Intact TP-53 function is essential for sustaining durable responses to BH3-mimetic drugs in leukemias

Venetoclax has emerged as a new and effective treatment option for patients with chronic lymphocytic leukemia (CLL)¹  or acute myeloid leukemia (AML).^2-4  Venetoclax belongs to a novel BH3-mimetic class of small molecules that selectively target BCL-2, MCL-1, or their prosurvival relatives, activating the apoptosis effectors BAX and BAK to drive mitochondrial outer membrane permeabilization (MOMP) and cell death.⁵ 

The tumor suppressor TP-53 coordinates the cellular response to noxious stress, including DNA damage. TP-53 activation leads to increased expression of genes encoding BH3-only proteins (PUMA, NOXA) that promote apoptosis. TP-53 operates upstream of the BCL-2 protein family, supporting the notion that defects in TP-53 activity should not compromise the action of BH3-mimetic drugs.⁵  Consistent with this, TP53-mutant CLL cells are highly sensitive to venetoclax in vitro,⁶  and clinical responses to venetoclax are similar for both wild-type and TP53-mutant CLL,^1,6  in contrast to inferior outcomes of conventional chemotherapy in patients with TP53-mutant disease. However, recent studies challenge the assumption that BH3-mimetic activity is agnostic to TP-53 status. Genetic screening in AML cell lines has linked venetoclax resistance to TP-53 loss.^7-9  In clinical studies, despite initial responses to venetoclax, TP-53 abnormalities predispose patients to earlier disease recurrence of CLL¹⁰  or AML, despite combination with chemotherapy or hypomethylating agents.^3,11,12 

To reconcile these apparently contradictory observations, we investigated the impact of TP-53 dysfunction on response of hematological malignancies to BH3-mimetic drugs. Across multiple models and corroborated by clinical observations, we observed that, in the presence of a sublethal concentration of BH3-mimetic drugs targeting either BCL-2 or MCL-1, TP-53 deficiency conferred a competitive advantage to surviving cells, impairing activation of proapoptotic BAX/BAK and enabling cells to outgrow their TP-53–intact counterparts. Although combining BH3-mimetics with cytotoxic drugs failed to avert outgrowth of TP-53–deficient cells, sustained responses could be achieved when BH3-mimetic agents were combined to cotarget BCL-2 and MCL-1, resulting in potent activity, irrespective of TP53 mutation status.

Our findings have significant implications for how BH3-mimetic drugs are ideally used in the clinic. Our results highlight the role of defective TP-53 in promoting treatment failure if a sublethal BH3-mimetic dose is used. Furthermore, our results support the deployment of novel therapeutic approaches with TP-53–independent activity, such as a tandem BH3-mimetic strategy to improve treatment outcomes in leukemias with defective TP-53.

Full details are provided in supplemental Materials, available on the Blood Web site.

Samples were obtained after the patients provided informed consent. Bone marrow (BM) samples were from patients participating in the phase 1 dose-escalation study of the BCL-2 inhibitor S55746 as a monotherapy in AML and higher risk myelodysplasia (Protocol CL1-055746-002). Additional samples were collected from patients participating in the Chemotherapy and Venetoclax in Elderly AML Trial (CAVEAT).¹²  All experiments were conducted in accordance with institutional requirements and the Declaration of Helsinki.

Mouse experiments were conducted with institutional approvals. Cells (4 × 10⁵ MOLM-13; 1 × 10⁵ MV-4-11) were injected IV into nonirradiated NOD-SCID IL2Rγ-null (NSG) mice. Primary AML leukemic blasts (1 × 10⁶) were injected into NOD-IL2Rcγ^null (NRG-SG3) mice. Mice were treated with various dose regimens of the BH3-mimetics targeting BCL-2 and MCL-1.

Virally delivered CRISPR/Cas9 technology was used to derive gene KO variants of diverse blood cancer–derived cell lines.¹³  Whole-genome KO screenings to identify resistance factors to the MCL-1 inhibitor (MCL-1i) S63845 were performed by transducing Cas9-expressing Eµ-Myc lymphoma cells with the mouse YUSA library.¹⁴ 

Control WT TP53 GFP⁺ cells were seeded at various ratios with test KO TP53 BFP⁺ cells and treated with doses of the BH3-mimetics or dimethyl sulfoxide (DMSO) as the control. The relative proportion of viable GFP⁺ or BFP⁺ cells was evaluated by flow cytometry (CytoFLEX or LSR II; BD). Similar experiments were performed with Eµ-Myc mouse lymphoma cells.

BH3-mimetic (or DMSO)–treated human cancer cells were fixed, permeabilized, and stained with antibodies to activated BAX (biotinylated rat monoclonal clone 1E5; Walter and Eliza Hall Institute), activated BAK (mouse monoclonal clone G317-2; cat. no. 556382; BD Biosciences) and antibodies against cytochrome c (clone REA702, cat. no. 130-111-368; Miltenyi Biotec). Cells were analyzed on a FACSFortessa (BD Biosciences).

We initially examined our records for patients with TP53-aberrant AML who had been treated with only BCL-2 inhibition. A patient with monosomal karyotype (45,XX,add(11)(p12),-17) AML refractory to standard induction chemotherapy received S55746,^15,16  a selective BCL-2 inhibitor equivalent to venetoclax.¹⁷  At screening, the leukemia harbored a dominant TP53-null clone (monosomy 17 and a P151H mutation). After S55746 for 21 days, the patient achieved near-complete remission with marked reduction in the variant allele burden of the TP53-mutated clone, along with concordant reduction in IDH1 mutation (Figure 1A; supplemental Table 1). Two further courses of S55746 were administered before the patient underwent an elective allogeneic hematopoietic stem cell transplant.

Analogous to the response to venetoclax in TP53-mutant CLL,^1,6  this AML case illustrates that defective TP53 does not preclude a robust clinical response to BCL-2 inhibition. This prompted us to further explore BCL-2 inhibition in patients with TP53 abnormalities.

To gain further insight into the clinical impact of venetoclax on TP53-mutated AML, we compared the leukemic burden in BM before and after a 7-day prephase of venetoclax, with doses ranging from 50 to 600 mg/d (maximum dose approved in AML), given before the addition of cytarabine and idarubicin as part of a phase 1b clinical trial (CAVEAT).¹²  Nine patients (supplemental Table 2) harboring TP53-mutant AML, displayed a variable response to venetoclax (Figure 1B). Although BM blast reductions were noted in some patients, we focused attention on 4 venetoclax nonresponsive cases. Paired BM samples before and after 7 days of venetoclax were characterized by single-cell targeted DNA sequencing (supplemental Table 3). In all 4 cases, expansion of TP53-mutant relative to TP53-wild-type clones was noted, even at the highest venetoclax dose (600 mg; Figure 1C-F). In 2 patients (CAL-012 and CAL-030; Figure 1C-D), expansion of TP53-mutant clones was propagated by commutations known to promote cell growth (N-RAS, K-RAS, and PTPN11). Therefore, although prior reports suggest that TP-53 deletion in AML cell lines is associated with venetoclax resistance in vitro,⁸  our clinical findings suggest that early response of TP53 mutant AML clones to venetoclax may be variable.

To explore the consequences of TP53 deficiency on response to BH3-mimetic drugs, TP53 was deleted by CRISPR/Cas9 in 2 human AML cell lines: MOLM-13 and MV-4-11 (Figure 2A; supplemental Figure 1A). As expected, TP53 loss rendered both cell lines resistant to apoptosis by an MDM2 inhibitor (supplemental Figure 1B). In contrast, the impact of TP-53 loss on venetoclax sensitivity in short-term (24 hour) assays was variable, with the 50% inhibitory concentration (IC₅₀) increased by approximately twelvefold to eightfold in MV-4-11 cells (supplemental Figure 1C) to only approximately threefold in MOLM-13 cells (Figure 2B). These results prompted further exploration into how TP-53 loss may influence the response to venetoclax.

As resistance to venetoclax imparted by TP-53 loss in short-term assays was only modest, we next examined the impact of TP53 loss in longer term assays spanning several weeks. Fluorescently marked TP53-wild-type (GFP⁺) and TP53-deficient (BFP⁺) MOLM-13 cells were mixed in equal ratio and their proportions tracked during continuous exposure to venetoclax (Figure 2C). Although TP53-deletion did not alter outcomes using an IC₈₀ dose of venetoclax, with cells of either genotype killed within 48 hours (not shown), an IC₂₀ of venetoclax led to marked outgrowth of TP53-deficient relative to TP53-wild-type cells after 1 week of therapy, whereas spontaneous outgrowth of TP53-deficient cells in control media was more gradual (Figure 2D). Outgrowth of TP53-deficient over wild-type cells seeded at a 1:20 ratio was also evident using an IC₅₀ of venetoclax (Figure 2E). Collectively, these results suggest that short-term measures of venetoclax drug sensitivity (Figure 2B) may conceal the potential for dominant outgrowth of TP53-deficient cells exposed to suboptimal concentrations over an extended period (Figure 2D-E). Short-term cytotoxicity experiments performed weekly during the long-term cell competition assays revealed no significant change in the IC₅₀ sensitivity of TP53-deficient MOLM-13 cells to venetoclax, thus ruling out the possibility of acquired drug resistance (Figure 2F).

To determine the relevance of these findings in vivo, an equal ratio of TP53-wild-type and -deficient MOLM-13 cells were inoculated into NSG mice, and 3 days later, treatment with either vehicle or venetoclax (75 mg/kg per day) for a 2-week period commenced (Figure 2G). By the completion of treatment, the proportion of TP53-deficient to wild-type MOLM-13 cells was significantly higher in the venetoclax-treated cohort (Figure 2H). These findings confirm the potential for venetoclax to favor expansion of TP53-deficient, compared with wild-type AML populations in vivo (Figure 1C-F).

Similar findings were made with TP53-deficient MV-4-11 cells.¹¹  Moreover, MV-4-11 cells harboring a spontaneous TP53 mutation (R248W) also exhibited a competitive advantage over wild-type counterparts when exposed to a sublethal IC₂₀ of venetoclax in extended cultures (supplemental Figure 1D). Collectively, these findings reveal a competitive advantage for TP-53–defective AML populations if treated for extended periods with a suboptimal concentration of venetoclax.

We extended our experiments to determine whether defective TP53 could present a barrier to venetoclax efficacy in models of B-cell malignancy. Deleting TP53 in the human B-ALL cell line RS4;11 (Figure 3A) impaired killing by an MDM2 inhibitor (supplemental Figure 2A), but had no impact on venetoclax sensitivity in short-term (24 hour) assays (Figure 3B). However, in extended duration cell competition assays, TP53-deficient RS4;11 cells expanded more rapidly during continuous exposure to sublethal vs higher doses of venetoclax (Figure 3C-D). Furthermore, there was no evidence of acquired increased drug resistance during the 4-week culture period (Figure 3E). Similar findings were observed for the diffuse large B-cell lymphoma (DLBCL) cell line OCI-LY19 (Figure 3F; supplemental Figure 2B).

In considering how TP-53–deficient leukemic cells out-compete WT cells in response to venetoclax, we initially explored whether venetoclax could indirectly activate TP-53. DNA damage activates TP-53 to drive a transcriptional program promoting apoptosis in hematopoietic cells.⁵  It has been proposed that venetoclax may cause DNA damage, potentially accounting for impaired apoptosis in TP-53–deficient leukemic cells.¹⁸  To avoid the confounding effect of DNA damage resulting from caspase activation in the later stages of apoptosis, we used BAX/BAK-deficient cells known to be resistant to venetoclax-induced cell death.^19,20  Agents known to directly cause DNA damage in the absence of BAX/BAK (eg, cisplatin and fludarabine) triggered dose-dependent DNA damage, as revealed by an increase in γH2AX staining. In contrast, no impact on γH2AX staining resulted from inhibition of BCL-2, ruling out the potential for venetoclax to induce direct DNA damage (Figure 4A).

Venetoclax has been reported to enhance metabolic vulnerability by impairing oxidative phosphorylation in AML cells.²¹  We therefore considered whether TP-53 deficiency could alter mitochondrial metabolism, reducing venetoclax sensitivity.⁷  We compared real-time metabolic flux in response to venetoclax in wild-type and TP53-deficient cells. Consistent with previous reports, venetoclax substantially impaired the basal oxygen consumption rate, as well as the glycolytic reserve in MOLM-13 (Figure 4B) and MV-4-11 cells (supplemental Figure 3A). TP-53 deficiency, however, neither enhanced the basal respiratory capacity of leukemic cells, nor impaired the reduction in glycolytic capacity induced by venetoclax compared with TP53 wild-type cells (Figure 4B; supplemental Figure 3A).

It has also been postulated that TP-53 deficiency could render leukemic cells less sensitive to induction of apoptosis via MOMP.⁸  We assessed the sensitivity of mitochondria to MOMP by BH3^BIM peptides and found it was unaffected by TP-53 loss (supplemental Figure 3B). Taken together, these findings preclude direct induction of DNA damage or altered mitochondrial function, as explanations for why TP-53 loss confers a competitive advantage to leukemic cells treated for extended periods with venetoclax.

BH3-mimetics trigger activation of BAX/BAK, leading to MOMP, cytochrome c release, and then caspase activation.^22,23  We next investigated whether TP-53 loss could impair the integrity of early (BAX/BAK activation) vs later stages (caspase activation) of apoptosis. After venetoclax exposure, the magnitude of caspase activation was not affected by TP-53 status (supplemental Figure 3C). In contrast, levels of conformationally active BAX and BAK were both distinctly lower in TP53-deficient, compared with wild-type RS4;11 cells after a 6-hour treatment with venetoclax (Figure 4C).⁵  Of note, total BAX/BAK levels were unaffected by TP-53 loss (supplemental Figure 5C). At 6 hours, TP53-deficient cells also appeared relatively more resistant to venetoclax than TP53-wild-type counterparts. However, these differences diminished over time (16 hours, supplemental Figure 3D; 24 hours, Figure 3B). Overall, we conclude from our mechanistic experiments that deficient TP-53 impairs the earliest stage of apoptosis induction as measured by reduced activation of BAX and BAK, in response to sublethal concentrations of venetoclax.

Having demonstrated that TP-53 loss resulted in gain of fitness for leukemias exposed to venetoclax, we sought to determine whether this observation extended to inhibitors of other prosurvival proteins. We examined BH3-mimetics targeting MCL-1, as clinical trials have recently commenced in hematological malignancies.^23-26  To identify factors important for efficacy of MCL-1 inhibitors, a genome-wide CRISPR/Cas9 loss-of-function screening (Figure 5A) was conducted in mouse Eµ-Myc lymphoma-derived cell lines known to be highly MCL-1 dependent.^23,27  The top positively selected hits (loss of genes conferring resistance) with the MCL-1 inhibitor S63845 (MCL-1i) were sgRNAs targeting Trp53 and Bax (Figure 5B; supplemental Figure 4A).

To validate these hits, we deleted Trp53 in 3 independent Eµ-Myc lymphoma lines (supplemental Figure 4B) and confirmed resistance to MDM2 inhibition (supplemental Figure 4C). Sensitivity to MCL-1i was only minimally affected by Trp53 loss (supplemental Figure 4D). In contrast, deleting Bax rendered the lymphoma cells profoundly resistant to the MCL-1i (>20-fold increase in IC₅₀; supplemental Figure 4E). In long-term cell competition experiments, deletion of Trp53 favored outgrowth in all 3 B-cell lymphoma lines cultured with suboptimal doses of MCL-1i (Figure 5C).

Given these findings, we extended our experiments to human leukemia and lymphoma cell lines. In MOLM-13 cells, TP53 deficiency resulted in an approximate fivefold decrease in sensitivity to an MCL-1i, whereas no change in the IC₅₀ was apparent in BL2 Burkitt lymphoma or RS4;11 cells (supplemental Figure 4F-H). In extended duration competition assays, TP53-deficient MOLM-13 or BL2 cells outcompeted wild-type counterparts when exposed to IC₂₀ MCL-1i drug doses (Figure 5D). This finding was confirmed in vivo after transplantation of MOLM-13 cells into NSG mice and treatment with either MCL-1i or vehicle (Figure 5E). A competitive advantage for TP53-defective variants treated with MCL-1i was further confirmed in MV-4-11 and RS4;11 cell line models in vitro (supplemental Figure 4G-H).

As with venetoclax, there was no evidence that MCL-1 inhibition induced DNA damage in lymphoma cells (supplemental Figure 5A). However, after a 6-hour exposure to a sublethal concentration of MCL-1i, activation of BAX/BAK was impaired in TP53-deficient models of leukemia/lymphoma (Figure 5F; supplemental Figure 5B), akin to our observations with venetoclax.

TP-53 can enhance expression of BH3-only members NOXA and PUMA directly and BIM indirectly, to promote apoptosis.^28-31  Deficiency of TP-53 and reduced endogenous baseline BH3 activity may impair the efficacy of exogenously administered BH3-mimetic drugs, particularly at suboptimal doses. To confirm this, we generated Eµ-Myc lymphoma cells that were deficient in Noxa, Puma, and Bim (triple BH3-KO cells) and observed that after a 6-hour treatment with MCL-1i, these cells had impaired BAX/BAK activation, compared with wild-type cells (Figure 5F). Despite similar sensitivities of Noxa/Puma/Bim triple KO, wild-type and Trp53 KO cells to MCL-1i in the 24-hour viability assays (supplemental Figure 6A), a marked competitive advantage was evident for Noxa/Puma/Bim triple-KO cells, compared with either wild-type (supplemental Figure 6B) or Trp53 KO (supplemental Figure 6C) cells in long-term assays cultured in the presence of a sublethal concentration (IC₃₀) of MCL-1i. Of note, a competitive advantage for Noxa/Puma/Bim triple-KO cells was also apparent in control DMSO-treated cultures. This result may be related to MYC-induced predisposition to apoptosis,³²  which was reduced by depletion of all 3 BH3-only proteins.^31,33 

Collectively, these findings suggest that defective TP-53 may increase the threshold for BAX/BAK activation, particularly for sublethal doses of BH3-mimetics administered in single-agent format. We therefore explored the potential for enhancing response to BH3-mimetics by using a combination of drugs to target both BCL-2 and MCL-1 simultaneously, at doses considered individually suboptimal against TP-53 defective leukemia.

In clinical practice, venetoclax is approved for use in AML in combination with hypomethylating agents or low-dose cytarabine. Analogous to observations with BH3-mimetics, we noted relative expansion of TP53-deficient MOLM-13 cells in 14-day competition assays, despite continuous exposure to either cytarabine or decitabine (Figure 6A). Furthermore, the competitive advantage of TP53-deleted MOLM-13 cells persisted despite combination of either cytarabine or decitabine with sublethal (IC₂₀) concentrations of BH3-mimetics targeting either BCL-2 or MCL-1 (Figure 6B-C). In 7-day cell competition assays, BCL-2 or MCL-1 inhibitors (at 100 nM), when used as monotherapy, could not restrain outgrowth of TP53-deficient cells (Figure 6D). In contrast, when BCL-2 and MCL-1 inhibitors were combined at one-tenth the dose (10 nM), TP53-deficient and wild-type cells survived in equal proportions for 3 days, with both populations extinguished by day 7 (Figure 6D). These findings suggest that combined targeting of BCL-2 and MCL-1 may counteract the competitive advantage associated with TP53-deficient cells when exposed to BH3-mimetics as single agents at sublethal doses.

To model the therapeutic effects of combined BCL-2 and MCL-1 targeting in vivo, NSG mice were engrafted with either MV-4-11 or OCI-AML3 cells comprising an equal ratio of TP53-wild-type and -deficient cells (Figure 7A-B). In the MV-4-11 xenograft model, combined venetoclax/MCL-1i, resulted in significantly longer survival, compared with BH3-mimetics targeting either BCL-2 or MCL-1 alone (Figure 7A). In the OCI-AML3 model, combined BCL-2 and MCL-1 targeting improved survival to a similar extent in the TP-53 WT and KO cohorts, compared with vehicle treatment. This outcome suggests the combination was effective at overcoming the negative impact conferred by loss of TP-53 (Figure 7B).

Next, we examined the efficacy of dual BCL-2 and MCL-1 targeting in a PDX mouse model of TP53 Y126D mutant AML (loss of TP-53 function; ClinVar [VCV000265333.7]). After 10 days of therapy, suppression of BM disease burden was highest with combined BH3-mimetic therapy (Figure 7C).

To confirm these observations, a PDX with 2 TP-53 missense variants (TP-53 V173M and V143M; both nonfunctional, ClinVar [VCV000233951.8] and ClinVar [VCV000142657.5]) were treated with vehicle, venetoclax, an MCL-1i or combination BH3-mimetic therapy for 4 weeks. After 2 weeks of treatment, femoral aspirate was harvested from anesthetized mice and TP53 mutation burden assessed by targeted NGS (Figure 7D). The TP53 variant allele frequency (VAF) increased fourfold in cohorts receiving BH3-mimetic monotherapy, suggesting TP53 LOH had evolved for both variants. In contrast, combined venetoclax and MCL-1i therapy suppressed TP53 mutation burden (Figure 7D) and significantly prolonged survival (Figure 7E).

In contrast to BH3-mimetics targeting either BCL-2 or MCL-1 alone, which showed reduced activation of BAX/BAK in TP53 KO RS4;11 cells, a combined BH3-mimetic approach engaging both BCL-2 and MCL-1 simultaneously restored and enhanced early BAX/BAK activation (Figure 7F; supplemental Figure 7). Collectively, these findings show that combined targeting of BCL-2 and MCL-1 overcomes the early apoptotic defect-associated deficient TP-53, facilitating enhanced disease suppression and prolonged survival in diverse models of TP-53–defective AML.

BH3-mimetic drugs that directly trigger apoptosis in cancer cells reliant on BCL-2 or its prosurvival relatives have emerged as powerful agents for treating hematologic malignancies.⁵  After receiving regulatory approvals of venetoclax for CLL and AML, numerous studies^34-37  are exploring the clinical impact of venetoclax in other blood cancers. BH3-mimetic drugs targeting MCL-1 have also entered clinical trials for patients with AML, DLBCL, and plasma cell myeloma.²⁴ 

Clinical experience regarding the impact of TP-53 defects on response to venetoclax in AML and CLL have been paradoxical. In AML, despite a response rate of 55% in older patients with TP53-mutant disease treated with a combination of azacitidine and venetoclax, long-term survival remains poor.³⁸  At relapse, acquisition of a second TP53 mutation or loss of heterozygosity of the wild-type TP53 locus has been observed.¹¹  All these patients received venetoclax in combination with other cytotoxic drugs; therefore, it remains to be determined whether enrichment of TP53 mutation at progression is mediated by resistance to cytotoxic drugs, venetoclax, or both.

In this article, we present a patient who achieved a good clinical response after monotherapy with a BH3-mimetic targeting BCL-2, despite the AML harboring a TP53 P151H mutation and monosomy 17 (Figure 1A). Furthermore, the TP53-mutation burden also diminished with therapy. Although we cannot exclude the possibility that a coexisting IDH1 mutation conferred BCL-2 dependency, akin to that seen with mutant IDH2,³⁹  this case illustrated that TP-53 deficiency does not preclude responses to BCL-2–targeted therapy in AML. In contrast, examination of 4 other patients without a reduction in marrow blasts revealed interval expansion of TP53-mutant clones upon exposure to venetoclax relative to TP53 wild-type populations. This suggests that, in some cases, defective TP-53 may confer resistance to BCL-2–targeted therapy.

TP53 loss as a cause of venetoclax resistance was previously identified from a genome-wide CRISPR screen in AML.^7,8  Using a similar approach, we show for the first time an association between TP53 loss and resistance to MCL-1 inhibitors, indicating a potential class effect in relation to TP53 integrity and BH3-mimetic activity. These findings are unexpected, as the prosurvival targets of BH3-mimetics lie downstream of TP-53.

Using engineered TP-53–deficient leukemic models, we have made the surprising finding that TP-53 loss does not impair short-term killing upon exposure to either BCL-2 or MCL-1 inhibitors. Instead, we show a more subtle phenotype, with TP-53 deficiency only revealing a competitive advantage relative to wild-type counterparts after extended exposure to sublethal BH3-mimetics targeting either BCL-2 or MCL-1. In contrast, Bax deletion, also identified from the MCL-1i CRISPR screening, rendered cells profoundly resistant to BH3-mimetics (supplemental Figure 4E).

Mechanistically, the critical question is how TP-53 deficiency impairs the activity of BH3-mimetics which are thought to have their function downstream. We found no evidence that venetoclax activated upstream DNA damage signals sensed by TP-53. There was also no impact of TP-53 deficiency on basal mitochondrial respiration or venetoclax-induced suppression of oxidative phosporylation.^40,41  In contrast, we make the unexpected observation that TP-53 deficiency impairs the ability of a suboptimal dose of BH3-mimetics targeting either BCL-2 or MCL-1 to activate BAX and BAK, critical effectors of apoptosis. TP-53 can directly or indirectly induce the expression of several BH3-only proteins (NOXA, PUMA, and BIM)^28-31  Upon administration of venetoclax, endogenous BH3-only proteins already bound to BCL-2 may be released, allowing for activation of BAX and BAK. In the setting of TP-53 deficiency, we propose that the amount of endogenous BH3-only proteins that are released from BCL-2 by a sublethal dose of venetoclax is reduced, impairing efficiency of BAX or BAK activation by a suboptimal dose of venetoclax and raising the early apoptotic threshold.

To highlight the importance of endogenous BH3-only proteins regulated by TP-53, we showed that lymphoma cells triple-deficient in Noxa, Puma, and Bim also displayed impaired early activation of BAX and BAK in response to suboptimal BH3-mimetic doses, resulting in a competitive advantage in long-term assays over WT cells. Interestingly, triple BH3-only KO cells also outcompeted TP-53-deficient cells, perhaps because expression of BIM and even PUMA can be regulated via TP-53-independent processes.^31,42  Impaired BAX/BAK activation in TP-53 KO cells was remedied by combining sublethal doses of BH3-mimetics so that BCL-2 and MCL-1 could be targeted simultaneously. This strategy led to enhanced short-term activation of BAX/BAK and greater long-term control of TP-53-deficient AML. In contrast, BH3-mimetics combined with cytotoxic drugs failed to suppress the long-term competitive advantage conferred by TP-53–deficient cells.⁴³ 

Our results have several important clinical implications. First, prolonged exposure of TP-53-deficient malignant cells to a sublethal dose of a BH3-mimetic targeting either BCL-2 or MCL-1 should be avoided. Second, our results suggest that targeting BCL-2 and MCL-1 simultaneously with BH3-mimetics in TP53 mutant AML may be more effective than either drug in combination with conventional cytotoxic or hypomethylating agents. Third, if the activity of MCL-1 inhibitors in TP-53–deficient AML can be enhanced through combination with venetoclax, the need to maximize the dose of MCL-1 inhibitors to obtain a clinical response should be lessened, thereby widening the therapeutic window for this emerging drug class. Last, our findings illustrate the dual importance of TP-53 as a facilitator of optimal proapoptotic drug activity, not only via its role as a sensor of upstream cell stress (eg, DNA damage), but also via its role in regulating the earliest stages of apoptosis induction by BH3-mimetics targeting downstream prosurvival BCL-2 family members.

Our work provides a rationale for exploring BH3-mimetics in combination as an approach to subverting the competitive advantage associated with TP-53–defective leukemic cells and to enhancing both short, and also long-term outcomes in patients with TP53 mutant disease. Indeed, several phase 1b trials combining inhibitors of MCL-1 and BCL-2 in patients with AML are in progress, suggesting that this dual BH3-mimetic targeting approach may be clinically feasible (registered on www.clinicaltrials.gov as #NCT03672695, #NCT03797261).

The authors thank the patients who enrolled in the venetoclax clinical trials; S. Cory for helpful suggestions; M. Herold and A. Kueh for assistance with CRISPR/Cas9 screening; and S. Monard, N. Sprigg, S. Wilcox, C. White, and the Australasian Leukaemia and Lymphoma Group (ALLG) tissue bank for technical assistance and provision of vital samples and reagents. The visual abstract was created using BioRender.com.

This work was supported by fellowships and grants from the Australian National Health and Medical Research Council (NHMRC; Program Grants 1016701 [R.M.K., A.S., and D.C.S.H.] and 1113577 [A.W.R.]; Research Fellowships 1139607 [A.K.], 1140851 [D.A.S.], 1020363 [A.S.], 1079560 [A.W.R.], and 1156024 [D.C.S.H.]; Investigator Grants 1176175 [F.C.B.], 1174902 [A.W.R.], and 1157263 [C.B.]; Project Grants 1140906 [D.A.S.], 1162809 [A.H.W.], and 1086291 [G.L.K.]; and Ideas Grants 2002618 and 2001201 [G.L.K.]); the Leukemia and Lymphoma Society of America (Fellowship 5467-18 [R.T.]; Specialized Center of Research [SCOR] grant 7015-18 (R.M.K., A.S., A.W.R., A.H.W., G.L.K., and D.C.S.H.); a Cure Cancer and Cancer Australia grant (1186003 [R.T.]); Victorian Cancer Agency (Mid-career Research Fellowship [MCRF] 17028 [G.L.K.], MCRF Fellowship 19011 [D.M.M.], and grant 15018 [I.J.M., A.W.R., and A.H.W.); Cancer Council Victoria grants-in-aid 1086157 and 1147328 (G.L.K.), 1124178 (I.J.M.), and 1141740 (D.M.M.); Leukaemia Foundation of Australia (the Bill Long Charitable Trust PhD Clinical Scholarship [E.C., A.S., and G.L.K.]); fellowships from Novartis Foundation for Medical-Biological Research and the Swiss National Science Foundation (P400PM-180807 [S.S.G.]); the Felton Bequest (I.J.M.); the estate of Anthony (Toni) Redstone OAM (A.S. and G.L.K.); the Craig Perkins Cancer Research Foundation (G.L.K.); the Dyson Bequest (G.L.K.); Medical Research Future Fund grant 1141460 (A.H.W.); Tour de Cure Foundation (RSP-212-2020); and the Australian Cancer Research Foundation. This work was made possible through Victorian State Government Operational Infrastructure Support (OIS) and Australian Government NHMRC Independent Research Institute Infrastructure Support (IRIIS) Scheme.

Contribution: R.T., S.T.D., D.M., F.C.B., A.S., A.W.R., D.C.S.H., G.L.K., and A.H.W. designed the research; R.T., S.T.D., D.M., E.C., F.C.B., M.X.S., M.A.D., V.L., S.M., N.S.A., B.R., S.S.G., T.M.D., C.D.R., C.C., Z.X., T.M., G.P., C.B., and L.T. performed the research; C.B., S.W.L., B.J.A., R.M.K., A.S., M.S., S.B., A.K., D.A.S., and A.B. contributed vital new reagents or analytical tools; C.F., M.C., E.C., R.T., S.T.D., S.S.G., D.M., F.C.B., I.J.M., R.M.K., A.S., A.W.R., D.C.S.H., G.L.K., and A.H.W. analyzed and interpreted data; R.T., S.T.D., D.M., F.C.B., A.S., A.W.R., D.C.S.H., G.L.K., and A.H.W. performed the statistical analysis; R.T., S.T.D., D.M., A.S., A.W.R., D.C.S.H., F.C.B., G.L.K., and A.H.W. wrote the manuscript; and all authors read and approved the manuscript.

Conflict-of-interest disclosure: M.S. and S.B. are employees of Servier. R.T., S.T.D., E.C., C.F., M.X.S., M.A.D., T.M.D., C.D.R., C.C., Z.X., R.M.K., I.J.M., A.S., A.W.R., D.C.S.H., and G.L.K. are employees, and N.A. and A.H.W. are former employees, of the Walter and Eliza Hall Institute, which receives milestone and royalty payments related to venetoclax. A.H.W. and A.W.R. have received research funding from Servier and Abbvie. D.C.S.H. has received research funding from Genentech. A.S., A.W.R., D.C.S.H., G.L.K., and A.H.W. have received research funding from Servier. The remaining authors declare no competing financial interests.

The current affiliation for B.R. is the Department of Biochemistry and Molecular Biology, Biomedicine Discovery Institute, Monash University, Melbourne, VIC, Australia.

The current affiliation for B.J.A. is the Department of Pediatric Oncology, Dana-Farber Cancer Institute, Boston, MA.

Correspondence: Andrew Wei, Australian Centre for Blood Diseases, Monash University, 99 Commercial Rd, Melbourne, VIC 3004, Australia; e-mail: andrew.wei@monash.edu; Gemma Kelly, Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, Melbourne, VIC 3052, Australia; e-mail: gkelly@wehi.edu.au; Fiona Brown, Australian Centre for Blood Diseases, Monash University, 99 Commercial Rd, Melbourne, VIC 3004, Australia; e-mail: fiona.brown2@monash.edu; and David Huang, Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, Melbourne, VIC 3052, Australia; e-mail: huang_d@wehi.edu.au.