 to kill mature parallel. We found a striking similarity in the numbers of overall BMMNC compared with the CD34+ subpopulation. When patients colonies especially in high-risk and sAML samples (Figure 4c). were grouped according to the r-WPSS, significant differences A significant and very similar reduction in the progenitor potential were detected between (very) low/intermediate-risk samples was observed after treatment with either compound. compared with (very) high-risk MDS samples or compared with Collectively, these data showed that ABT-737 or ABT-199 sAML samples (Supplementary Figure 3A). Similar results were targeted the stem/progenitor cell-containing subpopulation in obtained when patients were grouped by the IPSS or the WHO high-risk MDS/sAML but not in lower-risk MDS or healthy controls. category (Supplementary Figures 3B and 4A). For the cytogenetic Both compounds exhibited a very similar activity in blocking MDS risk score of Schanz et al.,28 statistical significant differences were progenitor potential. The lack of toxicity in healthy progenitors only detected comparing healthy controls to MDS samples supports the notion that this class of compounds might be safely (Supplementary Figure 4B). used in MDS patients as it avoids toxic effects on the healthy We also measured apoptosis following treatment with the hematopoiesis. BCL-2 selective compound ABT-199, which is currently in clinical development and holds promise for treatment of various B-cell MCL-1 expression conveys resistance against ABT-737 or ABT-199 neoplasias and de novo AML. Elevated apoptosis induction by in MDS cells ABT-199 was observed in stem/progenitor cells from higher-risk We next sought to define the molecular mechanisms of resistance samples but not in lower-risk samples and exhibited a similar to ABT-737 or ABT-199. We therefore subjected a cohort of MDS pattern to that observed after treatment with ABT-737. However, samples to flow cytometry-based intracellular staining for protein the overall capacity of the BCL-2-selective compound ABT-199 to expression of BCL-2, BCL-XL and MCL-1. In addition, we treated the induce apoptosis was lower compared with ABT-737 in all risk same samples with either compound for 72 h to directly compare categories (Figures 3b, d, f and h), suggesting some protective cell death induction of each individual sample to the respective effect by pro-survival BCL-xL. Nevertheless, ABT-199 killed r-WPSS- protein expression. In line with our data from the liquid culture

healthy (n=7) low (n=6) int (n=5) high (n=5) sAML (n=4) 80 80 100 150 200 * CFU-GM BFU-E 80 CFU-GEMM cells cells cells cells 60 cells 60 150 4 4 4 4 4 100 60 40 40 100 40 50 20 20 50 20 # colonies/1x10 # colonies/1x10 # colonies/1x10 # colonies/1x10 # colonies/1x10 0 0 0 0 0

DMSO DMSO DMSO DMSO DMSO ABT-737 ABT-737 ABT-737 ABT-737 ABT-737

healthy (n=7) low (n=4) int (n=6) high (n=4) sAML (n=3) 80 40 100 60 200 * CFU-GM BFU-E 80 cells cells cells cells cells 60 150 CFU-GEMM 4 4 4 4 4 40 60 40 20 100 40 20 20 50 20 # colonies/1x10 # colonies/1x10 # colonies/1x10 # colonies/1x10 # colonies/1x10 0 0 0 0 0

DMSO DMSO DMSO DMSO DMSO ABT-199 ABT-199 ABT-199 ABT-199 ABT-199

** 200 DMSO * ABT-199 cells

4 150 ABT-737 * 100 * 50

# colonies/1x10 0

sAML (3) int.-risk (3) low-risk (3) high-risk (2) healthy BM (2) Figure 4. BH3 mimetics ABT-737 or ABT-199 kill colony-forming stem/progenitor cells from high-risk MDS/sAML patients but not from healthy or low/intermediate (Int.)-risk patients. (a) BMMNCs (1 × 104) were plated in growth factors-enriched methylcellulose after 72 h of treatment with ABT-737 (1 μM). Numbers of colony-forming units (CFU) of multi-potential granulocytic-erythroid-macrophagic-megakaryocytic lineage (CFU-GEMM), CFU-granulocytic-macrophagic lineage (CFU-GM), burst-forming units-erythroid (BFU-E) and total number of colonies were scored after 14 days for the indicated patient samples. Shown is the mean of the total colony numbers as stacked bar chart of the single colony types for the indicated risk groups/sAML. Replicates as shown in the figure. Error bars denote standard deviation. Samples are classified using r-WPSS. P-values as shown in the figure. (b) BMMNCs (1 × 104) were plated in growth factors-enriched methylcellulose after 72 h of treatment with ABT-199 (1 μM). Numbers of CFU-GEMM, CFU-GM, BFU-E and total number of colonies were scored after 14 days for the indicated patient samples. Shown is the mean of the total colony numbers as stacked bar chart of the single colony types for the indicated risk groups/sAML. Replicates as indicated in the figure. Error bars denote standard deviation of total colony number. Samples are classified using r-WPSS. P-values as shown in the figure. (c) Shown in the direct comparison in the total number of colonies from two healthy age-matched control samples, three low-risk MDS, three int.-risk MDS, two high-risk MDS and three sAML samples treated in parallel with ABT-737 or ABT-199 as described in a or b, respectively. Experiments were performed in duplicates and error bars denote standard deviation. Samples are classified using r-WPSS. P-values: *Po0.05 and **Po0.005.

and the colony formation assays, we found that most low-risk MDS MDS risk groups (Figure 5, compared #1 to #3 with #6 to #8). The samples remained unaffected by treatment with ABT-737 or critical role of MCL-1 in treatment resistance can also be ABT-199, whereas most high-risk MDS samples induced apoptosis appreciated in an individual high-risk MDS samples with elevated substantially better (Figure 5). Interestingly, we found that the MCL-1 expression, which proved insensitive to the treatment with resistance to cell death induction by ABT-737 or ABT-199 was either BH3-mimetic despite being collected from a high-risk largely defined by the expression of the pro-survival BCL-2 family patient (Figure 5, #9). In addition, we observed that the expression member MCL-1 (Figure 5, compared #1 with #9). This is not of BCL-XL, albeit lower than MCL-1 expression, was capable of unexpected since both compounds are unable to inhibit largely blocking cell death induction by ABT-199 but not by MCL-1. MCL-1 expression was found to be substantially higher ABT-737, which is best exemplified in samples #2 or #5 and in lower-risk MDS samples compared with high-risk samples, corresponds to the previously reported binding affinities of this indicating that differential expression of this protein likely explains compound (Figure 5). Of note, expression of BCL-2 was lower the discrepancy in treatment response to BH3 mimetics in diverse compared with the expression of BCL-XL or MCL-1, suggesting that

Leukemia (2016) 112 – 123 © 2016 Macmillan Publishers Limited Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 119 ID viable cells (%) (total/CD34) BCL-2 BCL-xL MCL-1 1.2K 0.9K ABT-199 72.5/47.2 0.6K stained #1

ABT-737 63.7/34.6 count isotype sensitive 0.3K 0 0104 0104 0 104 3.0K

2.0K ABT-199 77.1/84.0 #2

ABT-737 63.9/50.0 count 1.0K high-risk

0 0104 0104 0104 3.0K

2.0K ABT-199 77.3/85.3 #3

ABT-737 79.2/86.8 count 1.0K

0 0104 0104 0104 3.0K

2.0K ABT-199 81.4/80.2 #4

ABT-737 77.8/75.1 count 1.0K

0 0104 0104 0104 3.0K

2.0K ABT-199 87.5/92.0 #5

ABT-737 75.4/78.8 count int-risk 1.0K

0 0104 0104 0104 3.0K

2.0K ABT-199 91.0/91.2 #6

ABT-737 84.2/76.6 count 1.0K

0 0104 0104 0104 3.0K

2.0K ABT-199 97.45/99.7 #7

ABT-737 86.4/76.3 count 1.0K

0 0104 0104 0104 3.0K low-risk

2.0K ABT-199 99.3/101.5 #8

ABT-737 89.2/99.3 count 1.0K

0 0104 0104 0104 3.0K

2.0K ABT-199 100.5/103.9 resistant #9

ABT-737 101.0/103.0 count 1.0K high-risk

0 0104 0104 0104 Figure 5. Elevated protein levels of MCL-1 convey resistance to ABT-737 or ABT-199 in primary MDS samples. Protein expression levels of BCL-2, BCL-xL and MCL-1 were measured by intracellular ﬂow cytometry before treating the primary human MDS cells with ABT-737, ABT-199 or DMSO for 72 h. Shown are isotype control (gray) and expression level of the indicated proteins (red) for each patient sample. The samples are listed from sensitive to resistant according to the viability of MDS BMMNC after 72 h of ABT-199 treatment. Viability is shown as percentage of viable cells after inhibitor treatment and viable cells after vehicle treatment for 72 h. The risk category was deﬁned using the r-WPSS. Cells were gated for CD34 expression and stained for viability as in Figure 1.

© 2016 Macmillan Publishers Limited Leukemia (2016) 112 – 123 Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 120 even low levels of this protein might be able to protect MDS cells define the differential sensitivity of MDS samples to the treatment (Figure 5). with ABT-737 or ABT-199. To check whether protein expression levels measured by intracellular flow cytometry might serve as biomarkers for the Stromal cells only partially protect high-risk MDS/sAML cells from response of MDS samples to BH3 mimetics before treatment, we cell death correlated protein abundance with cell survival. The strength of Mesenchymal stromal cells closely interact with hematopoietic the association between protein abundance measured by the stem/progenitor cells within the BM compartment.44–46 The fl mean uorescence intensity and the viability under inhibitor interaction is, at least in part, referred to as the hematopoietic treatment was calculated by the Spearman rank correlation. niche and supports stem/progenitor cell survival.47,48 Overcoming Subsequently, the functional relationship was described by linear the stromal cell protection of MDS cells is therefore considered an regression analysis (Supplementary Figure 5). This analysis showed important challenge for MDS therapy.49,50 To understand whether that the abundance of BCL-XL and MCL-1 clearly predicted stromal cells protect high-risk MDS/sAML progenitor cells from resistance to ABT-199, whereas elevated levels of BCL-2 predicted ABT-737-induced apoptosis, we co-cultured CD34+ BMMNC from good responses of ABT-199 (Supplementary Figures 5A and B). MDS patients in the presence of the stromal cell line EL08-1D2. For ABT-737, the level of MCL-1 predicted resistance when Despite the limitation of using a murine stromal cell line that directly compared with the levels of BCL-2 plus BCL-XL in the might protect human MDS cells only partially, EL08-1D2 cells have same sample (Supplementary Figure 5C). This linear regression been shown to support the maintenance of primitive human analysis supported the notion that the protein levels measured by cobblestone area-forming cells,51 facilitate human embryonic fluorescence-activated cell sorting are functionally relevant and stem cell differentiation into hematopoietic cells and support can therefore be used as biomarkers to predict response patterns their survival in vitro.31,52,53 against BH3-mimetic compounds. When comparing the effect of stromal cells in a cohort of high- Together, our intracellular flow cytometry data showed that the risk/sAML samples, we observed a marginally protective effect protein expression level of individual BCL-2 family members against ABT-737-induced killing (Figure 6a). Similar to our findings

*** high-risk #1 *** 100 high-risk #2 sAML 100 80 80 60 60 40 40 viable cells (%) 20 viable cells (%) 20

0 0 ABT-737: + + + ABT-737: + + Stromal cells: + + + Stromal cells: + - n: 6 3 9 high/ Risk group: healthy low sAML

Stromal cells only 100

40 ABT-737

viable cells (%) DMSO 20

0 0 244872 time (h) Figure 6. Stromal cells only marginally protect high-risk MDS cells against apoptosis induced by ABT-737. (a) CD34+ BMMNCs from six healthy donors and nine patients with high-risk MDS (denoted ‘high’) or sAML were cultured in the presence of the stromal cell line EL08-1D2 and treated with ABT-737 (1 μM) or vehicle control for 72 h. Each circle represents the ratio between viable cells after ABT-737 treatment and viable cells after vehicle treatment for 72 h. Cells were stained for viability as in Figure 1. Shown is the mean and error bars denote standard deviation. Samples are classiﬁed using r-WPSS. One-way ANOVA (Po0.001) and post-hoc pairwise t-tests as shown (***Po0.0005). Healthy vs low-risk: mean difference = 2.632, 95% CI (-10.01 to 15.28); healthy vs high-risk/sAML: mean difference = 38.48, 95% CI (29.06 to 47.90); low-risk vs high-risk/sAML: mean difference = 35.85. 95% CI (23.93 to 47.77). (b) CD34+ BMMNCs from two patients with high-risk MDS and one patient with sAML were cultured in the presence or absence of the stromal cell line EL08-1D2 and treated with ABT-737 (1 μM) or vehicle control for 72 h. Each circle represents the ratio between viable cells after ABT-737 treatment and viable cells after vehicle treatment for 72 h. Cells were stained for viability as in Figure 1. No statistical signiﬁcant difference could be detected comparing the group with stromal support vs without stromal support (P = 0.3264 using Student's t-test). (c) Viability of EL08-1D2 cells after treatment with ABT-737 (1 μM) is shown for the indicated time points. Viable cells are represented as percentage of viable cells at the start of the experiment and compared with vehicle control. Experiment was performed in triplicates and error bars denote standard deviation.

Leukemia (2016) 112 – 123 © 2016 Macmillan Publishers Limited Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 121 from the suspension culture, BMMNCs from nine high-risk MDS/ that will likely respond to treatment with BH3 mimetics. This sAML patients co-cultured on stromal cells underwent apoptosis supports the notion that apoptotic priming might serve as a significantly better than six healthy controls co-cultured on parameter independent of the molecular, cytogenetic or clinical stromal cells (Figure 6a). features present in individual patients. In addition, our samples When comparing the survival on stromal cells with the were collected irrespective of any treatment regimen thereby survival in liquid culture conditions, we found a minor protection including samples from patients at diagnosis, under first-line of stromal cell culture conditions in high-risk MDS samples treatment or after treatment failure. The largely uniform pattern of (Figure 6a: mean survival high/sAML with stroma = 54.98% response to ABT-199 or ABT-737 indicates that this class of compared with Figure 3a: mean survival high-risk without compounds exerted its potency primarily depending on the level stroma = 50.32%). However, a partial protection was observed of mitochondrial priming but not depending on parameters such in sAML samples when comparing the survival on stromal as previous treatment regimens. cells to liquid culture (Figure 6a: mean survival high/sAML with Colony formation showed that the pool of stem/progenitor cells stroma = 54.98% compared with Figure 3a: mean survival sAML in MDS efficiently induces apoptosis in response to ABT-737 or without stroma = 26.45%). ABT-199. This is an important feature as it suggests that BH3 To directly compare the effect of stromal cells, we subjected mimetics are capable of targeting the MDS cell of origin likely three high-risk MDS/sAML samples to ABT-737 with or without residing within the CD34+ progenitor cell compartment.38 Similar stromal cell support (Figure 6b). A partial protection by stromal fi cells was observed in only one sample, whereas the induction of ndings were reported in a MDS mouse model, where ABT-737 was capable of prolonging the survival of mice by targeting apoptosis proceeded irrespective of stromal cell support in two − + + 54,55 additional samples (Figure 6b). leukemia-initiating cells and primitive Lin /Sca1 /Kit cells. To exclude any toxicity of ABT-737 directly in stromal cells that However, to accurately recapitulate the high level of clinical and might cause the release of danger-associated molecular pattern molecular heterogeneity of human MDS, it is of critical importance molecules, which might in turn facilitate cell death in MDS cells, to test the effects of BH3-mimetic compounds on a large cohort or we measured apoptosis specifically in the stromal cells. In line with primary patient samples. a cell-intrinsic effect, we did not observe any relevant induction of Similar to previously published data that show MCL-1 expres- apoptosis in EL08-1D2 cells alone (Figure 6c). sion as the primary resistance mechanism to ABT-737 or ABT-199 56 Despite the limited sample size, our data suggest that stromal treatment in B-cell neoplasia, we found that MDS samples that cells only partially protected MDS cells from ABT-737-induced cell express high levels of MCL-1 were resistance to the treatment with death, implying that the inhibition of BCL-2 proteins primarily either compound. Most low-risk MDS samples exhibited high executed apoptosis in a cell-intrinsic manner and that niche MCL-1 levels rendering those samples resistant to the treatment. protection in the BM might not severely obstruct the clinical In line with our viability data from the liquid culture, most high- efficacy of pro-apoptotic compounds in patients. risk samples showed low MCL-1 levels and accordingly responded better to ABT-737 or ABT-199. Of note, individual resistant high- risk samples exhibited high MCL-1 levels indicating that MCL-1 DISCUSSION serves as the primary resistance factor irrespective of the disease Deregulated apoptosis is a key feature of MDS and an acquired risk category. These inter-individual differences in BCL-2 protein apoptotic resistance in the BM compartment has been associated expression were also found in AML, where BCL-2 expression varied 2,6–12 with higher-risk disease. This is likely caused by additional substantially between samples from a given French-American- molecular and cytogenetic aberrations acquired during the course British subgroup.27 Given the critical role of MCL-1 as a resistance of the disease, which is characteristic for the transition from biomarker, the detection of protein expression by intracellular 2,12 fi lower-risk MDS to higher-risk MDS. Our ndings show that flow cytometry might serve as a reasonable way of screening re-activation of mitochondrial apoptosis using compounds that fi patients before BH3-mimetic treatment. inhibit the pro-survival members of the BCL-2 family is speci cally Recent data obtained primarily from AML or CML cell lines or toxic to cells from higher-risk MDS or sAML patients and supports primary patient samples suggested that ABT-199 efficiently killed the notion that these cells experience an elevated level of blast cells.26,57 This finding is represented in our data by the loss of apoptotic stress.3 viability observed in sAML samples harboring an elevated number This feature of high-risk MDS or sAML cells is referred to as of blasts (Figure 3). This feature of BH3-mimetic together with the elevated priming for mitochondrial apoptosis, which is reminis- fi cent of the level of apoptotic priming previously reported for de previously described ampli cation of toxicity to azacitidine or novo AML blasts.40–42 Using a methodology termed BH3 profiling, mitogen-activated protein/extracellular signal-regulated kinase the therapeutic effect of chemotherapy was shown to correlate inhibition supports the notion that pro-apoptotic intervention in MDS alone or in combination should be further evaluated in the with the level of apoptotic priming in leukemic blasts. Poorly 26,27,58 primed and chemotherapy-resistant cells showed a striking clinical setting. resistance to apoptosis in response to synthetic BH3 peptides.42 Our data argue that pro-apoptotic drug treatment reduces In contrary, highly 'primed’ myeloblasts efficiently underwent the disease burden in higher-risk MDS or sAML patients by apoptosis after BH3-peptide treatment, which also correlated with selectively killing leukemic progenitors as well as blast cells the response of patients to first induction chemotherapy.42 This without significantly affecting the healthy progenitor cell finding and our data suggest that the resistance to apoptosis population. BH3-mimetic compounds might therefore represent imposed by anti-apoptotic BCL-2 family members is a defining an interesting approach to treat higher-risk MDS patients to hallmark in AML and also in MDS. The killing observed after delay progression into sAML or to perform a bridging therapy ABT-737 or ABT-199 treatment primarily in high-risk MDS/sAML for patients awaiting allogeneic stem cell transplantation, supports the notion that the clinical parameter 'high-risk’ in MDS especially since current therapeutic strategies such as azaciti- might be comparable to the parameter ‘mitochondrial priming’ dine have an only short responsedurationinthispatient – in AML.40 42 population.59 We validated the correlation between cell death and disease stage in four major MDS risk classification systems. Irrespective of the scoring system used, we found that the clinical category ‘high- CONFLICT OF INTEREST risk MDS’ or ‘sAML’ is a valid parameter to identify patient samples The authors declare no conflict of interest.

© 2016 Macmillan Publishers Limited Leukemia (2016) 112 – 123 Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 122 ACKNOWLEDGEMENTS 18 Davids MS, Letai A. ABT-199: taking dead aim at BCL-2. Cancer Cell 2013; 23: We thank T Haferlach from the Munich Leukemia Laboratory (MLL) for providing 139–141. gene expression data. We thanks T Haferlach from MLL and M Zingerle from the 19 Oltersdorf T, Elmore SW, Shoemaker AR, Armstrong RC, Augeri DJ, Belli BA et al. Gemeinschaftspraxis Hämato-Onkologie Pasing for providing MDS samples and An inhibitor of Bcl-2 family proteins induces regression of solid tumours. Nature 435 – clinical data and J Tuebel from the Klinik für Orthopädie und Sportorthopädie, 2005; :677 681. 11 Klinikum rechts der Isar for technical support. PJJ was supported by a Max Eder- 20 Letai A. BCL-2: found bound and drugged!. Trends Mol Med 2005; : – Program grant from the Deutsche Krebshilfe (program #111738), a Human Frontiers 442 444. Science Program grant (program #RGY0073/2012), a German Jose Carreras Leukemia 21 Cragg MS, Harris C, Strasser A, Scott CL. Unleashing the power of inhibitors of oncogenic kinases through BH3 mimetics. Nat Rev Cancer 2009; 9: Foundation grant (DJCLS R 12/22), and a research grant from the Deutsche 321–326. Forschungsgemeinschaft, Forschergruppe FOR2036 and Novartis for travel support. 22 Khaw SL, Huang DC, Roberts AW. Overcoming blocks in apoptosis with RAJO was supported by the German Research Foundation (DFG grants OO 8/5, OO BH3-mimetic therapy in haematological malignancies. Pathology 2011; 43: 8/9, and FOR 2033)). RAJO and KSG were supported by the German Jose Carreras 525–535. Leukemia Foundation grant (DJCLS R 11/12). KSG was supported by the German 23 Souers AJ, Leverson JD, Boghaert ER, Ackler SL, Catron ND, Chen J et al. ABT-199, a Research Foundation (Go 713/2-1) and the Deutsche Konsortium für Translationale potent and selective BCL-2 inhibitor, achieves antitumor activity while sparing Krebsforschung (DKTK) of the German Cancer Center (DKFZ). We thank Abbvie for platelets. Nat Med 2013; 19:202–208. supplying ABT-199. 24 Roberts AW, Seymour JF, Brown JR, Wierda WG, Kipps TJ, Khaw SL et al. Substantial susceptibility of chronic lymphocytic leukemia to BCL2 inhibition: AUTHOR CONTRIBUTIONS results of a phase I study of navitoclax in patients with relapsed or refractory disease. J Clin Oncol 2012; 30:488–496. PJJ conceived and supervised the project, analyzed the data and wrote the 25 Wilson WH, O'Connor OA, Czuczman MS, LaCasce AS, Gerecitano JF, Leonard JP manuscript. SJ performed the experiments, analyzed the data and wrote the et al. Navitoclax, a targeted high-affinity inhibitor of BCL-2, in lymphoid malig- manuscript. KG provided primary samples and clinical data and gave nancies: a phase 1 dose-escalation study of safety, pharmacokinetics, pharma- 11 – conceptual advice. VR, JK, UH, OG and CH performed experiments. JS, BS, RB, codynamics, and antitumour activity. Lancet Oncol 2010; : 1149 1159. 26 Pan R, Hogdal LJ, Benito JM, Bucci D, Han L, Borthakur G et al. Selective BCL-2 CM-T, H-JK, RAJO, JR and CP provided primary samples and clinical data, and Inhibition by ABT-199 Causes On-Target Cell Death in Acute Myeloid Leukemia. gave conceptual advice. Cancer Discov 2014; 4: 362–375. 27 Bogenberger JM, Kornblau SM, Pierceall WE, Lena R, Chow D, Shi CX et al. BCL-2 REFERENCES family proteins as 5-Azacytidine-sensitizing targets and determinants of response in myeloid malignancies. Leukemia 2014; 28: 1657–1665. 1 Tefferi A, Vardiman JW. Myelodysplastic syndromes. N Engl J Med 2009; 361: 28 Schanz J, Tuchler H, Sole F, Mallo M, Luno E, Cervera J et al. New comprehensive 1872–1885. cytogenetic scoring system for primary myelodysplastic syndromes (MDS) and 2 Corey SJ, Minden MD, Barber DL, Kantarjian H, Wang JC, Schimmer AD. oligoblastic acute myeloid leukemia after MDS derived from an international Myelodysplastic syndromes: the complexity of stem-cell diseases. Nat Rev Cancer database merge. J Clin Oncol 2012; 30:820–829. 2007; 7: 118–129. 29 Oostendorp RA, Medvinsky AJ, Kusadasi N, Nakayama N, Harvey K, Orelio C et al. 3 Raza A, Galili N. The genetic basis of phenotypic heterogeneity in myelodysplastic Embryonal subregion-derived stromal cell lines from novel temperature-sensitive syndromes. Nat Rev Cancer 2012; 12:849–859. SV40 T antigen transgenic mice support hematopoiesis. J Cell Sci 2002; 115: 4 Ma X, Does M, Raza A, Mayne ST. Myelodysplastic syndromes: incidence and 2099–2108. survival in the United States. Cancer 2007; 109: 1536–1542. 30 Oostendorp RA, Robin C, Steinhoff C, Marz S, Brauer R, Nuber UA et al. 5 Shetty V, Hussaini S, Broady-Robinson L, Allampallam K, Mundle S, Borok R et al. Long-term maintenance of hematopoietic stem cells does not require Intramedullary apoptosis of hematopoietic cells in myelodysplastic syndrome contact with embryo-derived stromal cells in cocultures. Stem Cells 2005; 23: patients can be massive: apoptotic cells recovered from high-density fraction of 842–851. bone marrow aspirates. Blood 2000; 96: 1388–1392. 31 Ledran MH, Krassowska A, Armstrong L, Dimmick I, Renstrom J, Lang R et al. 6 Raza A, Gezer S, Mundle S, Gao XZ, Alvi S, Borok R et al. Apoptosis in bone marrow Efficient hematopoietic differentiation of human embryonic stem cells on stromal biopsy samples involving stromal and hematopoietic cells in 50 patients with cells derived from hematopoietic niches. Cell Stem Cell 2008; 3:85–98. myelodysplastic syndromes. Blood 1995; 86:268–276. 32 Haferlach T, Kohlmann A, Wieczorek L, Basso G, Kronnie GT, Bene MC et al. 7 Raza A, Alvi S, Broady-Robinson L, Showel M, Cartlidge J, Mundle SD et al. Clinical utility of microarray-based gene expression profiling in the diagnosis and Cell cycle kinetic studies in 68 patients with myelodysplastic syndromes subclassification of leukemia: report from the International Microarray Innovations following intravenous iodo- and/or bromodeoxyuridine. Exp Hematol 1997; 25: in Leukemia Study Group. J Clin Oncol 2010; 28:2529–2537. 530–535. 33 Zekavati A, Nasir A, Alcaraz A, Aldrovandi M, Marsh P, Norton JD et al. 8 Parker JE, Mufti GJ. Ineffective haemopoiesis and apoptosis in myelodysplastic Post-transcriptional regulation of BCL2 mRNA by the RNA-binding protein syndromes. Br J Haematol 1998; 101:220–230. ZFP36L1 in malignant B cells. PLoS One 2014; 9: e102625. 9 Parker JE, Mufti GJ, Rasool F, Mijovic A, Devereux S, Pagliuca A. The role of 34 Kutuk O, Letai A. Regulation of Bcl-2 family proteins by posttranslational apoptosis, proliferation, and the Bcl-2-related proteins in the myelodysplastic modifications. Curr Mol Med 2008; 8:102–118. syndromes and acute myeloid leukemia secondary to MDS. Blood 2000; 96: 35 Ruvolo PP, Deng X, May WS. Phosphorylation of Bcl2 and regulation of apoptosis. 3932–3938. Leukemia 2001; 15:515–522. 10 Albitar M, Manshouri T, Shen Y, Liu D, Beran M, Kantarjian HM et al. Myelodys- 36 Vardiman JW. The World Health Organization (WHO) classification of tumors of plastic syndrome is not merely ‘preleukemia’. Blood 2002; 100:791–798. the hematopoietic and lymphoid tissues: an overview with emphasis on the 11 Bogdanovic AD, Trpinac DP, Jankovic GM, Bumbasirevic VZ, Obradovic M, myeloid neoplasms. Chem Biol Interact 2010; 184:16–20. Colovic MD. Incidence and role of apoptosis in myelodysplastic syndrome: 37 Malcovati L, Della Porta MG, Strupp C, Ambaglio I, Kuendgen A, Nachtkamp K morphological and ultrastructural assessment. Leukemia 1997; 11:656–659. et al. Impact of the degree of anemia on the outcome of patients with myelo- 12 Greenberg P, Cox C, LeBeau MM, Fenaux P, Morel P, Sanz G et al. International dysplastic syndrome and its integration into the WHO classification-based Prog- scoring system for evaluating prognosis in myelodysplastic syndromes. Blood nostic Scoring System (WPSS). Haematologica 2011; 96: 1433–1440. 1997; 89: 2079–2088. 38 Woll PS, Kjallquist U, Chowdhury O, Doolittle H, Wedge DC, Thongjuea S et al. 13 Invernizzi R, Pecci A, Bellotti L, Ascari E. Expression of p53, bcl-2 and ras Myelodysplastic syndromes are propagated by rare and distinct human cancer oncoproteins and apoptosis levels in acute leukaemias and myelodysplastic stem cells in vivo. Cancer Cell 2014; 25: 794–808. syndromes. Leuk Lymphoma 2001; 42:481–489. 39 Pang WW, Pluvinage JV, Price EA, Sridhar K, Arber DA, Greenberg PL et al. 14 Strasser A, Cory S, Adams JM. Deciphering the rules of programmed cell death to Hematopoietic stem cell and progenitor cell mechanisms in myelodysplastic improve therapy of cancer and other diseases. EMBO J 2011; 30: 3667–3683. syndromes. Proc Natl Acad Sci USA 2013; 110:3011–3016. 15 Adams JM, Cory S. The Bcl-2 apoptotic switch in cancer development and 40 Konopleva M, Contractor R, Tsao T, Samudio I, Ruvolo PP, Kitada S et al. therapy. Oncogene 2007; 26: 1324–1337. Mechanisms of apoptosis sensitivity and resistance to the BH3 mimetic ABT-737 16 Chipuk JE, Moldoveanu T, Llambi F, Parsons MJ, Green DR. The BCL-2 family in acute myeloid leukemia. Cancer Cell 2006; 10:375–388. reunion. Mol Cell 2010; 37:299–310. 41 Ni Chonghaile T, Sarosiek KA, Vo TT, Ryan JA, Tammareddi A, Moore Vdel G et al. 17 Pop C, Salvesen GS. Human caspases: activation, specificity, and regulation. J Biol Pretreatment mitochondrial priming correlates with clinical response to cytotoxic Chem 2009; 284: 21777–21781. chemotherapy. Science 2011; 334: 1129–1133.

Leukemia (2016) 112 – 123 © 2016 Macmillan Publishers Limited Blockade of BCL-2 induces apoptosis in high-risk MDS patients S Jilg et al 123 42 Vo TT, Ryan J, Carrasco R, Neuberg D, Rossi DJ, Stone RM et al. Relative mito- potent supporters of hematopoietic stem cell activity. Blood 2002; 99: chondrial priming of myeloblasts and normal HSCs determines chemotherapeutic 1183–1189. success in AML. Cell 2012; 151: 344–355. 53 Parmar A, Marz S, Rushton S, Holzwarth C, Lind K, Kayser S et al. Stromal niche 43 Jaiswal S, Ebert BL. MDS is a stem cell disorder after all. Cancer Cell 2014; 25: cells protect early leukemic FLT3-ITD+ progenitor cells against ﬁrst-generation 713–714. FLT3 tyrosine kinase inhibitors. Cancer Res 2011; 71:4696–4706. 44 Zhang J, Niu C, Ye L, Huang H, He X, Tong WG et al. Identiﬁcation of the 54 Omidvar N, Kogan S, Beurlet S, le Pogam C, Janin A, West R et al. BCL-2 and haematopoietic stem cell niche and control of the niche size. Nature 2003; 425: mutant NRAS interact physically and functionally in a mouse model of progressive 836–841. myelodysplasia. Cancer Res 2007; 67: 11657–11667. 45 Calvi LM, Adams GB, Weibrecht KW, Weber JM, Olson DP, Knight MC et al. Osteo- 55 Beurlet S, Omidvar N, Gorombei P, Krief P, Le Pogam C, Setterblad N et al. BCL-2 blastic cells regulate the haematopoietic stem cell niche. Nature 2003; 425:841–846. inhibition with ABT-737 prolongs survival in an NRAS/BCL-2 mouse model of AML 46 Raaijmakers MH, Scadden DT. Evolving concepts on the microenvironmental by targeting primitive LSK and progenitor cells. Blood 2013; 122: 2864–2876. niche for hematopoietic stem cells. Curr Opin Hematol 2008; 15: 301–306. 56 Choudhary GS, Al-Harbi S, Mazumder S, Hill BT, Smith MR, Bodo J et al. MCL-1 and 47 Wilson A, Trumpp A. Bone-marrow haematopoietic-stem-cell niches. Nat Rev BCL-xL-dependent resistance to the BCL-2 inhibitor ABT-199 can be overcome by Immunol 2006; 6:93–106. preventing PI3K/AKT/mTOR activation in lymphoid malignancies. Cell Death Dis 48 Scadden DT. The stem-cell niche as an entity of action. Nature 2006; 441:1075–1079. 2015; 6:e1593. 49 Li X, Deeg HJ. Murine xenogeneic models of myelodysplastic syndrome: an 57 Mak DH, Wang RY, Schober WD, Konopleva M, Cortes J, Kantarjian H et al. essential role for stroma cells. Exp Hematol 2014; 42:4–10. Activation of apoptosis signaling eliminates CD34+ progenitor cells in blast crisis 50 Raaijmakers MH, Mukherjee S, Guo S, Zhang S, Kobayashi T, Schoonmaker JA et al. CML independent of response to tyrosine kinase inhibitors. Leukemia 2012; 26: Bone progenitor dysfunction induces myelodysplasia and secondary leukaemia. 788–794. Nature 2010; 464: 852–857. 58 Konopleva M, Milella M, Ruvolo P, Watts JC, Ricciardi MR, Korchin B et al. MEK 51 Kusadasi N, Oostendorp RA, Koevoet WJ, Dzierzak EA, Ploemacher RE. inhibition enhances ABT-737-induced leukemia cell apoptosis via prevention of Stromal cells from murine embryonic aorta-gonad-mesonephros region, ERK-activated MCL-1 induction and modulation of MCL-1/BIM complex. Leukemia liver and gut mesentery expand human umbilical cord blood-derived 2012; 26:778–787. CAFC(week6) in extended long-term cultures. Leukemia 2002; 16: 59 Muller-Thomas C, Rudelius M, Rondak IC, Haferlach T, Schanz J, Huberle C et al. 1782–1790. Response to azacitidine is independent of p53 expression in higher-risk myelo- 52 Oostendorp RA, Harvey KN, Kusadasi N, de Bruijn MF, Saris C, Ploemacher RE et al. dysplastic syndromes and secondary acute myeloid leukemia. Haematologica Stromal cell lines from mouse aorta-gonads-mesonephros subregions are 2014; 99: e179–e181.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu)