Bcl-2 Is a Therapeutic Target for Hypodiploid B-Lineage Acute Lymphoblastic Leukemia

B-cell acute lymphoblastic leukemia (B-ALL) is the most common cancer in children and remains one of the top causes of cancer-related mortality for children 1–15 years old (1). Important prognostic factors for ALL include age, white blood cell count at diagnosis, somatic genetic abnormalities, and early response to treatment.

Among the highest risk subtypes for relapse is hypodiploid B-ALL, which is characterized by a chromosomal content <44 chromosomes. This subtype is classified on the basis of the modal chromosome number: near diploid (44–45 Chr.), high hypodiploid (40–43 Chr.), low hypodiploid (32–39 Chr.), and near haploid (24–31 Chr.; refs. 2, 3). While near diploid (ND) and high hyperdiploid (HH) are not associated with poor prognosis, patients with low hypodiploid (LH) and near haploid (NH) B-ALL presenting with detectable minimal residual disease (MRD) levels after treatment are associated with event-free survival (EFS) of <45% (2–4). A common occurrence in hypodiploid B-ALL is the doubling of the chromosome content (e.g., a 27 Chr. NH B-ALL can double its content to 54 Chr.), which may mask the near haploid or low hypodiploid nature of the initiating clone. It is important to distinguish patients who have masked NH or masked LH from patients who have hyperdiploid leukemia, as the latter has a significantly more favorable prognosis with contemporary therapy (2, 5). Despite receiving high intensity chemotherapy and hematopoietic stem cell transplant (HSCT), patients with hypodiploid B-ALL frequently experience relapse (2, 6, 7).

Our group published a comprehensive genomic analysis of leukemic blasts from 124 patients with hypodiploid B-ALL, which identified distinct patterns of recurrent mutations and gene expression profiles between LH and NH subtypes. LH ALL was characterized by alteration of TP53 in virtually all cases (>91% of LH B-ALL patients; refs. 3, 8, 9), with up to 40% of these mutations observed in nontumor cells (3), and recurrent somatic deletions in IKZF2 (53%). NH tumors were characterized by Ras/MAPK pathway activation via NRAS mutations (15%) or homozygous deletions in NF1 (44%) or IKZF3 (13%). In addition, the CDKN2A/B loci were deleted in approximately 25% of LH and NH cases.

While inhibition of MEK did not affect proliferation, PI3K inhibition did reduce proliferation in a hypodiploid cell line in vitro (3). It remains unknown whether these or any other molecular alterations can be exploited therapeutically in vivo for hypodiploid B-ALL. To shed light on this question, we first performed a biochemical characterization by assessing total and active levels of signaling proteins from selected pathways, to explore how the underlying mutations, in conjunction with the chromosomal losses, affect the biochemical blueprint of these leukemic cells. We thus identified the Bcl-2 pathway as a promising therapeutic avenue and pursued Bcl-2 inhibition in preclinical trials.

The following antibodies were used: ATM (CST, 2873S), Actin-HRP (CST, 5125), BAD (Abcam, ab32445), BAX (Bio-Rad, MCA2738), Bcl-2 (Abcam, ab32124), Bcl-xL (Abcam, 32370), Bcl-w (LSBio, LS-C382259), BIM (Bio-Rad, AHP933), BMF (Abcam, ab181148), BIK (CST 4592S), BAK (Abcam, ab32371), BID (CST, 2002T), cleaved caspase-3, (Cell Signaling Technology, 9664L) c-CBL (Cell Signaling Technology, 2747), pCDK2 (Abcam, ab76146), pERK (T202/Y204; Cell Signaling Technology, 9101), cleaved PARP (Abcam, ab32064), PTEN (Abcam, ab32199), PUMA (LifeSpan Biosciences, LS-C98860-400), PARP (Abcam, ab32138), HRK (Bio-Rad, AHP1178T), MCL-1 (Abcam, ab32087), pRaf1 (S338; Cell Signaling Technology, 9424), pS6 (S235/236; Cell Signaling Technology, 2211), pSTAT5 (Y694; BD Biosciences, 611964), pmTOR (S2448; Cell Signaling Technology, 2976), pTyr (Cell Signaling Technology, 9411), p14 (Cell Signaling Technology, 2407S), p16 (BioLegend, 675602), p21 (Cell Signaling Technology 2947S), p27 (Cell Signaling Technology, 3688T), p53 (Bio-Rad, MCA1701), p63 (ProteinTech, 12143-1AP), and p73 (Abcam, ab197040).

The following surface markers were used: CD19-PECy7 (Tonbo, 60-0199-T100), hCD45-APC-Fire750, (BioLegend, 368518), and mCD45-FITC (Tonbo, VWR 10050-904).

The following inhibitors were used: ABT-199 was kindly provided by Abbvie, ABT-737 (Selleck Chemicals, S1002), PP2 (Calbiochem, 529576), INCB-18424 (JAK-I) Calbiochem (420099), GDC0941(Calbiochem, 509226), PP242 (Calbiochem, 475988), PD0325901 (Calbiochem, 444968), PI-90, PI-103, p110α, p110β, p110δ, (were kindly provided by prof. K. Shokat, University of California San Francisco, Los Angeles, CA), AKT VIII (Calbiochem, 124018), dexamethasone (Calbiochem, 265005), etoposide (Calbiochem, 341205), and cisplatin (Calbiochem, 232120),

The following reagents were used: d-Luciferin (potassium salt; Gold Biotechnology, LUCK-1G). CellTiter-Glo Luminescent Cell Viability Assay (Promega, G7571), Caspase-Glo 3/7 (Promega, G8091), and Thymidine (Sigma Aldrich, T1895).

We have validated all our cell lines through the UC Berkeley Human Cell Line Authentication service (https://ucberkeleydnasequencing.com). We also test all our cell lines in culture every 3 months or when a new lot is put in culture using the LookOut Mycoplasma PCR Detection Kit (Sigma-Aldrich #MP0035).

Cells were collected and centrifuged at 1,300 rpm for 5 minutes. Supernatant was removed and pellet was snap frozen and stored at −20°C overnight or lysed immediately. Pellets were lysed in NP-40 lysis buffer (with NaF, Aprotinin, Leupetin, and phenylmethylsulfonylfluoride), vortexed, and kept for 30 minutes at 4°C in a rocker. Protein quantification was done using a BCA Protein Assay kit (#23225 Thermo Fisher Scientific). For Western blotting, 25 μg of whole-cell lysate was loaded per lane in Criterion TGX gels 4%–20% (Bio-Rad). For protein ladder, 3 μL of Precision Plus Protein Dual Color (Bio-Rad 161-0374) was loaded. Gels were run in TGS buffer (#161-0772 Bio-Rad) for 30 minutes at 200 V. Transfer was done using Immobilon-P PVDF membrane (#IPVH00010 Millipore) in TG buffer (#161-0771 Bio-Rad) supplemented with FLASHblot (#R-03090-D50 Advansta) for a faster transfer (25 minutes) at 320 mA. Polyvinylidene difluoride membranes were then blocked overnight in Advanblock-Chemi (R-03726-E10).

Proteins levels were detected using unconjugated primary antibodies, wash buffer, and secondary-horseradish peroxidase (HRP) antibodies using the iBind Flex system (#SLF2000 Thermo Fisher Scientific) for 2 hours 30 minutes. HRP signal was developed using Westernbright ECL (#K-12045-D20 Advansta), Westernbright Quantum (#K-12042-D20 Advansta), or Radiance Plus (#AC2103 Azure Biosystems) for high, medium, or low expressed proteins, respectively. Proteins were visualized using a Bio-Rad Chemidoc Touch Imager.

Cells were collected and centrifuged at 1,300 rpm for 5 minutes. Supernatant was removed leaving 1 mL of volume and pellet was resuspended. For fixation and permeabilization, cells were fixed in 2% paraformaldehyde (PFA) by adding 143 μL of PFA 16% (#15710 Electron Microscopy Sciences) to 1 mL of cells and left at room temperature for 10 minutes. Cells were washed in PBS and resuspended in 500 μL of 85% MeOH for permeabilization and left at −20°C until the cells were ready to use. Cells were then rehydrated by washing 2× in PBS and then resuspending in 500 μL of FACS buffer with Fc block (1:200) for 1 hour before incubating with antibodies at optimal concentrations for 1 hour. Cells were washed and incubated with secondary antibody (40 minutes). Cells were then washed and resuspended in 100 μL of PBS. Cells were run on a FACS Verse (BD Biosciences) and analyzed using FlowJo (Treestar. Inc). For heatmaps, files were analyzed using Cytobank.

One-hundred microliters of cells containing 20,000 cells (2 × 10⁵ cells/mL) were seeded in a 96-well plate. Cells were incubated in triplicates with drugs at increasing concentrations for 24 hours at 37°C in 5% CO₂. After 24 hours, the cells were subjected to CellTiter Glo (proliferation) or Caspase Glo (apoptosis). Luminescence was read using a Tecan Infinite M200pro.

NALM-16 cells were cultured in the presence of compounds at 1 μmol/L or DMSO vehicle in a 384-well plate (3,150 cells/well) for 72 hours. Cell number was quantified using the CellTiter-Glo assay. Triplicate wells were averaged and normalized to the mean value from DMSO-treated wells. Error bars indicate the 95% confidence interval for the mean.

Tumor cells from patient primagrafts were sequenced by whole-exome and/or whole-genome sequencing for identification of mutations and by SNP6 microarray (Affymetrix) to evaluate copy number alterations.

Cell lines were derived from NH B-ALL patient samples serially passaged through NSG mice and adapted to tissue culture using culturing techniques intended to mimic the tumor microenvironment in vitro. Cell lines were authenticated via short tandem repeat (STR) fingerprinting, karyotyping, and Sanger sequencing of known point mutations identified in the parental sample.

Five small guide RNAs (sgRNA) of varying lengths (17–20 bp) were designed against BCL2L11 (BIM) or BAD. Two additional 17-bp and 20-bp random sequence sgRNAs were designed to serve as nontargeting controls. All sgRNAs (except NTCs) were designed to target genomic regions before the BH3 domain. Guide design was assisted by the BROAD sgRNA design tool, Zi-Fit Targeter, and the MIT CRISPR design tool.

sgRNAs were introduced via Golden Gate cloning into lentiCRISPRv2 (Addgene plasmid # 52961), and modified to express either eGFP or mCherry in place of puromycin. LentiCRISPRv2, psPAX2 packaging plasmid, and pCMV-VSV-G envelope plasmid were transfected into HEK-293T cell lines using Mirus LT-1 transfection reagent. Viral supernatant was collected after 36 hours and centrifuged at 20,000 × g for 90 minutes at 4°C to pellet the virus. The viral pellet was resuspended in RPMI media with 10% FBS and spinoculation was performed at 1,000 × g for 60 minutes at room temperature at an MOI of 1 using retronectin-coated plates and/or polybrene at a concentration of 4–8 μg/mL.

All in vivo experiments were conducted following Institutional Animal Care and Use Committee (IACUC) guidelines as described in our IACUC-approved protocol (AN163079-02F, “Biologic and Pre-Clinical Studies in Mouse Models of Leukemia”) and approved by IACUC review board. NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ mice were transplanted with 1 × 10⁶ live tumor cells by tail vein injection. Tumor cells were previously transduced with lentiviral vCL20SF2-Luc2aYFP vector for stable expression of luciferase for live imaging.

To assess patient-derived leukemia engraftment, we collected 10 μL of tail vein blood from patient-derived xenograft (PDX) mice into EDTA tubes containing 200 μL of PBS, spun down for 5 minutes at 1,300 rpm, and red cells were eliminated using ACK buffer for 5 minutes. Cells were washed in PBS, spun down for 5 minutes at 1,300 rpm, and resuspended in 50 μL of antibody staining mix [mCD45-FITC (1:300), hCD45-APC-Fire (1:300), and CD19-PECy7 (1:500) for 30 minutes]. Cells were washed with PBS, resuspended in 200 μL of PBS and analyzed using a BD FACSVerse cytometer.

We prepared a 200× luciferin stock solution (30 mg/mL) in sterile water. We injected each mouse with 100 μL (3 mg) luciferin intraperitoneally 10 minutes before imaging. Dorsal and ventral images were taken using a Xenogen IVIS Spectrum Imager and images analyzed with Living Image Software.

Xenotransplanted human hypodiploid or control leukemia cells were obtained from the spleens of NSG mice treated with ABT-199 or DMSO control. Viability was assessed by staining with cisplatin as described in Fienberg and colleagues (10). Cells were fixed with PFA (Electron Microscopy Sciences) to a final concentration of 1.6% for 10 minutes at room temperature. Cells were barcoded as described in ref. 11, then pelleted and washed once with cell staining media (CSM; PBS with 0.5% BSA, 0.02% sodium azide) to remove residual PFA. Blocking was performed with purified human Fc Receptor Binding Inhibitor (eBioscience) following manufacturer's instructions. Surface marker antibodies were added and samples were incubated at room temperature for 30 minutes. Cells were pelleted and washed with CSM before permeabilization with 4°C methanol for 10 minutes at 4°C and then washed with CSM and stained with intracellular marker and phospho-specific antibodies in 50 μL for 30 minutes at room temperature. Cells were washed once in CSM, then stained with 1:5,000 ¹⁹¹Ir/¹⁹³Ir DNA intercalator (Fluidigm) in PBS with 1.6% PFA for 20 minutes at room temperature. Cells were washed once with CSM and then twice with water prior to analysis in presence of normalization beads by CyTOF (Fluidigm).

Data were normalized (12), and files were debarcoded as described previously (11). Single cells were gated using Cytobank software (www.cytobank.org) based on event length and ¹⁹¹Ir/¹⁹³Ir DNA content (to avoid debris and doublets) as described in ref. 13. Single-cell protein expression data were transformed using the inverse hyperbolic sine (arsinh) function with a cofactor of five. Expression of proteins in each population of interest was determined by calculating the median level of expression after arsinh transformation. Data were visualized using ViSNE (14).

Beck-1732 cells were plated at 1.5 mol/L/mL. Thymidine was added to a final concentration of 2 mmol/L. Cells were incubated for 18 hours in a tissue culture incubator. After 18 hours, thymidine was removed by washing cells with 1× PBS and then resuspending in fresh culture media. Cells were incubated for 9 hours in regular media (RPMI 20% FBS). After 9 hours, thymidine was added to final concentration of 2 mmol/L and cells were incubated for 15 hours. Cells were then washed with 1× PBS and then resuspended in fresh media. Cells were allowed to rest for 1–4 hours before collection.

Leukemic cells often harbor mutations in genes related to signaling pathways critical for proliferation and survival. Additional pathways that lack genomic alterations may also be activated as a result of these mutations and may contribute to leukemia development and maintenance. We reasoned that studying the biochemical circuitry may reveal how genomic alterations and aneuploidy in hypodiploid leukemia cells contribute to the dysregulation of signal transduction and allow for identification of protein targets with an essential role in leukemic cell survival.

To assess dysregulated signaling pathways across hypodiploid B-ALL samples, we analyzed primary and xenografted LH and NH patient samples (listed in Supplementary Table S1) by subjecting them to a signaling profiling panel (Supplementary Fig. S1A) measuring over 20 proteins from pathways commonly dysregulated in leukemia. As shown in Fig. 1A, we analyzed phosphorylated (active) levels of receptor tyrosine kinase (RTK)-associated proteins (pTyr, pSTAT5, pSYK, and pBTK), constituents of the RAS/MAPK (p-c-RAF, pERK), and PI3K (mTOR, p4EBP1, pS6) pathways. In addition, we measured total levels of proteins relevant for survival (Bcl-2, Bcl-xL, and MCL-1), and apoptosis [BAD, BAX, BIM, and cleaved poly (ADP) ribose polymerase (PARP)]. We also measured the levels of other proteins commonly mutated or dysregulated in leukemia (PARP, PTEN, p53, and CBL-1). This characterization was done at baseline in unstimulated conditions and protein levels were normalized to levels in mononuclear cells obtained from healthy controls.

This signaling panel confirmed hyperactivation of the MAPK and PI3K pathway previously observed in both the LH and NH patient samples irrespective of RAS mutation status (3), but showed normal to decreased levels of RTK-associated proteins (Fig. 1A). In addition, we observed high levels of the prosurvival protein Bcl-2 across all leukemias (Fig. 1A; Supplementary Fig. S1B), intermediate levels of MCL-1, and low levels of Bcl-xL (Fig. 1A). The observed protein levels correlated with gene expression levels observed in the GEO2R database comparing hypodiploid patient samples (LH, mLH, NH, and mNH) to nonhypodiploid patient samples from a total of 118 patients analyzed (Supplementary Fig. S1C–S1E; https://www.ncbi.nlm.nih.gov/geo/geo2r/?acc=GSE27237). A recent study from Dr. Mullighan's laboratory provides additional gene expression data from multiple leukemia subtypes including hypodiploid B-ALL (15). Interestingly, both transcript and protein expression of the proapoptotic BIM were greatly elevated in LH cells relative to NH cells (Fig. 1A; Supplementary Fig. S1F). Expression of the proapoptotic BAX and BAD proteins was also elevated across all samples. High Ki67 levels confirmed the proliferative capacity of the cells, whereas cleaved PARP levels indicated low level of apoptosis in baseline conditions. The panel also showed that p53 was highly expressed in some LH samples (Fig. 1A; Supplementary Fig. S1B) without an associated increase in transcript expression (Supplementary Fig. S1G), consistent with the reported presence of NLS domain mutations in TP53 that are known to stabilize the mutant conformation of the protein (3).

After identifying dysregulated pathways in patient-derived hypodiploid samples, we assessed the relevance of each individual pathway to cell proliferation and apoptosis. We subjected the human NH cell line NALM-16 to increasing concentrations of several drugs targeting such dysregulated pathways: PP2 (Src family inhibitor), INCB18424 (JAK/STAT inhibitor), PD0235901 (MEK inhibitor), GDC-0941 (PI3K inhibitor), PP-242 (mTOR inhibitor) and ABT-263 (Bcl-2/Bcl-xL inhibitor). We measured proliferation and apoptosis at 24 (Fig. 1B and C), 48, and 72 hours. We found that only the Bcl-2/Bcl-xL inhibitor ABT-263 profoundly reduced proliferation of NALM-16 at nanomolar concentrations (Fig. 1B). Similarly, only ABT-263 induced high levels of apoptosis at nanomolar concentrations, as measured by cleaved caspase-3/7 (Fig. 1C).

We then expanded the number of inhibitors for each pathway and validated our results using cell proliferation assays (Supplementary Fig. S2A). We also compared the induction of apoptosis by ABT-263 with apoptosis-inducing chemotherapy agents in NALM-16 cells. ABT-263 induced higher levels of apoptosis, as measured by cleaved caspase-3 at 24 hours, relative to any of the chemotherapy agents tested (Supplementary Fig. S2B). To maximize the possibility of identifying effective drugs to treat hypodiploid B-ALL, we also subjected NALM-16 cells to an independent drug screen containing 94 compounds, including FDA-approved drugs and commonly used chemotherapy agents. This screen independently confirmed ABT-263 as a highly active drug in this cell line (Supplementary Fig. S2C and S2D).

Given the high levels of Bcl-2 and the low levels of Bcl-xL in all hypodiploid cells analyzed (Fig. 1A), we hypothesized that the effect of ABT-263 is through the inhibition of Bcl-2 activity. To confirm this hypothesis, we treated NALM-16 and MHH-CALL2, a masked NH cell line (16), with ABT-263, ABT-737 (Bcl-2/Bcl-xL inhibitors), and ABT-199 (a Bcl-2–specific inhibitor). All three inhibitors showed overlapping effects to inhibit growth and induce apoptosis (Fig. 2A and B; Supplementary Fig. S2E and S2F), suggesting the effect of ABT compounds observed in hypodiploid ALL arises from the inhibition of Bcl-2. We then tested four primary patient-derived hypodiploid B-ALL samples to increasing concentrations of ABT-199. Results showed over 80% reduction in viable cells at low nanomolar concentrations at 24 hours (ED₅₀ 17.5 nmol/L; Fig. 2C). We expanded these studies to nine xenografted patient samples ex vivo using a single dose of 100 nmol/L over 24 hours, which also showed a profound antiproliferative effect of Bcl-2 inhibition accompanied by induction of apoptosis as shown by increased levels of cleaved PARP (Supplementary Fig. S3A and S3B).

Given these promising in vitro findings, we set out to investigate the efficacy of ABT-199 in vivo. We established PDXs from eight representative hypodiploid B-ALL patient samples [four low hypodiploid (LH1–LH4), and four near haploid (NH1–NH4)] that had undergone genetic characterization (Supplementary Fig. S3C; Supplementary Table S1 and Holmfeldt and colleagues; ref. 3). These leukemia cells were transduced to express luciferase and YFP to enable quantitation of tumor burden and response to therapy by bioluminescence imaging (BLI). With serial passaging in mice, the tumors maintained the overall patterns of aneuploidy observed in the diagnostic samples (Supplementary Fig. S3D). Prior to conducting the preclinical study, we confirmed the feasibility of tracking xenotransplanted hypodiploid leukemias in NSG mice. BLI demonstrated to be a sensitive approach to detect presence of leukemia in internal organs, whereas flow cytometry measurement of hCD45 showed to be a technique mostly suitable to identify the time of appearance of leukemic blasts in peripheral blood (Supplementary Fig. S4A).

We then conducted a preclinical trial to assess the efficacy of ABT-199 in a group of 10 leukemias: four low hypodiploid (LH1–LH4), four near haploid (NH1–NH4), and two nonhypodiploid ALL controls consisting of a Philadelphia chromosome positive (Ph⁺) and a Ph-like (ETV6-ABL1) leukemia ALL (Supplementary Fig. S4B) using a dose-escalating regimen to avoid the occurrence of tumor lysis syndrome (Supplementary Fig. S4C).

To assess the efficacy of ABT-199 in vivo, we performed weekly measurements of PB blast counts using BLI and the expression of human CD45 as measured by flow cytometry. Treatment was initiated when >15% hCD45 presence in PB was detected.

Following ABT-199 treatment, there was a rapid and significant reduction in leukemic blasts in PB to undetectable levels within 24–72 hours (Figs. 3A; Supplementary Fig. S4D). This reduction of leukemic blasts was durable in 80% (14/18) of all hypodiploid B-ALL mice treated with ABT-199 as shown by undetectable leukemic blasts in peripheral blood throughout the 60-day trial. Only 4 of 18 mice, all from LH leukemia, showed residual 1%–5% leukemic blasts in peripheral blood (Fig. 3A). This profound reduction in peripheral blood was mirrored by a concurrent reduction in leukemic burden in internal organs within the first week, as measured by BLI (Fig. 3B); although, in some cases, complete body clearance was achieved within 2–3 weeks (Supplementary Fig. S4E).

This significant response to ABT-199 treatment in vivo was observed in both LH and NH hypodiploid B-ALL, with some variation between individual leukemia samples. In particular, LH1, a leukemia with a particularly aggressive phenotype (Fig. 3C; Supplementary Fig. S4E), showed a reduction in leukemic burden in PB within 2–6 days (Fig. 3A; Supplementary Fig. S4D) and increased lifespan to 41–48 days (Fig. 3C). LH2 and LH4 showed a rapid and prolonged reduction in leukemic burden in peripheral blood and survived the 60-day trial period (Fig. 3A and C). A more striking response to ABT-199 was seen in all mice xenografted with NH leukemia. All four NH leukemias showed a sustained reduction in circulating peripheral blasts, with all 10 treated mice showing undetectable peripheral blood blasts throughout the 60-day period of the trial (Fig. 3A and D). Remarkably, 85% of all mice engrafted with either LH or NH leukemia treated with ABT-199 survived the 60-day trial (Fig. 3C and E). Importantly, control leukemias (Ph⁺ and Ph-like) did not show any reduction in leukemic growth or clearance between vehicle-treated and ABT-199 treated mice and died at a median of 19 days irrespective of the treatment. (Fig. 3A, C, and D). Overall, daily administration of ABT-199 significantly extended the lifespan of all mice xenografted with hypodiploid leukemias (Fig. 3C and E).

To further assess the efficacy of ABT-199, we measured the number of blasts in bone marrow and spleen by flow cytometry at the end of trial. Consistent with the peripheral blood blast levels measured throughout the trial, leukemic blasts were, in most cases, undetectable at the end of trial in peripheral blood (Fig. 3D). Consistent with BLI data, however, some leukemic blasts were detected in the spleen of LH and NH B-ALL, as measured by flow cytometry (Supplementary Fig. S4F).

We extracted residual hypodiploid B-ALL cells from spleen and treated them with ABT-199 ex vivo. We then compared IC₅₀ values with those obtained at time of xenotransplant. Leukemia cells harvested from the spleens of the majority (12 of 16) treated mice remained sensitive to ABT-199, as indicated by similar IC₅₀ values between these cells and treatment-naïve cells (between 5 and 75 nmol/L, n = 6; Supplementary Table S2). Only three LH samples and one NH leukemia showed increased IC₅₀ levels >100 nmol/L, suggesting an acquired resistance to ABT-199.

Although the number of resistant leukemias was too small to study potential mechanisms of resistance, we undertook a series of analyses as shown below to characterize persistent clones. Using single nucleotide polymorphism arrays, we compared residual leukemia cells with initially xenografted samples. Data showed a high degree of clonal conservation in all leukemias analyzed (Supplementary Fig. S5). We also used mass cytometry (CyTOF) to determine whether leukemic cells had undergone a change in their differentiation status as a result of treatment. All persistent leukemic blasts isolated from the spleen of these xenografted mice expressed CD45, CD179a, CD179b, CD43, and CD58 at the end of trial (Supplementary Table S3), indicating a common pro-B-cell stage of differentiation. These data together suggest that ABT-199 treatment did not induce detectable genomic alterations or a drift in cellular immunophenotype.

In an effort to identify potential changes in signaling as well as to better understand the cellular mechanisms modulating sensitivity to ABT-199, we searched for alterations in protein levels of the BCL2 (survival) pathway and apoptosis regulators that could represent biomarkers of response using hypodiploid cell lines as well as the xenografted samples collected at the end of the in vivo trial.

For this purpose, we first determined whether sensitivity to ABT-199 was B-lineage dependent. We exposed 18 cell lines from different hematopoietic lineages to ABT-199 and asked whether expression of BCL2 family proteins correlated with ABT-199 sensitivity. Cells were treated with 100 nmol/L of ABT-199 for 24 hours, proliferation was measured (Fig. 4A), and cells were collected for biochemical interrogation of the BCL2 family (Fig. 4B).

In this 18-cell line comparison, we assessed whether Bcl-2 levels alone correlated with ABT-199 sensitivity. Maximal sensitivity was associated with the highest levels of Bcl-2 regardless of leukemia subtype. Analysis of protein levels of the prosurvival and proapoptotic BCL2 family members revealed that low levels of Bcl-xL and high levels of BAX showed the strongest correlation with sensitivity to ABT-199, whereas low levels of Bcl-2, together with high levels of Bcl-xL or MCL-1, correlated with poor response to the drug.

To further confirm the ABT-199 response signature, we reduced the number of cell lines and extended the number of proteins. Thus, CCRF-CEM (T-cell line) insensitive to ABT-199 was used as a negative control and HL-60 (AML line; moderately sensitive to ABT-199) was used as positive control and compared with hypodiploid cell lines (NALM-16 and MHH-CALL2). Hypodiploid cell lines and HL-60 (all sensitive to ABT-199) showed high Bcl-2 and low/intermediate Bcl-xL and MCL-1 and high levels of the proapoptotic proteins, BAX, BIM, and BAD (Fig. 4C), whereas CCRF-CEM displayed the opposite profile, with low Bcl-2, high Bcl-xL, and high MCL-1 expression.

We then extended this analysis to the xenografted samples collected at the end of the trial. When looking at the prosurvival proteins of the BCL2 family, LH and NH B-ALL xenografts showed a sustained and homogeneous pattern of elevated to very elevated levels of Bcl-2 compared with the control leukemias (insensitive to ABT-199; Fig. 5A). Although levels of Bcl-xL or MCL-1 expression did not have a particular correlation with hypodiploidy or ABT sensitivity, in LH2-5 and NH3-6, the leukemias in which two mice showed persisting blasts and the highest IC₅₀ at the end of trial (Supplementary Table S2S), MCL-1 levels were significantly elevated. Because MCL-1 has been largely reported to mediate resistance to ABT-199, we knocked down MCL1 in NALM-16 cells, which show intermediate levels of MCL-1. Our data show a 10-fold increase in sensitivity to ABT-199 of NALM-16 cells transduced with shRNA targeting MCL-1 (10 nmol/L IC₅₀) versus NALM-16 transduced with a scrambled shRNA (120 nmol/L IC₅₀; Supplementary Fig. S6A and S6B).

Interestingly, when looking at the proapoptotic proteins of the BCL2 family, high levels of BIM were observed in LH leukemias (LH1–2) and one NH leukemia, whereas the three other NH leukemias expressed only trace levels of BIM (NH2–4). These data are consistent with the gene expression data (Supplementary Fig. S1F) indicating significantly elevated mRNA levels of BIM in the majority of LH leukemias and only a minority of NH leukemias. In those leukemias with low or no BIM expression, protein expression of the proapoptotic sensitizers BAD and PUMA were elevated (Fig. 5). Conversely, control leukemias, which failed to respond to ABT-199, expressed low levels of these proapoptotic mediators and variable expression of BAD. These data are consistent with the hypothesis that high levels of Bcl-2 coupled with high levels of proapoptotic mediators sensitize cells to BH3 mimetics. We then set out to identify the apoptotic regulator(s) responsible for mediating ABT-199–induced apoptosis in hypodiploid cells.

BIM is a major regulator of the intrinsic apoptotic pathway due to its ability to directly activate BAX and BAK to induce cytochrome c release, resulting in apoptosis (17, 18). Bcl-2, through its BH3 domain, can bind to BIM and prevent the induction of apoptosis (19). Accordingly, ABT-199 has been reported to act as a BH3 mimetic, disrupting the interaction between Bcl-2 and BIM (17, 20, 21). On the basis of these data, we expected BIM to mediate ABT-199 activity in hypodiploid cells with high BIM levels. To test this hypothesis, we used CRISPR/Cas9 genome editing to delete BCL2L11 (BIM) in NALM-16 cells (Supplementary Fig. S6E–S6G). Deletion of all three BIM isoforms effectively prevented apoptosis in NALM-16 cells upon exposure to ABT-199 (Supplementary Fig. S6G).

Surprisingly, despite the results above we observed that some hypodiploid cells lacking endogenous BIM expression (Fig. 5A) are highly sensitive to ABT-199. This led us to investigate an alternative mechanism for ABT-199–induced apoptosis. First, we evaluated whether ABT-199 can induce apoptosis with similar efficacy in cells with or without BIM. For this purpose, we compared cleaved caspase-3 levels in NALM-16 (BIM positive) and Beck-1732, a cell line lacking endogenous BIM expression. Beck-1732 is a patient-derived masked NH cell line (Supplementary Fig. S7A and S7B) that lacks BIM expression at the protein level (Fig. 5B; Supplementary Fig. S7C). ABT-199 induced apoptosis in both cell lines as shown by induction of cleaved caspase-3 and cleaved PARP at increasing concentrations of drug (Supplementary Fig. S5B). Surprisingly, Beck-1732 had a higher sensitivity to ABT-199 than NALM-16 with an IC₅₀ of 5–10 nmol/L (Fig. 5B and C). We thus used Beck-1732 cells to identify the proteins from the BCL2 survival and apoptosis pathway that mediate ABT-199–induced apoptosis in the absence of BIM expression.

Given that cell fate between survival and apoptosis is tightly regulated by the balance between prosurvival and proapoptotic BCL2 family members, we analyzed the protein levels of several members of this pathway in all three hypodiploid B-ALL cell lines currently available. We compared these levels with those of nonhypodiploid B-ALL cells (KOPN8 and Reh), the T-ALL cell line CCRF-CEM, and the AML cell line U937. The latter two cell lines are resistant to ABT-199 (Supplementary Fig. S7C). Of the prosurvival and proapoptotic proteins analyzed, Bcl-2, Bcl-w, and MCL-1 were the most significantly elevated in the three hypodiploid cell lines relative to the nonhypodiploid cells, while Bcl-xl levels were lower.

To identify the essential mediators of ABT-199 response in Beck-1732, we evaluated changes in the level of survival and apoptosis proteins from as early as 1 hour up to 24 hours of exposure to ABT-199. Apoptosis was observed as early as 1 hour only at 10 μmol/L [as measured by cleaved caspase-3 and cleaved PARP, and starting at 4 hours at a much lower concentration (100 nmol/L) with a concomitant reduction in c-Myc (Fig. 6A)].

An intriguing observation was that at concentrations below 100 nmol/L of ABT-199 over a course of 24 hours, induction of p21 was observed without induction of cleaved caspase-3, suggesting a cytostatic effect (cell arrest; Fig. 6A). At concentrations higher than 100 nmol/L, however, p21 levels decreased as an increase in levels of cleaved caspase-3 and cleaved PARP were observed, in line with a cytotoxic effect (apoptosis induction) of ABT-199. To further confirm this cytostatic to cytotoxic transition and to identify the mechanisms mediating both processes, we pursued an extended dose experiment ranging from 0.1 nmol/L to 10,000 nmol/L at 2 and 6 hours. Data in Fig. 6B confirmed the induction of p21 at concentrations below 100 nmol/L as well as an increase in p27, p53, and ATM expression, all proteins associated with the genotoxic stress pathway and cell arrest (Supplementary Fig. S7D; ref. 22), in the absence of cleaved caspase-3.

At concentrations above 100 nmol/L, our data showed a sharp reduction in p21 levels concurrent with an increase in cleaved caspase-3 and cleaved PARP (Fig. 6C). This increase in active caspase correlated with an increase in p14 and p16 (proteins associated with G₁–S cell-cycle checkpoint), p63 and p73 (proteins of the p53 alternate pathway), and BAD, BMF, PUMA, and BIK (apoptosis sensitizers). Notably, BAD and p14 mirrored the cleaved caspase levels from the 2-hour time point, whereas BMF, p63, and p73 correlated with cleaved caspase-3 induction at later time points (6 hours). Levels of BID, however, the other BH3 only apoptotic activator, were reduced with low drug concentrations at all time points. Taken together, these data demonstrate that ABT-199, in addition to its reported role as a BH3 mimetic, also actively induces upregulation of cell arrest and apoptosis mediators in a dose-dependent manner.

Given that BAD levels were elevated in cells with low/no BIM (Fig. 5A) and were induced upon ABT-199 exposure (Fig. 6C), we investigated whether BAD plays a major role in mediating ABT-199–induced apoptosis in the absence of BIM in hypodiploid leukemias. We used CRISPR/Cas9 editing to delete BAD in Beck-1732 cells. Knockout of BAD greatly reduced ABT-199–induced apoptosis, as determined by a reduction in cleaved caspase-3 (Fig. 7; Supplementary Fig. S7E and S7F).

Our study identified Bcl-2, a protein whose posttranscriptional dysregulation could not be inferred by genomic sequencing analysis, as an effective therapeutic target in hypodiploid B-ALL. The BCL-2 family of proteins is critical for controlling cell survival in mammalian cells, a cellular process identified as one of the six hallmarks of cancer (23). This pathway contains six antiapoptotic proteins (Bcl-2, Bcl-xL, MCL1, Bcl-W, Bfl-1, and Bcl-B; ref. 24) and several proapoptotic proteins that maintain the balance between cell death and survival. Bcl-2, in particular, is dysregulated in multiple human cancers (25–31).

Our results provide evidence that hypodiploid B-ALL displays a unique biochemical signature not observed in healthy cells. This signature includes overexpression of the prosurvival protein Bcl-2, both at the protein and mRNA levels, without any detected mutations or amplifications of the gene, usually accompanied by very high levels of proapoptotic proteins (BIM or BAD in a mutually exclusive manner) and low levels of Bcl-xL, which renders them sensitive to Bcl-2 inhibition.

Given the large chromosomal imbalance in hypodiploid cells, multiple mechanisms that may include haploinsufficiency, loss of particular genes, or alterations in epigenetic regulators may be responsible for the transcriptional or translational regulation of this survival protein. Similarly, expression of any of the 12 miRNAs reported to inhibit transcriptional activation of Bcl-2 (19) may be reduced by the chromosomal losses characteristic of hypodiploid leukemia (3). At a translational level, many mRNA stabilizers may also increase the half-life of Bcl-2 in these cells (19).

Moreover, BCL-2 is located in chromosome 18, a chromosome that is never lost in LH cases and is retained in most cases of NH leukemia (3), whereas chromosomes 20, 9, and 1 (encoding the other prosurvival proteins Bcl-xL, Bfl-1 and MCL-1, respectively) are largely lost in both LH and NH leukemias. This chromosomal imbalance may be responsible for the strong dependency of hypodiploid B-ALL on Bcl-2, as it may increase the ratio of Bcl-2 to Bcl-xL expression.

Our study identifies for the first time upregulated Bcl-2 as a molecular marker of hypodiploid cells as well as an effective drug target, underscoring the importance of combining protein profiling with genomic data to identify potential novel therapeutics. Accordingly, our in vitro data demonstrate the efficacy of ABT-199 in reducing cell proliferation and inducing cell arrest and apoptosis in hypodiploid cell lines and primary cells.

Our preclinical studies demonstrate the therapeutic benefit of using ABT-199 as a single agent in hypodiploid leukemia. Although ABT-199 was unable to fully eradicate all tumor cells in each xenograft as a single agent, the rapid rate of tumor burden reduction and durable response demonstrates Bcl-2 inhibition is a promising strategy for treatment of hypodiploid B-ALL.

In addition to the in vivo efficacy, our study also sheds light into the cellular mechanisms underlying the response to ABT-199. While evidence from other reports shows that sensitivity to ABT-199 requires not only high levels of Bcl-2 but also low levels of Bcl-xL and/or MCL-1 (32), our study indicates that elevated levels of either the apoptosis activator BIM or the apoptosis sensitizer BAD are also needed. While the role of BIM in mediating ABT-199–induced apoptosis had already been described (33, 34), BAD was thought to primarily assist BIM. Furthermore, the absence of BIM has been generally considered to mediate resistance to BH3 mimetics (35–38). Our work, however, shows that in hypodiploid B-ALL cells lacking baseline BIM expression, BAD can play a similar role as BIM in mediating ABT-199–induced apoptosis.

This study also shows for the first time that ABT-199 can mediate both cytostatic and cytotoxic effect depending on the dose, and we identified the proteins involved in both effects. Thus, at concentrations below 100 nmol/L, the drug induces a cytostatic effect correlating with an increase in ATM, p53, and p21 levels. At concentrations higher than 100 nmol/L, cell-cycle arrest is blunted, correlating with p53 degradation and a concomitant induction of apoptosis, as evidenced by an increase in p63 and p73 as well as an increase in p14, p16, BAD, and active caspase-3, all of which may represent biomarkers for ABT-199 response.

These data indicate that while p53 is relevant in the cytostatic effect of ABT-199, it is not required in mediating cell death. This observation explains the remarkable antileukemic efficacy exerted by ABT-199 against LH B-ALL cells, despite the high incidence of TP53 mutations observed in this B-ALL subtype. The p53 pathway, however, through its other two members p63 and p73 is shown to be involved when the drug reaches cytotoxic concentrations in leukemic cells. According to published pharmacokinetic data (US Venetoclax label, FDA, April 2016), ABT-199 showed a steady-state maximum concentration of 2.1 μg/mL (2.3 μmol/L) at the 400 mg once-daily dose in patients. Those values would be in the cytotoxic range seen in our study.

In conclusion, this study identifies Bcl-2 as a promising therapeutic target in hypodiploid B-ALL despite the lack of genomic alterations in this gene. ABT-199 is a potent and safe Bcl-2 inhibitor currently in clinical trials for the treatment of leukemia. We propose a protein signature comprised of high levels of Bcl-2/BIM or high Bcl-2/BAD and low levels of Bcl-xL and MCL-1 at baseline as biomarkers for predicting and mediating ABT-199 sensitivity. This study provides the intracellular mechanism of ABT-199 in hypodiploid cells lacking BIM, involving ATM-p53-p21 or p14, p73, and BAD providing new biomarkers for ABT-199 response. In addition, these biomarkers may provide additional therapeutic targets for which additional drugs may be or become available.

Finally, combinatorial therapies of additional targeted agents with ABT-199 may provide better disease control for these highly aggressive subtypes of ALL. Our group is conducting follow-up studies to identify promising novel targets and combinatorial approaches.

E. Diaz-Flores reports receiving a commercial research grant from Abbvie. C.G. Mullighan reports receiving a commercial research grant from Loxo Oncology, Pfizer, and Abbvie; has received speakers bureau honoraria from Amgen and Pfizer; and is a consultant/advisory board member for CPRIT and St. Justine Hospital. M.L. Loh is a consultant/advisory board member for MediSix Therapeutics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: E. Diaz-Flores, K. Wu, B.S. Braun, C.G. Mullighan, M.L. Loh

Development of methodology: E. Diaz-Flores, K. Beckman, K.L. Davis, K. Wu, J. Akutagawa, R. Marino, C.G. Mullighan, M.L. Loh

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E. Diaz-Flores, E.Q. Comeaux, K.L. Kim, E. Melnik, K. Beckman, K.L. Davis, K. Wu, J. Akutagawa, O. Bridges, M. Wohlfeil, B.S. Braun, C.G. Mullighan, M.L. Loh

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Diaz-Flores, E.Q. Comeaux, K. Beckman, K.L. Davis, K. Wu, J. Akutagawa, B.S. Braun, C.G. Mullighan, M.L. Loh

Writing, review, and/or revision of the manuscript: E. Diaz-Flores, E.Q. Comeaux, K.L. Davis, K. Wu, B.S. Braun, C.G. Mullighan, M.L. Loh

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Diaz-Flores, K.L. Kim, K. Wu, O. Bridges, B.S. Braun

Study supervision: E. Diaz-Flores, M.L. Loh

The authors would like to express their gratitude to Feng Zhang and Bradley Heller for providing the modified CRISPR/Cas constructs. We also want to thank Su Kim (Abbvie) for providing us ABT-199. Research was supported by the American Cancer Society Individual Research Grant (ACS IRG-14-191-16 to E. Diaz-Flores), the Abbvie preclinical funding program (to E. Diaz-Flores), the Team Connor Foundation (to E. Diaz-Flores and M.L. Loh), and the William Lawrence & Blanche Hughes Foundation (to E. Diaz-Flores and M.L. Loh). The research in the laboratory of E.Q. Comeaux was supported by Henry Schueler 41&9 Foundation, the research of K.L. Davis was supported by the St. Baldrick's Foundation Scholar grant, and the research of B.S. Braun was supported by an NIH grant (R01 CA173085). M.L. Loh is the UCSF Benioff Chair of Children's Health and the Deborah and Arthur Ablin Professor of Pediatric Molecular Oncology. We express our gratitude to the core facilities at UCSF, in particular the Animal Facilities core (LARC) for mouse housing and handling, the Preclinical Therapeutics core for the assistance in mouse dosing and Xenogen imaging, and the Mouse Pathology Core for the tissue dissection, staining, and analysis.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.