The BCL2 inhibitor venetoclax induces high rates of durable remission in patients with previously treated chronic lymphocytic leukemia (CLL). However, despite continuous daily treatment, leukemia recurs in most patients. To investigate the mechanisms of secondary resistance, we analyzed paired pre-venetoclax and progression samples from 15 patients with CLL progression enrolled on venetoclax clinical trials. The novel Gly101Val mutation in BCL2 was identified at progression in 7 patients, but not at study entry. It was first detectable after 19 to 42 months of therapy, and its emergence anticipated clinical disease progression by many months. Gly101Val reduces the affinity of BCL2 for venetoclax by ∼180-fold in surface plasmon resonance assays, thereby preventing the drug from displacing proapoptotic mediators from BCL2 in cells and conferring acquired resistance in cell lines and primary patient cells. This mutation provides new insights into the pathobiology of venetoclax resistance and provides a potential biomarker of impending clinical relapse.

Significance:

Why CLL recurs in patients who achieve remission with the BCL2 inhibitor venetoclax has been unknown. We provide the first description of an acquired point mutation in BCL2 arising recurrently and exclusively in venetoclax-treated patients. The mutation reduces venetoclax binding and is sufficient to confer resistance.

See related commentary by Thangavadivel and Byrd, p. 320.

This article is highlighted in the In This Issue feature, p. 305

Venetoclax is a potent and highly selective BCL2 inhibitor (1) approved for the treatment of patients with previously treated chronic lymphocytic leukemia (CLL), as monotherapy (2–5) or in combination with rituximab (6, 7). Venetoclax is taken continuously once daily until disease progression. Clinical responses are rapid, with ∼80% of patients achieving substantial cytoreduction (2, 3). Complete remissions are seen in ∼20% of patients with monotherapy and 20% to 50% when combined with rituximab (2, 3, 6–8). Further, a minority achieve remissions that are measurable residual disease (MRD)–negative by highly sensitive multicolor flow cytometry (MFC; refs. 7, 9). However, disease relapse ultimately occurs in the majority of patients who were heavily pretreated prior to commencing venetoclax. The median response duration in patients with deletion 17p CLL receiving venetoclax monotherapy is 33.2 months, highlighting the significant challenge secondary resistance poses (9).

Disease progression on venetoclax may manifest as transformation to diffuse large B-cell lymphoma [Richter transformation (RT)] in a small proportion of patients, but usually presents as a recurrence of CLL (CLL-type progression) without morphologic or immunophenotypic change (10, 11). The underlying biological mechanisms leading to CLL progression in venetoclax-treated patients remain largely unknown. Acquired abnormalities in BTG1, TP53, CDKN2A/B, SF3B1, and BRAF have been reported in a small cohort of 8 patients with deletion 17p CLL progressing on venetoclax, 4 with RT (12). However, a causal relationship has yet to be established, particularly for lesions that are common in CLL (e.g., TP53 and SF3B1), and the mechanisms that result in CLL-type progressions on venetoclax therapy may well be distinct from those emerging as RT.

Venetoclax inhibits BCL2′s function by binding its critical hydrophobic groove, the same site that sequesters its physiologic ligands, BH3 domain–containing proapoptotic proteins (13). If unconstrained by BCL2, these BH3-domain proteins (e.g., BIM, BAX, and BAK) are free to drive apoptosis. Acquired mutations in the BH3-binding groove of BCL2 have been reported in a murine cell line where venetoclax refractoriness was induced through continuous drug exposure (BCL2 Phe101Cys and Phe101Leu, analogous to human BCL2 Phe104Cys and Phe104Leu; ref. 14). However, BCL2 mutations have yet to be described in patients treated with venetoclax.

Using a cohort of 67 patients (10) with relapsed CLL treated with venetoclax on 3 early-phase clinical trials, we have performed a focused genomic evaluation in those with CLL-type progressions, specifically excluding those with RT (where genomic heterogeneity is well recognized). Among the CLL progressions on venetoclax, we have discovered and functionally characterized a recurrent novel BCL2 mutation [NM_000633.2:c.302G>T, p.(Gly101Val)] present at disease progression but not at treatment initiation. Gly101Val markedly reduces the affinity of venetoclax for BCL2 and confers acquired resistance in vitro and in vivo in patients with CLL.

Twenty-one patients treated with venetoclax in the study cohort (10) have experienced CLL-type progression after achieving an initial response, whereas 18 developed RT. CLL-type progression manifested clinically as a gradual increase in morphologic CLL burden over months to years, typically after a period when subclinical disease was initially detected in the bone marrow (and subsequently blood) by MFC. The median time on venetoclax to CLL-type clinical progression in this cohort was 36 (range, 6.5–73) months.

Genomic Assessment of Patient Samples Pre– and Post–Venetoclax Treatment

Fifteen of 21 patients with CLL-type progression had suitable pre-venetoclax and progression specimens available for analysis. Targeted amplicon next-generation sequencing (NGS) was performed on paired samples assessing the entire coding region of BCL2. In 4 patients (CLL2, CLL3, CLL5, and CLL12), a single-nucleotide variant was detected in BCL2 [NM_000633.2:c.302G>T, p.(Gly101Val)] in samples at, or following, CLL-type progression (Fig. 1A and B). This variant was undetectable in paired pre-venetoclax samples. Digital droplet PCR (ddPCR) was used to characterize the emergence of the mutation in serial archived samples from these patients (Fig. 1C) alongside bone marrow disease burden measured by MFC. Gly101Val was first detected at low variant allele frequency (VAF) after 19 to 42 months on venetoclax, up to 25 months earlier than when standard disease progression criteria were met.

CD19+-selected tumor cells of 92.8% purity collected at CLL progression from CLL3 had a Gly101Val VAF of 32.49% by ddPCR. No copy-number loss or acquired uniparental disomy at the BCL2 locus was observed (Supplementary Fig. S1A). This is consistent with Gly101Val being present in the heterozygous state in the majority (∼70%) of CLL cells in this patient at the time of progression. Expression of the Gly101Val mutation was confirmed in this patient by RNA sequencing (Supplementary Fig. S1B). Using sample CLL burden (determined by MFC) and ddPCR VAF, we estimated the proportion of CLL cells carrying Gly101Val in patients CLL2, CLL5, and CLL12 (assuming heterozygosity) to be approximately 26%, 46%, and 60%, respectively (Supplementary Table S1).

The observation of subclonality and detection of Gly101Val in preprogression samples at frequencies lower than detectable by targeted amplicon NGS led us to test whether it could be detected in other CLL-type progressors using ddPCR. Gly101Val was detected in 3 other patients (CLL6, CLL8, and CLL14) at low VAF (0.01%–0.4%) in post-venetoclax samples (Supplementary Table S1). The CLL disease burden assessed by MFC in these post-venetoclax samples ranged from 3.3% to 57.5% of leukocytes. Therefore, the estimated proportion of CLL cells carrying the Gly101Val mutation in these patients at progression was 1.4% to 4.3%. Importantly, Gly101Val was not detected in paired pre-venetoclax samples from any of the 7 Gly101Val-mutated patients using ddPCR (limit of detection 0.01%). After cessation of venetoclax, a stable proportion of CLL cells bearing Gly101Val persisted for at least 6 months in 5 patients who received BTK inhibitor therapy (Supplementary Fig. S2).

Targeted amplicon NGS of 32 other genes commonly mutated in CLL and lymphoid malignancy was also performed on pre- and postprogression samples (Supplementary Table S1). In 8 of 15 patients with CLL-type progressions, different sequence variants were detected (not including BCL2 Gly101Val) at progression compared with baseline. Five had TP53 variants at baseline that were not detected at CLL-type progression, 3 had new TP53 variants at progression; variants in NOTCH1, KRAS, and BIRC3 were lost in single patients at progression, and there were acquisitions of SF3B1 or KRAS variants in single patients.

BCL2 Gly101Val was not found in searches of either general population or cancer patient public databases (Fig. 1D). Importantly, review of BCL2 sequencing data from 96 patients with venetoclax-naïve CLL (63 untreated, 33 with relapsed/refractory disease; 23 with previous fludarabine-based treatment) did not reveal any cases bearing the Gly101Val mutation (Fig. 1D), and it was not observed in the limited number of biopsies analyzed from patients who progressed with RT (n = 5). Further, Gly101Val was not detected in targeted amplicon sequencing data from 301 patients with other B-cell malignancies. Nor did we detect any cases with the BCL2 Phe104Leu mutation analogous to that reported to confer venetoclax resistance in murine cell lines (14). Given the observation of acquisition of Gly101Val only in the setting of CLL-type progression on venetoclax treatment, we hypothesized that this variant contributes to venetoclax resistance.

Patient CLL Cells with Gly101Val at Progression and Gly101Val-Expressing Cell Lines Are Intrinsically Less Sensitive to Venetoclax

Freshly collected CLL cells from progression on venetoclax for the 4 patients with significant proportions of leukemic cells harboring Gly101Val were much less sensitive to venetoclax killing in vitro than their CLL cells collected at study entry (Fig. 2A). A similar acquisition of reduced sensitivity was observed when the cells were exposed to ABT-737, a dual BCL2/BCLxL inhibitor with a similar chemical scaffold (ref. 15; Supplementary Fig. S3).

To determine whether the Gly101Val mutation alone is sufficient to confer resistance to venetoclax, the mutant was overexpressed in two human B-lineage cell lines. RS4;11 Gly101Val cells were approximately 30-fold less sensitive to venetoclax than RS4;11 wild-type (WT) BCL2 cells (Fig. 2B and D), and similar results were observed using KMS-12-PE (Fig. 2C and D). The Phe104Leu mutation also induced similar resistance in these cell lines (Fig. 2B and C).

BCL2 Prosurvival Function Is Maintained Despite the Gly101Val Mutation

We next tested whether the prosurvival function of BCL2 was affected more generally by the Gly101Val mutation. High levels of BCL2 are associated with resistance to DNA-damaging cytotoxics and other anticancer drugs, as demonstrated when WT BCL2 was overexpressed in RS4;11 cells (Fig. 2E). Gly101Val similarly reduced in vitro sensitivity to each of etoposide, cytarabine, fludarabine, and dexamethasone, as did Phe104Leu (Fig. 2E). Following the same pattern, Gly101Val overexpression was as effective as WT BCL2 overexpression in KMS-12-PE cells at protecting against killing by bortezomib (Supplementary Fig. S3B). Thus, the normal function of BCL2 to block apoptosis is preserved despite the Gly101Val mutation.

Venetoclax Binding to BCL2 Is Markedly Reduced by the Gly101Val Mutation

Our functional studies in cell lines and the detection of the Gly101Val mutation exclusively in venetoclax-treated patients developing progressive CLL imply that the effect of this mutation is likely to be highly specific for the inhibition of BCL2 by venetoclax. To test this, we examined how the Gly101Val mutation affects the ability of BCL2 to bind either venetoclax or its physiologic ligands, the key proapoptotic proteins such as BIM, BAX, and BAK.

The Gly101Val mutation did not significantly affect the ability of BCL2 to bind a BIMBH3 peptide (Fig. 3A and B), consistent with our findings that the mutant broadly retains WT prosurvival activity (Fig. 2E). However, we saw a striking difference when we assessed the ability of venetoclax to compete with BIM for BCL2 binding (Fig. 3D and E). Competition surface plasmon resonance (SPR) experiments showed venetoclax binds avidly to WT BCL2, preventing BIM binding. This is markedly reduced (∼180-fold) with the Gly101Val mutant, as indicated by the flattening of the sigmoidal curves toward the hyperbola in the case of the mutation compared with WT protein (Fig. 3E). A modest reduction in BAXBH3 binding was also observed (Fig. 3A and C). Although the magnitudes varied, similar trends were observed with the Phe104Leu mutant. Given that the principal impact of these mutations is reduction in venetoclax binding to BCL2, we predicted that venetoclax would be less able to displace proapoptotic molecules from BCL2 Gly101Val in cells.

To investigate this, we assessed the ability of venetoclax to free proapoptotic BIM, BAX, or BAK already bound to BCL2 in cells. BIM binding to BCL2 (both WT and mutant) was very tight and was largely unaffected by the addition of venetoclax (Fig. 4A; Supplementary Fig. S4). In contrast, venetoclax could free BAX and BAK from WT BCL2, but was ineffective when these proapoptotic molecules were bound to the Gly101Val or the Phe104Leu mutants. Thus, in the biochemical and cellular assays, we found venetoclax to be less effective against the Gly101Val mutant, and this probably accounts for the reduced sensitivity of the cells to venetoclax. Interestingly, we also observed increased binding of BAK to the Gly101Val mutant, and reduced BIM and BAX binding in cells. Potentially, the impact of the mutation on the propensity of BCL2 to bind the various BH3 domain– containing proapoptotic proteins might also contribute to the resistance observed.

The Gly101Val Mutation Confers a Selective Advantage during Continuous Exposure to Sublethal Concentrations of Venetoclax

As our biochemical and short-term cellular studies highlight how the Gly101Val mutation impairs the sensitivity of BCL2 to venetoclax, we next explored the consequences of continuous exposure to venetoclax (as used clinically) in long-term cultures. In the absence of venetoclax, the WT and Gly101Val-expressing cells grew at similar rates. However, in the presence of a sublethal dose of venetoclax, the latter outcompeted WT cells. This occurred rapidly in KMS-12-PE cells and more slowly in RS4;11 (Fig. 4B), reflecting the 25% shorter doubling time and greater sensitivity to venetoclax of parental KMS-12-PE cells compared with RS4;11.

We next sought to understand how the presence of the Gly101Val mutation might interact with exogenous factors that also protect CLL cells from killing by venetoclax in vivo. Using an established model of the CLL microenvironment (16), we cocultured CLL cells from progression on CD40 ligand–expressing stromal cells. For comparison, we cocultured CLL cells with exclusively WT BCL2 from 3 venetoclax-naïve patients. As shown in Fig. 4C, freshly collected CLL3 cells were substantially less sensitive than venetoclax-naïve CLL cells prior to coculture. After 1 week on stroma, venetoclax-naïve CLL cells were less sensitive to venetoclax, displaying similar sensitivity to fresh CLL3 cells prior to coculture. With 1 week of coculture, cells from CLL3, CLL2, and CLL5 (Supplementary Fig. S5A) became markedly resistant to venetoclax, even at concentrations between 1 and 10 μmol/L, the latter higher than achievable clinically (2).

Additional Disease Resistance Mechanisms Can Coexist with the Gly101Val Mutation

Unlike cell line cultures, the patient samples showed significant subclonality for the Gly101Val mutation (e.g., 70% of malignant cells in CLL3 and 25% in CLL2). Cytotoxicity experiments in mixed populations of BCL2 WT and BCL2-mutant cell lines confirmed that the presence of subclones can shift the observed LC50 of the whole population (Supplementary Fig. S5B) by similar extent to that observed between study entry and progression for 3 patients (CLL3, CLL5, and CLL12; Fig. 2A). However, for CLL2, where the Gly101Val mutation–containing subclone comprised only 25% of the resistant leukemic population, the LC50 shift was ∼450-fold, markedly higher than observed in either the cell lines or the other 3 patients. We investigated prosurvival BCL2 family expression in CLL-progression cells from this patient by mass cytometry and observed bimodal expression of another prosurvival protein, BCLxL, not observed in the other patient samples (Fig. 4D; Supplementary Fig. S6). Cells with high BCLxL expression had much lower BCL2 expression than in either the BCLxL-low subpopulation or other CLL samples. When cells from CLL2 were flow-sorted based on BCLxL expression, the Gly101Val mutation was almost exclusively confined to the population with low BCLxL expression, indicating that distinct mechanisms of resistance were operating in these two populations (Fig. 4E). Notably, CLL2 was observed to be more sensitive to a BCLxL-specific inhibitor (A-1331852; ref. 17) than other CLL samples. Modeling of the data was consistent with the presence of two subpopulations, one sensitive to BCLxL inhibition and the other insensitive (Fig. 4F). Mass cytometry did not identify significant alterations in MCL1 or BCLxL expression in CLL cells in the other patient samples tested (CLL3, CLL5, and CLL12).

We have identified and functionally characterized a novel recurrent BCL2 mutation (Gly101Val) emerging in a cohort of patients with CLL-type progressions treated with venetoclax. Our data demonstrate that the Gly101Val mutation impairs binding of venetoclax to BCL2, confers resistance to venetoclax in both patient leukemia cells and engineered cell lines, and provides a selective growth advantage over WT cells when maintained in the presence of the drug in vitro.

CLL cells are universally addicted to the high levels of BCL2, which keep proapoptotic relatives such as BIM, BAX, and BAK in check by direct sequestration through its BH3-binding groove, thereby permitting their inappropriate survival (13, 18). By binding BCL2 tightly in the same groove, venetoclax effectively displaces these proapoptotic proteins, freeing them to trigger apoptosis of the CLL cells in vitro and in vivo (1, 19). Through competitive binding experiments, we showed that venetoclax has markedly decreased binding (∼180-fold) to Gly101Val, thereby compromising its ability to displace proapoptotic proteins such as BAX (Fig. 4A) and hence the induction of apoptosis (Figs. 2A–C and 4G). Because BCL2 binding to proapoptotic BIM and BAX was only marginally compromised, the ability of the Gly101Val mutant to function normally to protect cells from apoptosis in the absence of venetoclax was unaffected (Fig. 2E). Further investigation is required to formally establish whether Gly101Val has any modest gain or loss of function when assessed under different physiologic or pathologic circumstances.

Thus, the Gly101Val mutation specifically affects the action of venetoclax (Fig. 4E) and compounds with a similar chemical scaffold, such as ABT-737 and navitoclax (ref. 13; Fig. 2B and C; Supplementary Fig. S3A). Surprisingly, venetoclax does not directly contact Gly101, at least in the available structure of a venetoclax analogue bound to BCL2 (ref. 1; Fig. 1B). This residue is highly conserved among BCL2 prosurvival family members, and the binding groove is malleable such that it can accommodate multiple ligands, including a variety of BH3 domains and BH3-mimetic drugs (15, 20). We speculate that the mutation must indirectly affect the binding groove to selectively impair venetoclax binding. Gly101Val represents an early example of a mutation that induces resistance to a targeted therapy by modifying a protein–protein interaction site, rather than an enzyme active site (e.g., mutations in BCR–ABL1, ref. 21; and BTK, ref. 22).

Although our experimental data indicate that the Gly101Val mutation is sufficient to confer resistance to venetoclax and outgrowth of mutation-bearing clones in the presence of continuous venetoclax exposure in vitro and in patients, our data also indicate that other mechanisms for development of resistance are clinically important. The wide range of subclonality of Gly101Val in our cohort (1.4%–70%) suggests that other acquired changes must confer resistance in non–BCL2 mutation–bearing subclones. Further research is required to identify these alternative mechanisms. However, in 1 patient with cryopreserved cells available for analysis, we were able to demonstrate increased BCLxL expression in a major subclone without the Gly101Val mutation. Increased BCLxL expression has been described recently as a mechanism of resistance to venetoclax in mantle cell lymphoma (23, 24) and appears to be a major contributor to resistance in patient CLL2. Whether changes in expression of BCLxL or other BCL2 family proteins are common in CLL progression on venetoclax remains to be established. Of broad relevance, we also demonstrated that exogenous microenvironmental support amplified the intrinsic resistance of Gly101Val mutation–bearing CLL cells in all samples available for testing. This combined effect probably explains the scenario observed in our patients where BCL2-mutant clones with LC50 values of 100 nmol/L to 1 μmol/L accumulated despite ongoing exposure to the typical low micromolar (approximately 1–3 μmol/L) steady-state concentrations of venetoclax observed in vivo with continuous therapy (2).

Like others (12), we observed genomic changes between treatment commencement and progression, particularly acquisition and loss of TP53 variants, but no compelling candidates that could directly account for resistance to venetoclax. We have previously shown that venetoclax-mediated CLL cell death in vitro is independent of TP53 mutation and function (19). Most likely, the association of TP53 aberrations with venetoclax resistance is functionally indirect, reflecting the underlying genomic instability in CLL cells from this heavily pretreated, relapsed/refractory cohort of patients (10). Transcriptomic or epigenomic profiling of suitable clinical specimens when they become available may assist in the identification of additional causes of venetoclax resistance, as may broader genomic profiling with whole-exome and whole-genome sequencing. Although further research is required into the development of RT as an avenue to clinical resistance to venetoclax, we have not observed BCL2 mutations to be contributory. Reassuringly, recent clinical data suggest that RT is infrequently observed on venetoclax in less heavily pretreated patients (7).

Our data support the concept that venetoclax resistance in CLL-type progression is a heterogeneous phenomenon, both between and within individual patients. As we have observed disease progression in every case where Gly101Val was found and that this mutation could be detected before overt clinical relapse in patients through sensitive allele-specific ddPCR, this discovery may have clinical application. Detection of the Gly101Val could potentially serve as an early biomarker of impending disease progression, allowing early therapeutic intervention such as the addition of another drug with a different mechanism of action.

See also Supplementary Materials for additional details.

Clinical Cohort

The patient cohort of the first 67 consecutive patients with relapsed CLL/small lymphocytic lymphoma treated with venetoclax on clinical trials at our two institutions has been previously described (10). All provided written informed consent, and the studies were conducted in accordance with the Declaration of Helsinki and after Human Research Ethics Committee approval. Patients were enrolled from June 2011 to March 2015 and were reviewed at a minimum every 3 months until progression, death, or discontinuation from trial. Outcome data were updated to June 1, 2018.

NGS

Sequences of targeted regions within 33 genes (listed in Supplementary Methods) were analyzed. Indexed, amplicon-based libraries were prepared using Access Array methodology (Fluidigm) and sequenced to a depth of ∼1,000× on a MiSeq instrument using v2 chemistry (Illumina). Alignment, variant calling, and annotation were performed using an in-house pipeline. Variants were evaluated using multiple functional and quality filters to identify likely pathogenic variants. Sanger sequencing was used to confirm selected variants. Sample CLL3 was sequenced with a hybridization-based (Agilent SureSelect) NGS panel targeting 363 genes with mean target read depth of ∼600×, as described previously (25). Whole-transcriptome sequencing was performed using TruSeq RNA Library Prep Kit v2 on patient CLL3 using RNA extracted from CD19+ selected tumor cells.

Digital Droplet PCR

A ddPCR assay to detect the BCL2 NM_000633.2:c.302G>T, p. (Gly101Val) variant using forward (5′-CTGGACATCTCGGCGAAG-3′) and reverse (5′-HEX-CC+G+G+CGAC+GA-IABkFQ-3′) oligonucleotide primers with locked nucleic acid probes against WT and mutant sequence was designed. PCR reactions were carried out with 1× ddPCR Supermix for Probes (no dUTP), 2 μL of 360 GC enhancer (Applied Biosystems), primers (900 nmol/L), probes (250 nmol/L), and 50 ng of DNA template to a final volume of 25 μL. PCR cycling conditions were activation (95°C, 10 minutes), 40 cycles of denaturation (94°C, 30 seconds), annealing/extension (60°C, 1 minute), and enzyme deactivation (98°C, 10 minutes). All steps had a ramp rate of 2°C/second. Droplets were generated using the Automatic Droplet generator QX200 AutoDG (Bio-Rad) and read on the QX200TM Droplet Reader with data acquired and analyzed by QuantaSoft software (Bio-Rad).

Cell Biology

Cell Lines.

KMS-12-PE (sourced from DSMZ in 2013; cat. #ACC606) and RS4;11 (sourced from ATCC in 2014; cat. #CRL-1873) were cultured with HTRPMI/10% fetal bovine serum. Early passages (P5–P7) after purchase were cryopreserved and thawed for the experiments. Cells beyond passage 15 were not used. Monthly tests for Mycoplasma were consistently negative (MycoAlert mycoplasma detection kit; Lonza). In vitro killing assays (19) and coculture (16) were performed as previously described, with minor adjustments (see Supplementary Materials).

Plasmids, Retrovirus Production, and Infection.

WT BCL2 construct was reported previously (26); the mutants were introduced by PCR using primers with the desired mutation. cDNAs encoding either WT or mutant FLAG-tagged BCL2 were inserted into the MSCV-IRES-hygromycin retroviral construct; retroviral production and generation of cell lines was performed as previously described (27).

Cell Viability Assays.

Sensitivity of the engineered cell lines to cytotoxics was determined in cell viability assays (CellTiter-Glo assay; Promega, cat. #G9241) as previously described (27). For the competition experiments, WT (GFP+) or Gly101Val BCL2 (BFP+) cells were mixed in a 95:5 ratio, and their growth in 100 nmol/L venetoclax or DMSO was monitored by flow cytometry.

SPR Binding Experiments

Experiments were performed using a BIAcore 4000 using a SA sensor chip (GE Healthcare) immobilized with biotinylated BIMBH3 or BAXBH3 peptides, with BIMBH3-4A peptide as a nonbinding reference. Peptide affinities were determined by direct binding with BCL2 (0–63 nmol/L) as the analyte. Venetoclax affinity was determined by competition against immobilized BIMBH3 peptide, using BCL2 (0–250 nmol/L) premixed with venetoclax (0, 20, 40, 60 nmol/L) as the analyte. Direct binding experiments were fitted to a 1:1 binding site model and competition to a steady-state competition model; see also Supplementary Methods.

Coimmunoprecipitation and Mass Cytometry

FLAG-tagged WT or mutant BCL2 was immunoprecipitated from equivalent lysates prepared from venetoclax-treated or control cells with the rat anti-FLAG antibody (clone 9H1, WEHI). The coprecipitated proteins were detected with antibodies to BIM (clone 3C5, WEHI), BAX (clone 21C10, WEHI), or BAK (clone 7D10, WEHI).

For mass cytometry, fresh or thawed cryopreserved patient cells were incubated with cisplatin to allow detection of nonviable cells, then fixed with paraformaldehyde (PFA; Electron Microscopy Sciences), pelleted, washed, and stored at −80°C.

Cell samples for batch analysis were barcoded using the 20-plex palladium barcoding kit according to the manufacturer's instructions (Fluidigm). Following barcoding, cells were pelleted, washed, and then permeabilized at 4°C with methanol for 10 minutes. After washing, cells were incubated with specific antibody–metal conjugates (see Supplementary Methods for specific listings), then washed and stained with 125 nmol/L 191Ir/193Ir DNA intercalator (Fluidigm) overnight. After washing, cells were resuspended with EQ normalization beads immediately before analysis using a Helios mass cytometer (Fluidigm). Throughout the analysis, cells were maintained at 4°C and introduced at a constant rate of ∼300 cells/second. Data concatenation, normalization, and debarcoding were done with the Helios software, version 6.7.1014 (Fluidigm). Single cells were gated using the FlowJo (version 10.5) and Cytobank software (http://www.cytobank.org) based on event length and 191Ir/193Ir DNA content to avoid debris and doublets. Following single-cell gating, CD5+CD19+ live nonapoptotic cisplatinlo cells were analyzed for levels of the various metal conjugates.

The Walter and Eliza Hall Institute receives milestone and royalty payments related to venetoclax, and employees may be eligible for benefits related to these payments. M.A. Anderson is a hematologist at the Department of Haematology, Royal Melbourne Hospital and Peter McCallum Cancer Centre; is an employee of the Walter and Eliza Hall Institute which receives milestone and royalty payments related to venetoclax; and has received honoraria from the speakers bureau of AbbVie. J.-n. Gong, R. Thijssen, R.W. Birkinshaw, C.E. Teh, Z. Xu, C. Flensburg, and I.J. Majewski have received remuneration from the Walter and Eliza Hall Institute. T.E. Lew is an employee of the Walter and Eliza Hall Institute of Medical Research and receives milestone and royalty payments related to venetoclax. D.H.D. Gray reports receiving commercial research support from Servier Pharmaceuticals and has received other remuneration from the Walter and Eliza Hall Institute. C.S. Tam has received honoraria from the speakers bureau of AbbVie. J.F. Seymour reports receiving commercial research support from AbbVie, Celgene, Janssen, and Roche; has received honoraria from the speakers bureaus of AbbVie and Roche; and is a consultant/advisory board member for AbbVie, Acerta, Celgene, Janssen, Morphosys, Roche, Sunesis, and Takeda. P.E. Czabotar reports receiving a commercial research grant from Anaxis, is a consultant/advisory board member for Anaxis, and has received other remuneration from the Walter and Eliza Hall Institute. D.C.S. Huang reports receiving a commercial research grant from Servier, has received honoraria from the speakers bureau of AbbVie, and has received other remuneration from the Walter and Eliza Hall Institute of Medical Research. A.W. Roberts receives milestones and royalties related to venetoclax from the Walter and Eliza Hall Institute of Medical Research, reports receiving commercial research grants from AbbVie and Janssen, and has received other remuneration from the Walter and Eliza Hall Institute of Medical Research. No potential conflicts of interest were disclosed by the other authors.

Conception and design: P. Blombery, M.A. Anderson, R. Thijssen, R.W. Birkinshaw, D.C.S. Huang, A.W. Roberts

Development of methodology: P. Blombery, M.A. Anderson, J.-n. Gong, R. Thijssen, R.W. Birkinshaw, E.R. Thompson, C.E. Teh, T. Nguyen, D.H.D. Gray, D.C.S. Huang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Blombery, M.A. Anderson, J.-n. Gong, R. Thijssen, R.W. Birkinshaw, C.E. Teh, T. Nguyen, Z. Xu, I.J. Majewski, J.F. Seymour, D.C.S. Huang, A.W. Roberts

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Blombery, M.A. Anderson, J.-n. Gong, R. Thijssen, R.W. Birkinshaw, E.R. Thompson, Z. Xu, C. Flensburg, T.E. Lew, I.J. Majewski, D.H.D. Gray, C.S. Tam, J.F. Seymour, P.E. Czabotar, D.C.S. Huang, A.W. Roberts

Writing, review, and/or revision of the manuscript: P. Blombery, M.A. Anderson, J.-n. Gong, R. Thijssen, R.W. Birkinshaw, E.R. Thompson, Z. Xu, T.E. Lew, I.J. Majewski, D.H.D. Gray, D.A. Westerman, C.S. Tam, J.F. Seymour, P.E. Czabotar, D.C.S. Huang, A.W. Roberts

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.A. Anderson, R.W. Birkinshaw, T. Nguyen, T.E. Lew, D.C.S. Huang, A.W. Roberts

Study supervision: D.A. Westerman, P.E. Czabotar, D.C.S. Huang, A.W. Roberts

The authors are grateful to the patients who enrolled on the venetoclax clinical trials, and for assistance from Naomi Sprigg with collection and curation of samples, Dr. Andrew Mitchell for mass cytometry maintenance/operation, and Tania Tan for sample preparation for mass cytometry. The mass cytometry was performed in part at the Materials Characterisation and Fabrication Platform at the University of Melbourne and the Victorian Node of the Australian National Fabrication Facility. This research was supported by grants from the Snowdome Foundation (P. Blombery and M.A. Anderson), Vision Super and the Wilson Centre of Lymphoma Genomics (P. Blombery), the Leukemia and Lymphoma Society [fellowship to R. Thijssen (5467-18); SCOR (7015-18) to D.C.S. Huang and A.W. Roberts], the Cancer Council of Victoria (grants 1146518 and 1102104 to D.H.D. Gray; 1124178 to I.J. Majewski), the Victorian Cancer Agency (fellowship to I.J. Majewski), the CLL Global Foundation (C.S. Tam), the National Health and Medical Research Council (NHMRC) of Australia [fellowships to C.E. Teh (1089072), D.H.D. Gray (1090236), P.E. Czabotar (1079700), D.C.S. Huang (1043149), and A.W. Roberts (1079560)], and grants to P.E. Czabotar (project 1141874), D.C.S. Huang (program 1113133), and A.W. Roberts (program 1113577). This work was made possible through Victorian State Government Operational Infrastructure Support, Australian Government NHMRC IRIISS, and financial support for mass cytometry by the Victorian Comprehensive Cancer Centre.

1.
Souers
AJ
,
Leverson
JD
,
Boghaert
ER
,
Ackler
SL
,
Catron
ND
,
Chen
J
, et al
ABT-199, a potent and selective BCL-2 inhibitor, achieves antitumor activity while sparing platelets
.
Nat Med
2013
;
19
:
202
8
.
2.
Roberts
AW
,
Davids
MS
,
Pagel
JM
,
Kahl
BS
,
Puvvada
SD
,
Gerecitano
JF
, et al
Targeting BCL2 with venetoclax in relapsed chronic lymphocytic leukemia
.
N Engl J Med
2016
;
374
:
311
22
.
3.
Stilgenbauer
S
,
Eichhorst
B
,
Schetelig
J
,
Coutre
S
,
Seymour
JF
,
Munir
T
, et al
Venetoclax in relapsed or refractory chronic lymphocytic leukaemia with 17p deletion: a multicentre, open-label, phase 2 study
.
Lancet Oncol
2016
;
17
:
768
78
.
4.
Coutre
S
,
Choi
M
,
Furman
RR
,
Eradat
H
,
Heffner
L
,
Jones
JA
, et al
Venetoclax for patients with chronic lymphocytic leukemia who progressed during or after idelalisib therapy
.
Blood
2018
;
131
:
1704
11
.
5.
Jones
JA
,
Mato
AR
,
Wierda
WG
,
Davids
MS
,
Choi
M
,
Cheson
BD
, et al
Venetoclax for chronic lymphocytic leukaemia progressing after ibrutinib: an interim analysis of a multicentre, open-label, phase 2 trial
.
Lancet Oncol
2018
;
19
:
65
75
.
6.
Seymour
JF
,
Ma
S
,
Brander
DM
,
Choi
MY
,
Barrientos
J
,
Davids
MS
, et al
Venetoclax plus rituximab in relapsed or refractory chronic lymphocytic leukaemia: a phase 1b study
.
Lancet Oncol
2017
;
18
:
230
40
.
7.
Seymour
JF
,
Kipps
TJ
,
Eichhorst
B
,
Hillmen
P
,
D'Rozario
J
,
Assouline
S
, et al
Venetoclax-rituximab in relapsed or refractory chronic lymphocytic leukemia
.
The N Engl J Med
2018
;
378
:
1107
20
.
8.
Roberts
AW
,
Stilgenbauer
S
,
Seymour
JF
,
Huang
DCS
. 
Venetoclax in patients with previously treated chronic lymphocytic leukemia
.
Clin Cancer Res
2017
;
23
:
4527
33
.
9.
Stilgenbauer
S
,
Eichhorst
B
,
Schetelig
J
,
Hillmen
P
,
Seymour
JF
,
Coutre
S
, et al
Venetoclax for patients with chronic lymphocytic leukemia with 17p deletion: results from the full population of a phase II pivotal trial
.
J Clin Oncol
2018
;
36
:
1973
80
.
10.
Anderson
MA
,
Tam
C
,
Lew
TE
,
Juneja
S
,
Juneja
M
,
Westerman
D
, et al
Clinicopathological features and outcomes of progression of CLL on the BCL2 inhibitor venetoclax
.
Blood
2017
;
129
:
3362
70
.
11.
Davids
MS
,
Hallek
M
,
Wierda
W
,
Roberts
AW
,
Stilgenbauer
S
,
Jones
JA
, et al
Comprehensive safety analysis of venetoclax monotherapy for patients with relapsed/refractory chronic lymphocytic leukemia
.
Clin Cancer Res
2018
;
24
:
4371
9
.
12.
Herling
CD
,
Abedpour
N
,
Weiss
J
,
Schmitt
A
,
Jachimowicz
RD
,
Merkel
O
, et al
Clonal dynamics towards the development of venetoclax resistance in chronic lymphocytic leukemia
.
Nat Commun
2018
;
9
:
727
.
13.
Czabotar
PE
,
Lessene
G
,
Strasser
A
,
Adams
JM
. 
Control of apoptosis by the BCL-2 protein family: implications for physiology and therapy
.
Nat Rev Mol Cell Biol
2014
;
15
:
49
63
.
14.
Fresquet
V
,
Rieger
M
,
Carolis
C
,
Garcia-Barchino
MJ
,
Martinez-Climent
JA
. 
Acquired mutations in BCL2 family proteins conferring resistance to the BH3 mimetic ABT-199 in lymphoma
.
Blood
2014
;
123
:
4111
9
.
15.
Oltersdorf
T
,
Elmore
SW
,
Shoemaker
AR
,
Armstrong
RC
,
Augeri
DJ
,
Belli
BA
, et al
An inhibitor of Bcl-2 family proteins induces regression of solid tumours
.
Nature
2005
;
435
:
677
81
.
16.
Thijssen
R
,
Slinger
E
,
Weller
K
,
Geest
CR
,
Beaumont
T
,
van Oers
MH
, et al
Resistance to ABT-199 induced by microenvironmental signals in chronic lymphocytic leukemia can be counteracted by CD20 antibodies or kinase inhibitors
.
Haematologica
2015
;
100
:
e302
6
.
17.
Leverson
JD
,
Phillips
DC
,
Mitten
MJ
,
Boghaert
ER
,
Diaz
D
,
Tahir
SK
, et al
Exploiting selective BCL-2 family inhibitors to dissect cell survival dependencies and define improved strategies for cancer therapy
.
Sci Transl Med
2015
;
7
:
279ra40
.
18.
Del Gaizo Moore
V
,
Brown
JR
,
Certo
M
,
Love
TM
,
Novina
CD
,
Letai
A
. 
Chronic lymphocytic leukemia requires BCL2 to sequester prodeath BIM, explaining sensitivity to BCL2 antagonist ABT-737
.
J Clin Invest
2007
;
117
:
112
21
.
19.
Anderson
MA
,
Deng
J
,
Seymour
JF
,
Tam
C
,
Kim
SY
,
Fein
J
, et al
The BCL2 selective inhibitor venetoclax induces rapid onset apoptosis of CLL cells in patients via a TP53-independent mechanism
.
Blood
2016
;
127
:
3215
24
.
20.
Ku
B
,
Liang
C
,
Jung
JU
,
Oh
BH
. 
Evidence that inhibition of BAX activation by BCL-2 involves its tight and preferential interaction with the BH3 domain of BAX
.
Cell Res
2011
;
21
:
627
41
.
21.
Gorre
ME
,
Mohammed
M
,
Ellwood
K
,
Hsu
N
,
Paquette
R
,
Rao
PN
, et al
Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification
.
Science
2001
;
293
:
876
80
.
22.
Woyach
JA
,
Furman
RR
,
Liu
TM
,
Ozer
HG
,
Zapatka
M
,
Ruppert
AS
, et al
Resistance mechanisms for the Bruton's tyrosine kinase inhibitor ibrutinib
.
N Engl J Med
2014
;
370
:
2286
94
.
23.
Jayappa
KD
,
Portell
CA
,
Gordon
VL
,
Capaldo
BJ
,
Bekiranov
S
,
Axelrod
MJ
, et al
Microenvironmental agonists generate de novo phenotypic resistance to combined ibrutinib plus venetoclax in CLL and MCL
.
Blood Adv
2017
;
1
:
933
46
.
24.
Chiron
D
,
Bellanger
C
,
Papin
A
,
Tessoulin
B
,
Dousset
C
,
Maiga
S
, et al
Rational targeted therapies to overcome microenvironment-dependent expansion of mantle cell lymphoma
.
Blood
2016
;
128
:
2808
18
.
25.
Ryland
GL
,
Jones
K
,
Chin
M
,
Markham
J
,
Aydogan
E
,
Kankanige
Y
, et al
Novel genomic findings in multiple myeloma identified through routine diagnostic sequencing
.
J Clin Pathol
2018
;
71
:
895
9
.
26.
van Delft
MF
,
Wei
AH
,
Mason
KD
,
Vandenberg
CJ
,
Chen
L
,
Czabotar
PE
, et al
The BH3 mimetic ABT-737 targets selective Bcl-2 proteins and efficiently induces apoptosis via Bak/Bax if Mcl-1 is neutralized
.
Cancer Cell
2006
;
10
:
389
99
.
27.
Gong
JN
,
Khong
T
,
Segal
D
,
Yao
Y
,
Riffkin
CD
,
Garnier
JM
, et al
Hierarchy for targeting prosurvival BCL2 family proteins in multiple myeloma: pivotal role of MCL1
.
Blood
2016
;
128
:
1834
44
.

Supplementary data